Freely Available British and Irish Public Legal Information
[Home] [Databases] [World Law] [Multidatabase Search] [Help] [Feedback]
Scottish Court of Session Decisions
You are here: BAILII >> Databases >> Scottish Court of Session Decisions >> McTear v. Imperial Tobacco Ltd [2005] ScotCS CSOH_69(2) (31 May 2005)
URL: http://www.bailii.org/scot/cases/ScotCS/2005/CSOH_69(2).html
Cite as: [2005] ScotCS CSOH_69(2)
[New search] [Help]
McTear v. Imperial Tobacco Ltd [2005] ScotCS CSOH_69(2) (31 May 2005)
[5.189] Professor Sir Richard Doll, aged 91, had not provided a CV and reference was made instead to items in his entry in Who's Who. He said that he was still at work and gave lectures. He had recently returned from lecturing in Japan. He was awarded the OBE in 1956. He was made a Knight in 1971 and a Companion of Honour in 1996 for services to medicine. He became a Fellow of the Royal Society in 1966. His current post was Honorary Consultant, Cancer Research UK Cancer Studies Unit, Radcliffe Infirmary, Oxford, a post he had held since 1983. He was an honorary member of the Clinical Trial Service Unit and Epidemiological Studies Unit, Nuffield Department of Clinical Medicine, University of Oxford. He was Warden of Green College, Oxford from 1979 to 1983, the first head of this college. He had been instrumental in its establishment as a college with a special interest in clinical medicine. He qualified in medicine at St Thomas's Hospital Medical School, University of London. In the course of his subsequent career he was Regius Professor of Medicine at the University of Oxford from 1969 to 1979. Previously he was Teacher in Medical Statistics and Epidemiology at University College Hospital Medical School, London, from 1963 to 1969. He explained that he had been an epidemiologist from 1946 onwards, which required some knowledge of medical statistics, and this was an honorary title given to him when he took his Medical Research Council Statistical Research Unit to University College Hospital Medical School. He had been a member and chairman of a number of committees and sub-committees and had received a number of honours. He had been awarded honorary degrees by fourteen or fifteen universities. Among honorary fellowships, he was an Honorary Fellow of the Royal College of General Practitioners, the Royal College of Surgeons and the Royal College of Radiologists. He was a Fellow of the Royal College of Physicians. He had been awarded a number of prizes and gold medals, including the Gold Medal of the British Medical Association and the Royal Medal of the Royal Society. He said that he had not published many books, but had published just over 500 articles, mainly on the aetiology of lung cancer, leukaemia and other cancers. Some of his early papers were on clinical therapeutics, because he worked as a clinician as well as an epidemiologist for about twenty years.
[5.190] Sir Richard was next asked to consider IARC 1986. He agreed that this monograph was more than 400 pages long. He was chairman of the working group which wrote it. He did not know how the members were selected. IARC chose people on the advice of its own staff, and sought to have representatives from many different countries. He knew personally about half the members of the working group. They were leaders in their field. DL Davis, Director of the Tobacco and Health Research Institute at the University of Kentucky, was the principal person to have done chemical analyses of tobacco smoke in the United States. N Gray was Director of the Anti-Cancer Council of Victoria, Australia, which had been working for many years on publicising the harmful effects of smoking and getting the Victorian Government to do something about it. HJ Evans, of the MRC Clinical and Population Cytogenetics Unit at the Western General Hospital, Edinburgh, was a leading biologist working on chromosomes. Y-T Gao, Director of the Shanghai Cancer Institute, had done some work in confirming the same findings in China as in Britain. T Hirayama was a leading Japanese epidemiologist and had organised a very big cohort study, bigger than the one Doll and Hill had organised on British doctors, somewhat smaller than the big American studies, on some 280,000 Japanese residents, getting details of their smoking habits and following them up to find out what diseases they developed. AB Miller, Director, Epidemiology Unit, National Cancer Institute of Canada, at the University of Toronto, was the leading epidemiologist in Canada. S Moolgavkar was a statistician in the United States, very interested in time-exposure relationships between cigarette smoking and lung cancer. NP Napalkov was Director of the Research Institute of Oncology in Leningrad, and was a basic scientist whom Sir Richard knew well. R Peto was Sir Richard's colleague at Oxford and was now Director of the Clinical Trial Service Unit and Epidemiological Studies Unit, Nuffield Department of Clinical Medicine, University of Oxford. MAH Russell, Reader in Addiction, Institute of Psychiatry, The Maudsley Hospital, London, was a psychiatrist with a special interest in addiction. L Teppo, of the Finnish Cancer Registry, was a leading epidemiologist from Finland. NJ Wald, Department of Environment and Preventive Medicine, the Medical College of St Bartholomew's Hospital, London, had been a member of Sir Richard's department in Oxford and became Professor of Preventive Medicine at St Bartholomew's. EL Wynder, President of the American Health Foundation in New York had published a report with very similar findings to those of Doll and Hill 1950, associating cigarette smoking with lung cancer, though in that initial report he only suggested that it might be a cause, he did not regard the evidence as conclusive. Sir Richard was unable to recall some of the other members of the working group.
[5.191] Sir Richard agreed that the working procedures of IARC were as set out at pp.16-17 of IARC 1986 quoted in Professor Friend's evidence at [para.37]. One of the topics considered by the working group was epidemiological studies of cancer in humans. In their conclusions and evaluations, at p.312, they stated:
"Lung cancer is believed to be the most important cause of death from cancer in the world, with estimated total deaths in excess of one million annually. The major cause of the disease is tobacco smoking, primarily of cigarettes. Risk of lung cancer is particularly dependent on duration of smoking; therefore, the earlier the age at initiation of smoking, the greater the individual risk. Further, the longer the time period during which a major proportion of adults in a population have smoked, the greater the incidence and mortality from the disease in that population. Risk of lung cancer is also proportional to the numbers of cigarettes smoked, increasing with increasing cigarette usage. In populations with a long duration and heavy intensity of cigarette usage, the proportion of lung cancer attributable to smoking is of the order of 90%. This attributable proportion applies to men in most western populations; in populations in which women are increasingly using cigarettes, the attributable proportion in women is also approaching this level."
Sir Richard said that this was the conclusion of all twenty-seven members of the working group, with no dissent.
[5.192] At p.314 the working group set out their evaluations:
"There is sufficient evidence that inhalation of tobacco smoke as well as topical application of tobacco smoke condensate cause cancer in experimental animals.
There is sufficient evidence that tobacco smoke is carcinogenic to humans.
The occurrence of malignant tumours of the respiratory tract and of the upper digestive tract is causally related to the smoking of different forms of tobacco [...]."
Asked whether, after all the work they had done, this was quite a modest way of putting their conclusions, Sir Richard said that it made them clear for any reader to understand. So far as he was aware there had never been any challenge to these evaluations.
[5.193] Asked to comment on IARC 2004, Sir Richard explained that the monograph had not yet been published: publication was expected in early 2004. He was aware that a press release had been published, but said that he had to claim ignorance that part of the report, containing the summary of data reported and evaluation, was also available. He agreed that it looked as if this part of the report had been published on the Internet, which was something he was not really accustomed to dealing with himself. He agreed that the statement in IARC 2004 that the major cause of lung cancer was tobacco smoking, primarily in cigarettes, and that in populations with prolonged cigarette use, the proportion of lung cancer cases attributable to cigarette smoking had reached 90%, was consistent with IARC 1986. His attention was drawn to para.5.3, in which it was stated:
"The most compelling evidence for a positive carcinogenic effect of tobacco smoke in animals is the reproducible increase observed in several studies in the occurrence of laryngeal carcinomas in hamsters exposed to whole tobacco smoke or to its particulate phase."
He said that this was part of the material that he dealt with. At para.5.4 it was stated:
"Tobacco smoking is addictive, and nicotine has been established as the major addictive constituent of tobacco products."
He said that this was a conclusion which the working group had reached, though it was not a new one. In the evaluation at para.5.5 it was stated: "There is sufficient evidence in humans that tobacco smoking causes cancer of the lung [...]", and the overall evaluation was that: "Tobacco smoking and tobacco smoke are carcinogenic to humans." Peto was the only person whom Sir Richard could recall, apart from himself, who had been involved in the preparation of both IARC 1986 and IARC 2002. All the members of the working group which produced the latter were leading people in their fields, he said.
[5.194] Sir Richard's attention was next directed to a biennial lecture delivered by him at Green College in 1997, entitled "Tobacco: A medical history", the text of which was reproduced in UKHC 2000, Vol.II, pp.19-35 as Appendix 1 to a memorandum by the Health Education Authority: Doll 1997. In this lecture Sir Richard reviewed the medical history of tobacco, with particular reference to the question whether cigarette smoking caused lung cancer. Asked about the current teaching in medical schools and medical textbooks, he said that he had not read a recent medical textbook, but he knew from talking to senior colleagues that they taught the extreme dangers of cigarette smoking, and its importance as a cause of lung cancer.
[5.195] He described 1950 as a watershed, because five case-control studies were reported in that year. All of them showed a close association with smoking. Two studies stood out because of their size. One had been initiated by Wynder in 1948, on the basis of knowledge that the burning of tobacco would lead to the formation of cancer-causing chemical compounds. Analysis of results obtained from interviewing patients led to the conclusion that excessive and prolonged use of tobacco, especially of cigarettes, seemed to be an important factor in the induction of bronchogenic cancer. The results were published by Wynder and Graham in 1950.
[5.196] Of the other, which was published by himself and Professor Bradford Hill, also in 1950, he said (at p.23):
"In the other, which had been initiated by the British Medical Research Council's conference in 1947, detailed consideration of the possibility of confounding, the consistency of the findings in different studies, the biological relationships with amount and duration of smoking, the size of the estimated relative risk, and the relationships over time and place and for each sex led the authors to conclude that (I quote) 'cigarette smoking is a factor, and an important factor, in the production of carcinoma of the lung' (Doll & Hill, 1950).
Reaction to findings
This conclusion was accepted by Sir Harold Himsworth, who had become secretary of the Medical Research Council, but not generally by medical or statistical scientists and certainly not by the British Department of Health's Standing Advisory Committee on Cancer and Radiotherapy [...]. Most accepted that an association had been shown, but not that it implied cause and effect. Some, however, were even more sceptical, including Berkson [...] the leading American medical statistician who suggested that the findings were an artefact due to the combination of lung cancer and smoking leading to a greater chance of a patient's admission to hospital than when the disease occurred in a non-smoker. Other sceptics were the representatives of the tobacco industry, who, in Britain, sought an interview with the Medical Research Council and were referred to Professor Hill. The conclusion that cigarette smoking was a cause of disease was, they argued, unsustainable for three reasons: the international correlation between cigarette consumption and the mortality from lung cancer of about 0.5 was too low, smoking histories were too unreliable to use as a basis for an association with disease, and lung cancer, in any case, was obviously due to atmospheric pollution. To this Hill replied that a correlation of the size observed with crude international statistics was, in his experience, unusually high and supported a causal relationship rather than the reverse: that if smoking histories were unreliable, this would have weakened a true association rather than have created a false one; and if they had thought that atmospheric pollution was the main cause of lung cancer they should go away and prove it, for Hill and I couldn't."
[5.197] Sir Richard said that the theory that atmospheric pollution was the main cause of lung cancer was the one which was most widely held when the study was started and one that he and Hill were very conscious of trying to test in it. About twenty years earlier the idea that tobacco might be a cause had been considered and some English pathologists had tried unsuccessfully to produce cancer in animals with tobacco tar and decided that tobacco could not be a cause of human lung cancer because of this. It was subsequently shown they had not applied the tar for long enough to produce the effect. He and Hill kept their minds open to all the things that they thought could possibly be an explanation and tried to investigate them all. Very rapidly it became clear that the only one that stood up was tobacco smoking. At the outset, if he had to put money on it, he would have said it was something to do with motor cars and road tar, which was known to contain powerful human carcinogens. In their survey they could not find any relationship whatsoever with frequency of exposure to car exhausts or length of time spent on roads.
[5.198] In the lecture, at p.24, Sir Richard discussed the evidence that led to wide acceptance of major harm from smoking. He started by referring to the early cohort studies. He stated:
"Evidence of a different type was, however, clearly needed, if reactions were to be changed, as, for example, by recording the smoking habits of large numbers of people and following them up to see if the risk of lung cancer could be predicted from the information about the individual's level of smoking."
Sir Richard said that Professor Bradford Hill, who taught him epidemiology, taught him that the scientist's responsibility was always to try to disprove his own work, not to try to look for evidence in support of it. He had the idea that if they learnt about the smoking habits of British doctors and divided them according to the amount smoked, they would see if they could disprove the relationship between cigarette smoking and lung cancer by following up the non-smokers and the heavy smokers. Of course, he said, the opposite happened: it greatly strengthened the conclusion. The study concluded on 1 November 2001, fifty years after they started obtaining information from British doctors and following them up, with frequent reports about changes in smoking habits over time.
[5.199] The lecture continued:
"The evidence from the 'cohort' study of British doctors mounted quickly, and within two and a half years the findings with regard to lung cancer had confirmed those predicted from the case-control studies. This is shown in Table 8, which gives the relative mortality rates for different levels of smoking, as estimated from the final results of the British case-control study based on 1,357 deaths from lung cancer in men [...], and the first results of the cohort study based on only 36 such deaths [...]. With so few deaths in this second study, the confidence limits of the mortality rates were wide, but even so the trend in mortality with smoking was significant [...].
Altogether, however, 789 deaths had been recorded and it was possible to examine the relationship between smoking and several other diseases. [...]
Two years later these results were confirmed with larger numbers [by Doll and Hill, writing in 1956]. More importantly, they were also confirmed in the much larger study that the American Cancer Society had started in 1952 specifically, as the principal investigator told me, to disprove the relationship between smoking and lung cancer that had been observed in the case-control studies [...]."
Sir Richard said that this principal investigator was Hammond, who believed that the relationship was incorrect and strongly believed in the disproof, but within a couple of years he found he was wrong and admitted this. The lecture continued: "The results, based on nearly 5,000 deaths in the 190,000 American men followed for two years, are shown in Table 9 for lung cancer [...]." These investigators concluded inter alia that regular cigarette smoking caused an increase in the death rate from cancer of the lung.
[5.200] The next passage of the lecture related to proof of causation. It started:
"The conclusion that cigarette smoking was a major cause of the disease had not been easy to accept, as the evidence was observational in humans and unconfirmed by animal experiment."
Sir Richard said that this was the position in the 1950s. There had been animal experiments since then which had shown that tobacco tars were carcinogenic. The direct demonstration of carcinogenicity to the lung had not really been possible because experiments which had been started were stopped on humane grounds. They did not wish to expose the animals to tobacco smoke, it was considered improper to do so. But that had never been a significant objection to a carcinogen being a cause of cancer. IARC had on several occasions concluded that a carcinogen caused cancer in humans without experiments to produce the same cancer in animals.
[5.201] The lecture continued by stating that two leading statisticians remained unconvinced. One of these, in the USA, was Berkson, who wrote in 1958 that he was disturbed that the relationship with smoking held to some extent across the board with a variety of conditions.
"In Berkson's opinion this raised the suspicion that there must be something wrong with the method of enquiry and he suggested that they were the result of the interplay of various subtle and complicated biases or that they had a constitutional basis, people who were non- or relatively light smokers, being the kind who were biologically self-protective and that this (I quote) 'correlated with robustness in meeting mortal stress from disease generally.'
In making this criticism, Berkson [...] took no account of the great difference in the relative risks of different diseases among heavy cigarette smokers compared to non-smokers, varying in Doll & Hill's [...] study from 24 to one for lung cancer to 1.01 to one, [nor] of the fact that tobacco smoke was not a pure chemical entity, but a mixture of many chemicals, subsequently shown to number more than 4,000."
Sir Richard added that of course, as we knew now, there were other specific chemicals in tobacco smoke that specifically caused some of the cancers that were caused by smoking, so it was scientifically perfectly straightforward now that smoking did have so many effects.
[5.202] The lecture continued:
"In the UK, Fisher, the most eminent theoretical statistician worldwide, was disturbed that the original finding (Doll & Hill, 1950) that smokers with lung cancer reported inhaling less often than smokers without the disease [...] weighed against causation, unless it were also concluded that (I quote) 'inhaling cigarette smoke was a practice of considerable prophylactic value in preventing the disease' [...] and he argued that secular changes in smoking habits could not be related to the increase in lung cancer since 'lung cancer has been increasing more rapidly in men relatively to women' and that 'it is notorious, and conspicuous in the memory of most of us, that over the last 50 years the increase of smoking among women has been great, and that among men (even if positive) certainly small' [...].
Neither objection was valid. The effect of inhaling was impossible to predict without knowing where the smoke droplets would be deposited and this was uncertain because tobacco aerosols swell under warm and moist conditions and might, if inhaled deeply, deposit in the alveoli rather than on the bronchi [...]. Doll & Hill [...], moreover, found that while inhaling was associated with a diminished risk of cancer in the large bronchi, it was associated with an increased risk of developing cancer in the periphery of the lung, which made biological sense."
Sir Richard explained that most lung cancers occurred near to where the bronchi separated from the trachea at the top end of the lungs, but some occurred right out in the very small bronchi near the periphery. When reviewing the records of all the patients with lung cancer, he had kept a note of whereabouts in the lungs the cancers occurred. When they did their second analysis with a greater amount of data, they found that when subjects said that they inhaled, this was associated with a greater risk of cancer right out in the periphery, in the small bronchi approaching the alveoli, whereas the opposite was true of the cancers right up near the trachea in the big bronchi.
[5.203] The lecture continued:
"As for the evidence of secular changes, Fisher [...] was just wrong; for he had ignored the cohort effects whereby the risks among successive cohorts are determined not only by their recent smoking history but also by their smoking habits in the distant past. When comparisons are made at appropriate ages and times, the trends in the sex ratio of the disease mimic the trends in cigarette consumption by sex over the relevant periods [...]."
Sir Richard said that Peto and he showed that very clearly in some later analyses they were doing for another purpose.
[5.204] Sir Richard said that he understood that Fisher had changed his mind just before he died. A friend of Sir Richard's, Sir Walter Bodmer, now Principal of Hertford College, Oxford, and previously the head of the Imperial Cancer Research Fund, had been a statistical student of Fisher's and he called on Fisher shortly before he died in Australia. According to Bodmer, Fisher said that he recognised that he had been wrong and was intending to publish a statement to that effect, accepting that cigarette smoking was a co-factor in the production of lung cancer. It was the evidence that had led him to change his mind.
[5.205] The lecture continued, at p.25:
"Difficulty in reaching a conclusion about a causal interpretation of the evidence also arose, because different people gave different meanings to 'cause'. In saying that a particular factor is a cause of disease, epidemiologists have in mind a situation in which, for example, prolonged cigarette smoking results in a rare disease becoming 10 times as common as it would have been in the absence of smoking. Cigarette smoking is not then a necessary cause nor a sufficient cause; but it can be an important cause (as few people would have developed the disease if they had not smoked) and this is not contingent on the absence of other causes. What was claimed was that for several diseases causation in the sense described was proved beyond reasonable doubt. The detailed evidence that led to this claim has been reviewed many times and I note here only the extraordinary strength of the association with lung cancer, with increased risks of more than 20 fold in heavy cigarette smokers which alone made the alternative explanation of confounding virtually impossible, the diminution of risk with cessation of smoking, and the consistency of the findings with different methods of investigation and in different countries and different cultures."
Sir Richard said that it was his position that causation had been proved.
[5.206] The lecture continued:
"During the 1950s, this epidemiological evidence, which had been supplemented by many other studies, was supported by the experimental demonstration that tobacco tars were carcinogenic when applied regularly for a long time to the skin of laboratory animals [...] and by the identification of known carcinogens in tobacco smoke [...]. Expert committees appointed to review the evidence were consequently able to reach positive conclusions. Between 1956 and 1959, the Netherlands Ministry of Social Affairs and Public Health (1957), the British Medical Research Council (1957), a study group appointed jointly by the US National Cancer Institute, the National Heart Institute, and the American Cancer Society (Study Group on Smoking and Health, 1957), the Swedish Medical Research Council (1958), and the US Public Health Service (Burney, 1959) all reported that cigarette smoking was a cause of lung cancer, and a year later an expert committee of the World Health Organization (1960) did so too."
So beside Sir Richard's own cohort studies, he said, these various bodies were reaching the same conclusion.
[5.207] The lecture then turned to the topic of the public acceptance of causality. In it Sir Richard said:
"Despite their provenance these reports had little lasting impact on the general public and the situation did not change materially until after the reports by the Royal College of Physicians of London in 1962 and the Advisory Committee to the US Surgeon General in 1964. The first was short and aimed at interested laymen. The second was long and detailed and was particularly newsworthy, because the tobacco industry had been privileged to veto any member of the Committee who had publicly expressed any views about the subject. Both reports nevertheless agreed that smoking was a major cause of lung cancer. [...]
Following these reports, the idea that smoking was a major cause of lung cancer ceased to be seriously challenged. Even the tobacco industry in the UK agreed not to deny the causal relationship on the advice of Geoffrey Todd, their senior statistician. Todd had been a representative of the industry who had visited Doll and Hill in 1952 and had sought to persuade them that their conclusion was wrong; but he had become convinced that it was right. In the USA, however, the industry continued to maintain that all that had been shown was a statistical association and the causality had not been scientifically proven: that is, until recently when the smallest manufacturer broke ranks and accepted that smoking was a cause of the disease."
[5.208] Asked when it was generally accepted in the scientific community that cigarette smoking was a cause of lung cancer, Sir Richard said that this was in the late 1950s, by the end of the reports that he had referred to. The MRC 1957 effectively settled the issue as far as England was concerned. But the public of course did not pay much attention to these reports, because, when the media reported the conclusions of, say, the Medical Research Council, they also always reported a statement by some representative of the tobacco industry that the issue was controversial and that it could not be regarded as proven. This of course was very misleading to the general public. It was not until the early 1970s, 1972 to 1973, when the media became convinced and, for example, television announcers would no longer smoke cigarettes, that the British public started reacting much more sharply to the evidence and a lot more people started giving up smoking.
[5.209] Asked whether ITL had scientists who were advising on these issues in the 1950s, Sir Richard said that he did not know who they had apart from their statistician, Todd. They should have had scientists, he could not recall their names. They did not have any epidemiologists, certainly.
[5.210] At p.26 of the lecture, in a passage relating to the current knowledge of the effects of cigarette smoking, under the heading "Harmful effects" Sir Richard said:
"The morbid effects that are caused in part by cigarette smoking are listed in Tables 10-13. Those that are five or more times more common in cigarette smokers than in non-smokers are marked with an asterisk".
He was asked to look at Table 10, in which there was an asterisk against lung cancer (among other cancers). Under the heading "Total effect on risk of death" Sir Richard said:
"In retrospect, it may be surprising that resistance to the idea that smoking caused so much disease was initially so strong. Three factors, at least, contributed to it. One was the ubiquity of the habit, which was as entrenched among male doctors and scientists as among other men and had dulled the sense that tobacco might be a major threat to health. Another was the novelty of the epidemiological techniques, which had not previously been applied to any important extent to the study of non-infectious disease. The findings were consequently undervalued as a source of scientific evidence. A third was the primacy given to Koch's postulates for determining causation. The evidence that lung cancer occurred in non-smokers, was consequently taken to show that smoking could not be the cause and the possibility that it might not be a cause was inappropriately doubted. The manner in which lung cancer was linked to smoking was not, however, unique. All the other major diseases related to smoking were found to be so by epidemiological enquiry and laboratory evidence of physiological effects that provided plausible mechanisms by which smoking might cause them was obtained only later and, in some instances, is still awaited."
Sir Richard said that laboratory evidence was not available. He stood by everything that he had said in the lecture and by all the articles that he had written relating to lung cancer and its association with smoking.
Cross-examination of Sir Richard Doll
[5.211] Mr Jones started his cross-examination of Sir Richard by asking him about a passage in USSG 1964, at p.20:
"Statistical methods cannot establish proof of a causal relationship in an association. The causal significance of an association is a matter of judgment which goes beyond any statement of statistical probability."
Sir Richard said that this statement was correct and he agreed that it was a question of judgment whether a causal relationship was proved beyond reasonable doubt, on the basis of all the evidence. A conclusion had to be formed on the evidence as a matter of judgment, as in a court of law. In addition to the strength of the association all other relevant evidence had to be considered by an epidemiologist, in the same manner as a judge would do.
[5.212] Sir Richard said that he and Hill started thinking about their retrospective study in 1947, began work on it in 1948 and began writing it up in 1949, before it was published as Doll and Hill 1950. He gave up smoking in 1949 because he thought that the evidence that cigarette smoking caused lung cancer was sufficiently strong. He stopped without difficulty after having been a smoker for about twenty years. Sir Richard concluded in 1950 that smoking causes lung cancer in women. His wife smoked about forty cigarettes a day and found it very difficult to stop. She reduced the amount regularly year after year and still ended up smoking about forty a day. She eventually stopped, several years later; Sir Richard helped her by bribing her.
[5.213] Sir Richard was asked about Doll 2000, in which he gave an account of the work which led to the publication of Doll and Hill 1950 and its reception at the time. At p.5 he wrote:
"The paper obtained much less publicity than we had expected. Senior clinicians and cancer research workers advising the Ministry of Health were, for the most part, unconvinced of the causal relationship and they advised against publicizing the findings for fear of scaring people. Further evidence of a different type was clearly needed if our conclusion was to be taken seriously and this, we thought, could be obtained by seeking information about people's smoking habits and then following them up to see if the mortality from lung cancer varied, as we predicted it would, with the amount they smoked."
Sir Richard said that this was how he recollected it.
[5.214] He was next asked about Doll 1998, in which he wrote, at p.98:
"To Doll and Hill, it seemed clear that evidence of a different type would have to be obtained, if [critical] reactions to the findings in the case-control studies were to be changed."
Accordingly, they decided to obtain evidence from a prospective "cohort" study of British doctors. At p.99 he wrote that with the publication of the results of prospective studies by Doll and Hill in 1954 and 1956 and by Hammond and Horn in 1954, "scientific opinion rapidly changed". Two leading statisticians, however, remained unconvinced. Berkson in the USA had initially suggested that the first results of the two studies might be biased. In Sir Richard's view, Berkson was wrong. At p.100 he wrote:
"Fisher, in the UK, was the other outstanding statistician who questioned a causal interpretation. In his view, Doll and Hill's original finding that reports of inhaling by smokers with lung cancer (62% of whom reported inhaling) were less common than by smokers without lung cancer (67% of whom did so) weighed heavily against causation, unless it were also concluded that 'inhaling cigarette smoke was a practice of considerable prophylactic value in preventing the disease'. He preferred instead the idea that there was some common factor that was responsible both for the individual's smoking habits and his risks of developing the disease, which Fisher postulated was genetic, and he supported his hypothesis by showing that the smoking habits of monozygous pairs of twins were more similar than those of dizygous pairs and (on small numbers) appeared to be similar irrespective of whether they had been raised together or apart. [...] Neither objection was, in fact, valid."
At p.103 he referred to RCP 1962 and USSG 1964 and wrote:
"Following these reports, the idea that smoking was a major cause of lung cancer ceased to be seriously challenged, except by the tobacco industry outside the UK (where it had been quietly accepted) and by a few eccentric individuals such as Burch who, however, raised no material objections."
Asked by counsel for the identity of other "eccentric individuals", Sir Richard spoke of an Irishman, an American and an English psychiatrist; he could not remember their names or those of any of the others. He said that by "material objections" he meant objections which most scientists took seriously. Sir Richard agreed that this passage was similar to that in Doll 1997 at p.25, referred to earlier.
[5.215] Sir Richard was next asked about Sir Ronald Fisher. He said that he had done some genetic work but no epidemiological work. Fisher had suggested that the results of the epidemiological studies could be explained by there being some genetic factor which caused people both to want to smoke and, quite independently, cause cancer of the lung. Sir Richard said that it seemed a pretty bizarre idea to him and to a lot of people. He was not aware of any epidemiologist, anybody experienced in his field, who took it seriously. There could have been some American geneticists who took him seriously, but how they came to do it he could not imagine.
[5.216] Fisher 1959 was a pamphlet incorporating letters by Sir Ronald Fisher to the editors of the British Medical Journal and Nature in 1957 and 1958, two lectures published by the Centennial Review in 1958 and a previously unpublished passage on inhaling. Sir Richard agreed that Fisher had achieved a formidable reputation amongst statisticians for his pioneer work in this field during the previous forty years. His particular achievement had been in the development of statistical methods appropriate to biological research. During his brilliant career in academic and research work, many honours had come to him. He had been awarded the Royal, Guy, Darwin and Copley Medals of the Royal Society, of which he was a Fellow. He was a Foreign Associate of the United States National Academy of Science, a Foreign Honorary Member of the American Academy of Arts and Sciences, a Foreign Member of the Royal Swedish and Royal Danish Academies of Sciences, and a Foreign Member of the American Philosophical Society. He held degrees from the Universities of Ames, Chicago, Harvard, Calcutta and Glasgow; he was a Fellow of Gonville and Caius College, Cambridge, and a former Arthur Balfour Professor of Genetics in the University of Cambridge. He was a very eminent geneticist. He had also been Galton Professor of Eugenics in University College, London. He had a reputation for frank and outspoken contributions to many statistical debates.
[5.217] Sir Richard disagreed with the statement about Sir Ronald, under the heading "The author" (in a passage from which most of the foregoing information is derived): "This pamphlet is a fair-minded assessment of the value of the statistical evidence relating to the incidence of lung cancer in smokers." Sir Ronald thought that he himself was "fair-minded", but nobody else did. He was the leading theoretical statistician of his day, and an eminent geneticist, but that did not make him right. In his researches Fisher had obtained data from Doll and Hill's 1950 case-control study but, according to Sir Richard, he would not accept the data from the 1952 publication.
[5.218] At p.21 Fisher wrote:
"Before I stop, in fact, I hope I shall make it clear that there is a case for further research, and I shall only mention two areas which would seem to be profitable for investigation. I would stress the importance of what could be done comparatively easily with rather little expense, namely, to ascertain unmistakably what the facts are about inhaling. If inhaling is found to be strongly associated with lung cancer, it would be consonant with the view that the products of combustion, wafted over the surface of the bronchus, might induce a pre-cancerous and thence a cancerous condition. But if there is either no association at all or a negative association, we should have to reject altogether that simple theory of the causation of cancer."
Sir Richard said that of course Fisher was ignoring a lot of published data and he was quite wrong in making that conclusion in any case. He refused to look at the data when he made that remark. Sir Richard agreed, however, that this was the view that Fisher expressed.
[5.219] At p.22 Fisher wrote:
"For my part, I think it is more likely that a common cause supplies the explanation. Again, we do not know. I do not put forward any explanation as proved, but as requiring investigation. The obvious common cause to think of is the genotype. We are all different genotypes. [...] If one studies cancer in mice [...], if one examines any of the many [...] inbred lines of mice [...] - if you take, then, any two such lines of differing genotypes, they will, I believe, invariably be found to differ in the frequency, in the age incidence, and in the type of cancer which those mice suffer from. Consequently if there is any genotypic difference between the different smoking classes, we may expect differences in the type or frequency of the cancers that they display."
Fisher then went on to develop a proposal for a line of research directed to the possibility that there was a genetic component which distinguished the different smoking classes: non-smokers, cigarette smokers, pipe smokers and cigar smokers. Sir Richard said that this was a suggestion which Fisher was entitled to make, but he regarded it as a long shot. He himself did not think that it was worth researching but research was done to test it. His view was that it was not a sufficiently worthwhile proposal to spend money on. He and Hill had found, for example, that when doctors gave up smoking the incidence of lung cancer in the population was reduced. If it had been a genetic factor, the fact that some gave up smoking would not have affected the overall incidence. So they had evidence quite soon that it just was not a worthwhile topic to study. He had concluded in his own mind that it was smoking that caused lung cancer.
[5.220] At p.39, Fisher wrote:
"The association observable between the practice of cigarette smoking and the incidence of cancer of the lung, to which attention has been actively, or even vehemently, directed by the Medical Research Council Statistical Unit, has been interpreted, by that Unit, almost as though it demonstrated a causal connexion between these variables.
The suggestion, among others that might be made on the present evidence, that without any direct causation being involved, both characteristics might be largely influenced by a common cause, in this case the individual genotype, was indeed rejected with some contempt by one writer, although I believe that no one doubts the importance of the genotype in predisposing to cancers of all types.
It seemed to me that, although the importance of this factor had been overlooked by the Unit in question, it was well within the capacity of human genetics, in its current state, to examine whether the smoking classes to which human beings assign themselves, such as non-smokers, cigarette smokers, pipe smokers, cigar smokers, etc., were in fact genotypically differentiated, to a demonstrable extent, or whether, on the contrary, they appeared to be genotypically homogeneous, for only on the latter view could causation, either of the disease by the influence of the products of combustion, or of the smoking habit by the subconscious irritation of the postulated pre-cancerous condition, be confidently inferred from the association observed."
Sir Richard said that this was a view which Fisher was entitled to take and, when it was taken, it was shown to not be relevant.
[5.221] Fisher continued on pp.39-40:
"The method of inquiry by which such differentiation can be recognised is the same as that by which the congenital factor has been demonstrated for several types of disease, namely, the comparison of the similarities between monozygotic (one-egg) and dizygotic (two-egg) twins respectively; for any recognizably greater resemblance of the former may be confidently ascribed to the identity of the genotypes in these cases."
Sir Richard said that this was a view that Fisher was entitled to take. Fisher then referred to information provided by researchers in Germany about the results of an inquiry into the smoking habits of adult male twin pairs. He wrote, on p.40, that in all cases the monozygotic twins showed closer similarity and fewer divergences than the dizygotic. He continued:
"There can therefore be little doubt that the genotype exercises a considerable influence on smoking, and on the particular habit of smoking adopted, and that a study of twins on a comparatively small scale is competent to demonstrate the rather considerable differences which must exist between the different groups who classify themselves as non-smokers, or the different classes of smokers. Such genotypically different groups would be expected to differ in cancer incidence; and their existence helps to explain such oddities as that pipe and cigar smokers should show much less lung cancer than cigarette smokers, while among the latter, the practice of inhaling is associated with less rather than with more cancer of the lung."
Sir Richard said that he did not accept that Fisher was entitled to form a view on the matters of genetics that he addressed because he had not looked through all the human evidence on lung cancer. It was just a theoretical concept in which he had not taken into account the existing evidence. He was not entitled to express the view contained in the last sentence quoted above. Counsel made it clear that his questions were directed to Fisher's views about genetics, not about inhaling. Sir Richard said that Fisher was proved to be wrong, but he suspected he was wrong in any case. It was not a view that would be at all widely accepted by geneticists at the time that genotypically different groups would be expected to differ in cancer incidence. He described Fisher as an "ignorant geneticist" in expressing this extreme view which other geneticists would not have taken. In relation to smoking, Fisher was an ignorant geneticist.
[5.222] Sir Richard was then asked about Dr Joseph Berkson. He said that Berkson was concerned that both the American Cancer Society and his and Hill's study found that there was a relationship between smoking and a whole range of diseases, not just lung cancer and heart disease, and he said that this was a very surprising thing and must make one think that there was something wrong with the study. He did not take into account the fact that the risks for the different diseases were very different, some greatly increased risks and some very small increased risks, and he did not take into account the fact that tobacco smoke contained many, many different chemicals and perfectly reasonably might affect many different diseases. Hill said that it was as if Berkson had said that milk could not spread an infectious disease when, in those days it did spread many infectious diseases, such as diphtheria, typhoid fever and scarlet fever and therefore it could not cause any of them because it caused a lot. Sir Richard said that this was a good argument of Bradford Hill's. He agreed that milk did spread a large number of infectious diseases. He described Berkson's views as "silly nonsense".
[5.223] Sir Richard agreed that Berkson worked at the Mayo Clinic, Rochester, Minnesota and was an Associate Member of the Institute of Biological Research at Johns Hopkins University, Baltimore, Maryland. He was a very distinguished statistician, eminent in his field. He was Professor of Biometry in the University of Minnesota and at the Mayo Clinic he was Head of Biometry and Statistics. Berkson 1959 was a paper by him referring to the cohort studies of the British doctors and Hammond and Horn's study of 200,000 Americans. At p.448 Berkson wrote:
"Firm opinions have been published to the effect that, on the basis of accumulated evidence, it is scientifically established beyond reasonable doubt, that smoking is an important cause of cancer of the lung. I am a member of a committee that has sponsored some of the most important of the published studies and, owing to this circumstance, have felt the responsibility to make a fairly careful study of this evidence. My own conclusions are quite different.
In the first place, virtually all the evidence is obtained from statistical studies in the ordinary connotation of the term 'statistical.' We are not dealing with the results of laboratory experiments, or even with placebo-controlled clinical trials."
Sir Richard said that he disagreed with this because by 1959 there was experimental evidence from animals and there was the detection of benzo[a]pyrene in tobacco smoke. Benzo[a]pyrene was a very strong carcinogen but, at best, was only a minor cause of lung cancer. He would not call it a weak carcinogen in any circumstances. The amount present in tobacco smoke, however, was not sufficient to cause very large risks. It was one of the strongest carcinogens known to humans, but present only in small quantities. He did not believe that it was the component of tobacco smoke which caused the great majority of cases of lung cancer.
[5.224] Berkson continued:
"Nor is the conclusion based on a synthesis, by a 'chain of reasoning,' of relevant scientific knowledge from many different sources. Such statistical evidence, for a question like the identification of a cause of a disease, at best, can be only presumptive."
Sir Richard said that Berkson was wrong there again. Berkson continued:
"But even as statistical investigations, I do not find the published studies so sound or convincing as they apparently have widely been assumed to be. In the studies that have been called 'retrospective,' as well as in those called 'prospective,' I find questionable and even paradoxic elements."
Sir Richard said that he was not entitled to that view. On the same page Berkson suggested for consideration three explanations of the observed associations. The second of these was:
"The observed associations have a constitutional basis. Persons who are non-smokers, or relatively light smokers, are the kind of people who are biologically self-protective, and biologically this is correlated with robustness in meeting mortal stress from disease generally."
Sir Richard said that Berkson was here putting forward a constitutional hypothesis.
[5.225] In Berkson 1960 at p.968 reference was made to Dr Harold Dorn. Sir Richard said that Dorn was employed by the National Cancer Institute in the US and was Secretary of the International Organization of Cancer Societies. When Sir Richard knew him he was a statistician employed by the Public Health Service. Berkson wrote:
"I spent considerable time with Dr Dorn and with some veteran experimental workers of the Public Health Service and have corresponded with others. I cannot speak didactically of their views, but I can say that they seemed in general agreement with mine, as regards the tentative character of the evidence. Surely, I had the impression that Dorn believes that much more work in pharmacology, pathology, and other associated sciences must be done before anything definite can be said with regard to the explanation of the statistical findings."
Sir Richard said that while Berkson had this impression, it was not the impression he himself had from talking to Dorn. Berkson continued:
"Certainly, the word 'cause' does not appear in any of his writings on the subject, and he has recently expressed himself elaborately on the complexities of interpreting statistical results for their etiological significance."
At p.969 Berkson wrote:
"As I have repeatedly pointed out, the prospective studies show more excess deaths among the smokers from other causes than from lung cancer, and these other causes are distributed among all classes of disease. If the statistical results are not spurious but reflect some real biological effect, this points to a generalized, not a local, effect. I have presented a plan [in Berkson 1959 and other papers] for an experimental epidemiologic study to explore this lead. Why has not the Public Health Service set up this experiment, or some modification of it?
Sir Ronald Fisher, who has been critical of the analysis of the English data on smoking and lung cancer [in Fisher 1959], early remarked, 'The question seems to be a serious one; when is serious investigation going to begin?' His question is timely even now."
Sir Richard said that he disagreed with Berkson. Further research was done, though Sir Richard thought the evidence was quite strong enough to show that it was not necessary. A lot of money was spent on it and it showed what he expected it would show.
[5.226] Reference was next made to Berkson 1964. In this paper Berkson explained that he had prepared written questions and answers as a basis for a televised interview which he had been asked to give. He subsequently rescinded his acceptance of the invitation to appear on the programme because "the broadcasting people would select the parts actually to be presented." The written questions and answers he had prepared were incorporated in the paper. At p.327 Berkson explained why he had expressed doubt that "some statistical studies have shown fairly conclusively that smoking is the chief cause of lung cancer". He wrote:
"The way you have put the question reflects a widespread misunderstanding regarding the facts. You have heard the adage that the most misleading sort of untruth is a half truth. If it is said that these studies showed smokers to have a higher death rate from lung cancer than nonsmokers, this is about a 15% truth. What these studies disclosed is that, in the populations under investigation, the smokers have a higher death rate from all causes together - lung cancer, other cancers including, for instance, cancer of the pancreas, of the bladder and of the prostate, heart disease and diseases other than any of these - in short, all classes of disease. Lung cancer constituted only about 10 to 15% of the total excess deaths. The problem presented by these findings has been put as well as need be by Cuyler Hammond, the senior author of the American Cancer Society's reports, from whom I quote:
'In my opinion, the most important finding in all these studies is that overall death rate increases with the amount of smoking. I want to emphasize that I said "overall death rate," not "lung cancer." The relationship between smoking and lung cancer should be viewed in the light of this broader picture.'"
Sir Richard agreed that at the time Berkson made these observations, he was entitled to do so on the basis of the American Cancer Society Study, and that Hammond was also entitled to say what Berkson quoted him as saying.
[5.227] On p.328, after referring to the statistics from the studies by Hammond and Horn and by Dorn, Berkson wrote:
"We are not concerned with indicting or exonerating smoking [...] but to try to find the scientific explanation of these statistical results. The idea that cigarette smoking causes all these many deaths from all these many causes does indeed seem seriously questionable. There is not any scientifically known pharmacologic or physical explanation for so widespread and multifarious an effect. If we extrapolate the results to the general population, we must believe that there are some 250,000 deaths annually from smoking-induced diseases, without any of them having been individually noted as such from independent clinical or pathologic evidence. Only by their numbers are they known."
Sir Richard said that this had been done by other methods since. This was one of the first observations that indicated that smoking was responsible for so many more diseases, and it took quite a long time to establish which ones smoking actually caused and which were statistical flukes. The American Cancer Society finding about an increase in the death rate from cancer of the prostate turned out to be just an odd finding which was not repeated in other studies. But of course, he said, in subsequent years the explanation for these findings all became clear. Counsel pointed out that he was asking Sir Richard to look at the texts in the context of the time in which they were written, and he was not asking whether any of the views held good today in the light of subsequent knowledge.
[5.228] At p.329 Berkson wrote that it was the distinct and real possibility that the results stemmed from a statistical fallacy in the data. He had put forward several other explanations:
"Another explanation, which has been advanced also by other scientists, is that the difference in death rate from all causes reflects a constitutional difference between smokers and nonsmokers. The ideas is that nonsmokers or light smokers are of a constitutional type marked by self-protective habits, and one aspect of the constitutional makeup is that they have generally lower death rates. [...]
The findings of Seltzer seem especially pointed. An anthropologist, he investigated the morphologic characteristics of 922 college students as reflected in a number of anthropometric measurements, in relation to their smoking habits as determined 13 years after graduation. He found, not only that the indices of body build clearly differentiated the smokers and nonsmokers, but that smokers were clearly differentiated according to type of smoking. He concluded:
'Smoking behavior appears to be in part a reflection of the biological or genetic makeup of the individual.'"
Sir Richard agreed that Berkson was here suggesting a possibility, but did not feel able to say whether the suggestion was sensible without reading the whole paper. These statements were very complex. Seltzer was the American he had referred to earlier as being one of the "eccentrics".
[5.229] Sir Richard was next asked about Yerushalmy 1962. He agreed that Yerushalmy had been a statistician at the National Institute of Health in the United States, that in the early 1940s he assumed the Directorship of the Division of Statistical Research of the Children's Bureau in the US Department of Labor, and that, after working as a statistician in the Public Health Service, in the late 1940s he joined the University of California at Berkeley and was Professor of Biostatistics at that university. At p.212, under the heading "Specificity of association", Yerushalmy wrote:
"The main doubt about the validity of the epidemiologic evidence stems [...] from the fact that the groups possessing or not possessing the characteristic under suspicion have been self-selected. Consequently, the association which has been established epidemiologically between the factor F and the disease D-1 may not reflect a causal relationship between F and D-1, but be due to other factors and characteristics by which the two groups differ. To fix the ideas - the association observed between cigarette smoking and cancer of the lung may indicate a cause-effect relationship, but it may also be due to differences between smokers and non-smokers in characteristics other than smoking. It is at least possible that these other factors and characteristics, rather than smoking per se, are behind the observed association. It is therefore necessary to search for methods of investigation which would help elucidate, if not eliminate this dilemma."
Sir Richard said that he knew Yerushalmy's views, but he had not read the article and would rather not comment on it until he had had an opportunity of reading it. Counsel pointed out that he was asking Sir Richard about his evidence that nobody had seriously challenged the causal relationship between cigarette smoking and lung cancer. Sir Richard said that counsel was picking out the two or three who did, and Yerushalmy certainly would be one he would expect to have done so; he was a statistician who did not know a great deal about epidemiology.
[5.230] At p.214, under the heading "Cigarette smoking and health", Yerushalmy wrote:
"A strong association has been demonstrated between cigarette smoking and cancer of the lung. Many studies, both retrospective and prospective, have confirmed the fact that a larger proportion of persons who suffer from cancer of the lung are cigarette smokers than is true of a comparable group not suffering from this disease. Similarly, cigarette smokers have a higher death rate from cancer of the lung than non-smokers. Moreover, heavy smokers suffer higher mortality from cancer of the lung than light smokers and past smokers who stopped the habit have lower rates than those who continue to smoke [...]."
Sir Richard said that he agreed with these statements.
[5.231] Yerushalmy continued:
"The investigations by which these associations were established presented at first some interesting and subtle sampling problems which could have biased the results. These were discussed by a number of investigators, especially Berkson [...]. However, if the association were due entirely to these sampling biases they would not have persisted for many years after the initiation of the investigation. The fact that cigarette smokers continued to have higher rates in the second and third year of observation would indicate that the association between cigarette smoking and cancer of the lung is not a resultant of these sampling peculiarities alone. Consequently, the association itself is accepted as definitely established. The question turns to the interpretation of this association in terms of causation."
In the next paragraph, Yerushalmy wrote:
"The main difficulty in evaluating such association stems from the fact that the individuals observed have made for themselves the crucial decision whether they are smokers, non-smokers, or past smokers."
Sir Richard said that he did take objection to this, but really could not comment further without reading the whole article. At pp.220-221 Yerushalmy wrote:
"These findings can perhaps more easily be explained on the basis that smoking acts as an index to differentiate smokers from non-smokers on a number of different characteristics rather than as indicating a causal relationship."
Sir Richard said that he could not comment on this: he could not understand what Yerushalmy was saying until he had read the paper.
[5.232] Sir Richard was also asked about Yerushalmy 1972. At p.279 it was stated:
"A comparison of smokers and nonsmokers showed that the two differed markedly along several environmental, behavioral, and biologic variables. [...] These findings raise doubt that cigarette smoking acts as an exogenous factor which interferes with the intrauterine development of the fetus. The findings give equal support to the hypothesis that smokers represent a group of people whose reproductive experience would have duplicated the observed patterns whether or not they smoked. In other words, the observed differences in incidence of low-birth-weight infants may be due to the smoker, not the smoking."
Sir Richard said that although he was not familiar with the paper, he was familiar with Yerushalmy's general view on the effect of smoking on birth weight and he generally agreed with him. He did not feel able to make more detailed comments without an opportunity of reading the paper. I gave Sir Richard an opportunity to read the two Yerushalmy papers, and having read them he subsequently agreed that Yerushalmy also was a worker who questioned the causal hypothesis.
[5.233] Counsel next asked Sir Richard about HJ Eysenck. He said that he was familiar with the name. He was the psychiatrist Sir Richard had referred to earlier. He died in 1977. He did a lot of research for the tobacco industry. He was on Sir Richard's list of eccentrics; indeed he would go further and come straight out, he was a liar. They had direct experience of that with him saying one thing on one day and then denying and saying the opposite the next day. He was a psychologist and a Professor at the London Institute of Psychiatry, he worked at the Maudsley Hospital and at the Institute of Psychiatry for most of his professional life. He was editor-in-chief of a journal, the name of which Sir Richard could not remember. He was one of the leaders of clinical psychology as a profession in Britain. He had a particular interest in researching the role of personality types, explored concepts of introversion and extraversion and was particularly interested in the relationship between personality types and the development of lung cancer. He challenged the causal hypothesis (that cigarette smoking caused lung cancer), and then he withdrew, and then he challenged it again and then he withdrew. "If you got him in front of the public and cross-examined him, he would withdraw and then he would go back again and say the same thing a week later." There was a deathbed recognition of where he received his funding. Sir Richard agreed that he was suggesting that Eysenck's integrity was compromised by the source of his funding. He knew that Eysenck had written a lot of papers.
[5.234] Eysenck 1965 contained an introductory passage which concluded with this sentence:
"Above all, Dr Eysenck's book is an urgent plea for more fundamental research and for recognition of the fact that the issue is far more complicated than most official and journalistic statements would have us believe."
Sir Richard said that he would not pay any attention to what Eysenck said in relation to that. At p.15 Eysenck wrote:
"This book is concerned then with the relationships between three variables: smoking, disease, particularly lung cancer and coronary thrombosis, and personality. All the evidence to be considered is circumstantial; there is no direct evidence which does not rely on statistical investigation and logical interpretations. These are difficult matters and it is very easy to go wrong. As Dr R Doll, himself one of the scientists most prominently associated with the promulgation of the theory that smoking causes cancer, has pointed out:
'When the nature of the disease makes it impossible to carry out logically conclusive experiments there is always room for honest difference of opinion. In the case of smoking it is particularly hard to envisage how a conclusive experiment could be carried out and no such experiments have been made.'
Doll goes on to quote a famous saying of Claude Bernard, to the effect that 'There are no false theories and true theories, but only fertile theories and sterile theories.'"
Sir Richard did not recall having said this, but agreed that he might well have done. He did not take exception to it.
[5.235] At pp.50-51, in a chapter entitled "The critics hit back", Eysenck wrote:
"The work of Doll and Hill in England, of Horn and Hammond in America, and of all the other investigators who have taken up the trail of the disease-producing effects of cigarette smoking, has, of course, not gone unchallenged. I shall not in this chapter deal with all the arguments that have been advanced in an attempt to rebut their conclusions. Many of the criticisms are themselves unsound, and it would be a waste of time to discuss them in detail. Others, while reasonable at the time when they were made, have since been answered by the original authors, either in further analyses of their data, or by new studies. In going through the writings of Doll and Hill, and of Horn and Hammond again for the purpose of this book I was struck, as I had been when I had read them originally, by the great care which had been taken in the carrying out of these investigations, by the way in which the authors took note of published criticism and tried to answer it by adducing further factual information, and by their wholly admirable refusal to be side-tracked from strictly scientific argument. I believe that the case which they make out can be criticized but that is merely to say that scientific investigators, even the most eminent, are only human; whatever the truth of the criticisms here presented the work of these investigators will always remain as a fine example of scientific detective work."
Sir Richard said that he did not disagree with any of this, though he described the concluding part of the passage as "a bit smarmy".
[5.236] At p.75, in a chapter entitled "Personality and constitution" Eysenck wrote:
"We have already referred several times to the so-called constitutional theory of lung cancer and smoking, i.e. the hypothesis first put forward by Sir Ronald Fisher that people of a certain constitutional type are particularly prone to lung cancer and are also particularly predisposed to take up cigarette smoking. In this form the theory is too weak to be very useful because it is so indefinite that no specific test can be conducted to support or disprove it."
Sir Richard said that this was not actually true: studies were done to test it and in fact disproved it. The studies of monozygous twins that were carried out in Sweden, the United States and Finland all showed that it just was not true. Few people took Fisher's suggested hypothesis seriously. Eysenck was one of them. Sir Richard challenged his good faith and also pointed out that what Eysenck suggested had been demonstrated to be untrue. He could not be sure whether this was so when the book was published in 1965.
[5.237] Eysenck went on:
"What is maintained in effect is that there are certain types of people who smoke; that this type of person has acquired his particular personality through hereditary causes, and that this particular type of person is also more likely to develop cancer."
Sir Richard said that this was shown to be untrue and Eysenck knew it perfectly well, though perhaps not in 1965. Pressed on this, Sir Richard said that he did not say that in 1965 Eysenck knew it was untrue and it was misleading if he did say that. He repeated it later at a time when it was known to be true, at about the period when there were proceedings at the Royal Statistical Society in 1978, when there was more evidence. Eysenck continued: "Clearly, therefore, psychological studies are needed to link up both smoking and cancer proneness with specific personality types." Sir Richard said that it was wrong of Eysenck to make that suggestion. He agreed that he himself had no expertise in the concept of personality in psychology.
[5.238] Asked about Eysenck et al. 1960, a report of a study commissioned by the Tobacco Manufacturers' Standing Committee and published in the British Medical Journal, Sir Richard said that he must make it clear that he did not accept any of Eysenck's proposed findings "in the light of knowledge we gained later about his behaviour". He was corrupted, though Sir Richard could not say whether this was so in 1960. Sir Richard had not read this paper. He was not prepared to look to see what Eysenck and his colleagues had said so that he might be asked for his view on it, because he would not believe it.
[5.239] Sir Richard agreed that Professor Philip Burch was one of the people he described as "eccentrics". He was Professor of Medical Physics at the University of Leeds and was Director of the Medical Research Council Environmental Research Unit, but "not for long, because I attended a meeting at the Medical Research Council in which he sought funds for such a unit and he was flatly turned down". He was not aware of Burch having had a unit, but was aware of having "described himself as some such title". So far as he knew, Burch did not get support for such a unit. What Burch wrote should be regarded with some care. He wrote "a terrible lot", and some of it was "nonsense".
[5.240] Counsel pointed out to Sir Richard, under reference to Burch 1964, that the author was writing at that time from the Medical Research Council Environmental Radiation Research Unit at the Department of Medical Physics, University of Leeds. Sir Richard agreed that Burch was a member of that unit at the time, under a professor at Leeds, but said that he was talking about a time when Burch sought to be the Director of the unit and that was what the Medical Research Council turned down.
[5.241] Sir Richard's attention was next directed to a debate which was conducted on the pages of New Scientist in 1974. It started with Burch 1974a, which began:
"In their 1971 report, the Royal College of Physicians stressed:
'Many countries have set up authoritative committees and commissions to study the cause of this modern scourge [lung cancer]. All have concluded that it is almost entirely due to cigarette smoking.'
On the other hand, the late Sir Ronald Fisher, who has been described as the greatest statistician who ever lived, feared that this conclusion would be seen in retrospect to be a 'catastrophic and conspicuous howler'. Where then does the truth lie? With a statistician-logician-geneticist of genius, or with 'authoritative committees and commissions'?
During the past year or so I have reiterated Fisher's views in The Lancet and supplemented them with new analyses. Surprisingly, these arguments and demonstrations have provoked few responses and nothing to challenge Fisher's position. Having at one time accepted that most cases of lung cancer are caused by smoking, I shall outline in this article some of the evidence that forced me to change my mind."
Sir Richard said that by this time he had not responded to Burch because there was no controversy with him, he just made wild statements which people ignored. Of course it was quite untrue that Fisher's position had not been disproved, it had been by 1974. Sir Richard did not enter into debates with Burch. He had ignored him intentionally.
[5.242] At p.463 Burch wrote:
"For the reasons outlined here, and for many others that limitations of space do not allow me to describe, I am unable to sustain the hypothesis I once held: that lung cancer 'is almost entirely due to cigarette smoking'. At the same time, I am unable to refute Fisher's constitutional hypothesis which offers a plausible and well supported interpretation of numerous otherwise paradoxical findings."
Sir Richard agreed that the scientific method required that a hypothesis may be enunciated and then the scientist worked to try to disprove the hypothesis, and that this method was being invoked by Professor Burch.
[5.243] Burch's article was followed, on p.463, by a reply by Sir Richard, Doll 1974, which he said that he had been advised to write. The response to it was Burch 1974b, which generated correspondence in the letters pages, where among other letters were two by Burch, Burch 1974c and Burch 1974d. In the last of these, Burch wrote:
"[...] Fisher's constitutional hypothesis is, in principle, eminently falsifiable: many testable consequences follow from it. If 'smoking genotypes' generally associate positively with 'lung cancer genotypes', then appropriate studies of the first-degree relatives of (a) lung cancer probands and (b) suitably matched controls should reveal such associations. For example, the frequency of smokers among the first-degree relatives of non-smoking lung cancer probands should be higher than among the corresponding relatives of non-smoking matched controls. This and other predictions of Fisher's hypothesis have all been verified by Dr George K Tokuhata [...]."
Sir Richard agreed that Burch was here suggesting a methodology to test the hypothesis, but he added that there were "heaps of other methodologies" and that actually the hypothesis was untestable because it did not account for an increase in the disease; genotypes do not change, so it was never a starter.
[5.244] Sir Richard was next asked about Carl C Seltzer. He said that Seltzer was one of "a rather peculiar group of people who took some odd views". He did not accept the causal hypothesis, and a lot of other things as well. Sir Richard was unable to recall what had led him to describe Seltzer as one of a rather peculiar group of people, but he confirmed that he was one of the people he had previously described as "eccentrics". In Seltzer 1963a the author reported on a study of 922 college men relating their morphological characteristics as students to their subsequent histories of smoking thirteen years after graduation, in order to ascertain the extent to which the different classes of nonsmokers, cigarette smokers, pipe smokers and cigar smokers were phenotypically and genotypically conditioned. Significant differences in physique were found between smokers and nonsmokers and in accordance with the form of smoking adopted. Smokers were consistently greater than nonsmokers in height and weight and in the dimensions of the head, face, shoulders, chest, hip, leg and hand. He claimed that the findings delineated constitutional differences between smokers and nonsmokers and among the several varieties of smokers. Smoking behaviour thus appeared to be in part a reflection of the biological or genetic make-up of the individual. At p.644 Seltzer wrote:
"Given the strong likelihood of a constitutional factor in smoking behavior, consideration should also be given to the possibility of a constitutional factor in patients with lung cancer. (This would bear upon the subject of the association of smoking and cancer of the lung.)"
[5.245] Sir Richard said that while it was a view that he always held that consideration should be given to the possibility of a constitutional factor in patients with lung cancer, he accepted only to a trivial extent that this would bear upon the subject of the association of smoking and lung cancer. This was because of the increase in the incidence of the disease, so even if there was some constitutional factor it was relatively unimportant. Asked whether there were some who still did take issue with the proposition that there had been an increase in the incidence of the disease, Sir Richard said: "There was still a Flat Earthers Society at that time." Those who suggested that the reported increase in the incidence of the disease might be artefactual were Flat Earthers. There was no question that part of it was certainly artefactual, but by that time the suggestion that it was wholly artefactual was quite unacceptable. He had gone on record as saying that he accepted that part of the increase at least was artefactual, without question, and that one simply had to exercise one's judgment about how much. Here there was a thirty- or forty-fold increase in incidence that was not artefactual, and the incidence increased differently in the two sexes and in different countries. In the early years there were some who thought differently from him, and Seltzer must have, but it was a pretty extreme position for him to take in 1963.
[5.246] Seltzer continued:
"That there is a relationship between constitution and disease is generally accepted in medical circles. [...] Accordingly, a logical line of future research is indicated. Attention should be focused on an investigation of constitutional characteristics in patients with lung cancer."
Sir Richard said that this was unreasonable; the hypothesis was a non-starter because of the increased incidence, but nevertheless it was done, with tobacco industry support, quite properly.
[5.247] In Seltzer 1963b, in reply to a correspondent who had written that his speculations would be used by promoters of cigarette smoking as ammunition in their campaign against efforts to control "this addiction", Seltzer wrote:
"[D]oes Dr Flick wish to hold researchers personally responsible for what promoters, publicists, and others might do with their published works?"
Sir Richard said that this was a fair comment.
[5.248] The last paper by Seltzer which was referred to by counsel was Seltzer 1967. Seltzer had been asked by the committee responsible for a conference to review the status of knowledge of the subject of human constitution and genetics as related to smoking. He found that very little new information had been added to this area since the appearance of USSG 1964. At p.324 he noted "that a recent reevaluation of the Harvard Study [Seltzer 1963a] by Livson and Stewart has failed to undermine its basic results." Sir Richard said he had not been at the conference and was not clear what was being referred to.
[5.249] Reference was next made to KA Brownlee of the University of Chicago. Sir Richard said that he had heard of Brownlee, who was a statistician of British origin, and a fellow of the Royal Statistical Society of London, who had gone to work in the United States. In Brownlee 1965 he contributed an invited review article on USSG 1964 to the Journal of the American Statistical Association. At p.733 the author wrote:
"The fact that the hypothesis that cigarette smoking is a cause of cancer appears by and large to be in conformity with the data, and hence that this hypothesis is acceptable, does not rule out the possibility that there are other hypotheses also in conformity with the data. [...] The main alternative to the smoking-causes-cancer hypothesis is the genetic hypothesis, and there are several odd pieces of information that give plausibility to it."
On the same page there was quotation from a paper by Sir Ronald Fisher. Sir Richard accepted that Brownlee was someone who, in 1965, was not prepared just to accept the causal hypothesis, and who turned to Fisher and apparently took his views seriously.
[5.250] Burch 1978 was a paper read before the Royal Statistical Society in that year, followed by a discussion of the paper by some of those present and contributions subsequently received in writing. According to the summary:
"Methodological problems that arise in epidemiology are briefly reviewed. Associations between smoking and lung cancer, and their connexion with sex and country, are described. Secular trends in recorded mortality from lung cancer in England and Wales, 1901 to 1970, are derived from the Registrar General's statistics. Estimates are made of the expected trends assuming that the associations found between the various types of smoking and lung cancer reflect causation. Observed and expected trends conflict. The secular trends in mortality from lung cancer in British male doctors are also examined. From the foregoing and other types of evidence it is argued that no definitive conclusions can be reached, as yet, about the extent of any causal link between smoking and lung cancer. It is very doubtful, however, whether the entire association observed between smoking and lung cancer in occidental males should be interpreted in causal terms."
Sir Richard said that this was consistently Burch's view.
[5.251] At p.438, in para.2.1, Burch said:
"Epidemiology, like astronomy, is largely an observational rather than an experimental science and hence the scope for planned and controlled intervention by the investigator is usually either limited or absent."
Sir Richard said he did not disagree with this. In the same paragraph Burch said: "We cannot randomize for smoking." Sir Richard said that it was correct to say that as a general approach one could not randomise for smoking; attempts had been made to do so, not very successfully. At para.2.2 Burch said that many studies had shown that cigarette smokers differed, on the average, from non-smokers with respect to morphology (reference to three papers by Seltzer), personality (reference to papers by Heath, Lilienfeld, Eysenck and Thomas et al.) and genetic markers. Sir Richard said that he did not know the papers to which reference was made in respect of genetic markers, but he did know that no genetic markers had been recognised for lung cancer. At para.8.4, on p.456, Burch said that, having once shared the view that lung cancer was almost entirely due to cigarette smoking, he found himself forced back to the verdict in Fisher 1959 that "the data so far do not warrant the conclusions based upon them". Sir Richard agreed that this was a familiar theme of Burch.
[5.252] The first contributor to the discussion of Professor Burch's paper was Professor P Armitage of Oxford University, whom Sir Richard described as a close colleague of his with whom he had collaborated. At p.458 Armitage said:
"Professor Burch's own views have been widely publicized in the past, although this is the first time they have been presented in a paper to the Society. He expresses a hope [...] that rigorous methods of statistical inference will help to resolve the issues that concern him. In this I think he is mistaken, unless one takes a much wider view of the principles of statistical inference than is customary. The issues seem to me to be concerned with the reliability of sources of data, the possible effects of confounding variables and the plausibility of alternative biological hypotheses. These problems are basically for the epidemiologist, rather than the statistician, to grapple with [...]."
Sir Richard agreed that what Armitage was saying here was not that what Burch proposed was nonsense, but that it might be misguided. At p.460 Armitage concluded:
"It will be clear that I am not wholly in sympathy with Professor Burch's point of view, but he has provided a clear statement of his case and given us an opportunity to review it in detail."
[5.253] The next speaker was Dr PD Oldham of the MRC Pneumoconiosis Unit in Glamorgan; Sir Richard said he knew of him, describing him as a very solid, middle-of-the-road statistician with a lot of epidemiological experience. After referring to Doll and Hill 1950, Oldham said, at p.460:
"The consequence is that, 28 years later, we still do not know how cigarettes cause lung cancer, nor even, if we are particularly rigorous in our use of scientific logic, whether they do."
Sir Richard said that this was not a statement with which he would have disagreed. At p.462 Oldham said:
"In one respect at least I find myself in wholehearted agreement with the author [Burch], and believe that this would have been shared by Fisher. This is in the ridicule with which he describes the steady flow of theories of the causation of diseases based on epidemiological associations. There is nothing wrong with this method of formulating hypotheses, but too often the authors of such studies succeed in implying that a tenable hypothesis, developed from a careful epidemiological survey, is a proven one. Attribution of this step, by the public, to a statistical argument is a certain means of discrediting our science."
Sir Richard said that he thought that this was true.
[5.254] Professor HJ Eysenck spoke next. Sir Richard said that he would not be interested in his contribution. Asked to give his reasons for this attitude, Sir Richard said that, leaving aside the fact that Eysenck did not admit that his research was funded by the tobacco industry, which only came out after his death, it was the way in which he would change his opinions in a debate. When you were debating with him, and if you gave him good clear evidence of something, he would agree with you and then he would go back in the next meeting or in a television broadcast and say the opposite. This was why he could not take him seriously. So far as Sir Richard was concerned Eysenck concealed the fact that he was funded by the tobacco industry until after his death and this was dishonourable and quite probably tainted his views. If he had acknowledged that he had been funded by the tobacco industry, it would have made a difference; one would have known where one was then. In major research, the source for the funding should always be disclosed. Sir Richard did not know that the fact that Eysenck had been funded by the tobacco industry in the years between 1963 and 1967 was a matter of public record; all he could say was that he did not know it, nor did any of his colleagues.
[5.255] At p.462, GJA Stern of Imperial College, London, whom Sir Richard did not know, said:
"The author seems to confirm Fisher's objections to the dogma that smoking causes lung cancer. Fisher's point that inhaling correlates negatively with lung cancer is important, as is the rather suspicious silence and apparent reluctance to investigate this matter adequately on the part of medical statisticians."
Sir Richard said that this was quite untrue, it was investigated in detail, he himself had advanced a suggestion to explain the inhalation anomaly, and did some experiments with the use of radioactive isotopes to see where tobacco smoke went, but the experiments failed. At p.463 Stern said:
"Fisher observes that lung cancer has increased far more in women. This, and Professor Burch's analysis, makes the Royal College of Physicians' claim that the secular trend argument is the chief reason for rejecting a genetic hypothesis hardly believable. The identical twin analysis seems rather to demonstrate an important genetic component."
Sir Richard said that he did not agree with any of this; Stern was just incorrect about the increase in women.
[5.256] Professor H Gwynne Jones of the University of Leeds, whom Sir Richard knew, said at p.463:
"This paper represents a very valuable bringing together of the strands of a complex but empirically based argument by which Professor Burch has consistently failed to modify medical opinion concerning the relationship between smoking and lung cancer. Nevertheless, in my view, the popular causal theory is clearly falsified in true Popperian fashion by his careful analysis of secular trends, sex differences and smoking habits. Equally clearly, any direct influence of smoking on the aetiology of lung cancer is shown to be minimal."
Sir Richard said that he rejected this, it was just nonsense. Gwynne Jones continued:
"On the other hand, Professor Burch does present an admittedly complex biological model of diseases of this type which derives from quite different sources but is consistent with all the data reviewed. This model certainly merits more attention than he has so far received."
Sir Richard said that he would disagree with this. This was the model he was referring to earlier, considered by a special committee of the Medical Research Council and rejected. So his own disagreement was backed up by the committee.
[5.257] Sir Richard was asked to note that Dr CC Seltzer of Harvard University attended the meeting and made a contribution to the discussion which was supportive of Burch's approach. Dr Frank Hansford-Miller of the Inner London Education Authority expressed gratitude to Burch for opening up the question of the causative factors of lung cancer and suggested that the search should be started for causes other than smoking which Burch had shown must make a far bigger contribution than smoking. Sir Richard agreed that this speaker supported Burch. The last substantive contribution to the discussion was made by Joy L Townsend of the University of Essex, who reached a different conclusion from that of Burch.
[5.258] Among the written contributions, at p.468 was one from Professor AR Feinstein of Yale University, whom Sir Richard described as an eminent but peculiar scientist who tended to have a different view to anybody else on any subject. He was well known for this. He was a Professor of Medicine and Epidemiology at the Yale School of Medicine and was Director Emeritus of the Robert Wood Johnson Clinical Scholars Programme at Yale. He had "a clutch" of honours and awards. He was a professional critic and a professional author. Sir Richard was honestly not sure whether Feinstein was a sincere critic. He knew him quite well and had spoken in his department, but had difficulty making up his mind about him. "He liked to do things because it was naughty, I think."
[5.259] Feinstein wrote:
"The history of medical science contains many instances of thoroughly accepted, paradigmatic 'gospel' that was later rejected as fallacious. During the past century, statistical evidence was used to support many ideas about causes of disease that were eventually discarded as erroneous. Among such ideas were the beliefs that cholera was due to high atmospheric pressure, that pellagra was infectious, and that retrolental fibroplasia was produced by a vitamin deficiency. Recalling the long history of pitfalls in medical aetiological reasoning, cautious scientists may wish to keep at least a slightly open mind about the currently well-accepted hypothesis that cigarette smoking causes lung cancer. Although supported by a large collection of positive evidence, the hypothesis is not as securely established as the vigour with which it is argued by epidemiological authorities."
Sir Richard said that he did not accept this as a legitimate view. If he took the view, which he had expressed, that the evidence was proved beyond reasonable doubt, he did not have any doubt about that. Feinstein went on to write about "loose strands in the fabric of the argument", which he enumerated; Sir Richard said that they were not loose now. It was incorrect of Feinstein to state that "about 8 - 10% of patients with lung cancer have never smoked cigarettes"; this was not true of men, it was true of women at that time.
[5.260] Counsel next asked Sir Richard about Passey 1962. He said that he knew this paper well. Malcolm Pike and he wrote an article following it, pointing out the fallacies in Passey's argument. Passey challenged the causal hypothesis, but Sir Richard regarded the challenge as extremely fallacious. He added that Passey "unfortunately died of lung cancer". Passey was a pathologist by the time he wrote the article. Like Sir Richard, he was ageing. He had been interested in the cause of lung cancer for many years. He had done some experiments in animals in the 1920s with tobacco tar and had failed to produce cancer of the skin in mice, and Sir Richard thought that had influenced him. He did produce skin cancer in mice with lots of things, but he did not succeed in doing it with tobacco tar. At the time of writing he was Emeritus Professor of Experimental Pathology in the University of Leeds. In the article, Passey suggested that lung cancer might well be a sequel to tissue damage of a non-specific nature rather than to the specific action of carcinogens. He continued, at p.107:
"The smoker (Doll and Hill 1950) and the townsman [...] are more prone to lung cancer than the non-smoker and the countryman. We know that tobacco smoke and the air of our industrial towns contain small quantities of the polycyclic aromatic hydrocarbons [...], some of which are carcinogenic. Hence it is natural to suppose that cancer of the lung is the result of these carcinogens. But is this necessarily so? May not lung cancer be the result of altered conditions arising out of disease or damage of the respiratory system in which carcinogens need have played no part?"
Sir Richard said that this was rather a bizarre question.
[5.261] Passey continued:
"This damage may have been induced by (a) the irritating properties of tobacco or other smoke, (b) chronic respiratory disease, or (c) other injuries. Perhaps that is all that one can or should say; but if pressed I could, with some reluctance, add a suggestion that the altered conditions which appear to be important, are those associated with the presence of excess of mucus, and its accumulation and stagnation."
Sir Richard agreed that this was the view expressed by Passey in 1962. At p.111 Passey stated, as part of his conclusions:
"Lung cancer is not the result of carcinogens. It is a 'natural' form of cancer - the result of changed conditions in a damaged respiratory system."
Sir Richard agreed that Passey was prepared to express this view publicly in The Lancet in 1962.
[5.262] Asked to look at Hinshaw and Garland 1963, Sir Richard said that this was an American textbook which he had not seen before. He was asked to note that in the preface it was stated that in the six years since the first edition of the book was completed there had been developments, including the accumulation of much experience with bronchogenic carcinoma. These developments had necessitated extensive revision as well as the inclusion of entirely new material in the second edition. Chapter 20 was entitled "bronchogenic carcinoma". It started, at p.340:
"Bronchogenic carcinoma is of paramount interest to every physician and to every student of the cancer problem. While the causes are unknown, there are more stimulating theories and fragments of etiologic evidence than is true of many other forms of cancer. An increasing incidence of the disease challenges each physician to maximal efforts of prevention and control. Sex and age incidences suggest where we might find the condition with greatest frequency."
At pp.343-344 the authors stated, under the heading "Etiology" and the sub-heading "Cigarette smoking and lung cancer":
"There is a widespread belief that excessive smoking is the major cause of human lung cancer. If and when this opinion is verified, it will be the duty of physicians to warn persons against heavy cigarette smoking. The authors are keenly conscious of their responsibility in attempting to summarize the evidence on both sides of this important and controversial issue."
At p.346, under the heading "Evidence that lung cancer is not due to excessive cigarette smoking" and the sub-heading "Experimental evidence" the authors wrote:
"Despite literally hundreds of careful and prolonged experiments designed to produce lung cancer by exposure of mice or rats to prolonged breathing of cigarette smoke, no bronchogenic carcinomas have been produced."
Sir Richard, having noted the earlier passages, said that this passage was correct.
[5.263] The authors continued:
"The epidermis of the mouse is not the same as the epithelium of the human bronchus; the cigarette smoke condensate applied to the mouse was in very high concentrations, quite dissimilar to the concentration of polycyclic hydrocarbons in inhaled cigarette smoke. Incidentally, the condensate contains no substance demonstrated as being carcinogenic to man."
Sir Richard said that this last statement was untrue in 1963, they got it grossly wrong.
[5.264] Sir Richard said that he had heard of Dr R H Rigdon. In Rigdon 1965 he said that he was a physician who taught and practised pathology and was an experimental pathologist. For the previous fifteen years he had been interested in lung cancer. At p.600 he stated that he was unable to transfer data about cancer, as obtained from experimental animals, to man. Sir Richard said that this was a view he himself held, and he thought most people did the opposite, they transferred observations made in animals to the human situation. He had always personally been sceptical of the justification for doing so. In the next paragraph Rigdon said that genetic factors must be considered in cancer. Sir Richard said that this was not a controversial view, but it was a very minor factor. Rigdon continued:
"The cause of cancer is not known. By this we mean that we do not know why a cell in the body of one person will become a cancer and the corresponding cell in another person will not become a cancer."
Sir Richard said that he thought that at the time this was true.
[5.265] At p.601 Rigdon said:
"I hope that it is obvious from my remarks why I was classified in the Surgeon General's report [USSG 1964] on cigarette smoking on p.179 as 'though accepting the existence of an association, have questioned its significance in terms of a causal hypothesis.' In regard to the Surgeon General's report on smoking and health, I think it is an excellent summary of one side of this controversy. If equal time and effort had been given to the opposite view by this committee, I think the report would have been twice the size and would have been a report that scientists in the future could review to see really what the state of knowledge was in 1963."
Sir Richard said that of course he disagreed with Rigdon about this.
[5.266] In the next paragraph Rigdon said that he would consider a statement on cigarette labels that "Cigarette smoking may be dangerous to health" to be consistent with the facts as they were known, but also a number of other products such as peanuts and charcoal broiled steaks should be similarly labelled because they might contain carcinogenic substances. Sir Richard said that he did not think that this was a justifiable remark because there was a different quantitative effect.
[5.267] Rigdon was referring to different things, some very tiny and theoretical and some very large and evident. Rigdon continued:
"To summarize my views, may I say that, first, I believe that there is no satisfactory experimental evidence to establish a connection between cigarette smoking and lung cancer. Cancer of the lung has not been produced experimentally in any animal by the inhalation of cigarette smoke. The fact that tobacco tars will produce a tumor on the skin of a specific strain of mouse does not, in my opinion, lead to the conclusion that cigarette smoke will produce cancer in the lungs of man. As I continue my work in the field of experimental carcinogenesis, I find puzzling variations in the responses observed from one species of animal to another with respect to the development of cancer. This is why I say that data concerning cancer in animals should not be considered proof of a similar effect in man."
Sir Richard said that he agreed with Rigdon in this passage.
[5.268] In Rosenblatt 1969 the abstract stated:
"An eminent clinician takes a stand that indicts the concept of an absolute increase in bronchogenic carcinoma. He presents compelling arguments, pointing to the progressive decline in rate of increase concurrent with the growth of more accurate diagnosis. The author examines critically the paradoxical trends, inaccuracy of mortality statistics, faulty certifications of death and unrealistic classifications of malignancy that have combined to lead, more hysterically than historically, to the 'illusion'."
Sir Richard said that there were those who would have shared Rosenblatt's opinion in the 1950s, but he found it surprising that as late as 1969 there was anyone who took that view.
[5.269] The article concluded, at p.38:
"The prodigious increase in lung cancer during the past three decades is not due to the exposure of the population to an alleged carcinogen but is the natural consequence of the widespread use of diagnostic techniques not previously available. The intense interest in lung cancer has also produced a tendency toward over-diagnosis of the disease on the basis of radiologic, biopsy, and cytologic findings which are often not substantiated by autopsy."
Sir Richard said that he took issue with the proposition contained in the first sentence in this passage; practically everybody did, this was a "very far-out viewpoint" to be expressed as late as 1969. Asked about the tendency towards over-diagnosis on the basis of these findings, Sir Richard said that they did studies in the UK and there was some over-diagnosis, but there was also a compensatory under-diagnosis, and they just about cancelled each other out in the studies they did on autopsies and clinical diagnoses. So while it was true there was an over-diagnosis, it was also true there was under-diagnosis. With notice he could point to the studies in question.
[5.270] Referred to Wells and Feinstein 1988, Sir Richard said that he was also interested in detection bias in the diagnostic pursuit of lung cancer. At p.1025 the authors wrote:
"These results, which clearly demonstrated the existence of detection bias in the diagnostic pursuit of lung cancer in living patients, are entirely consistent with what has previously been noted at necropsy. The necropsy results showed that lung cancers are particularly likely to be undetected during life in patients who do not cough, are women, and do not smoke."
Sir Richard said that this was not consistent with what he had found.
[5.271] The authors continued:
"The existence of the pattern is not surprising. Physicians are particularly likely to suspect lung cancer and to plan a diagnostic workup if the disease has produced appropriate pulmonary manifestations (such as cough) and if the patient is in a demographic category (such as male sex or cigarette smoker) in which the disease is thought to have a relatively high incidence."
Sir Richard said that this was contrary to his experience, but it seemed to be the result of the investigation of these authors.
[5.272] Wells and Feinstein continued:
"On the other hand, despite an apparent economy in the ordering of tests, detection bias can have three important adverse effects. [...] A third consequence of detection bias is the effect on statistics for the cigarette smoking-lung cancer association. The magnitude of this association will be falsely elevated if the lung cancers are often not suspected and not diagnosed when present in nonsmokers. Although our results show a striking bias in the ordering of sputum Pap [Papanicolaou] smears for the antemortem diagnosis of lung cancer, the total effect of detection bias in the cigarette smoking-lung cancer relation cannot be quantified from our study."
Sir Richard said that he tried to quantify it in his. They continued:
"Separate research is needed to determine the effect of this bias when physicians order the many additional technologic tests used in diagnosing lung cancer today."
Asked whether he had reason to doubt the results of Wells and Feinstein's work, as reported in this article, Sir Richard said that all he could say was that it did not apply in the UK.
[5.273] In USSG 1964 at p.179 it was stated that a number of investigators, although accepting the existence of an association between lung cancer and smoking, had questioned its significance in terms of a causal hypothesis. A number of references were given for this statement. One of these was a paper by Cohen and Heimann (1962) entitled "Heavy smokers with low mortality", which Sir Richard had not read. Another was by Eastcott (1956), entitled "The epidemiology of lung cancer in New Zealand". Sir Richard remembered this paper, but did not know that Eastcott was a proponent of the constitutional hypothesis. Another reference was Haag and Hanmer (1957), entitled "Smoking habits and mortality among workers in cigarette factories". Sir Richard did not know this paper. Another reference was to Little (1969). Sir Richard recognised Little as "the man that worked for the tobacco industry". This, he said, eventually invalidated views that he might hold, because initially he was regarded as a good experimentalist, but later there came to be "some very odd questions about his work". This was for more reasons than because he was funded by the tobacco industry, but Sir Richard could not recall the details of what Little did. He agreed that Little had been managing director of the American Society for the Control of Cancer, now known as the American Cancer Society. He accepted that he had been President of the University of Michigan, but did not know about other posts he had held. He knew that he had been regarded as a senior scientist, that was really all that he could say. He became Scientific Director of the Council for Tobacco Research in the United States. It was not this feature that caused Sir Richard to doubt his views, but his subsequent behaviour in that position. Sir Richard only had second-hand knowledge about this, but sufficient for him to form a view about Little's views.
[5.274] Reference was next made to Tokuhata 1972. Sir Richard was aware of Dr George Tokuhata, who had worked with Lilienfeld at Johns Hopkins University. He was Director of the Division of Research and Biostatistics of the Pennsylvania State Health Department and held adjunct appointments as Professor of Biostatistics at the University of Pittsburgh Graduate School of Public Health and Associate Professor of Community Medicine - Epidemiology at Temple University College of Medicine. At p.5, having stated that despite the fact that statistical methods could not establish proof of a causal relationship in an association, USSG 1964 had concluded that cigarette smoking was a major causal factor in human lung cancer, Tokuhata continued:
"However, a number of investigators, though accepting the existence of a statistical association, have questioned its significance in terms of causal hypothesis. Some of these doubts have been on the basis of a possible genetic underlay which might determine both smoking and lung cancer. Others have supported some type of constitutional theory or have claimed that the observed associations are 'spurious' because of selection biases in the design of study. Also, many experiments on inhalation of cigarette smoke in animals have failed to produce a single cancer similar to the most prevalent type of lung cancer in humans." (I have omitted the references in the original.)
Sir Richard did not take issue with this as an accurate historical account, but said that the statement about biases was made by Berkson, who did not continue to make it because he recognised it was wrong.
[5.275] At p.17 Tokuhata wrote:
"Although [the] smoking habit is considered by many observers as being influenced by environmental factors, there is sufficient evidence to suggest that genetic factors may also play a significant role in the induction of such habits [...]. At present neither the acquired nor the inherited factor in smoking is completely understood."
Sir Richard agreed that genetic factors might play a statistically significant role, though a small one, and did not take issue with the second sentence in this passage.
[5.276] At pp.27-28 Tokuhata referred to Fisher's "common genotype" hypothesis, and wrote:
"Fisher's theory has been challenged by some investigators on the ground that the twin data on which his argument was based were biased, and that the history of cancer in twins whose smoking habits are known has not been documented sufficiently. More data on smoking habits and medical histories are needed regarding other siblings, offspring, and parents. Our data on the familial study of lung cancer and smoking, as discussed earlier, are not consistent with the 'common genotype' hypothesis."
Sir Richard agreed that Tokuhata was taking Fisher's hypothesis seriously; he would have expected him to, from the work he did. His work, however, was not consistent with the hypothesis. Reminded that he had given evidence earlier that Tokuhata was a statistician who did not know much about epidemiology, Sir Richard said that this was at the time he wrote his first paper. He became experienced and eventually he did know quite a lot about the subject.
[5.277] The next published work on which Sir Richard was asked to comment was a discussion by WC Hueper of Fletcher 1966. Sir Richard said that he was familiar with Hueper and that he was the founding chief of the environmental cancer branch of the American National Cancer Institute. He did a great deal of work on cancer-causing agents in the environment. There were other people who did not perhaps get as much credit as Hueper did, but he did a great deal on it. The discussion by Hueper started:
"I am afraid that I am one of the ignorant or biased individuals who have not been persuaded as yet that from 80 to 95% of all lung cancers affecting males are induced by cigarette smoking. Among the various types of so-called scientific evidence often advanced in support of this well promoted concept and possessing, in my opinion, rather dubious scientific merits is the proposition that the well-known and startling differences in lung cancer incidence prevailing in Great Britain and the United States are attributable to the fact that the English cigarette smoker inhales larger amounts of carcinogenic chemicals than the American one because he smokes his cigarette to a smaller butt which contains an accumulation of carcinogenic matter."
Sir Richard remembered that this was postulated as one of the contributory factors about this time. He did research on it himself, based on actually measuring the lengths of the butts.
[5.278] Hueper continued:
"The argument is based entirely on statistical data measuring coincidental factors. As a matter of fact epidemiologic studies on variations in lung cancer incidence of numerous metropolitan areas in the United States have established that there exist within different regions of similar character in the United States fluctuations of lung cancer incidence which are even more striking than those reported for English and American populations."
Sir Richard said that he would think that was true, though he did not know the evidence. Hueper continued:
"It surely cannot be maintained that the cigarette smoking habits as far as the length of the cigarette butt is concerned are significantly different in different parts of USA."
Sir Richard said he had thought this was quite reasonable; he did not have the data for it and it was not something he would wish to postulate himself.
[5.279] In the next paragraph, Hueper said:
"It is moreover a fallacy to contend that there exists among American scientists any degree of unanimity as to the role which cigarette smoking plays in the causation of lung cancer."
Asked whether he agreed that Hueper was in a position to have a view on this matter, Sir Richard said not actually on cigarette smoking. He really did not look into that evidence. He knew Hueper well and he was very careful in his many observations on occupational hazards. But he attended committees with him and talked to him about this problem and he really did not understand it at all. He did not question Hueper's view about the lack of unanimity among American scientists.
[5.280] Hueper then said:
"The opponents of the cigarette theory are merely less vociferous than its proponents and have less official and organized support."
Sir Richard said this was nonsense. Hueper said:
"They do not appreciate moreover the repeatedly advanced accusation that their scientific judgment is colored either by their chain smoking habit or, worse, by some financial considerations supplied by cigarette companies. I am speaking here from personal experience.
The obviously exaggerated estimates as to the causal role of cigarette smoking in regard to lung cancer tend also to defeat quite effectively compensation claims of workers afflicted with well-recognized occupational lung cancer and to paralyze the organization of a well-balanced program of investigation and control of pulmonary cancer hazards."
Sir Richard said that this was the point he was making earlier about Hueper's enthusiasm for occupational hazards, which he grossly exaggerated. Asked whether this was another scientist who was simply calling for the application of the scientific method and more research into the question, Sir Richard said that he would not say that in relation to Hueper. Asked whether this was because of Hueper or because of the words quoted here, he said that it was because of Hueper. Hueper continued:
"It is time that the bandwagon atmosphere which has controlled for some time the campaign of the lung cancer causation from cigarette smoking be abandoned for a more sober and realistic approach to this serious problem."
Sir Richard said that this was not a view to which Hueper was entitled.
[5.281] Sir Richard was asked about further passages in Fletcher 1992. At p.535 the interviewer asked:
"I seem to recall from Sir Richard Doll that the Ministry kept asking the MRC to do more research to confirm the findings [of Doll and Hill 1950], and the MRC refused because they said that they were satisfied with the conclusion and no more research was needed. What made you start thinking that you should do something like getting the Royal College of Physicians to publish a report?"
In his reply Fletcher referred to having lunch with George Godber, Deputy Chief Medical Officer, who
"made it clear to me that his chief, Sir John Charles (the Chief Medical Officer), was dead against the Ministry being involved in any action on smoking because he thought it would cause trouble. I asked him if he thought it would help if we asked the Royal College of Physicians to produce a report on smoking, sort of side-stepping John Charles."
Sir Richard said that he knew this was true, he had personal knowledge about the position of Sir John Charles.
[5.282] Fletcher went on to explain how Sir Robert Platt, who had just become President of the Royal College of Physicians, decided that the College should become involved and formed a committee, one of whom was Fletcher, who edited the report, RCP 1962. Sir Richard said that the idea was to make clear to Members of Parliament what the evidence was: they were the audience. In the interview Fletcher said that from the beginning he realised that the voice of doctors expressed through the report must influence the politicians, they were the ones who had to act. There was tremendous coverage of a press conference to advertise the report; the media recognised that what was new was that a group of practising doctors were saying that smoking was dangerous. Asked (at p.536) what the response of the politicians was, he said:
"Neglect. Enoch Powell was Minister of Health and he agreed that the Ministry of Information should produce some posters, but he said that any action on this must be taken by local authorities through their health education money."
[5.283] Sir Richard was asked to note the following passage in RCP 1962, pp.53-55, under the heading "Possible action by the Government":
"112. Some decisive steps should be taken by the Government to curb the present rising consumption of tobacco, and especially of cigarettes. This action could be taken along the following lines:-
113. Public education. Much more imagination, effort and money should be devoted to drawing the attention of the public to the hazards of smoking. Special attention should be paid to effective education of schoolchildren, but use should also be made of every modern method of advertising, including posters, press notices and short items on radio and television. The attention of parents should continually be drawn to their responsibility for dissuading and discouraging their children from smoking. Such public education might also advise safer smoking habits for those whose addiction is too strong to be broken. Appropriate surveys of smoking habits should be organised periodically to ensure that accurate information about the effects of education by various means, especially of schoolchildren, is obtained in order to discover and implement the most effective of them.
114. An educational campaign among children might be supported by more effective restrictions on the sale of tobacco to children. [...]
115. Restriction of advertisement of tobacco. Any increase in public education concerning the risk of smoking would at present be in conflict with the vast expenditure on advertising tobacco. While it may be doubted whether advertisement does much to initiate the smoking habit, and it is predominantly designed to attract smokers towards the advertiser's particular brand rather than to increase overall consumption [...], legislation to prevent or at least to restrict the advertisement of a habit which causes such widespread injury to health would be reasonable and would provide evidence of official acceptance of the reality of the hazard. [...]"
Other proposed measures were wider restriction of smoking in public places, taxation, smoke analyses on cigarette packets and anti-smoking clinics.
[5.284] Sir Richard was asked about a confidential Cabinet Office Report by a committee of officials who had been asked to consider the legal and administrative implications of the recommendations of RCP 1962. In this report the committee made various recommendations about education of the public, sale of tobacco to children, tobacco advertisements, smoking in public places, taxation, smoke analyses on cigarette packets and anti-smoking clinics. Sir Richard said that he had not seen the report (which was marked "confidential"), and would rather not express an opinion, in light of it, on whether he shared Fletcher's view that the Government's response to RCP 1962 was "neglect". He vaguely recalled that the Government had circularised local health and education authorities, calling on them to make the conclusions of RCP 1962 widely known in order to advance public education, but this was not a subject he had really been personally concerned with; his concern had always been with the science of the matter, not with what was done about it.
[5.285] Counsel then turned to a different topic, the evidence given by Sir Richard to the effect that experiments had shown that tobacco tars were carcinogenic to the animals exposed to them. In Wynder et al. 1955 at p.448, under the heading "significance of animal-cigarette research", the authors wrote:
"The ultimate proof of whether or not a given agent causes cancer in man rests exclusively on clinical, statistical, and incidence data obtained from man. [...] An animal experiment cannot significantly add and certainly cannot detract from these human observations. The purpose of the laboratory experiment is chiefly to identify active carcinogenic components which must be in cigarette tar. Such a procedure can only be done through laboratory investigations. It can only be assumed that the carcinogen(s) isolated for the animal will also be the carcinogen active for man."
Sir Richard said that there were differences of opinion about animal experiments, and it was not everybody's belief that a carcinogen isolated for an animal will also be active for man, but it was his belief.
[5.286] In Wynder 1959 the author wrote, at p.317:
"The importance of laboratory work is not to prove that smoking is a cause of cancer in man. Such proof can only come from human epidemiological investigation."
Sir Richard said that he believed that. IARC did not, but now he believed it. Wynder continued:
"Laboratory research can, however, contribute to and give a logical explanation for the human findings."
Sir Richard accepted that experiments on rats, mice, hamsters and various other animals continued beyond experiments on dogs, which he said were stopped on humane grounds because of objections to the methodology of using a tracheostomy for smoke inhalation.
[5.287] Reference was next made to Stewart and Herrold 1962. Sir Richard said that he knew Stewart very well. He was at the Laboratory of Pathology at the National Cancer Institute in Bethesda, Maryland. The paper began:
"Were it possible regularly to induce pulmonary cancers in laboratory animals by exposure to inhalation of cigarette smoke, this would constitute good evidence that cigarette smoking could cause cancer of the lung in man."
Sir Richard said that he would not agree with that wholly, but it would contribute to the evidence.
[5.288] RCP 1962 stated at p.26, para.37:
"Skin cancer can be produced in mice by applications of tar condensed from tobacco smoke but the results obtained by various investigators have not been uniform and exposure of animals to tobacco smoke in inhaled air has failed to produce lung cancers."
Sir Richard said that this was correct, squamous cell carcinomas had not been produced, though Wynder had produced adenocarcinomas. RCP 1962 continued:
"Moreover the amount of cancer-producing substances in the smoke itself does not seem likely to be sufficient to account for the large number of cases of cancer associated with the habit."
Sir Richard said that he did not agree with this because he did not think they knew all the carcinogenics, it was probably an unwise statement. It would have been correct if the statement had been that the amount of benzo[a]pyrene present was not sufficient, but it did not recognise that there were so many other carcinogens in tobacco smoke, the effects of which had to be taken into account. That was not done at the time.
[5.289] IARC 1986, in the chapter entitled "Conclusions and evaluations", stated at p.309:
"Cigarette smoke has been tested for carcinogenicity in experimental animals by inhalation and by topical application of condensate and in other ways. Exposure of hamsters and rats to whole smoke results in the induction of malignant respiratory-tract tumours."
Sir Richard was taken to the passage of the report which related to this conclusion. At p.127 the report stated:
"Although the evidence for carcinogenicity of tobacco smoke emerged first in humans, there was need for an inhalation model in experimental animals in which the carcinogenicity of different types of tobacco and tobacco products could be studied. The availability of animal models also permits the comparison of different products and the investigation of various modifying factors in the development of respiratory cancer. [...] The discovery that inhalation of tobacco smoke caused carcinomas in the larynx of hamsters established a model system in which the carcinogenicity of tobacco smoke for hamsters has been confirmed repeatedly [...]. Few bioassays have been conducted, however, in longer-lived mice, rats and dogs, because of low priority and high cost; therefore the spectrum of possible tumour responses in animals to inhalation of tobacco smoke is little known."
[5.290] On succeeding pages papers relating to experiments on mice, rats, hamsters, rabbits and dogs were reviewed. The papers relating to an experiment on dogs were by Auerbach and others and by Hammond and others, both published in 1970. IARC 1986 noted that in this report of an experiment, in which groups of male beagle dogs, trained to inhale cigarette smoke through tracheostomata, were exposed to smoke from different types of cigarette, while a non-exposed group served as controls, no data were given on specific smoking parameters or measures of exposure. It was also noted that a small number of control dogs were used and there was an unusually high incidence of lung tumours in these animals, since these were tumours that rarely occurred spontaneously in dogs. The focal inflammatory lesions usually found in the lungs of animals exposed to smoke were not mentioned in the report. Examination of the upper respiratory tract and other organ systems was not reported. The authors' interpretation of the photomicrographs as representing neoplasia was considered to be not entirely convincing. Sir Richard agreed that some aspects of the dog experiment were "getting the thumbs down" from the IARC working group (of which he was chairman).
[5.291] At p.194 of IARC 1986 it was stated:
"Studies involving inhalation of smoke are hampered by difficulties in reproducing the exposure of humans. Technical problems occur in the generation of smoke and its delivery to animals; moreover, the respiratory systems of animals and humans differ. Rodents are obligatory nose breathers, and the structure of their nasal turbinates is more complex than that of humans. Unlike humans, experimental animals smoke involuntarily, with shallow, hesitant breathing patterns. Other difficulties are caused by the toxicity of nicotine and carbon monoxide. Despite these problems, however, informative data have been obtained concerning the carcinogenicity of whole smoke and its gaseous phase."
Sir Richard said that he believed these statements to be correct.
[5.292] The report continued:
"In some experiments in mice, exposure to whole cigarette smoke resulted in the induction of lung tumours."
Sir Richard agreed that these were not described as "malignant" lung tumours. The report also stated:
"In one study involving long-term exposure of rats to cigarette smoke, tumours of the respiratory tract were induced."
Again, he agreed that these tumours were not said to be "malignant". By contrast, the report stated:
"In hamsters, various experiments demonstrated reproducibly the induction of laryngeal carcinomas."
Sir Richard agreed that these were stated to be malignant. The report continued:
"Studies in rabbits and dogs of whole cigarette smoke were inadequate for evaluation. No treatment-related tumour other than in the respiratory tract has yet been produced by smoke inhalation in experimental animals."
Sir Richard was asked to contrast the statements on p.194 with the statement on p.309:
"Exposure of hamsters and rats to whole smoke results in the induction of malignant respiratory-tract tumours."
Sir Richard agreed that it was implied on p.194 that the tumours of the respiratory tract were not malignant. He said that he would need to look at the original paper.
[5.293] He was referred to Dalbey et al. 1980, which related to the chronic inhalation of cigarette smoke by rats of a particular strain. The authors stated that their choice of the rat was heavily influenced by the induction of squamous carcinomas in the respiratory tract after intratracheal instillations of relatively small amounts of polycyclic hydrocarbons or after bronchial implantation of pellets containing cigarette smoke condensate. Sir Richard agreed that the rats had been chosen because of the ability to induce squamous carcinomas in their respiratory tracts. Sir Richard was asked to consider Table 2 on p.386, in which percentages were given for rats with tumours at specific sites. There were columns for the untreated and sham-exposed groups of controls, and for the smoke-exposed rats. One rat, representing 1% of the combined controls, had a tumour of the respiratory tract, while seven rats, representing 9% of the smoke-exposed group, had this tumour. A footnote stated that the figures for the smoke-exposed group were (statistically) significantly different from combined controls at P<0.05. In the text it was stated:
"We observed 10 respiratory tumors in 7 smoke-exposed rats. Nasal tumors occurred as 1 early adenocarcinoma and 1 squamous cell carcinoma. The pulmonary tumors were 5 adenomas, 2 alveologenic carcinomas, and 1 squamous carcinoma."
Sir Richard was asked how many of these ten tumours were malignant and how many were not, including the nasal ones, and he said there were five.
[5.294] The article continued:
"One alveologenic carcinoma was observed in controls. Two pulmonary adenomatoid lesions were observed in the controls and 1 was found in the smoke-exposed group. Also, 1 large, highly keratinizing squamous carcinoma was found in the head, including the nasal cavity, in each of 2 rats, 1 from the control group and 1 from the smoke-exposed group. These tumors are not included in Table 2 because their origin could not be determined. [...] Besides the description of laryngeal neoplasms in smoke-exposed hamsters [...], the present work is the only study in which an unequivocal tumor response in the respiratory tract resulted from long-term tobacco smoke exposure."
Sir Richard agreed that the authors were here not contending for an unequivocal "malignant" tumour response. He accepted that this was the only rat study in which statistically significant numbers of tumours were produced.
[5.295] The authors continued:
"The tumors in the respiratory tracts of smoke-exposed rats consisted of 5 adenomas and 5 adenocarcinomas, alveologenic carcinomas, or squamous carcinomas as compared to 1 alveologenic carcinoma in the control animals. This amounted to 10 tumors in 80 smoke-exposed animals as opposed to 1 tumor in 93 controls. All but 2 of the tumors in the smoke-exposed animals were in the lungs. Previous studies on increased tumor incidence in rats after tobacco smoke exposure had been much less definite."
[5.296] The authors concluded, on p.398:
"In summary, our studies with SPF F344 rats showed that significant damage could be induced in various parts of the respiratory tract by lifetime exposure to tobacco smoke. This damage consisted of hyperplastic and metaplastic epithelial lesions in the upper airways and of focal alveolitis and alveolar fibrosis in the lungs. The incidence of respiratory tract tumors was significantly elevated in smoke-exposed rats."
Sir Richard agreed that the authors did not contend that the incidence of malignant tumours in the respiratory tract was significantly elevated in smoke-exposed rats. If they were intent on attempting to induce carcinomas in the respiratory tract, one would expect them to say if they had succeeded. But of course, he said, the adenomas were regarded as a preliminary stage in this sort of experiment. They were a stage towards the development of, and frequently in these studies taken as equivalent to, malignant tumours. He accepted, however, that the authors did not contend that the adenomas were in any sense malignant, they contended that the incidence of respiratory tract tumours was elevated, of which five were benign and five malignant.
[5.297] Counsel asked Sir Richard to consider the guidelines set out in McConnell et al. 1986, bearing in mind that they were published after the meeting of the IARC working group which produced IARC 1986. In this paper the authors stated that a question frequently raised in the work of the National Toxicology Program (NTP) was whether certain neoplasms should be combined for overall assessment of rodent carcinogenesis data. If there were neoplasms in the same organ or tissue, or in different organs or tissues in which the morphology of the tumours was comparable, combining or not combining them might affect the carcinogenic evaluation of a given chemical, because combining them might produce statistical significance when not combining them would not:
"For example, 4 benign and 4 malignant neoplasms of the same cell type are found in a group of 50 treated animals versus none in 50 control animals. When the incidence data are analyzed separately, the increase in benign and malignant neoplasms does not reach significance with the use of Fisher's exact probability test [...]."
On p.286, Table 1 set out guidelines for combining neoplasms in the F344 rat and a particular strain of mouse. Among the guidelines relating to the respiratory system, it was stated that in the nasal cavity squamous cell neoplasms and glandular cell neoplasms should not be combined, and in the lung squamous cell neoplasms and broncheoalveolar neoplasms should not be combined. Sir Richard agreed that if the guidelines were applied to the results shown in Dalbey et al. 1980, one adenocarcinoma and one squamous cell carcinoma found in the nasal tract would not have been combined with the other tumours, the squamous carcinoma and the alveologenic carcinomas of the lung would not have been combined, and tumours of the bronchi would not have been combined with those found in the peripheral lung. Sir Richard added the comment that IARC did not necessarily adopt these guidelines.
[5.298] Counsel asked Sir Richard about the hamster studies. Reference was first made to Dontenwill et al. 1973. At p.1791 the authors wrote:
"[I]n earlier investigations, we could not observe carcinoma infiltration in the cartilage of the trachea or the larynx. Tumor growth spread almost always between the bridges of the cartilage and the area of the pars membranacea."
Sir Richard agreed that this related to neoplasms of the larynx in hamsters. At p.1801 the authors wrote:
"We were chiefly interested in the substances in the particulate phase of the smoke. Since man inhales via the oral cavity, it may be assumed that, in spite of complete inhalation, only about half the smoke particles remains in the respiratory tract. As already explained, part of the smoke is exhaled with the next expiration, and a further portion is carried away by the lung clearance mechanisms. Because of the intervals of 58 seconds between each 2-second inhalation, inhaled smoke particles can be removed by exhalation and lung clearance."
Sir Richard said that this was correct, and there was nothing he regarded as contentious in this passage.
[5.299] The authors continued:
"The situation in animals was completely different with our experimental techniques. The animal breathed from a vacuum-free system a smoke-air mixture that remained at a constant ratio of 1:15. The animal was continuously exposed to this mixture for 10-minute periods not more than 3 times a day. After a few breaths, the lungs were totally filled with the equivalent smoke-air mixture.
Since the inhaled mixture was constant, there was no interval between successive smoke inhalations and hence no cleansing exhalation or intervals for lung clearance to take place. The hamster normally has a respiratory frequency of 75 breaths/minute. According to previous investigations, respiratory frequency increased during exposure to smoke. On the assumption that the hamster's respiratory frequency was 100/minute during exposure to smoke, the animal inhaled the smoke-air mixture ≈ 1000 times during the 10-minute exposure with almost no lung clearance during that time. Since the flow characteristics of air in the upper regions of the respiratory tract are different in the golden hamster and in man, the highest deposition of smoke particles per surface unit occurred in the region of the larynx. The concentration there was ≈ 300 times that in the lungs and bronchi [...] under theoretical conditions of an equal distribution of smoke at the surface.
The dose per surface unit has a significant and important influence on the effects. Also, clearance is more difficult in the region around the larynx where the epithelium is not ciliated, and the laryngeal squamous epithelium is possibly more sensitive than the cylindrical epithelium of the bronchi and trachea [...]. Interpretation of all the factors shows that it is almost impossible to calculate an exact dose comparison between man and experimental animals. However, the effective dose acting on the larynx of the experimental animal is many times greater than the effective dose per surface area in the respiratory tract of man. These calculations do not affect the comparison between 2 experimental groups for which the same methods and techniques were utilized to show which dose of type of modified cigarettes had a stronger or weaker effect."
Sir Richard agreed that in this paper there was no report of carcinoma of the lung or any part of the respiratory tract other than the larynx in the hamsters which had been exposed to cigarette smoke in this experiment. He agreed that, subject to intervention in the course of the experiment, cancer involved growth, invasion, metastasis and finally death of the experimental animal. He further agreed that while the paper gave an explanation as to why it was thought that laryngeal carcinoma was induced in the experimental hamsters, there was no explanation as to why laryngeal carcinoma had not been induced in mice or rats exposed to inhalation.
[5.300] Reference was also made to Dontenwill et al. 1977, in which the authors carried out a further study on the effects of chronic cigarette smoke inhalation in Syrian golden hamsters. At p.11 of a translation from the German, prepared for the purposes of the proof, the authors stated that in summary the findings of the second smoke-exposure experiment confirmed the results of the first experiment to a large extent.
"The results show that inflammatory, pre cancerous and tumorigenic effects depend in number and intensity upon the dose and time. A reduction in the quantity of inhaled condensate reduces the frequency of inflammations and tumors in the respiratory tract."
Sir Richard agreed that, according to this paper, the authors had again succeeded in inducing laryngeal cancers in hamsters. This was subject to the suggestion that the strain of hamsters used in the experiment might be a relevant factor.
[5.301] Counsel next asked Sir Richard about Bernfeld et al. 1974, which was also referred to in IARC 1986. The article started by referring to the work of Dontenwill et al. 1973, in which "randombred Syrian golden hamsters" were used. The authors used a model system which they stated, at p.1150, differed from that reported by Dontenwill et al. in several important respects, using two groups of inbred hamster lines. At p.1150 they stated:
"The point of greatest practical importance to emerge from our work is the striking differences among various lines of hamsters with respect to susceptibility to acute toxic effects of smoke and to hyperplastic response of the larynx to smoke. Animals of the inbred BIO 15.16 line have both the highest resistance to smoke or nicotine toxicity and the greatest laryngeal susceptibility - qualities greatly increasing the sensitivity of the model. Further studies with larger numbers of animals will be necessary to ascertain the significance of the laryngeal changes in different strains."
While noting these statements, Sir Richard said that he thought that questions about them should be put to one of the pathologists on the IARC working group. Although he had been chairman of the working group, he did not set himself up as an expert on the subject-matter of these papers.
[5.302] Finally, Sir Richard was asked about benzo[a]pyrene. He agreed that this substance was produced by everything that was burnt and was present in the air that was breathed. As he had previously said, smoking thirty cigarettes a day was not equivalent to exposure to a strong carcinogen. He had also said that whether cigarette smoke acted even as a weak carcinogen was more difficult to determine, though it was necessary to be careful in the use of these terms, because the quantity of a carcinogen was a factor as well as its strength or weakness: "The poison is in the dose."
[5.303] Sir Richard agreed that he was giving evidence without fee (but in the hope of having his fares covered) in order to help the public health as well as to help the court. He had been a supporter of ASH over the years and was not sure whether he was president of it (though in fact he was).
Re-examination of Sir Richard Doll
[5.304] In re-examination Sir Richard said that by the late 1950s only a very small minority of scientists did not accept that cigarette smoking caused lung cancer. The standard of proof he looked for was proof beyond reasonable doubt. In the case of infectious diseases, Koch's postulates were applied to establish proof of causation. It was inappropriate to apply these postulates to chronic diseases, which required a very much more difficult procedure. A mass of evidence had to be taken into account, chance, bias and confounding had to be excluded and one then had to go on to look at the positive evidence in support of causality. Bradford Hill had drawn up nine guidelines, which grew out of his and Sir Richard's study of smoking and lung cancer. On the basis of such evidence, one could only say that one believed that a relationship was a causal relationship beyond reasonable doubt. One could not go further than that. In the case of occupational hazards, if a chemical agent was removed and the disease from which the workers were suffering disappeared, the causal relationship was confirmed. As regards smoking, if smoking was reduced, the diseases due to smoking became less common. "So what you have to do is decide that something is proved beyond reasonable doubt and then test it in practice, that your conclusion is correct." This was the standard he had used throughout his career as an epidemiologist.
[5.305] Professor Hastings, PhD, BSc, aged 49, was Professor of Marketing at the University of Strathclyde. He had been Director of the Centre for Social Marketing and the Centre for Tobacco Control Research there since 1987. He said that the Centre for Social Marketing was previously known as the Advertising Research Unit and was founded in 1980 by his predecessor. Once Professor Hastings had taken over the directorship, in 1992, to reflect the broadening of its advertising to look at the marketing process more generally, its name was changed to the present one. The Centre for Tobacco Control Research was founded in the late 1990s, with funding from the charity Cancer Research UK and other non-profit and Government organisations. He said that the two research centres really provided a channel for his research activity as an academic, which was an important part of his job.
[5.306] Professor Hastings served as Head of Department in the Department of Marketing from 1996 to 1992. From 1980 to 1987 he was a Research Fellow in the Advertising Research Unit. From 1978 to 1980 he was a Research Executive with Purchaser and Consumer Studies Ltd, a market research agency in London. He said that, having done a degree which covered social research methods, it was a natural progression to go and try and put some of those theoretical skills into practice in market research. Purchaser and Consumer Studies Ltd was a small company, working for a variety of different clients, mainly in the fast moving consumer goods area, and was interested in advertising and marketing. His first degree was in sociology and social research and his doctoral subject was social marketing. In his CV he listed numerous publications, including Government and other reports, of which he was author or joint author.
[5.307] Among his public and other appointments, Professor Hastings had been Special Advisor to the House of Commons Select Committee on Health and United Kingdom representative on the Comité Scientifique et Technique du Fonds Communautaire du Tabac. He explained that this was a European Commission body that was concerned with the control of tobacco products, and he had been asked by the Department of Health in London to take on that role. He was a member of the Health Technology Board for Scotland, and had been the Chair of the Evaluation of the CRC Care Education Research Group, which had been put together to see whether a particular research centre was doing a sensible job, an instance of academic peer review. He had been Assessor for the Department of Health Smoking Research Initiative, as a member of the panel who commented on tenders on research to do with smoking. He had been Convenor of the Health Education Authority Expert Group on the Mass Media and Social Depravation. He explained that this was one of a series of working groups which had been set up to look at the issue of the "health divide", a phenomenon of particular note in the United Kingdom where degree of deprivation and health status were strongly correlated. He was a founding member of the International Network of Tobacco Control Organisations. He said that these were research organisations, involved in doing research on tobacco control issues. It was an attempt to try to avoid duplication of effort and share learning and generally take forward concerns and research in this area on an international basis. As an example, they at Strathclyde were the leaders in the United Kingdom in an international study of adult smokers looking at their response to different policy interventions of one sort and another. The research was being done simultaneously in Canada, the United States and Australia and there were principal investigators in each country. It was very difficult to measure the effect of national policy. The classic way in which one would measure such effects would be to have some sort of control group, and it was difficult to find a control group in Britain for example. He was the invited expert at the Government Tobacco Summit. He described this as "very exciting". It was when the present Government first came into power and they had a strong commitment to do things about the problems of smoking. If they were to have a serious public health agenda, then tobacco had to be dealt with as well as possible, because it contributed such a big toll. The Government had called the summit to talk through the issues and "get learning from overseas". UKWP 1998 was one of the things that came out of it. Tobacco had to form a big part of any public health strategy.
[5.308] Professor Hastings said that in addition to having been Special Advisor to the House of Commons Health Committee on the Tobacco Industry and the health risks of smoking, he was currently Special Advisor in connection with an inquiry into obesity. He offered advice to the Committee about what things they should be investigating, what they should be concerned about, and who might be called as witnesses. The process operated by the Committee calling in people who were expert in the field and asking them questions and to provide written statements and written evidence. They needed to collect that evidence together, synthesise it and draw conclusions from it. The role of the expert was to advise them in that process and suggest, for example, questions they might ask of the witnesses. His advice had been about tobacco advertising and marketing. He suggested to the Committee what documents they should seek from the advertising agencies that worked for tobacco companies, such as creative briefs and market research reports.
[5.309] Professor Hastings's report continued:
"The health evidence about tobacco was therefore well known and publicised by the 1960's. For Mr McTear, however, the effects of this health information have to be balanced against three powerful counter-veiling [sic] forces emanating from the industry in general and Imperial Tobacco in particular:
1. Tobacco advertising
2. Health risk denial
3. A reluctance to introduce health warnings"
[5.310] Professor Hastings, who was born in 1954, remembered that his father smoked a pipe and, although his parents were very concerned for the health and welfare of their children, it did not seem to occur to them that this was an odd thing to do. He was asked about an article in The Herald by Ron Ferguson, dated 9 October 2003. In it the writer stated that attitudes to smoking had changed dramatically within the past three decades. He grew up at a time when nearly everybody smoked. Both his parents smoked, as did most of the adults he knew. His friends and he were determined to be fashionable when they started smoking. "To be a non-smoker was to be a freak with a headache." Professor Hastings agreed with the idea that young people in particular would take up a habit because it looked cool and fashionable. This was still true of young people in their relationship to tobacco. There was a glamour that people could associate with tobacco. The generation of moods and association was so much part of the promotion of tobacco, and tobacco advertising had long been predicated on "those sorts of levers" of encouraging the smoking process. Also the ubiquity of tobacco smoke and smoking was important. It was now recognised that one of the key policies one could adopt in order to reduce tobacco consumption was to control smoking in public places and in the work place.
[5.311] A few years previously they had done some research into smoking in deprived areas. There was a strong correlation between social class and smoking, which had been increasing over the years, and "the further down the social grade you are, the more likely you are to smoke." There was a concern that the very poorest in the multiply deprived areas had much inflated smoking prevalence levels, and because of that they were commissioned to do a survey in Castlemilk in Glasgow, which was one such area. They found that smoking rates there were about 56%, rather than 25% to 30% as they were in the population as a whole. One of their research techniques was to talk to people in focus groups. It was difficult to find non-smokers. There were ex-smokers who typically had some sort of health reason for not smoking. Previous research had been done by Alan Marsh earlier in the 1990s. He put the prevalence of smoking among poorer people down to "the sheer disappointment of being poor in Margaret Thatcher's Britain." He tied it in very closely with the problems of having very low aspirations, hopes and capabilities. Professor Hastings said he did not know about the reference to Margaret Thatcher, but he thought the principle was a good one. He had done focus groups in Jarrow in Northumberland with single mothers in very straitened circumstances, and tobacco provided one of the few pleasures they had in a life that was otherwise fairly bleak. "So it seems perverse but it does seem to provide people with some sort of support and tobacco promotion has actively exploited that."
[5.312] Advertising had now gone, he continued, but for many years there had been strong promotional efforts put into the cheapest brands to try and make people feel better about smoking what were really very cheap cigarettes and, therefore, inevitably lower quality cigarettes, to reassure them that they were OK and it was a reasonable thing to do.
[5.313] Professor Hastings's report continued:
"At the time Mr McTear took up smoking the tobacco companies, including Imperial Tobacco, were putting enormous resources into the marketing of their products. A combination of advertising, point of sale material, price promotions and pack design were used to enhance the attractiveness of the product. As now, this material emphasised images and associations rather than hard fact, building positive and powerful brand identities like Player's."
[5.314] Professor Hastings was asked to comment on some documents relating to expenditure on advertising. ITL had prepared, for the purposes of the proof, a table showing their expenditure on advertising and marketing in the UK on cigarettes and cigarette hand-rolling tobacco in each of the years 1960 to 1975. For the year ending 31 October 1960 the amount was £8,182,748, and for the year ending 31 October 1975 it was £15,802,000. In RCP 1962, p.6, para.10, it was stated:
"There have been impressive increases in expenditure on the advertising of tobacco goods in recent years. [...] During a period (1955-1960) when total expenditure on advertising had not quite doubled, expenditure on tobacco advertising has increased threefold. The total expenditure in 1960 was approximately £11,000,000."
Professor Hastings commented that expenditure by ITL made up quite a large percentage for that total in 1960, "so they were big players in the advertising market."
[5.315] RCP 1971, pp.16-17, para.1.11 stated:
"Substantial competition between the major manufacturers dates chiefly from 1955, when leaf supplies became more plentiful after post-war scarcity. Since then, expenditure on advertising of other goods has not quite doubled, while expenditure on tobacco advertising has more than trebled. After the end of July 1965, advertising of cigarettes was banned on television, and subsequently there was a moderate rise (greater than the increase in sales) in advertising of pipe tobacco and cigars. There were few gift-coupon schemes till 1963, but expenditure on this form of sales promotion is now larger than that devoted to direct advertising [...]."
[5.316] Reference was made to Table 1.1 on p.17, where figures were given for the UK expenditure on sales promotion of cigarettes, tobaccos and cigars in the years from 1955 to 1968, with totals both actual and adjusted for the increase in cost over the years, taking 1960 as the base for this purpose. Professor Hastings said that he supposed that an index of media costs, which could vary from general inflation, had been used for the adjusted totals. To compare what was being spent by the tobacco companies on promotion with what was being spent by the Government, it was necessary to look at the actual costs. He also commented that after 1965 expenditure on cigarette advertising as a whole continued to rise, notwithstanding the ban on television advertising, and there was now a lot of research evidence that this was precisely why partial bans on advertising simply did not work. The continued television promotion of pipe tobacco and cigars seemed slightly extravagant, when compared with the increase from sales, and one had to assume that part of the agenda here was to promote the idea of smoking and tobacco use.
[5.317] On p.18 of RCP 1971, in para.1.12, it was stated that despite the prohibition of television of advertising, the total expenditure on advertisements in 1968 amounted to £17 million and on coupon schemes to over £35 million, making a total of £52 million. At p.20, in para.1.15, this figure was contrasted with the £100,000 per annum spent by the newly established Health Education Council on publicity about the dangers of cigarette smoking. Professor Hastings commented that there was clearly an enormous divergence in the buying power of the two sides of the debate, and indeed that disparity had continued down the years.
[5.318] In RCP 1977, at p.20, it was stated:
"Expenditure on tobacco sales promotion by advertising in various ways and by gift coupons in cigarette packs has risen steadily in recent years through something in excess of £80 million in 1975 [...]. Until 1973, these increases just kept pace with the falling value of the pound, but by 1975 the pound had fallen so steeply in value that the effective expenditure appeared to have fallen too."
Reference was made to Table 1.1a on p.22, in which the total estimated advertising and promotional expenditure, including expenditure on coupons and sports sponsorship, was £82.9 million in 1975. This was made up of £25.4 million on advertising in the press and through posters, television and cinema, £52.5 million on coupons and £5 million on sports sponsorship, the latter figure being an estimate provided by ITL. Professor Hastings said that he thought that this expenditure of £5 million was in addition to the ITL figure of £15,802,000 referred to above. This was an example of compensating for restrictions on main media advertising. RCP 1977, in Table 1.1b on p.22, gave figures for Government anti-smoking publicity expenditure. In the year 1975 to 1976 expenditure by the Health Education Council was £707,000 and by the Scottish Health Education Unit was £223,000.
[5.319] Professor Hastings was next asked about a series of advertisements. During the course of this, objection was taken on behalf of ITL on the ground that there was no record for any line of evidence going to the general nature of cigarette advertising and any line of evidence going to the effects of cigarette advertising. Article IX of condescendence starts:
"The defenders have spent millions of pounds since 1960 in the promotion of smoking of their cigarettes and tobacco. They have instructed their advertisers to project an image of cigarette smoking where it is seen by the general public as 'very much the thing to do' and as a means of reducing stress. They instructed their advertisers to project it as a completely safe thing to do."
In the course of discussion about this objection, Mr McEachran sought leave to amend the pursuer's pleadings by adding averments in these terms:
"The defenders' advertising since 1960 has projected an image of cigarette smoking where it is seen by the general public as 'very much the thing to do' and as a means of reducing stress. A combination of advertising, point-of-sale material, price promotions and pack design have been used to enhance the attractiveness of the product. The material emphasises images and associations rather than hard fact, building positive and powerful brand images for their products. The defenders' tobacco advertising since 1960 has encouraged people, including Mr McTear, to take up and continue smoking."
This motion was opposed, and after further discussion I came to the view that the proposed amendment came too late and I refused the motion. I then sustained an objection to the line of evidence, on the ground that there was no record for it.
[5.320] This decision clearly gave difficulty to both Mr McEachran and Professor Hastings, and on a number of occasions, in the face of repeated objections, questions were asked or answers were given which appeared to me to ignore my decision. I made it clear that I would pay no heed to evidence about the matters referred to above for which there was no record. The summary of Professor Hastings's evidence which follows omits reference to such evidence.
[5.321] Professor Hastings said that it was highly likely that ITL would have employed advertising agents in the period from 1960 onwards. It was likely that they would have been given instructions as to how to proceed by ITL. There was a great deal of money involved and there would be planning documents which would be carefully thought through, and options would be researched to target audiences to make sure they were doing what they hoped they were doing. Asked whether inferences could be drawn from tobacco advertisements as to the sort of instructions ITL were likely to have given their advertisers, Professor Hastings said that he thought it was fairly apparent that they must have been seeking to attach certain values to the product, particularly values relating to the brand. Much as was done today, careful, creative briefs would be developed and careful designs would be used to get across, not overt messages, but associations between a product and certain values. The process of seeking to attach particular values would not only be the subject of instruction, but also careful research, particularly with the target groups for that material. Advertising was built on an understanding that communication was not a one-way process, it was not just about an active communicator telling things to a passive audience who then responded in a particular way, the consumer was actively involved in the process of communication and therefore good advertising was designed, using careful research, to ensure that language, images, feelings and associations were being used that would resonate with the target group.
[5.322] Professor Hastings was asked to comment on some documents obtained from ITL's advertising agency relating to creative briefs for the advertising of two new brands of cigarettes in the late 1990s, Richmond and Regal. He said that the need was not just to produce a physical cigarette with tobacco and paper, but also to imbue that with brand value. Branding was something that pervaded all areas of fast-moving consumer goods these days. It was a recognition that people bought things, not just for their physical practical use, but because they wanted to satisfy emotional needs, to express themselves, to show that they were fashionable, up to date, rebellious or whatever it might be. The documents were trying to pin down something which was really very difficult to pin down, because it was about impressions and images and associations, trying to get people to feel in a certain way about the product. People had mixed feelings about buying and consuming tobacco. Advertising could reassure and help them feel less ambivalent about it, and the advertising agencies cleverly recognised that. Professor Hastings would have expected ITL to employ advertising agents in the 1960s. The process of advertising had always been one of purposeful communication. Large sums of money were involved and he would have expected the process to have been done carefully and systematically.
[5.323] Professor Hastings drew attention to various statements, reported in UKHC 2000, in support of his view that the tobacco industry in general and ITL in particular had denied the health risks associated with smoking. Among these was a statement by the World Health Organization that for decades tobacco companies, including those in the United Kingdom, had "denied or minimised the overwhelming scientific evidence of the dangerous effects of tobacco." He said that the position adopted by Mr Davis in his evidence to the Committee (referred to in my account of Mr Davis's evidence at paras.[2.10] to [2.14]) confirmed "this Canute-like stance".
[5.324] Finally, on the question of a reluctance to introduce health warnings, Professor Hastings wrote in his report:
"Over the years the tobacco industry has also shown some reticence about health warnings. A British American Tobacco paper from 1974 states:
'If Government insists on warning notices on packs it is essential that the warning notice should be attributable to the Government or some other official body.'
A later BAT (1976) document states:
'If to comply with legislation warning notices have to be printed on packs, Companies must endeavour to ensure their attribution to Government or some other official body [...].'
If Imperial Tobacco were genuine in their desire for Mr McTear to know about the health consequences of smoking they would not only have refuted this stance, but also moved to put health warnings on their products and advertising as soon as the health risks were established. The Royal College of Physicians Report in 1962 would have been a suitable time, for example."
Professor Hastings said that he was not an expert in the epidemiology of smoking, but his position would be that, when extremely well-reputed bodies like the Royal College of Physicians, and indeed the Government and its American equivalent, were beginning to declare that this was beyond all doubt now established, then a responsible company would see it as part of its duty to inform its customers about the problems. He thought that a message coming from the company would have greater resonance than one coming from the Government or an official body. In reality they did nothing "until compelled to do so by primary legislation in 1971". By then it was too late for Mr McTear.
[5.325] He concluded his report as follows:
"There was undoubtedly considerable publicity about the health consequences of smoking during the 1960s, and Mr McTear must have been aware of it. However, in my opinion, its impact on his decisions about smoking would have been severely reduced by Imperial Tobacco's advertising, public denials of these health consequences and unwillingness to use health warnings until they were obliged by the legislation to do so."
Asked about the statement that Mr McTear must have been aware of the publicity, Professor Hastings said that it would surprise him if he was not aware that there was debate about this issue. His experience with the media was that they were not really interested in stories that had only one side. Debate was an important part of how they covered things. The media were covering the issue of smoking and health to a certain extent in a spontaneous way, and the Government was making ad hoc statements. But there was not what he would regard as a coherent campaign designed to move people in a certain direction until the late 1960s, when the Health Education Council and the Scottish Health Education Unit were established. This should be contrasted with a situation where tobacco advertising was very much in the form of a coherent campaign that was carefully designed to meet certain objectives and push one side of the debate, which was that smoking was acceptable, a fine and pleasant thing to do. Looking at it from Mr McTear's point of view, he would have received far more pro-tobacco advertising messages than anti-smoking media messages, particularly when account was taken of the fact that not all the messages coming from the media were anti-smoking. For most people, even now, tobacco was a small part of their lives, which were difficult and complex. They had to make challenging decisions day in and day out. The media coverage would make an impact, whether it be pro-tobacco advertising or anti-tobacco advertising or other sorts of media mention. Advertising was likely to be more successful because it was specifically designed to engage in that sort of persuasion. For many years research had been done with young people and one of the findings was that in the past young people were inclined to agree that tobacco smoking could not be that bad, otherwise the Government would not allow people to advertise it. The television advertisements for cigarettes, which were permitted until 1965, must have been quite a major influence on people. Even after 1965, other forms of tobacco continued to be advertised on television. If the media had an influence on people in terms of alerting them to the health risks of smoking, then surely it followed logically that the media used to promulgate the opposite message would also have an effect, it was as simple as that. Somebody like Mr McTear would be receiving a lot more pro-smoking messages than anti-smoking ones.
Cross-examination of Professor Gerard Hastings
[5.326] In cross-examination, Professor Hastings said that the only documents he had been given for the purposes of preparing his report were some advertisements, copies of which were attached to it. He had not seen any of the documentary productions relating to Mr McTear, the report of the commission to take Mr McTear's evidence, or the transcript of the evidence given by Mrs McTear. The statement at the conclusion of his report that Mr McTear must have been aware of the considerable publicity about the health consequences of smoking during the 1960s was not based on anything particular to Mr McTear. It was based on the obvious media coverage that would have been around at the time. He agreed that any member of the public must have been aware of the publicity that was being given to the health consequences of smoking.
[5.327] Asked to explain what social marketing was, he said that it was predicated on the idea that when commercial companies used marketing, they used a variety of techniques. They used advertising, combined with product development, distribution and price, a series of variables that they used to try to influence customers and specifically to influence consumer behaviour. Social marketing simply took that idea and said that if commercial companies could influence consumer behaviour, maybe social and health bodies could use it to influence social and health behaviour. If people's behaviour could be influenced in a positive way, then that would bring benefits. A review produced in the early 1990s in the United States came to the conclusion that over half the premature deaths there were caused by people's own behaviour. So any technology that could help to understand this better and to influence it had a potential to bring enormous benefits. Really that was all social marketing did. Part of the point was to persuade people to change their behaviour. Health campaigns of the sort run by the Scottish Health Education Unit which sought to persuade people not to smoke might be described as social marketing. The end point of social marketing was behaviour change.
[5.328] Media advocacy was a slightly different phenomenon in that it was purely focused on media activity. It was the strategic use of the mass media for advancing a social or public policy initiative. It involved using a variety of strategies to stimulate media coverage of an issue: it was focused on the media, rather than on other activities. A helpful way of thinking about media advocacy was that it was the social and public version of public relations. It was similar to what large corporations did in the commercial sector.
[5.329] Professor Hastings agreed that he had been widely quoted in the media on the subject of advertising bans. He did not regard himself as an active campaigner against the tobacco industry. He saw himself as an academic who had done research in this area and had reviewed the research of others and had come to certain conclusions. He thought that as an academic he had a duty to stand up and say what conclusions he had reached, but he did not see himself as a campaigner. People came to campaigning with a pre-set ideology, while he was saying that he had looked at the evidence and had drawn a conclusion. In his CV he stated:
"Academics exist to improve our understanding of the world and ensure that this understanding makes it a better place."
It was his view that the world would be a better place if people did not smoke tobacco. He regarded himself as an advocate for greater measures of tobacco control. He had said previously that the tobacco industry was to lung cancer what the mosquito was to malaria.
[5.330] Professor Hastings agreed that in addition to the appointments he had mentioned in his CV, he was a member of the Advisory Council of ASH, which was a campaigning body. He gave advice to ASH on his understanding of how marketing worked and the need to fully comprehend how multifaceted it was. He believed that ASH was funded through the Government. He did not know the precise origins of ASH. It was a body which engaged in advocacy and lobbying activities. He had contact with both the United Kingdom body and ASH Scotland. ASH Scotland, like the United Kingdom body, was a campaigning organisation which engaged in advertising and lobbying, it was exactly the same sort of organisation, just the Scottish version of it. He thought that it was simply the case that ASH UK, despite its name, concentrated primarily on England and Wales, and Northern Ireland, whereas ASH Scotland concentrated on Scotland. He was not absolutely sure how that worked, but in general terms ASH Scotland was a separate body and operated in that way on a day to day basis.
[5.331] In Hastings et al. 1998, of which he was one of the authors, it was stated, at p.55, under the heading "Combating tobacco marketing":
"In the UK, smoking is becoming increasingly associated with low-income and deprivation. Unskilled workers are two to three times more likely to smoke than professionals, and those who live in rented accommodation are much more likely to smoke than owner-occupiers [...]. Smoking is also strongly linked to a stressful lifestyle, with high smoking rates among the unemployed, those who are divorced or separated, and lone parents [...]."
Professor Hastings said that these were still valid statements.
[5.332] MacAskill et al. 2002, of which Professor Hastings was also one of the authors, was a report of an investigation carried out in eight Glasgow communities with the highest levels of social deprivation, over a period of three years. One of the aims of the study was to assess the level of motivation and readiness to quit of people living in these communities. Motivation was an important factor in quitting smoking. In the abstract it was stated:
"Understanding the consumer reveals that tobacco meets many needs in the struggle to cope with limited income intensified by poorly resourced local infrastructure and limited opportunities."
Professor Hastings agreed that this was one of their findings.
[5.333] At pp.21-25 the following statements were made:
"Tobacco plays an extremely important role in the lives of people in low-income communities. For many respondents in our study, the struggle to cope on a limited income, often in tandem with caring for children and other family members, was intensified by a poorly resourced local infrastructure, high levels of crime and drug use, and severely limited opportunities for recreation or respite escape from the immediate environment. Where recreation was available and affordable, it centered around drinking, drug use, and especially smoking. The latter in particular was easily accessible, relatively cheap, and, for many, their main pleasure.
Smoking offered both a respite from, and a means of coping with, this stressful and unrewarding environment. [...]
Many smokers noted how their consumption levels had risen steadily over their lifetime in response to increasing demands and worries. [...] Smoking also provided a means of coping with the frustration and demotivation of widespread unemployment. Not working had become a way of life for many, and the poor income offered by training schemes and low-paid work provided little incentive to try to move off welfare benefits. Lack of training and qualifications, combined with the stigma of living in a community whose residents were labeled as unemployable, had put many at such a disadvantage in the job market that they found it had to imagine working again. [...]
It has been suggested that people give up smoking 'for reasons connected to optimism' [...]: actual or anticipated improvements in life circumstances, health, or feelings about oneself. For many respondents in our study, there were few causes for optimism and little motivation to quit. [...] Men in particular envisaged little likelihood of a future improvement in either financial circumstances or employment, and felt little desire or confidence in their ability to change one of the few things - smoking - which helped them cope with this lack of prospects [...].
A [...] barrier to smoking cessation in low-income communities is the importance of smoking as a means of coping with stress. In our study, smoking helped respondents cope with daily anxieties and relieve the boredom of unemployment."
At p.31 the conclusion began with the statement:
"Smoking continues to be entrenched among those living in disadvantaged communities, as shown by the absence of a decline in prevalence and consumption relative to the general population. Clearly, tobacco meets a multitude of needs in these community members' lives - respite, a means of coping, and a key to social interaction and cohesion."
Professor Hastings agreed that these were the findings of the study, though he thought it important to recognise that within the paper there was discussion of the problems created by the tobacco industry's marketing activities, which presented a considerable barrier to people quitting.
[5.334] Professor Hastings was asked to consider various passages in the advertising brief for the ITL project to launch a new brand in the ultra low price sector. This brand, Richmond, was to compete with Gallahers' brand Mayfair. He agreed that the aim of the advertising was to attract smokers away from Mayfair to Richmond, by offering them a more valuable product that would give them more utility. He was shown a series of newspaper advertisements for cigarettes dating from 1964 and 1965. He agreed that he had previously written that forming associations between brands and evocative and symbolic images helped customers to differentiate between some of their products. This was part of the reason for branding.
[5.335] In his examination-in-chief, Professor Hastings had said that debate was an important part of how the media covered things. It was crucial in the news and media: if they had only one side of the story, it did not last for very long. He agreed that the media reported on events and other matters which were considered to be of interest to readers, listeners or viewers. In the 1950s and early 1960s it would be of interest to many members of the public that smoking was being linked with fatal diseases. It would be of particular interest to smokers and to people who had friends and family members who smoked, though some people would take it on board and respond to it and others might well deliberately avoid it. There was a lot of evidence that smokers really did not want to know about the health consequences of smoking, because it was so "dissonant". Awareness of a risk was a very simple way of expressing what was actually a very complex phenomenon. People often did things that involved an element of risk. People engaged in reasonable risks and indeed life without risk would be unbearable. People took risks for a variety of reasons, for example that the activity which involved the risk had some enjoyable effect. The problem emerged when people took risks that were really stacked against them. The problem with tobacco which distinguished it from all other examples was that the risks were so great. One in two long-term smokers would die as a result of that habit.
[5.336] Professor Hastings was asked about a long series of newspaper reports published in the period from 1950 to 1964. I have already referred to many of these reports in the section relating to public awareness, and I see no need to repeat the detail here. The subject matter of these reports included discussion of the conclusions reached in Doll and Hill 1950, MRC 1957, RCP 1962 and USSG 1964. Professor Hastings did not dispute that the general public would be aware, as a result of reading newspaper reports such as these, that an association had been established between cigarette smoking and lung cancer and that authoritative opinions were being expressed that the association was one of cause and effect. The newspaper reports also gave publicity to the contrary view, expressed by tobacco manufacturers and others, that a causal relationship had not been established and that further research was required. Professor Hastings said that until the early 1990s the media dealt with the issue as a debate to which there were two sides. In addition, the general public were exposed to tobacco advertising.
[5.337] A number of points were made repeatedly by Professor Hastings. He said that because the newspaper coverage presented the issue as one to which there were two sides, people reading the newspapers would not have the skills to make judgments about the risks associated with smoking, when weighed against all the other risks they encountered in their lives. It was very complex and difficult to make such decisions. In addition, it was not just a rational process. In addition to the difficulties of interpretation, individuals were exposed to all the other influences, such as advertisements for tobacco products. If a smoker felt threatened and upset by the possibility that smoking was going to cause serious illness, he would cling to any expression of doubt as a reassurance. Smokers were very uncomfortable when the health dangers of smoking were pointed out to them and they wanted to try and distance themselves from such a warning, either by ignoring it altogether or by rationalising it in some way. People did not want to take such information on board because it had very uncomfortable repercussions, particularly in relation to a habit that was so difficult to break. People did not want to know such information, so they would fend it off in all sorts of ways. There was a lot of research evidence to show that smokers were very threatened by messages of this sort and looked for get-out clauses. He agreed that readers of newspapers might be expected to have a certain degree of sophistication and to look at the source from which any particular statement came, as well as the content of the statement itself. He agreed that he had written that audiences, especially young ones, were extremely sophisticated consumers of the media. What they would also do, however, was to look for reassurance and an escape route, to help them feel a little bit easier about what was a very ambivalent consumption behaviour.
[5.338] Professor Hastings agreed that, particularly following publication of RCP 1962, the Government had taken steps to educate the public about the risks associated with smoking. He also agreed that from 1962 onwards the tobacco companies agreed not to advertise cigarettes on television before 9pm and that certain kinds of cigarette advertisements would be excluded from commercial television broadcasts, and that in 1965 cigarette advertising on television ceased. From 1968 onwards the Scottish Health Education Unit publicised the risks associated with smoking. From 1971 onwards cigarette packets bore Government health warnings, the first being "Smoking can damage your health." In 1961 an anti-smoking clinic was established in Paisley and there was publicity about it in the local press. The local Medical Officer of Health was in 1965 said to be engaged in "a fairly intensive campaign to instruct the public in the dangers of tobacco smoking". Professor Hastings agreed that doctors would to a certain extent tell their patients about the risks associated with smoking, and teachers would instruct their pupils likewise. While not disputing that any of these measures had been taken, Professor Hastings repeatedly made the point that members of the public would also be exposed to contradictory messages from the tobacco manufacturers whose promotional devices, such as sponsorship of Formula 1 motor racing, were intended to make smoking appear glamorous.
[5.339] Dr Keith Kerr, BSc (Hons), MB, ChB, FRC Path, aged 47, was a Consultant Histopathologist at Aberdeen Royal Infirmary, a post he had held since 1989. His specialist area of expertise was pulmonary pathology. He acted as Consulting Pathologist to the Medical Boarding Centre (Respiratory Diseases) for Scotland and Northern Ireland. He was a member of the UK Interstitial Lung Disease Pathology Panel and a member of the International Association for the Study of Lung Cancer. He served on an international expert panel of six pathologists, with two others from Europe and three from the USA, for the ongoing trials of screening detection of early lung cancer. He had published over fifty scientific research papers, book chapters and invited review articles, and many abstracts relating to research into, and the diagnosis of lung disease, particularly lung cancer and its development. He sat on the Editorial Consensus group for the forthcoming revised World Health Organization (WHO) Classification of Lung Tumours, of which he was a co-author.
[5.340] Dr Kerr explained that his work involved the examination of cells and tissues, and making diagnoses. He would see something of the order of 500 to 700 patients' cases per month. He was actively involved in teaching at his home university and by invitation in other centres throughout the United Kingdom, and also overseas. He taught that there is a strong association between smoking tobacco and lung cancer and that there were things in tobacco smoke which caused lung cancer.
[5.341] Dr Kerr provided a report dated 28 August 2003 on his examination of slides relating to Mr McTear. There were six stained slides of cytological material including smear and cytospin preparations, and three slides, each with three sectioned levels of a bronchial biopsy specimen, stained with Haematoxylin and Eosin. He reported that the cytology preparations were of variable quality and some showed fading of the staining. In most slides there were groups of large pleomorphic malignant-looking cells with irregular large nuclei and variable amounts of cytoplasm. The bronchial biopsy sections showed several tissue fragments showing tumour and elements of bronchial mucosa. In at least two of the larger fragments there was clear evidence that the bronchial mucosa was directly invaded by the tumour. The tumour consisted of large, relatively pleomorphic malignant epithelial cells with large irregular nuclei and moderately abundant eosinophilic cytoplasm. There was a tendency to cell stratification and there were intercellular bridges. The features were those of rather poorly differentiated squamous cell carcinoma. Some of the small fragments of bronchial mucosa were lined by pseudostratified respiratory type epithelium, which showed a transition to squamous-type epithelium, which was atypical. This atypia showed sufficient cytological abnormality, architectural disarray and abnormal mitotic activity to warrant grading as moderate squamous dysplasia. The presence of dysplasia indicated changes that had taken place in a squamous epithelium. The change from respiratory type to squamous type epithelium was one of the very early changes but, thereafter, in the development of carcinoma of the lung, atypia developed within the cells in the squamous epithelium in a step-wise manner, it was believed.
[5.342] Dr Kerr interpreted his findings by saying that there was unequivocal evidence of squamous cell carcinoma in the biopsy. There was also evidence of moderate squamous dysplasia. The latter was part of the recognised spectrum of pre-invasive changes, which occurred in the bronchial mucosa during the development of carcinoma in the central bronchi. He said that he had no doubt that Mr McTear suffered from primary bronchogenic squamous cell carcinoma (squamous cell carcinoma of the bronchus).
[5.343] Dr Kerr said that he had seen thousands of biopsies like this from other patients. Squamous cell carcinoma, he said, was one of two types of lung cancer which were most strongly associated with a history of smoking tobacco. When specimens were submitted to him for examination, he might or might not be told the smoking history of the patient. If he was given any history at all, it would more often be in the unusual situation where such a specimen was sent to him in a non-smoker, simply to highlight the fact that he was being asked whether the patient had lung cancer and ought to know that the patient was a non-smoker. This sort of information was highlighted, though not very often, because it was so unusual that there would be a suspicion of lung cancer in someone who did not smoke. Specifically, with squamous cell carcinoma, 98% of cases were seen in smokers. This was based on the literature, but also his own experience: it was most unusual to see a squamous cell carcinoma of the lung in a non-smoker. He could give a figure of 98% based on his own patients. Lung cancer, he went on, was rare in non-smokers, and when it did occur, the squamous cell type was very infrequent. The pre-invasive squamous dysplasia, the precursor lesion of invasive squamous cell carcinoma, was also frequently found in the bronchi of smokers and much less so in non-smokers.
[5.344] In support of these views, Dr Kerr referred to Colby et al. 1995. At p.93 the authors wrote:
"Cigarette smoking is the major cause of lung carcinoma in the United States and around the world [...]. The lung carcinoma rate parallels smoking prevalence. There is a dose-response association between the number of cigarettes smoked and the risk of carcinoma of the lung [...].
There are some variations in the histologic subtypes among smokers and non-smokers (Table 7-4), with squamous cell carcinoma and small cell carcinoma showing the highest association with smoking. It is possible that the recent decrease in the frequency of squamous cell carcinoma is in part attributable to changing smoking habits [...]."
Table 7-4 gave figures (modified from Rohwedder and Weatherbee 1974) for five histologic subtypes of carcinoma in smokers and non-smokers. For squamous cell carcinomas, 98% were stated to have been in smokers and 2% in non-smokers. Dr Kerr said that it was from this table that he derived his figure of 98%.
[5.345] Reference was also made by Dr Kerr to Auerbach et al. 1961. He said that Oscar Auerbach conducted a series of very extensive studies in the mid to late 1950s on autopsy material, taking a very large number of tissue sections and examining by microscopy almost the entire tracheobronchial tree and the extent of abnormalities in the airway epithelium, and associated these changes with smoking habits. At p.267 the authors wrote:
"It is known that cigarette smoke contains irritants, toxic substances and agents that are carcinogenic to the skin of laboratory animals. The application of irritants to epithelial tissue usually produces hyperplasia; toxic substances can destroy cells, and carcinogenic agents produce atypical changes in cells. In this study, we found hyperplasia, destruction of cilia (or destruction of ciliated cells) and the occurrence of atypical cells all to increase greatly with the amount of cigarette smoking. We cannot escape the conclusion that a direct causal relation exists between the degree of exposure to cigarette smoke and the frequency and degree of these changes in bronchial epithelium. [...]
In our opinion the histologic evidence from this study greatly strengthens the already overwhelming body of epidemiologic evidence that cigarette smoking is a major factor in the causation of bronchogenic carcinoma."
Dr Kerr said that he agreed with this.
[5.346] He had read a number of other papers by Auerbach and colleagues which essentially repeated the findings in this paper. In Auerbach et al. 1975 a study was made of histologic type of lung cancer in relation to smoking habit, year of diagnosis, age and sites of metastasis. Of 662 autopsies, 35.2% were epidermoid (squamous cell) carcinoma. The six non-smokers of the series were all found to have had adenocarcinoma. Dr Kerr made the point, therefore, that all of the squamous cell carcinomas occurred in patients who were smokers. Dr Kerr said that the same situation prevailed in Auerbach and Garfinkel 1991. At p.1976, Table 4 showed figures for histologic type of lung cancer by smoking habits. Squamous cell carcinoma was found in no patient who had never smoked, all of the patients with squamous cell carcinoma having smoked tobacco of some form. The figure of 98% in Colby et al. 1995 was accordingly consistent with the findings of Auerbach and colleagues.
[5.347] Dr Kerr was asked about Spain et al. 1970, a paper relating to metaplasia of the bronchial epithelium. Without referring to the text, he explained that the term metaplasia referred to the change in one type of tissue to another, usually under a stimulus, where the change in tissue type was a response to irritation or injury. It was an adaptive mechanism that the human body had everywhere. The development of squamous metaplasia in the lining of the airway was a relatively common change seen in response to chronic irritation in the airway. The notion would be that squamous epithelium developed instead of the normal columnar epithelium because it was better able to resist the noxious irritant, whatever it happened to be. Squamous metaplasia developed in squamous-type epithelium. Metaplasia was a stage before dysplasia, a relatively early change in the step-wise progression towards the development of cancer. Dysplasia was the development of cancer-like changes in the cells within the metaplastic epithelium. When these were seen in the cells of a squamous metaplastic epithelium, they were referred to as squamous dysplasia, or squamous atypia, a synonymous term. If dysplasia was diagnosed, this was a pre-cancerous stage, but a recognised stage on the way to the development of cancer. Accordingly, metaplasia occurred when there was irritation in the lung. It might or might not move on to the next stage, which was dysplasia, and this was a warning sign that cancer may develop.
[5.348] Finally, Dr Kerr referred to Nasiell et al. 1987. At p.207 the authors wrote:
"There is increasing evidence that bronchogenic carcinoma, particularly squamous (epidermoid) cell cancer, is preceded by epithelial abnormalities [...] and develops through successive stages characterized by progressively more abnormal morphological and cytochemical features induced by the long lasting action of irritating agents and carcinogens on the respiratory mucosa [...]. It is of great importance to find out which alterations or at which stage the precancerous lesions are reversible, which ones are irreversible, and which ones ultimately lead to invasive cancer.
The bronchial epithelium was studied by cytological, histological, and cytochemical techniques in human and experimental dog material in order to obtain information about the problems mentioned above. [...] It was found that atypias in metaplastically altered epithelium occur which show great similarity to dysplastic/premalignant alterations of the human cervical epithelium and that these changes in the bronchi possessed similar cytochemical properties as the more or less obligatory precancerous lesions which were studied earlier in the uterine cervix [...]."
At p.217 the authors wrote:
"Consequently, it seems that a severe chronic injury of the bronchial mucosa plays a prominent part particularly in the development of squamous-undifferentiated carcinoma and probably also in adenocarcinoma, resulting in a disturbance of the morphology of bronchial epithelium in the subjects developing lung cancer."
Dr Kerr said that he referred to these passages, and to the data in Table 8.1 on p.217, but he was not asked to give further evidence about the information contained in the table.
Cross-examination of Dr Keith Kerr
[5.349] The cross-examination of Dr Kerr started with an invitation to him to demonstrate, with the aid of a model, the anatomy of the organs within the human thorax. He explained that the respiratory tract passed through the larynx and the trachea to the point where it divided at the carina into two bronchi, each of which entered the lung at the hilum. Squamous cell bronchial carcinoma often arose centrally, towards the carina. Within the lung the bronchi divided many times, eventually becoming bronchioles. The space in the thorax between the pleural sacs containing the lungs was the mediastinum, where the heart and major blood vessels, and other structures including the oesophagus, lay.
[5.350] Dr Kerr agreed that the word "cancer", in colloquial use, referred to a variety of different types of tumour. In its general sense, the tumour was any lump or swelling, but in a more technical sense it might be used to describe a neoplasm. This was an abnormal growth of tissue that had become independent of any outside stimulus. It might be benign, such as a wart, which was a benign epithelial neoplasm, or it might be malignant. A malignant neoplasm was characterised by, among other things, invasion of the surrounding tissue. A squamous cell bronchial carcinoma might invade through the bronchial wall and into the mediastinum. It could involve the pericardium, the membrane surrounding the heart, and the great vessels which passed through the mediastinum. Simply by virtue of its growth, it could impinge upon these great vessels or on the oesophagus, in a manner which could lead to their destruction. Any of these events could result in the death of the patient. Another feature of malignant neoplasm was that it might metastasise to sites removed from the primary seat of the tumour. Dr Kerr had written on the significance of lung cancer to the mediastinal lymph nodes. Primary tumours in the mediastinum were relatively rare.
[5.351] Carcinoma, he said, was a malignant neoplasm arising in the epithelium, the tissue covering the external and internal surfaces of the body. Like the rest of the body, it was made up of cells. Epithelial cells might be squamous, tuboidal or columnar. Squamous epithelium was composed in part of flattened, plate-like cells. It lined the upper respiratory tract, such as the inside of the mouth. In the lower respiratory tract, in the trachea and the bronchi, there were columnar cells which included ciliated cells. Carcinoma could arise in any part of the respiratory tract, but carcinoma of the trachea was very rare and carcinoma of the larynx was less common than bronchial carcinoma. There were many different types of carcinoma of the lung. Part of Dr Kerr's expertise was in seeking to distinguish between these types. So far as management was concerned, the important distinction was between small cell carcinoma and non-small cell carcinoma. The latter included squamous cell carcinoma, adenocarcinoma and a variety of other types of carcinoma. The WHO had classified lung tumours by histological type in a volume first published in 1967, with subsequent editions in 1981 and 1999. Dr Kerr was currently associated with this work. The classification was intended to promote the adoption of a uniform terminology to facilitate communication among cancer workers.
[5.352] Lung cancers frequently showed histological heterogeneity, by which was meant that there might be variations in appearance and differentiation from one microscopic field to another and from one histological section to another. The current edition of the WHO classification stated that almost 50% of lung carcinomas exhibited more than one of the major histological types. This fact had important implications for lung tumour classification and had to be kept in mind, especially when interpreting small biopsies. When tumours were typed on the basis of small biopsies, there was some scope for inaccuracy. Dr Kerr had been involved in the preparation of national clinical guidelines on the management of lung cancer in Scotland. According to these guidelines, there was considerable inter-observer variability in subtyping non-small cell carcinoma, especially on cytological and brochoscopically derived material. Only 10% - 15% of patients with lung cancer would ultimately have the tumour removed and the pre-operative tumour classification confirmed. Dr Kerr had participated in work which sought to promote the use of the non-small cell lung cancer category by means of using a more general type rather than a more specific type to achieve a greater degree of accuracy. He was co-author of an article published in 2000, in which it was stated that it was of concern that if 85% - 90% of patients had a diagnosis made on pre-operative specimens alone, with the likelihood of some degree of inaccuracy in tumour classification from such specimens, then epidemiological studies might be based on flawed data. He said that he wrote this, and had also written that non-small cell lung cancer was a useful and more appropriate classification to use, reflecting the view that it was sometimes better to be more accurate by being less precise in those situations where it was not possible to be more precise. He wrote that epidemiological data would be more useful if based upon more robust histological diagnosis. In another work he had written that, despite minimal changes in the WHO classification, some researchers had reported decreases in the number of squamous cell carcinomas as well as increases in the number of adenocarcinomas. He had also written that there was anecdotal evidence that in the past, and perhaps still, it was likely that where histologic typing was difficult, some pathologists might have opted for the commonest histological type, which was squamous cell carcinoma.
[5.353] In reporting on the slides relating to Mr McTear, Dr Kerr relied predominantly on the biopsy specimens because there was more information to be gained from them than from the cytology specimens. In a cytology specimen all one had were individual cells, one could not see where they were located in the bronchus or their orientation with respect to one another. He himself had written that the importance of cytological atypical cells could be difficult to determine. The point of a histopathological examination was to report on what was found in that process, which would be a large component of the job in hand. In most circumstances, it was possible to make an adequate and accurate diagnosis without any patient's history. In some situations the history was required, but he would not necessarily be given one. A biopsy specimen was in effect a small piece of the bronchus and a histological examination of this sort was possible only when a bronchial biopsy had been performed. Also, in situations where a whole tumour had been surgically removed, it was available to be examined. In such cases pre-operative diagnosis might be confirmed by looking at the whole tumour. A biopsy would not be carried out on a patient unless there was a good medical reason for it, that is to say a suspicion that the patient had a lung disease which required an invasive investigation.
[5.354] In reporting on the slides, Dr Kerr noted two phenomena, a squamous cell carcinoma and a moderate squamous dysplasia. He could not say for sure that the squamous cell carcinoma in this biopsy arose from the area of dysplasia. It was well-recognised that, when carcinoma of the lung, particularly squamous cell carcinoma, occurred, dysplasia was frequently observed in the airways surrounding the carcinoma. It was found that there was an association between the presence of carcinoma and the presence of dysplasia. He knew of no evidence to support the assertion that the dysplasia might occur in response to the carcinoma. He agreed that they might have arisen entirely independently of one another.
[5.355] Dr Kerr said he was sure that we did not know everything about the genesis and development of squamous cell lung cancer. He believed that we knew a great deal about how the tumour came about, but he would not claim that we knew everything. "How do we know what we do not know?" He himself had previously said that the origins of this common disease were not well understood and that little was known about the rate and risks of progression of squamous dysplasia to CIS (carcinoma in situ) and ultimately invasive disease. He described squamous dysplasia as a precursor lesion of invasive squamous cell carcinoma. "Lesion" was another generic term used to describe an area of abnormality. Squamous metaplasia would be described as a lesion. It was not a normal feature of the bronchus: it was an abnormality of the bronchus when squamous metaplasia occurred. It was a common occurrence during bronchial damage and chronic irritation and repair.
[5.356] Dysplasia was only one of a number of changes which might be found on the histopathological examination of a bronchial biopsy. Other changes would include hyperplasia, a thickening of the cell layer due to an increase in the number of cells, and metaplasia, a change in the epithelium, for example a finding of squamous epithelium where ciliated epithelium would be expected. Dysplasia involved a variable degree of atypia. International classification divided it into three categories: mild, moderate and severe. These changes did not show up on x-ray and most of them were not visible on a conventional bronchoscopy, so they would not normally be detected without a histopathological examination of a biopsy specimen or a resected lung. It was not possible to tell from the histopathology how any of these changes came to be there, or what would have happened to it if it had not been removed; but there was evidence to suggest that the greater the degree of atypia or dysplasia, the greater was the likelihood that the disease would progress to invasion and therefore the development of carcinoma. This evidence came from biopsies. When he used the expressions "pre-invasive" and "precursor", he was implying that there was a risk that the lesion would progress if there was no intervention, although there was no guarantee that this would be so. It was possible that all such changes were capable of regression, although it was believed that the more atypical or abnormal they became, the less likely regression was to occur. It was not known which, if any, of these lesions were obligatory for the development of squamous cell carcinoma or any other type of lesion.
[5.357] Asked whether it was the case that we did not know what proportion of the population, who exhibited no clinical symptoms and who never went on to develop lung cancer or any other condition requiring biopsy, had lesions in their lungs, Dr Kerr said that there was arguably some evidence in the papers of Auerbach and colleagues. In the autopsy studies, some patients had lung cancer, some did not, some patients were smokers and some were never-smokers, so in a sense these were studies of samples of the population at large and gave some indication of how frequent or otherwise these changes might be in the airways. A variety of the changes were found in patients who did not have lung cancer, but the changes most likely to be associated with the development of lung cancer were very rarely found in patients who were never-smokers as opposed to those who were smokers. He agreed that these were autopsy studies carried out on people who died in hospital, but said that they would be more or less typical of the population, depending on the parameter that was being examined. Auerbach worked at a veterans hospital, but Dr Kerr thought that the general population were treated at such a hospital, it was not just for ex-soldiers.
[5.358] Dr Kerr had not been familiar with Spain et al. 1970 before coming to court to give evidence. According to the abstract of that paper, in 500 apparently "healthy" adults who died suddenly and unexpectedly and who lived and worked within the environs of a large urban area noted for its air pollution, autopsy studies revealed that 50% of all adult non-smokers had metaplasia of their tracheobronchial epithelium. Metaplasia was present in 70% of male cigarette smokers. Women in each category revealed half the frequency of metaplasia present in the men. Metaplasia was present in less than 20% of men under the age of 30 years, rose to a peak of 80% at age 50 years, and gradually declined in frequency after the age of 60 years.
[5.359] The text of the paper started with the statement:
"Regenerative hyperplasia or metaplasia of the tracheobronchial epithelial lining may be caused by a variety of physical, chemical, or infectious materials."
Dr Kerr said that this was correct. He also agreed with the statement, also on p.1331, that:
"It is obviously impossible to collect adequate samples of tracheobronchial mucosa on a random selection of a healthy live population, so one must attempt to develop this information from autopsies."
The authors stated that previously, autopsy studies designed to determine the distribution of tracheobronchial metaplasia, as for example in the reports of Auerbach and colleagues, were on sick individuals who died in hospitals.
[5.360] The authors carried out a simplified study on the morphological changes in the tracheobronchial mucosa of an autopsied, apparently "healthy" population made up of individuals who died suddenly or unexpectedly from accident, suicide, homicide, or within minutes of a first acute attack of coronary heart disease. Only full-fledged metaplasia was recorded. At p.1334 they commented:
"The association of cigarette smoking with a high frequency of tracheobronchial mucosal metaplasia has been demonstrated several times in the past 15 years. Our findings now confirm this association in a different and less highly selected population. More important, however, is our observation that varying degrees of tracheobronchial metaplasia are present in a surprisingly high proportion (50%) of 'healthy' adult men who have never smoked cigarettes, cigars, or pipes. In this group the frequency is as high as 60% in those who are more than 50 years old. [...] The quality of the metaplasia changes was indistinguishable from that found in the cigarette smokers. In the nonsmokers though, it seemed to develop at a somewhat later age than it does in the cigarette smokers.
This relatively common occurrence of metaplasia in nonsmoking, apparently healthy, men who are otherwise free of pulmonary alterations cannot be explained as simply being a manifestation of the aging process, because even for those below the age of 31 years the frequency was 20%. Furthermore, there was a considerable decline in the frequency after the age of 65 years. It also can not be explained on the assumption that in a certain percentage of normal individuals there are scattered foci of metaplastic epithelium in the tracheobronchial mucosa. The curve of distribution at different age levels [...] negates this. The age distribution curve is entirely consistent with increasing time of exposure to noxious materials that might be inhaled from the external environment. [...] Because this particular population group was derived mainly from the environs of a large metropolitan area noted for its air pollution, it is tempting to incriminate the polluted air as a likely cause. However, this possibility must be regarded as speculative until such time as additional data can be developed from similar studies in other geographic areas - some of which should be free of significant air pollution."
[5.361] Dr Kerr said that the expression "full-fledged" metaplasia was not a technical term, but he inferred that it described a situation where the epithelium was easily identifiable as a squamous epithelium from its lower to its upper part. In some cases where the squamous metaplasia was in evolution, the change was not complete within the epithelium from the base to the top, and degrees of squamous change may occur in the hyperplastic zone of cells in the basal area of the respiratory epithelium. It was believed, he said, that in many cases of squamous metaplasia there was a hyperplastic change which occurred in the tracheobronchial epithelium and the squamous change took place within the hyperplastic zone. He would assume that metaplasia would be described as not full-fledged by these authors in those areas where there was squamous change within a hyperplastic zone but not a fully developed squamous epithelium from top to bottom. No one had actually observed changes going on as they occurred in the human lung, that was not possible, so experimental models had to be used; the change, apart from anything else, was believed to take a long time to develop.
[5.362] Dr Kerr said that it was possible to find dysplasia in the lungs of non-smokers. Auerbach et al. 1975 and Nasiell et al. 1987 were both studies which set out to investigate whether or not there was an association between lesions in the lung and smoking. Dr Kerr could not bring to mind any work which addressed whether or not there was an association between lesions in the lung and a family history of lung cancer. He was not aware of any similar work addressing whether or not there was an association between lesions in the lung and a history of alcohol use. Asked whether he was aware of any work which addressed whether or not there was an association between lesions in the lung and residence in an urban rather than a rural area, he said that it made a difference to his answer whether "lesions in the lung" referred to metaplasia or dysplasia. In Spain et al. 1970 the possibility of an association between urban living and metaplasia was raised speculatively, but there was no mention of dysplasia, and he knew of no work relating dysplasia to urban living. He was not aware of any work which addressed whether or not there was an association between lesions in the lung and a history of stress and depression. He agreed that the development and maintenance of the normal epithelial lining of the respiratory tract depends on an adequate supply of vitamin A, and that vitamin A deficiency might lead to abnormal cellular differentiation and proliferation. He was aware that there had been some pieces of work, with which he was not familiar, in which an association was made between the development of malignancy and vitamin A deficiency. He had no knowledge whether alcohol abuse might deplete vitamin A levels, it was not his area of expertise.
[5.363] Dr Kerr said that he was not familiar with Winternitz et al. 1920, in which there was a report of autopsies carried out on people who had died in the influenza epidemic of 1918. The authors reported, at pp.15-16:
"Sooner or later, with the subsidence of the irritating agent, repair begins in the bronchus or bronchiole. If its walls have been destroyed and the lesion has extended into the surrounding alveoli to form an abscess of greater or lesser extent, or if the necrotizing process has been superficial and confined to the epithelium in large part, the reparative process is very much the same. Mitotic figures in the fibroblasts and in the endothelial cells of the capillaries abound in the young granulations [...]. However, the granulation tissue does not have an unrestricted path of growth, for if a remnant of epithelium remains, this is stimulated to grow probably in this disease as in no other. Mitotic figures are common and the young epithelial cells stretch across the denuded submucosa or granulation [...] and extend downward into the surrounding alveoli, not only as strands, but also as solid nests of cells [...]. The bronchioles, therefore, show changes dependent upon the extent of the damage suffered by their walls. The vast majority, in all probability, will be restored; but if the wall has actually been necrotized, the bronchioles may be converted into small, saccular, bronchiectiatic cavities [...], or obliterating bronchiolitis may result from the organization of the exudate within their lumina [...]. The importance of the epithelial proliferation cannot be ignored; in many cases, it invades the surrounding lung tissue and a typical, histological picture results - an infiltrating, malignant, epithelial neoplasm [...]."
Dr Kerr said that it was correct that certain changes in the lung occurred as part of the normal process of repair, but he would not agree with the interpretation placed by these authors on their findings, indeed he probably doubted whether the pathologists found what they said they found in their autopsy studies. He said that it was possible in cases of respiratory infection, where the airway was repairing itself, that occasional atypical cells might be seen, but he thought it most unlikely that a squamous epithelium would be seen to be organised in such a way that it would be regarded and described as being a dysplastic squamous epithelium as the term was now understood and used.
[5.364] Dr Kerr was asked about Table 8.1 on p.217 of Nasiell et al. 1987, relating to the incidence of cytological changes. Among other figures, the table stated that 23% of non-smokers with chronic inflammatory lung diseases displayed atypical metaplasia. Dr Kerr agreed that atypical metaplasia would be dysplasia, as he had defined it. He said, however, that the direct comparison was difficult because the table referred to cytological changes, where cells were being described which were separated and not cohesive and related one to the other in the form of an epithelium. The description of squamous dysplasia which he made earlier was based upon the examination of the intact epithelium, where it was possible, as well as to appreciate the atypical nature of the individual cells, to appreciate the architecture and structure of the atypical epithelium and the way each cell related to the other, and also the localisation of the mitotic figures, which was also a crucial part in defining squamous dysplasia. The table accordingly required to be read subject to the qualification that it related to cytological changes. Where there were cases of atypical metaplasia in non-smokers with chronic inflammatory lung disease, that might not denote dysplasia. He could not tell, because the cells were dissociated one from the other, and the WHO definition of dysplasia, which he was applying and using, required the making of a histological examination of the tissue, not the disassociated cells.
[5.365] Dr Kerr was next asked about his evidence that, with squamous cell carcinoma, 98% of cases were seen in smokers. The reference for that was Table 7-4 in Colby et al. 1995, relating to histologic subtypes of carcinoma in smokers and non-smokers. The information in the Table was said in a footnote to have been "modified from reference 9", which was Rohwedder and Weatherbee 1974. It was pointed out to Dr Kerr, however, that this paper did not contain the information in question, and the reference should have been to reference 10, Rosenow and Carr 1979. The authors both held positions at the Mayo Clinic, Rochester, Minnesota. At p.233 they stated:
"The Mayo Lung Project of the Mayo Foundation [in Rochester, Minnesota] is conducting a controlled, prospective lung cancer screening program in males 45 years and older who smoke at least 20 cigarettes per day. The study group has a chest x-ray and pooled three-day sputum analysis every four months, and the controls are urged to have the same tests yearly."
At p.234 it was stated that Table 2 listed the prevalence of the five most common lung cancers in men and women, smokers and never-smokers "as they have presented at the Mayo Clinic". One of these cancers was epidermoid carcinoma, also called squamous cell epithelioma or squamous cell carcinoma in the text. Figures given in respect of epidermoid carcinoma in Table 2 on p.235 showed that out of a total of 992 cases, 892 were in male smokes, 7 in male never-smokers, 80 in female smokers and 13 in female never-smokers. The figures of 98% and 2% given in Colby et al. 1995 could be seen to be derived from these statistics. Dr Kerr agreed that the only passage in the text which suggested how the diagnosis was made was by exfoliative sputum cytology, at least in the males in the study group.
[5.366] Dr Kerr was asked repeatedly about the conclusions that could be drawn from the above figures without knowledge of the proportion of smokers and never-smokers in the general population. His reply, under reference to the table, was: "In patients who have squamous carcinoma of the lung, 98% of them are cigarette smokers, 2% of them are never-smokers. That is it." He said he did not agree with the assertion that in order to assess whether or not these figures had any significance, one would need to know the relative distribution of smokers and never-smokers in the population at large. Asked, however, whether he agreed that whether or not these figures were of significance depended on whether or not the patients who formed the data set were representative of the general population at large, he said: "Yes. If we are to draw an inference from these figures and apply it to the general population, then that would be the case, yes." He agreed that in order to assess this, information would be needed about the subjects of the study. He pointed out that it was not possible to tell from the text whether the smokers in the screening programme were included in the statistics. Asked whether, if these people were included within the sample, that would bias the sample towards the smokers, he said that this was not a case of bias, it was a case of the study population having many more lung-cancer-bearing patients who were smokers than it did those who were never-smokers. He said that the only response he could give was that in any study that he would be aware of, or any group of lung cancer patients that he could imagine from his experience, where histories were taken from the patients and their smoking habits were considered, a larger proportion of patients in a group who had lung cancer would be found to be smokers than would be the proportion of smokers in the general population. He thought the paper showed that squamous carcinoma in the lung was much more likely to be found in someone who had been a smoker rather than in someone who had been a non-smoker.
"That is the piece of information that I think comes from this paper, and, in my interpretation of this, the rates of smoking in the general population have nothing to do with this paper. It is quite irrelevant to the interpretation of these data."
[5.367] Dr Kerr agreed that Auerbach et al. 1975 reported on 662 autopsies of men, of whom six were non-smokers, and Auerbach and Garfinkel 1991 reported on 505 cases of lung carcinoma, the number of patients who had never smoked being ten.
[5.368] Dr Kerr was asked about his evidence that "there are things in tobacco smoke which cause lung cancer". He said he was not aware, and very much doubted, that there had been any direct studies on living human beings of any of the recognised carcinogenic agents which were in tobacco smoke. Asked then what studies formed the basis of this proposition, he said that he could not quote direct studies, but
"I am reflecting an opinion and a belief that I have over many years of reading the medical literature, of understanding and reading reports that there are a whole variety of different chemical agents within tobacco smoke, some of which are recognised in test situations using either laboratory animals or human tissue cells in culture, where those chemicals are recognised as carcinogens because they can be shown experimentally to induce tumours in laboratory animals, and also to induce changes in human cells, in tissue culture, for example; changes which we recognise as being associated with the development of the malignant process."
Asked whether it was the case that studies which sought to produce squamous cell bronchogenic carcinoma in animals by the inhalation of tobacco smoke had failed to produce such tumours, he said that he did not know the literature on the animal studies well enough to comment.
[5.369] Dr Kerr had described a theory of carcinogenesis which involved the irritation of the epithelium by some external agent. The generally accepted theory of carcinogenesis involved multiple stages and multiple changes within in a cell population before malignant behaviour eventually supervened. He agreed that ultimately carcinogenesis was a genetic process, and that in order to understand it, one would need to understand what happened at the genetic level. He agreed that bladder and cervical cancer were associated with cigarette smoking, and did not see why the irritation theory could not apply to these cancers. He said that the association with bladder and cervical cancer was less strong than with carcinoma of the lung, for example, but it was a matter of speculation that chemicals that were inhaled in tobacco smoke entered the blood stream and could therefore be delivered to those sites. He agreed that this was a very active field of scientific investigation.
Re-examination of Dr Keith Kerr
[5.370] In re-examination, Dr Kerr said that he was not wholly familiar with the details of the experiments, but he believed that studies had been done to test chemicals to see if they could induce changes in the DNA of human cells in tissue culture in the laboratory. Certain chemicals could be recognised as carcinogenic, or as mutagenic.
[5.371] Asked whether he knew what percentage of the population of the UK were smokers, Dr Kerr said that he thought the last figures he saw suggested that across the board approximately 30% of the population smoked, but he could not give a figure for the percentage of population who either were or had been smokers, his figure was for active smokers.
[5.372] In his personal experience Dr Kerr had dealt with biopsies of people with respiratory infection, and he had never found dysplasia, in the sense in which he used that term in his report, an abnormal squamous-type epithelium lining the bronchus, fulfilling the criteria laid down by the WHO. It was believed that precursor lesions carried a risk of progression, a step on the way to carcinoma. The data were very soft and difficult to interpret, but there was evidence to suggest that, as the degree of dysplasia in the cell population increased, the statistical likelihood of progression to malignancy increased over a time period of several years.
PART V: THE EXPERT EVIDENCE
[5.373] Professor Gray, aged 69, was Emeritus Professor in the Department of Psychology at the Institute of Psychiatry, University of London. He described this as a tertiary education and research establishment, generally regarded as one of the best, and perhaps the best, in the world, in the field of psychiatry.
[5.374] He qualified initially in modern languages, being awarded a BA degree at Oxford University in 1957. He obtained a second BA degree at Oxford in 1959 in psychology and philosophy, followed by a qualification in clinical psychology in 1960 and a PhD in experimental psychology in 1964 at the Institute of Psychiatry in London. He returned to Oxford in 1964 as a lecturer in psychology and in 1965 became a Fellow of University College. In 1983 he succeeded Hans Eysenck as Professor and Head of the Department of Psychology at the Institute of Psychiatry. He remained there until retiring at the age of 65 in 1999, and was thereafter Emeritus Professor.
[5.375] He said that Hans Eysenck founded the Department of Psychology at the Institute of Psychiatry in the 1950s. He was generally recognised as the most important scientist to date in the field of the experimental psychology of personality, that is, differences between different individuals in their behaviour and psychology. During his lifetime he was the most highly cited living psychologist. His most important contribution to psychology was his work in personality, in which he developed what today would be the scientific model of that field. Next in importance was his work in the field of psychological treatment, particularly the development of behaviour therapy and, subsequently cognitive behavioural therapy. He did not himself do much research in that field, but provided, in the Department of Psychology at the Institute of Psychiatry, the environment for others to work in that field, and he also did a great deal to "spread the word" that this was the way to do psychology. This was now the standard approach through the developed world. It was now accepted that, with many psychiatric disorders, the beneficial effects of these forms of psychological treatment were at least as good as the effects of drug treatment. Eysenck died in 1997.
[5.376] In addition to the foregoing posts, Professor Gray had been Visiting Professor at the Université Paris VI from 1979 to 1980, the Virginia Military Institute in Virginia in 1983 and the Collège de France, Paris in 1999. He was a Fellow at the Center for Advanced Study in the Behavioral Sciences at Stanford University from 2001 to 2002. He had served as President of the Experimental Psychology Society from 1996 to 1998, and as a member of the United Kingdom Government's Gambling Review Body in 2000-2001. He became a Fellow of the British Psychological Society in 1993. He received the Presidents' Award of that Society in 1983: he described this as a "mid-career award" given to scientists in the field of psychology for the distinction of the work they had done up until that time. He became an Honorary Member of the Experimental Psychology Society in 1999 and an Honorary Fellow of Goldsmiths College, London in 2002. He was awarded an Honorary Doctorate at Washington and Lee University, Virginia in 2000.
[5.377] He said that a most general description of his work was that it was concerned with the relationship between brain activity and behaviour. Much of it had been with animals, particularly with rats, but also mice and New World monkeys. He had been particularly interested in aspects of brain function and behaviour that related to psychiatric problems. In the first part of his career he mainly concentrated on problems to do with anxiety and its treatment. Subsequently he worked in the field of schizophrenia, again looking at "animal models", that is, behaviour in animals which bore a conceptual resemblance to the equivalent behaviour in human beings. Thereafter, at the Institute of Psychiatry, he began to work more with human subjects and patients suffering from the conditions he was interested in, such as schizophrenia. More recently, with his colleagues at the Institute, he developed methods of implanting neural stem cells into the damaged brain, in order to repair it. In parallel with that, he began to develop, with students and colleagues, methods of treating anxiety and depression by computerised versions of cognitive behavioural therapy.
[5.378] As part of Professor Gray's research he had investigated dependence on both opiates and anti-anxiety drugs, and psychological treatments for anxiety, depression and opiate dependence. He said that the effects of drugs had figured repeatedly and prominently in his areas of research. As part of his programme, he had published extensively on the effects of nicotine and smoking on both human and animal behaviour, and on nicotine's neurochemical effects on rats and on cultured nerve cells in vitro. Reference was made to a list of his publications, of which there were about 460, including seven books, and to a separate list of those of his publications which related to nicotine and smoking, of which there were about forty. He said that his most important research contributions had been concerned with the biological basis of human personality; the behavioural and brain systems that mediated anxiety and schizophrenia; and the development of neural stem cells as a means to repair the damaged brain. In preparation for giving evidence as an expert witness, he had read or re-read "between 500 and 1,000 scientific papers, nearer 1,000".
[5.379] Professor Gray said that there was a distinction between the effects of nicotine and the effects of smoking. Some of the effects produced by smoking tobacco could also be produced by the administration of nicotine, but some could not be mimicked in experiments looking at the effects of nicotine. He described smoking behaviour as "complex". It was complex both in terms of the multiplicity of chemical compounds in tobacco, of which 4,000 or so has been identified in tobacco and tobacco smoke. It was complex also in respect of the behaviour of smokers. Smokers in general reported quite a wide range of different effects of smoking and they could smoke in different circumstances and in different ways. The effects they obtained from smoking depended upon the circumstances and the ways in which they smoked. So there was no single effect of smoking, there was no single route by which smoking produced its effects and there was no single reason why an individual smoker smoked. The chemicals in tobacco and tobacco smoke contributed in a variety of ways to the satisfaction derived from the smoking habit, e.g. by influencing the taste, smell and other sensory characteristics of a cigarette. Isolating the role played by any one compound was exceptionally difficult and it was unlikely in any case that any single compound played a role in isolation from those of others. The influence that these constituents of tobacco smoke had on an individual smoker was itself modified by the type of cigarette he smoked and the way in which he smoked it. A smoker might vary the way he smoked by varying the frequency and the duration of puffing, and in doing so he altered the exact composition of the constituents of the smoke he inhaled. The environment in which the cigarette smoker was smoking and his psychological state were factors that in turn influenced the way he smoked. For example, the smoker smoked in a different manner under conditions of stress than when he was relaxed. So there was a chain in which the environmental circumstances and the psychological state altered the way in which a cigarette was smoked and that in turn altered the effects of the cigarette upon the future psychological state and behaviour of the smoker.
[5.380] The smoker would adjust his smoking behaviour according to the nicotine and also the non-nicotine (e.g. "tar") contents of a cigarette. Adjustments would be made depending on what that particular smoker was used to smoking: he would attempt typically to adjust to the level of nicotine and the level of tar that he was most familiar with. There was evidence from laboratory studies that the "tar" yield of a conventional cigarette determined the amount that was smoked independently of the effects of its nicotine yield. Thus, when nicotine yield was controlled for, smokers of lower-tar cigarettes puffed more smoke from their cigarettes than smokers of higher-tar cigarettes, i.e. they compensated for the reduced amount of tar by puffing and inhaling a greater volume of smoke. Conversely, cigarettes from which the nicotine had been removed reduced self-reported "craving" and increased ratings of smoking satisfaction and psychological reward. In the same experiment, moreover, an intravenous injection of nicotine failed to increase either satisfaction or reward.
[5.381] Reference was made to Rose et al. 2000. Professor Gray was asked to consider the text of this paper in some detail, but for present purposes it is sufficient to quote the abstract, which was in these terms:
"To dissociate the sensorimotor aspects of cigarette smoking from the pharmacologic effects of nicotine, smokers rated the subjective effects of nicotine-containing or denicotinized cigarettes, and intravenous (IV) nicotine or saline infusions. Three groups of participants (n = 20 per group) received either: (1) continuous nicotine, (2) pulsed nicotine, or (3) saline. Each group was exposed to an IV condition once while smoking a denicotinized cigarette and once while not smoking, in a 3 x 2 mixed design. A fourth group (n = 20) received saline while smoking their usual brand of cigarette. The dose and rate of nicotine administration were individualized based on previous measures of ad lib smoke intake. Denicotinized cigarette smoking significantly reduced craving and was rated significantly more satisfying and rewarding than the no-smoking conditions. IV nicotine reduced craving for cigarettes, and increased ratings of light-headedness and dizziness. However, no significant satisfaction or reward was reported after IV nicotine. The combination of IV nicotine and denicotinized cigarette smoke produced effects similar to those of smoking the usual brand of cigarette. The results suggest that sensorimotor factors are critical in mediating the immediate subjective response to smoking, and that the immediate subjective effects of nicotine administered in doses obtained from cigarette smoking are subtle. Thus, addressing smokers' needs both for the sensorimotor aspects of smoking as well as for the direct CNS [central nervous system] effects of nicotine may be critical in enhancing smoking cessation treatment outcome."
Professor Gray said that in his view this was the best controlled study to date investigating the issue of the extent to which the satisfaction and reward aspects of cigarette smoking were mediated by way of either nicotine or the rest of the events involved in smoking. The results showed clearly that, as far as the pleasure and satisfaction of smoking were concerned, the principal route did not involve nicotine, but lay somewhere else in the complex behaviour that was smoking a cigarette. The variable called "craving" was influenced by nicotine to some extent, but it was also influenced by smoking a denicotinized cigarette. So there was a joint effect of these two pathways for that variable; but for most of the positive aspects of smoking cigarettes - satisfaction and psychological reward - the critical pathway did not involve nicotine.
[5.382] Professor Gray went on to say that to add to the complexity, different smokers sought different outcomes from smoking. Individuals gave many different answers to the question why they smoked, but it was possible to organise the answers by methods of statistical analysis of the correlations between them and to reduce the overall number of answers to a small range of underlying factors. The most important of these factors were concerned with relaxation and sensory enjoyment. A second cluster of reasons was mood control: the reduction of depression, irritability and so on. Thirdly there was a set of cognitive reasons: increasing alertness, being able to perform a boring job for longer, that sort of thing. Most smokers could then be classified according to the predominance of one of those clusters of reasons in their behaviour, but most smokers would also show some degree of recognising the other reasons as well. So far example there might be a smoker who predominantly smoked under most circumstances for the pleasurable aspects of smoking, but who recognised that on some occasions he smoked in order to control his mood. Sometimes smoking was used as a means of oiling the wheels of social interaction. When smoking was more widely accepted as a respectable activity it was common to exchange cigarettes to increase social interaction. Used in this manner, smoking was essentially independent of any of the chemical constituents of tobacco. These different outcomes from smoking were not mutually exclusive. A smoker might report simultaneously, for example, that he felt more relaxed and more alert, paradoxical as this might seem at first sight. The same smoker was likely to smoke for different reasons at different times: to change his mood if he felt depressed, to increase his level of alertness if he had been drinking alcohol, and so on in many combinations. There was, however, a notable absentee from the list of effects of smoking reported by smokers: it was hardly ever reported to cause feelings of intense pleasure and euphoria ("rush") that were commonly reported after ingestion of opiates such as heroin or stimulants such as cocaine.
[5.383] Professor Gray said that while self-report was an excellent starting point from which to embark upon scientific studies, the conclusions suggested by that route required confirmation by other means. It was, however, very far from straightforward to test the behaviour and effects of smoking in the laboratory. In any scientific investigation, one wanted to eliminate anything which changed when the variable of interest was changed and which might therefore spuriously produce effects which would be attributed to the main variable. This was the problem known as "confounding". Ideally a study group and a control group would be selected at random and the variable, nicotine for example, would be applied to the study group and not to the control group, as for example by the authors of Rose et al. 2000, who applied either nicotine or saline through an intravenous line without the subjects knowing which they were receiving. There were, however, difficulties in connection with cigarette smoking. Only existing cigarette smokers could be included in a study group, as it was unacceptable to take people at random and induce some of them to start smoking; and it was known that non-smokers and smokers differed systematically so that they were not in all relevant aspects similar apart from smoking.
[5.384] Laboratory studies of the behavioural effects of smoking supplemented self-report by taking objective measurements of the effects of smoking, but, like self-report itself, they were susceptible to the potential confounds of placebo and withdrawal effects. The behavioural or psychological effects of smoking might be due, not to the inhaled smoke itself, but to the smoker's beliefs about what those effects should be, i.e. placebo effects. The observed effects might indeed be due to the inhaled smoke, but because of reversal of withdrawal symptoms rather than direct effects that would occur also in a person who had never smoked. Furthermore, any differences observed between non-smokers and smokers (whether the latter smoked or not during the experiment) might reflect, not the effects of past or present smoking, but rather pre-existing differences between people who did or did not take up smoking (smokers and non-smokers differed from one another genetically and in personality). For these reasons, he said, scientists would prefer to find some way of study the smoking habit without having people smoke at all.
[5.385] To achieve this difficult ambition scientists studied, instead, the effects of nicotine. They made a simplifying hypothesis that the effects of tobacco smoking were due to the nicotine contained in the tobacco and then they studied the effects of the nicotine. In that way, all these confounds could be eliminated. But there were problems with this approach. It would work if, and only if, it was indeed the case that all of the effects of tobacco smoking were produced by nicotine, and this was not necessarily the case. If it were to be supposed, however, that particular observed effects of nicotine (the same in non-smokers and smokers alike, and shown not to be placebo artefacts) took the same general form as those of smoking, then one could reasonably conclude both that these were likely to be real effects of smoking tobacco and that they were caused by the nicotine absorbed from tobacco smoke. Further support for this conclusion could be derived from studies with animals. If, in animals that had never before been exposed to any psychoactive compound, the administration of nicotine had effects that resembled those seen in smokers smoking tobacco, then one could again reasonably conclude that these were real effects and due to the nicotine in tobacco smoke.
[5.386] Professor Gray went on to say that a number of effects that met the criteria of having been reported by smokers as beneficial and as having been confirmed in laboratory studies with human and/or animal subjects consisted in improvements in performance and cognitive function. This was the area of psychological performance which was furthest away from emotion and closest to simple problem-solving. These effects included increased levels of arousal and alertness, improved capacity to maintain performance during long and boring tasks, improved capacity selectively to attend to task-relevant information, improved capacity to maintain concentration of attention, faster learning and enhanced "working" memory.
[5.387] There were also confirmed effects of smoking tobacco and the administration of nicotine on mood and emotion. Prominent among such negative moods was depression. However, it was controversial whether the anti-depressant effects of smoking were due to an intrinsic effect of tobacco as such, or whether they reflected a reversal of depression occurring as a withdrawal symptom in abstinent smokers. Increased depression was certainly reported by many smokers during the period immediately after they quitted smoking. It could not, however, be deduced from this sequence of events that quitting smoking led to depression as a withdrawal symptom, that is, a symptom that would not have occurred had there not been a history of smoking followed by abstinence, in the way that diarrhoea formed part of the opiate withdrawal symptom. There was good evidence that smokers differed as a group from non-smokers. Part of these differences lay in their personality and their susceptibility to mood disorder, including depression. It could be the case that smokers started out in life prior to becoming smokers with an elevated tendency to be depressed and that, when they became smokers, their smoking habit reduced their level of depression so that, when they stopped smoking, their depression returned to what, for them, would have been a normal constitutional baseline. So in that case smoking would not be alleviating depression by reversing a withdrawal symptom, but rather not smoking was unmasking a constitutionally high level of depression in the smoker. It was not possible directly, so far as Professor Gray was aware, to test that hypothesis experimentally in smokers themselves.
[5.388] The possibility of intrinsic anti-depressant action was supported by several lines of evidence. People with a history of depression were more likely to smoke than those who lacked such a history. An important twin study indicated, furthermore, that this relationship between lifetime smoking and lifetime major depression results solely from genes that predisposed to such conditions. In addition, it had been shown, in both animal studies (using behavioural models of depression) and human non-smokers treated with a transdermal nicotine patch, that nicotine reduced depression. Nicotine, then, clearly could reduce depression that was not due to smoking withdrawal; and it was likely therefore that smoking had similar effects.
[5.389] The twin study referred to above by Professor Gray was Kendler et al. 1993. In this study of personally evaluated female twins from a population-based register, average lifetime daily cigarette consumption was strongly related to lifetime prevalence and to prospectively assessed 1-year prevalence of major depression. Using the cotwin control method, the authors evaluated whether the association between smoking and lifetime major depression was causal or noncausal. While the relative risk (95% confidence interval) for ever-smoking giving a lifetime history of major depression was 1.48 (1.30 to 1.65) in the entire sample, it was 1.18 (0.88 to 1.47) and 0.98 (0.71 to 1.26) respectively in dizygotic and monozygotic twin pairs disconcordant for a history of major depression. The relative risk for a history of major depression given ever-smoking was 1.60 (1.39 to 1.83) in the entire sample, while in dizygotic and monozygotic twins discordant for smoking it was 1.29 (0.87 to 1.74) and 0.96 (0.59 to 1.42) respectively. Controlling for personal smoking history, family history of smoking predicted risk for major depression; controlling for the personal history of major depression, family history of major depression predicted smoking. The best-fitting bivariate twin model suggested that the relationship between lifetime smoking and lifetime major depression resulted solely from genes that predisposed to both conditions. These results suggested, according to the authors, that the association between smoking and major depression in women was not a causal one but arose largely from familial factors, which were probably genetic, that predisposed to both smoking and major depression.
[5.390] Professor Gray said that similar considerations applied to the greater calmness and reduced irritability commonly reported by smokers. There were the same possible confounds as previously discussed. There was much less evidence about irritability than there was about depression, but there was evidence from humans that nicotine was likely to be involved in that particular effect of tobacco, because irritability had been observed to be reduced in withdrawing smokers who were given nicotine replacement therapy. So it was likely that the effects of smoking in reducing irritability were mediated by nicotine. The question then arose whether the nicotine was working by way of alleviation of withdrawal, or as a direct effect in reducing irritability, with or without withdrawal. The evidence that it could be through an intrinsic route of action was that in animals aggressive behaviour had been reported in several studies to be reduced by injection of nicotine, so again there was a direct effect similar to that reported by smokers. So the effects reported by smokers might be real, rather than the control of a withdrawal symptom.
[5.391] The situation regarding anxiety was much less clear cut than with regard to depression. Most of the experimental work had been carried out on animals. There was a very substantial literature, but it was also very complex. There were many reports of injections of nicotine increasing behaviour in animals which was generally taken to have an important conceptual resemblance to human anxiety, but there were also many reports in which nicotine did exactly the reverse. The equivalent laboratory evidence in human was very scanty. It was a problem ethically to induce serious anxiety in the human laboratory, so there were few data to go on. The scant information about human studies was again inconsistent in the direction of the obtained effects. It was likely that both these effects were real: nicotine did indeed both increase and decrease anxiety, depending upon the nature of the environmental circumstances. On the basis of the animal literature, Professor Gray said that nicotine could have the effect of reducing anxiety by acting on some systems in the brain, while also being able to increase anxiety by acting upon others. Smoking could therefore have an anxiolytic effect, which was real rather than simply the control of a withdrawal symptom.
[5.392] As an overview, Professor Gray said that nicotine and smoking tobacco had a remarkably broad spectrum of activity in alleviating negative mood and emotion, including depression, irritability and anxiety.
[5.393] Professor Gray then explained that the functional view of smoking is that smokers smoked to obtain the broad spectrum of benefits referred to above, with some smokers smoking more for some of these effects, others smoking more for others of them, and most smokers smoking for a mixture of them. In that manner, smoking did not differ very much from most of the complicated behaviour that people engaged in. This functional view reflected his own view.
[5.394] Professor Gray next referred to what he called the "addiction" view. He said that in contrast to the complexity of smoking behaviour already noted by him, it had become almost a point of dogma to treat the continuation of tobacco smoking as entirely dependent upon one particular compound: nicotine. At its simplest (the level, indeed, at which it was most often expressed), this view was summed up in two bullet points on the back cover of RCP 2000:
" |
|
Professor Gray said that he did not agree with the view that smokers do not smoke out of choice but because they become addicted to nicotine. He explained the "addiction" model in this way. It contained a number of key features, each of which had to be considered in turn and put together to make the entire model. The first point was that there was in the brain a reward system upon which all of the drugs that people self-administered acted, and they acted in such a manner as to become "super rewards". They activated that system to an unusual degree or in an unusual manner. According to this model, the effect of the drug on the reward system was subjectively experienced as euphoria of the kind that cocaine users reported as being a "high" or a "rush". The next point the model made was that there was a development of tolerance to the repeated use of the substance in question, meaning that the effect of the substance, in producing the euphoric state, reduced and that the user of the substance, in the effort to re-attain the initial euphoric experience, increased the dose of the substance in a process known as "dose escalation". At the same time the body in general adapted to the increasing amounts of the drug that the user was ingesting, and that adaptation served the well-known function of homeostasis. This was the process by which the body attempted always to revert to the status quo after there had been any serious shift in the balance of its activities from the equilibrium point. After tolerance had developed, there were two opposing sets of effects: the original set of direct drug effects, and the counteracting set directed at maintaining homeostasis. If the drug was not further ingested, only the counteracting set of effects remained, which were experienced as unpleasant physical withdrawal symptoms. The drug-taker now had two strong compulsions to continue taking the drug: craving for its euphoric effects and the need to avoid unpleasant withdrawal symptoms.
[5.395] Application of the "addiction" view to nicotine, Professor Gray said, given these assumptions, was accomplished as follows. First, the neurochemical signature of activation in the reward system was taken to be the release of dopamine in the nucleus accumbens of the brain. This could be readily measured in laboratory animals and there was no question that nicotine did produce that effect, in rats for example. It was also correct that the same release of dopamine was seen in animals injected with opiates such as heroin and stimulants such as cocaine. These facts were not in dispute. The interpretation placed upon those facts was that nicotine was a drug of abuse because it had those effects on the brain. Professor Gray said that he did not share this interpretation. The same effect of the release of dopamine in the nucleus accumbens could be produced in rats by an electric shock to the feet, by a tone that the animal had learned to recognise as a warning of an electric shock to the feet, or even by a tone that simply told the rat that it was going to be stimulated by a light. All of these effects had been demonstrated in his own laboratory. None of them was consistent with inferring, from dopamine release in the nucleus accumbens, that the event which had produced that dopamine release was a reward, let alone a euphoriant. So he accepted the facts about nicotine giving rise to dopamine in the nucleus accumbens, but not the inference from those facts that nicotine was producing the kind of euphoric state that the addiction model associated with those observations.
[5.396] According to the "addiction" model, the motivation to smoke tobacco and the difficulties smokers had in giving up the habit were due to the twin compulsions of craving for nicotine and avoidance of withdrawal symptoms. In Professor Gray's view, there were flaws in the model. First, there were several serious weaknesses in the "euphoria" side of the "addiction" view. Common experience, whether that of smokers themselves or those who observed smokers, gave no hint of a "rush" or "high" as described by users of cocaine or heroin. Consistent with this common-sense appraisal, in controlled studies smokers did not report euphoric responses to any significant degree to either tobacco or the intravenous administration of nicotine. This was supported by the findings of Rose et al. 2000.
[5.397] The contrary view rested upon a report that intravenous nicotine and inhaled tobacco smoke increased ratings on a "euphoria" scale on which high ratings were given also to morphine and amphetamine: Henningfield 1984. Professor Gray said that the data given in Figure 5 on p.29 of this paper, "Scores on the euphoriant scale for compulsive behaviors", had been repeatedly cited in the literature. He explained that there had previously been developed a number of self-report scales to measure various different ways in which drug users described their experiences of taking drugs. A range of compounds had been studied, as shown in Figure 4 on p.28, including morphine and benzedrine, and the point of the study was to compare the responses of such subjects to intravenous nicotine, using these scales, with the responses to the other compounds that had previously been studied. As stated on p.28:
"In these studies, volunteers are given a range of doses of the test compound and placebo under double-blind conditions. Individuals with histories of drug abuse are used as subjects because they can accurately discriminate compounds with a potential for abuse and can compare the effects of the compounds to those of abused drugs."
Professor Gray said that subjects with a history of drug abuse were not representative of the general population in two respects. First, they differed systematically from the general population in the personality characteristics which led them into taking illegal drugs on a regular basis. Secondly, it was unknown whether the taking of these other drugs would have altered the way in which they responded to a test compound, nicotine in the present case. So it was not possible to generalise these results to smokers who, for the most part, were not takers of illegal drugs. Most smokers simply smoked cigarettes and were a different population from the population of illicit drug users.
[5.398] Henningfield went on to describe how doses of the test compound and placebo were administered to volunteers who reported their responses. The responses were measured according to two scales: the Liking Scale (Single Dose Questionnaire) and the Morphine Benzedrine Group (MBG) scale of the Addiction Research Center Inventory (ARCI). Figure 5 showed responses to "nicotine, tobacco, other drugs of abuse and simulated gambling". The author stated, at p.29:
"The ARCI data are consistent with the Liking Scale data, confirming that nicotine, given by both routes of administration, was a euphoriant. When asked to identify the injections from a list of commonly used and abused drugs, subjects more frequently identified nicotine injections as cocaine."
Figure 5 set out the findings on the ARCI scale. Professor Gray said that these were the findings which had been most widely cited. He did not regard the Liking Scale as relevant to the question whether there was a euphoric response: one might like many things which did not produce euphoria. The data in Figure 5, however, were specifically concerned with a comparison between the effects of nicotine on the ARCI and those of the drugs in respect of which the scale was first formulated, morphine and benzedrine (a form of amphetamine which at that time was in common use). The argument had been used that on this scale nicotine had the same effects. Figure 5 showed nine histograms each showing two bars, one for placebo (P) and the other for the "drug" under investigation (D) or simulated gambling (SG). In all nine histograms the D or SG bar was higher than the P bar. Professor Gray said that the significance was said to be that nicotine increased scores on this scale to the same degree as that seen for morphine and amphetamine, for example, and from the way the figure was drawn that appeared to be the case. It has subsequently been pointed out by Warburton, however, that the scales on the vertical axes of each of the histograms were quite different from each other. This was standardly regarded as very poor scientific methodology and should have been pointed out by the referees of the paper at the outset; it was something that graduate students were taught at an early stage not to do. For morphine for example, the P value was below 4 and the D value nearly 10; the same for amphetamine. For nicotine, the P value was 5 and the D value just below 7. But the scales had been set so that the difference between P and D for nicotine appeared to be as large as that for amphetamine and morphine, though this was simply not the case.
[5.399] Professor Gray described Professor Warburton, who had been professor in the Department of Psychology at Reading University, as a leading researcher internationally in the field of psychopharmacology. He had recently received a lifetime award for his contribution over the entire span of his career to that field of research and was a very highly respected scientist. In Warburton 1988a at pp.31-34 there was discussion of Henningfield 1984. Warburton noted that there were remarkable differences in the scales shown in his Figure 3.1 (the same as Figure 5 in Henningfield 1984) and the results took on a different picture when plotted on the same scale, as in his Figure 3.2. If the scores for the difference between assessment of the substance and placebo were plotted it was possible to derive a ranking of euphoria, or "abuse potential" to use Henningfield's term. It could be seen that nicotine injections and smoking were low on the euphoriant scale, and had low "abuse potential" on the scale that Henningfield used for assessment. On this evidence there was not strong support for Henningfield's conclusion. From the data it seemed that nicotine was, at best, a weak euphoriant and did not act like other compounds in the maintenance of other kinds of substance self-administration, i.e. it is not like morphine in opium use.
[5.400] Warburton went on:
"In the laboratory of the Department of Psychology, University of Reading, Henningfield's work has been followed up by considering the concepts of 'euphoria' and 'pleasurable well-being' in terms of two separable experiences, pleasurable-stimulation and pleasurable-relaxation. Tests were made on 139 subjects for their recall of their experience of different substances and activities. They were asked to rate these substances and activities on a scale from zero to ten, where ten was the maximum imaginable. The sample was a set of subjects who had some experience of using a variety of different substances. In this way, ratings of the pleasurable-stimulation and pleasurable-relaxation of a set of substances and activities, including tobacco, were obtained. These ratings enabled a comparative ranking of the substances and activities on these two kinds of euphoria to be derived [...].
The most interesting comparisons were those of tobacco use with other substances and activities. Alcohol, amphetamines, amyl nitrite, cocaine, heroin, marijuana, and sex were significantly more stimulating than tobacco. Sleeping tablets and tranquillizers were significantly less stimulating than tobacco, while there were no statistically reliable differences between tobacco and caffeine or chocolate in terms of pleasurable-stimulation.
On the pleasurable-relaxation dimension, alcohol, heroin, sex, sleeping tablets, and tranquillizers were significantly more relaxing that tobacco. Amphetamine, amyl nitrite, cocaine, caffeine and glue were significantly less relaxing that tobacco, while there were no statistically reliable differences between tobacco and chocolate in terms of pleasurable-relaxation."
[5.401] Warburton accordingly disagreed with the statement in Henningfield 1984 at p.30:
"The results of these studies provide direct evidence that nicotine, in doses comparable to those delivered by cigarette smoking, is an abusable drug. That is, nicotine meets the critical criteria of being psychoactive, producing euphoriant effects, and serving as a reinforcer. These findings suggest that the role of nicotine in cigarette smoking is similar to the roles played by other drugs in the maintenance of other kinds of substance self-administration, e.g., morphine in opium use, tetrahydrocannabinol (THC) in marijuana smoking, cocaine in coca leaf use, and ethanol in alcoholic beverage consumption."
A further criticism of Henningfield's histograms, according to Professor Gray, was that there were no indications of statistical significance. There were no standard error bars, so it was not possible to state whether the modest difference in the D and P values for IV nicotine and tobacco was a statistically significant difference or not.
[5.402] Returning to the issue of self-report, Professor Gray said that smokers did, in contrast, report reduced "craving", increased smoking satisfaction and increased psychological reward when, after overnight abstention from tobacco, they smoked cigarettes from which the nicotine had been removed: Rose et al. 2000. Thus it was misleading to liken the pleasure derived from smoking cigarettes to that occasioned by cocaine or heroin; and the pleasure so derived did not depend principally upon nicotine.
[5.403] Professor Gray then turned to the question of self-administration. He said that other evidence cited in support of the notion that nicotine had directly rewarding ("reinforcing") or quasi-euphoriant effects was that animals would work (e.g. by pressing a bar) to self-administer nicotine intravenously. There were indeed many demonstrations of nicotine self-administration in animals. He said that these had been accepted rather uncritically, as was thoroughly analysed in Frenk and Dar 2000, but despite that critique he believed that the evidence was that animals would self-administer nicotine. However, the conditions under which they did so were much narrower than those in which animals self-administered heroin or cocaine. It was harder to train them to do it and once trained they did not show the extreme avidity for nicotine that in other experiments had been shown for heroin and cocaine. There were studies in which 30% of animals had kept on working to ingest the drug to the point of death and in the case of cocaine as many as 90% had done so. He had never seen a report that an animal self-injecting nicotine produced that kind of deleterious effect upon its own life or its general health.
[5.404] Professor Gray explained that Frenk and Dar 2000 came to the view that there was no good evidence that animals would self-administer nicotine. They did a very effective analysis of the great majority of experiments that had been published in this field. They demonstrated a number of potential artefacts which were not controlled for in these studies. For example, in most experiments, the possibility was not controlled for that when animals self-injected nicotine, the rate of bar-pressing went up as a consequence of the nicotine increasing activity, rather than because the animal was working in order to obtain the nicotine. Professor Gray's own view of the literature, however, was that there was evidence that animals would self-administer nicotine.
[5.405] Critically, he said, there was no reliable evidence that nicotine as such (when given by routes other than the smoking of tobacco) was liked, reported as pleasurable, or self-administered by human subjects, whether smokers or not; and some evidence, especially in non-smokers, that it was actively disliked. It was notable in this context that the various nicotine replacement therapies (such as nicotine-impregnated chewing gum, transdermal patches or nasal sprays) had only weak or no effects in aiding attempts to quit smoking, and had not themselves become a focus for drug-seeking behaviour. Abstinent smokers given nicotine replacement therapy did not normally continue to take nicotine for long periods, in sharp contrast to abstinent heroin users treated with the opiate substitute, methadone, who typically continued to take this drug for years. And, when patients were treated with nicotine for other medical conditions, such as ulcerative colitis, they did not exhibit nicotine-seeking behaviour, whether they were non-smokers or ex-smokers; nor did the ex-smokers revert to smoking as a result of this treatment.
[5.406] Professor Gray said that in the case of other drugs it was well-established that if, for example, a heroin user had not been taking heroin for a long period of time, exposure to heroin re-instated a new round of high level heroin use. Tobacco did this for smokers, but nicotine did not. When I asked him whether this kind of relapse was attributable to "addiction", Professor Gray said that the vocabulary of addiction could be used descriptively or as an attempt at causative explanation. His view was that as an attempted method of explanation the language of addiction was not helpful. It did not explain. It made sense to him to describe certain forms of behaviour as "addictive", but to describe someone as "addicted" in no way provided an explanation for his behaviour.
[5.407] Professor Gray continued by saying that, given these weaknesses in the evidence that nicotine made a major contribution to the pleasurable effects of smoking, one must consider the possibility that the appeal of cigarettes derived, at least in part, from other sources. There was little evidence that nicotine was involved in the pleasurable aspects, as distinct from the cognitive aspects and the mood-improving effects, of tobacco smoking. That raised the question to what else one might attribute the satisfaction of smoking, its pleasurable effects. Nicotine was the only individual constituent of the whole cluster of things that made up tobacco and tobacco smoking that had actually been investigated in the laboratory to any significant degree. So the rest of smoking had to be treated in a much more general manner. In doing this, one could either consider the whole of the sensorimotor aspects of smoking or the "tar", which was a conglomerate of much of the other constituents of tobacco smoke once the nicotine had been removed. One could ask questions such as whether the pleasure of smoking was in part due to the tar, or was in part due to the sensory aspects of smoking, or the motor habits involved in taking a cigarette - lighting it up and smoking it and so on - or all of those together. There was evidence that tar was involved because, when scientists altered the tar content of cigarettes, while holding the nicotine content constant, smokers chose to compensate their smoking to obtain the level of tar that they were used to in their accustomed brand of cigarette. There was other evidence from other experiments that the sensory aspects of smoking were involved, and so the pleasure probably came from a combination of those two routes.
[5.408] Professor Gray next addressed the question of tolerance. He said that the addiction model focused in particular upon the development of tolerance to the euphoric effects produced initially by a drug. Nearly all effects produced by drugs could be subject to tolerance to some degree. But in the case of the addiction model, tolerance to the euphoric effects should be such as to lead the substance user on to taking more and more of the substance. In the case of smoking, there was no euphoric effect as such, though there were pleasant effects of smoking, but no evidence that these underwent tolerance to any significant degree.
[5.409] Professor Gray went on to say that in considering the withdrawal side of the "addiction" view, two related distinctions needed to be borne in mind. First, the term "withdrawal symptom" was used both descriptively and causatively. In its descriptive use it was sufficient to observe that certain changes followed upon discontinuance of taking a substance. What smokers reported when they had recently given up smoking was correctly described as a withdrawal symptom. The term did not add to the description. Causatively however it was used with additional meaning. It implied that the observed changes were due to the absence of the discontinued substance and reflected neural and/or bodily adaptations to the substance when it was present, the withdrawal symptoms being attributable to the homeostatic counteracting process. There were further surplus meanings on top of the descriptive use when the concept of withdrawal was applied causatively. The first additional meaning was about how the withdrawal symptoms were themselves caused; the second was about what the withdrawal symptoms did to the individual who was experiencing them. It was said that the withdrawal symptoms caused the individual to take up again the habit that was being discontinued. Care needed to be taken to make the distinction, because there was no doubt that, after smoking was discontinued, smokers reported a range of changes from the state they were in when they were smoking. Professor Gray questioned the additional causative meanings thereafter. The fact that a change was observed after a drug was discontinued does not justify the inference that the change was caused by the discontinuance, still less that it was caused in just the way implied by the causative sense of withdrawal. In order to justify that inference, further evidence was needed.
[5.410] The most compelling evidence turned upon a further distinction between withdrawal symptoms, in the descriptive sense, that were or were not specific to the substance whose use had been discontinued. Normally, physiological adaptations to an ingested substance were antagonistic to the changes the substances produced directly. So the best evidence that a (descriptive) withdrawal symptom was indeed caused by drug withdrawal was that (a) it has specifically related to a change directly produced by the drug, and (b) took a direction opposite to that change. A symptom of withdrawal from heroin that clearly met this evidential criterion of causation was that of diarrhoea. Heroin, like other opiates, caused constipation to which the body adapted by homeostasis. Discontinuance of the drug unmasked the adaptation to this effect, and diarrhoea resulted. This symptom was "specific" in relation both to the discontinued drug (diarrhoea did not occur, for example, during cocaine withdrawal) and to a particular direct effect of the drug (constipation). Symptoms that did not meet these criteria were "non-specific". In general, specific symptoms were probably "withdrawal symptoms" in the full causative sense; non-specific symptoms probably were not.
[5.411] There were marked differences between the "withdrawal symptoms" described upon quitting smoking and heroin, respectively. As well as diarrhoea, symptoms of heroin withdrawal included tears from the eyes, bleeding from the nose, nausea or vomiting, sweating, fever, muscle pain, joint pain and lethargy. The symptoms described by smokers were almost entirely psychological, above all irritability, anxiety and depression, and also difficulty in concentrating, increased appetite and decreased heart rate. The two latter symptoms were probably specific, in that smoking reduced appetite and nicotine increased heart rate. Such self-reported symptoms, in the case of quitting smoking, usually peaked at about three to ten days and disappear by two to four weeks. Both "syndromes" included restlessness and disturbed sleep, but otherwise they have little in common. Quantitative comparison between two such different conditions was difficult and, to Professor Gray's knowledge, had not been attempted. From their respective descriptions, common sense would certainly judge heroin withdrawal to be much more unpleasant. This judgment was confirmed by the willingness of heroin users to undergo prolonged anaesthesia in order to get through withdrawal speedily and without actually feeling the symptoms, under a treatment in which an opiate antagonist was administered to the patient. USSG 1964 at p.352 stated that many heavy smokers might cease abruptly and, while retaining the desire to smoke, experienced no significant symptoms or signs on withdrawal; and Professor Gray said that this had been documented and had happened to people of his own acquaintance.
[5.412] As noted, he said, heroin withdrawal included a range of physiological effects, whereas quitting smokers largely reported psychological changes. The latter resembled those reported also by people who gave up cherished habits of many kinds, including those that did not depend upon any substance at all, e.g. gambling. Clinical treatments for gamblers had included measurements of those changes they reported when they gave up gambling, which were essentially identical to those reported by smokers, including increased irritability, increased anxiety, increased depression and difficulty in concentrating. One therefore needed to consider the possibility that the causal chain which led to the withdrawal symptoms in the person who quitted smoking might not involve nicotine at all, but might involve merely the habit of smoking: of carrying the cigarettes, and of smoking under certain conditions which formed an integral part of the smoker's life. In Professor Gray's view, however, the symptoms observed in a smoker who was quitting were due both to the loss of the nicotine and the loss of the habit. Some of the effects of quitting smoking could be reversed by giving nicotine, but others could not. This is "tricky territory": one had to distinguish whether the descriptively ascertained withdrawal symptoms were causative and also between the descriptive withdrawal symptoms that could be reversed by nicotine and those that could not. The psychological changes reported by quitting smokers were clearly non-specific and it could not be inferred that they were caused by the absence of tobacco or of any ingredient, including nicotine, found in tobacco. The exceptions to this generalisation were the increased appetite and the decreased heart rate observed after quitting smoking, which were both opposite in sign to the decreased appetite and increased heart rate caused directly by nicotine. However, neither increased appetite nor decreased heart rate were unpleasant, and so they could not fulfil the role putatively played by "withdrawal symptoms" in driving the quitting smoker back to smoking. With these exceptions, the behavioural and psychological changes observed in quitting smokers were more reasonably interpreted as reflecting non-specific dysphoria consequent upon disruption of a habit and loss of the pleasure or other benefits the habit provides.
[5.413] Furthermore, the addiction view of smoking focused not on withdrawal from tobacco but withdrawal from the putative addictive substance in tobacco, nicotine. If one could show changes in animals with nicotine that resembled the changes in humans with tobacco smoking, that would be good evidence that the changes in the case of tobacco smoking case reflected something to do with nicotine. At first sight, reports of a withdrawal syndrome in rats and mice that had been given nicotine offered support for the notion that the human withdrawal symptoms, after smoking tobacco had been discontinued, resulted from a withdrawal from nicotine. But the withdrawal syndrome in rodents was extremely different from the symptoms in human beings: gasping, writhing, chattering of the teeth, chewing, drooping of the eyelids, shakes, tremors and yawns in no way resembled those seen in quitting smokers, even though these were all changes that could be easily observed in human beings were they to exist. In addition, the withdrawal symptoms described in human beings who had received nicotine by way of skin patches, chewing gum or other such routes were minimal.
[5.414] If the addiction model were to apply to smoking, it would be expected that withdrawal symptoms would be a significant part of the causal chain that led the quitting smoker to resume smoking, and that in turn led to the prediction that the major time at which smokers who were quitting would resume smoking would be when withdrawal symptoms were at their strongest. This simply did not occur. In smokers who had quitted, the resumption of the habit ("relapse") did not correlate well with self-reported "craving" or withdrawal symptoms. This was borne out by evidence. Furthermore, most resumption of smoking took place well after the first few weeks of abstinence (the period when withdrawal symptoms and craving were at their greatest). Nicotine replacement therapy did reduce (as the nicotine addiction view predicted) self-reported symptoms, and did aid smokers to quit, but the improvement produced by it (relative to placebo controls) in successful quitting was modest, that is, of the order of a doubling from 10% to 20%, about the same as that produced by cognitive-behavioural counselling. Moreover, the resumption of smoking did not follow the time-course that would be expected if it were due to withdrawal severity; nor did the reduction in resumption of smoking produced by nicotine replacement therapy follow the time-course expected if it were due to alleviation of withdrawal. That is, neither the former nor the latter was especially pronounced during the period when withdrawal symptoms were greatest.
[5.415] Professor Gray said that the same point was made in RCP 2000 at p.143, where it was stated:
"NRT [nicotine replacement therapy] alleviates withdrawal discomfort. However, as has been pointed out, the severity of withdrawal is only a weak predictor of success in stopping smoking."
He disagreed, however, with a statement in the same document, at p.xiv: "The major psychological motivation to smoke is the avoidance of negative mood states caused by withdrawal of nicotine." This was because there was more than one psychological motivation to smoke, and he did not think the data warranted the selection of any one of these as "the major psychological motivation". He accepted that smoking helped the smoker to improve negative mood states. But it was not established that the smoking was leading to the "avoidance" of a negative mood state that had been caused by the withdrawal of nicotine. There was evidence of a direct intrinsic anti-depressant effect of nicotine and there was therefore no certainty that should allow the formation of such a sweeping statement. In addition, he did not agree with the statement on the same page: "Most smokers do not smoke out of choice, but because they are addicted to nicotine."
[5.416] Professor Gray's conclusions on the "addiction" view were as follows. The two central claims of the view were (1) that "cigarettes are highly efficient nicotine delivery devices" and (2) that "smokers do not smoke out of choice but because they become addicted to nicotine." He did not question the first of these claims; nicotine reached the brain in about ten seconds from a puff on a cigarette (RCP 2000, p.36). However, given the flaws in the arguments and evidence as outlined by him, he said that one could not infer that the habit of smoking was wholly or even largely based upon this fact. As to the second claim, there was at the heart of the statement a false dichotomy between two ways of viewing human behaviour. One view could be described as mechanistic: it supposed that behaviour could always be attributed to a causal chain and that therefore what an individual did at a particular moment was entirely beyond his choice. If this determinist claim was correct, then it applied, not only to the smoking of cigarettes, but to everything human beings did, because there was always a causal chain. The other view was the normal humanist view of life that we all adhered to in our everyday activities and grew up believing, namely that when we did something which it was in our power to control, we chose to do it: we chose to do anything that we were not coerced to do by some obvious means.
[5.417] Professor Gray believed the dichotomy to be false, and that it went to the heart of the issues before this court. Within the scientific investigation of human behaviour, it would now be widely accepted that the brain should be considered as an intellect set of feedback systems which attempted to control variables in a way that was good for behaviour and that was adaptive. At the highest level of the feedback systems were those that helped to obtain rewards and to avoid punishments. A reward system was concerned with maximising the benefit to the individual and minimising the costs. Such a system could be described in mechanistic terms, but there was no contradiction with the language of choice, because rational choice was precisely about choosing that which as good for us and avoiding that which was bad for us. On one understanding of the statement that smokers did not smoke out of choice but because they became addicted to nicotine, it simply meant that they smoked because they wanted nicotine and that was in fact exercising their choice. However, it was not nicotine that smokers wanted, they wanted cigarettes. So when a smoker chose to take a cigarette, it did not change any of the facts to re-describe that behaviour as the smoker was "addicted" to cigarettes. It just shifted the focus from using one of these languages of choice to another of determinism.
[5.418] A more specific reason for disputing the statement was that if it meant that it was impossible for a smoker to choose not to smoke, that was simply contradicted by the data. Millions of people who were smokers had given up smoking, in particular when the cost and benefits changed with the understanding that they were running risks to their health. In the United States some ten to twelve million people had given up smoking over a period of twelve years. So smokers could quit, they could choose to quit and, if that was true for large numbers of smokers, there was no reason to say that any individual smoker was forced to continue smoking and could not choose to give up. Reference was made in this connection to USSG 1988, in which it was stated, at p.466:
"According to the 1985 National Health Interview Survey (NHIS), there are approximately 41 million former smokers in the United States. Approximately 90 percent of former smokers report that they quit smoking without formal treatment programs or smoking cessation devices [...]."
[5.419] Professor Gray said that the claim that smokers did not smoke out of choice but because they became addicted to nicotine was sharply contradicted by the epidemiological data. What distinguished those who succeeded in quitting from those who did not was at present largely a matter of conjecture, with the exception that high socio-economic class and membership of the medical profession were strongly associated with quitting. None of the proposed behavioural or psychological predictors of quitting (such as self-reported "craving" or intensity of withdrawal symptoms, number of cigarettes smoked a day, and so on) accounted for more than a small part of the variance. The "addiction view" resolved this conundrum by stating that those who did not succeed in smoking were "addicted" and "had no choice". But this statement lacked all explanatory power. If failed even to account for the association between quitting and socio-economic class or being a doctor, as there was no reason to suppose that these variables were inversely related to vulnerability to addiction. In some, to say that a smoker was "addicted" was based upon just that observation - that a smoker had not quitted - that it purported to explain.
[5.420] Professor Gray explained that "variance" was a technical statistical term which quantified the degree to which values were spread out around their average or mean. If the addiction model of quitting smoking were correct, from supposed measurements of the degree of addiction, such as self-reported craving or withdrawal symptoms, it should be possible to predict those individuals who would find it more or less difficult to quit. To say that those variables accounted for a very small part of the variance meant that, if they were taken into account, the scatter of measurements on whether or not people quitted remained very little changed. It could not be predicted with any accuracy who was going to quit more or less easily. That being so, the normal scientific inference would be that the hypothetical model was false as it did not apply to the data.
[5.421] Professor Gray went on to say that there was an alternative, more nuanced, account of smoking behaviour, one that offered a marked contrast to the simplistic "addiction" view. This "functional" account was based upon the fact that, by smoking tobacco, people derived a variety of benefits and help in coping with everyday life. Reference was made to Warburton 1988b. In this paper, with which Professor Gray said that he agreed, the author adopted a pharmacopsychoecological approach to smoking behaviour which proposed that this behaviour could only be understood by considering the ecological context of the person. This had been discussed in previous publications. He said that smoking was adopted because it enabled the smoker to control his psychological state. At any moment this would be the outcome of both exogenous and endogenous factors, the situation and the individual. The author presented evidence to show how situations determined the individual's smoking behaviour. At p.76 he stated:
"One of the propositions of the functional model is that smoking behaviour must be interpreted within the ecological context of the person. Consequently, nicotine use is seen as a purposive, coping activity for the needs of everyday life. Smoking is maintained because it has a function and satisfies these needs. Consequently, smoking is not an irrational behaviour but the behaviour is adopted because it enables the user to effectively control their psychological state. It can be seen as a psychological resource that the person uses [...].
As a smoker may have more than one motive, nicotine use may have different functions in different situations. However, smoking behaviour can be the outcome of not only the characteristics of the situation, but the personality of the individual. Personality characteristics will determine the way in which the person interacts with situations. Consequently, differences in smoking behaviour will be the result of the interaction of the individual with the situation, the person and their ecology."
[5.422] Professor Gray commented further on RCP 2000. He said that the conclusions in this report, and other such reports, formed part of an attempt to change behaviour, part of a health action, and as such, in interpreting the conclusions from a scientific point of view, one must always bear that context in mind. So he would not necessarily take the conclusions in the report as simply straightforward accounts of scientific evidence or conclusions. In addressing oneself to a scientific question the data that carried most weight were those in the primary experimental or empirical reports: the so-called archival data. So one would go, wherever possible, to those primary reports. One might then, or in advance of that, rely upon reviews of a particular field, in order to see what the general conclusions were, but also as a means of orientating oneself to the primary reports. One would not normally start with a book so far removed from the primary data as was RCP 2000.
[5.423] Warburton 1988b mentioned the importance of personality in smoking behaviour. There had been investigation into whether smokers tended to have personality characteristics which differed from those of non-smokers. Professor Gray elaborated on this. He referred to the general theory of personality, which in turn was based upon particular methods of data-gathering and data analysis. If many individual observations on large groups of individuals were collected and it was then considered whether, if individuals were ranked on one of these observations, this would predict how they would rank on other measures, and the overall set of correlations so obtained was reduced in order to obtain the simplest description of the underlying dimensions of variation, all of the research agreed that there was a dimension of introversion/extraversion so that individuals might be more or less socially withdrawn (introverted) or socially outgoing and ready to engage with others (extraverted). There was also universal agreement that there was a dimension of neuroticism. At one end of that scale, people were anxious, depressed, moody, irritable and so on, and at the other end they were calm and phlegmatic and did not have many mood changes. Then there was what Professor Gray called "the P dimension". "P" stood for "psychoticism" or "psychopathy". The simplest description was that "P" stood for high tendencies to engage in antisocial behaviour at the high end of the scale, or tendencies towards psychotic behaviour. If these three broad axes of variation were imagined as a three-dimensional space, all individuals fitted somewhere within it. If smokers were randomly drawn from the entire population, they would have average scores on each of these dimensions, with scatter similar to the rest of the population. In fact however smokers tended to be somewhat extraverted, somewhat neurotic and somewhat high on the P scale. Their positions on these scales were not necessarily immutable over the years, because they were themselves a response to social circumstances. This was because, as described by Warburton 1988b, smoking was not an immutable fixed effect of personality but was an interaction between the personality and the situational and cultural effects prevalent at the time. Spielberger and Jacobs 1982 found that smokers had significantly higher scores than non-smokers on the Eysenck Personality Questionnaire (EPQ) extraversion, neuroticism and psychoticism scales. Previous studies had led to the conclusion that smokers were more extraverted and had more antisocial tendencies than non-smokers, and the evidence was reasonably convincing that smokers were also more impulsive. Professor Gray said that introversion- extraversion, neuroticism and psychoticism were features which had a genetic base, and were determined by genetic factors to an extent of 30 to 50% of the variance, depending on the particular studies.
[5.424] On the question "Who smokes?", Professor Gray said that the incidence of smoking in the United Kingdom population had varied considerably over time. However, at any one time, smokers were not a random sample of the population, but rather had distinctive characteristics of their own, which were of several kinds: genetic, family environment, personality and social environment. Chronologically speaking, the first factor to come into play consisted in the inheritance of a genetic predisposition that favoured smoking. This term ran the risk of misinterpretation. It did not imply lack of choice. Nobody in the field of the genetics of human behaviour and psychology believed that a single gene determined irrevocably that someone would have a certain characteristic, but so-called genetic determinism had become sufficiently widespread, as a "straw man" set up by its opponents, that it was necessary to state the matter clearly. A "predisposition" was just that: a tendency which, other things being equal, made it more likely that the predisposed event (in this case, smoking cigarettes) would come to pass. Other things, of course, were rarely if ever equal; and the final outcome, in respect of this and all other behaviour, reflected an interaction, usually highly complex, between a variety of different factors that predisposed towards or against it. The genetic influences on smoking behaviour were in part mediated by the smoker's personality.
[5.425] Predisposing influences from the early environment worked in the same general manner as genetic predisposition, that is to say, they increased or decreased the probability that the individual would be a smoker. A particularly important role was played by the family environment: children whose parents smoked were more likely to smoke themselves. Interpretation of this correlation, however, was not straightforward. It might arise because: (i) parents passed on to their children genes that predisposed to smoking; (ii) children modelled their smoking behaviour on that of their parents; (iii) both parents and children were influenced by factors in the general environment in which the family was situated (e.g. within a culture in which smoking was prevalent); or (iv) a combination of these causal pathways. However, statistical genetic research had developed powerful methods for distinguishing them.
[5.426] It was usually impossible to establish all the actual factors involved (e.g. specific genes, or specific features of the family or general environment) or how they interacted with one another. What was possible, however, was to calculate the relative contributions of genetic factors and environmental factors. These relative contributions were described in statistical terms as the proportions of variation, technically "variance", in the relevant behavioural trait for which they accounted.
[5.427] It had been estimated, from studies of the similarities in smoking behaviour between identical and fraternal twins, that the genetic route accounted for about half the variance in taking up smoking. A further 30% of the variance was accounted for by shared family environment. Putting these two factors together, then, the family background accounted for about 4/5ths of the variance in taking up smoking, by way jointly of shared genes and shared early environment. Genes played an even greater role in the continuation of smoking than in taking it up, accounting for as much as 70% of the variance, with the influence of shared family environment falling to zero. Reference was made to True et al. 1997. In this paper the authors estimated the magnitude of genetic and shared environmental contributions to risk of initiation and maintenance of smoking. Genetic models were fitted to data from male monozygotic and dizygotic twin pairs. The best fitting model allowed for both genetic and shared environmental effects on smoking initiation, accounting for 50% and 30% of the variance in risk, but allowed for only genetic effects, accounting for 70% of the variance in risk, on persistence in smoking among those who had become regular smokers. They stated that this finding of a major genetic influence on smoking persistence confirmed similar results from twin studies in Scandinavia and Australia. The results yielded by these studies were consistent with the conclusion that there was an important genetic contribution to the risk in men of becoming long-term smokers, with heritability estimates ranging from 47% to 74%.
[5.428] A fourth factor relevant to the question "Who smokes?" was, Professor Gray said, social environment. Aspects of social interaction relevant to smoking behaviour included peer pressure, social rituals of sharing and exchange, the social setting and the self-image that the individual sought to project. There was considerable interaction between an individual's personality and these social factors. For example, an extravert would seek out social gatherings. Under conditions where social gatherings involved a large amount of tobacco smoking, and where rituals of exchange in sharing formed part of what went on at them, then an extravert was both more likely to encounter them and to be influenced by them. The level of an individual's score on neuroticism scales taken relatively early in life predicted the subsequent diagnosis of a range of psychiatric conditions, most notably anxiety and depression. Smoking had an antidepressant effect and might have an anxiolytic effect. Against that background the association between neuroticism and smoking could be explained by reference to the functional model of smoking. A person who was high in the dimension of neuroticism was likely frequently to experience moments of anxiety, depression and also irritability, general loss of calm and feelings of tension. Such a person, upon smoking a cigarette, was likely to discover that these moods were reversed rather rapidly. So an individual would discover that smoking a cigarette was a useful way of making himself feel better.
[5.429] Warburton 1988b found that an individual, by smoking, was able to reduce negative emotional feelings occurring during times of stress and to increase concentration. In an experimental vigilance session, individuals who had high degrees of extraversion and neuroticism improved their performance after smoking. At p.74 the author stated:
"It can be argued that the improved performance after smoking was a consequence of the tranquillising effects of nicotine, which reduced situational anxiety."
Professor Gray said that this provided an excellent example of the manner in which there was a constant interaction between the individual's personality, the demands that a situation placed on the individual, and the effects of smoking.
[5.430] At pp.74-75 Warburton wrote:
"Coping strategies can deal directly with the environmental challenge or indirectly to reduce its impact on the organism temporarily. Smoking can have both of these effects. In a challenging work situation, nicotine can enhance performance and experience of this enhanced efficiency will reduce an individual's anticipatory concerns about their behavioural competence [...]. When there is an uncontrollable life event, the nicotine can calm the person and improve mood generally. In both cases, the impact of the stressor on the person is reduced. In this regard, it is interesting to compare nicotine with alcohol, which is often used to calm and improve mood. The action of alcohol is to escape from stressors, while nicotine enables the person to confront them."
Professor Gray said that the different outcomes from smoking were not mutually exclusive. A smoker might report simultaneously, for example, that he felt more relaxed and more alert, paradoxical as this may seem at first sight. The same smoker was likely to smoke for different reasons at different times: principally to change his mood if he felt depressed, but principally to increase his level of alertness if he had been taking alcohol, and so on in many combinations. A person with a high P score, who was inclined to be aggressive, might find smoking useful because it could calm him down. This again fitted with the functional model. He also said that the reason why individuals who had quitted smoking might return to it well after the disappearance of any withdrawal symptoms could be explained by reference to the functional model. Returning smokers often cited periods of stress, depression and negative mood changes as the background for the resumption of smoking.
[5.431] Professor Gray was asked to consider averments in the pleadings for Mrs McTear that after commencing smoking, Mr McTear quickly became addicted to cigarettes, that in about 1971 he attempted unsuccessfully to give up smoking and thereafter attempted on numerous occasions to give up smoking but, due to his addiction, he was unable to do so. In order to comment on these averments, he had read the transcripts of the evidence given by Mr McTear on commission and by Mrs McTear at the proof, and records relating to Mr McTear's medical history, employment history and criminal record. He said that during the period when Mr McTear smoked, smokers were relatively high on extraversion, low on neuroticism and high on the P dimension. There was not a complete picture of Mr McTear's personality, but his early history demonstrated the kind of active social life one would expect from a relatively extraverted young man. This picture was strengthened by his Army records, which stated that "he has no interest other than in his own personal experience", a typical extravert characteristic. A high level of neuroticism was suggested by the diagnosis he was given in later life of suffering from an anxiety state. Also suggestive of high neuroticism was his need to drink alcohol to "increase his confidence" (Dr Lind's letter 23 October 1990). Evidence of antisocial behaviour suggestive of an elevated P score could be found at several points in his records, particularly his Army records. His frequent and poorly motivated changes of employment were testimony to the accuracy of his Commanding Officer's judgment that he appeared not to know where he was going in life, and were again consistent with a high P score. Criminal offences were especially associated with high P scores. At later ages, it was impossible to exclude his excessive drinking as a factor in the high rate of offending throughout much of his life. Excessive drinking was, however, unlikely to have been responsible for all of the offences committed by him, even in the form of acute intoxication. Furthermore, excessive drinking was itself associated with high P scores.
[5.432] Among the functional effects of smoking, the ones that figured more prominently in Mr McTear's testimony were concerned with mood control. He said that he liked, enjoyed or even loved smoking. He described the use of smoking to control negative moods. As previously stated by Professor Gray, smoking tobacco might alleviate depression, not because it had a direct antidepressant effect, but rather because, for a habitual smoker, deprivation of smoking had a depressant effect. This type of explanation perhaps applied to Mr McTear's descriptions of the use of smoking to control mood. He emphasised the period when he had temporarily given up smoking and in consequence could have been especially depressed and irritable. It was impossible to determine in Mr McTear's individual case whether the depression he experienced when he temporarily gave up smoking reflected a withdrawal symptom or the return to a natural relatively depressed baseline mood which smoking elevated. In an individual case, it was impossible to make a determination whether symptoms such as those described by Mr McTear when he tried to give up smoking reflected a return to a natural relatively depressed baseline mood, which smoking might have elevated, or withdrawal symptoms, or the effect of other factors such as alcohol, or indeed a combination of some or all of these factors.
[5.433] In Professor Gray's opinion there was nothing which prevented Mr McTear from giving up smoking at an earlier time than when he in fact decided to do so about a year before he died. The decision whether to continue or to cease smoking was made by the individual smoker from time to time in the light of all of the costs and benefits. Mr McTear did stop smoking when his balance sheet changed with the discovery that he had a diagnosis of cancer. The only change was in his state of knowledge about his medical condition. Had he formed a similar opinion earlier on, that would have increased the costs relative to the benefits of smoking and he could have stopped smoking at any earlier time. In general, he said, people did what they were motivated to do. Motivation depended upon the perceived costs and benefits of action that one could choose to perform or avoid. It was easy to understand, within such a motivational framework, that Mr McTear's discovery that he had cancer should alter his perception of the costs and benefits attached to smoking so that he at once quitted the habit. Common-sense would describe such a change as one in the willingness to give up smoking, not in the ability to do so.
[5.434] Professor Gray referred to a paper of which he was one of the authors, Gray et al. 1994. In that paper it was acknowledged that part of the authors' research programme had been supported by the UK Medical Research Council, the Wellcome Trust (a private charity providing similar support on a somewhat larger scale), RJ Reynolds Tobacco Company, the Council for Tobacco Research, BAT (British American Tobacco) and Schering AG Berlin (a pharmaceutical company). The first time Professor Gray received funding from the tobacco industry was in the mid-1970s. There was a shortage of funding from the Medical Research Council. He was interested in the functions of the cholinergic system in relation to Alzheimer's disease and, because of the role of nicotine in acting upon the brain as though it were that same system, the tobacco industry seemed to be a possible source of funding. So he obtained funding from RJ Reynolds Tobacco Company. It was very common for scientists to seek and receive funding for research from relevant sectors of industry. At that time it was not unusual for scientists to receive funding from the tobacco industry. No tobacco company had ever prevented him from publishing his research. It had, however, become unusual in the UK for medical or scientific research to be funded by the tobacco industry and indeed had latterly become impossible because universities and hospitals, in which this type of research was conducted, had formulated a policy not to accept funding from the tobacco industry. Both the Medical Research Council and the Wellcome Trust had adopted policies under which they declared that they would not fund research for either a scientist or his institution if he accepted funding from any part of the tobacco industry. In order to accept funding from these bodies, the acceptance of support from the tobacco industry was precluded.
Cross-examination of Professor Jeffrey Gray
[5.435] In cross-examination, Professor Gray was asked about his conclusion on the question why Alfred Mr McTear smoked. In his report he wrote:
"[He] smoked for a variety of reasons, to obtain a variety of different benefits, and to cope in a variety of ways with the problems and stresses of everyday life. In this respect, he behaved just like the majority of other people in his time and place, especially those from similar social groups, in similar social settings and with a similar family background."
He agreed with the suggestion that Mr McTear was much like the average person who smoked.
[5.436] Asked about funding he had received from the tobacco industry, Professor Gray said that this was for only the minority of his research, dealing with nicotine. The greatest part, and possibly all, of the articles he had written about nicotine were funded in part by the tobacco industry. Asked about his remuneration in connection with the present case, he said that he was being paid for his time spent as being an expert witness. He was paid by the hour and the total would "reach five figures".
[5.437] Shown cigarette packets with the legend "Smoking kills", he said that this was not his field of expertise. He was not medically trained. He had a personal view, but not a view as an expert. His personal view was that smoking had been established as a strong risk factor for a number of diseases. This did not mean that smoking killed. Ask about the statement in UKWP 1988, in the preface by the Prime Minister, that in Britain today more than 120,000 were going to die over the next year from illnesses directly related to smoking, he said it was not his field of expertise, he did not know. As an individual, he believed that smoking carried risk. Asked whether it would be extremely irresponsible for a Prime Minister to put out a message like this in a White Paper unless there was extremely strong evidence to back it up, he said he did not know, his understanding was that he had been called in this case as an expert witness. He was not as an expert entitled to any view on the degree of responsibility that fell upon the shoulders of the Prime Minister when he made a particular statement, though as a citizen of the United Kingdom he believed that our Prime Minister normally weighed carefully his words and the reasons for writing them, at least he hoped that that was the case. This was not a matter on which he had any expert opinion.
[5.438] Asked about the statement in the White Paper at para.1.8 that smoking was addictive, he said that he accepted a use of the word "addictive" in which it was descriptive of certain types of behaviour. Used descriptively, it was not unreasonable to talk of smoking in that way. But to say that smoking was "addictive" provided no explanation for the fact that people engaged for large parts of their lives in smoking behaviour. He would accept the statement that smoking was addictive if it implied that it was a former of behaviour in which people engaged frequently and repetitively over long periods of their lives. The definitions in DSM-IV were mutually contradictory as between addiction in general and addiction to nicotine in particular to the point that it was not possible to say it was one thing or the other. So Professor Gray would not agree that it was possible clearly to refer to nicotine as addictive, or to smoking as addictive behaviour, with reference to those sets of definitions. The critical causative sense of addiction, with which he did not agree, was that it caused smoking and smoking continuation. He could agree with a statement that nicotine could be seen as addictive, if what is meant by this was that it was capable of creating some of the dependence and withdrawal symptoms that were described in DSM-IV and ICD-10, so long as it was taken descriptively but not causatively. In common parlance the word "addictive" was typically used in response to the descriptive characteristics of a form of behaviour, but at the same time to give the misleading impression that one had understood the nature of the mechanisms that gave rise to that behaviour. It was ambiguously used in common parlance. Some smokers had a strong urge to smoke because of the multiplicity of effects that they found beneficial in smoking a cigarette, as previously described. Professor Gray knew many people who had stopped being smokers and in general they had told him that giving up smoking was somewhat unpleasant but not too difficult.
[5.439] Professor Gray did not agree that the statistics demonstrated that not a large proportion of smokers managed to give up. One of the "key points" in RCP 2000 at p.xiv stated that only about 2% of smokers succeeded in giving up in any year. Professor Gray said he thought that that number was probably correct, because if accumulated over twenty years it would correlate with the forty million or so US smokers who gave up smoking over that period. He saw no reason to dispute the further statement that 50% of young adult smokers would still be smoking when they were 60. He said that to know why 50% gave up by the age of 60 and 50% did not, would mean that there would be some way of predicting in advance who would quit smoking and who would not. None of the purported individual level predictors of quitting or failing to quit were able to predict to more than a very slight degree. The variables that were predictive were to do with social class. The reasonable inference that he would make from those figures was that a greater understanding of behaviour and the reasons for it and the particular costs and benefits attached to it was most important in accounting for quitting or failure to quit. The only solid predictor was socio-economic class, and beyond that it was a matter of inference. The statement in RCP 2000 that over two-thirds of smokers said they would like to quit and about one-third tried to quit in any year, but only 2% succeeded, meant, he said, that 6% of smokers who tried to quit succeeded in quitting in one year. This was rather low in relation to a large number of published clinical trials in which the number who tried to quit and who stayed quit was typically 10% without any additional help, rising to 20% with additional help. So in his view the statement in RCP 2000 rather strongly understated the percentage of smokers who succeeded in quitting and it did so in order to convey a particular kind of health message.
[5.440] There had been less substantial investigation of what happened over repeated attempts to quit. The evidence to date was that the chances that a particular individual would succeed in quitting on any individual attempt remained constant over a series of attempts; so that if there was a 10% chance of quitting on each of these occasions unaided, by the tenth attempt it became 100%. It was not possible to say, in respect of those who had not quitted by the age of 60, how many had tried unsuccessfully to quit. He had no reason to disagree with the statement that about two-thirds of smokers said they would like to quit and about one-third tried to quit. As for the reasons, the data tended to come from studies in which a fairly large cohort of smokers were asked rather simple questions and it was unlikely that any further probe could have been undertaken to establish how much they would like to quit, their reasons for wishing to do so and so on. The present state of the scientific art was that measurement of the behaviour of individual smokers prior to a quit attempt was a very poor predictor of which individuals would succeed. So he could not say why it was that on a particular occasion one smoker succeeded in quitting while another did not. The reasons were most likely, he thought, to reside in the smoker's individual circumstances and his perception and evaluation of those individual circumstances at the time he made each quit attempt. If the quitting were quitting from nicotine use, rather than smoking, then there was no problem whatsoever: the quitting rates from nicotine delivered by any other mode were essentially 100% immediately. Smokers smoked cigarettes, as he had previously explained, for a variety of reasons.
[5.441] Professor Gray was asked about a paper entitled "Motives and incentives in cigarette smoking" by William L Dunn, Jr of the Philip Morris Research Center, Richmond, Virginia. The paper was described in the text as a "presentation". It related to a conference on the Caribbean island of St Martin in 1972, attended by twenty five scientists, described as "pharmacologists, sociologists, anthropologists and a preponderance of psychologists". It was not clear from the text to whom the presentation was made. Mr McEachran told me that he thought it was "an internal report", but this did not explain who were the audience to whom it was addressed. At all events, especially because I only allowed it to be lodged on the morning of the day on which Professor Gray was cross-examined, he had not had an opportunity of considering it previously. Professor Gray agreed with a statement in the paper that the primary incentive to cigarette smoking was the immediate salutary effect of inhaled smoke upon bodily function. He said that his view was that there was something about the interaction between nicotine, on the one hand, and the rest of what was in tobacco, on the other, which underlay its great popularity. The paper contained criticisms of Eysenck's theory of personality as part of his explanation of smoking. Professor Gray said he agreed with the comment that it was absolutely impossible for the concentration of level of nicotine required to induce neural depression to be attained by means of smoke inhalation. He did not agree with the statement that to postulate both activating and sedating effects was to defy the documented universality of the activating, physiological effect of smoke inhalation. Under reference to Warburton 1988a, Professor Gray said that all of the effects that he had been talking about in relation to nicotine were produced by way of its action on the brain. Warburton had shown experimentally that the more extraverted the subject was, the more he would wish to do something to reverse the decline in alertness towards the end of a boring task, and one way of achieving this was the nicotine which produced the desired change in the level of arousal.
[5.442] Professor Gray was asked about passages in MacAskill et al. 2002. The authors stated that in Scotland there was in 1988 a marked social class gradient in respect of smoking between AB groups and DE groups, 13% of the former and 49% of the latter being smokers. He would expect Mr McTear to fit into Group D, "partly skilled/unskilled". He would not dispute that tobacco played an extremely important role in the lives of people in low-income communities, and that for many smoking was their main pleasure. He accepted that smoking offered both a respite from and a means of coping with a stressful and unrewarding environment. He had no reason to dispute that studies had shown that disadvantage was inversely associated with cessation success: the more disadvantaged, the less likely a person was to succeed in quitting smoking. He agreed that this would be a factor to take account of in the case of Mr McTear. Another factor, he agreed, would living in a poor neighbourhood and being a member of a disadvantaged community with limited exposure to non-smoker or ex-smoker "role models". He accepted that a barrier to smoking cessation in low income communities was the importance of smoking as a means of coping with stress. He also agreed that fair points were that smokers in such communities had low confidence regarding cessation and an unsupportive environment for it. But the evidence, to which he had previously referred, was that the family environment contributed to the initiation of smoking but not, apart from the genetic route, to the persistence of smoking once commenced. He did not, however, disagree with what was said in the document about barriers to quitting.
[5.443] Professor Gray was asked about various passages in USSG 1988 to the effect that nicotine had addicting properties. He agreed that he was "out on a limb" by comparison with many distinguished scientists and doctors. He did agree that nicotine was a psychoactive drug with actions that reinforced the use of tobacco. He would agree with the use of the word "addicting" descriptively, but not if it was intended to be explanatory of the activity of smoking cigarettes. A similar description could be applied to gambling. A report such as this was essentially part of a strategy aimed at changing people's behaviour and was supplying messages concerning health and the measures that people could take. So it was a mixture of propaganda and the scientific evidence upon which it was based, but with the interpretations and the language used to describe those data biased towards the interests of the message that the report was conveying. With regard to the scientific standard of proof it was a biased report.
[5.444] His attention was drawn to a passage in the introduction, on p.5:
"This Report was developed by the Office on Smoking and Health, Center for Health Promotion and Education, Centers for Disease Control, Public Health Service of the US Department of Health and Human Services [...].
The scientific content of this Report reflects the contributions of more than 50 scientists representing a wide variety of relevant disciplines. These experts, known for their understanding of and work in specific content areas, prepared manuscripts for incorporation into this Report. The Office on Smoking and Health and its consultants edited and consolidated the individual manuscripts into appropriate chapters. These draft chapters were subjected to an extensive outside peer review [...] whereby each chapter was reviewed by up to 11 experts. Based on the comments of these reviewers, the chapters were revised and the entire volume was assembled. This revised addition of the Report was resubjected to review by 20 distinguished scientists inside and outside the Federal Government, both in this country and abroad. Parallel to this review, the entire Report was also submitted for review to 12 institutes and agencies within the US Public Health Service. The comments from the senior scientific reviewers and the agencies were used to prepare the final volume of this Report."
Professor Gray agreed that this was an appropriate way to go about making a report on a matter like this, but his understanding was that the actual writing of the report was done by the Office on Smoking and Health and its consultants, it was they who edited and consolidated the individual manuscripts into appropriate chapters. He could vouch for the efficiency with which the scientific information of the time was gathered together in the chapters, but the final messages which had been quoted to him from the introductory material, as distinct from the individual scientific chapters, were put together as part of a health message by civil servants in the Office on Smoking and Health. Summaries, such as the one in this report from which Mr McEachran had quoted, mainly contained the views of those who had a particular health message that they wished to be heard.
[5.445] Professor Gray said that prior to this report the scientific literature was not concerned with the question whether one word or another - "habitual" or "addictive" - should be applied to one form of drug use or another. The report substituted a previous classification which separated heroin and alcohol on the one hand from smoking and other activities on the other, but the change was not based upon new scientific evidence of any kind. It was based upon a health message decision that it would be better to include them all as "addicting" drugs. The clinical community had preferred to use the expression "dependence" rather than "addiction", but the report prepared to define drug dependence and drug addiction as the same thing. None of these changes rested upon scientific or medical knowledge. In his view, the reason why smokers found it difficult to give up was partially habit-based, partially nicotine-based and partially the seeking of the positive aspects of smoking, such as taste and
enjoyment, which were not mediated by nicotine and were not simply the habit, but the consequences of the habit; the sensory characteristics. He did not agree that most smokers did not smoke out of choice but because they were addicted to nicotine. This might be the conclusion of USSG 1988, but it did not alter his own assessment of the evidence. He did not form his opinions on the basis of the authority of public health organisations.
[5.446] RCP 2000 stated at p.100:
"On current evidence, we can conclude that cigarettes are properly categorised among the most addicting substances as this form of nicotine delivery maximises the addictive effects of the drug. The fact that nicotine is of low abuse potential in controlled dosage form such as the transdermal nicotine patch or nicotine gum supports the conclusion that the form of delivery is an important determinant of its addiction potential."
Professor Gray said that he did not agree with this because exactly the same evidence supported the conclusion that nicotine was not an addicting substance; if it was, the change of the form of delivery would not have such an important effect. The final conclusion in this passage was a good example of the way in which the conclusions had to fit both the scientific evidence in such a report but also the health message that was being conveyed.
Re-examination of Professor Jeffrey Gray
[5.447] In re-examination Professor Gray pointed out that Warburton had been involved in the preparation of USSG 1988. In Warburton 1988b he referred to criticisms he had published of the conclusions of the editors of the report. Professor Gray said that the view he had reached on his reading of the evidence was not dissimilar, as regarded the addiction model, to that of Professor Warburton. Frenk and Dar 2000 expressed views with which he also agreed about the inappropriateness of describing nicotine as addicting.
[5.448] Dr James, aged 40, had been senior lecturer in Forensic Pathology at the University of Wales College of Medicine since 1977. He was Honorary Consultant in Forensic Pathology to Cardiff and Vale NHS Trust, also since 1997. In the latter post his responsibilities were to run the hospital consent autopsy service, in association with other pathologists, to teach the postgraduate pathology trainees and carry out other teaching duties, and to liaise with other clinicians on matters relating to mortality and post mortem examination of their patients. He was a Home Office registered pathologist since 1994. In that post he performed autopsies on behalf of the coroner and, in dealing with suspicious deaths, in conjunction with the coroner and the police. He held the degrees of MB, ChB, the Diploma in Medical Jurisprudence of the Society of Apothecaries (in forensic pathology) and he was a member of the Royal College of Pathologists. He sat on the Forensic Sub-Committee of the Specialist Advisory Committee in Histopathology of the Royal College of Pathologists. He was an examiner and the specialty training advisor in forensic pathology of that college.
[5.449] Dr James's interests included death certification and the investigation of death. He had written a number of papers about this and had recently given evidence to the Smith Inquiry (the Shipman Inquiry), to the Review of Death Certification and to the Review on Death Certification and Coroners (the Luce Report). The last mentioned of these was a fundamental review of the coroner system in England and Wales, also encompassing matters relating to death certification. He gave evidence on death certification, on cremation certificates, on the function of the coroner system and the role of the pathologist and of the medical advisor within it. The Review of Death Certification was a departmental review within the Home Office. He submitted to the Shipman Inquiry a document based on his previous documents to the Luce Inquiry and also relating to other matters more particularly addressed by the Shipman Inquiry. He was then asked for further input and took part in a series of seminars chaired by Dame Janet Smith in 2003.
[5.450] Dr James was one of the signatories of a letter published in the British Medical Journal on 3 October 1992, James et al. 1992. The letter started:
"The reasoning behind the decision that doctors can now put smoking as a cause of death on death certificates without the death having to be reported to a coroner is obscure."
Later in the letter the authors wrote:
"That smoking is associated with several potentially fatal diseases is not in dispute; the difficulty lies in applying an epidemiological and statistical association to individual cases. How is a doctor to determine that smoking is of sufficient causal importance to be separated out from other possible risk factors in those diseases - particularly coronary artery atherosclerosis - regarded as multifactorial in aetiology? Even those diseases with the highest smoking related mortality ratios - lung cancer and chronic obstructive pulmonary disease - can occur in non-smokers. [...] Faced with the spectre of possible future litigation, will doctors feel sufficiently confident in their knowledge of the complexities of epidemiological association to specify smoking as the underlying cause of death?"
Dr James said that he remained of this view. His interest was in being sure that death certificates were fit for use. In the letter he and his co-authors stated that the accuracy of clinical death certification, without validation by necropsy, was known to be suspect. He said that this was a widely held view. There were many papers in the international medical literature which showed that doctors seemed to find accurate formulation of causes of death difficult. It seemed inevitable that doctors would differ over the precise role that smoking might have had in individual cases.
[5.451] Publication of the letter led to Dr James's being approached by solicitors for ITL. He was asked by them to report whether the increase in recorded lung cancer incidence between 1900 and 1950 was real. He reported:
"My opinion is that it cannot be known to what extent the reported rise in incidence was real or artefactual as a result of the changes in medicine and society discussed below.
This opinion has been formed from an extensive review of the scientific literature, my specialist knowledge of death certification, cause of death and autopsy pathology, and my knowledge of the history of medicine."
[5.452] Dr James went on to say that there was a reported rise in lung cancer incidence between 1900 and 1950. Reference was made to MRC 1957 in which, under the sub-heading "The increase in lung cancer" the Medical Research Council stated:
"In its annual report for 1948-50 the Council drew attention to the very great increase that had taken place in the death rate from lung cancer over the previous 25 years. Since that time, the death rate has continued to rise, and in 1955 it reached more than double that recorded only 10 years earlier (388 deaths per million of the population in 1955 compared with 188 in 1945). Among males, the disease is now responsible for approximately 1 in 18 of all deaths. Although the death rate for females is still comparatively low, it also has shown a considerable increase in recent years and the disease is now responsible for 1 in 103 of all female deaths.
Three comments may be made on these figures. In the first place, the trend over the last few years indicates that the incidence has not yet reached its peak. Secondly, the figures are not to be explained as a mere reflection of the introduction and increasing use of improved methods of diagnosis, but must be accepted as representing, in the main, a real rise in the incidence of the disease to an extent which has occurred with no other form of cancer. Thirdly, only a small part of the rise can be attributed to the larger number of older persons now living in the population; in the last ten years the lung cancer death rates among both men and women have risen at all ages from early middle life onwards."
Dr James said that his investigation was concerned with the question whether the rise in lung cancer incidence over the years 1900 to 1950 was real or apparent.
[5.453] In RCP 1962, at p.12, it was stated:
"23. During the past 45 years lung cancer has changed from an infrequent to a major cause of death in many countries. This increase has been most serious in men and women in late middle age, when family and professional responsibilities are at their height. [...] While death rates from lung cancer have been increasing, those from other forms of cancer, and other respiratory diseases have been declining or, like bronchitis, remaining stationary.
24. The experience of chest physicians and surgeons in the past 30 years leaves no doubt in their minds that there has been a very large and real increase in incidence of lung cancer, though some pathologists consider that the disease used to be more common than mortality figures suggest and that much of the increase may be due to improved accuracy of diagnosis on death certificates."
Reference was made to Table II on p.14, which showed average annual numbers of death from various causes during five-year periods from 1916 to 1959 in men and women aged 45 to 64 in England and Wales. Dr James said that the information in the Table had come from death certificates.
[5.454] At the end of the passage quoted above from RCP 1962 there was a reference to Willis 1961, the author of which was a tumour pathologist. In the article he asked the question whether carcinoma of the lung was increasing in frequency, and reviewed lung cancer records before 1900, from 1901 to 1930 and from 1931 to 1960. Although Dr James was taken through much of the text, I propose only to quote from the "Concluding comments on the incidence of carcinoma of the lung" on p.436:
"From the foregoing outline the following main points emerge. In the later decades of last century, pathological records show that bronchial carcinoma was not a rare disease, although it was rarely identified as such by pathologists and was almost unknown to clinicians. In the first three decades of the present century, pathologists became increasingly aware of the characters of bronchial cancers, and the common 'oat-cell' tumours were for the first time clearly recognised as carcinomas and not sarcomas. The fourfold increase in the number of lung-cancer death certificates which took place during this period is readily attributable to these advances in pathological knowledge. The last three decades have witnessed many further advances, not only in pathological knowledge, but also in the establishment of special chest clinics and thoracic surgery units with special diagnostic facilities and techniques - radiology, bronchoscopy and cytological and biopsy studies; and, even more important, clinicians as well as pathologists have become lung-cancer-conscious instead of lung-cancer-ignorant, as most of them were in the earlier period. Under these circumstances it was inevitable that increasing numbers of this clinically misleading and elusive disease should have been detected and correctly identified; and this progressive improvement in diagnosis certainly accounts for a large part of the greatly increased numbers of lung-cancer deaths registered during recent years. Whether or not improved diagnosis accounts for the whole of this increase we shall probably never know, for it is impossible to assess retrospectively just what proportion of clinical and pathological misdiagnoses were made at any given past time, and even now this proportion is still substantial, in spite of our greatly improved diagnostic armamentarium and our awareness of the disease."
Dr James concurred with the content and the conclusions of Willis 1961.
[5.455] Dr James said that the reported rise in lung cancer incidence between 1900 and 1950 was based largely on death certificates and autopsy results; it was supported by doctors' subjective impressions of lung cancer frequency. It could not be known what part of this supposed rise was real because of changes in: the age distribution of the population; the ability to diagnose; the ability to treat; medical knowledge and practice; the role of hospitals; the availability and character of medical care; death certification practice; and medical statistics generally. There was a perception in the early 1950s that lung cancer was rare in 1900 but was common by 1950. Of the evidence that was cited to support this perception, he said that he would address: clinical diagnosis having risen many-fold; autopsy diagnosis having risen many-fold; deaths attributed to lung cancer on death certificates having risen many-fold; and similar findings having been reported world-wide.
[5.456] Dr James said that there was an impression amongst doctors in the late 19th century and early 20th century that lung cancer was a rare disease, but that careful reviews of 19th century literature did not confirm this view and demonstrated that those 19th century physicians familiar with lung cancer were convinced that it was a much more common disease frequently missed. He referred to a series of papers. In Boyd 1886 the author expressed the view that the existence of thoracic cancer was frequently overlooked. In Adler 1912 the author expressed the view that misdiagnosis did occur, with what he believed to be lung cancer being diagnosed as tuberculosis; and that whilst many clinicians might believe lung cancer to be rare, he disagreed. Reference was then made to Weller 1929, in which the author stated, at p.482:
"As is true of all cancer statistics, those dealing with primary carcinoma of the lungs and bronchi are of value only to the extent that the material units have been critically selected. The authenticity of the individual case is the crucial test. Obviously, only those reputed examples of primary carcinoma of the lung are of value on which a complete autopsy has been done. Certainly, too, there must be microscopic verification of each case, if results are to be fully accepted. Routine microscopic examination reveals many errors in the conclusions of even the most experienced gross pathologists."
Finally, in Onuigbo 1971 the author discussed diagnostic difficulties, the influence of preconceived notions, post mortem diagnosis and confusion with tuberculosis among other topics as having a bearing on the recognition by 19th century practitioners that lung cancer was being under-diagnosed. Reference was also made to Horsfall 1966a, which Dr James said was relevant to the extent that it indicated that in the United States the effective treatment of infectious diseases had been such as to help explain why so many more people went on to survive into late middle-age, the age at which cancer was more likely to appear.
[5.457] Dr James said that at the turn of the 20th century lung cancer was widely regarded as being a secondary cancer - having spread from a primary cancer in another organ - rather than originating in the lungs themselves. This was a view disseminated by Rudolf Virchow, a 19th century doctor regarded then as one of the leading lights of medicine and now as the father of modern pathology. Reference was made to Adler 1912 and to Rigdon and Kirchoff 1952. In the latter paper the authors referred to a number of factors which in their view pointed to there being insufficient evidence to prove that the reported increase in lung cancer over the previous fifty years was either apparent or actual. The factors mentioned by them included the questionable authenticity of many death certificates, improved methods in diagnosis, a progressive increase in the number of individuals reaching the age at which cancer might develop and the recognition as primary epithelial tumours of what had previously been classified as metastatic sarcomita.
[5.458] Dr James went on to say that the age distribution of the general population was changing markedly; the group most at risk of lung cancer - those over fifty years old - tripled in number between 1851 and 1921. Without a change in age-specific lung cancer rates, a physician in 1920 would be expected to see three times as many cases, compared with his counterpart in 1850, thus getting a subjective, though false, impression of a real increase. The frequency of diagnosis of any disease depended in part on the availability of medical care. Between 1900 and 1950 the number of people per doctor fell from 1,400 to 900 in England and Wales. This was a phenomenon widely linked with an increase in the diagnosis of cancer in general and lung cancer in particular: Rigdon and Kirchoff 1952 discussed this in relation to various states in the United States. Smithers 1953 said that much of the recorded rise in England and Wales could be attributed for by changes in age distribution and improvement in diagnostic techniques:
"What the clinician so readily attributes to his diagnostic skill the research worker is apt to believe may well be due to the action of a specific carcinogen. Both need to take particular notice of the way their natural bias is apt to lead them and to remember their responsibility to an anxious and ill-informed public when they publish their views on serious problems of fatal disease."
Dr James added that the population receiving medical care was also changing; men received more frequent medical care, probably partly as a result of the National Insurance Acts from 1914 giving benefit to those in work and the Workmen's Compensation Acts from 1918 requiring certain workers to have medical examinations.
[5.459] He went on to say that the discovery of diseases, their frequency and the attribution of death to them were dependent upon hospitalisation, because it had always been through the hospital system that advances in knowledge, practice and provision of facilities were made and disseminated in the literature, and hospitals were where autopsies were performed. From 1900 to 1950 there was considerable development of hospitals with a steady increase in admissions throughout that period and a change in the "type" of patient. In 1900 hospitals had not yet shaken off their role as "refuges for the sick poor" which selected their patients very differently from the "centres of specialised diagnosis and treatment" which they were to become by 1950. Not only would hospitalised patients in 1900 reflect the profile of diseases prevalent in lower socio-economic classes, but also the low life expectancy in these classes would further distort lung cancer incidence, since lung cancer's peak incidence was in later life.
[5.460] Lung cancer, he continued, was extremely difficult to diagnose in life in the absence of modern diagnostic techniques, or at death in the absence of an autopsy. It was in the hospitals that new diagnostic techniques became available and awareness of lung cancer first increased. It was also in the hospitals that autopsies could be performed. Since access to medical care was restricted by limited availability, the number of admissions to hospital was an important determinant of the frequency of diagnosis of lung cancer. During the period from 1900 to 1950 the nature of hospitals and the number of patients being treated in them changed considerably. For example, between 1894 and 1928 the number of admissions to sixteen British teaching hospitals went up by a factor of 2.5; this reflected the increase in the accessibility of medical care. Moreover the proportion of men to women was about 2 to 1 which would favour an unequal distribution of lung cancer diagnosis between the sexes, exaggerated by the 25% lower female mortality (since there would be a relatively greater proportion of autopsies in men).
[5.461] The ability to diagnose and treat lung cancer was extremely limited as of 1900 but had increased massively by 1950. As of 1900 doctors had only the basic skills of history taking, inspection, palpation, auscultation and percussion. There were no x-rays, no bronchoscopy and bronchography, no effective sputum cytology or biopsy, no exploratory thoracotomy and little medical specialisation. There was little awareness of lung cancer, partly because it was widely held to be rare. There was no effective treatment beyond palliation with morphine and therefore doctors had no incentive to diagnose it. By 1950, radiology had become a speciality in its own right during the 1930s and chest x-ray was a simple, effective and available test to discriminate between different chest diseases and a reliable means of diagnosing lung cancer. It was used in mass screening programmes to detect tuberculosis and this facilitated lung cancer diagnosis. Simple chest x-rays were supplemented during the 1930s with lipiodol bronchography which outlined obstruction of the airways by tumour. Bronchoscopy had been well validated but little used by 1930 but its use then took off, very rapidly supplemented by the use of biopsy for precise diagnosis. Sputum cytology using the Papanicolaou method was introduced in the 1940s and was an effective and simple means of diagnosis. Sputum examination was also used for tuberculosis; the failure to detect tubercle bacilli in a case of chest disease was an indication for further investigation. By 1925 there were only about eighty-five case reports of exploratory thoracotomy in the literature but, when directed by radiology and coupled with histological examination of biopsies, it was such a successful means of diagnosis that during the 1930s it became a commonly used procedure. The development of antibiotics in the 1930s, and the resulting improved cure rates for pneumonia, allowed the detection of underlying diseases (a frequent presentation of lung cancer was pneumonia caused by obstruction of an airway). In previous times the patient might have died of pneumonia without the realisation that he or she had an underlying lung cancer. Histopathology had become a speciality and considerable advances were made in tissue diagnosis with repercussions for biopsy and autopsy diagnosis. More accurate and useful classifications for disease were produced; for example, tumours of the mediastinum previously described as sarcomas were recognised as lung cancers. Treatment options for lung cancer prior to 1930 were virtually non-existent. There had been the occasional attempt at lobectomy or pneumonectomy from about 1908 but these were unsuccessful until Sauerbruch performed a lobectomy in 1926, and Graham a pneumonectomy in 1933, by which time radiation was being used with some success for palliation. The realisation by the mid-1930s that cure was possible convinced doctors of a need to diagnose the disease and the unsuitability of the majority of cases for surgical treatment due to their advanced disease led to calls for earlier diagnosis. The availability of a treatment only for early disease encouraged clinicians to rule out lung cancer as a possible diagnosis. This was facilitated by the use of mass radiography (primarily as a means of diagnosing tuberculosis but having a "spin-off" of lung cancer diagnosis).
[5.462] The propositions contained in the previous paragraph were vouched by references in the course of Dr James's evidence. Reference was made to Rosenblatt 1969, in which the author discussed the effect of improved diagnostic techniques in hospitals in the United States in the diagnosis of lung cancer. He related improvements in diagnosis to the progressive decline in the rate of increase of the disease, and at p.32 stated:
"It is impossible to reconcile a declining rate of increase in lung cancer mortality with the epidemic concept. The deceleration of the rate of increase implies that there has been a relatively fixed prevalence of the disease in the population and that expansion of diagnostic facilities has resulted in the recognition of a greater number of cases each year, leaving proportionately fewer cases undiagnosed. The trend toward progressive decline of the rate of increase is crystal clear and will ultimately produce a standardization of the age-adjusted death rate of lung cancer. Inasmuch as lung cancer is a disease of older individuals, the total number of cases may continue to rise because of the increasing longevity among all groups of the population."
[5.463] Dr James went on to say that the effects of these changes individually and collectively were not measured (and, probably, were not measurable) at the times they occurred and could not be assessed with any accuracy in retrospect. It was possible, however, to match the approximate time at which these changes had practical effect (not necessarily when they were first developed or introduced) to the recorded increase in lung cancer. He referred to a graph in Steiner 1953. In this, a comparison was made between the annual number of deaths attributed to lung cancer in males in England and Wales during the period from 1900 to 1950 and the methods available for the diagnosis of lung cancer during life in that period. The methods in question were history and physical examination, which Dr James said were all that was available in about 1900, followed by introduction of the examination of pleural fluid and the biopsy of occasional distant metastases between 1905 and 1910. Between 1920 and 1930 the use of x-rays became more general, and during that period aspiration biopsy was introduced, followed by bronchoscopic biopsy in about 1930. In about 1935 lobectomy and pneumonectomy were introduced, followed by cytological examination of sputum and bronchial washings in about 1940 and exploratory thoracotomy in about 1945. Dr James said that this graph showed that the changes which would be expected to result in a greater number of diagnoses of lung cancer occurred in conjunction with the steepest parts of the incidence curve, thereby presenting a satisfactory explanation for the increase in recorded incidence without the need for a "real" change in incidence having happened. He said that he broadly agreed with Steiner's model.
[5.464] In support of the foregoing views, Dr James referred to various papers. In Johnson et al. 1949 the authors had reviewed the records of 384 patients operated upon by one of them, and stated:
"Our experience with thoracic disease in recent years has convinced us that exploratory thoracotomy carried out as a diagnostic measure is a valuable and important procedure that is not appreciated generally but which can contribute greatly to improvement in results of treatment of pulmonary disease."
In Barnard 1926 the author demonstrated the advantages of the use of microscopic among other techniques in support of his view that tumours then usually called "oat-celled sarcomata" of the mediastinum were primarily carcinomas of the lung.
[5.465] Dr James said that it was not only the improvement in diagnosis that affected mortality statistics and thereby the incidence of deaths attributed to lung cancer. There were changes in collection of mortality data involving improved completion of the medical certificate of cause of death and improved coding of the cause of death. A clinical diagnosis was thus more effectively and reliably translated into an accurate mortality statistic. Four revisions of the International Classification of Disease up to 1950 had progressively refined the classification of lung cancer, and medical statistics had developed as a speciality with increasingly detailed data being provided by the Registrar-General's reports. Death certificates and the quality of diagnoses in general showed improvements in precision through the late 19th century and the first half of the 20th century. The precise degree to which improved death certificate diagnosis altered disease incidences between 1900 and 1950 was difficult to determine, but it was instructive to note that in a study of autopsies carried out in a Japanese city between 1961 and 1987 improvement in the death certificate diagnoses of various cancers was not only still ongoing but was responsible for 60% of the recorded rise in incidence of those cancers.
[5.466] It had been recognised since the 19th century that lung cancer might be missed because physicians misdiagnosed it as one of the other, far more common, respiratory diseases. In the view of Adler 1912, lung cancer was common and a doctor might come across it any day in his practice but, for the symptoms of lung cancer, tuberculosis was a comfortable and satisfactory diagnosis. Even in 1950 there were nearly 90,000 people a year in England and Wales dying with respiratory symptoms, many times more than those who were dying of lung cancer. It was well-recognised that the symptoms of lung cancer were non-specific and commonly present in other more frequently diagnosed diseases; this was an explanation for why nearly all lung cancer diagnosed prior to the First World War was diagnosed in the post mortem room. In various studies looking at autopsy-diagnosed lung cancer it was discovered that between 20% and 36% had been misdiagnosed in life as tuberculosis, 15% as syphilis, 18% as pneumonia and 23% as cancer of another organ.
[5.467] In support of these views reference was made to the following papers. Wells 1923 concluded that the necropsy statistics under discussion "show convincingly that we have at the present time no reliable statistics relative to the frequency of cancer as a cause of death." Anonymous 1950 stated:
"One of the first questions likely to be asked in any discussion of the aetiology of cancer of the lung is whether there has been any real increase in the disease since, say, the beginning of this [the 20th] century. The answer is that we do not know, because there are no figures either here or in any other country on which the actual death-rate from intrathoracic cancer of forty years ago can be compared with that of today."
[5.468] Gilliam 1955 discussed the trends of mortality attributed to carcinoma of the lung in the United States between 1914 and 1950 and the possible effects of the faulty certification of deaths to other respiratory diseases, particularly tuberculosis. The author said that there was abundant evidence attesting to the erroneous diagnosis of lung cancer as tuberculosis. In addition, use of antibiotics in recent years had permitted the survival until correct diagnosis of an unknown number of patients who formerly would have died of pneumonia or some other respiratory cause before discovery of the underlying cancer of the lung. From published data there was no ready way to estimate directly the frequency with which this might have occurred and thus no way to determine precisely how well the recorded mortality measured the magnitude of the true increase. Looking at tuberculosis on its own, Gilliam noted that a 10% error in diagnosis in 1914 decreasing to a 2% error in 1950 would account for a reduction in the apparent increase in lung cancer from 26.4-fold to 2.2-fold in white men and from 6.7-fold to 0.8-fold in white women. Furthermore, if 2% of cases diagnosed as respiratory diseases other than lung cancer between 1914 and 1950 were, in fact, really cases of lung cancer, then the apparent increase in lung cancer incidence, between those dates, in white men, would be reduced from 26.4-fold to 4.5-fold and a 6.7-fold increase in white women would be reduced to 1.3-fold.
[5.469] It was, Dr James said, in that context, particularly interesting that Auerbach 1949 described a series of fifty lung cancer autopsies derived from a tuberculosis hospital having been mistakenly diagnosed. During the time that these lung cancer cases were being accrued, more than 2,000 tuberculosis autopsies were performed, indicating an error rate in 1948 (when sanatoria and specialised means of diagnosis were commonly available) of approximately 2.5%. Tuberculosis was not the only disease with which lung cancer might have been confused; the list of other possible diseases was long, encompassing many acute and chronic respiratory disorders together with malignancies of other sites. The low level of histological confirmation of diagnosis (38%) or autopsy (11%) in 3,280 cases of lung cancer in 1945 and 1946 discussed in one study indicated a potential for such misdiagnoses to cause a significant distortion of the figures. Since these figures related to a time when the major improvements in diagnosis had occurred, it suggested that earlier in the century the misdiagnosis rate might have been considerably higher.
[5.470] Examination of series of autopsies might be used as an alternative way of looking at the changing incidence of a disease and might also be used to act as a means of checking death certificate data. A problem with such series was that they represented a highly selected population - people who had consulted a doctor, been admitted to hospital, died in hospital, and been selected for autopsy. Where series had been done in a single location and where the autopsy rate was high, some of this bias was cancelled out. Bonser 1929 could find no increase in the proportion of lung cancer cases to all cancer cases in the period from 1891 to 1927 in Leeds with an autopsy rate of 83.6%. Passey and Holmes 1935 found no evidence of an increase in lung cancer (when corrected for an increase in hospital admissions) in twelve out of sixteen teaching hospitals in the United Kingdom between 1894 and 1923 and in eight out of sixteen up to 1928. In these two studies the absolute numbers had increased, and this accounted for the subjective impression amongst physicians and pathologists of a real increase in incidence. Jaffe 1934, in large series from Vienna (1915-18) and Chicago (1927-33) was unable to demonstrate any change in incidence of lung cancer (expressed as a percentage of all cancer), there being 11.47% in the first series and 10.73% in the second. Steiner 1953 reviewed the work of others and could find no evidence that the incidence at autopsy of lung cancer was out of proportion to other cancers and concluded that it was not possible to tell if any real increase had occurred. Smithers 1953 stated:
"Even to-day, however, the accuracy of the diagnosis of bronchial carcinoma in the country as a whole and the reliability of death certification from this cause must still be under grave suspicion."
[5.471] Dr James said that autopsy series might also be used to assess the accuracy of clinical diagnosis (from which, for the most part, death certificate data were based). Those series in which such a comparison of clinical diagnosis with autopsy diagnosis was possible indicated a false negative rate (i.e., clinical diagnosis = not lung cancer; autopsy diagnosis = lung cancer) of 97% in 1904, 91% in 1923 and 42% in 1948. The average figure for the period 1930 to 1950 would be about 30%. During the later part of the period from 1900 to 1950 this false negative rate would have been off-set to a large extent by the false positive rate (i.e. clinical diagnosis = lung cancer; autopsy diagnosis = other than lung cancer) but around 1900 there was no evidence of false positives (it was not a diagnosis often made). It should be apparent that if, in 1900 to 1920, only 3% to 9% of cases from lung cancer were attributed to lung cancer on a death certificate, compared with well in excess of 70% between 1930 and 1950, the validity of the trend was seriously questioned. (References were made to Smithers 1953, Wells 1923 and Willis 1948, among other papers, in support of the foregoing figures.)
[5.472] Dr James concluded that the death certificate data constituting the reported rise in lung cancer incidence between 1900 and 1950 were not reliable. Widespread, detailed knowledge and awareness of lung cancer developed hand-in-hand with effective and widespread means of diagnosis and treatment options. The times at which these developments became significant in lung cancer diagnosis (as opposed to their introduction) were strongly associated with the steepest part of the graph and a reduction in the autopsy-derived false negative rate (indicating better diagnosis and fewer missed cases). The earlier part of the graph, less associated with diagnostic change, rose only slowly (and equally, male and female), at a time when the false negative rate was extremely high, suggesting massive under-diagnosis.
[5.473] Dr James added that many scientists and major journals were cautious about drawing conclusions from the incidence data presented by 1950. Most of the data described by him were available to them and most of the problems in differentiating a real increase from an apparent increase had been identified. Many found no reason to suppose that the increase was real. Reference was made to Anonymous 1942 and Anonymous 1950. Fried 1948 concluded at p.12:
"The author is of the opinion that the more frequent occurrence of this disease [cancer, particularly cancer of the lung] today as compared with the past, can be explained on the basis of the following factors: (1) Progress in medicine in the past five decades resulting in improved methods of diagnosis; (2) Progress in public health resulting in increased longevity (more people reach the 'cancer age'); (3) Vastly developed means of hospitalization; (4) Rise in the social and economic levels leading to more people seeking medical advice; (5) Increased attention to the disease. The increase, therefore, is in all likelihood more apparent than real."
Willis 1948 expressed the opinion that it was not possible either to affirm or deny that there had been a real increase of carcinoma of the lung. Auerbach 1949 said:
"A review of the literature makes it appear that much of the increase in bronchial carcinoma is apparent. How much is real will probably never been known."
Dr James stated:
"In conclusion it is my opinion that it cannot be determined how much of the apparent rise in lung cancer mortality between 1900 and 1950 was real because of changes in population; in diagnosis and treatment; in medical knowledge and practice; in the role of hospitals; in the availability of medical care and in death certification and general statistical methods."
[5.474] Referring to RCP 1962, para.24, in which it was stated that the experience of chest physicians and surgeons in the previous thirty years left no doubt in the authors' minds that there had been a very large and real increase in the incidence of lung cancer, Dr James said that in his review of the literature he had found no pathologist who did not consider that the disease used to be more common than mortality figures suggested, and that it was not difficult to see why cancer of the lung should have become so much more frequently diagnosed. Lung cancer was a peculiar disease and was idiosyncratic amongst other cancers in terms of its discovery, presentation, diagnosis and treatment and its impression upon physicians and surgeons.
Cross-examination of Dr Deryk James
[5.475] In cross-examination Dr James said that he was an occasional smoker. His only contact with ITL had been through lawyers acting on their behalf. He understood that his remit was to look at what measures there were of the incidence of lung cancer at the beginning of the period from 1900 to 1950 and through that period, in order to see if it could be known what the incidence really was and, therefore, whether there was a rise of any degree. One of the problems was to put a figure on how much, if any, the rise was.
[5.476] Asked about the statement "Smoking kills" on modern cigarette packets, he accepted the message for what it was, a public health warning, but not that smoking directly killed, because it was neither necessary nor sufficient to kill someone. For that reason he said that smoking did not cause lung cancer. It depended in part on the interpretation of the word "cause".
[5.477] Dr James agreed that there were many scientists and doctors who considered that during the period in question there had been a real rise in rates of lung cancer. There were statements to that effect in some of the papers referred to by him in his evidence-in-chief, for example Passey and Holmes 1935, Smithers 1953, Rosenblatt 1969 and Auerbach 1949. When asked "so there was a lot of material, a lot of doctors and a lot of scientists saying that the increase was real?", he answered in the affirmative.
[5.478] Mr McEachran asked Dr James about RCP 1962, which contained statistics showing a "massive increase" in lung cancer in men during the period in question, by contrast with a much smaller increase in other cancers. Dr James said that in 1950 the reporting of lung cancer was not as good as it was now, but much more to be relied upon than in 1900. He regarded the figures for 1946 to 1950 as still being an underestimate, but probably not a significant one. The figures for the next decade were reasonably accurate in the sense that when "cancer of the lung" was written on a death certificate, that was likely to be a correct diagnosis. But other cases might have been missed. Figures given in RCP 1971 for lung cancer deaths in the 1960s were likely to be reasonably accurate, probably within perhaps 20% or 30%, judging by the overall accuracy of death certificates in that period, but Dr James had not considered the period after 1950 in detail. The figures given in RCP 2000 for the incidence of lung cancer in the 1990s were likely to be reasonably accurate. Asked about the statement in IARC 1986 p.203 that about 94% of deaths attributed to lung cancer were confirmed by hospital diagnosis in a study based on the Third National Cancer Survey in the USA, Dr James said that lung cancer was a diagnosis that was made virtually always in hospital and always had been, so this was a completely unsurprising statement and did not really mean anything about the accuracy of diagnosis.
[5.479] Dr James was asked to comment on Figure 4.1 at p.49 of RCP 1971. This was a graph showing death rates from lung cancer, other forms of cancer, tuberculosis of the lung and bronchitis in men aged 45 to 64 from 1916 to 1965. He commented that this was the same as Figure 7 in RCP 1962, p.15, with the addition of years after 1959 and a line for "all cancer". He said that if there had been an "all cancer" line in the earlier graph, it would have been flat apart from a slight rise at the end, occurring around 1950. He regarded the whole statistics as flawed and he looked at RCP 1962 with distrust, because it was talking about rises in lung cancer incidence of up to about fifty-fold. That made him immediately think that people like Bonser, who were doing autopsy series from around 1900 at an autopsy rate of 80% to 90%, so that they were very representative samples from one place, were finding 1% of autopsies were lung cancer and about 8% of all cancer was lung cancer. If this had gone up fifty-fold, they should have been reporting every other autopsy as lung cancer and a massive proportion of cancer as lung cancer, and this clearly had not happened. So something was very wrong, and this graph showed what it was. This was that as lung cancer appeared to increase, diagnoses of other forms of cancer were decreasing, because there were misdiagnoses which were being corrected. By 1950 to 1960 the correct diagnoses were "kicking in" and the graph was then altering. But that occurred very much later than many people in the 1950s were saying. Lung cancer was a disease that was poorly understood in the 19th century and by 1968 to 1970 was very common. There were an enormous number of changes in society and in medicine between those dates, so working out when the real increase started was immensely difficult. The graph in RCP 1971 showed strong support for the view that the increase did not really start until about 1950, but the data were poor, especially the earlier data, and we were never really going to know. There might have been real increases prior to 1950. The fact that there had been increases since 1950 did not demonstrate that there had been increases prior to that.
[5.480] Professor Idle, aged 53, was Professor in Medicine and Molecular Biology at the Institute for Cancer Research and Molecular Biology, Medical Technical Research Centre, Norwegian University of Science and Technology, Trondheim. He had held that post since 1995. He was awarded the degree of BSc in applied chemistry in 1972 and the degree of BSc with First Class Honours in Medicinal Chemistry in 1973. He was awarded a PhD in biochemistry at St Mary's Hospital Medical School, University of London, in 1976. He was a Chartered Chemist and became a Fellow of the Royal Society of Chemistry in 1987. He was a Chartered Biologist and became a Fellow of the Institute of Biology in 1999. He became a European Chemist in 2000, meaning that his qualifications as a chemist in the United Kingdom were recognised by the European Union. He became a European Professional Biologist in 2000 also, his qualifications in biology in the United Kingdom thereby being recognised across the European Union.
[5.481] Professor Idle was Lecturer in Biochemistry in 1976, Lecturer in Biochemical Pharmacology from 1976 to 1983, Wellcome Trust Senior Lecturer from 1983 to 1985 and Reader in Pharmacogenetics from 1985 to 1988, all at St Mary's Hospital Medical School. He was Professor of Pharmacogenetics from 1988 to 1995, Chairman of the Department of Pharmacological Sciences from 1992 to 1995 and Head of the School of Clinical Medical Sciences, Faculty of Medicine, all at the University of Newcastle.
[5.482] He was a Visiting Scientist with Curtis C Harris at the Laboratory of Human Carcinogenesis, National Institutes of Health (NIH) and National Cancer Institute (NCI), USA, in 1982. He described carcinogenesis as the process by which a normal cell becomes a cancer cell, in lay terms, but more correctly as the process by which an epithelial cell becomes a carcinoma. He was World Health Organization (WHO) Special Adviser and WHO Committee Chairman (genetic predisposition to toxic effects of chemicals) in 1989. He said that the remit of this Committee was to read, discuss and digest a large amount of data over the period of about one week, as a group, and to produce a report for the WHO, which went out to all Member Governments on the United Nations on the subject in question. He was a member of the Cancer Research Advancement Board of the Irish Cancer Society from 1986 to 1991 and from 1995 to 1996. He said that this was the grant-giving arm of a charity, the Irish Cancer Society, and its role was to consider grant applications from universities around Ireland and to make recommendations for the award of grant funding. He was a member of the Scientific Advisory Board of The Council for Tobacco Research - USA, Inc., New York from 1986 to 1991. He said that the function of this Board was to consider grant applications from scientists and to score them on the basis of their scientific merit and their relevance to smoking and health, and to make recommendations regarding funding. The scoring system was identical to that used by studies sections of the NIH in the USA. When he joined the Board it was Chaired by Leon Jacobson who had been the Dean of Medicine and Professor of Medicine at the University of Chicago and was a Member of the National Academy of Sciences, USA, which was roughly equivalent to the Royal Society in the United Kingdom. Several other members of the Board were members of the National Academy of Sciences; they were all eminent scientists and/or eminent academics. They were drawn from all over the United States, plus one from Canada.
[5.483] Professor Idle was Founding Editor and Editor-in-Chief of Pharmacogenetics from 1991 to 1998. He said that this journal published original scientific articles and some review articles and occasionally correspondence on the subject of pharmacogenetics, mainly human pharmacogenetics, and all the issues that surrounded the role of genes and the response of individuals to chemicals and drugs. He was on the editorial boards of two other journals. He was Consultant in Medical Genetics and Head of Department at the Regional Teaching Hospital, Trondheim from 1996 to 1998. He was Visiting Professor at the University of Bern, Switzerland, with effect from 1 January 2003. He was Consultant in Metabolism, Laboratory of Metabolism, NCI, NIH from 2002. Additional society memberships were of the British Pharmacological Society, the American Association for the Advancement of Science, the American Society for Pharmacology and Experimental Therapeutics, the American Society for Biochemistry and Molecular Biology and the American Chemical Society. His CV listed numerous refereed papers, book chapters and review articles, communications, abstracts, letters and editorials.
[5.484] Professor Idle was asked by ITL's solicitors to give an opinion based upon his own area of scientific expertise in answer to the following five questions:
"(1) In your opinion was it 'almost universally accepted by scientists' in 1964 [as averred on behalf of Mrs McTear] that cigarette smoking can cause lung cancer?
(2) In your opinion had cigarette smoking been established as a cause of lung cancer as at 1964?
(3) In your opinion had cigarette smoking been established as a cause of lung cancer as at 1971?
(4) In your opinion had cigarette smoking been established as a cause of lung cancer as at September 2003?
(5) In your opinion did cigarette smoking cause Mr McTear's lung cancer?"
He said that his report was structured into a number of time periods: the period up to 1964, the period from 1965 to 1971 and the period from 1972 until the completion of the report in September 2003. In constructing the report he had reviewed the scientific evidence involving the various theories of the causation of cancer in general, and of lung cancer in particular; the toxicological experiments that comprised the administration to experimental animals of cigarette smoke or its derived condensates; and the chemistry of cigarette smoke and its derived condensates. He took a chronological and historical perspective of the published laboratory data that he judged to be of relevance to answering the above questions. He reviewed approximately 1,600 original scientific articles, books, reports and scientific communications. On and off, this took about six years. He concluded by answering the first four questions in the negative and, in response to the fifth question, stated that it could not be determined whether or not cigarette smoking caused Mr McTear's lung cancer.
[5.485] In Doll 1997H Sir Richard Doll said at pp.25-26:
"In the three subsequent decades [i.e. subsequent to the 1950s], cigarette smoking has been found to be positively associated with nearly 50 diseases or causes of death and to be negatively associated with eight or nine. In a few instances the associations are due to confounding with other factors, but the great majority arise because tobacco smoke is a contributory cause. Pace Berkson, this is not surprising; not only because of the complexity of tobacco smoke, but also because many of the diseases are different clinical manifestations of common processes, such as DNA damage, vascular occlusion, and damage to small airways."
Professor Idle said that the complexity of tobacco smoke was something he had investigated in the course of researching for his report, as was DNA damage, which indeed he dealt with in his normal academic working life.
[5.486] RCP 2000, Table 1.2, p.17, gave figures for deaths from disease estimated to be caused by smoking in the United Kingdom in 1997. Diseases caused in part by smoking were said to include cancer at seven specified sites and also in unspecified sites, and myeloid leukaemia. Among other diseases were ischaemic heart disease, cerebrovascular disease, aortic aneurysm, myocardial degeneration and atherosclerosis, in the category of circulatory diseases; and ulcer of the stomach or duodenum, in the category of digestive diseases. Professor Idle said that in preparing his report he had given some thought to the biological plausibility of cigarette smoking being responsible for some of this range of diseases.
[5.487] In Berkson 1959, the author stated at p.445:
"For myself, I find it quite incredible that smoking should cause all these diseases. It appears to me that some other explanation must be formulated for the multiple statistical associations found with so wide a variety of categories of disease. And if we are not crassly to violate the principle of Occam's razor, we should not attribute to each separate association a radically different explanation.
One explanation is that the associations have a constitutional basis. This hypothesis has been formulated - and rejected - by Doll and Hill in the following terms:
'... it has been suggested that constitutional and psychological factors might have such an effect.... that persons of a certain "make up" are peculiarly liable to lung cancer and to smoke. We know of no published evidence to this effect.'"
To the quotation from Doll and Hill Berkson added a footnote, also at p.445:
"Cancer is a biologic, not a statistical, problem. Statistics can soundly play an ancillary role in its elucidation. But if biologists permit statisticians to become the arbiters of biologic questions, scientific disaster is inevitable."
(The expression "Occam's razor" is defined in the Oxford English Dictionary as: "The leading principle of the nominalism of William of Occam, that for purposes of explanation things not known to exist should not, unless it is absolutely necessary, be postulated as existing; usually called the Law of Parsimony".) Professor Idle said that during the course of his researches he had read work that indicated that "persons of a certain make up" were "peculiarly liable to lung cancer". Berkson's view was strong and in some respects surprising, because Berkson himself was a very senior statistician. Here he was at least being even-handed, rather than protecting his own constituency. The content was absolutely right: statistics often pointed the way but, as a laboratory scientist, Professor Idle would, he said, stick absolutely to the view that it was necessary to carry out, with the scientific method, scientific experiments in order to demonstrate proof.
[5.488] In USSG 1964 at p.20, under the subheading "Criteria of the epidemiologic method", the authors of the report stated:
"Clinical, pathological and experimental evidence was thoroughly considered and often served to suggest an hypothesis or confirm or contradict other findings. When coupled with the other data, results from the epidemiologic studies can provide the basis upon which judgments of causality may be made.
In carrying out studies through the use of the epidemiologic method, many factors, variables, and results of investigations must be considered to determine first whether an association actually exists between an attribute or agent and a disease. Judgment on this point is based upon indirect and direct measures of the suggested association. If it be shown that an association exists, then the question is asked: 'Does the association have a causal significance?'
Statistical methods cannot establish proof of a causal relationship in an association. The causal significance of an association is a matter of judgment which goes beyond any statement of statistical probability."
Professor Idle said, under reference to this passage, that epidemiological studies could point the way, other data could be brought into the equation and, taking all of that together, a judgment could be made. The scientific method entailed, first of all, making an observation and then constructing a hypothesis or series of hypotheses that might explain it. A hypothesis might and should be used to predict other phenomena that would occur out there in the world or in a laboratory, and should also be able to predict a quantitative outcome in a particular experiment. So it should have a qualitative component and, if appropriate, also a quantitative process. Then the process was set in train to carry out experiments to test if the hypothesis was correct or not. It was usually necessary to run through a fair number of iterations of the third and fourth points until everything fitted into place, until all observations fitted the final hypothesis. So the hypothesis was something which had to be continually modified until all the observations could be explained by it.
[5.489]This led to the creation ultimately, over a period of time, and in collaboration with other scientists, of a model for the world and how it operated. In carrying out experiments the object was both to establish that the hypothesis was correct and to establish that it was wrong. The final hypothesis must explain all the positives and the negatives. He agreed that "statistical methods cannot establish proof of a causal relationship in an association".
[5.490] Reference was next made to Hill 1966. Professor Idle described the author, Professor Bradford Hill, as a very eminent statistician. He was asked to comment on passages in Chapter XXIV, entitled "Statistical evidence and inference". At p.302 Hill wrote:
"The aim of this book has been to make clear in non-mathematical terms some of the techniques that the statistician employs in presenting and in interpreting figures. Much attention has been paid to two basic problems:-
(1) The 'significance,' or reliability in the narrow sense, of a difference which has been observed between two sets of figures - be those figures averages, measures of variability, proportions or distributions over a series of groups.
(2) The inferences that can be drawn from a difference which we are satisfied is not likely to be due to chance."
Professor Idle said that this passage set out a two-stage process: first, to establish whether there was statistical significance in an association; and, secondly, to determine the nature of the relationship.
[5.491] After discussing means of determining whether an observation was statistically significant, Hill wrote, at p.306:
"Faced then with a clear and significant association between some form of sickness and some feature of the environment, what ought we specifically to consider in drawing conclusions about the nature of the relationship - association or causation?"
He then went on to discuss the strength of association, consistency, specificity, the relationship in time, the biological gradient, biological plausibility, coherence of the evidence, the experiment and reasoning by analogy. At pp.313 to 314 he wrote:
"This interpretation of statistical data turns, it should be seen, not so much on technical methods of analysis but on the application of common sense to figures and on elementary rules of logic. The common errors [...] are not due to an absence of knowledge of specialised statistical methods or of mathematical training, but usually to the tendency of workers to accept figures at their face value without considering closely the various factors influencing them - without asking themselves at every turn 'what is at the back of these figures? what factors may be responsible for this value? in what possible way could these differences have arisen?' That is constantly the crux of the matter. Group A is compared with Group B and a difference in some characteristic is observed. It is known that Group A differed from Group B in one particular way - e.g. in treatment. It is therefore concluded too readily that the difference observed is the result of the treatment. To reject that conclusion in the absence of a full discussion of the data is not merely an example of armchair criticism or of the unbounded scepticism of the statistician. Where, as in all statistical work, our results may be due to more than one influence, there can be no excuse for ignoring that fact. And it has been said with truth that the more anxious we are to prove that a difference between groups is the result of some particular action we have taken or observed, the more exhaustive should be our search for an alternative and equally reasonable explanation of how that difference has arisen."
Professor Idle said that he agreed with this passage. It was still as valid today as it was when Professor Bradford Hill wrote. It was a tremendous problem for scientists to avoid becoming slaves of numbers without any logical interpretation. The discussion in the passage about Group A and Group B was equally applicable where the difference was in exposure. The conclusion of the passage quoted was a very important aspect of the scientific method.
[5.492] Professor Idle then turned to the biology of cancer. He said that it was quintessentially a biological entity. It was a general term for more than 100 malignant diseases that were characterised by uncontrolled, abnormal growth of cells and the spread of these cells locally and through the bloodstream and lymphatic system to other parts of the body, by a process known as metastasis. The understanding today of what was this biological entity, in its nature, in its behaviour, and in its ferocity, rested upon a mountain of both related and unrelated observations collected across the decades, in a wide range of disciplines, together with a broad range of clinical specialities such as oncology, surgery, radiotherapy, internal medicine, paediatrics, endocrinology, haematology, pathology and numerous others. These different disciplines were very disparate, both intellectually and in many respects geographically, but in almost every academic environment in a university or hospital there was a group working on cancer of some type or other, and they came together through the scientific literature and through scientific conferences, which were tremendously important in bringing minds together.
[5.493] The term "lung cancer", he said, also represented a group of diseases, those malignancies found in the lung. They can be either primary lung cancers, those originating in lung tissues, or secondary lung cancers, those resulting from metastasis of a tumour originating in a different organ, such as the colon. Primary lung cancers were sub-divided upon the basis of both their characteristics and their cellular origins. The principal lung cancer types were squamous cell carcinoma, adnocarcinoma (both currently grouped together as "non-small cell lung cancer") and small cell carcinoma. There were other more minor categories. A small cell carcinoma was so called because the cell had shrunk and almost all of it was made up of the nucleus. Among the minor categories was benign adenoma, which grew out of glandular tissues, but without the characteristics of a malignant tumour; it would stop growing by itself and often regress and disappear, and it did not metastasise. Benign adenomas frequently arose in experimental animals, yet rarely, if ever, in humans.
[5.494] One could not understand lung cancer without first comprehending cancer as a biological phenomenon. Any consideration of the biology of lung cancer must, therefore, be built on a foundation of cancer biology in general.
[5.495] In the period prior to 1965 there were a number of competing theories regarding the origins of cancer. These theories included the ideas that cancer arose after chronic irritation, or from embryonic tissue, from a filterable invisible virus, from macroparasites, from perturbations in cellular metabolism, from ionising radiation, or from chemicals in the environment. Each of these theories was offered as an explanation of cancer and the scientific literature of the day was replete with references to "the mechanism" of cancer causation. The competition between these theories was quite vigorous at times.
[5.496] According to the scientific method, progress from hypothesis to substantiation should be made through a series of rigorously controlled and reproducible experiments at the laboratory bench. In an effort to determine whether or not one or more of the competing theories of cancer causation could evolve to a position of pre-eminence, carefully planned and executed studies had needed to be undertaken in experimental animals. But the scientific method had not always been rigorously applied, he said. In the 19th century, the use of animals in a laboratory setting had become commonplace. For example, Louis Pasteur employed sheep, cattle, guinea pigs and rabbits.
[5.497] The idea that cancer was caused by chronic irritation, articulated by Virchow in the mid 19th century, dominated thinking for over half a century. By the end of 1964 irritation was still seen to play a role in carcinogenesis. Professor Idle made it clear that throughout his report and his evidence, the terms "carcinogen", "carcinogenic" and "carcinogenesis" were used exclusively in the context of appearance of carcinomas in experimental animals. Carcinomas were malignant new growths arising in epithelia, those layers of cells that cover the internal and external surfaces of the body, including the lining of vessels and other small cavities.
[5.498] Professor Idle next mentioned the virus theory. He said that the first clues that a virus could be an infectious cause of cancer came from the influenza epidemic of 1918. Winternitz et al. 1920 drew attention to the proliferative changes in the lung that resembled those of malignant neoplasms in survivors of this epidemic based on post mortem examinations of a large number of cases of influenza. They reported that in many cases the epithelial proliferation invaded the surrounding lung tissue and a typical, histological picture resulted of an infiltrating, malignant, epithelial neoplasm. They observed in abundance, in the injured bronchiolar lining, mitotic figures, a term used to describe chromosomes that were visible under a microscope when a cell was dividing and thus an indication of cell division.
[5.499] Viral carcinogenesis had been extensively studied in animals. Under the leadership of Dr Clarence Cook Little, staff at the Jackson Laboratories in Maine, USA, demonstrated that a virus was the cause of the high incidence of mammary tumours in a breed of mice and its transmission from mother to daughter in the mouse. Rabson and colleagues successfully produced "epidermoid carcinoma" (squamous cell carcinoma) in Syrian golden hamsters after inoculation with polyoma virus. Viruses were discovered that could cause malignancies in many other species, although these were largely of haematopoietic (e.g. lymphomas) or connective tissue (e.g. sarcomas) origin, rather than arising from epithelial tissues. In 1964, Kotin reported the results of work on a strain of mice that had been administered a combination of influenza virus and city smog extract in the form of ozonised gasoline aerosols. The only group of mice to develop squamous cell carcinomas of the lung were those exposed to both the virus and the smog. By the end of 1964, influenza virus was suspected as a cause of lung cancer in humans and, in combination with smog, had produced squamous cell carcinomas in the mouse lung. Additionally, a number of other viruses had been shown to cause cancer in experimental animals.
[5.500] During the period from 1965 to 1971, the fields of chemical and viral carcinogenesis were beginning to cover less. Workers were beginning to recognise that there might be some common threads in the mechanisms by which viruses and chemicals might cause cancer. Reference was made to various experiments. In one, addition of extracts of city smog to cultures of hamster or rat cells infected with leukaemia virus caused malignant transformation of the cells. In another experiment, mice were treated with the synthetic chemical carcinogen 20-methylcholanthrene. Cells of the resultant tumours displayed properties very similar to certain cancer-causing viruses and it was postulated that the chemical interacted with some endogenous virus present in every mouse cell. This foreshadowed the later discovery of the oncogenes. Horsfall 1966b stated that several thousands of viruses were then known and several hundreds of them could infect human beings. At p.181 Horsfall concluded:
"[T]he cancers of animals do not seem to be different in any significant way from those of human beings. They are of the same types and may occur in the same sites. In some animal species they develop with frequencies similar to those seen in human beings. The cancers of animals are, frequently, attributable to viruses. Indeed, it appears that, in those instances in which the primary incitant of naturally occurring animal cancers has been demonstrated, it has turned out to be a virus.
The number of cancer-inducing viruses and the number of species affected by them are sufficiently large to suggest that, unless one takes a rather forced point of view and insists that cancers of human beings are unique, it would be difficult to defend the idea that viruses do not have anything to do with some human cancers."
Professor Idle said that the views expressed in this passage were still current.
[5.501] Whilst other authors wrote in general terms about how viruses could cause cancer, Dulbecco, writing in 1967, envisaged specific mechanisms which required the interaction of viral genes with host DNA (he subsequently shared a Nobel Prize for these insights). In 1970 independent pivotal discoveries were made of an enzyme within certain tumour virus particles which became known as "reverse transcriptase", which could make DNA copies of viral RNA genes. Prior to these reports it was thought that RNA could only be synthesised from DNA, and not vice versa. This discovery suggested a mechanism by which certain RNA viruses could cause cancer.
[5.502] During this period, researchers were also examining the possible role of viruses specifically in connection with lung cancer. Leuchtenberger and colleagues reported in 1965 that they had found that a strain of influenza virus transformed cells from mouse kidney and bronchus and yielded transplantable tumours. Professor Idle said that one of the hallmarks of a malignant tumour was that it could be transplanted from one animal into another of the same strain and that it would grow and continue to grow there, so the fact that the tumours were transplantable suggested that they were malignant. Mohr and colleagues reported in 1969 that they had isolated a transforming virus-like particle from human alveolar cell carcinoma, suggesting that it was a causative agent. Professor Idle said that the history of viruses as causes of human cancer rested to a particular extent on the detection and then isolation of virus particles in human tumours, which had now been done on a number of different occasions. The isolated particles were grown up in a host cell in the laboratory and these were used to re-infect animals. It was first necessary to see them, and they were usually discovered under the electron microscope in a tumour cell. Staemmler and colleagues reported in 1970 that, using a small amount of a strain of influenza virus, they were able to produce tumours in very high numbers in the bronchi and the alveoli of mice. The tumours appearing early were mostly of the squamous cell type whereas the later ones predominantly showed the features of adenocarcinoma, though often transition to the squamous cell type was observed. By the end of 1971, therefore, the evidence suggested that viruses could cause human cancer and that respiratory virus infection might be a cause of human lung cancer.
[5.503] Defendi and Jensen reported in 1967 what Professor Idle called an interesting reaction between radiation and viruses. They irradiated three different viruses and were able to destroy their infectivity, without nullifying their tumorigenicity (their ability to cause tumours, whether benign or malignant, to form). One of the viruses was unable to cause cancer itself in the animal model but after irradiation with ultraviolet light it was, and the assumption therefore was that a change had occurred in the genetic material of the virus - possibly mutation - that led to its carcinogenic properties. Thus, he said, it was possible that radiation, chemicals and viruses might interact in causing cancer. In a review article, Miller and Miller, writing in 1968, stated in summary that no carcinogenic mechanism, be it incited by radiation, virus, or a chemical, was understood at that time in molecular terms. Professor Idle described this statement of the position as at 1968 as being "both accurate and ex cathedra". Gelboin and Bates, writing in 1966, were convinced that cancer was the result of modification of key genes by carcinogens. This view was shared by Huebner and colleagues, writing in 1971, who envisaged chemical carcinogens and viruses sharing a common mechanism by working through the genes of the host organism to cause cancer. Huebner was a laboratory chief at the NCI in the USA.
[5.504] In the early 1970s, researchers reported the discovery of viral proteins in cells from chickens and mice that had not been infected by a virus. These chickens and mice were kept in sterile conditions in a laboratory animal facility and in addition it was by then possible to look at their tissues and detect the presence of virus. It was very surprising to workers at the time, Professor Idle said, that proteins could be found which originated from a virus when the virus was making more viruses in large numbers before it burst out of the cell. These were found in the cells of mice and chickens which had not been infected with viruses, so they must therefore come from the hosts' own genetic information. It led to a theory that animal cells contained genes related to viral genes, which became known as "virogenes". The first of these genes were discovered in various animal species by Varmus and Bishop in 1976; they became known as "proto-oncogenes". Professor Idle said that proto-oncogenes were genes present in all cells of all animals that play a role in "housekeeping". They were involved in the normal growth of cells and in maintaining the nature of a cell, a process known as "differentiation", so making sure that a liver cell, for example, stayed like a liver cell and the next time it divided it did not go into bone. These processes were all regulated by housekeeping genes, of which the proto-oncogenes formed one group. He described the name as "unfortunate", because of the suggestion of a link with cancer, but said that they were so named because of the way they were discovered in association with cancer.
[5.505] This discovery, that the animal genome contained endogenous genes that closely resembled viral genes that could cause tumours in animals, led to the "oncogene theory" of cancer. This involved the idea that cancer originated from these genes, already within the cells, that had normal cellular functions, but somehow the gene became activated and did things that it did not normally do in a normal cell. This formed part - maybe even a early part - of the process that set a cell on a pathway to becoming a malignant or cancer cell. The pathways below oncogene activity were very complex and involved a number of different signalling molecules and other proteins that set different reactions in train; but, in general, they upset the process by which a cell recognised where it was in what was called a "cell cycle". The cell went through a cycle when it grew; at a certain point it divided into two, and at another point, for example, it rested in a "resting phase". The cell needed to have sensors to know where it was in the cell cycle at any time so that it could switch on and off different genes because, when a cell was dividing, it needed to do different things from when it was resting. Oncogenes seemed to play a role in the recognition of where a cell was in its cycle and also in maintaining the differentiated integrity of the cell. The proto-oncogene might then be thought of as having lost its housekeeping function and as becoming a sort of rogue, allowing the cell to get out of control so far as growth and differentiation were concerned.
[5.506] The oncogene discovery propelled cancer research into the fields of virology and genetics. In 1972, there was beginning to emerge evidence that viral particles could be detected in certain human tumours. Very rapidly the question became not whether viruses could cause human cancer, but what proportion of cancers did viruses cause. The NCI continued to receive large appropriations from the United States Congress for its Special Virus Cancer Program, which was designed to develop a vaccination programme against cancer, on the proposition that cancer could be eliminated in the way that many infectious diseases, such as polio, had been eradicated. The use of viruses to study the process of carcinogenesis was very attractive because, unlike the host, the virus contained within a very small number of genes all the information necessary to cause cancer. The Program was later disbanded because it required special containment facilities, very special operating procedures and buildings. Also the nature of the Program had changed; many virologists who worked in it still held senior positions in direct research in the fields of viruses and cancer, but not in an attempt to produce a vaccine.
[5.507] Estimates of the proportion of human cancers induced by viruses ranged from 10% to 25%. Anonymous 1986 stated, at p.1431: "[A] reasonable estimate suggests that at least 25% of all human cancer worldwide is likely to be virus-induced." Specifically, Professor Idle stated, with respect to lung cancer, two human viruses and a number of animal viruses had been proposed as aetiological agents: first, the T-cell leukaemia virus, HTLV-1, which was endemic in certain regions, was found to be associated with small cell lung cancer in Japan; secondly, several studies had reported a high incidence of lung cancer in HIV-positive patients; and, thirdly, human papillomavirus (HPV) had been suggested to contribute to the pathogenesis of human lung cancer based upon the finding of statistically significantly elevated HPV in the blood of 47.7% of 149 lung cancer patients, compared with 12.6% of 174 non-cancer controls in a study conducted in Taiwan: Chiou et al. 2003.
[5.508] The situation today, Professor Idle said, was that viruses were believed to play a role in the aetiology of a broad range of human cancers, including lung cancer. The molecular genetic techniques available today to detect the presence of viruses in human and animal tissues had been developed only over the past decade or so. There were also vastly improved laboratory methods for the study of virus life cycles and this work continued in the hope of finding viral particles in more types of tumour and thus aiding an understanding of the role of viruses in human cancer.
[5.509] Professor Idle then turned to the macroparasite theory. He said that Fibiger reported in 1913 the occurrence of carcinoma in the stomach and oesophagus of rats that had fed on cockroaches infected with a parasitic worm, and was subsequently awarded a Nobel Prize for the claim that he had discovered that this worm could cause the cancers he observed. His work was, however, discredited by investigators within a decade, the consensus being that the gastric cancer he reported was due to vitamin deficiency. Nevertheless, by the end of 1964, the macroparasite theory was still being considered as a possible cause of some human cancers, for example in certain types of bladder cancer, particularly in Third World countries.
[5.510] Next mentioned was the chemical theory. In the period from approximately 1870 to 1930, coal and the products manufactured from it evolved into the mainstay of the chemical industry and provided many of the familiar household products of the day. The production of oils and tars from coal and shale, involving pyrolysis and distillation, produced new chemicals, such as the simple aromatics, which could be used in the manufacture of synthetic dyestuffs and pharmaceuticals. Whereas combustion involved a rapid chemical reaction between substances that was usually accompanied by a generation of heat and light in the form of flame, and in most cases oxygen comprised one of the reactants, pyrolysis was the destruction of organic materials by heat in the absence of oxygen. Pyrolysis of an organic material would yield destructive products, typically comprising a mixture of tars, oils and gases. Distillation was used to separate volatile oils, such as paraffin oil, from non-volatile residues, such as tars. Professor Idle explained, with the aid of a sketch, that the compounds referred to as "aromatic" were formed of six carbon atoms usually, but not always, joined in a ring, each with one hydrogen atom attached.
[5.511] The bulk production of oils and tars was associated with sporadic cases of skin cancer in workers exposed to these products. Workers exposed to paraffin oil developed skin tumours in areas of their bodies that normally came into contact with it. Reports such as this stimulated the scientific community of the time to ask the question whether chemicals could cause cancer, about which there was some scepticism. This question was addressed, in the first half of the 20th century, largely by one individual, Dr Kennaway, and his research group, working in London. In 1918 Yamagiwa and Ichikawa produced experimental cancer on the rabbit ear using a chemical mixture, specifically coal tar. This was the beginning of a period when coal tar was painted onto the skin of animals in an attempt to reproduce an occupational disease in humans. The purpose of the experiments was to produce an animal model upon which substances could be tested to see what was in the coal tar, with the hope of being able possibly to modify the process and remove the substance.
[5.512] Professor Idle said that it was important to appreciate the differences between skin tumours and skin cancer. A tumour was an abnormal mass of tissue, resulting from excessive cell division, that served no physiological purpose. A benign tumour simply stopped growing. A skin cancer was a malignant tumour which did not stop growing, had unregulated cell growth, invaded the surrounding tissues, metastasised to distant organs and, without any other intervention, would kill the animal or individual. Certain of the studies discussed in his report, particularly the earlier studies, referred only to "tumours", without specifying that they were largely benign growths which resolved after the treatment ceased. Later papers reported the incidence of both papillomas and epitheliomas (squamous cell carcinomas of the skin), the latter reportedly resulting from malignant transformation of the former. This was simply observational science. The appearance of papillomas on the skin of a rabbit ear or the back of a mouse, most of which disappeared, followed by the appearance of squamous cell carcinomas, led to the assumption that one followed the other. But this was not substantiated by any particular experiments and remained in the realm of hypothesis.
[5.513] Leitch, writing in 1922, reported that he painted the skin of mice, rats, guinea pigs and rabbits with different fractions of shale oil distillate; only mice ever developed malignant tumours. After conducting his animal studies in London, where he was the director of the Cancer Hospital Research Institute, he concluded, despite the observation of very few tumours, that in relation to human exposure to paraffin oil "we are dealing with an actual, in contradistinction from a predisposing, cause of cancer." Professor Idle said that he thought that this was one of the earliest observations in toxicology of a qualitative species difference in response to a challenge.
[5.514] An important figure at this time was Kennaway, also working at the Cancer Hospital Research Institute in London. His aim, as stated by him in 1924, was to identify "the cancer-producing substance in coal tar". His interest was in the development of methods that used mouse skin as a bioassay which would help in his search for this. Prior to this, the only discrete chemical that was thought to cause cancer in humans was arsenic. Kennaway pursued various fractions and individual chemical components of coal tar and realised rather quickly that this was not going to lead him to the answer. Professor Idle explained that coal tar was a material that could be sub-fractionated not only by the physical process of distillation but by extraction with various solvents. These fractions were then tested on animals. In a series of experiments Kennaway attempted to mimic coal tar. He pyrolysed substances at high temperatures under a stream of hydrogen gas to yield pyrolysis products with a simpler composition than that of coal tar. Starting with the simple hydrocarbon isoprene, a tar which he called "isoprene tar" was collected from the pyrolysis apparatus. Studies in which Kennaway painted this substance on the backs of mice led him to conclude that the coal tar carcinogen must be a hydrocarbon, since isoprene comprised only carbon and hydrogen, and was probably aromatic (the molecules being composed of closed rings) rather than aliphatic (the molecules being in open chains), like isoprene itself. Reference was made to a figure in Professor Idle's report showing the effect of high temperatures on simple aliphatic molecules to produce complex aromatic compounds. Kennaway then repeated this high temperature pyrolysis under a stream of hydrogen gas with a variety of other organic substrates, including other simple chemicals, one of which was human skin. All of these yielded fractions which he tested for carcinogenicity on mouse skin. Pyrolysis of human skin yielded a tar which was found to be carcinogenic when painted on mouse skin. Kennaway, writing in 1925, was perhaps the first to realise that the temperature at which substances were pyrolysed critically altered the biological properties of the resultant products. He found that products were not carcinogenic if the pyrolysis was at either too low or too high a temperature.
[5.515] One of Kennaway's collaborators was Dr Izrael Hieger who used ultra-violet spectroscopy to shine a beam of ultra-violet light through solutions of hydrocarbons and diffract the resulting fluorescent emissions into spectral bands, recording these on glass photographic plates. Kennaway was able to track down the carcinogenic constituents of his tars because, fortuitously, the carcinogenic fractions had a characteristic fluorescence spectrum. Schroeter in 1920 had shown, in an experiment in which the hydrocarbon tetralin was treated with aluminium chloride, that a fraction was produced, which became eponymous with him. When tested by Kennaway, this was found to be a mouse skin carcinogen. It was also found to produce a strong blue-violet fluorescence with three characteristic bands on the photographic plate, which became known as the "Schroeter bands". The Kennaway laboratory then embarked on a series of experiments that were based on the premise that when a compound displayed a particular pattern of Schroeter bands, that compound would be a mouse skin carcinogen.
[5.516] Together with Hieger, Kennaway screened by ultra-violet spectroscopy the large number of pure polycyclic aromatic hydrocarbons that had been prepared from the distillation of coal tar and from the pyrolysis of organic materials. Having identified the compound with the best match to the Schroeter bands, Kennaway then sought to modify its structure chemically to find an even better match to the Schroeter bands, and this resulted in the de novo chemical synthesis and subsequent testing on mouse skin of a number of compounds. One of these, 1,2,7,8-dibenzanthracene was reported to be carcinogenic to mouse skin and thus became the first synthetic chemical reported to be a carcinogen. Kennaway was also able to identify a large number of polycyclic aromatic hydrocarbons as being non-carcinogenic to mouse skin.
[5.517] The discovery of the first synthetic chemical carcinogen occurred because the researchers had correlated the particular physico-chemical characteristic (the Schroeter bands) of a chemical with carcinogenicity on mouse skin. They still had the problem of finding what it was in coal tar that caused carcinogenicity in the mouse. Kennaway's laboratory continued screening a large number of synthetically produced polycyclic aromatic hydrocarbons. They were able to find that some were carcinogenic, others were mildly carcinogenic and the majority were non-carcinogenic. They started to put together some ideas of how the structure of the compounds might dictate their biological activity, but, at the end of the day, this could not go on forever and they had to stop. Kennaway and his colleagues therefore returned to their original 1924 aim of identifying the carcinogenic agent in coal tar itself. Starting with two tons of coal pitch, they isolated crystals of a highly carcinogenic substance with the exact Schroeter bands which, by comparison with authentic chemically-synthesised material, turned out to be 1,2-benzpyrene (now referred to as benzo[a]pyrene, B[a]P). (Professor Idle explained that benzpyrene existed in two distinct chemical forms, benzo[a]pyrene (known then as 1,2-benzpyrene) and benzo[e]pyrene (known then as 3,4-benzpyrene). Only benzo[a]pyrene was carcinogenic on mouse skin. It was, he said, a point of great confusion that at some point in history the numbering system was changed and 1,2-benzpyrene became 3,4-benzpyrene and vice versa. Thus, today, the carcinogenic benzo[a]pyrene was also known by the trivial name of 3,4-benzpyrene. The biological properties of this chemical, first reported as a constituent of a carcinogenic mixture (coal tar) in 1933, were still being investigated. Benzo[a]pyrene seemed to occupy a special place in the minds of scientists in the field of experimental carcinogenesis. This featured also in research into smoking and lung cancer: reference was made to RCP 1962, in which it was stated, at p.10, para.19, that some sixteen different substances capable of initiating cancer in experimental animals had hitherto been identified in tobacco smoke; among the references were papers relating to the presence of benzpyrene and other polycyclic hyrocarbons in cigarette smoke.
[5.518] Work continued to focus on benzo[a]pyrene in the 1930s and 1940s and a vast number of other polycyclic aromatic hydrocarbons were synthesised in the laboratory. Many of these were tested in Kennaway's laboratory on mouse skin and subsequently in other bioassays, primarily to understand the relationship between chemical structure and carcinogenesis. There was an underlying belief that chemicals that could produce cancer must have some common motifs in their chemical structure: it was not the belief then that chemicals randomly could produce cancer. So they were trying to understand what in the chemical structure might be the determining motif in the carcinogenesis process. The testing of single agents on mouse skin continued until the early 1940s. In the course of these experiments, it was reported that different strains of mice were differentially sensitive to the skin painting protocol. Professor Idle said that when different strains of mice demonstrated different biological properties, it was almost invariably due to genetic differences between the strains. This was, in his view, all part and parcel of evolutionary differences.
[5.519] By the end of 1964, it had been established that chemicals could be carcinogenic to mice and rabbits and thus might cause cancer in humans. The chemical theory of carcinogenesis prompted considerable research and this continued in subsequent years. It had been proposed, in the period prior to 1965, that when a polycyclic aromatic hydrocarbon, such as benzo[a]pyrene, was painted on to mouse skin it was converted to a metabolite that exerted the carcinogenic effect of the benzo[a]pyrene. The role of metabolism in chemical carcinogenesis became the subject of enquiry in the subsequent period. Professor Idle said that it was the subject of a very vigorous debate at that time. Metabolism was the process of modification of the chemical structure of things within the body, both endogenous and exogenous compounds. There were those that proposed that the carcinogenic activity of a substance like benzo[a]pyrene resided in a metabolite produced from it, although this was not a universal view at the time. Boyland was a professor at what was then known as the Chester Beatty Research Institute in London, previously the Laboratory of the Cancer Research Hospital; this was the premier cancer research establishment in the United Kingdom at that time. Boyland, who was a biochemist, developed the idea that polycyclic aromatic hydrocarbons required conversion to an epoxide in order to exert their carcinogenic effects. That view was considered as risible at the time; people really did not believe it at all. Epoxides were reactive metabolites. Reactive metabolites of chemical carcinogens were formed in a reaction between the chemical and the host organism. Reactive metabolites possessed the property of being able to react chemically and spontaneously with other molecules, particularly the macromolecules of the cell, such as proteins, RNA and DNA (deoxyribonucleic acid).
[5.520] Polarisation in the debate was largely between the Chester Beatty Institute and the McArdle Laboratories in Wisconsin, where the view of James and Elizabeth Miller at that time was that polycyclic aromatic hydrocarbons were per se carcinogenic, requiring no metabolism to exert their effects. The McArdle Laboratories were the first cancer research institute in the United States to be built on a university campus and grew into a very formidable and important research institute; it had several Nobel Prizes. Miller and Miller were amongst the first individuals to study chemical carcinogenesis and were very highly regarded scientists. Conney was a PhD student studying with the Millers at the time of a report in 1957. Working with rats, he reported that if they were pre-treated with benzo[a]pyrene, their livers were mashed and a fraction was made from them that contained an enzyme which he called "benzpyrene hydroxylase", the benzpyrene hydroxylase activity of animals so treated was much higher than in untreated animals, and he deduced that benzo[a]pyrene could induce the metabolism of itself and other related chemicals. By the end of 1964, the role of metabolism in carcinogenesis had become an active area of research.
[5.521] Professor Idle went on to discuss the debate that took place between proponents of the "protein loss theory of carcinogenesis" at the McArdle Memorial Laboratory and the "DNA theory" of carcinogenesis at the Chester Beatty Research Institute. The protein loss theory was based upon an observation in bacteria that the gene which coded for an enzyme known as lactate dehydrogenase (which converted lactate acid to pyruvic acid), was regulated by a protein in the cell. A relationship was discovered between the cell and the DNA: as described by Professor Idle, there were "on/off switches" that were regulated by proteins. At the McArdle Laboratory the theory was that, if there was a key protein that regulated something in the genetic regulatory apparatus of the cell and a chemical damaged the protein in some way, then that altered the way in which the whole cell apparatus was regulated, hence the protein loss hypothesis of carcinogenesis. The DNA theory was somewhat similar, except that it omitted the protein intermediary. Under this theory, there would be an effect of the chemical directly at the genetic level on the gene, disrupting the information process and causing the cell not to operate properly and perhaps go too fast or get out of control. At that time this theory was rudimentary. There was considerable tension between the two theories that lasted up to the end of 1964, and beyond.
[5.522] In Boyland 1952 the author, working at the Chester Beatty Research Institute, advanced the theory that carcinogens interacted with nucleic acids. At p.81 he stated:
"Although some carcinogenic compounds may produce their specific effects by enzyme inhibition, the hypothesis that the results are due to the more direct action on chromosome nucleic acid seems more likely in some cases. Because the carcinogens produce specific damage to chromosomes and chromosomes contain desoxyribosenucleic acid, it is tempting to assume that the carcinogens act by producing abnormalities in the nucleoprotein."
Researchers at the McArdle Memorial Laboratory continued to insist that carcinogenesis did not result from an interaction of carcinogens with DNA, but rather with proteins. Aware of this debate, Brookes and Lawley, from the Chester Beatty Research Institute, published evidence in 1964 that polycyclic aromatic hydrocarbons bound to DNA in mouse skin and suggested that non-carcinogenic polycyclic aromatic hydrocarbons did not bind, whilst carcinogenic ones did. Thus the two phenomena, binding to DNA and carcinogenesis, were reported to be related. They also suggested that the polycyclic aromatic hydrocarbons must first undergo metabolism to enable this binding. They concluded, however, that the results obtained in their work at that time did not give a conclusive indication of the nature of the significant cellular receptors of chemical carcinogens. Professor Idle commented that this dispute, between two of the foremost cancer research institutes in the world, indicated that there was no consensus amongst cancer researchers about the cellular target of chemical carcinogens.
[5.523] In the same period, Nordling published in 1953 what he called a new theory on the cancer-inducing mechanism, that cancer arose in cells by mutation. A mutation is a permanent change in the genetic information in the DNA of a cell, which may be passed on to progeny cells. Nordling proposed a multi-step process in carcinogenesis involving serial mutations. Burdette, writing in 1955, in an extensive review of the role of mutations in carcinogenesis, concluded that evidence for the existence of cancer susceptibility genes was compelling and extensive. He proposed that there was no justification for a general correlation between mutagenesis and carcinogenesis. Whereas Nordling had seen carcinogenesis as depending upon mutagenesis, Burdette concluded that a carcinoma could arise without the intervention of changes in the DNA sequence. He also tended to dismiss somatic mutation, a permanent mutation that occurred in a cell that could be passed on to progeny. Brues, in agreement with Nordling, dismissed the idea of a single mutation leading to cancer, and stated that there was a threshold below which neither chemicals nor radiation caused cancer. The single discovery that most radically altered the understanding of how living systems worked was the description of the structure and function of the DNA double helix by Watson and Crick in 1953. This showed the molecular structure of DNA and its genetic implications, and was a breakthrough which was to have a profound and long-term effect upon cancer research.
[5.524] As at the end of 1964, there was an unresolved debate between the "DNA" and the "protein" proponents. There was also an ongoing discussion in the scientific literature of the role that might, or might not, be played by genetic mutation in cancer.
[5.525] Turning to the topic of chemical structure and metabolism during the period from 1965 to 1971, Professor Idle said that considerable progress was to be made in this period in answering the questions concerning the chemical structure of carcinogens in relation to their biological properties and, in particular, whether or not a carcinogen was active per se, or required conversion to some reactive intermediate by the host's own endogenous metabolic processes. Miller and Miller, writing in 1965, noted that their present knowledge of the metabolism of chemical carcinogens, like that of their structures, was of no great predictive value in evaluating the possible carcinogenic hazards of a new chemical structure. They had not been able to understand how the structure of a particular polycyclic aromatic hydrocarbon might predict the biological outcome. Then they had gone on to look at some of the metabolites of polycyclic aromatic hydrocarbons and they still were unable to predict the outcome. They also were of the view that the polycyclic aromatic hydrocarbons were carcinogenic per se and did not require metabolic activation, but that the nitrosamines did require the intervention of host metabolism to generate carcinogenic molecules. Nitrosamines, Professor Idle said, were a ubiquitous group of chemicals that were formed very readily from substances known as "secondary amines" and their reaction with either the oxides of nitrogen in the atmosphere or nitrate and nitrite, which were two inorganic ions.
[5.526] Meanwhile, a contrary idea had developed that the polycyclic aromatic hydrocarbons did require metabolic activation and that this probably occurred in the form of metabolites known as epoxides. This had originally been suggested in Boyland 1950, but there was no experimental evidence for such an effect. The Millers had tested polycyclic aromatic hydrocarbons and found that they were no more nor less carcinogenic than some of the synthetic epoxides that had been made. Other scientists now entered the arena who made theoretical calculations and synthesised other chemicals. They carried out additional experiments which led further towards the idea that epoxides required to be made to render these molecules carcinogenic. Pullman and Pullman, working in Paris, identified part of the polycyclic structure as the motif which caused cancer, and coined the expression "K region epoxides". Miller and Miller tested a number of K region epoxides on mouse skin and found them to be non-carcinogenic, thus reinforcing their earlier view that polycyclic aromatic hydrocarbons were carcinogenic per se. Other researchers tested other K region epoxides and found them to have very low carcinogenicity or to be only weakly carcinogenic.
[5.527] Professor Idle said that despite these results Dipple and colleagues in 1968 offered the important and insightful conclusion that the distinction between carcinogens and non-carcinogens would most likely be a consequence of the chemical structure of the reactive intermediate rather than of the structure of the parent compound. They produced experimental data from the testing of a series of polycyclic aromatic hydrocarbons to support this contention. In the same year Miller and Miller modified their earlier view regarding the metabolism of polycyclic aromatic hydrocarbons, stating that an examination of the literature suggested strongly that most, if not all, chemical carcinogens either were, or were converted in vivo to, reactive electrophilic derivatives which combined with nucleophilic groups in crucial tissue components, such as nucleic acids and proteins. A chemical entity was said to be "nucleophilic" when it attacked other molecules through the parts of their chemical structures which were low in electron density. Nucleophiles, in general, had areas of their molecules which were abundant in electrons. In turn, nucleophiles were attacked by electrophiles, which is what Miller and Miller believed all chemical carcinogens to be. Thus the Millers now agreed with Boyland 1950 that the polycyclic aromatic hydrocarbons required to be metabolised before they became carcinogenic.
[5.528] The epoxide metabolites proposed by Boyland 1950 for polycyclic aromatic hydrocarbons were electrophiles, but the existence of epoxides as biological metabolites was merely theoretical until a report in 1968 that their occurrence had been observed for the first time. It was reported in 1969 that benzo[a]pyrene required metabolism by enzymes known as microsomal hydroxylases before binding to DNA, and in 1970 evidence was provided that the K region epoxides were probably the metabolites that bound to DNA and protein. In 1971 it was reported that an epoxide was formed when another polycyclic aromatic hydrocarbon underwent metabolism, and there was a further report that several K region epoxides of polycyclic aromatic hydrocarbons were able to transform rodent cells in culture to a malignant phenotype, and that epoxides were indeed the metabolites of polycyclic aromatic hydrocarbons. Magee and Barnes, writing in 1967, stated that nitrosamines were mutagenic by virtue of their metabolism by an unknown process by host enzymes. Professor Idle said that it was unusual to find exactly the same enzyme in two different species, such as a rat and a mouse, so the process of metabolism of any compound by enzymes was likely to vary from species to species, though it would not be known whether this was so until experiments were carried out.
[5.529] At this time, another important discovery was made by Nebert and Gelboin at the NCI. They discovered that different mouse strains responded to the induction response in different ways. If an animal was pre-treated with a polycyclic aromatic hydrocarbon like benzo[a]pyrene, the enzyme that metabolised this and related compounds reached a higher level in the tissues of the animal. Nebert and Gelboin found that there were strains of mice that were poorly induced by polycyclics and strains that were readily induced by them; they bred these animals and showed that this was a genetic characteristic and varied from strain to strain, due to a gene. The question that then emerged was whether or not there existed in the human population a similar genetic variation in the induction of polycyclic aromatic hydrocarbon metabolism. This question was to be addressed in the next time period and still occupied researchers today. Professor Idle said that there was little doubt that homo sapiens was the most variant of all the mammalian species, with tremendous variation in most biological processes. These were covered, unseen variations in metabolic pathways. This had been learnt through the field of pharmacogenetics. There were tremendous intersubject and interethnic variations in metabolic processes in the human species. Metabolic processes might vary between the genders: many enzymes were regulated by hormones, including sex hormones, so there might be a process which was at a very high level in females and at a very low level in males.
[5.530] Taken together, he said, all these observations pointed to a role for endogenous metabolic activation in the carcinogenicity and mutagenicity of two classes of chemical carcinogens, polycyclic aromatic hydrocarbons and nitrosamines. It appeared that the structure of the metabolite, rather than the original chemical, was what determined its carcinogenicity. For example, in the case of benzo[a]pyrene, it was not the benzo[a]pyrene molecule itself which was carcinogenic, but rather the product or products of its metabolism within the tissues that were. Having considered, until this time, that metabolism was a process that removed toxic foreign molecules from the body, scientists were now faced with a paradigm shift, whereby metabolism could contribute materially to the carcinogenic process. Professor Idle thought that this did not surprise people working in the field of drug metabolism as much as it did people in other branches of medicine who were not familiar with this field.
[5.531] By the end of 1971, it was known that some but not all carcinogens needed to be metabolised to exert their carcinogenic effects. The occurrence of genetic factors that might control this metabolism was beginning to be explored. This exploration continued throughout the subsequent period.
[5.532] A continuing question during the period from 1965 to 1971 concerned the molecular mechanism of attack of a chemical carcinogen or its reactive metabolite on cellular constituents: did the carcinogen strike at protein molecules or DNA molecules? Miller and Miller stated in 1965 that the cellular and molecular mechanisms were not known for any carcinogen and, while there had been reports of interactions with DNA and RNA (the molecules of cell information transfer) there were really no correlations with carcinogenesis. Pitot and Heidelberger had advanced in 1963 a novel theory involving protein-carcinogen interactions. Brookes, writing in 1966, cast some doubts on this "protein-deletion theory", but stated that in the past when several rival theories had each received strong support, the truth had ultimately been shown to embrace all points of view, and asked whether it was possible that this might be so for the obviously complex process of carcinogenesis. Horsfall came out in favour of the DNA theory in 1966. Prehn, writing in 1964, used his "clonal selection hypothesis" to support the protein-deletion theory. Professor Idle said that this was unlike any other theory of cancer that had gone before. Prehn stated that, within a population of normal cells, there were some that were already programmed and destined to become cancer cells. He saw chemicals as killing off the least fit cells, in a Darwinian sense, and that the fittest, pre-programmed cells would then be selected for and grow into malignant clones. He distinguished between an instructional process, where an outside signal instructed a cell to become a tumour cell, and a non-instructional one, where the cell was simply selected for. In 1967 Goshman and Heidelberger said that the question whether DNA, RNA or protein was the primary cellular receptor of chemical carcinogens remained unanswered; it was evidence that, although binding of carcinogens to DNA might be necessary for carcinogenesis, the appearance of tumours depended upon a series of cellular events which had not yet been elucidated.
[5.533] By this stage, Professor Idle said, the debate about the activation of carcinogens had been settled. The DNA/protein target debate was still continuing and evolving, and he thought that in many ways it converged during this period. In 1969 Brookes and Heidelberger generated further experimental evidence for the interaction between polycyclic aromatic hydrocarbons and DNA, but were unable to characterise the interaction. Miller wrote in 1970 that neither the DNA nor the protein-deletion hypothesis, nor any other hypothesis, had yet received unequivocal experimental support in any incidence with any carcinogenic agent. Professor Idle said that this was consistent with his reading of the literature of that period. In the context of the search for the cause of causes of lung cancer, it was clear to him at least that scientists were struggling to define and agree upon a mechanism by which cancer, as a phenomenon, could occur. There was little attention to individual cancers such as lung cancer. The big question was what cancer was and how it occurred. This was the time when he entered into the drug metabolism field and he remembered these issues. In 1971 Huberman and colleagues provided in vitro evidence for DNA as the molecular target for chemical carcinogens. They concluded that there was a good correlation between chemical reactivity, mutagenicity, and binding to DNA in the test tube and in cells. Their work established a better correlation between metabolic activation of hydrocarbons and mutagenic activity than was previously reported, but did not establish a critical association between the processes of mutagenesis and carcinogenesis. Professor Idle said that this was an accurate reflection of the state of the art as at 1971. In relation to nitrosamines, Magee and Barnes discussed in 1967 the competing DNA and protein-deletion hypotheses. They favoured the former because of the proven interactions between nitrosamines and the DNA molecule, but were forced to conclude that there was no experimental proof that reaction with these macromolecules rather than other cellular components was related to carcinogenesis. Other researchers in 1968 attributed tumorigenesis to enzyme inhibition; Professor Idle described this as a novel theory unrelated to the aforementioned principal theories of chemical carcinogenesis, but without support from other investigators.
[5.534] He said that by the end of 1971, there still existed a lively debate as to whether the critical site of action in the cell was protein or DNA or something else. At a cellular or molecular level, the mechanism of action of chemical carcinogens remained unknown. The period from 1972 to date had seen the coalescence of a number of diverse threads of evidence into the current paradigm of chemical carcinogenesis. Nowak and colleagues, writing in 2002, said that the concept that tumours developed through the accumulation of genetic alternations of oncogenes and tumour suppressor genes was now widely accepted as a fundamental principle of cancer biology. Professor Idle explained that tumour suppressor genes were originally called "anti-oncogenes", because they appeared to serve the opposite purpose to that of oncogenes. They were also involved in the housekeeping functions that regulated how cells grew and divided and sometimes how they died, but they provided a "hub", through which a whole series of pathways was triggered. Where a tumour suppressor gene was removed from the circuit, then a number of different processes failed: DNA repair, repair of modifications to DNA, the ability of the cell to enter into the process known as apoptosis and various other integral processes concerning the well-being of the cell. Apoptosis was one of the great discoveries of the last two or three decades. Cells were always thought to die by damage from the outside, undergoing a certain transformation during death that could be seen under the microscope, a process called necrosis. It was then observed that cells died, not just by necrosis, but by what was now called apoptosis. A signal was sent from the nucleus of the cell, usually to itself but also to other cells, saying that it was time to die. The cell shrank and a dissolution took place, enabling its materials to be used by other cells. Interference with either an oncogene or a tumour suppressor gene affected the division, differentiation and death of a cell.
[5.535] DNA repair was possibly the most important housekeeping function of any cell, because damage occurred to DNA all the time. The simple use of oxygen for respiration by all organisms created very reactive intermediates which damaged DNA. DNA itself was an unstable molecule: it could hydrolyse, reacting with water to break some of its bonds. So DNA became modified in the normal everyday process of all life, and these modifications had to be repaired. If they were not, then the gene contained erroneous information that could be passed on to another generation in the case of a bacterium, for example, or that could alter the functioning of the liver cells of a human. There were 130 discrete DNA repair enzymes that had been described in humans. Each of them repaired a different type of modification of the DNA structure, and some of them backed up the effects of others. There were tens of thousands of modifications to DNA in every single cell of the human body every day. These had to be repaired if the cell was to function normally and throughout a normal life span of an animal or an individual. There were tens of thousands of DNA repair events every day in a mammalian cell. The processes of misuse of oxygen molecules, known as "oxidative stress", and of hydrolysis gave rise to at least 60,000 discrete events requiring a similar number of repair events per cell per day.
[5.536] Professor Idle said he thought that the view of Nowak and colleagues was partly correct. What the evidence showed clearly was that tumour suppressor genes and oncogenes became modified through genetic alteration, leading to an increased speed of division of cells and certain other characteristics. He did not agree that it was a fundamental principle of cancer biology, if the authors meant that this led to cancer. It certainly was part of some of the early transformations that might occur on a pathway that might or might not lead to cancer, bearing in mind that cells had the ability to kill themselves off by apoptosis and the immune system could also remove them. Other factors had been recognised as relevant in carcinogenesis. Professor Idle thought that viruses clearly could play a part. Radiation was sometimes described as the classical carcinogen, by damaging DNA randomly. Oxidative stress and many other factors had been identified.
[5.537] Professor Idle said that the tumour suppressor gene which was of the greatest interest and which had been the most extensively studied over the past twenty years was p53 (the protein coded for by this gene having a molecular weight of 53,000). The p53 gene served as a hub to contribute to the control of apoptosis, DNA repair, the cell cycle and differentiation pathways: many of the housekeeping and important functions of a cell. Robles and colleagues, writing in 2002, stated that since the discovery of this gene an overwhelming volume of literature had accumulated supporting its role as a central regulatory node in the protein network that mediated cellular responses to endogenous and exogenous stresses. Disabling this pathway through mutation or silencing of its components led to self transformation and was thought to contribute to tumorigenesis.
[5.538] Professor Idle said that the assorted elements of the chemical theory carcinogenesis as it involved oncogenes and tumour suppressor genes could be itemised as follows: (i) certain chemical carcinogens entered into cells and participated in metabolic processes carried out by the enzymes within the tissues that converted them into chemically reactive metabolites; (ii) chemically reactive metabolites remained in the tissues chemically bonded to DNA, in a process known as "covalent binding", and survived the processes of DNA repair, becoming "persistent DNA adducts" that were considered to be detrimental to the processes of transcription and replication within the cell; (iii) persistent DNA adducts were then seen to cause DNA mutations; and (iv) if these mutations happened to occur by chance in oncogenes or tumour suppressor genes, these led to perturbations of cell function and ultimately to cancer. Overall, he said, these jigsaw pieces were seen as forming a picture of chemical carcinogenesis. Whilst some of the component pieces of the current paradigm were uncontroversial, others might or might not turn out to be valid. There were points of the theory that were worthy of careful consideration.
[5.539] First, it was generally accepted that certain chemicals could be converted by the tissues into both inactive water-soluble metabolites, in a process known as "detoxication", and short-lived chemically reactive metabolites, in a process known as "activation". Some of the reactive metabolites of polycyclic aromatic hydrocarbons had half lives of a second or so, so that the longest lived ones probably had lives of the order of one minute. What often distinguished a toxic from a non-toxic compound was the balance between detoxication and activation. Most, but by no means all, chemical carcinogens formed reactive metabolites that themselves were often further metabolised to relative inert and excretable metabolites. Boyland 1950 first proposed that the chemically inert benzo[a]pyrene was metabolised, within the mouse skin upon which it had been painted, to a more chemically reactive epoxide derivative. The metabolic activation of chemicals was currently accepted as the first step in chemical carcinogenesis and there existed ample evidence that it occurred. This evidence was found from the administration of chemicals either to living animals or to cells derived from living animals and sometimes from humans in vitro.
[5.540] Secondly, it was also generally accepted that reactive metabolites of chemicals could bind to nuclear DNA. The current paradigm focused on this, in contradistinction to the "protein loss" theory. However, these reactive metabolites also bound to other cellular macromolecules, including proteins, RNA, and mitochondrial DNA. Mitochondria were organelles that occurred in almost all cells and which generated almost all the energy in a cell. They were small discrete cellular structures responsible for energy production and cellular respiration. The mitochondrial DNA differed from nuclear DNA in that all of it was used to code for RNA molecules that were used in the making of proteins, whereas in nuclear DNA about 90% was not currently thought to have any function. Thus, any disruption or mutation of mitochondrial DNA sequences would be likely to have effects upon the physiology of the cell. In addition, nuclear DNA had multiple processes by which damage could be repaired, whereas mitochondrial DNA had a limited repair system. It has been reported that polycyclic aromatic hydrocarbons, among other substances, bound to mitochondrial DNA very much more avidly than to nuclear DNA in experimental animals. Professor Idle said that the current paradigm of chemical carcinogenesis did not take account of the binding of chemically-reactive carcinogen metabolites to mitochondrial DNA. It was highly focused on nuclear DNA and particular on oncogenes and tumour suppressor genes and very surprisingly had failed to focus on the interesting observations with mitochondrial DNA, given its special characteristics and that while there might only be handful of genes in each of the mitochondria, there were often many thousands of mitochondria per cell, so the amount of DNA present in a cell in the mitochondria was quite considerable and in addition it was not repaired when there were modifications to the sequence. So in his opinion this was a major oversight.
[5.541] Thirdly, the formation of persistent DNA adducts could result in the death of the cell. Fourthly, almost all insights into the relationship between adduct formation and mutagenesis derived from studies using bacteria, rather than investigations in mammals. Fifthly, the finding of mutations in the proto-oncogenes and the tumour suppressor genes had generated particular interest in recent years. The obvious interpretation of this observation was that the upsetting of the role of these housekeeping genes could lead to carcinogenesis. The current theory of chemical carcinogenesis was built upon the mutation and activation of proto-oncogenes and the mutation and inactivation of tumour suppressor genes. The tumour suppressor gene of greatest current interest was p53, which Lane, one of its co-discoverers, has named "the guardian of the genome". The observed response of p53 to cellular imbalance had been to switch on DNA repair, and when repair has been incomplete and had not solved the problem, it then told the cell to die by apoptosis. Thus, the effect of mutated and thus activated oncogenes would be neutralised by the switching on of the p53 gene, the resulting synthesis of the p53 protein and the subsequent benefits to the cell. If, however, p53 was also mutated, then what has been called "the oncogenic stress pathway" was seen as running unchecked, with the consequence of unregulated cell growth ultimately leading to cancer. The aforementioned chain of events, involving the mutation and activation of oncogenes and the mutation and inactivation of tumour suppressor genes, had so far been established only in colorectal cancer.
[5.542] Under reference to a figure in his report, Professor Idle said that it had been reported that the p53 gene contained mutations in most types of human cancer, ranging from a frequency of between 13% and 14% in prostate carcinoma, up to more than 70% in small cell lung cancer. The frequency in non-small cell lung cancer, which included mainly adenocarcinoma and squamous cell lung cancer, with a frequency of about 40%, placed this group in the middle of the range, so it was no different from other tumours essentially. Professor Idle said that it was noteworthy that, while 40% of non-small cell lung cancer tumours were reported to have had p53 mutations, 60% did not. The data were drawn from an international database held in France and associated with IARC; they came from a relatively small sample of the population. The observation that certain proto-oncogenes and tumour suppressor genes were mutated in various human tumours had become a major focus of the current paradigm of chemical carcinogenesis.
[5.543] Sixthly, activated ras oncogenes were thought to occur on average in about 30% of human cancers, most notably pancreatic and colon carcinomas. Proto-oncogenes were always counterparts of viral genes, ras was the cellular counterpart in mammalian cells for a virus known as the rat sarcoma virus. Even in experimental animals where tumours had been induced by the administration of large doses of carcinogens, a mutated and thus activated ras and a mutated and thus inactivated p53 were usually not found together, never mind in the same cell. This was inconsistent with this plank of the current paradigm that dictated that these mutations should occur together in the same cell. Some workers had commented that pathways independent of ras activation and p53 inactivation were involved in the experimental tumours they had produced. These findings had led researchers to propose that not just any kind of p53 mutation is associated with cancer formation; rather, mutations in certain regions of the gene were required.
[5.544] Genes contained the information necessary for coding for a protein that had some cellular function, whether it be architectural or metabolic. They were always thousands of bases long and were divided into exons, units that coded for part of the protein, separated by usually larger regions of introns, which were thought to be just "garbage" in biochemical terms, nonsense information. So a gene was a very long piece of DNA with some useful and some non-useful sequences in it. Mutations could occur anywhere at any single base along the many thousands of bases in a line, and researchers had proposed that mutations in certain positions on that gene were the key thing, not just mutations anywhere within the gene. Genetics was a very new subject and was moving very rapidly. New genes had been discovered to be hidden inside what was thought to be "garbage", and there were still many secrets to be discovered. Over 90% of human DNA had a function that could not be ascribed to anything, and in the case of only a small fraction of the 30,000 genes that had emerged in the human genome project was their function known.
[5.545] It might be seen therefore, Professor Idle said, that not all of the four consecutive elements of the current model of chemical carcinogenesis had been demonstrated to occur for human cancers, including lung cancer. The first element, that chemicals were activated through metabolism, had been the best studied and was generally accepted. The second element, the formation of persistent DNA adducts, had certainly been well-studied in experimental animals and adducts had been detected in various human tissues. That this might follow from the first element was uncontroversial. However, the third element, which required that a particular DNA adduct led to a specific DNA mutation, had not been demonstrated in relation to any human cancer, most of the experiments having been performed in bacteria. Finally, the fourth element of the current paradigm, that mutations in certain proto-oncogenes and tumour suppressor genes, as they were found in some human cancers, were the cause of these cancers, had also not been demonstrated. The only exception to this statement was the understanding of the serial genetic steps that led to certain forms of human colorectal carcinoma. Mutations in oncogene and tumour suppressor genes might explain how cells could be stimulated by chemicals to grow faster, but that did not explain the features of invasion and metastasis. In Professor Idle's opinion, the pathways that led to cancer required considerably more research before a robust paradigm could emerge.
[5.546] Professor Idle then turned to the radiation theory. Muller was awarded the Nobel Prize in 1946 for his discovery that x-rays could cause inheritable genetic damage in fruit flies. Gates and Warren, writing in 1961, commented that "epidermoid carcinoma" (squamous cell carcinoma) of the lung had been produced in rats and mice, using alpha-radiation, beta-radiation and gamma-radiation emitting radioisotopes, as early as 1938. They concluded that the increase in the particulate and gaseous radioactivity in the atmosphere in the previous decade, and the expanding industrial uses of artificial radioisotopes, had made it desirable to define more explicitly the potential dangers of these new additions to the environment. Professor Idle said that by the end of 1964 there existed concern that environmental radiation might be a cause of human cancer, including lung cancer.
[5.547] In the period from 1965 to 1971, the question was being asked whether or not x-rays and other types of radiation could, by breaking chemical bonds, generate cancer. There was an understanding of the amount of energy required to break different kinds of chemical bond. Ionising radiation, for example gamma-radiation or x-rays, possessed enough energy to break chemical bonds in the irradiated material, in contrast to visible or infra-red radiation, which were of lower energy. In 1966 De Villiers and Gross reported success in producing squamous cell carcinoma of the lung in hamsters using x-rays. They stated that under these circumstances, the reproducible initiation of squamous cell carcinomas by external radiation, a total incidence of 12% among deaths occurring within twenty post-irradiation weeks, in the lungs of hamsters, was unique and highly significant. Professor Idle said that these were among the first observations of researchers producing squamous cell carcinoma in animals with radiation. He was almost certain that what was occurring was that in the period from the 1930s until the 1960s, radiation was a topic of public concern. It was then a new science, and the ability to work with radioisotopes in the laboratory was new. Then it was radiation, today it was genetics. Trends came and went, and he thought activity in this area had died out to some extent and had been replaced by other activities.
[5.548] The next theory discussed by Professor Idle was that of personality, emotion and stress. He said that this might be the most long-lived of the all the theories. There was still an interest today in the role of the mind in medicine: psychosomatic medicine was an important area of activity and very important in cancer treatment. All the oncologists he knew had always said to him that people's belief that they would recover and survive was an important component. The idea had originated with the Ancient Greeks. Among more recent writers, Sir James Paget, the famous surgical pathologist, wrote in 1870 that it could hardly be doubted that mental depression was a weighty addition to the other influences that favoured the development of the cancerous constitution. The first laboratory investigations of stress were by Selye who, in 1936, described experiments in which rats, when suddenly confronted with a critical situation, developed a stereotyped syndrome characterised by enlargement and hyperactivity of the adrenal glands, atrophy of the thymus gland, lymph nodes, spleen and liver, and the appearance of gastrointestinal ulcers. Critical situations that could precipitate this syndrome were cold, infection, trauma, excessive muscular exercise, and intoxications with sub-lethal doses of drugs and chemicals. He termed this syndrome the "general alarm reaction"; in 1956 he referred to this response to stress as the "general adaptation syndrome". An example of a similar process in humans, Professor Idle said, was the general recognition that gastric ulcers could be caused by stressors. Selye was the President of the International Institute of Stress until his death in 1982 and became the leading figure researching the effects of stress on the endocrine and immune systems and its predisposition to disease. LeShan, writing in 1959, was one of the first to point to personal loss as a cause of cancer in humans, when he wrote that the most consistently reported, relevant psychological factor had been the loss of a major emotional relationship prior to the first-noted symptoms of the neoplasm.
[5.549] In the early 1960s Dr David Kissen in Glasgow began a detailed clinical psychological study of individuals diagnosed with lung cancer, which he reported in Kissen 1963a and Kissen 1963b. Professor Idle said that he himself was neither a psychologist nor a psychiatrist, and his knowledge was a general one and came from the reading of Kissen's papers and other papers in this field, but what he took from Kissen's papers were the conclusions that lung cancer patients had a lower rate of "childhood behaviour disorders", a finding "in keeping with the hypothesis that lung-cancer patients have a diminished outlet for emotional discharge", and that statistically significantly more lung cancer patients than controls stated that they tended to repress emotional problems.
[5.550] Professor Idle said that in the period prior to 1965, there had been a long historical interest in psychological factors as they might predispose to cancer. As Sir William Osler said in 1906, it was much more important to know what sort of a patient had a disease, than what sort of disease the patient had. Before 1965, studies had commenced on psychological factors in both childhood and adulthood that might lead to lung cancer. These psychological and emotional factors became the foundation in the next time period for the discussion of the role of stress in cancer causation. Green reported in 1996 a study of patients newly diagnosed with leukaemia or lymphoma and concluded that to the extent that psychological factors might prove to be relevant to the pathogenesis of leukaemias and lymphomas, it was a matter of the individual's running out of psychological resources. Kissen continued his studies in Glasgow until his death in 1968, by which time he was recognised by his peers as the leading authority on psychosomatic factors leading to lung cancer. In a study reported in 1967 of male lung cancer patients and non-cancer control patients he found that the lung cancer patients reported significantly more adverse life situations, such as the death of a parent, an unhappy home (mainly due to drunkenness), and a higher incidence of problems with interpersonal relationships, work upsets, financial difficulties, sexual difficulties and a miscellaneous group of adverse life situations that included trouble with the police and a police record. He concluded that long-standing adult difficulties of more than ten years' duration appeared to have a greater influence on the genesis of lung cancer than those of shorter duration.
[5.551] In Kissen and Rao 1969, Kissen's previous work was summarised. It was stated that lung cancer patients had diminished outlets for emotional discharge compared with non-cancer patients. Among the empirical evidence supporting this was the finding that lung cancer patients had a lower mean score on the neuroticism scale of the Short Maudsley Personality Inventory than did non-cancer patients. This was taken to indicate a lesser ability to discharge emotion and was independent of smoking habits. Significant differences between the same lung cancer patients and the controls were later found in the reported incidence of certain environmental and social factors such as adverse events in childhood and adulthood. In this paper, the results of an investigation of certain markers of the endocrine system in lung cancer patients were reported. As described by Professor Idle, they were interested in having objective measurements of stress. One such measurement would be the activity of the adrenal cortex to produce and release a hormone, cortisol, the metabolites of which could be measured in the urine of patients. The authors said that the endocrine system was the obvious choice since it was well known that it was influenced by emotional factors through the hypothalamus, the hypophysis (the pituitary gland) and the autonomic nervous system. The authors predicted that the patterns of urinary steroids would fluctuate in response to the stress experienced by patients during the first days of admission to hospital. They reported that the urinary steroid levels showed markedly different patterns in a group of lung cancer patients compared with a group of controls and they stated, at p.476:
"The results of the present study showed that lung-cancer patients have some abnormalities in adrenal function, compared with control patients, and these abnormalities are related to certain psychological measures that are characteristic of lung-cancer patients."
The authors recognised, however, that urinary steroid levels in lung cancer patients could have been a result of the disease.
[5.552] Rao refined this work, and in 1970 reported on the ratio of two corticosteroid metabolites in the urine, which was approximately half the value in a group of inoperable lung cancer patients compared with groups of healthy controls and hospital controls. The ratio was also reported to be unchanged two weeks after resection of the lung cancer for lung cancer patients undergoing surgery, suggesting that the presence of the tumour was not the cause of the abnormal ratio. This assertion was further confirmed by Rao in 1971, with the additional interpretation of the data that steroid excretion after removal of tumours would include the response to the physical and emotional stress after surgery. Rao's work therefore suggested that lung cancer patients had an altered endocrine system response to both hospitalisation and, when resection was possible, to the surgery itself.
[5.553] Professor Idle said that the overall work of Kissen and Rao during that period showed that both childhood-acquired and adult-acquired psychological factors modified the emotional structure of certain individuals in such a way as to leave them with a poor outlet for emotional discharge, which might be related to the role of corticosteroids, whose release by the adrenal glands was normally associated with stress. The link with lung cancer might be through the immune system, which was impaired after corticosteroid release. Research of this nature continued into the next time period.
[5.554] In the period from 1965 to the end of 1971, therefore, clinical psychologists had taken an increased interest in aspects of an individual's personality and emotional make-up that might be linked to cancer development. One theme that was beginning to emerge, and was developed thereafter, was that internally generated "stress" might lead to perturbations of immune function and consequent cancer.
[5.555] In addition to a continuing effort by psychologists in the period from 1972 onwards to discover and define the personality and emotional factors that might lead to cancer, there started to emerge a large body of work on the relationship between neurological and immunological functioning. A particular focus of this work was to understand the mechanisms by which internal and external stresses might modify immune function and thus lead to cancer. Pointing to the problem of the ignorance of different scientific and medical disciplines of each others' work, Holden wrote in 1978 that potential research had been greatly impeded by the lack of communication between, for example, biochemists and psychologists. There were independent reports in 1979 of findings in humans that pointed to a relationship between stress and the immune system, in relation to cancer causation. Psychologists also began to turn to animal models of stress to investigate these relationships.
[5.556] Professor Idle explained that the immune system provided a very important defence against external and internal foreign agents, not only viruses and bacteria and so on, but also newly developing cancer cells. These could be destroyed by an antibody produced by an immune cell, if the body recognised the cancer cell as foreign, as it sometimes did, or by cytotoxic T cells and natural killer (NK) cells. Visintainer and colleagues described in 1982 the effects of electric foot shock, comparing rats that were able to avoid the shock with rats that were not, on the rejection of transplanted tumour cells: the latter were only half as likely to reject the tumour and twice as likely to die as rats receiving escapable shock or no shock. The authors offered a mechanistic explanation that when an animal was exposed to a stressor, the arousal released adrenocorticotrophic hormone and adrenocortical activation occurred, which in turn caused immunosuppression. The more aroused the animal, the greater the immunosuppression. Subjects receiving inescapable stress might experience greater arousal, as indicated by emotional responses and physiological changes, than subjects receiving escapable stress.
[5.557] Glaser and colleagues reported in 1986 that stress, in the form of examinations in medical students, depressed the amount of interferons produced by white blood cells and also decreased the activity of NK cells, both by a large margin. In their earlier work they suggested that stress might contribute directly to carcinogenesis by impairing DNA repair processes. They concluded that longer-term stress-related alterations in cellular immunity might carry an increased risk for immunodeficiency disorders and malignant and infectious disease. A study in 1989 stated that NK cell activity was significantly reduced in a group of depressed patients, melancholic subtype, compared to sex- and age-matched controls. Another study, in 1990, stated that both major depressive disorder and the presence of threatening life events in control subjects were independently associated with a 50% reduction of NK cytotoxicity. A decrease in natural cytotoxicity was significantly associated with depressive symptoms but not with age, alcohol consumption, or tobacco smoking. Rosch, writing in 1995, said that many brain transmitters that could influence malignant growth also participated in the humoral response to stress, including melatonin, serotonin, dopamine, and the endorphins, and that lower immune defences clearly predisposed to the development of malignancy. Professor Idle explained that the humoral response meant that where stress was experienced a number of different substances were released which passed round the body from one place to another, including the hormones mentioned above, and local hormones. Most recently, Fortes and colleagues reported in 2003 on a study of depression in residents of a home for the elderly over a period of four years. They determined the numbers of various immune cell types in the blood of the subjects, who had also been evaluated on a depression rating scale, and reported that the results suggested an adverse effect of depressive symptoms on immune response. They suggested that immune cell killing of target cells might be impaired, placing these individuals at risk of various diseases.
[5.558] Professor Idle concluded by stating that there was now a burgeoning body of work that had linked personality, emotional problems and stress to the occurrence of cancer and, moreover, it was believed that these factors operated through the immune system. One of the many proposed mechanisms was that cancer might arise when personality, emotion and stress influenced the immune system by reducing number and function of NK cells, together with the neuromodulators, such as interferons, that regulated their activity.
[5.559] Professor Idle next turned to endogenous factors and the theories relating to them. He said that during the period up to 1965, the question arose as to whether cancer was caused by something inside, rather than outside, the body. In 1933, Cook first proposed that aberrant metabolism of an endogenous steroid might explain spontaneous tumours in animals. Cook and his group had previously heated an endogenous steroid known as deoxycholic acid under pyrolysis conditions. He wondered whether the body could carry out some metabolic reactions that would generate a carcinogenic or pro-carcinogenic material from an endogenous steroid. Cramer, a researcher at the Imperial Cancer Research Fund Laboratories in London, wrote in 1934 that the failure to account for the apparently arbitrary selection of victims of cancer was to be found in the fact that until then comparisons had been made taking into account only extrinsic factors and neglecting altogether the complicating influence of the intrinsic factor of susceptibility. He was of the view that the total burden of cancer in different populations was essentially the same, but that endogenous factors determined the organ distribution of the disease. By the end of 1964, therefore, it was appreciated that endogenous factors played a role in cancer and this became a fertile area of future research.
[5.560] In the next period, Coombs and Croft investigated in 1969 how a group of polycyclic aromatic hydrocarbons, the cyclopenta[a]phenanthrenes, might be manufactured within the body itself. Building upon the original ideas of Cook, they proposed that these chemicals could be produced within the body from bile acids, like deoxycholic acid, and they wrote extensively on the chemistry and biochemistry of this process. But Professor Idle had been unable to find literature that showed the end of this line of research, and in particular whether or not this group of carcinogens was produced within the body itself. Separately, Magee and Barnes, the leading researchers in nitrosamine carcinogenesis, questioned whether the bacteria living in the gut might synthesise nitrosamines under certain conditions.
[5.561] This led Professor Idle to introduce the additional topic of host factors. He said that it had long been known that endogenous factors could play a role in cancer. Of particular importance were genetic factors that were associated with disease outcomes in individuals, including cancer. Genetic differences could influence the individual's response to exposure to the ubiquitous chemicals that were found in foodstuffs, pharmaceuticals, cosmetic products and drinking water, for example. These genetic differences were manifested as genetic polymorphisms. Genetic polymorphism was the existence in the same location of two or more forms of a characteristic, where the form of lowest incidence was greater than 1%. Different blood groups, eye colours or skin colours were all characteristics which were almost invariably genetically determined, and each formed a part of a polymorphism, or multiple types of a particular characteristic. Human characteristics such as height and blood pressure were the product of a large number of genes (and their environmental interactions) and were thus not subject to polymorphism.
[5.562] Asked further about genetic polymorphism, Professor Idle explained that the 30,000 genes in the human genome contained sequence variations between different individuals, so that there were literally thousands and thousands, perhaps millions, of genetic differences in the human genome between different individuals. This was not recognised twenty or thirty years ago, but certainly was today, as the result of the human genome project. So, at almost any genetic locus, there existed multiple alleles, alternative forms of the gene, in the population. These included not only such matters as the colour of eyes, hair or skin, or blood groups, but all the biochemical characteristics that made up our cells: the enzymes, the structural proteins, even lipids and carbohydrates, were there by virtue of genetic characteristics. No two persons, with the exception of identical twins, were genetically identical.
[5.563] In 1968, Nebert and Gelboin at the NCI reported that the activity of an enzyme, which they renamed aryl hydrocarbon hydroxylase (AHH), could be induced by the treatment of rodent cells in culture with a polycyclic aromatic hydrocarbon such as benz[a]anthracene. They reported that different strains of mice had different degrees of induction of AHH. This work represented the first study on the genetics of the cytochrome P450 (CYP) system. Professor Idle explained that cytochrome P450s were a family of enzymes that occurred in all animals and plants that had so far been studied, including humans. These enzymes carried out the metabolic oxidation of virtually every foreign chemical that entered the body, including many nutrients, and were extremely important in the body's defence against small molecules. P450, in adding oxygen, generally rendered the molecule more water-soluble, or created a handle on the molecule, to which the body added a water-soluble substance, such as a sugar molecule. Several things were therefore produced overall, in one or two phases: the pharmacological or toxicological activity of the material was reduced, its water solubility was increased and its ability to be excreted from the body was increased. So generally, but not always, water-soluble, excretable metabolites are produced. The levels of the enzyme could be increased by an external stimulus by a factor of between two- and ten-fold, depending on the circumstances. The experiments had been carried out almost exclusively on laboratory rodents. The increase in the amount of the enzyme in general had the effect of increasing the rate and extent of metabolism of chemicals that were substrates for that enzyme. It appeared to be AHH that metabolically activated polycyclic aromatic hydrocarbons, such as benzo[a]pyrene, to reactive, carcinogenic metabolites, and it was observed that the polycyclic aromatic hydrocarbons themselves could induce AHH and thus their own metabolic activation.
[5.564] The report by Nebert and Gelboin that there were mouse strains for which AHH was inducible and others for which it was not led to the hypothesis that chemical carcinogenesis might be genetically controlled and gave rise to the proposition that different people might react differently to chemical carcinogens. The first report of human studies of this paradigm came from Kellermann, Shaw and colleagues, who concluded that susceptibility to bronchiogenic carcinoma was associated with the higher levels of inducible AHH activity. But over the subsequent two decades, no other research group was able to duplicate these findings and their work was discredited.
[5.565] Professor Idle said that there was broad agreement that what first energised the field of what might now be called cancer pharmacogenetics was the discovery by his own group of the so-called debrisoquine 4-hydroxylation polymorphism and its subsequent application to lung cancer. They were interested in the early to mid-1970s at St Mary's Hospital Medical School in London in the reasons why some patients responded to the drug debrisoquine, which was a new antihypertensive agent, and some patients did not. They could see clearly that this might be a metabolic phenomenon and they worked in an environment of drug metabolism. They set about doing a study in medical students and some members of staff in which debrisoquine was given to every subject and measurements were made of the amount of debrisoquine and its principal metabolite, the pharmacologically-inactive 4-hydroxy-debrisoquine. They made a ratio of the debrisoquine divided by 4-hydroxy-debrisoquine and expressed this in a histogram of the population. They found that 3% had very high values of this ratio, while the rest all had low values. They proposed then that these were two different phenotypes which they called "extensive" and "poor" metabolisers. They then carried out family studies on the individuals with the high values to see if there was a clustering of other poor metabolisers in their families and were able to demonstrate that the character known as "poor metabolism" was inherited, and inherited in a recessive fashion.
[5.566] Their research developed in many directions, one of which was to carry out a study in lung cancer using the technology that they had developed for typing people as either extensive or poor metabolisers. A study was carried out on lung cancer patients at the Whittington Hospital in London, where there was a large and active chest clinic, and on controls who were matched for age, gender and smoking history. They carried out the debrisoquine hydroxylation test on the subjects and measured their debrisoquine metabolism in the laboratory. They found that the lung cancer patients differed from the controls in two major respects. First, the cancer group comprised many fewer subjects who were poor metabolisers than did the control group; this suggested that poor metabolisers, who made up almost one-tenth of the United Kingdom population, were at considerably reduced risk of lung cancer. Secondly, as shown on a histogram to which Professor Idle made reference, the metabolic ratio values were shifted to the left in the cancer group, suggesting that the persons with the lowest values of metabolic ratio (fastest metabolism) were at the highest risk of the disease. Professor Idle's group did not at that time include an epidemiologist, so the terminology of epidemiology was not used by them. Their early investigations had now been repeated and refined by them and many others in numerous populations and patient groups. Other groups had not, however, always found the same result and some had reported no difference between lung cancer patients and controls. The position today was that there was still a debate as to what role this enzyme might play in lung cancer, it was quite uncertain.
[5.567] Many other host factors had since been identified, he said, some of which were derived from enzymes thought to be involved in chemical carcinogen activation and detoxication and others which were related to DNA repair enzymes. For example, there were now known to be a number of P450 enzymes that displayed genetic polymorphism and these were beginning to be evaluated as factors in human cancer, including lung cancer. Debrisoquine hydroxylase, know called CYP2D6, was one of a large family of enzymes, among which were other enzymes of which humans and animals were either extensive metabolisers or poor metabolisers. Investigators were thus provided with far more tools for their studies. Moreover, other groups of non-P450 enzymes that metabolised chemicals foreign to the body had been reported to be polymorphic. Professor Idle said that as science delved deeper into the genetics of these enzyme systems it was becoming apparent that most persons differed from each other in their complement of chemical metabolising enzymes. Additionally, other genetic polymorphisms related to DNA repair enzymes, oncogenes and tumour suppressor genes were starting to emerge. The various aspects of the field of molecular epidemiology of cancer were now very active, involving hundreds of research groups. Approximately 300 genes were thought to be involved in the cancer process, so it was not surprising, he said, that workers looked for differences at these gene loci in cancer patients in an attempt to understand who was at greatest risk of the disease. There was now an enormous compendium of data from such studies. Some of the research had been directed at uncovering associations between certain polymorphisms and lung cancer, but no gene had yet been discovered that explained why certain persons developed lung cancer whilst others did not.
[5.568] The next topic was the stages of cancer development. Professor Idle said that Berenblum was the first to report, in 1941, that certain irritants, notably croton oil, turpentine and xylene, augmented the carcinogenicity of benzo[a]pyrene on mouse skin. Later, Friedewald and Rous, writing in 1944, coined the terms "initiation" and "promotion" of skin carcinogenesis to explain the co-carcinogenic action of benzo[a]pyrene and chloroform, respectively. Professor Idle explained that co-carcinogenesis occurred when two substances acted together, often in a multiplicative rather than an additive manner, to increase the incidence of experimental tumours. In those days there was little understanding of the molecular mechanisms and the theories were based on good observation. It was observed that if polycyclic aromatic hydrocarbons, specifically benzo[a]pyrene and 20-methylcholanthrene, were applied to a rabbit ear, for example, and some cells were initiated to become cancer cells, the subsequent application of chloroform appeared to promote the appearance of cancer in these cells. These polycyclic aromatic hydrocarbons were labelled as initiators and described as "calling forth" the skin tumours. Solvents and irritants were labelled as promoters, in that they hastened the formation of the skin tumours on rabbit ears after the application of the initiator, but were not carcinogenic when applied in isolation. Some chemicals were believed to possess the qualities of both tumour initiators and promoters and were thus known as "complete carcinogens", meaning that they could be carcinogenic alone, without the intervention of any other agent. Hieger reported in 1962 that croton oil, formerly thought only to be a promoter, produced sarcoma in a certain strain of mice, which suggested that croton oil might have been misclassified as a tumour promoter. This demonstrated that it was not always easy to distinguish between an initiator, a promoter and a complete carcinogen. Boutwell, after carrying out a large number of detailed experiments with turpentine on mice, concluded in 1964 that carcinogenesis might be separated into three stages, involving initiation, conversion and propagation. Conversion was seen by him as the change of an initiated cell which appeared to be normal into a malignant-appearing cell, which then went on to form a tumour through the process of propagation.
[5.569] Based on observations of the results of skin painting in laboratory animals, by the end of 1964 the two-stage theory of carcinogenesis, that is initiation and promotion, had been broadly accepted. Professor Idle added that the only two-stage animal models were mouse skin, bladder and thyroid, and rat liver. Nevertheless, the model was erroneously referred to when discussing other tumour types in other organs. There was, he said, no established two-stage lung cancer model and today very few researchers accepted a two-stage model. Today, in general, researchers would speak of a multi-stage model. The vast majority of experiments, in the model itself, as used today, were on mouse skin. Even with mouse skin, researchers would speak of there being multiple stages, not just simply two or perhaps three.
[5.570] In the subsequent period, although the two-stage model had only been shown to operate in skin cancer in experimental animals, Wynder and Hoffmann proposed in 1967 a three-stage mechanism for human lung cancer comprising a disruption of cilia-mucus defence, followed by initiation and promotion. This theory was pursued by these authors in subsequent papers. Professor Idle said that certainly in the 1970s and 1980s there was a growing recognition that there must be more than two or three stages involved in the carcinogenic process and experiments were producing data consistent with these developing ideas.
[5.571] The next concept which Professor Idle introduced was that of anticarcinogens. The expression "anticarcinogenic" was coined by Berenblum in 1931 to relate to a phenomenon first reported by him in 1929. He found that the production of "warts" (papillomas) on the skin of mice by coal tar could be almost completely prevented by the addition of a small proportion of mustard gas to the tar. This was a surprise, because both of these substances were known to be carcinogenic. He later demonstrated that a number of other skin irritants had marked anticarcinogenic actions with respect to coal tar. Lacassagne and others reported in 1945 that weakly carcinogenic polycyclic aromatic hydrocarbons could inhibit the effects of highly carcinogenic polycyclic aromatic hydrocarbons on mouse skin. Falk and others reported in 1964 that non-carcinogenic polycyclic aromatic hydrocarbons and crude mixtures containing a spectrum of polycyclic aromatic hydrocarbons could inhibit the appearance of sarcomas in mice injected with the polycyclic aromatic hydrocarbons benzo[a]pyrene and dibenz[a,h]anthracene. Accordingly, as at the end of 1964, Professor Idle said, it was known that experiments with chemicals in isolation could not predict their biological activity when they occurred in combination with other chemicals. Reference was made to Falk et al. 1964. At p.178 the authors summarised their findings in these terms:
"The activity of carcinogenic hydrocarbons can be inhibited by closely related pure compounds or by crude mixtures including a spectrum of polycyclic aromatic hydrocarbons when both carcinogen and anticarcinogen are administered simultaneously. Though the effect is maximum when the agents are administered simultaneously, inhibition can also be observed when the anticarcinogen is administered several days before or after the administration of the carcinogen."
The authors said that in the absence of an understanding of the mechanism of carcinogenesis on the intracellular level, any explanation of inhibition would be compounding a theory upon a theory.
[5.572] On the topic of genetic differences, Professor Idle said that the period from 1965 to 1971 witnessed the emergence of a belief that genes, either alone or in concert with other factors, could cause human cancer. In addition, this period saw the development of the field of pharmacogenetics, in which it was recognised that genetic differences could have effects upon the metabolism of chemical carcinogens. Heston in 1965 discussed the role of genes in the origin of all tumours. He was particularly convinced that the so-called "spontaneous tumours" in animals were due to genetics. Gelboin and Bates, writing in 1966, were convinced that cancer was the result of modification of key genes by carcinogens. This view was shared by Huebner and colleagues, writing in 1971, who envisaged chemical carcinogens and viruses sharing a common mechanism by working through the genes of the host organism to cause cancer. This opened the way for consideration of gene-environment interactions in cancer causation. Koller, writing in 1967, was convinced of the central role played by genetics in carcinogenesis. Lynch wrote in 1968 that although the organism interacted with the environment, the critical point was that the genotypic endowment of the host might determine resistance as well as susceptibility to cancer; the aetiology and pathogenesis of cancer in man were, he wrote, extremely complex and for the most part remained an enigma. Keith and Brown in 1970 analysed the differences in cancer incidences between monozygotic (identical) and dizygotic (non-identical) twins and remarked that environmental factors appeared to be more important than genetic factors, but the data were not available to make a clear judgment. Lynch and Krush, in relation to colon cancer, remarked in 1971 on the importance of acquiring family histories from cancer patients. It had not previously been common practice to take family histories from such patients, and this limited the understanding of the role of genetics in cancer. During this period, Knudson published in 1971 one of the most important papers on cancer genetics. He analysed familial retinoblastoma, a childhood cancer of the eye known to have a dominant inheritance (meaning that all affected persons had at least one affected parent) and calculated that the genesis of the tumour required two mutations, one inherited and the other occurring in the early years of life. This theory led some cancer researchers to the view that cancer did not require a number of mutational events as other researchers had earlier claimed.
[5.573] In this period, research aimed at investigating the influence of genetics on lung cancer was in its infancy. Severi had observed in 1965 that the appearance of lung tumours in mice was under the influence of at least 10 known genes. In the same year, Hueper studied the role of genetics in resistance to benzo[a]pyrene in mice, but his results were equivocal. Burch commented in 1964 on the report by Tokuhata in the same year that lung cancer deaths among non-smoking first-degree relatives of lung cancer cases were four times greater than in the non-smoking first-degree relatives of controls without lung cancer. Burch recognised that the proportion of lung cancer cases in the population that arose due to a "specific predisposition" could not be accurately estimated, but he nevertheless used an estimate of 50%, which allowed him to calculate that the frequency in the population of any particular allele predisposing to lung cancer was likely to be appreciably less than 3%. (An allele was any one of two or more different genes that occupied the same position (locus) on a chromosome, such as the alleles which determined eye colour.) So, by the end of 1971, Professor Idle said, researchers were exploring the possible role of genetics in human cancer, including lung cancer.
[5.574] Professor Idle summarised his research into the period prior to 1965 by saying that there were competing theories regarding the origins of cancer. Among these theories were those concerned with the possible roles of chronic irritation, viruses, macroparasites, chemicals, radiation, personality, emotion and stress, embryonic tissue and cellular metabolism in cancer causation. It was also postulated that endogenous factors played a role, albeit undefined, in cancer causation. It was in this period that it was first proposed that carcinogenesis occurred in two, or possibly three, distinct stages. There was also considerable debate concerning the cellular target of chemical carcinogens, either protein or DNA. Such a debate was only made possible by the discovery of the structure and function of DNA by Watson and Crick in 1953. Finally, chemicals were identified that had anticarcinogenic properties. In 1964, scientists who were struggling to identify the causes of cancer were also seeking to define the mechanisms involved in cancer formation. Boutwell, who was a senior investigator at the McArdle Memorial Laboratory, wrote in Boutwell 1964, at p.207:
"The problem of cancer has received more intensive research effort than any other single medical problem, yet the essential facts needed for understanding and controlling the disease continue to elude us."
At p.211 he wrote:
"The ultimate comprehension of the problem of cancer is dependent upon elucidating the chemical processes responsible for the biological phenomenon known as cancer. However, the voluminous efforts of the biochemists have not been successful, partially because the biological descriptions are still incomplete."
[5.575] Professor Idle said that in his opinion, as at the end of 1964, none of the competing theories of cancer causation had gained general acceptance. The cause or causes of cancer were unknown and there was little understanding of the mechanisms whereby cancer in general, and lung cancer in particular, developed. During the period from 1965 to the end of 1971, competing theories of cancer causation continued to evolve, and new theories emerged. Substantial research was devoted to investigating how viruses might cause human cancer. Progress was made in understanding the metabolism of chemicals. The possible role of radiation as a cause of cancer continued to be explored. Researchers continued to investigate whether or not the body itself might produce cancer-causing chemicals. Continued research was carried out into the roles of personality, emotion and stress in cancer causation. Wynder and Hoffmann, building on the two-stage model of carcinogenesis, proposed a three-stage model of human lung cancer. Debate continued on the nature of the cellular target for chemical carcinogens. Finally, researchers started to consider the role of genes in cancer susceptibility.
[5.576] During this period, there emerged no single hypothesis as to how cancer was caused. Horsfall, a virologist and Director of the Sloan-Kettering Institute, one of the premier cancer research establishments in the United States, wrote in 1966 that the necessary and sufficient causes of cancer were still unknown and that their elucidation was the pressing priority for cancer research. There were those who said that knowledge of the nature of cancer had not advanced in 25 years. Indeed, in 1967, Magee and Barnes, the world experts on nitrosamine mutagenesis and carcinogenesis, were still proceeding on the historical hypothesis that cancer followed a mutation in a somatic cell. As with cancer in general, during the period 1965 to 1971 there was no single hypothesis as to how lung cancer was caused. The importance of the biological approach to cancer research was underscored by Roe, who was a pathologist working in London, associated with the Chester Beatty Research Institute and its attendant hospitals. In Roe 1966 he stated, at p.101:
"The epidemiologist can only find associations, he cannot approach very close to proving cause-and-effect relationships. He has no control over many possible relevant factors in the environment of the populations he studies. The experimentalist, on the other hand, can control many more factors in the environment of his test animals. He can come much closer to proving that exposure to a particular agent causes a particular effect. Secondly, the experimentalist is in a position to investigate the mechanism by which the effect is brought about."
Professor Idle added that he did not believe that at the present date the necessary and sufficient causes of cancer, referred to by Horsfall in 1966, were yet known.
[5.577] In summary, Professor Idle said, in the period up to 1972, there were competing theories regarding the origins of cancer. These included viruses, chemicals, radiation, endogenous factors, and personality, emotion and stress. The nature of a chemical carcinogen in structural terms was not understood, but it was accepted that chemical carcinogens needed to be metabolised to exert their effects. The proposition that such metabolism was under genetic control was being investigated, and this research would continue into the next period. Similarly, the molecular target for such carcinogens had also not been agreed, and there was continuing debate whether this target was protein or DNA. The role of genetics in cancer was beginning to be appreciated. In his opinion, as at the end of 1971, none of the competing theories of cancer causation had gained general acceptance. The causes of cancer were unknown and there was little understanding of the mechanisms whereby cancer in general, and lung cancer in particular, developed.
[5.578] Throughout the period 1972 up to the present, Professor Idle went on to say, a number of competing theories had been proposed regarding causation of cancer, including lung cancer. The scientific views that had been advanced during this period involved viruses, failure of host immunity, personality, emotion and stress, failure of DNA repair, host genetics, chemical carcinogenesis, mutations in oncogenes and tumour suppressor genes, oxidative stress, epigenetics and aneuploidy. It could be seen from this list that some of the theories that were being explored in the earlier periods continued to be explored in this period while some had been discarded, and other theories had emerged during the most recent period.
[5.579] He said that this period had witnessed an explosion of scientific activity and publications on the subject of the biology of cancer. Both basic and clinical scientists from fields as diverse as chemistry, biochemistry, physiology, haematology, immunology, bacteriology, virology, endocrinology, cell biology, molecular biology, pharmacology, toxicology, genetics, oncology, therapeutics, radiology, internal medicine, surgery, psychology and psychiatry had come together under the single umbrella that was cancer research today. Additionally, we had witnessed in the recent past the creation of new subject areas both in research and in undergraduate curricula, which provided both a focus and a framework for the evolving concepts of biology that could be utilised to tackle the as yet unsolved questions of cancer causation. Such new subject headings included the hybrid disciplines of pharmacogenetics, genomics, proteomics, metabonomics, biochemical and molecular epidemiology, molecular virology and molecular toxicology. Professor Idle explained that genomics was a subject that dealt with the genome, using information derived from the human genome project of genetic sequencing. Proteomics was the second stage after genomics and was a subject which dealt with variation and behaviour in protein molecules and what that told us. Metabonomics was the study of the complete complement of small molecules and their flux within a cell. Epidemiology underwent a transformation some years ago with the recognition that it would be useful to make objective measurements, rather than simply use the instrument of questionnaire. Biochemical markers were measured, for example in blood or urine, and this formed the field of biochemical epidemiology, for example in a case-control study. Molecular epidemiology used molecular markers in the same setting, particularly genetic factors measured in case-control or cohort-type studies. This re-ordering of the landscape and constituencies of biological research and teaching had become necessary with the realisation that questions regarding the nature of cancer had produced yet more complex questions that defied investigation, within the confines of a single traditional subject area. The multidisciplinary and often multicentre approach to cancer research today was a testament to the increasing complexity of the question of cancer causation.
[5.580] Since 1953 research centred around DNA had steadily increased and, by today, dominated much of the activity in cancer research. Professor Idle explained that DNA was the molecule that passed to the next generation the information of how to build and operate a new organism. It was found in the nucleus, the cell body that contained the chromosomes, and, to a lesser extent, the mitochondria, the small intracellular bodies responsible for energy production and cellular respiration, of living cells. DNA comprised four "bases": two purines, adenine (A) and guanine (G); and two pyrimidines, thymine (T) and cytidine (C). The bases were joined in a chain by a "sugar-phosphate backbone", a polymeric thread of sugars and phosphates which were joined together, with the bases attached to them. When two strands of the single-stranded DNA polymer coiled together to form double-stranded DNA, known as the "double helix", the structure was held together by obligatory pairings between G and C, and between A and T. In this way, the second strand was a mirror image of the primary strand, containing within it all the information the cell needed to synthesise a new primary strand. The copying, from the secondary strand, of every A into T, every T into A, every G into C and every C into G, produced an identical copy of the original primary DNA strand sequence. The primary and secondary strands were known as "complementary" DNA strands. When a cell divided, the primary strand was used as a template to make a complementary copy, and the secondary strand was also used to make a complementary copy, the two newly synthesised strands re-combining to form a second double helix, to be used by the daughter cell. Professor Idle referred to a figure illustrating this process, part of the primary strand having the sequence TAGCTACG, and part of the secondary strand having the sequence ATCGATGC. When the primary strand was copied, by a process called "replication", into a complementary strand, its sequence would be the same as for the original secondary strand, and when the original secondary strand was replicated, the resulting complementary strand would have the same sequence as the original primary strand. Thus, the two new single strands of DNA were able to pair into a double helix, having replicated the complete genome of the organism, and so providing the genetic basis for mitosis, or cell division.
[5.581] The scientific era since 1972 had also witnessed a number of discoveries in the development of revolutionary laboratory techniques that had radically altered the way in which cancer research could be pursued today. One of the most important techniques that emerged in the mid-1980s was the polymerase chain reaction (PCR). PCR enabled tiny amounts of DNA to be detected by amplification of particular sequences of interest, so that an individual might be identified from the DNA contained in a single cell. The discovery of the proto-oncogenes and the tumour suppressor genes prompted new lines of research on cancer mechanisms. More than anything, the observation of single mutations in these genes led scientists to hypothesise that the carcinogenic process was a set of clearly discernible cellular events, each of which could be defined in molecular terms. It had not, however, been possible to reduce cancer to a linear set of simple genetic alterations.
[5.582] In 1982, Weinberg made the landmark discovery of the first mutated and activated oncogene in a human tumour. In the following three decades, research demonstrated that there were approximately 300 genes which might be involved in human cancer. About half of these were oncogenes and tumour suppressor genes. The other half coded for substances known as cytokines, growth factors and their receptors, which were also involved in the regulation of cell growth. Professor Idle said that cytokines were another group of messenger molecule which passed from one cell to another, often within the immune system. Despite these major advances in knowledge, however, human cancer, as a clinical entity, still evaded scientific definition and understanding. Twenty years after his discovery of the first activated oncogene in a human cancer, Weinberg wrote in 2002:
"For those who believe in the simplification and rationalization of the cancer process, the actual course of research on the molecular basis of cancer has been largely disappointing. Rather than revealing a small number of genetic and biochemical determinants operating within cancer cells, molecular analyses of human cancers have revealed a bewilderingly complex array of such factors."
[5.583] One of the new theories in the period after 1972 was the failure of host immunity theory. Professor Idle explained that the body's key defence system was the immune system. This was composed of many different cell types that collectively protected the body from bacterial, parasitic, fungal and viral infections and from the growth of tumour cells. The immune response was based upon the paradigm of self and nonself, i.e. that there was an immune response to anything recognised as foreign or nonself. Thus, invading organisms and tumour cells were recognised through their foreign proteins ("antigens"), and part of the body's response to antigens was to synthesise "antibodies" that neutralised them. This was the principle, for example, of vaccination, which introduced a specific antigen into the body to stimulate the production of a specific antibody.
[5.584] The Australian immunologist and subsequent Nobel Laureate, Sir Frank Macfarlane Burnet, evolved the principal theory in this field. Stated simply, in his hypothesis of immunological surveillance, clones of malignant cells were thought to arise at regular intervals in all individuals; but such cells hardly ever survived because they were rejected as allografts (foreign), unless some unspecified defect in the immune response permitted these spontaneously arising clones of malignant cells to be tolerated. This theory was favoured in the early 1970s but, as Laroye wrote in Laroye 1974 at p.1099:
"All arguments about immunological surveillance tend to postulate that neoplastic cells acquire tumour-associated antigens which subsequently can become susceptible to immunological attack. This does not seem to be true in all instances: many tumours, indeed, especially those induced by chemical carcinogens, seem to be devoid of any significant tumour antigenicity."
Professor Idle said that this observation was based upon studying these types of cells in vitro and by transplant experiments into animals, and the recognition that tumours, particularly those caused by chemicals, could sometimes give themselves an "immune privilege" around themselves. Epidemiologists concluded that there was some evidence that failing immunosurveillance might be related to skin cancer and adenocarcinomas of the stomach and lung.
[5.585] One of the strong arguments in favour of the immune surveillance theory of cancer, he said, was the mounting incidence of cancer with increasing age, an approximate doubling every five years after the age of 25, and the parallel decline in immune competence with increasing age. Added to the age factor of the observations in the immune deficiency syndromes, whereby mortality rates for cancer in immune deficiency patients exceeded by 100 times the expected rates, there were inherited immune deficiency syndromes in addition to the acquired immune deficiency syndrome (AIDS). There were those who had observed a lower death rate from cancer in persons who were given BCG vaccination against tuberculosis at birth. Reference was made in this regard to Rosenthal 1984, in which it was reported that in various studies it had been found that BCG vaccination at birth lowered the mortality from leukaemia and, in the author's studies, to all forms of cancer followed over a period of twenty years. The author proposed that the mechanism of cancer mortality reduction in those vaccinated with BCG at birth was at least in part due to the detection and destruction of embryonic cells or components thereof. He wrote that in the embryo thousand of errors were made in the development, most of which were corrected before birth but thousands of which remained. These foetal remnants acted as foreign bodies to the host and were removed by the lymphoreticulo-endothelial system, which could be markedly stimulated by BCG, which thus helped appreciably in removing foreign cells from the body. It had been proposed in another paper that BCG vaccination enhanced the tumour cell killing efficiency of two types of white blood cell, macrophages and lymphocytes. Kripke 1988 stated that the immune surveillance theory, once highly controversial, had survived over three decades and had been modified as our understanding of the immune system developed. Nevertheless, there were those who did not accept that "immune senescence" contributed to this increased risk of cancer with increasing age.
[5.586] Professor Idle said that during the 1990s, there was still much discussion of the immune surveillance theory of cancer. The theory was restated in simple terms by Chowdhury and others in 1991, with the concluding statement that the survival and growth of the tumour cells depended crucially on how the tumours evaded the immune surveillance mechanism of the host and escaped the killer cells as well as the antibodies.
[5.587] The development of understanding of the immune system had led to the awareness that three immune cell types were involved in the control of neoplasia. First, the NK cell was a type of white blood cell that attacked tumour cells by binding to them and releasing lethal chemicals. NK cells were thought to play a key role in cancer prevention by killing abnormal cells before they multiplied and grew. Secondly, the macrophage was a phagocytic cell that played an important role in killing tumour cells and releasing substances that stimulated other cells of the immune system. Phagocytic cells engulfed foreign particles, such as bacteria, and destroyed them using bursts of reactive oxygen species. Thirdly, the tumour-infiltrating T cell (or cytotoxic T cell) was a type of white blood cell that could recognise and destroy abnormal cells, such as tumour cells and virus-infected cells. All three immune cell types had been observed in vitro, using human cells. Immune response to a foreign organism or to a tumour cell, therefore, was an integrated response involving multiple cell types. It had been reported that cancer patients may have a deficiency in their T cell responsiveness and that injection into the skin of lung cancer patients with bacterial or chemical antigens demonstrated a clear immune impairment. Reference was made to Gorski et al. 1994.
[5.588] There were those, nevertheless, who rejected the immune surveillance theory of cancer, which had been premised on the concept that the body's immune system erroneously recognised a tumour as "self" and thus did not attack and destroy it, allowing the tumour to "sneak through" immune surveillance. Fuchs and Matzinger, writing in 1996, argued for rejection of the self/nonself paradigm. They proposed that this paradigm provided a satisfactory explanation in the context of infectious disease originating from foreign organisms. However, they postulated that the default response of the immune system to new antigens, including those produced by a tumour, was to ignore them, in the absence of a secondary "danger signal", arising from tissue damage or distress. This, they explained, was why the administration of the BCG tuberculosis vaccine, which supplied the "danger signal", seemed to be an effective therapy for certain tumours. As if to demonstrate the lack of a scientific consensus on this topic, Kavanagh and Carbone catalogued in 1996 all the possible defects in immune surveillance that could permit a tumour, including specifically a lung tumour, to "sneak through".
[5.589] Meanwhile, new mechanistic theories of how certain types of tumour cells, including lung carcinoma cells in culture, could become antigenic were continuing to be advanced. The elevated incidence of lung cancer in HIV-positive patients and the fact that AIDS patients could have a 3.8-fold elevated risk of lung cancer compared with persons without AIDS merely added weight to the immune surveillance theory. It appeared that lung tumours could utilise a variety of mechanisms to evade immune detection. It had also been argued that an individual's human lymphocyte antigen (HLA) type, known commonly as "tissue type", might predispose that individual to cancer. The HLA type could be thought of as the white cell equivalent of ABO blood groups, except that there were many more HLA types than ABO blood types, running into the thousands. People with certain HLA types appeared to be "cancer prone", further suggesting a role for the immune system in the development of cancer. Furthermore, it had been proposed by O'Connell and colleagues, writing in 1999, that tumours could generate an environment of "immune privilege", through a process called the "Fas counterattack". Fas was a receptor present on cells of the immune system. In the ordinary process of the work of the immune system, the Fas receptor interacted with the "Fas-ligand", which had been produced by either that cell itself or some other immune cell. The Fas-ligand signalled to the cell that it was time to die, the process of apoptosis. The immune system, therefore, having mobilised immune cells to fight an infection or neoplasia, terminated this assault when these cells were no longer required. Certain tumours themselves had the capacity to produce the Fas-ligand, which engaged and destroyed immune cells, thus creating an environment of immune privilege for the tumour cell.
[5.590] Several factors are known to impair the functioning of the immune system and, in so doing, might place the individual at greater risk of cancer. The immune system had been shown to be impaired by the presence of various diseases, including common diseases such as the common cold and mononucleosis. Many drugs could lower the immune defences, and some were indeed frankly immunosuppressant. Experiments in animals, together with clinical observations, demonstrated that immunosuppression occurred as a result of malnutrition, and that nutrition was a critical determinant of immunocompetence and risk of both illness and death. Another important factor was stress. Overall, the relationship between tumours and the immune system had been described elegantly, in Professor Idle's view, by Pettit et al. 2000, at p.1903:
"In summary, the process of immune selection may be considered to occur in three non-discrete phases [...]. During the first phase, initiation, proliferation and diversification of tumour occurs against a background of immune ignorance. The second phase is entered when an anti-tumour immune response is activated. Consequent immune selection pressure allows a process of microevolution to lead to the gradual acquisition of adaptations that permit immune evasion. The third phase of immune escape is reached when tumour cells acquire sufficient adaptations to allow proliferation unchecked by immunological control."
Professor Idle said that the situation today was that much of Burnet's original concepts, having been ignored for some time in the past, were now again part of mainstream immunological thinking. Thus, there were many cancer immunologists who held fast to the concept that tumours might develop means to evade detection and that this was an essential component of the process of human cancer. It still remained to be determined to what extent stress might impair the immune system and contribute to lung cancer.
[5.591] Professor Idle next discussed the failure of host DNA repair theory. He said that many types of chemicals modified the structure and sequence of cellular DNA and its functions. The repair of DNA modifications corrected most of these changes. The chemicals that modified DNA were both endogenous and exogenous and the DNA sequence changes they caused were normally repaired by DNA repair enzymes. Changes of sequence might involve a piece of a chromosome moving to another place in the chromosome, or even to a different chromosome, in a process known as "translocation", or the substitution of one base for another, a process known as "point mutation". Most DNA repair occurred during cell division. The apparatus recognised that the strands did not agree, and the mismatches were repaired and the correct sequence produced. DNA repair enzymes were really good at removing bases that had been modified in their chemical structure and replacing them quickly with new ones, a process which happened tens of thousands of times in each cell every day. This was a tremendously important process which happened constantly to maintain the fidelity of DNA division and cell division. Endogenous DNA modification occurred continuously, not only through the effects of oxidative stress on DNA, but due to spontaneous hydrolysis of the DNA molecules, a phenomenon that occurs in the order of 10,000 events per cell per day (Rouse and Jackson, writing in 2002). It had now been hypothesised that cancer arose when a modified cell by-passed the system that detected when the cell's DNA was modified and signalled this to the DNA repair enzymes. How these checkpoints normally relayed information to the correct DNA repair system was unknown, and Rouse and Jackson had concluded that the question of how lesion-specific repair factors were initially directed to DNA lesions also represented a major challenge for future research.
[5.592] As a major defence against modification of DNA, processes of DNA repair were present in all organisms that had been examined, including bacteria, yeast, fruit flies, fish, amphibians, rodents and humans. A viable cell needed to be able to detect DNA modifications and signal this to the DNA repaired apparatus. Different enzymes repair different types of DNA modification; a total of 130 discrete enzymes had been shown to participate in human DNA repair. Among these were, first, the synthesis of aberrant proteins arising from the modification of DNA that might be toxic to the cell: the potentially deleterious effects of such DNA modifications could be seen when they were inherited, such as in genetic diseases, for example cystic fibrosis. A second consequence might be the persistence of the DNA modification which, if it occurred in a germ cell (a cell specialising in the production of ova or sperm), would be passed on to subsequent generations. This could possibly give rise to a teratological outcome, the formation of a foetus with some gross abnormality. Thirdly, there might be errors in the process of replication involved in the course of mitosis, which errors would be passed onto the daughter cell and subsequent generations. Fourthly, chromosomal rearrangement might be triggered that scrambled the cell's genetic information, a process that was associated with cancer. Fifthly, there might be apoptosis.
[5.593] Abnormalities in the processes of DNA repair had been associated with both cancer and ageing. One of the commonest cancers in the western world was colorectal carcinoma, 5% to 8% of which comprised an inherited sub-group known as hereditary non-polyposis colon cancer (HNPCC). This was exclusively a disease of deficient repair of a specific type of DNA modification that occurred during replication of the DNA molecule. The most deleterious type of DNA modification to the cell was the double-strand break that could be caused by radiation and by certain chemicals. A double-strand break was damage to the double helix of DNA such that both strands had been severed, thus breaking the continuity of the genetic sequence. In inherited disorders of DNA repair, such as ataxia telangiectasia, patients had an increased incidence of cancer. The commonest cancer in persons of European origin was skin cancer, which had been associated with exposure to UVB radiation from sunlight. Persons with the inherited disease xeroderma pigmentosum, who had a deficiency in a type of DNA repair mechanism, were extremely sensitive to both sunlight and strong artificial light with a UVB component. They readily developed squamous cell carcinomas of the skin in areas, such as the head and neck, which were exposed to the light.
[5.594] The role of DNA repair enzymes in the cell's defence against carcinogenesis was complex and not fully understood. For some time it had been possible in the laboratory to make a so-called "gene knock-out mouse", using embryonic stem cells from which genes had been deleted (a technique that had not yet proved possible with any other species). This enabled direct observations to be made of the exact function of each gene. A knock-out mouse lacking the xeroderma pigmentosum group A gene (XPA) had been made; these animals were deficient in a specific type of DNA repair mechanism. Irradiation of these mice and control mice with UVB showed the importance of DNA repair in the genesis of skin tumours. Additionally, these studies showed that the development of skin tumours in these DNA repair-deficient mice also required an element of immunosuppression that was caused by the UVB radiation. Thus, skin cancers in xeroderma pigmentosum patients might arise due to a perturbation of two distinct defence mechanisms, DNA repair and immune surveillance.
[5.595] A further DNA repair system, called XRCC1, might play a role in squamous cell lung cancer. It was reported by Park and colleagues in 2002 that an inherited deficiency in XRCC1 was associated with squamous cell carcinoma of the lung. Other studies had also shown associations between deficiencies in DNA repair and lung cancer. Attenuated DNA repair in combination with urban air pollution had been proposed as the source of DNA modifications.
[5.596] Professor Idle concluded this passage by saying that deficiencies in DNA repair had been recognised as playing an important role in a number of human cancers and might play a role in lung cancer. Investigation of DNA repair was one of the fastest growing fields in cancer research. In fact, in a paper published in September 2003, Paz-Elizur and others reported that there was 18% lower activity of the DNA repair enzyme OGG (which repaired DNA modifications caused by oxidative stress) in a group of patients with non-small cell lung cancer as compared with a group of non-cancer control subjects. This represented a substantial impairment of DNA repair defences against oxidative stress.
[5.597] Oxidative stress occurred in a cell when there existed an excess of chemically-reactive by-products of oxygen metabolism. These were molecules known as "reactive oxygen species" (ROS), "oxygen free radicals", "free radicals" or, simply, "radicals". ROS were clusters of atoms, one of which contained an unpaired electron in its outermost shell of electrons, known as a radical or free radical. This was an extremely unstable molecular configuration, and radicals quickly reacted with other molecules or radicals to achieve the stable configuration of electrons in their outermost shell. If not removed by the cell's antioxidant system, ROS were thought to lead to a range of detrimental cellular events. The studies which had been carried out had been in vitro, almost always involving individual cells in culture. There were problems of extrapolation from these types of system when talking about tissue damage, organ damage, or damage to an animal or human. There were various theories and hypotheses about the meaning of the detrimental events that could be seen when a cell was exposed to an excess of ROS. Professor Idle referred to a figure which illustrated that approximately 95% of the oxygen breathed by humans was utilised in the normal processes of metabolism and converted into water. However, the residual part of the oxygen passed down an alternative pathway, the formation of ROS, such as hydrogen peroxide and the hydroxyl radical.
[5.598] ROS had been seen to play a role in the cellular changes associated with both ageing and cancer. The molecular target of ROS was DNA. The principal modification brought about by ROS was to add an oxygen atom into the normal guanine (G) in DNA, thereby creating 8-hyroxyguanine (8-HG) adducts. This was a very small modification and hard to detect. If these modifications were not repaired by the enzyme OGG, then they caused a change in the base sequence of the DNA from a G to a T. This was a mutation known as a "G ý T transversion", which was commonly found in the p53 and K-ras oncogenes of lung and renal tumours. This type of mutation could be provoked by ROS in the p53 gene and bore clear similarities to the type of mutation caused in vitro by the benzo[a]pyrene metabolite BPDE. Thus, the normal physiological processes that utilised oxygen, such as respiration, could lead to the generation of ROS and produce DNA modifications that resembled those that had been observed when cells were incubated with such mutagens as BPDE.
[5.599] In addition to the ROS mentioned by him, Professor Idle said that there were a number of other endogenously-formed free radicals that were capable of modifying DNA. The essential characteristic of free radicals was that they involved themselves like a set of "chemical domino effects"; there were chain reactions triggered by one chemical and the electron was then passed from one chemical to another, so it was not surprising that other chemicals became involved. For example, the result might be nitrogen-containing free radicals, such as nitric oxide, and lipids (fats) which were converted to lipid hydroperoxides; and even sulphate, which was an endogenous material, produced a sulphate radical.
[5.600] The foregoing, he said, showed that there might be a connection between cancer and the endogenous processes that produced oxidative stress and DNA modification and repair. Some authors believed that there was sufficient evidence to establish that oxidative stress furnished a comprehensive and unifying theory of human cancer. In this regard, exogenous chemicals and their metabolites were perceived as being able to generate ROS and this process was proposed by these authors to underlie the cancer-causing potential of a broad range of agents, including radiation, quinones, polycyclic aromatic hydrocarbons, phenols, aromatic amines, N-nitroso compounds and alkylating agents. However, the views expressed by Kovacic and Jacintho in 2001 required comprehensive investigation and substantiation. It was, however, pretty well established in radiotherapy that radiation generated oxygen free radicals in the environment of a tumour which killed the tumour, and the relationship between radiation and oxidative stress had been appreciated for a considerable time.
[5.601] The generation within each cell of large numbers of highly chemically reactive oxygen radicals, by processes that were an obligatory consequence of the use of oxygen, placed an oxidative stress upon the cell. The ROS were capable of interfering with cellular structure and function. Failure by the cell to prevent or to repair such impairments could have profound consequences for the cell, including the triggering of pathways that might lead to cancer. In Professor Idle's judgment, whether or not oxidative stress played a role in human lung cancer was not established, though work continued on this question.
[5.602] Professor Idle's next topic was epigenetics. This term was coined by the eminent biologist CH Waddington, who was Professor of Genetics at the University of Edinburgh. He did not believe that genetics, embryology and evolution were separate sciences, and attempted to understand the link between the genotype and the phenotype of an organism. A phenotype was a genetically-determined characteristic of an individual, such as eye colour or blood group. The genotype was the genetic information held in those genes that coded for the phenotype, so that the latter was a manifestation of the former. Epigenetics was the study of the processes by which the genotype gave rise to the phenotype, how the phenotype was manifested from the genetic information in the cell.
[5.603] The human genome was now known to encompass about 30,000 genes that made up just a small part of the 3 billion base pairs that comprised human DNA. In any given cell type, for example a sensory nerve cell in the skin, only a tiny fraction of the 30,000 genes was transcribed into RNA and then protein. When a gene was expressed in a particular tissue, it was copied into an RNA molecule that carried all the genetic information. This then found its way into the cell itself and interacted with part of the cell known as "ribosomes", which were small, spherical structures attached to an inner membrane of the cell. Ribosomes were the places where proteins were made; the genetic information in the form of RNA sat on the ribosome and the ribosome read each of three base pairs, forming triplets known as codons. A protein was made up of amino acids. There are twenty different amino acids, so a huge number of possible different proteins could be made. Each amino acid was coded for in this process by a codon. The codons were contained one after another, like a series of words, in the RNA molecule. This molecule attached itself inside the cell on top of a ribosome and the ribosome read each of the codons. By this process, known as "translation", the RNA molecule added different amino acids to produce a chain which would be the primary protein. In this way, the protein was specified in order to perform a particular function.
[5.604] The vast majority of genes were silenced by epigenetic mechanisms which were only just beginning to be understood; they were kept in a quiescent form so that they were not expressed into RNA and ultimately into protein. What made a sensory neuron different from a liver cell, and indeed different from a neuron in the brain, was the pattern of gene transcription which translated inherited genotypes into functional phenotypes. The state that underlay this distinction between different cell types in the body, often in close proximity to each other, such as an epidermal cell and a keratinocyte in the skin, was often referred to as "differentiation". Processes that controlled how genes were "switched on and off" and caused one tissue to differ outwardly and functionally from another were described as epigenetic processes. Epigenetic processes, whilst inherited, were not represented in the coding sequence of the DNA.
[5.605] The field of epigenetics was centred principally around two biochemical processes, methylation of DNA by the addition of a methyl group, and acetylation of proteins called histones by the addition of an acetyl group. Professor Idle explained that it so happened that within a sequence of DNA there were areas within certain genes, or close to certain genes, that were very rich in GC sequences. It seemed that the purpose of these was to control the epigenetic nature of the gene, to act as a "switch". The body seemed to have an ability to add a methyl group to the C bases. When an area of the gene rich in CG sequences, or an area close to the gene rich in these sequences, was methylated, then the gene could not be expressed. So when the apparatus for transcription into RNA reached this part, it skipped over it, it was not able to make an RNA copy and, therefore, a protein from that gene. Acetylation of histones was a grosser, larger process. The double helix was tightly coiled. Histones sat on top of the DNA sequence and helped it pack into a tight ball inside the nucleus. These proteins would prevent the transcription of an area of DNA that sat underneath them. In order for a gene to be copied into RNA, the histone had to move out of the way. When this underwent acetylation, the histone broke free from the DNA and lifted up, opening a "trap door" in the chromosome, enabling the apparatus to make an RNA copy and a protein. When the histones were deacetylated, the DNA could return to its tightly coiled state that inhibited gene transcription. The acetylation-deacetylation process was a continuous and reversible one that formed part of the circuitry within which genes were transcribed.
[5.606] It had recently been proposed by Urnov, writing in 2002, that dietary insufficiencies could affect the epigenetic methylation pathways in the nucleus, leading to genome misregulation. Epigenetic changes were frequently observed in cancer. Certain key genes were often "switched off" in tumours, permitting cell cycle checkpoints to be by-passed that would normally, when triggered, lead to apoptosis and the death of the tumour cell. For example, aberrant DNA methylation and histone deacetylation were associated with the silencing of the enzyme known as death-associated protein kinase (DAPK), which brought about apoptosis in response to various signals in colorectal and gastric cancers.
[5.607] With respect to lung cancer, rates of DNA methylation had been measured in readily accessible cells, those from the buccal mucosa, the mucous membrane lining the inside of the cheek. Using assays that measured the rate of incorporation of methyl groups into C bases in nuclear DNA, and the detection of the methylated product in situ, researchers had reported in 2002 that changes in global methylation in buccal mucosal cells might reflect changes in tissues at high risk of developing lung cancer. These authors also reported on studies where human squamous cell carcinoma of the lung was associated with a deficiency within the tumour of two vitamins, folate and B12, together with reduced DNA methylation. In contrast, an excess of vitamin C was found in squamous cell carcinoma of the lung and larynx compared with adjacent uninvolved tissue. This phenomenon appeared to be associated with increased methylation of DNA. Professor Idle said that these studies suggested that diet acted upon epigenetic mechanisms that might lead to lung cancer.
[5.608] An epigenetic mechanism that had been proposed to lead to lung cancer was that of "tumour acquired promoter hypermethylation". The DNA sequence that controlled the transcription of a gene was known as the promoter (to be distinguished from the expression "tumour promoter"), and when CpG sequences in the promoter were hypermethylated, the gene was silenced, i.e. not transcribed in that tissue. It had been found recently that "tumour acquired promoter hypermethylation" was a phenomenon that occurred in the promoter region of many tumour suppressor genes in, for example, non-small cell lung cancers. The promoter regions in nine separate tumour suppressor genes had been reported to be hypermethylated in non-small cell lung cancer relative to unaffected tissue from the other lung. The theory was that the hypermethylation silenced the tumour suppressor genes and was thus part of the multistage mechanism of lung cancer. But these very recent findings had yet to be confirmed and the interpretation of them was still preliminary.
[5.609] In summary, Professor Idle said, changes to the regulation of the genome that were not genetic (i.e. not involving mutations), rather than those that were epigenetic, might be involved in the genesis of cancers, including lung cancer. There was some evidence that DNA hypermethylation, the process that silenced gene expression, might be influenced by cellular vitamins and, by association, diet.
[5.610] Professor Idle next turned to aneuploidy, genome instability and the "mutator phenotype". He explained that aneuploidy was the state of having the wrong number of chromosomes in the cells, either too many or too few. The complement of chromosomes in a cell was known as the karyotype. A normal cell had twenty-three pairs of chromosomes. In certain situations, a cell might have an abnormal karyotype, which was always associated with a disease process. When the karyotype was grossly abnormal, due to the addition or deletion of chromosomes, the situation was referred to aneuploidy and the cell was known as "aneuploid". Professor Idle referred to a figure in his report showing a normal karyotype and an abnormal karyotype from a cancer cell, including both missing and extra chromosomes that might result from fusions between pieces of different chromosomes. He said that there existed very little agreement among scientists as to how cancer cells acquired such a large number of genetic and epigenetic abnormalities. At what stage in the development of cancer, whether early in the mechanistic process or later, the cell became aneuploid, was unclear.
[5.611] According to Duesberg and co-workers, reported in Li et al. 2000, gene mutation was an insufficient explanation of the cause of cancer, except when the mutation was so gross as to be characterised as aneuploid. In other words, simple mutations, even within a large number of genes such as tumour suppressor genes or oncogenes, were not sufficient to cause malignancy. Cancer, according to these authors, must be viewed at the level of the chromosome not the gene. They stated in the abstract at p.3236:
"The aneuploidy hypothesis predicts the long latent periods [of cancer] and the clonality [of tumours] on the basis of the following two-stage mechanism: stage one, a carcinogen (or mutant gene) generates aneuploidy; stage two, aneuploidy destabilises the karyotype and thus initiates an autocatalytic karyotype evolution generating pre-neoplastic and eventually neoplastic karyotypes. Because the odds are very low that an abnormal karyotype will surpass the viability of a normal diploid cell, the evolution of a neoplastic cell species is slow and thus clonal, which is comparable to conventional evolution of new species."
Professor Idle explained that clones were organisms or cells that were genetically identical by descent, meaning that the progeny all had the identical genetic make-up of the parents. A diploid cell was one in which all chromosomes, except sex chromosomes, were two in number and were structurally identical with those of the species from which the culture was derived.
[5.612] The 26 July 2002 issue of the premier scientific journal Science was dedicated to genome instability. One author wrote:
"Cancer cells are chock-full of mutations and chromosomal abnormalities, but researchers can't agree on when and how they come into play. Inside a cancer cell is a veritable gallery of horrors: inactivated genes, extra or missing chromosomes, and a host of other genetic abnormalities, large and small."
Professor Idle said that the origin of the aneuploidy that was proposed to generate genomic instability as a forerunner of cancer was not known. Given that aneuploidy required the complete loss of at least one chromosome or the addition of one new one, the genetic change must be quite drastic.
[5.613] Somewhat in parallel to the aneuploidy theory of cancer of Duesberg and co-workers was the "mutator phenotype" theory championed by Loeb and colleagues. This theory started with the premise that tumours, once detected either experimentally or clinically, contained far too many mutations (tens of thousands, in some cases) to have occurred by normal processes of mutation, and postulated that, early in its life, the tumour required a so-called mutator phenotype. This characteristic of the cell permitted multiple random unrepaired mutations to occur during development of the tumour. Loeb and colleagues had proposed that it was oxidative stress that drove the mutator phenotype and thus genomic instability, so much so that they had proposed (in 2001) the use of antioxidants to reduce DNA modifications by ROS and delay the progress of cancer.
[5.614] In summary, Duesberg's group had made a strong case that aneuploidy could be the only explanation for carcinogenesis as it was understood today, and it was certainly the case that tumour cells isolated ex vivo or developed in cell culture always had an abnormal karyotype and were aneuploid. These views were not universally shared, especially by Weinberg's group, who pointed to a small number of defined mutations, as few as three, as the first step of carcinogenesis, or by Loeb and other proponents of the mutator phenotype theory. Nevertheless, whether it was a small number of mutations or a much larger number due to a mutator phenotype and genomic instability, or even gross karyotypic changes such as aneuploidy, that "kick-started" the process of cancer, these theories were the subject of an active and current debate amongst some of the most senior figures in cancer research.
[5.615] Professor Idle summarised the competing theories of cancer causation as follows. He said that there had been competing theories advanced by senior scientists and that each theory was credible. Nevertheless, there remained little agreement today as to how cancer arose. As with any area of scientific endeavour, this represented the normal process of proposition, counter-proposition and wider scientific debate, that might ultimately lead to a consensus view, or, through the discovery of new facts, to yet further proposition and debate. The cancer question was one of the most complex that medical science had yet faced. An enormous store of cellular and molecular data had been accumulated, the exact meaning of much of which was still unclear to the cancer research community. Mortality from cancer today was largely equivalent to what it was fifty years ago (though there had been variations in the occurrence of cancer at different sites). As he had previously stated, most of the understanding about the mechanisms by which lung cancer arose was built on the knowledge base of cancer in general. In this section of his evidence, he had focused on theories of cancer causation that in his judgment could be the most relevant to lung cancer. These included viruses, chemicals, failure of host defence mechanisms and various processes that operated at the level of genes and chromosomes. Additionally, there was now a major effort to understand how the nervous and immune systems might be inter-related and mediate the effects of personality, emotion and stress in the cause of cancer. However, in his view, for the reasons given, none of these provided an adequate basis for explaining how or why lung cancer arose.
[5.616] The next main section of Professor Idle's report and evidence related to laboratory investigations with tobacco derivatives and public health reports up to the end of 1964. He said that, as previously described by him, in the period up to the end of 1964 cancer researchers were concerned with the big questions of what cancer was and by what mechanisms it arose. In this same time period, a statistical association between lung cancer in males and heavy cigarette smoking was reported. A number of researchers began laboratory work on tobacco derivatives on the view that, if this association was causal, it ought to be possible to identify and remove the constituent or constituents of tobacco smoke responsible for causing lung cancer. This concept had evolved from the work, previously discussed, on shale oil and coal tar.
[5.617] Professor Idle next discussed condensate and derivatives of tobacco. He said that it was pertinent to consider the differences between cigarette smoke, total particulate matter, tobacco smoke condensate and tobacco "tar", terms that appeared in the literature of the period prior to 1964. Cigarette smoke was a dynamic mixture consisting of a fine aerosol of coated particles, known as the particulate phase, and a gas phase, also referred to as the "vapour phase". Many tobacco smoke constituents existed in both phases in a dynamic equilibrium between the two. The total particulate matter (TPM) was the material that could be trapped by passing the cigarette smoke through a Cambridge glass fibre filter. Gas phase was that material which escaped such filtration. If the cigarette smoke was passed through a series of glass vessels chilled to a very cold temperature, such as -60oc, then there condensed out a material known as "tobacco smoke condensate", which was sometime referred to, perhaps erroneously, as "tobacco tar". The "tar" yield referred to on a packet of cigarettes was the TPM captured on a Cambridge filter minus the water and nicotine content. It was not really an entity, it was a number. It was different from tobacco smoke condensate. It was the result of an experiment and then a calculation. A smoker was never exposed to tobacco smoke condensate because this material was created in a laboratory by cold-trapping cigarette smoke. In addition, a number of the studies involved the separation of tobacco smoke condensate into various fractions for biological testing. Extraction of the tobacco smoke condensate with various solvents and other fractionation processes led to the production of three main fractions, the acidic, the basic and the neutral fractions. Benzo[a]pyrene and the other polycyclic aromatic hydrocarbons were found in the neutral fraction.
[5.618] Against the background of previous studies, Wynder and others reported in 1953 a study in which tobacco smoke condensate was painted on the skin of mice. Their experimental method, as described by Professor Idle, was first to produce tobacco smoke condensate using a smoking machine, to take a solution of that material in a solvent and to paint it with a brush onto the backs of mice in places from which the hair had been clipped, repeating this a number of times per week for many weeks. The experiments used a particular strain of mice. Some additional animals were painted with a solution of 20-methylcholanthrene. All of these developed skin carcinomas in less than five months, thus demonstrating that this strain were particularly sensitive to skin carcinogens. The test animals were divided into two groups. One group was treated with tobacco smoke condensate in the solvent acetone. A second, smaller group, were treated with acetone alone to act as a control group. The treatment continued, with some important variations, three times a week for up to twenty-four months. The experimenters found that they were able to increase the concentration of the condensate in the acetone from one part in three eventually to equal parts. The problem was that when the experimenters started to paint the material on at a very high concentration, some of the animals died, so they gradually increased the amount over a period of time. Professor Idle commented that, when they had observed the appearance of papillomas on the animals' skin, they specifically painted only the papillomas once the animals "became weakened by old age or disease", rather than painting the whole shaved area uniformly. The authors reported data from eighty-one mice (from an original group of 112) that had been painted with the tobacco smoke condensate. The reported incidence of skin carcinomas in these animals was 44.4% and of papillomas was 59%. The reported incidence of carcinomas was dramatic in light of the results obtained by previous researchers, who had reported either no incidence or a very low percentage of incidence. The report by Wynder and others generated wide publicity and considerable scientific debate and controversy. The results were never replicated by them or by any other researchers. There were hundreds of studies carried out on skin painting of mice with tobacco smoke condensate, some of them by Wynder and his colleagues, but the same result was never achieved. Professor Idle said that he thought that the consensus view was probably that the material used by Wynder was somehow contaminated by having decomposed or not been stored correctly.
[5.619] Passey and others reported in 1954 that, in an attempt to replicate the results of Wynder and colleagues in 1954, they observed no tumours, and that the results of their experiments at most suggested that any carcinogenic properties possessed by the tars used were very attenuated. Kosak in 1954 catalogued the known and putative constituents of tobacco smoke, listing 43 known chemicals. Cooper and colleagues reported in 1954 that they had made new findings, namely the presence of small amounts of acenaphthylene and benzo[a]pyrene in tobacco smoke condensate. The same investigators, in a further report, identified the presence of various polycyclic aromatic hydrocarbons in condensed smoke from "cigarettes" made entirely of cigarette paper. Hamer and Woodhouse prefaced a study in 1956 with the statement that since the evidence that cigarette tars were carcinogenetic had come from only one group in one country, it seemed desirable that the results of other workers should be presented as soon as possible so that the position could be better evaluated. These British workers did not obtain skin tumours in either rabbits or mice that had been painted with tobacco smoke condensate obtained from a smoking machine. In addition, the tobacco smoke condensate did not initiate tumours in mice that were subsequently painted with the tumour promoter croton oil. In mice previously painted with benzo[a]pyrene, a tumour initiator, some tumours were reported, suggesting a possible tumour promoting effect of the tobacco smoke condensate. The authors discussed the possible reasons for the discrepancy between their findings and those of Wynder and colleagues, noting that the latter prepared a tobacco smoke condensate using cigarettes that were smoked at temperatures up to 960oC, in contrast with their own condensates which were prepared using cigarettes that were smoked at a temperature up to only 780oC which, they stated, was comparable to smoking cigarettes by mouth.
[5.620] Roe and co-workers, who had turned their attention to the study of initiation and promotion, concluded that since the effect of cigarette smoke condensate could be increased by the addition of initiators but not of promoters, the relative weakness of the carcinogenic action of cigarette smoke condensate was due to the paucity of initiators. Professor Idle said that in this period workers were expressing the view that cigarette smoke condensate did act weakly as a carcinogen. Roe et al. 1959 stated at p.631:
"Man, whether a smoker or not, is exposed to substances of proved tumour-initiating power in the polluted atmosphere, and in the products of some industrial processes and in some kinds of food. The correlation between smoking habits and lung tumour incidence may well be determined not primarily by the carcinogenic effect of tobacco smoke, but by its predominantly tumour-promoting action on the bronchial epithelium."
[5.621] Professor Idle summarised the laboratory studies involving skin painting in the period up to 1964 as follows. Prior to the publication in 1950 of the first reports of an association between lung cancer and cigarette smoking, there was relatively little laboratory research on cigarette smoke. But what research existed indicated that tobacco smoke condensate was weakly carcinogenic to mouse and rabbit skin. Following the 1950 reports, Wynder's laboratory began conducting studies in which tobacco smoke condensate was painted on the backs of mice. The 1953 report provoked considerable controversy among other scientists. Other laboratories reported that the findings of Wynder and colleagues could not be replicated. Moreover, Wynder's laboratory itself was never able to duplicate their original findings. In a number of subsequent publications, Wynder and his colleagues made clear their view that the purpose of their laboratory research was to identify the constituent or constituents of cigarette smoke that might account for the epidemiological association between lung cancer and cigarette smoking, thereby permitting its removal. They stressed that the mouse skin painting studies were not designed to prove that smoking caused lung cancer. In Wynder et al. 1957b, the authors wrote:
"As we have repeatedly stated, the primary proof that smoking is a cause of cancer in man must be based not upon animal experimentation but upon human epidemiological, pathological, and clinical studies. However, if animal data confirm the carcinogenic activity of a substance suspected to be carcinogenic to man, the evidence that this substance is active in man is strengthened.
The basic purpose of animal experimentation is to define the mechanism of carcinogenesis and to identify the specific carcinogens responsible for this activity."
[5.622] Professor Idle then turned to what he called "tobacco smoke condensate versus fresh whole cigarette smoke". He explained that cigarette smoke freshly produced by a cigarette was a highly changing and dynamic mixture of two phases: a fine aerosol coated with a liquid film and a gas or vapour phase, and that substances moved in and out of the liquid coating of particles and the vapour phase constantly. It was a dynamic, forever-changing chemical mixture, whereas the tobacco smoke condensate was a viscous liquid that had been collected at cold temperatures. He said that it was well established by 1965 that chemical changes rapidly occurred in the smoke after its formation in the burning zone of the cigarette and that these changes continued to occur in the collected tobacco smoke condensate. The sudden cooling of the smoke in the traps also allowed the formation of new chemical entities that were either not present in the smoke hitherto, or were present at a very low level. Cold-trapping permitted the gaseous and highly volatile constituents of the smoke to escape; these constituents comprising the great majority of the mass of the cigarette smoke. Thus, the four principal components of cigarette smoke by weight - nitrogen, oxygen, carbon dioxide and carbon monoxide - were not condensed in the cold-traps. Likewise, the other gases, such as nitrogen oxides and volatile components of the smoke, which were legion, were not condensed in the cold-traps. In summary, in his opinion, fresh whole cigarette smoke and tobacco smoke condensate differed in both their nature and chemical composition. This told us that we must extrapolate with great care from mouse skin painting experiments, using cigarette smoke condensate, to the effect of cigarette smoke on the lung.
[5.623] Professor Idle next turned to the question of the amount of cigarettes needed to produce condensate versus human smoking. He said that it was common practice in the mouse skin painting studies prior to 1965 for a mouse weighing 25-35g to be painted with an amount of tobacco smoke condensate that had been derived from at least one cigarette, and sometimes as many as five, that had been smoked on a smoking machine. Laboratory mice were not exposed to a dose of tobacco smoke condensate comparable to the number of cigarettes smoked by a typical cigarette smoker. It could be calculated from reports of the Wynder group that the number of cigarette equivalents received by a mouse in skin painting with tobacco smoke condensate per month equated to a human dose of 6,400 cigarettes per month, or over ten packs of cigarettes per day. On this basis, it could be calculated that mice were not exposed to a dose of tobacco smoke condensate equivalent to the number of cigarettes smoked by a cigarette smoker per day. In summary, in Professor Idle's judgment, the amount of tobacco smoke condensate applied to mouse skin each day was derived from a number of cigarettes far in excess of the number of cigarettes smoked by any human smoker in one day.
[5.624] The next comparison made by Professor Idle was that of mouse skin epidermal tissue versus respiratory epithelium. He said that the mouse skin, like most mammalian skin, was comprised of epithelial tissues, organised in layers. Within these layers were specialised tissues, such as various types of gland, hair follicles and so on. When the outer layers were damaged, they came off, and new layers rose up, hardened and became the outside surface. In the lung, while the tissue was epithelium, it was structurally very different. The cells were not arranged in flat layers, they were arranged in columns known as "columnar epithelium". On the top of each column, at least in the tracheobronchus, were little cilia, which wafted the mucus blanket up from the lung to the level of the larynx. Between these columns were goblet cells that produced the mucus. The epithelial cells of the skin arranged in layers, sometimes described as a "squamous" or "pavement" morphology. Thus, the epithelia of the skin were and the tracheobronchus were architecturally different and they served distinctly different functions for the host.
[5.625] There were further dissimilarities. The skin was keratinised, meaning that it had a hardened surface. The columnar epithelial cells in the lung were not hardened. Their purpose was not like that of the skin, which was to keep out the outside world, but to provide a structure on which the mucus could be produced, taken up and moved along to remove any particles or dead cells from down in the lung, up and to the level of the throat, where it could be swallowed or expectorated. Since the mucus was primarily an aqueous solvent, it would also trap water-soluble chemicals, aerosols and particular aggregates, expelling them by the same process. In a cigarette smoker, a proportion of the tobacco smoke constituents, therefore, was absorbed in the mouth and by the mucus of the tracheobronchial tree. In addition the lung was home to a type of white cell known as the pulmonary or alveolar macrophage, which was able to engulf and digest particles or micro-organisms. Like certain of the cells of the lung itself, these macrophages also were rich in enzymes that could metabolise and detoxicate chemicals. In summary, in Professor Idle's judgment, the results of experiments on mouse skin could not be carried over to respiratory epithelium.
[5.626] The next comparison was that of mouse versus man. He said that there were many differences at biochemical, anatomical, pathological and physiological level between mouse and man. Whilst we might share over 90% genetic similarity, nevertheless there were differences, both obvious and less obvious. It was an archaic point of view, no longer accepted, in toxicology to think of a mouse as simply a small human. Green, who was Chairman of the Department of Pathology at Yale University School of Medicine, testifying in 1957 before a sub-committee of the Committee on Government Operations in the United States House of Representatives, said that Wynder's experiments demonstrated that tobacco tar extracted by a special technique induced cancer on the skin of one strain of mice, and nothing more. Different results had been achieved with different strains of mice. The occurrence of such high differential susceptibility among most strains would suggest the existence of an even more pronounced difference between species, and an extrapolation of the findings to man would be absurd. There was no surety that a substance capable of inducing cancer in mouse skin would also induce cancer in mouse lung, to say nothing of human lung. Professor Idle said that he agreed with this conclusion.
[5.627] Professor Idle next considered the interaction of constituents of tobacco smoke. He said that in the period up to 1965, a number of studies were conducted in an effort to identify a constituent or constituents of tobacco smoke that might account for lung cancer in human smokers. Despite these efforts, no such constituent was identified. The studies showed that tobacco smoke condensate was active as a complex mixture in the mouse skin painting model and no binary combination of individual initiators and promoters was likely to explain this activity. Thus, by 1965, mouse skin painting studies had not identified any constituent or group of constituents, as they existed in tobacco smoke, that accounted for lung cancer in cigarette smokers.
[5.628] The next topic was inhalation. By 1965, researchers had begun to conduct studies in which laboratory animals were exposed to whole cigarette smoke by inhalation. The objectives of this research were to determine whether or not evidence could be produced that cigarette smoke could cause lung cancer in animals, and thereby investigate a possible relationship between human lung cancer and cigarette smoking. Thus, Stewart and Herrold, the first of whom was a prominent researcher at the NCI, stated in Stewart and Herrold 1962 that were it possible regularly to induce pulmonary cancers in laboratory animals by exposure to inhalation of cigarette smoke, this would constitute good evidence that cigarette smoking could cause cancer of the lung in man. Professor Idle said that this was in contrast to the objectives of mouse skin painting, which had been mainly to identify a constituent or constituents of tobacco smoke that might account for the occurrence of lung cancer in cigarette smokers with a view to removing it or them. So the purpose of inhalation experiments with laboratory animals was to induce lung cancer. Nevertheless, Hinshaw and Garland 1963 stated that despite literally hundreds of careful and prolonged experiments designed to produce lung cancer by exposure of mice or rats to prolonged breathing of cigarette smoke, no bronchogenic carcinomas had been produced. Professor Idle said that prior to 1965, none of the studies that exposed laboratory animals to cigarette smoke by inhalation reported a statistically significant increase in squamous cell carcinoma of the lung.
[5.629] The next topic was ciliastasis. Professor Idle said that certain investigators turned their attention to the gas phase of cigarette smoke, in contrast to the particulate phase which had been focused upon in relation to polycyclic aromatic hydrocarbons, such as benzo[a]pyrene. Hilding, writing in 1956, cited an article in the New York Times in 1955 that stated that Yemeni Jews never developed lung cancer despite smoking "a good deal of strong tobacco but passed the smoke through water in a special type of pipe." This author investigated the extent to which cigarette smoke, on entering the mouth, might be retained by the wetted surfaces of the mouth. Employing the tracheas and lungs of freshly slaughtered cows, he demonstrated that smoke components were deposited on the "mucous blanket" as a "visible scum of tar" and, at a high enough concentration, this was capable of causing the ciliary action (which persisted after death) to cease. He extended these experiments, using India ink as a marker of mucus flow, in a subsequent paper.
[5.630] Interest in the gas phase of tobacco smoke continued to grow in the late 1950s, particularly with respect to the ciliastatic properties of tobacco smoke constituents. Falk and others reported in 1959 on a study of tobacco smoke constituents in an effort to determine which of them might possess ciliastatic properties using the ciliated trachea of rats and rabbits and the ciliated epithelium of the frog oesophagus. They reported that cigarette smoke had ciliastatic properties and that nicotine elicited the greatest ciliastatic effect. They also determined that polluted atmospheric air (smog) had ciliastatic properties. Kensler and Battista reported in 1963 that several constituents of the gas phase of cigarette smoke, specifically hydrogen cyanide, formaldehyde, acrolein, ammonia and nitrogen dioxide, were ciliastats in experiments performed on the rabbit trachea. Wynder and others investigated in 1963 whether or not various constituents of tobacco smoke condensate were ciliastats, using the gill lamellae of the fresh-water mussel as their model. They reported that the phenols were the most active ciliastats.
[5.631] Thus, Professor Idle concluded, prior to 1965, three groups of researchers, and possibly more, had reported that individual constituents of cigarette smoke and tobacco smoke condensate had ciliastatic properties using various animal models. In his judgment, by 1965, it was unclear whether or not ciliastasis played a role in human lung cancer.
[5.632] In the period from 1965 to the end of 1971 the action of phenols as ciliastats began to attract the attention of investigators. A number of researchers investigated whether or not constituents of cigarette smoke had the ability to slow, or even halt, the movement of the cilia found in the tracheobronchus. It was thought that the ciliastats in cigarette smoke were found in the gas phase and particular attention was paid to certain groups of constituents, the phenols, the aldehydes and the acids. Wynder and Hoffmann, writing in 1967, were definitive in their view about the role of ciliastasis in lung cancer. They stated that the impairment of ciliary function and mucus transport in the respiratory system was clearly a phase preceding the manifestation of metaplasia in the bronchial epithelium. Wynder and his colleagues continued to perform experiments designed to measure the effect of constituents of tobacco smoke condensate or gas phase on the ciliary activity of the gills of clams and other molluscs. However, Professor Idle pointed out, this animal system was very different from the human tracheobronchus. In addition, the physiological function of the clam gill cilia was to extract oxygen from the water, in stark contrast to that of the mammalian tracheobronchial cilia, which was to move foreign matter up the ciliary escalator to the throat. Wynder and colleagues, writing in 1965, justified their use of the mollusc gill assay on the basis of its "simplicity, ease of operation and reproducibility".
[5.633] Other researchers preferred to use other animals. One group used mammalian trachea, from sheep, rabbit or rat, to conduct their studies. Battista and colleagues reported in 1967 that cigarette smoke applied to the larynx of the whole (live) chicken had little or no effect on ciliary action, but did result in a greater than four-fold increase in mucus secretion. Dalhamn from the Karolinska Institute in Stockholm, one of the leading researchers on ciliastasis, in a criticism of the methodologies employed to investigate it, wrote in 1970 that vertebrates, possibly mammals, were preferable to inferior species; this was in the context of a critique of using shellfish gills as a model to test ciliastasis. He also proposed other criteria for the investigation of ciliastasis, such as the preferential use of in vivo experiments, direct measurement of ciliary activity rather than indirect measurement on mucus, and the use of high-speed cameras for filming the cilia in in vivo experiments. Not many of the reported investigations met these stringent criteria.
[5.634] Pavia and others reported in 1970 on an investigation of ciliastasis using human volunteers, both non-smokers and smokers. The volunteers inhaled polystyrene pellets labelled with a radioactive isotope, and it was reported that they did not exhibit impaired tracheobronchial clearance (an in vivo measurement of ciliary function). The authors concluded that although the study had shown that the cilia were apparently functioning normally in spite of a lifetime of smoking, it was possible that temporary stasis occurred in the cilia after a smoke, with complete recovery between smokes and overnight. Camner and others reported in 1971 that the speed of mucociliary transport in a group of smokers was significantly higher during intensive cigarette smoking than when they were not smoking and that on the average tracheobronchial clearance diminished in the same persons when they had abstained from smoking for one week, although the decrease was not significant; accordingly, ciliastasis did not seem likely to occur in connection with ordinary smoking. Professor Idle commented that this remarkable finding suggested that smoking actually increased ciliary activity, rather than diminishing it, as had been claimed by others. Das and colleagues reported in 1970 that diluted cigarette smoke also stimulated ciliary activity in frog oesophagus.
[5.635] Professor Idle concluded that by the end of 1971 a number of volatile phenols, aldehydes and acids were reported to have ciliatoxic properties in the mollusc gill, but the relevance of these studies to humans was questioned. Direct investigation of ciliary activity in smokers versus non-smokers and in smokers in relation to patterns of smoking was unable to detect any ciliatoxic activity of cigarette smoking. The belief that ciliastasis brought about by smoking was part of a mechanism of human lung cancer had been virtually eliminated. To his knowledge this had never been resurrected.
[5.636] Professor Idle was next asked to comment on passages in various public health reports. In RCP 1962 it was stated at p.26, para.37:
"Skin cancer can be produced in mice by applications of tar condensed from tobacco smoke but the results obtained by various investigators have not been uniform and exposure of animals to tobacco smoke in inhaled air has failed to produce lung cancers."
Professor Idle said that these statements were accurate. The paragraph continued:
"Moreover the amount of cancer-producing substances in the smoke itself does not seem likely to be sufficient to account for the large number of cases of cancer associated with the habit."
Professor Idle said that he thought that the Royal College of Physicians were recognising at that point that workers had not been able to identify any substance or substances in the smoke in sufficient quantity to explain the statistical association between cigarette smoking and lung cancer. RCP 1962 stated at p.26, para.39:
"In addition to the known carcinogens which have been detected in tobacco smoke others as yet undetected may be present; possibly two or more in combination may reinforce each other in producing cancer. It is possible that tobacco smoke may contain substances which act in conjunction with substances generally present in the air we breathe to produce cancer, although neither substances might do so alone. Indeed the action of tobacco might be simply to produce chronic irritation which, as in other tissues, may increase liability of the lung to cancer. There is a wide field for further investigation here, but no ground for refuting the evidence from human experience."
Professor Idle commented that this passage contained speculation, unvouched by experimental evidence, and he agreed that there was scope for further investigation. In RCP 1962, at p.S5 of the summary, among the facts mentioned which might be considered to conflict with the conclusion that smoking was a cause of lung cancer was the fact that no animal had yet been given lung cancer by exposure to cigarette smoke. Professor Idle agreed that this statement was correct.
[5.637] Turning to USSG 1964, Professor Idle repeated that he agreed with the statement, at p.20, that:
"Statistical methods cannot establish proof of a causal relationship in an association. The causal significance of an association is a matter of judgment which goes beyond any statement of statistical probability."
[5.638] At p.33 of the report it was stated:
"Seven polycyclic hydrocarbon compounds isolated from cigarette smoke have been established to be carcinogenic in laboratory animals. The results of a number of assays for carcinogenicity of tobacco smoke tars present a puzzling anomaly: the total tar from cigarettes has many times the carcinogenic potency of benzo(a)pyrene present in the tar. The other carcinogens known to be present in tobacco smoke are, with the exception of dibenzo(a,i)pyrene, much less potent that benzo(a)pyrene and they are present in smaller amounts."
Professor Idle said that these statements accorded with his own view.
[5.639] In the same passage of the report, it was suggested that this discrepancy might possibly be due to the presence of cocarcinogens in tobacco smoke, and/or damage to mucus production and ciliary transport mechanism. Professor Idle referred to his previous evidence about the latter hypothesis. As to the former, he was asked to comment on a passage at p.58:
"The results of a number of such assays [of tobacco smoke tars for carcinogenicity by skin-painting] present a puzzling anomaly: the total tar from cigarettes has about 40 times the carcinogenic potency of the benzo(a)pyrene present in the tar."
This, he said, quantified the anomaly.
[5.640] At p.146 of the report it was stated:
"The amount of known carcinogens in cigarette smoke is too small to account for their carcinogenic activity. Promoting agents have also been found in tobacco smoke but the biological action of mixtures of the known carcinogens and promoters over a long period of time is not understood."
At p.59 the report stated: "Assessment of all conceivable synergistic effects presents a gigantic problem for exploration." Professor Idle said that this problem had never been solved.
[5.641] At p.167 of the report it was stated:
"The lungs of mice, rats, hamsters, and primates have been found to be susceptible to the induction of bronchogenic carcinoma by the administration of polycyclic aromatic hydrocarbons, certain metals, radioactive substances, and oncogenic viruses. The histopathologic characteristics of the tumors produced are similar to those observed in man and are frequently of the squamous variety."
Professor Idle said that this meant simply that it was possible to induce squamous cell carcinoma of the lung in the animals mentioned.
[5.642] At p.165 the report stated: "The production of bronchogenic carcinomas has not been reported by any investigator exposing experimental animals to tobacco smoke." Professor Idle said that his research in the literature led to the same result.
[5.643] Professor Idle concluded that as at the end of 1964, none of the competing theories had been established as the cause of human cancer. None of these theories provided an adequate basis for explaining how lung cancer arose. Some researchers conducted studies in which tobacco smoke condensate was painted on the backs of laboratory mice. The methodologies used in these studies were not standardised and, in addition, other variables contributed to the widely divergent incidence of mouse skin cancer reported in these studies. These studies did not result in the identification of a constituent or group of constituents that explained the occurrence of lung cancer in cigarette smokers. None of the studies in which laboratory animals were exposed to whole cigarette smoke by inhalation reported a statistically significant increase in squamous cell carcinoma of the lung in the exposed animals.
[5.644] By the end of 1964, three public health authorities had concluded that cigarette smoking was a cause of lung cancer. These judgments were based almost entirely upon an interpretation of epidemiological reports. These conclusions were not, however, based on a consideration of lung cancer as a biological phenomenon. For the reasons thus far given by him, he did not agree with these conclusions. In his opinion it was not "almost universally accepted by scientists" (as averred on behalf of Mrs McTear) in 1964 that cigarette smoking could cause lung cancer. Moreover, in his opinion, cigarette smoking had not been established as a cause of lung cancer as at the end of 1964. This opinion was expressed on the basis of the literature which he had considered.
[5.645] In the period up to the end of 1971, research continued on the factors thought to cause cancer and on the competing theories of cancer causation. Laboratory research on tobacco derivatives also continued in this period. This research involved both mouse skin painting and inhalation studies. Much of this effort continued to be directed towards the identification of constituents of tobacco smoke that might be responsible for lung cancer in smokers. The principal investigators included Day and colleagues at the Tobacco Research Council Laboratories in Harrogate. By 1965, it had been recognised that a number of the variables in the mouse skin painting studies might account for the disparate results obtained. These included the type of tobacco used to produce the tobacco smoke condensate, the preparation and storage of the tobacco smoke condensate, the method and schedule of application of the tobacco smoke condensate to the mouse skin, the strain of mouse used in the studies, and the criteria used to determine the biological end points of the studies. If an investigator were to determine whether a skin tumour had been produced on a mouse, then it was necessary to distinguish between benign skin tumours and various types of malignant skin tumour, whether a tumour appeared malignant under the microscope, behaved in a malignant manner in the mouse, had invaded underlying tissues and probably that it had metastasised to other organs. During this period, researchers at the Harrogate Laboratories investigated and standardised many of these variables. In their papers they stated clearly that there were difficulties that they recognised in comparing studies between different laboratories for these reasons, and so they set about an attempt to standardise as many of them as possible.
[5.646] The principal study was by Day, writing in 1967, who calculated that 8,000 mice would be required to perform a study that could distinguish between a tumour rate of 20% and a tumour rate of 30% with 98% confidence: this was a very much larger number than had been used in any previous study, for example those of Wynder. Day attempted to standardise a number of the experimental variables: females of a strain of mouse that were resistant to nicotine toxicity were used; large group sizes were used; a measured amount of condensate was applied, using an automatic pipette; the study continued until each animal died or became sick, rather than finishing at an arbitrary time point; the diagnosis of skin cancer was made by an animal pathologist who could make a diagnosis of an invasive skin cancer by ascertaining whether the skin was attached to underlying muscle by the tumour; and complete autopsies, including the histopathology of all major organs, were performed on all mice, including those without skin tumours. The autopsies constituted an important difference between this study and the other published skin painting studies in this period, because Day was able to report on the probable cause of death for each test animal. The author reported an overall incidence of 3% of malignant tumours in the painted area (in control groups as well as tested animals). This was the eleventh most common pathological finding at the time of death in these animals. He also reported on a dose-response relationship. In a further paper, published in 1968, Davies and Day commented on the divergent findings between Day's results, and those of Wynder and others, published in 1953. They observed that the differences were possibly due to the type of tobacco smoke condensate used and the method of application of the condensate.
[5.647] In this period, Professor Idle said, the results of the mouse skin painting studies were offered by some as evidence that cigarette smoking was a cause of human lung cancer. Thus, Dr Thomas Carlile, Chairman of the Committee on Tobacco and Cancer of the American Cancer Society and past President of the American Cancer Society, in testimony at the 1965 United States House of Representatives Hearings on Cigarette Labeling and Advertising, identified one of the points that the American Cancer Society had considered in reaching its decision that cigarette smoking was a major cause of the increase in lung cancer to be the fact that cancer had been produced repeatedly by the application of tar condensed from cigarette smoke to the skin of experimental animals. Leading medical scientists, however, disagreed with the view that the mouse skin painting data were relevant to human lung cancer. For example, Dr Harry S N Greene, Chairman of Pathology, Yale University School of Medicine, testified at the same hearings that it was true that cancer had been produced in the skin of mice by long-continued painting with tobacco condensates. It should be emphasised, however, he said, that cancer had also been produced in mice by a variety of common innocuous substances such as salt, sugar, egg white, and cellophane. Dr R H Rigdon, Professor of Pathology at the University of Texas at Galveston School of Medicine, also testifying at these hearings, stated that the fact that tobacco tars would produce a tumour on the skin of a specific strain of mouse did not, in his opinion, lead to the conclusion that cigarette smoke would produce cancer in the lungs of man. He said that as he continued his work in the field of experimental carcinogenesis, he found puzzling variations in the response observed from one species of animal to another with respect to the development of cancer, so data concerning cancer in animals should not be considered proof of a similar effect in man.
[5.648] Dr Clarence Cook Little gave evidence at the 1969 United States House of Representatives Hearings on Cigarette Labeling and Advertising: Little 1969. At that time he was Scientific Director of The Council for Tobacco Research - USA. He had been Director and thereafter Director Emeritus of the Jackson Laboratory for Research in Genetics and Cancer, and Managing Director of the American Society for the Control of Cancer (subsequently the American Cancer Society). He served two terms as President of the American Association for Cancer Research. He had been actively interested in cancer research as a biologist for more than 60 years. In the course of his evidence he said, at p.1106:
"When this [the painting of animal skin with tobacco smoke condensate] is done the painting is followed in some cases by abnormal growth of the skin and, in some of these cases, later by skin cancer. In this respect the condensate of tobacco smoke joins more than 500 other substances, many of which are known to be entirely harmless to man, in producing a more or less standardized effect on mouse skin. There is nothing unusual or peculiarly characteristic about the reaction of these animals to smoke condensate. However, an important fact which has been overlooked or ignored is that the relative cancer-producing effects of condensates of cigarette smoke, of pipe tobacco smoke and of cigar smoke do not follow the same order of activity or coincide with the recorded degrees of statistical association between cancer and these three types of smoking exposure in man. This discrepancy is basic. It shows that if we accept the statistics as a basis for a working hypothesis we must also admit that skin painting with condensates of tobacco smoke is not a significant or valid method of testing the relative cancer-forming potentials of such smoke required by the working hypothesis [that smoking causes lung cancer]. In other words, the employment of a different agent (condensate instead of smoke) on a different target organ (skin instead of lung) and on a different animal (mouse instead of man) is not a firm basis on which to develop a broad and comprehensive research program to determine the quantitative effects of smoke on man."
Professor Idle said that this passage reflected his own views.
[5.649] He concluded that by the end of 1971, it was recognised that continued screening of tobacco smoke condensate in the mouse skin bioassay in an effort to identify a constituent that might be responsible for the occurrence of lung cancer in cigarette smokers was unlikely to produce a result. Therefore, the testing of tobacco smoke condensate in the mouse skin model came to a close in the period immediately following 1971. In his opinion, the reported results of the various experiments in which tobacco smoke condensate was painted on the skin of mice could not answer the question of whether or not cigarette smoking caused human lung cancer. He said that the position in this regard remained the same now as it was then.
[5.650] Professor Idle's attention was next directed to IARC 1986. At pp.135-136, in discussion of the administration of tobacco smoke condensates, reference was made to the 1952 paper of Wynder and others. At p.136 it was stated:
"Animal studies conducted prior to 1964 provided an important measure of support for the epidemiological demonstration that cigarette smoke is an important human carcinogen."
Professor Idle said that he did not agree with this view, for the reasons already given by him. This statement, in IARC 1986, was at variance with Wynder's statements about the purpose of his studies, which was to identify constituents that might be carcinogenic and to remove them, rather than to prove a causal connection between cigarette smoking and lung cancer. For the reasons he had given, studies in which tobacco smoke condensate was painted on to mouse skin were not designed to answer, and could not answer, the question of whether or not cigarette smoking was the cause of human lung cancer. Thus, in the period 1965 to the end of 1971, there was an increased focus on experiments that exposed experimental animals to whole cigarette smoke by inhalation. This was seen, at the time, as a model which more closely resembled human smoking than the mouse skin painting model.
[5.651] Professor Idle then turned to experimental animal inhalation models. He said that there were four principal animal models that were used in the inhalation studies, the mouse, the rat, the hamster and the dog. It should be appreciated at the outset that the objective of these studies was to produce squamous cell carcinoma of the lung in animals exposed to whole cigarette smoke by inhalation.
[5.652] First, there were studies in mice, the methodology and results of which were summarised in his report. He was asked simply to state the conclusion, that in none of them was squamous cell carcinoma of the lung induced.
[5.653] Among studies in rats, Alexandrov and Raitchev reported in 1965 on the passive exposure of 107 rats to cigarette smoke for two years. They reported that six out of 107 developed adenomas and two out of 107 developed adenocarcinomas, both in the lung. They also reported that none of 113 controls developed lung tumours. Laskin and others reported in 1970 on a study in which rats were exposed to the irritant gas sulphur dioxide mixed with the polycyclic aromatic hydrocarbon benzo[a]pyrene. Some rats developed squamous cell lung tumours, and this established that the rat was a species susceptible to develop squamous cell lung carcinomas.
[5.654] Investigators at the Research Institute of the German Tobacco Industry were engaged in an undertaking to establish if hamsters might be a suitable model for cigarette smoke inhalation. Dontenwill reviewed this work in 1970. No lung tumours were reported. Progressive epithelial changes were found in larynx which Dontenwill described as "precancerous". Professor Idle said that what was observed was not therefore cancer.
[5.655] Auerbach and others, and Hammond and others, both reported in 1970 on a study of eighty-six male beagle dogs trained to smoke cigarettes through a tracheostoma. Before being asked to comment on these studies, Professor Idle gave evidence that in the period 1965 to the end of 1971, a number of studies were conducted in which mice, rats, hamsters and dogs were exposed to cigarette smoke by inhalation. In the studies with mice, rats and hamsters, no squamous cell carcinomas of the lung were reported. Only the study by Auerbach and Hammond reported the development of "invasive squamous cell carcinoma of microscopic size" in two dogs.
[5.656] In RCP 1971, p.37, para.3.4, it was stated:
"In many experiments animals have been made to inhale cigarette smoke for various periods. The efficiency of the filtration by the noses of animals used is such that a large proportion of the smoke is removed before it reaches their lungs [...]. Until recently there was only one report of an inhalation experiment with mice in which lung cancer had been induced and this was of a different type from human lung cancer [...]. Production of typical cancer of the larynx in hamsters by exposure to cigarette smoke has now been reported [a reference to a paper by Dontenwill and others then in press]. Furthermore, two out of twenty-four dogs taught to inhale cigarette smoke directly into the lungs by means of tubes inserted into their windpipes have now developed lung cancer, similar to the human disease, after two and a half years of smoking seven unfiltered cigarettes daily. Ten developed other types of lung cancer. Dogs which had smoked the same number of filtered cigarettes during the same period did not develop cancer though some showed precancerous changes."
[5.657] The information in the latter part of this paragraph was derived from a report by Auerbach and Hammond in 1970 entitled "Smoking gives dogs invasive lung cancer". Professor Idle said that the study of Auerbach and Hammond attracted much contemporaneous attention and was highly controversial. For example, Sterling wrote in 1971 that it appeared from the description of the experiment that the lungs of the dogs were infused daily with unfiltered air drawn directly into the lungs through holes in the trachea and so by-passing the usual protective mechanisms which kept out or removed dust, bacteria and other impurities. It was well known, he wrote, that many changes in lung tissue, including true metastatic cancers, could be observed as a consequence of introducing dust and impurities into the lungs of animals. Professor Idle added a number of comments of his own. First, the control animals were very small in number, and they were not true controls, inasmuch as they had at some time been exposed to cigarette smoke, being dogs which had been taught to smoke earlier and had been tracheostomised. The eighty-six dogs had been divided into four exposure groups and one control group of eight dogs. In a normal laboratory experiment the number of controls was at least equal to the number of test animals, and in many studies the number of controls was double that of exposed subjects. The small number of controls used by Auerbach and Hammond did not provide a basis for saying that what was observed was greater than would be expected from the control observations. There was, he said, an adage well known amongst laboratory investigators that an experiment was only as good as its controls, and he regarded this as axiomatic. If the controls had been taught to smoke earlier, then they had been exposed to cigarette smoke. Secondly, there were doubts about the interpretation of the reported results. The authors reported that twelve out of seventy-four dogs that had smoked non-filter cigarettes developed locally invasive bronchiolo-alveolar carcinomas. They also reported that two out of eight control dogs developed non-invasive bronchiolo-alveolar carcinomas. Furthermore, they reported that two out of twelve dogs with bronchiolo-alveolar carcinomas also developed "invasive squamous cell carcinomas of microscopic size". Professor Idle said that bronchiolo-alveolar carcinomas were rare in humans and, as stated by the authors, the bronchiolo-alveolar tumours found in their smoking dogs differed in several respects from bronchiolo-alveolar tumours found in man. These tumours also occurred in 25% of control dogs. Moreover, he said, the relevance to human lung cancer of the finding of "invasive squamous cell carcinomas of microscopic size" in the lungs of two dogs was unclear. Such microscopic lesions were not invasive and metastasising squamous cell cancers. They do not therefore fulfil many of the criteria for lung cancer per se.
[5.658] In Professor Idle's judgment, in the period prior to 1972, no animal exposed to fresh whole cigarette smoke developed squamous cell carcinoma of the lung. He then turned to consider the research into the constituents of cigarette smoke in the period from 1965 to 1971. Researchers continued, during this period, with an effort to identify a constituent of cigarette smoke that might be responsible for human lung cancer, so that it might be removed. Several constituents and groups of constituents were the focus of this endeavour.
[5.659] Many tobacco smoke constituents existed in both a particulate phase and a gas phase, in a dynamic equilibrium between the two. The total particulate matter was the material that could be trapped by passing the cigarette smoke through a Cambridge glass fibre filter. Gas phase was that material which escaped such filtration. By the end of 1971, approximately 1,000 constituents of tobacco smoke condensate had been identified. Fewer than 100 constituents of the gas phase had been identified by this time.
[5.660] Before discussing the biological properties of the constituents of tobacco smoke, and in particular their potential as a cause of human lung cancer, it was necessary to define the occurrence of such constituents. By 1971, little laboratory work, either chemical or biological, had been carried out with the total particulate phase. In almost every case, the experiments were performed with tobacco smoke condensate, which was used as a surrogate for tobacco smoke. Tobacco smoke was seen as sufficiently chemically complex that the scientific literature of the day was replete with phrases such as "thousands of compounds", which gave tobacco science an air of mystery. According to Wynder and Hoffmann, writing in 1967, the total mainstream smoke from a cigarette amounted to about 500mg, more than 85% of which was comprised of the atmospheric gases nitrogen, oxygen and carbon dioxide. The total particulate matter comprised only about 8% by weight. The gas phase, excluding the common and inert atmospheric gases, comprised about 1% of the smoke by weight. Thus, the chemical constituents of the particulate phase and gas phase combined, that had generated toxicological interest, comprised less than 10% by weight of the cigarette smoke.
[5.661] By the end of 1971, polycyclic aromatic hydrocarbons were among the most studied group of constituents in tobacco smoke condensate. Benzo[a]pyrene continued to be the focus of research interest. Professor Idle said that it was pertinent, in analysing whether or not the total polycyclic aromatic hydrocarbons, as they are found in tobacco smoke, might cause lung cancer, to ask to what extent their presence explained the observed carcinogenicity of tobacco smoke condensate on mouse skin. He divided this into two questions: first, whether there was enough present to explain the carcinogenicity of tobacco smoke condensate; and, secondly, to what extent might interactions with other constituents in the mixture influence the mouse skin carcinogenicity of the polycyclic aromatic hydrocarbons. As he had already explained, compounds acting in combination might either act as co-carcinogens or as anticarcinogens. As to the first of these questions, benzo[a]pyrene could be calculated to possess approximately 80% of the mouse skin carcinogenicity of the polycyclic aromatic hydrocarbon group of constituents and thus it might be used as a surrogate for the group of carcinogenic polycyclic aromatic hydrocarbons as a whole. Earlier studies had led to the conclusion that benzo[a]pyrene could not account for more than 2.5% of the carcinogenic activity of tobacco smoke condensate on mouse skin. The total polycyclic aromatic hydrocarbon fraction was therefore unlikely to contribute more than 3% of the total activity of the tobacco smoke condensate. Professor Idle agreed with the conclusions of Wynder and Hoffmann in 1967 and Van Duuren in 1968 that the polycyclic aromatic hydrocarbons did not exist in sufficient quantity in cigarette smoke to explain the carcinogenicity of tobacco smoke condensate on mouse skin. As to the second question, various studies, including that of Wynder and Hoffmann in 1967, had shown that the mouse skin carcinogenicity of benzo[a]pyrene, and indeed other carcinogenic polycyclic aromatic hydrocarbons, could be diminished by the presence, in the tobacco smoke condensate, of other members of this constituent group and indeed by other chemically-unrelated constituents. In addressing this question, the effect of interactions that might enhance the carcinogenicity of benzo[a]pyrene must also be considered. By the end of 1971, tumour promoters, particularly the phenols, that could enhance the mouse skin carcinogenicity of benzo[a]pyrene, had been identified in cigarette smoke, but in this period the interaction of polycyclic aromatic hydrocarbons and phenols was not understood. Accordingly, by the end of 1971, the polycyclic aromatic hydrocarbons, as they existed in cigarette smoke, had not been shown to be a cause of lung cancer in cigarette smokers.
[5.662] Having already considered phenols, aldehydes and acids in the context of his discussion of ciliastasis, Professor Idle next briefly mentioned the isotope polonium-210, which was present in the atmosphere, the soil and certain plants, including tobacco. Despite an initial difference of views, by 1971, he said, there was general agreement that polonium-210, as it existed in tobacco smoke, was not a cause of lung cancer in smokers. Another group of compounds that were being discussed in this period as a possible cause of lung cancer in smokers were the nitrosamines. By the end of 1971, most commentators were sceptical that nitrosamines actually occurred in tobacco smoke, believing that it was more likely that they were being formed as artefacts during the condensation process. The formation of nitrosamines from amines and nitrogen oxides took place in an exothermic reaction, in which heat was given out, so the cooling process in the traps in which tobacco tar condensate was formed provided an appropriate environment for the formation of nitrosamines, since the heat of the reaction was taken away in the cooling process.
[5.663] Among other constituents which were considered during this period were selenium in cigarette paper, pesticides and nickel. In each of these cases, the constituents were researched, often by only one laboratory, and then discounted as potential causes of human lung cancer. By 1971, there was general agreement that these other constituents, as they existed in tobacco smoke, were not a cause of lung cancer in smokers.
[5.664] By the end of 1971, Professor Idle said, no constituent or constituents had been identified in cigarette smoke that accounted for lung cancer in cigarette smokers. Dr William Lijinsky of the Oak Ridge National Laboratories in the United States gave evidence in the course of hearings before a subcommittee of the United States House of Representatives Committee on Government Operations in 1971 that "in spite of twenty years' work on tobacco smoking we cannot identify or nobody has been able to identify a carcinogen to explain the incidence of lung cancer in man."
[5.665] Professor Idle's conclusions were that, as at the end of 1971, none of the competing theories had been established as the cause of human cancer. None of these theories provided an adequate basis for explaining how lung cancer arose. In this period, researchers continued to paint tobacco smoke condensate on the backs of mice. Skin painting protocols were improved and standardised by the Harrogate Laboratories. None of the skin painting studies, or any other investigation, resulted in the identification of a constituent that explained the appearance of lung cancer in cigarette smokers. By the end of 1971, a number of researchers conducted studies in which several species of laboratory animals were exposed to whole cigarette smoke by inhalation in an attempt to determine if cigarette smoke could cause lung cancer in these species, and thereby provide data concerning a possible relationship between human lung cancer and cigarette smoking. These studies were not able to produce squamous cell carcinoma of the lung in the exposed animals. In his opinion, cigarette smoking had not been established as a cause of lung cancer as at the end of 1971.
[5.666] Professor Idle next discussed laboratory investigations with tobacco derivatives from 1972 to the present. He said that some researchers continued to focus their investigations on the small group of tobacco smoke constituents that had been reported to be carcinogenic in the prior periods. Others took the view that a complex mixture such as cigarette smoke must be evaluated in its entirety, due to the myriad interactions that might occur when an organism was exposed to that complex mixture.
[5.667] In this period, a number of human cancers that had been associated with specific chemical exposures were successfully modelled in experimental animals. By this he meant that experimental animals exposed to those chemicals developed the same tumour type in the same organ as that observed in humans exposed to these chemicals. Exposures and diseases in which this had been demonstrated included: asbestos and pulmonary mesothelioma; vinyl chloride monomer and liver angiosarcoma; bis-chloromethyl ether and small cell lung cancer; aflatoxin B1 and hepatocellular carcinoma; and β-naphthylamine and bladder cancer. Additionally, it had been stated in USSG 1964 that experimental animals had been reported to develop squamous cell carcinoma of the lung following the inhalation of a number of carcinogens, including several dusts, metals and radioactive materials. Several of the studies in this time period reported that cigarette smoke reached the lung during inhalation experiments in the mouse, the hamster and the dog, using a technique called "nose-only exposure" and the addition of tracer materials to experimental cigarettes. In view of these reports, a number of researchers conducted experiments in which animals were exposed to whole cigarette smoke, in the period after 1971.
[5.668] The most substantial study of cigarette smoke inhalation in animals was that funded under contract by The Council for Tobacco Research - USA, Inc, performed in mice by the contract toxicology laboratory, Microbiological Associates of Bethesda, Maryland. This company's laboratory was located close to the NIH, and it was their principal contractor. More than 10,000 hybrid mice, containing several strain backgrounds, were exposed to the smoke from over 800,000 cigarettes. In one experiment, a total of 2053 female mice were exposed (nose only) to fresh whole cigarette smoke, together with 1014 sham-exposed mice and 449 shelf controls. It was reported that the results of this study suggested that unfiltered experimental reference cigarette smoke had weak carcinogenic activity in mouse lung tissue. No bronchogenic squamous cell carcinomas were observed in any of these animals. Alveolar adenocarcinomas, as well as other alveologenic lesions, were observed in smoke-exposed animals, and the incidence was higher in the smoke-exposed animals than in the sham-exposed animals. The difference in incidence of these alveolar adenocarcinomas and other alveologenic lesions, however, never reached a statistical level of significance. In addition, 20-methylcholanthrene was administered to mice as a positive control and produced squamous cell carcinoma of the lung. Daily smoke administration was reported not to alter the incidence or distribution of the 20-methylcholanthrene lesions. Moreover, benzo[a]pyrene was administered intratracheally to a further 320 (smoke-exposed), 260 (sham-exposed) and 130 (shelf controls) mice. Daily smoke exposure for 110 weeks was reported not to alter the incidence or the latency of benzo[a]pyrene-induced lung tumours. In his foreword to the final report, CTR 1984, Dr Sheldon C Sommers, Scientific Director of The Council for Tobacco Research - USA, Inc, wrote:
"The results observed included these: none of the smoke-exposed animals developed pulmonary squamous cell carcinoma; mice pretreated with a pure carcinogen did not have significantly increased lung cancers after prolonged smoke exposure; and the overall numbers of pulmonary neoplasms identified were not statistically significantly different in the smoke-exposed mice, compared to the sham smoke-exposed or shelf control mice."
[5.669] Witschi and colleagues reported in 1997 that in a study carried out by them the incidence of pulmonary adenoma/adenocarcinoma was not statistically significantly different between two exposed groups of mice.
[5.670] Among studies in rats, Davis and colleagues, working at the Harrogate Laboratories, reported in 1975 that neither squamous cell carcinomas nor any other types of lung tumour were reported to occur in any of the groups of rats exposed to tobacco smoke condensate and the gas phase of cigarette smoke in various combinations. In another study, also reported in 1975, of 406 rats that had been exposed to cigarette smoke, four reportedly developed squamous cell lung tumours, only one of which was described as a squamous cell carcinoma. The authors reported that the incidence of squamous cell carcinoma in the exposed and control groups was not statistically significantly different. They also reported that the effects of a single administration of benzo[a]pyrene, followed by smoke exposure, were in general similar to those of smoke exposure only. Kendrick and colleagues reported in 1976 on an experiment in which 160 rats were exposed to the smoke of seven to ten cigarettes per day. Sham exposed and shelf controls were also used. No lung tumours were reported in any group.
[5.671] Dalbey et al. 1980 reported on an experiment in which eighty female rats were exposed to the smoke from seven cigarettes per day, five days per week, for the lifetime of the animals. Seven of the smoke exposed rats were reported to have developed eight lung tumours, including adenoma (5/80), alveologenic carcinoma (2/80) and squamous cell carcinoma (1/80). One of the thirty sham-exposed control animals, but none of the sixty-three untreated controls, was also reported to develop a lung tumour, alveologenic carcinoma. The authors stated that the occurrence of one squamous cell carcinoma of the lung in eighty rats was not statistically significant. However, Professor Idle commented, they combined the total number of lung tumours with two nasal tumours observed in the exposed animals and reported this total to be statistically significantly greater than the tumours observed in the combined control groups. They did not discuss the observation reported by them that tumours in the pituitary gland, uterus and ovaries, together with the lymphoid system, occurred statistically significantly more frequently in untreated and sham-exposed groups than in the smoke-exposed group.
[5.672] Wehner and colleagues reported in 1981 on the results of a study in which groups of eighty female rats were exposed to the smoke from eight cigarettes, seven days per week, up to twenty-four months. Sham-exposed and untreated controls were also employed. A single squamous cell lung carcinoma (reported by the authors as a "lung epidermoid carcinoma") was reported to have developed in a group exposed to low-tar, medium-nicotine cigarette smoke. Heckman and Dalbey reported in 1982 on the results of a study in which female rats were exposed to cigarette smoke. No tumours of any kind were reported to develop in any rat, even with the greatest dose and duration of smoke exposure. Heckman and Lehman reported in 1985 on a study in which female rats were exposed to the same dose of cigarette smoke as in the preceding study. No lung tumours were reported in any rat.
[5.673] Turning to studies in hamsters, Professor Idle mentioned first a report by Dontenwill and others in 1973 on a study in which large numbers of hamsters were actively exposed (nose only) to the smoke of various cigarette types. The authors reported that the incidence of "early invasive carcinoma" in the larynx, which did not invade the underlying cartilage, was greater in the smoke-exposed animals than in the controls. No statistical evaluation of these data was reported. In a study by Wehner and others, reported in 1974, on the exposure of hamsters to cigarette smoke for the lifetime of the animal, with a similar number of sham-exposed controls, neither lung nor laryngeal tumours were reported to develop in the smoke-exposed animals. Bernfeld and others reported in 1974 on a study in which hamsters were actively exposed (nose only) to cigarette smoke. No lung tumours were reported in the exposed groups of either of the strains that were used. The authors reported the appearance of "microinvasive" carcinomas in the larynges of the exposed hamsters. In another study by Dontenwill and others, involving the active exposure (nose only) of a large number of hamsters to cigarette smoke for various types of cigarettes, no lung carcinomas were reported in any animal. Bernfeld and others reported in 1979 on the results of another study carried out by them, in which no carcinomas of the lung were reported. Professor Idle commented that in all the aforementioned hamster smoke inhalation studies, the expected end point was bronchogenic squamous cell carcinoma. Such tumours were not found. However these studies did report various incidences of laryngeal carcinoma in the experimental animals. This had been attributed to the unusual anatomy of the hamster airways by Dontenwill and others, writing in 1973. They wrote that since the flow characteristics of air in the upper regions of the respiratory tract were different in the golden hamster and in man, the highest deposition of smoke particles per surface unit occurred in the region of the larynx. The concentration there was approximately 300 times that in the lungs and bronchi.
[5.674] Hecht and others reported in 1983 on a different kind of study carried out on hamsters, to examine the effects of cigarette smoke inhalation on the carcinogenicity of subcutaneously injected NNK, a tobacco-specific nitrosamine known in 1983 to be carcinogenic in hamsters to the nasal mucosa, trachea and lung. Hamsters were given an injection of increasing doses of NNK, and in addition identically sized groups were either sham-smoked or exposed to the diluted smoke from thirty cigarettes, twice daily for seventy-two weeks. It was reported that eight out of fifty-nine hamsters developed adenomas and adenocarcinomas of the lung after NNK administration alone, and that when animals that had been injected with NNK were also exposed to cigarette smoke inhalation, eighteen out of fifty-nine developed adenomas and adenocarcinomas of the lung. No statistical evaluation of these data was reported, and in addition no squamous cell carcinomas of the lung were reported.
[5.675] Among studies in dogs, an unpublished industrial report from Hazleton Laboratories America, Inc in 1971, cited by Coggins in a review article in 2001, described a study in which 240 male beagles were exposed to cigarette smoke by tracheostomy for 108 weeks (the same procedure as that used by Auerbach and Hammond). There were no reports of neoplasia.
[5.676] Professor Idle then offered a critique of the inhalation studies. He said that the negative results for squamous cell carcinoma obtained in the cigarette smoke inhalation studies in laboratory animals should be viewed alongside similar investigations of other inhaled substances. As expected by the researchers, the inhalation of certain dusts, metals and radioactive materials by mice, rats, hamsters and dogs elicited squamous cell carcinoma of the lung. These laboratory studies were prompted by the occurrence of squamous cell lung cancers in humans exposed in the work place to dusts, metals and radioactive materials.
[5.677] The largest and best design smoke inhalation study was that performed by Microbiological Associates, reported in CTR 1984. In this study, no squamous cell carcinomas were reported to develop in mice exposed to fresh whole cigarette smoke. Positive controls demonstrated that the mice used in the study were capable of developing squamous cell carcinoma of the lung when exposed to the carcinogen 20-methylcholanthrene. Inhalation studies, such as the Microbiological Associates study, demonstrated that the inhaled cigarette smoke alone did not cause squamous cell lung carcinoma in laboratory animals. However, the question as to whether or not inhaled cigarette smoke might be acting as a tumour promoter had also been tested in the laboratory. When known chemical lung carcinogens, such as 20-methylcholanthrene, benzo[a]pyrene or NNK, were first administered to animals, followed by long-term inhalation exposure to cigarette smoke, the carginogenicity of the chemical lung carcinogens was not enhanced. It was unlikely, therefore, that cigarette smoke was acting as a promoter of tumour initiating chemicals.
[5.678] Certain workers reported the development of laryngeal lesions, including laryngeal carcinomas, in hamsters that had been exposed to cigarette smoke inhalation through the nose. Hamsters were obligatory nose-breathing animals. Particulate matter from the cigarette smoke aerosol was deposited at high concentration on to the tiny surface area of the larynx due to the peculiar hamster physiognomy. Because the observations were made in the larynx, the hamster studies did not tell us anything about the origins of lung cancer. In Professor Idle's opinion, there was no good explanation as to why Syrian golden hamsters developed laryngeal carcinoma, but not lung cancer, after the inhalation of cigarette smoke, except that it was a species-specific phenomenon and highly strain-dependent.
[5.679] Critique of the controversial study reported by Auerbach and Hammond in 1970 continued into the current period. For example, Hoffmann and Wynder commented in 1976 that the forced exposure of dogs to cigarette smoke by means of a tracheostomy led to hyperplastic and metaplastic changes in the bronchi. The observed carcinomas in situ needed reconfirmation. Professor Idle said that the claims by Auerbach and his colleagues had since been dismissed in IARC 1986. Their expert working group, comprising Doll, Peto, Hoffmann, Wynder and others noted that Auerbach and his colleagues had failed to include measures of exposure to cigarette smoke, had included an inappropriately small number of control dogs, and had failed to mentioned focal inflammatory lesions or to give any information about the upper respiratory tract or other organs. They also described the authors' interpretation that the photomicrographs represented neoplasia as "not entirely convincing".
[5.680] Professor Idle said that the foregoing passage of his evidence demonstrated that laboratory animals exposed to whole cigarette smoke by inhalation did not develop squamous cell carcinoma of the lung.
[5.681] The next topic was short-term mutagenicity testing. Professor Idle was told that Dr Kerr had given evidence that "there are things in tobacco smoke which cause lung cancer", a view based on, inter alia, reports that chemicals in tobacco smoke had induced changes in human cells, in tissue culture for example, recognised as being associated with the development of the malignant process. Professor Idle said that he recognised that this was a process that occurred. Ames proposed in 1974 that carcinogens caused cancer by somatic mutation. He also suggested that a test system (which became known universally as "the Ames test"), that utilised frameshift mutagenesis in bacteria, be employed to detect carcinogens. For the first time, it appeared that the carcinogenic property of a chemical could be tested for in a rapid, simple and inexpensive laboratory assay that did not require recourse to long-term studies in large numbers of animals. The proposal was an attractive one and was enthusiastically, if somewhat uncritically, received by the cancer research community. Other short term mutagenicity tests followed. These were sometimes referred to as tests for genotoxicity and they were sub-divided into tests that employed sub-mammalian systems (other than the Ames test) or cultured mammalian cells, and utilised end points such as chromosomal abnormalities in mammalian cells, DNA damage and repair, and cell transformation or altered growth properties, all of which were conducted in vitro.
[5.682] Since the mid-1960s, Ames had been working on the genetics of the bacterium Salmonella typhimurium, with particular emphasis on the "histidine operon", a genetic element in the bacterial chromosome that regulated the bacterial synthesis of the essential amino acid histidine. Organisms required twenty separate amino acids for the synthesis of proteins, such as enzymes; some of these were synthesised endogenously, while others were obtained only from the diet. Humans could not, but bacteria could, synthesise histidine. By introducing mutations into the histidine operon, Ames developed bacteria that required exogenous histidine for growth. He then discovered that certain chemicals could reverse these mutations, and thus permit the bacteria to synthesise histidine again. This could be observed by the growth of colonies of bacteria on histidine-free nutrient in a dish. This rekindling of the ability of the bacteria to grow in the absence of exogenous histidine was the basis of the Ames test of mutagenicity of chemicals. Ames went on to liken the mutagenicity observed in his in vitro test to in vivo carcinogenicity, when he advocated his test to the cancer research community.
[5.683] Further genetic engineering of the bacteria took place. Ames developed new strains of bacteria lacking the usual impermeable coat and thus permitting large mutagenic molecules to enter the bacterium. These molecules introduced a new mutation into the bacterial genome at a particular site and switched on the bacterial apparatus that synthesised histidine. Next, the removal of a key DNA repair enzyme prevented the bacterium from repairing this new mutation. In combination, these two manipulations to the basic tests amplified the response of the bacteria and thus their sensitivity to mutagenic molecules. Finally, since almost all carcinogens required metabolic activation by an enzyme system which was lacking in the bacteria, Ames made a further refinement by adding fractions of rat or human liver homogenates to the culture. The test was now complete. Liver enzymes activated the added chemical, reactive metabolites entered the quiescent bacterium and switched the histidine synthesis control system to an active state, thereby permitting the bacterium to synthesise histidine, and to grow and divide, producing a visible colony. The more colonies, the more mutagenic the chemical and, according to Ames, writing in 1974, the more carcinogenic the chemical. This was a simple and quick test, and it was not long before a plethora of chemicals were tested in this assay and declared as mutagens with potential carcinogenic activity. These included polycyclic aromatic hydrocarbon epoxides, aromatic amines, aflatoxin and benzo[a]pyrene.
[5.684] Meanwhile the process of testing chemicals on rodents, particularly in the National Toxicology Program (NTP) in the United States, continued. Data were accumulated over a considerable period on chemicals that were carcinogenic or not in rodents, according to NTP criteria, and at the same time the same chemicals, in general terms, were being tested in the Ames test. Comparison of the two sets of results showed that there were chemicals that were carcinogenic in rodents but not mutagenic in the Ames test and, more importantly, that there were a considerable number of chemicals and groups of chemicals that were mutagenic in the Ames test, but for which no carcinogenicity could be demonstrated in laboratory rodents.
[5.685] Professor Idle said that to see the bigger picture of the impact of the Ames test on the investigation of human lung cancer in cigarette smokers, it was pertinent to examine the recent literature in this area. Ames had proposed at the outset, in 1974, that his assay could be used to determine the mutagenicity of human urine concentrates. A small number of papers subsequently reported that smokers voided a urine that was more mutagenic in the Ames test than that of non-smokers. In order to understand the origin of these bacterial mutagens in smokers' urine, Doolittle and colleagues studied the phenomenon in rats. High-dose intraperitoneal injections of cigarette smoke condensate in rats produced a urine that was two to three times more mutagenic than background. However, when rats were exposed to cigarette smoke, by nose-only inhalation, for one hour per day up to ninety days, no mutagenicity could be detected in the urine. Thus, direct investigation in rats of the effects of cigarette smoke and its derivatives cast doubt on the interpretation of bacterial mutagenicity of smokers' versus non-smokers' urine. Professor Idle said that in the absence of more detailed research into this subject it was not possible to draw any firm conclusion regarding the relevance of the Ames test to the occurrence of lung cancer in cigarette smokers.
[5.686] His conclusions regarding short-term mutagenicity testing were that bacterial mutagenicity as measured by the Ames test for a series of chemicals did not correlate with in vivo carcinogenicity in experimental animals. Mutagens were not necessarily carcinogens and carcinogens were not necessarily mutagens. Bartsch and Tomatis stated in 1983 that long-term animal tests were still the only ones capable of providing evidence of the carcinogenic effect of a chemical. In relation to short-term mutagenicity tests, they concluded that the conflicting results of experimental data published so far with regard to a possible quantitative correlation between the potency of a chemical carcinogen in animals and its activity in short-term mutagenicity tests did not yet sufficiently establish such a relationship for all classes of carcinogens to allow its general use for the confident prediction of carcinogen potency of new compounds. Professor Idle said that whatever the Ames test might demonstrate regarding the mutagenicity in bacteria of extracts of urine from animals and humans exposed to cigarette smoke, the fact remained that animals exposed to whole cigarette smoke did not develop squamous cell carcinoma of the lung. In his opinion, whilst tobacco smoke condensate was mutagenic in the Ames test, the mutagen itself had not been identified, and the Ames test was, in any event, merely a first screen for mutagenicity. Whether a mutagen was a carcinogen required investigation in long-term rodent carcinogenicity studies.
[5.687] Professor Idle then went on to consider research in the period 1973 to the present date into the individual chemical constituents of cigarette smoke with a view to understanding which constituent or group of constituents might cause lung cancer in cigarette smokers. Having reviewed various areas of research in this period, he concluded that as at September 2003, no constituent or constituents had been identified in cigarette smoke that accounted for lung cancer in cigarette smokers. As stated by Peto and Doll in 1985, thirty years of laboratory research had yet to identify reliably the important carcinogenic factors in cigarette smoke. In his judgment, nothing had appeared in the scientific literature since 1985 that would alter this conclusion.
[5.688] Professor Idle then discussed the challenges in analysing complex mixtures. After referring to chemical and biological interactions and ambiguous carcinogens, he summarised the position by saying that the problems in evaluating the toxicology of cigarette smoke from a component-based point of view (bottom-up) were also shared by other complex mixtures that had been studied, including air pollution, groundwater pollution and diesel exhaust. The dilemma existed on many levels. First, there were at least 4,000 chemical entities in cigarette smoke. Little was known about the propensity of these constituents to react chemically together both within and without the body to produce novel and undetected chemical entities. There had been very little attempt to elucidate new chemical entities in cigarette smoke for many years, but it had been suggested that there might be hundreds, and perhaps even thousands, of polycyclic aromatic hydrocarbons in coal tar, and scientists had stopped looking for them. Secondly, little was known of the myriad spontaneous chemical reactions that might occur in vivo between different constituents of tobacco smoke, between smoke constituents and endogenous molecules. Moreover, the biological ramifications of such chemical reactions were unknown, but might include the slowing or enhancement of biological processes, both by inhibition and induction of enzyme systems. We did know, however, that the entire encyclopaedia of organic molecules was represented within cigarette smoke and that spontaneous chemical reactions must therefore occur. In view of the foregoing, analysis of cigarette smoke that was predicated on an evaluation of the individual components of this complex mixture was not achievable.
[5.689] By at least the 1980s, there was an awareness of the problems inherent in the evaluation of complex mixtures using a component-based approach. These problems arose from attempts to model the toxic responses, particularly carcinogenesis, of a mixture based upon knowledge about the toxicity of its component parts. In the United States, this awareness culminated with the National Institute of Environmental Health Sciences asking the National Academies' National Research Council to evaluate the toxicity testing of mixtures and to make recommendations for improving that testing. A committee which was formed to consider methods for the in vitro toxicity testing of complex mixtures under the auspices of the Board on Environmental Studies and Toxicology of the National Research Council's Commission on Life Sciences concluded in 1988 that a new approach to toxicological testing of mixtures was needed. They observed that people were seldom exposed to single chemicals. Most substances to which they were exposed, whether naturally or artificially produced, were mixtures of chemicals. Mixtures that were of particular concern included chemicals generated in fire, hazardous wastes, pesticides, drinking water, fuels, and fuel combustion products. They wrote that testing complex mixtures presented a formidable scientific problem. The key to attacking this problem lay in the analysis and planning of the strategy or experimental approach. The extent and nature of testing should be guided closely by recognition of what was known and what needed to be learnt. If the question being posed was related to the effects of a mixture, the strategies invoked involved toxicity testing of the mixture itself. Professor Idle said that he agreed with this. In essence, the Committee's recommendations were that the mixture should be tested under circumstances that reflected human real life exposures and then, if consistent positive results had been obtained, the bioassay could be used to track down which constituent or group of constituents might be responsible for the activity of the mixture.
[5.690] The Committee stated that bioassay-directed fractionation, which was not used for single agents, was the most useful strategy for studying mixtures. Professor Idle said that experiments had shown that, when groups of unrelated chemicals with unrelated targets of toxicity were administered to rodents simultaneously (i.e. as a mixture) at doses equal to their subsequent NOAELs (no-observed-adverse-effect levels), no toxic effects were observed; each chemical acted independently, not additively or synergistically. The same was true for chemicals with the same target but with different mechanisms of action. Studies in which similar chemicals with similar mechanisms of action and similar targets were administered simultaneously indicated that antagonism, not additivity or synergism, was the usual outcome, thereby reducing the overall effect. Antagonism was the inhibition by one chemical of the biological properties of another.
[5.691] With respect to cigarette smoke, with 4,000 chemical components, the "factorial design" would call for 24000-1 individual binary interaction experiments. This number might also be expressed as 1.32 x 101204. To perform the binary interaction experiments at two doses of each component, which was the recommended procedure, the number of experiments required rose to 1.55 x 101908. Professor Idle said that clearly this was neither feasible nor desirable. A rigorous, bottom-up, component-based, toxicological analysis of tobacco smoke was impossible.
[5.692] However, what was not only feasible, but had already been accomplished, was a "top-down", mixture-oriented evaluation of cigarette smoke. To date, inhalation studies with fresh whole cigarette smoke conducted in mice, rats, hamsters and dogs had not produced squamous cell carcinoma of the lung. Because the mixture itself did not produce squamous cell lung carcinoma in the test animals, the testing of fractions of the smoke was not indicated by the top-down protocol proposed in Feron et al. 1998. In short, the end point of the toxicological testing of cigarette smoke, with respect to squamous cell carcinoma of the lung, had been reached and no further investigations of this type were indicated. Laboratory studies on individual constituents of cigarette smoke, while they had undoubtedly revealed much about the various properties of those individual constituents, did not, he said, inform us about the biological properties of the complex mixture. These studies did not take into account the chemical and biological interactions, and their consequences, which he had considered.
[5.693] According to Feron et al. 1998, to perform a top-down analysis the mixture should be available in its entirety for toxicity testing, adequate doses needed to be applied that were not acutely toxic, and the analysis should not require extrapolation from low to high doses, because the interactions between chemicals might be dose-dependent. Professor Idle referred to a figure derived from Feron et al. 1998. According to the path shown in this figure, where complex chemical mixtures were readily available for testing in their entirety, testing in their entirety was indicated, and if no adverse effects were found in adequate and relevant studies, the research should stop, as the data should be useful for risk assessment. Even if one constituent or group of constituents of a complex mixture, such as cigarette smoke, induced changes in cells in tissue culture, one could not properly conclude that the complex mixture would induce similar changes, because the biological and chemical interactions between individual constituents were not the same as they were in the complex mixture.
[5.694] In explaining his conclusions as at September 2003, Professor Idle said that it was his judgment that cigarette smoking had not been established as a cause of human lung cancer. Indeed, as explained by him, the cause of cancer was unknown. Moreover, the mechanisms by which lung cancer developed were not known. Researchers had not produced squamous cell lung carcinoma in laboratory animals by inhalation exposure to cigarette smoke. No constituent or group of constituents, as they existed in the complex mixture which was cigarette smoke, had been shown to be a cause of lung cancer in smokers. In view of this, it could not be determined whether or not smoking caused Mr McTear's lung cancer.
[5.695] Professor Idle said that he did not always hold the view that cigarette smoking had not been established as a cause of human lung cancer. He entered into the field of lung cancer as a young researcher from pharmacology, a background of pharmacogenetics, and he adopted the received wisdom of the day that smoking causes lung cancer. More than twenty years of association with studies and becoming familiar with the literature caused him to look at the subject in a serious way, and then to understand that there were so many components missing that he could not see where, in the history of the subject, it had been scientifically established by experiment using the scientific method. So his view had altered within the last ten years.
[5.696] Professor Idle said that he had studied the question of what risk might be occasioned by environmental tobacco smoke. In 1995 and 1996 he was a member of a European working group which was set up to evaluate whether or not there was evidence that passive smoking could be a cause of lung cancer. The study was funded by the tobacco manufacturers Philip Morris, British American Tobacco and Rothmans, but the committee had formed its own independent judgment. Overall the committee concluded that it had not been established that environmental tobacco smoke exposure (passive smoking) could be a cause of lung cancer, and there was no reason to suppose that it did. All the contrary views of which Professor Idle was aware were based upon a study of the Environmental Protection Agency in the United States, who categorised passive smoking as a lung carcinogen. As he understood it, the conclusions of the report had been nullified because of false methodology in reaching conclusions, but the consequences of it remained unchanged. Professor Idle's committee had reviewed all the relevant scientific literature before reaching its conclusion.
[5.697] Professor Idle was asked to comment on Table 1.2 at p.17 of RCP 2000 (reproduced in UKHC 2000 at p.xv). The Table was headed "Estimated number and percentage of deaths attributable to smoking by cause, UK 1997". Professor Idle said that he believed that these estimates were made on the basis of epidemiological studies, but he did not know the methodology by which particular estimates were based on such studies. He could only speculate about the use of the word "attributable", but believed that it came from odds ratios, so-called relative risks. He did not know the methodology by which odds ratios could yield the figures given in the table. A number of diseases were listed under the heading "Diseases caused in part by smoking" (the word "smoking" is omitted in UKHC 2000). From a biological point of view, Professor Idle said that he did not understand how one could attribute diseases caused in part by smoking.
Cross-examination of Professor Jeffrey Idle
[5.698] In cross-examination Professor Idle said that he was not a smoker. He had first been approached in 1997 by a firm of New York lawyers, acting for ITL, who asked him to review literature in relation to lung cancer. He had not been asked to give advice to the Board of ITL. He had given an oral presentation at a seminar at which Board members were present, in about 2001. The presentation took about one hour. He might have provided those present with a copy of his PowerPoint presentation. It was a general presentation on lung cancer. His view when he gave it was the same as when giving evidence. He had been approached by a scientist from a law firm in London, representing Philip Morris, to chair the committee on environmental tobacco smoke and lung cancer. He did not know whether it would have been known that he was someone who did not subscribe to the generally held view that tobacco smoking caused lung cancer. Each member of the committee billed his consultancy fees separately, for time and travelling expenses.
[5.699] When Professor Idle moved from the University of Newcastle to the University of Trondheim, he received a bridging grant of about $30,000 from January 1996 to help him to move from one place to another and start up research in a new place. The funding continued for one or two years. He did not know whether other people at the University of Trondheim received research funding from the tobacco industry. He believed that the Council for Tobacco Research in New York was funded by the American tobacco industry. He was a member of the Scientific Advisory Board from 1986 to 1991.
[5.700] Having first been approached in connection with the present case he reviewed scientific literature and prepared a briefing document for the lawyers regarding the scientific literature that he had read and analysed. This was sent to New York. He did not know whether it was made available to the Board of ITL. This was in 1998. He continued to work with the New York lawyers on another period of literature in 1999 and produced another report. He thought the intention was for it to be used to brief counsel in the present case on the scientific issues. He thought that he was aware that it was a case involving Mrs McTear and ITL from the outset of his involvement, in late 1997 or early 1998. It would not be fair to describe him as an advisor to ITL: he had never met anybody from ITL. He was not giving advice, but was reading the literature and producing a scientific analysis of it as a scientist. He had once previously given a deposition in a case brought against manufacturers of smokeless tobacco (chewing tobacco and snuff).
[5.701] He went on leave of absence from the University of Trondheim in about September 1998. He retained a portion of his salary from there and had supported himself by carrying out consulting work. He had not been funded by ITL. He had been paid for the time spent on preparing the report. He agreed that he had been paid a "six-figure sum". He was provided with a copy of the pleadings in the present case. He had not made any special study of Mr McTear's medical records. He was not made aware of ITL's position prior to writing his report. He thought he did know their position in relation to the question of smoking and lung cancer: he understood that they believed that it had not been established that smoking caused lung cancer, at least that was what he had read in the press.
[5.702] Professor Idle was asked questions about the evidence given by Mr Davis to the House of Commons Health Committee on 13 January 2000, as recorded in UKHC 2000, pp.238ff. He was not aware of this publication, or of the words chosen by Mr Davis to state the position of ITL. He said that he was aware that ITL believed that smoking had not been shown to be the cause of lung cancer and other diseases, but he did not know that ITL did not deny that smoking was such a cause. He was not told that by any of the lawyers. He was not aware that ITL had never challenged, or sought to challenge, the public health message, or indeed to challenge the figures, as stated by Mr Davis. He had come to give evidence not knowing the position of ITL in this detail. Asked by me whether it would have made any difference to his approach if he had known, he said it would not; he was a scientist interested in what scientific experimentation produced and what one took from the laboratory.
[5.703] Professor Idle said that his area of scientific expertise, in relation to the present case, was his scientific investigation for more than twenty years of risk factors in lung cancer specifically, and the genetics of lung cancer. His general background and training were as a chemist and biologist. His conclusions were reached against that background and his research into lung cancer over an extended period of time. Counsel said that it was not challenged on his side that, as stated by Professor Idle, the mechanisms by which lung cancer developed were not known. The professor said that this came from his biology and chemistry background. Another statement that was not challenged was that researchers had not produced lung cancer in laboratory animals. This came from Professor Idle's review of the literature. Again, it was not challenged that no constituent or group of constituents, as they existed in the complex mixture which was cigarette smoke, had been shown to be a cause of lung cancer in smokers. Professor Idle said that this came from his review of the literature and his own background in carcinogenesis, chemistry and biology. His judgment that cigarette smoking had not been established as a cause of human lung cancer came from his general training in these areas, and his reading of an extensive literature. He was not an epidemiologist.
[5.704] Apart from the above-mentioned references in his evidence, he had not considered the epidemiology. He did not refer to more up-to-date material with respect to the epidemiology of lung cancer because he was not a epidemiologist, he had to leave that to others. Asked whether there was not a great deal of evidence which established cigarette smoking as a cause of lung cancer, he said no, he believed that there was evidence which had established an association, if counsel was referring to epidemiology. There were over 4,000 constituents in tobacco smoke, and there might be many more. Asked whether these constituents, whether singly or in combination, were a perfectly plausible mechanism for the causation of lung cancer, he said that this was so, if applied individually in animals. Experiments in animals formed part of the judgment.
[5.705] Professor Idle agreed that there was a general view in science and the medical profession, about the relationship between smoking and lung cancer, that smoking was a cause of lung cancer. Shown statements on cigarette packets that "Smoking kills", he said that he did not know if smoking killed or not. Asked about a packet bearing the statement "Smoking causes fatal lung cancer", he said that this had not been established. His position was that he did not know: it might, but it had not yet been established. He did not know why these warnings appeared on cigarette packets. He was not aware of Directive 2001/37/EC of the European Parliament and of the Council of 5 June 2001 on the approximation of the laws, regulations and administrative provisions of the Member States concerning the manufacture, presentation and sale of tobacco products. He agreed that in adopting this Directive, which included that each unit packet of tobacco products must carry a general warning, one of which was that "Smoking kills", and an additional warning, one of which was that "Smoking causes fatal lung cancer", the Parliament and the Council must have been satisfied of these matters.
[5.706] Asked what was the teaching in the medical schools about lung cancer and its likely causes, Professor Idle said that in none of the medical schools in which he had worked had he been involved in that part of the curriculum, so he did not know. He would accept that in the medical schools of the United Kingdom the current teaching was that cigarette smoking caused lung cancer. He had not read the Oxford Textbook of Medicine, but would accept counsel's statement (without seeing the text) that it was suggested in this book that cigarette smoking was a cause of lung cancer. Asked about the statement in UKWP 1998 that smoking killed, he said he did not know if smoking killed or not. He did not agree that this had been known for years. He did agree that the epidemiological evidence linking smoking with lung cancer had been known for years. As for the position of other tobacco companies in the UK with respect to cigarette smoking and lung cancer, he said that he was aware of the position of Philip Morris and British American Tobacco, but did not know about the others. So far as these two were concerned, he thought they had stated that there was evidence that supported the contention that smoking caused lung cancer, so they went further than he did. He did not know if he was "out on a limb", or was one of a very small minority who held this view.
[5.707] Professor Idle did not believe that epidemiology could ever uncover a cause in the case of medical conditions found in association with environmental exposure. In the case of asbestos and mesothelioma there was an animal model that replicated the disease seen in the exposed humans in the workplace. He thought that the association was made on the basis of epidemiology, but cause and effect required various other criteria to be met. Epidemiology on its own could never be sufficient. He was aware of the writings of Sir Richard Doll. His stature in the medical and scientific community was high, because he showed the association between lung cancer and smoking a very long time ago. He did not refer to any of Sir Richard's material in his paper because it was epidemiology, and it was not helpful to him in his analysis. He was looking at experimental carcinogenesis, the laboratory-based studies, not the statistical studies. Asked about Sir Richard's evidence that MRC 1957 effectively settled the issue as far as England was concerned, he said that he did not accept that statement because epidemiology on its own could not demonstrate the cause of lung cancer and it was not generally accepted in the scientific community that it did. Many scientists were unhappy with the extrapolation from association to cause. He had not dealt in his report with the epidemiological evidence, but had dealt with the statements derived from it. He was looking to see whether there were chemical and biological data which would add to the epidemiology. There was a whole period of science, for example, when viruses were seen as principal candidates of much of human cancer, and these were dealt with in his report. There were various statements from people saying that epidemiology on its own could never lead to cause and effect.
[5.708] Professor Idle explained that the Medical Research Council (MRC) was a body that co-ordinated the funding of medical research in the United Kingdom, a body made up of doctors, scientists and administrators. They had a number of ad hoc committees which looked at grant applications and were populated by scientists and medical doctors from around British universities, many of them professors. Their views on medical matters and with respect to grant funding were respected. Counsel asked Professor Idle about various passages in MRC 1957. On p.1524, conclusions were set out, one of which was in these terms:
"4. Evidence from many investigations in different countries indicates that a major part of the increase [in the death rate from lung cancer] is associated with tobacco smoking, particularly in the form of cigarettes. In the opinion of the Council, the most reasonable interpretation of this evidence is that the relationship is one of direct cause and effect."
The witness agreed that this statement was being put out in 1957 as an interpretation of the epidemiology. He could not say whether in concluding that there was a relationship of "direct cause and effect" the MRC represented the majority of scientific opinion at that time.
[5.709] In the paragraph preceding the conclusions, the MRC stated:
"In scientific work, as in the practical affairs of everyday life, conclusions have often to be founded on the most reasonable and probable explanation of the observed facts, and so far no adequate explanation for the large increase in the incidence of lung cancer has been advanced save that cigarette smoking is indeed the principal factor in the causation of the disease. The epidemiological evidence is now extensive and very detailed, and it follows a classical pattern upon which many advances in preventive medicine have been made in the past. It is clearly impossible to add to the evidence by means of an experiment in man. The Council are, however, supporting a substantial amount of laboratory research which may throw more light on the mechanism by which tobacco smoke and other suspected causative factors exert their effect, and which may thus eventually add to the degree of proof already attained as a result of studies of human populations. It must be emphasized, however, that negative results from work with animals cannot invalidate conclusions drawn from observations on man."
Professor Idle agreed that it was impossible to add to the evidence by means of an experiment in man, and that laboratory research continued after 1957. He regarded the last sentence of the paragraph quoted above as a statement of the opinion of the MRC. He thought it was possible that animal experiments could overturn interpretations of a statistical nature.
[5.710] In the summary of RCP 1962, at p.S3, it was stated:
"There has been a great increase in deaths from [cancer of the lung] in many countries during the past 45 years. [...] Many comparisons have been made in different countries between the smoking habits of patients with lung cancer and those of patients of the same age and sex with other diseases. All have shown that more lung cancer patients are smokers, and more of them heavy smokers than are the controls. The association between smoking and lung cancer has been confirmed by prospective studies [...]. All these studies have shown that death rates from lung cancer increase steeply with increasing consumption of cigarettes. Heavy cigarette smokes may have thirty times the death rate of non-smokers."
Asked about this passage, Professor Idle said that he was aware that there had been epidemiological studies. Prospective studies were one of the methods used by epidemiologists. He did not quarrel with this. It was one of the available methods of looking at associations. He was aware of the statistics about the death rates of heavy smokes and non-smokers. He said it had not been established scientifically that smoking caused lung cancer.
[5.711] In the main report, in the section on smoking and cancer of the lung, the following conclusion was stated at p.27, para.41:
"The strong statistical association between smoking, especially of cigarettes, and lung cancer is most simply explained on a causal basis. This is supported by compatible, though not conclusive, laboratory and pathological evidence namely (a) the presence of several substances known to be capable of producing cancer in tobacco smoke; (b) the production of cancer of the skin in animals by repeated application of tobacco tar; and (c) the finding, in the bronchial epithelium of smokers, of microscopic changes of the kind which may precede the development of cancer. The conclusion that smoking is an important cause of lung cancer implies that if the habit ceased, the death rate from lung cancer would eventually fall [...]."
Professor Idle said that he accepted that several substances known to be capable of producing cancer were present in tobacco smoke, and that there had been some episodes of production of cancer on the skin of animals by repeated application of tobacco tar. He was not familiar with the literature relating to the finding in the bronchial epithelium of smokers of microscopic changes of the kind which might precede the development of cancer. He said that the RCP was a group that represented the profession of medicine in the United Kingdom and he agreed that when they put out a report like this, they were putting their name behind it. He also agreed that the RCP was "an extremely prestigious professional body".
[5.712] Counsel next invited Professor Idle's comments on passages in USSG 1964. At p.iii the Surgeon General's Advisory Committee on Smoking and Health were listed. Professor Idle said that he had attended a presentation given by one of them, Dr Farber, who was talking about diet and cancer in rats. At p.x, in the foreword, the Surgeon General stated that he had appointed a Committee, drawn from all the pertinent scientific disciplines, to review and evaluate both the new and older data and, if possible, to reach some definitive conclusions on the relationship between smoking and health in general. Professor Idle agreed that it was appropriate to proceed in this manner. At pp.ix ff the assistance of a number of individuals was acknowledged. Professor Idle said that he was aware, in broad terms, of "the depths to which this Committee went in looking at the available literature and seeking the assistance of people outside", and that it was extensive.
[5.713] Counsel next asked Professor Idle about passages in IARC 1986. At pp.9-10 the members of the IARC working group on the Evaluation of the Carcinogenic Risk of Chemicals to Humans: Tobacco Smoking were listed. Of these members, Professor Idle said that he knew of DM De Marini who worked on bacterial mutagenesis at the Institute of Environmental Health Sciences and was regarded as an expert. HJ Evans of the MRC Clinical and Population Cytogenetics Unit at the Western General Hospital, Edinburgh, was, he said, an expert and leader in the field of human genetics. T Hirayama, of the Epidemiology Division of the National Cancer Center Research Institute in Tokyo, Japan, was a leader in the field of epidemiology, perhaps one of the more prominent in Japan; Professor Idle had been at a meeting with him. He knew of D Hoffmann, of the Naylor Dana Institute for Disease Prevention, American Health Foundation, Valhalla, New York, as a specialist in chemical carcinogenesis, specifically tobacco and cancer issues. MAH Russell, Reader in Addiction, Institute of Psychiatry, The Maudsley Hospital, London: Professor Idle believed that he was a leading man in his field. M Sorsa, Institute of Occupational Health, Helsinki, Finland: a specialist in occupational health and occupational carcinogenesis, well-thought of in her field. P Vineis, Unit of Cancer Epidemiology, Istituto di Anatomia Patologica, Turin, Italy: an epidemiologist with some interest in pharmacogenetic epidemiology, considered to be one of the top epidemiologists in Italy. NJ Wald, Department of Environmental and Preventive Medicine, The Medical College of St Bartholomew's Hospital, London: a leading epidemiologist in the United Kingdom. EL Wynder, President, American Health Foundation, New York: a specialist in tobacco and health and the author of papers referred to in the evidence. Professor Idle was also asked to indicate in passing that he recognised the names of the Chairman of the working group, Sir Richard Doll, and another member, Sir Richard Peto.
[5.714] At p.16 it was stated: "The IARC Monographs are recognized as an authoritative source of information on the carcinogenicity of environmental and other chemicals." Professor Idle said that he agreed with this, and that he thought the whole series of monographs was a wonderful source of tabulated information, of a pulling together of data, a very useful encyclopaedic source. He consulted the published volumes from time to time.
[5.715] Counsel read out the passage from pp.16-17 quoted in Professor Friend's evidence at para.[5.37]. Professor Idle said that this was one way for such a body to proceed. He was next asked to look at the list of references at pp.315-375. He agreed this was "quite a lot of material to review" and that this would be an appropriate thing to do. He said he did not know whether the findings of this body had ever been challenged.
[5.716] In the chapter relating to biological data relevant to the evaluation of carcinogenic risk to humans, it was stated in the summary at pp.194-195, under the heading "Carcinogenicity Studies in Animals":
"Considerable effort has been devoted to developing experimental animal systems to study the carcinogenicity of cigarette smoke. Useful models have been developed for testing both whole smoke, by inhalation, and smoke condensate, by topical application.
Studies involving inhalation of smoke are hampered by difficulties in reproducing the exposure of humans. Technical problems occur in the generation of smoke and its delivery to animals; moreover, the respiratory systems of animals and humans differ. Rodents are obligatory nose breathers, and the structure of their nasal turbinates is more complex than that of humans. Unlike humans, experimental animals smoke involuntarily with shallow, hesitant breathing patterns. Other difficulties are caused by the toxicity of nicotine and carbon monoxide. Despite these problems, however, informative data have been obtained concerning the carcinogenicity of whole smoke and its gaseous phase.
In some experiments in mice, exposure to whole cigarette smoke results in the induction of lung tumours. In one study involving long-term exposure of rats to cigarette smoke, tumours of the respiratory tract were induced. In hamsters, various experiments demonstrated reproducibly the induction of laryngeal carcinomas. [...]"
Professor Idle said he agreed that the respiratory systems of certain species of animals and of humans differed, that rodents were obligatory nose breathers and that the structure of their nasal turbinates was more complex than that of humans. He did not know that experimental animals smoked involuntarily, with shallow, hesitant breathing patterns. It was correct that other difficulties were caused by the toxicity of nicotine and carbon monoxide. He agreed that in some experiments in mice, exposure to whole cigarette smoke resulted in the induction of lung tumours, and said that he was aware of one study involving long term exposure of rats to cigarette smoking, in which tumours of the respiratory tract were induced. He said that certain strains of hamster developed laryngeal carcinomas.
[5.717] At the end of this part of the summary, at p.195, this statement appeared: "Cigarette smoke contains many chemicals known to be carcinogenic to experimental animals and/or humans [...]." When this was put to Professor Idle, he said that the first half of the statement was correct: there were chemicals that were carcinogenic to experimental animals, but could not be evaluated in humans. Obviously one could not do an experiment administering chemicals to humans. Carcinogenicity was demonstrated by producing tumours in an animal, including the human, and if you could not do the experiment in humans, then your definition was restricted to experimental animals.
[5.718] In the chapter setting out the Working Party's conclusions and valuations, at p.309 under the heading "Conclusions" and the subheading "(b) Carcinogenicity in animals" the following paragraph appeared, which Mr McEachran put to Professor Idle:
"Cigarette smoke has been tested for carcinogenicity in experimental animals by inhalation and by topical application of condensate and in other ways. Exposure of hamsters and rats to whole smoke results in the induction of malignant respiratory-tract tumours. Cigarette smoke condensate induces skin cancers in mice and rabbits after application to the skin, and lung cancers in rats after intrapulmonary injection. Cigarette smoke contains many chemicals known to produce cancer in animals and/or humans."
Professor Idle said that he agreed with this paragraph, except for the concluding statement about humans, for the reasons already given by him. In the same part of this chapter, under the subheading "(f) Cancer in humans", the following passage appeared, which again was read to Professor Idle:
"Lung cancer is believed to be the most important cause of death from cancer in the world, with estimated total deaths in excess of one million annually. The major cause of the disease is tobacco smoking, primarily of cigarettes. Risk of lung cancer is particularly dependent on duration of smoking; therefore, the earlier the age at initiation of smoking, the greater the individual risk."
Professor Idle said that he did not agree with this. He did not doubt the association, but his problem was that the cause had not been demonstrated experimentally. He did not know whether the association was more clearly demonstrated, the earlier the age at initiation of smoking. He said he was not competent to give an expert opinion on epidemiology.
[5.719] In the evaluations, on p.314, it was stated:
"The occurrence of malignant tumours of the respiratory tract and of the upper digestive tract is causally related to the smoking of different forms of tobacco (cigarettes, cigars, pipes, bidis [small, hand-rolled, Indian cigarettes])."
Professor Idle agreed that this group of scientists and doctors had concluded that the occurrence of lung cancer was causally related to smoking. He said that it was a public health report destined for member governments of the United Nations. It was prepared by scientists of high standing in their particular fields. He did not know whether it was the case that this conclusion had never been challenged.
[5.720] In the chapter relating to epidemiological studies of cancer in humans, the section relating to cancer of the lung and the sub-section relating to factors affecting risk, under the subheading "Duration of smoking" the following statement appeared, at p.206:
"One of the key features of the relationship between cigarette smoking and lung cancer is the relevance of duration of regular cigarette smoking to lung cancer onset rates. For example, using a statistical model fitted to data from the British male doctors study, Doll and Peto (1978) estimated that the excess annual incidence rates of lung cancer after about 45, 30 and 15 years of cigarette smoking were in the approximate ratio 100:20:1 to each other [...]."
Professor Idle said that this ratio showed an association which was greater for 45 year exposure, less for 30 and less for 15, in the ratio of 100:20:1. He did not agree that if it was a very long period of smoking that increased the risk of lung cancer, then that sort of length of time could not be reproduced in animal experiments. He said that the data of Doll and Peto showed, so far as he understood them, that smoking for somewhere around half a human lifetime - 45 years - produced these figures. So, in animal experiments, the investigators sought to put the animals on a lifetime exposure. He accepted that the lifetime of animals was very short in terms of a human lifetime. He agreed that experimental animals did not live as long as humans did. The life expectancy of mice and rats was four or five years at most.
[5.721] In the same chapter, at pp.199-203, there was a summary of cohort (prospective) studies on smoking and cancer. Professor Idle said that he was aware of the existence of some of them. Mr McEachran put in figures of the numbers of subjects involved in some of these studies: in the American Cancer Society Nine-State Study, reported by Hammond and Horn in 1958, about 188,000 men were involved; in the Canadian Study, reported by Best and others in 1961 and Lossing and others in 1966, 92,000 war veterans; in the British Doctors Study, reported by Doll and Hill in 1964, Doll and Peto in 1976 and Doll and others in 1980, 3,440 male and 6,194 female British doctors who had responded to a questionnaire were followed up; and in the American Cancer Society 25-State Study, reported by Hammond in 1966 and Hammond and Seidman in 1980, more than 1 million subjects were enrolled. Professor Idle agreed that these were large groups of people. He did not agree that it would be fair to describe these studies as "the same sort of thing as laboratory studies done on animals". Counsel put to him that these were much larger populations being looked at in a prospective way and asked whether one was not going to get extremely accurate information from this. Professor Idle said this was not so, because the investigators usually did not meet any of the subjects. They trained people such as relatives or friends to fill in questionnaires and the data were all collected in an office somewhere and that was when the epidemiologists started, when they got the data to analyse; whereas, in the laboratory, the investigator was very close to what was going on and could investigate material by scientific method, which was not so for epidemiology, as he understood it. He had read that there were problems; he believed "confounding" was a problem in epidemiological studies, "but I am not competent to comment further on that."
[5.722] Returning to USSG 1964, counsel read out the first paragraph of chapter 1 on p.5, which was in these terms:
"Realizing that for the convenience of all types of serious readers it would be desirable to simplify language, condense chapters and bring opinions to the forefront, the Committee offers Part I as such a presentation. This Part includes: (a) an introduction comprising, among other items, a chronology especially pertinent to the subject of this study and to the establishment and activities of the Committee, (b) a short account of how the study was conducted, (c) the chief criteria used in making judgments, and (d) a brief overview of the entire Report."
Counsel next drew attention to p.6, where it was stated that since 1939 there had been twenty-nine retrospective studies of lung cancer, and on the same page the statement:
"During the decade 1950-1960, at various dates, statements based upon the accumulated evidence were issued by a number of organizations. These included the British Medical Research Council: the cancer societies of Denmark, Norway, Sweden, Finland, and the Netherlands: the American Cancer Society: the American Heart Association: the Joint Tuberculosis Council of Great Britain: and the Canadian National Department of Health and Welfare. The consensus, publicly declared, was that smoking is an important health hazard, particularly with respect to lung cancer [...]."
Professor Idle said that he was aware of these matters, as stated in the report.
[5.723] At pp.6-7 the report stated:
"The US Public Health Service first became officially engaged in an appraisal of the available data on smoking and health in June, 1956, when, under the instigation of the Surgeon General, a Scientific Study Group on the subject was established jointly by the National Cancer Institute, the National Heart Institute, the American Cancer Society, and the American Heart Association. After appraising 16 independent studies carried on in five countries over a period of 18 years, this group concluded that there is a causal relationship between excessive smoking of cigarettes and lung cancer."
Professor Idle said that he was not aware of this study group. Asked whether this would not impinge on his view that it was not established in the scientific community by 1957 that smoking was a cause of lung cancer, he said that he did not dispute the associations or the health warnings. He believed that it had not been scientifically established, then or now. He agreed that some bodies had concluded by 1957 that smoking caused lung cancer. Asked whether there were not a substantial number of bodies who had looked at a very large body of scientific evidence "to back this up", he said that there were scientific opinions such as these, and the literature also contained opposing opinions that questioned the interpretation of the associations.
[5.724] Asked by me about the reasons for his reluctance to accept the conclusions of the epidemiologists, Professor Idle said that his problem lay, for example, in a statement in IARC 1986, that this was "the simplest interpretation of these data". For him, unfortunately, not everything was that simple in medicine and in science. He agreed that the interpretation of the data, that there was causation, was plausible and might be right, but it also might not be right. For him the scientific method dictated that certain other types of experiment and hypothesis must be formed and answered. The epidemiology, in other words, was a pointer: it pointed towards the laboratory and said "Look, here is the hypothesis, now go and test it". "It does not matter for me how much epidemiology you have, it is still just a pointer to scientific investigation. That is my difficulty." Even if an absolute association were found, where all subjects who were exposed to a particular substance contracted a particular condition, he would describe that as a very strong indication that required relatively little experimental confirmation, and would be confirmed very rapidly by laboratory experiment. He said:
"It is only because of the nature of epidemiology, that one is dealing with statistics and how the numbers are collected by questionnaire and so on. There is obviously scope for error within the methodology, unlike an experiment with enzymes, animal cells or whatever, where the experimenter has it under very carefully controlled conditions, with control investigations to ensure that what he or she sees is real. I think one must always be adopting the Popperian view to try to disprove one's own hypothesis, and I find epidemiology is often - not always, but often - lacking in that approach: that you get a result and then it is interpreted 'in the simplest way possible', for example, to quote the IARC report, rather than to go to the laboratory and see if it is real or not. I am sorry for a complicated answer [.......], but this is a complex subject."
[5.725] Counsel returned to p.7 of USSG 1964:
"Impressed by the report of the Study Committee and by other new evidence, [the Surgeon General] issued a statement on July 12, 1957, reviewing the matter and declaring that: 'The Public Health Service feels the weight of the evidence is increasingly pointing in one direction: that excessive smoking is one of the causative factors in lung cancer.'"
Counsel then drew attention to a passage relating to the establishment of the Surgeon General's Advisory Committee on Smoking and Health in 1962. Professor Idle agreed that it was appropriate to form an advisory committee composed of "outstanding experts who would assess available knowledge in this area [smoking vs. health] and make appropriate recommendations". At p.8 the report referred to a meeting held in July 1962 between the Surgeon General and representatives of various organisations, including the American Cancer Society, the American College of Chest Physicians, the American Heart Association, the American Medical Association and the Tobacco Institute, Inc. The participants in this meeting compiled a list of scientists and physicians working in the fields of biology and medicine, and during the next month this was screened by the representatives of organisations present at the meeting in July. "Any organization could veto any of the names on the list, no reasons being required." Professor Idle said he agreed that this was an appropriate way in which to proceed, but he was not able to confirm that the Tobacco Institute Inc. represented the tobacco industry in the US, so that the tobacco industry were able to exercise a veto.
[5.726] The Committee was eventually composed of ten members and a chairman. Counsel asked Professor Idle about these: he had already indicated knowledge of Dr Emanuel Farber, and he recognised also the name of Dr Louis F Fieser, Professor of Organic Chemistry at Harvard University, who was appointed for his interests and competence in the field of carcinogenic hydrocarbons. Professor Idle had studied his book as an undergraduate, and said that he was a leader in his field. He agreed that there were "a variety of fairly outstanding people on this committee, on the face of it".
[5.727] Counsel next referred to a paragraph in chapter 2, p.13, relating to the conduct of the study, in which it was stated:
"With an enormous amount of assistance from 155 consultants, from members and associates of the supporting staff, and from several organizations and institutions, the Committee feels that a document of adequate scope, integrity, and individuality has been produced."
Professor Idle agreed that on the face of it the preparation of the report was a "huge operation". He also agreed that, as stated on p.14, it was appropriate for the Committee to take steps to review the relevant literature.
[5.728] Counsel drew attention to passages in chapter 3, entitled "Criteria for judgment". It was stated at p.19 that the Committee made decisions or judgments at three levels:
"I. Judgment as to the validity of a publication or report. Entering into the making of this judgment were such elements as estimates of the competence and training of the investigator, the degree of freedom from bias, design and scope of the investigation, adequacy of facilities and resources, adequacy of controls.
II. Judgment as to the validity of the interpretations placed by investigators upon their observations and data, and as to the logic and justification of their conclusions.
III. Judgments necessary for the formulation of conclusions within the Committee."
Professor Idle agreed that the making of such judgments was appropriate.
[5.729] In a passage discussing "Criteria of the epidemiologic method", on p.20, the Committee stated:
"Statistical methods cannot establish proof of a causal relationship in an association. The causal significance of an association is a matter of judgment which goes beyond any statement of statistical probability. To judge or evaluate the causal significance of the association between the attribute or agent and the disease, or effect upon health, a number of criteria must be utilized, no one of which is an all-sufficient basis for judgment."
These criteria included the consistency, the strength, the specificity, the temporal relationship and the coherence of the association. Professor Idle noted that these criteria were applied to the epidemiology.
[5.730] Counsel asked Professor Idle to note the discussion of "causality" on pp.20 and 21, including the statement at p.21:
"The word cause is the one in general usage in connection with matters considered in this study, and it is capable of conveying the notion of a significant, effectual, relationship between an agent and an associated disorder or disease in the host."
Professor Idle said that the Committee were defining their terms.
[5.731] Counsel next went to chapter 4, "Summaries and conclusions", under the heading "Associations and causality" on pp.30-31. The Committee stated:
"The array of information from the prospective and retrospective studies of smokers and non-smokers clearly establishes an association between cigarette smoking and substantially higher death rates."
Professor Idle agreed with this. The Committee went on:
"In this inquiry the epidemiologic method was used extensively in the assessment of causal factors in the relationship of smoking to health among human beings upon whom direct experimentation could not be imposed. Clinical, pathological, and experimental evidence was thoroughly considered and often served to suggest an hypothesis or confirm or contradict other findings. When coupled with the other data, results from the epidemiologic studies can provide the basis upon which judgments of causality may be made.
It is recognised that no simple cause-and-effect relationship is likely to exist between a complex product like tobacco smoke and a specific disease in the variable human organism."
Professor Idle said that he agreed with this last statement. The Committee went on:
"It is also recognized that often the coexistence of several factors is required for the occurrence of a disease, and that one of the factors may play a determinant role; that is, without it, the other factors (such as genetic susceptibility) seldom lead to the occurrence of the disease."
Professor Idle said he agreed with this statement also.
[5.732] Counsel then referred to the passage headed "The effects of smoking: principal findings" and the subheading "Lung cancer" on p.31, under which it was stated:
"Cigarette smoking is causally related to lung cancer in men; the magnitude of the effect of cigarette smoking far outweighs all other factors."
The following exchange then took place:
"Q So that is the conclusion they reach. Do you see that?
A I do.
Q And have they not gone about it in an entirely appropriate way; getting a body of eminent men, reviewing all the scientific literature up to that time and reaching a conclusion on it?
A There are two statements which we have read in the last fifteen minutes which contradict each other. One says that epidemiology can never lead to cause, and then they go on to say that reaching a cause relied on epidemiology. That is where my problem lies. The statement above says 'In conjunction with other experimental data' or 'with other data' and that is what I agree with. So they have gone about it the right way - to answer your question - but the interpretation of their findings I cannot agree with."
[5.733] Professor Idle was asked about a passage in UKHC 2000, Vol.I, paras.49 and 50, relating to the research facility at Harrogate. Professor Idle said that he had reviewed a number of papers relating to the research carried on there, mainly skin painting studies carried out with tobacco smoke condensate. He did not know if the facilities still existed, and if it did not, why not. Counsel drew attention to a passage in para.50:
"BAT [British American Tobacco] emphasised in its analysis that the purpose of Harrogate was not to establish whether tobacco condensate could produce cancer in animal test models, a fact already 'amply reported' in the scientific literature, but to identify 'the chemical constituents of tobacco smoke primarily responsible for the mouse skin tumorigenicity and to investigate cigarette design modifications which might reduce the specific tumorigenicity'."
Professor Idle said that this was his understanding of the research which had been conducted.
[5.734] Counsel then asked Professor Idle questions about IARC 2004. Professor Idle said that he was not aware of the press release and the publication of the conclusions in 2002. He agreed that, as stated in it, a scientific working group of twenty-nine experts had concluded its evaluation of the carcinogenic risks associated with involuntary smoking, with second-hand smoke also classified as carcinogenic to humans. He did not agree with it. In the summary of data reported and evaluation, at para.5.2, reference was made to IARC 1986, in which various cancers were identified as caused by cigarette smoking. "Many more studies published since this earlier Monograph support these causal links." Professor Idle did not agree with this latter statement: many more studies had been published, but they still did not establish a causal link. He was asked about other passages relating to animal carcinogenicity data, and gene mutations and chromosomal abnormalities in humans. His position on this was that he had no basis to comment on the conclusions reached by the working group, because he had not had access to the main body of the Monograph and was not aware which papers were referred to in the summary. He said that this was particularly so in respect of the statement, under the heading "Animal carcinogenicity data":
"The most compelling evidence for a positive carcinogenic effect of tobacco smoke in animals is the reproducible increase observed in several studies in the occurrence of laryngeal carcinomas in hamsters exposed to whole tobacco smoke or to its particulate phase."
He said that without references to the studies, he had no basis to comment on this statement.
[5.735] Counsel then turned to statements made in a series of articles. In Bennett et al. 1999, the introduction started with the statement:
"Classical cancer epidemiology has successfully identified populations at high cancer risk, e.g. users of tobacco products including cigarettes, cigars, and chewing tobacco. Molecular epidemiology of human cancer risk has the challenging goal of identifying high-risk individuals within these populations."
Professor Idle agreed that the authors were stating that classical cancer epidemiology had successfully identified populations of high cancer risk. In Biroš et al. 2001a the authors stated, in the introduction:
"Cigarette smoking is the most important single factor in causing lung cancer and is directly responsible for at least 90% of newly diagnosed cases [...]."
Professor Idle agreed that these researchers accepted the association and the causal effect. In Biroš et al. 2001b the authors stated, at p.410:
"Cigarette smoking is the most important single factor in causing lung cancer. [...] From this work [their research], we can consider a certain role of the p53 codon 72 single nucleotide polymorphism in susceptibility to lung cancer."
Professor Idle agreed that they were carrying on the research which he had mentioned. Denissenko et al. 1996 stated in the abstract:
"[T]argeted adduct formation rather than a phenotypic selection appears to shape the p53 mutational spectrum in lung cancer. These results provide a direct etiological link between a defined chemical carcinogen [benzo[a]pyrene] and human cancer."
The article started:
"Lung cancer is currently the leading cause of cancer death in the United States and is also the most common type of tumor world-wide. Tobacco smoking is the single most important risk factor for lung cancer."
The study ended, at p.432:
"Our study thus provides a direct link between a defined cigarette smoke carcinogen and human cancer mutations."
Professor Idle agreed that these authors were saying that tobacco smoking was the single most important risk factor for lung cancer, so here again were people working in a related field who came to a different conclusion from his.
[5.736] Hainaut and Pfeifer 2001 stated in the abstract:
"It is unquestionable that the major cause of lung cancer is cigarette smoking. p53 mutations are common in lung cancers from smokers but less common in non-smokers."
In the introduction they stated:
"Today, lung cancer is the leading cause of cancer death in the US for both women and men. It is estimated that lung cancer kills over one million people each year worldwide. It has long been recognized that the major cause of lung cancer is cigarette smoking [...]."
Professor Idle agreed that these authors accepted that cigarette smoking was a major cause of lung cancer. Hashimoto et al. 2000 stated in the abstract that their results indicated that two major subtypes of lung adenocarcinoma existed, one probably caused by tobacco smoke, and the other possibly due to spontaneous mutations. In the main text, on p.2133, they stated:
"Lung cancer constitutes one of the leading causes of cancer death in the world, and its incidence is increasing in Japan. [...] Exogenous factors, especially tobacco smoke, are established causes of squamous cell and small cell carcinomas, but other, as yet unknown, endogenous factors may be more important for adenocarcinomas."
Professor Idle agreed that these authors accepted that tobacco smoke probably caused one of the subtypes of lung adenocarcinoma and that it had been established that tobacco smoke was a cause of squamous cell and small cell carcinomas.
[5.737] Krawczak and Cooper 1998, the latter of whom was known to Professor Idle, referred to the claim in Denissenko et al. 1996 to have provided a "direct link between a defined cigarette smoke carcinogen and human cancer mutations". They wrote:
"Whilst the causal relationship between cigarette smoking and lung cancer is beyond any reasonable doubt, we nevertheless wish to take issue with Denissenko and his colleagues over the validity of their conclusions. The fundamental requirement for an epidemiological study to be sound is that a given phenomenon (in this case the spatial coincidence of in vitro benzo[a]pyrene-induced adduct formation with in vivo single base pair substitutions in the TP53 gene) which is claimed to apply to an exposed group (i.e. smoking lung cancer patients) is shown to be absent from the non-exposed group (i.e. non-smoking lung cancer patients)."
Professor Idle agreed that these were two further researchers working in his field who accepted that the causal relationship between cigarette smoking and lung cancer had been established. Pfeifer et al. 2002 stated in the introduction:
"Cigarette smoking causes 30% of all cancer deaths in developed countries [...]. In addition to lung cancer, cigarette smoking is an important cause of [other cancers]."
Professor Idle agreed that again these researchers accepted that cigarette smoking caused lung cancer.
[5.738] Rodin and Rodin 2000 stated in the abstract that it was "an almost consensus opinion that the major carcinogenic risk of tobacco smoke is in its direct mutagenic action on DNA of cancer-related genes." Professor Idle said that this was the current paradigm. The authors stated that results of their analysis of p53 mutations compiled from the IARC p53 database and from the literature pointed to a different causative link. Professor Idle said that this was "a terribly complex paper". Various different hypotheses were being mounted. Denissenko, Pfeifer and Rodin were all based in the same department at the Beckman Research Institute in Duarte, California. This was scientific debate at its most interesting, with academics at the same department debating different hypotheses in the public domain literature, a situation that was still not resolved. In summary, the authors concluded that physiological stresses (not necessarily genotoxic) aggravated by smoking were the leading risk factor in the p53-association etiology of lung cancer. Finally, counsel referred to Toyooka et al. 2003, which started with the statement:
"Lung cancer is the leading cause of cancer deaths in the world with over one million cases diagnosed every year [...], and the vast majority of cases are smoking related."
After reading this out, counsel asked whether Professor Idle saw that, to which the witness replied that he did. In concluding that cigarette smoking had not been established as a cause of lung cancer, Professor Idle agreed that his position was that he did not know.
Re-examination of Professor Jeffrey Idle
[5.739] In re-examination, Professor Idle said that the statement in MRC 1957, that the finding of carcinogenic agents in tobacco smoke was an important step forward, in that it provided a rational basis for the hypothesis of causation, might be contrasted with the position of Wynder, who had said that the importance of laboratory work was not to prove that smoking was a cause of cancer in man. EL Wynder, President, American Health Foundation, was a member of the IARC working group which prepared IARC 1986. To the extent that this report relied on the evidence from animal skin painting, in reaching a view on causation, they were at odds with Wynder, one of their participants. The working group were using a different approach to the evaluation of carcinogenicity in humans from the approach that they used in its evaluation in animals. In humans, they relied upon the epidemiology, the results of which were said to imply causality, and on nothing else. On Wynder's approach, the demonstration of the carcinogenicity of cigarette smoke condensate on mouse skin provided no evidence for determining whether or not cigarette smoking caused lung cancer in humans. The statement that there was sufficient evidence that inhalation of tobacco smoke as well as topical application of tobacco smoke condensate caused cancer in experimental animals came from studies of the inhalation of cigarette smoke by animals, particularly hamsters, mice and rats. The statement that there was sufficient evidence that tobacco smoke was carcinogenic to humans was based on epidemiology, as were similar statements that the occurrence of malignant tumours of the respiratory tract and other malignant tumours were causally related to tobacco smoking.
[5.740] In cases where there had been a specific association between a disease and a putative causative exposure, such as mesothelioma and exposure to asbestos, and angiocarcinoma of the liver and exposure to vinyl chloride monomer, the causal relationship had been established by experimentation on laboratory animals. If there were a highly specific association, one would expect to be able to establish it in the laboratory. Epidemiology, he said, pointed to experiments that should be carried out under carefully controlled conditions in a laboratory setting and, where this had been done, the hypothesis that smoking caused lung cancer in humans had failed.
[5.741] The statement in USSG 1964 that after appraising sixteen independent studies carried on in five countries over a period of eighteen years, the group concluded that there was a causal relationship between excessive smoking of cigarettes and lung cancer, was based on epidemiological studies.
[5.742] Professor Idle was asked about the suggestion that had been made, in connection with animal inhalation studies, that animals could not be compared with humans because of their much shorter life spans. He said that he was sure that researchers were aware of this and took account of it in their study designs. As previously stated by him, Laskin and colleagues exposed rats to the irritant gas sulphur dioxide mixed with the polycyclic aromatic hydrocarbon benzo[a]pyrene and reported that some rats developed squamous cell lung tumours. This report established that the rat was a species susceptible to develop squamous cell lung carcinomas. He thought that the investigators were interested in the two substances that they were applying, per se, but it could be interpreted from their paper that this operated as a form of control. In the study carried out under contract by Microbiological Associates for the Council for Tobacco Research - USA, Inc., and reported in 1984 and 1986, 20-methylcholanthrene was administered to mice as a positive control and produced squamous cell carcinoma of the lung. This led to the conclusion that if a carcinogen, something which could truly cause lung cancers, was applied to a mouse in a life time study, then it should be possible to see squamous cell carcinoma in these animals, upon the basis of the positive control experiments.
[5.743] Under reference to the most recent papers, Professor Idle said that since 1996 a controversy had developed about the role of p53 mutations, lung cancer and smoking. Denissenko et al. 1996 proposed that the pattern of p53 mutations observed in lung cancers could be reproduced in the laboratory by the incubation of normal human bronchial epithelial cells with the metabolite of benzo[a]pyrene, BPDE. They concluded that they had now established a causal link between a carcinogen and cigarette smoke and the occurrence of lung cancer. This paper generated a significant controversy and an ongoing debate in the scientific literature, which to the best of Professor Idle's knowledge was still continuing. Krawczak and Cooper 1998 wrote that Denissenko et al. and Denissenko et al. 1997 had not only failed to demonstrate that the claimed mutagenic action of benzo[a]pyrene was confined to smokers, but their data were also insufficient in general to prove that TP53 mutations associated with lung cancer were anything other than predominantly endogenous in origin. Rodin and Rodin 2000 were on the same side of the argument as Krawczak and Cooper. They proposed that oxidative stress or some other process might cause particular mutations at certain sites in the p53 gene that were observed with Denissenko et al., and these gave the cells in which they occurred a selective advantage. Rubin wrote in 2001 that there was currently sharp disagreement about the relative roles in human pulmonary carcinogenesis of the direct induction of mutations by polycyclic aromatic hydrocarbons in tobacco smoke versus the selection of endogenous mutations. The disagreement was based upon differing interpretations of base changes in codons of the p53 gene of human lung cancers. In none of the recent papers put to him in cross-examination was any evidence offered in support of statements to the effect that the causal relationship between smoking and lung cancer had been established: the authors should be regarded as treating this as the current paradigm, in the context of which their research could be assessed.
[5.744] Finally, reference was made to the statement in IARC 2004 that urine from smokers was mutagenic. While it could not be told on what research this conclusion was based, Professor Idle had given evidence that a handful of papers had reported that smokers voided a urine that was more mutagenic in the Ames test that non-smokers' urine. This, however, was at variance with experimental observations in rats, in which exposure to injections of cigarette smoke condensate produced urine that was more mutagenic by a factor of two or three than background, but when the rats were exposed to whole cigarette smoke they did not produce an elevated incidence of mutagenic urine.
[5.745] Dr Cohen, BSc, PhD, C.Chem, FRSC, MBIRA was aged 66. He was an Associate and thereafter a Fellow of the Royal Society of Chemistry (formerly the Royal Institute of Chemistry) and a member of the British Institute of Regulatory Affairs and had been admitted to the Register of Toxicologists of the Institute of Biology/British Toxicology Society and the register of Eurotox Registered Toxicologists. He explained that the British Institute of Regulatory Affairs was set up to deal with regulatory matters relating to pharmaceutical development and EU regulatory requirements when drugs were presented to the authorities for marketing authorisations. From 1963 to 1968 he headed the Editorial Department of the British Industrial Biological Research Association and was the Assistant Editor of its two publications, one of them the first European journal on Toxicology. In 1968 he served as head of the Medical Information Department of Schering Chemicals Ltd in the United Kingdom. In 1969 he founded Toxicology Advisory Services, a consultancy of which he was, and remained, sole practitioner. It provided an independent consultancy service in toxicology to industry, research organisations, government departments and international agencies in the field of pharmaceuticals, medical devices, food chemicals, agrochemicals, cosmetics and environmental and industrial chemicals. During his career he had been actively involved in the preparation of submissions and expert reports to European Union regulatory authorities on pharmaceuticals and pesticides. He explained that when a new drug was submitted to the European Union regulatory authority three expert reports were required, one of which related to toxicology. His involvement was with the toxicology programme and required the assessment of the data. From 1991 to 2002 he was a Trustee and Director of Marie Curie Cancer Care and for much of that period was a member of its Cancer Research Committee. He also served as a Director of Phogen Ltd, a joint venture biotechnology company set up between Marie Curie Cancer Care and Cantab Pharmaceuticals plc. He was also a member of various other societies including the Royal Society of Medicine and the European Society of Regulatory Affairs. He had published about thirty research papers and reviews, had been joint editor of two books on irradiated foods and had written many unsigned articles on toxicology.
[5.746] He said that he had been asked to evaluate the carcinogenicity studies reported in the scientific literature, in order to determine whether exposure of laboratory animals to cigarette smoke by inhalation or intratracheal administration resulted in the development of squamous cell carcinoma of the lung. Only those animal studies involving inhalation or intratracheal exposure to cigarette smoke for prolonged periods, i.e. six months or more, were considered by him. The six-month threshold was chosen because tumours normally were not expected until later in life, normally a year or even longer, and many studies were of only a few weeks' duration, where acute effects would have been seen. These would be totally irrelevant for the assessment of carcinogenicity. So six months was chosen as a reasonable period of exposure, so as to eliminate many short-term studies which would not have been relevant to his assessment. He reviewed reports of fifty-seven cigarette smoke inhalation and intratracheal exposure studies conducted in the mouse, rat, hamster, guinea-pig, rabbit and dog.
[5.747] Dr Cohen said that toxicity testing in laboratory animals entailed exposure of groups of animals to the test material and study of the biological response elicited under the conditions of the experiment. Typically, high doses were used in a toxicity study in an attempt to product an adverse biological response. This applied to the whole field of toxicity testing on laboratory animals. A wide range of chemicals had been submitted to toxicity testing, including drugs, food additives, pesticides and industrial and environmental chemicals, that were subject to regulatory control in many countries. In addition, lifestyle factors such as diet, alcohol, smoking and ultraviolet radiation had been submitted to toxicity testing. So the animal inhalation experiments he was considering were not unique, they formed part of a range of lifestyle animal experiments.
[5.748] He explained that there were various matters, which he described as key considerations, that needed to be considered when a toxicity study was started. An appropriate species had to be selected; and the route of exposure, the duration of exposure and the levels of exposure were dependent on the nature of the material being tested and the reason for doing the test. Control groups were very important because findings could occur in the exposed animals that were not due to treatment but due to spontaneous effects that could be picked up in the control group. For any given animal species the strain, gender, age, genetic makeup and health condition were important factors in determining the toxicological response. The route of exposure selected in a toxicity study should be dependent upon the intended human use of the test material in question or the route of ultimate human exposure. The length of a toxicity test should reflect the likely duration of exposure in humans and could range from a single dose study to daily administration over the animals' lifespan. In order to maximise the detection of carcinogenic activity of a test material, long-term exposure was necessary in animals, e.g. eighteen to twenty-four months in mice and twenty-four to thirty months in rats.
[5.749] He was asked how, if the onset of cancer in man typically occurred after the age of 50 years, this could be modelled effectively in mice and rats. He said that essentially the relative lifespans across the species were compared. In the case of a rodent with a lifespan of two years or just over, that would be deemed to be sufficient time for a carcinogenic response to develop. In the case of the dog, exposure would need to be for periods of up to ten years, and even longer with monkeys. A tumour is an age-related phenomenon linked to the age of the particular species. So fifty to seventy years in man was equivalent to one and a half to two years in the rodent and seven to ten years in the dog. He said that it was known empirically that cancers could form within the lifespan of the species in question.
[5.750] Dr Cohen went on to say that toxicity studies should be designed so as to determine the toxic potential of a test material over a wide dose range. Thus several groups, typically three, were exposed to a low, mid and high dose of the test material, the latter being well in excess of the potential human exposure level. This increased the chance of a toxic effect occurring and enabled it to be determined at what dose the effect either could not or could be seen. This helped to determine safety levels in humans, and increased the chance of an effect being manifest in the study. Various biological markers, for example general health, body weight, survival, haematology, blood chemistry, urine analysis and pathology were determined in test groups and compared with those in appropriate control groups.
[5.751] It was essential to include at least one negative control group in a toxicity study. This was a group which should be identical to the exposed group, other than receiving the treatment that the exposed group received. This provided a background measure against which the result in the test group was compared for any given toxicological end-point. A positive control group might also be included in a toxicity study in order to demonstrate the susceptibility of the animal model to a particular effect of a known agent, thereby validating the animal model for detecting the potential of the test substance to show the same effect. For example, the group might be exposed to a known carcinogen.
[5.752] There were essentially three types of negative control group: untreated control (also known as cage or shelf control), vehicle control and sham control. In untreated control the animals were left in cages, fed in the normal way and not handled in any way relating to the exposed group. In the vehicle control group, if for example what was being tested was a topically applied drug in a vehicle, the vehicle control group was exposed to the vehicle to see whether it was having any effect. In a sham control group, in the case of inhalation studies, for example, where the animals were handled in a more rigorous and robust manner than when the material is administered in the diet, they were exposed to the same procedural arrangements except that they were not exposed to the inhaled material, but were handled in exactly the same way as the exposed group. Thus any potential effect of the stressful procedure which could contribute the effects seen in the test group would be identified.
[5.753] Dr Cohen said that cancer was synonymous with a malignant neoplasm, i.e. a tumour that grew progressively, invaded local tissues, spread to distant sites (metastasis) and eventually led to death. In contrast, benign tumours grew slowly, might stop growing or regress, did not infiltrate into local tissues and did not metastasise. Benign tumours were invariably not fatal. In an adequately conducted carcinogenicity study, a statistically-significant excess or earlier onset of a particular malignant tumour in a treated group, when compared with an appropriate control group, would identify the chemical as a carcinogen in the animal model used.
[5.754] Dr Cohen was asked to comment on a passage in IARC 1986. At p.309, under the heading "Conclusions" and the subheading "(b) Carcinogenicity in animals" the report stated:
"Cigarette smoke has been tested for carcinogenicity in experimental animals by inhalation and by topical application of condensate and in other ways. Exposure of hamsters and rats to whole smoke results in the induction of malignant respiratory-tract tumours. Cigarette smoke condensate induces skin cancers in mice and rabbits after application to the skin, and lung cancers in rats after intrapulmonary injection."
This conclusion was related to a passage of text starting at p.127 under the heading "Biological data relevant to the evaluation of carcinogenic risk to humans" and the subheadings "1. Carcinogenicity studies in animals" and "(a) Inhalation of tobacco smoke". The IARC working group considered studies on mice, rats, hamsters, rabbits and dogs. Dr Cohen said that he considered all these studies "and a lot more", including studies which had been published before the Committee carried out its work in 1985. He had looked at more than twice the number of papers discussed in IARC 1986. He made it clear that his report focused purely on the inhalation studies.
[5.755] Dr Cohen was asked to consider a series of hamster studies. In Dontenwill et al. 1973, at p.1791 the authors stated:
"The fact that epithelial proliferations called 'early invasive carcinoma' [...] did not infiltrate the cartilage may be explained by the behaviour of tumor growth and cartilage."
He said that at no place in the report did the authors describe having found carcinoma without the qualification "early invasive". In Dontenwill et al. 1977 proliferative laryngeal lesions were graded as in Dontenwill et al. 1973. Stage 6 leukoplakia, the most advanced lesion, represented "early invasive carcinoma", also described as "carcinoma in situ". (These expressions are taken from a translation of the original German). The authors did not report the existence of carcinoma without the qualification "early invasive".
[5.756] In Bernfeld et al. 1974 instances were reported of laryngeal papilloma and laryngeal "microinvasive" carcinoma in two strains of hamster. Compared with either control group, laryngeal papilloma was more prevalent in the exposed groups of both strains and the increase in laryngeal "microinvasive" carcinoma was more marked in one strain. Dr Cohen said that the data were not statistically analysed. No lung tumours were reported in the exposed group of either strain; and increased incidence of non-neoplastic histopathological lesions was seen in the lungs in the exposed group of both strains. The authors did not report having found laryngeal carcinoma unqualified by the use of the word "microinvasive". In Bernfeld et al. 1979 the reported incidences of laryngeal papillomas and carcinomas (described as early carcinoma/carcinoma in situ or invasive carcinoma without distant metastasis) were set out in a table. Increased incidences in laryngeal papillomas and carcinomas were reported in the exposed groups but differences in tumour incidence between the control and exposed groups were not analysed statistically. Tumour incidence was not increased in the lungs or significantly increased in the trachea; increases in non-neoplastic histopathological changes were reported in the lungs and trachea of exposed animals.
[5.757] Dr Cohen summarised his discussion of the four hamster inhalation studies of cigarette smoke referred to above by saying that they reported an increase in laryngeal "early invasive" or "microinvasive" carcinoma (without distant metastasis). In his view, the production of laryngeal "early invasive" or "microinvasive" carcinoma in the hamster appeared to be a species-specific phenomenon; laryngeal tumours were not seen in the various mouse or rat studies, and the only species that showed laryngeal tumours was the hamster. Bernfeld also showed that one strain of hamster was more susceptible than the other in developing laryngeal tumours.
[5.758] In his evaluation of carcinogenicity studies of inhaled or intratracheally administered cigarette smoke in laboratory animals, Dr Cohen reported:
"None of the 57 inhalation and intratracheal carcinogenicity studies in the mouse, rat, hamster, guinea-pig, rabbit or dog, as reviewed in this report, showed a statistically-significant increase in squamous cell carcinoma of the lung following exposure to cigarette smoke and indeed none of the authors of these studies reported a statistically-significant increase for this carcinoma."
[5.759] Dr Cohen then turned to rat studies. IARC 1986 stated at p.194: "In one study involving long-term exposure of rats to cigarette smoke, tumours of the respiratory tract were induced." This was a reference to Dalbey et al. 1980. In this study the authors stated that their choice of a particular strain of rat was heavily influenced by the induction of squamous carcinomas in the respiratory tract after intratracheal instillations of relatively small amounts of polycyclic hydrocarbons or after bronchial implantation of pellets containing cigarette smoke condensate. They stated at p.387:
"We observed 10 respiratory tumors in 7 smoke-exposed rats. Nasal tumors occurred as 1 early adenocarcinoma and one squamous cell carcinoma. The pulmonary tumors were 5 adenomas, 2 alveologenic carcinomas and one squamous carcinoma [...]. One alveologenic carcinoma was observed in controls. [...]
Besides the description of laryngeal neoplasms in smoke-exposed hamsters [...], the present work is the only study in which an unequivocal tumor response in the respiratory tract resulted from long-term tobacco smoke exposure. The tumors in the respiratory tracts of smoke-exposed rats consisted of 5 adenomas and 5 adenocarcinomas, alveologenic carcinomas or squamous carcinomas, as compared to one alveologenic carcinoma in the control animals."
In Table 2 the authors stated that 9% of smoke exposed rats, i.e. seven individuals out of the original group of eighty, had tumours in the respiratory tract. Dr Cohen said that he believed that the authors should not have included certain tumours in that grouping, and should have shown the benign tumours and the malignant tumours in the lung not only separately but to indicate whether the animals that showed the adenomas also had adenocarcinomas. It was not known whether the lung adenomas and lung adenocarcinomas appeared in seven, six or five animals, from the data presented in the paper. The two nasal tumours should not have been included because the practice of the National Toxicology Program (NTP) guidelines at the time recommended that nasal tumours should not be lumped together with lung tumours in tumour analysis.
[5.760] The NTP guidelines were contained in McConnell et al. 1986. At p.284 the authors stated:
"The purpose of this report is to give the rationale and criteria for combining certain neoplasms and to provide guidelines for combining neoplasms and sites."
The guidelines were based mainly on lesions occurring in the F344 inbred rat and the B6C3F1 inbred mouse, both of which were widely used in carcinogenesis studies. At p.285 the authors stated:
"Table 1 contains a list of organs and tissues where combining neoplasms is or is not appropriate to obtain a better understanding of the evidence of carcinogenicity. This list comprises those organs and tissues in which neoplasia is most often observed in F344 rats and B6C3F1 mice and may or may not be appropriate for use in other strains or species."
On p.285, Table 1 set out various organ systems and neoplasms. It stated that, in respect of the nasal cavity, squamous cell neoplasms should not be combined and that, in the case of the lung, squamous cell neoplasms and bronchioalveolar neoplasms should not be combined. Dr Cohen said that the mixing of tumours of different cell types in Dalbey et al. 1980 would have been inconsistent with these guidelines, had they been published earlier.
[5.761] In IARC 1980, which set out the basic requirements for long-term assays for carcinogenicity, it was stated at p.68, in a passage relating to the analysis and reporting of categories of tumour:
"Since it appears that cancers arise independently in various parts of the body, it has become customary to treat each potential target site (brain, kidney, bladder, etc.) as a separate experiment for evaluation."
Dalbey et al. 1980 (which was submitted for publication in 1979) did not act in conformity with this custom in not treating nasal tumours and lung tumours separately. Not only did the authors mix cell types, they also mixed sites in order to get their final total. Dr Cohen also said that IARC 1980 stated at p.69 that where judgments about the grouping of lesions for statistical analysis were difficult, it was desirable to evaluate them in both ways, divided and combined. If each was statistically significantly different from its counterpart in control animals, interpretation was easier. For example, lung adenomas and lung adenocarcinomas should be analysed separately as well as combined. Dalbey et al. 1980 did not do that even though the NTP guidelines stated that for a ten-year period before their publication their approach had been widely accepted and had undergone peer review.
[5.762] Dr Cohen also criticised Dalbey et al. 1980 for not having carried out separate analyses of the effects of treatment on each separate type of tumour, contrary to the advice given in IARC 1980 at p.321 in relation to statistical tests for carcinogenicity. They had combined the control groups in order to achieve their result. He believed that in their study the comparisons should have been with the sham control group. This was closer to the exposed groups than the untreated control group, and the survival rate in the untreated control group dropped considerably, so that there were relatively few animals available in it to develop spontaneous tumours. Therefore this group was not truly representative and was not a true comparison with the exposed group in the latter three to four months of the study.
[5.763] Asked whether if the NTP guidelines had been followed this would have made a difference to the outcome of the study of Dalbey et al., in particular the finding of statistical significance of the existence of tumours in the respiratory tract, Dr Cohen said that if the two nasal tumours and the lung squamous cell carcinoma had been removed, on the basis of the guidelines, then three of the ten tumours would have been set aside, leaving seven tumours remaining, i.e. five adenomas in the lungs and two alveologenic carcinomas in the lungs. But it was not stated whether those seven tumours appeared in seven animals, six animals or five animals: it was possible that one animal had both an adenoma and an adenocarcinoma. If there were seven tumour-bearing animals and they were compared with the combined controls, there would still be statistical significance. But if the seven animals out of eighty in the group were compared, not with the combined controls but with the sham controls, Dr Cohen did not believe that the difference would have been significant. If the advice in IARC 1980 had been followed by the authors of Dalbey et al. 1980, Dr Cohen said we would have known how many animals in the Dalbey study had lung adenomas and how many had lung adenocarcinomas, and whether there were seven, six or five tumour-bearing animals. This was crucial because without that knowledge it was difficult to assess the significant difference.
[5.764] Dr Cohen said that he would have expected the working group who prepared IARC 1986 to have in mind the advice on practices set out in IARC 1980 when they came to evaluate the animal inhalation studies. Nevertheless, and despite criticism of other studies for not following such advice, they reported on Dalbey et al. 1980 without further comment. In his view, on the evidence that was before the working group and discussed in the monograph, they were not entitled to express the conclusion that exposure of rats to whole smoke resulted in the induction of malignant respiratory tract tumours.
[5.765] Dr Cohen concluded his evidence-in-chief by stating that Wynder, either on his own or as a member of a group, had not produced lung adenocarcinomas in any laboratory animal prior to 1962.
Cross-examination of Dr Arnold Cohen
[5.766] In cross-examination Dr Cohen said that he was not a smoker. He was being paid for his work as an expert witness, at his normal rate. He did not know whether, as was now stated on cigarette packets, "Smoking kills", because he was not an expert in human epidemiology. He did not know that a very large number of doctors and scientists agreed that smoking kills. He was not aware of the statement in UKWP 1998 that 120,000 people a year in Britain died from illnesses directly related to smoking. He had not read any of the scientific papers dealing with the human epidemiology. "I am not an expert in this field."
[5.767] Mr McEachran made it clear that he was not challenging the conclusion in Dr Cohen's report which stated that in his opinion "laboratory studies using whole cigarette smoke have not produced squamous cell carcinoma of the lung in experimental animals". He went on to ask Dr Cohen about IARC 1986. Dr Cohen said that he was probably aware of it in 1986. He said that IARC were a reputable organisation and were a body that disseminated information and evaluated chemicals for carcinogenicity. Their reports were put together by in-house people at IARC and also a committee that assessed the data. He agreed that the committee was made up of "top people". He was taken through the names of the members of the committee. He recognised three of the names. One of these was Sir Richard Doll, whom he described as "a major figurehead in world science". Another was Sir Richard Peto, whom he described as "a leading figurehead in epidemiology". He was taken through the description of the Committee's working methods, with particular reference to the identification of articles for evaluation and their subsequent valuation. He agreed that this was an appropriate way to approach and prepare a monograph. He explained that it was necessary to rely on animal studies of carcinogenicity because in a long-term study humans would not deliberately be exposed to the test material. He agreed that an extensive amount of work had been put into the monograph and that the Committee had "looked at a lot of research material". The references extended to sixty pages.
[5.768] Asked about his criticisms of Dalbey et al. 1980, Dr Cohen said that if only five instead of seven animals developed tumours, that would not be statistically significant even compared with combined controls. He had consulted a statistician and five out of eighty compared with one out of ninety-three would not be statistically significant at the probability level of less than 0.05 (P<0.05). In carcinogenicity studies the importance should be on the number of animals showing the tumour, rather than the number of tumours, because some animals could have multiple tumours. This was the practice accepted by IARC.
[5.769] Mr McEachran asked Dr Cohen questions about the epidemiological studies of cancer in humans considered in IARC 1986 at pp.199 to 270. He said that he knew "how epidemiology works" in broad terms. His attention was drawn to the very large numbers of people included in these studies. Counsel invited comparison with the relatively small numbers of animals used in the laboratory studies. Dr Cohen said that in the case of an animal study which was controlled one had a much better picture of what was going on. People were "exposed to all sorts of things out there", whereas an animal study was focused on a particular agent which was appropriately controlled, so when the result was obtained, it was the direct effect of that agent. Numbers were not necessarily the key to it. He did not know that less than 1% of smokers developed lung cancer in any year. Asked whether, if that was so, replication of that effect on animals would require a large number of animals, he said that carcinogenicity studies had been carried out on all sorts of chemicals and it was a standard procedure. The dose in the animals could be increased.
[5.770] In response to questions by Mr McEachran about the treatment in IARC 1986 of the relevance of duration of regular cigarette smoking to lung cancer onset rates in humans, Dr Cohen said that this was outside his area and that it was beyond his expertise to look at a statistical model. He could not comment, because he was not an expert in the field, on a finding that lung cancer incidence increased with the duration of cigarette smoking so that it was considerably higher after forty years than after fifteen. It was put to him that laboratory animals were not tested for such long periods. He said, however, that the majority of chemicals that had been shown to develop tumours in animals had been reported to show an increased risk in humans, and he could not think of any examples where the human data were in conflict with the animal data. The number of years was not relevant, it was lifespan that determined the outcome of a carcinogenic event. So two years in a rodent was the equivalent of fifty, sixty or seventy years in a human. It was in accordance with the carcinogenicity guidelines of most countries to use animals, and all the published guidelines explained that it was the lifespan of the animal in a species that mattered in determining carcinogenic potential. He did not agree that the reason why squamous cell carcinomas had not been induced in animal inhalation studies was that the exposure, albeit for most of the lifetime of the animals, was too short. Known carcinogens could induce cancers in experimental animals in a relatively short time. In one study, squamous cell carcinomas were induced in mice in twelve months with the use of 3-methylcholanthrene, a known carcinogen, in order to demonstrate whether that particular strain was capable of developing squamous cell carcinomas within the lifespan of the animals. Other compounds had produced squamous cell carcinomas in various species. Dalbey et al. 1980 mentioned in the introduction that the F344 rat was selected because it was capable of expressing and developing squamous cell carcinoma.
[5.771] Dr Cohen was asked about statements in IARC 2004 at para.5.3:
"In four out of five studies in rats, exposure to whole smoke led to modest increases in the occurrence of malignant and/or benign lung tumours. Similarly, in four of eight studies in mice with varying susceptibility to lung tumour development, exposure to whole smoke led to a modest increase in the frequency of lung adenomas. An increased incidence of lung 'tumours' has also been reported in dogs exposed to tobacco smoke [...]."
Dr Cohen said that he could not comment on these statements because he did not know which studies were referred to in the summary in IARC 2004 (which was all that had been published by then). He said that he understood "tumours" in this context to be doubtful neoplasia, which could be either tumours or pre-neoplastic lesions.
[5.772] Dr Lewis, aged 55, stated that he was born in Germany and studied biology in the USA from 1970 to 1974, gaining a Bachelor of Science degree. He then went on to study medicine in Germany and became a doctor of medicine, followed by epidemiology in the USA and Canada, gaining a diploma in epidemiology from McGill University. He gained his clinical experience in Ulm in Germany, in internal medicine, and earned a diploma in tropical medicine in Hamburg. He was Assistant Professor in the Department of Epidemiology and Biostatistics at McGill University in Montréal, Canada, from 1993 to 1996. From 1993 to 1995 he was Associate Director of the Potsdam Institute of Pharmacoepidemiology and Technology Assessment, leaving the Institute as its Scientific Director at the end of 1996. He explained that pharmacoepidemiology was a branch of epidemiology which dealt with associations between medicines and their outcomes and that technology assessment dealt with the infrastructure of health care within any given country. He was currently director of a commercial firm called EPES Epidemiology, Pharmacoepidemiology and Systems Research in Berlin, specialising in consultations and conduct of pharmacoepidemiology studies. He was adjunct Reader of Epidemiology and Biostatistics at the University of Potsdam. He explained that epidemiology dealt with populations and assessing means of designing studies and capturing populations for specific questions, whereas biostatistics dealt with the more technical issues of data analysis and analysing the proportions and the distributions one finds in the population.
[5.773] Dr Lewis had been involved in several multinational projects addressing
epidemiological and pharmacoepidemiological issues, including the WHO-MONICA project and the Transnational Study on Oral Contraceptives. He explained that the WHO-MONICA project was initiated by the World Health Organization and ran from about 1982 to about 1992 in forty countries. One component looked at cardiovascular risk factors within populations and their distribution, using a survey technique; the other looked at the annual incidence of myocardial infarction in each of these areas over time. The Transnational Study on Oral Contraceptives was a multinational study involving five centres in different countries and ran for four years to determine whether or not there was an association of venous thromboembolism with oral contraceptive use, dividing oral contraceptives into various types of components, and also myocardial infarction. He had published about 150 papers and reports in scientific journals and held the Certificate for Epidemiology of the German Epidemiological Association (DAE). He was a board member of the International Society for Pharmacoepidemiology from 1991 to 1998, and was its President from 1996 to 1997. He was on the editorial boards of the journals Pharmacoepidemiology and Drug Safety and Disease Management & Health Outcomes. From 1997 to 2002 he held a faculty appointment as Adjunct Professor at the Department of Preventive Medicine of the University of Kansas Medical Center, and became Visiting Professor at the European Institute of Health & Medical Services of the University of Surrey in 1999.
[5.774] In Article 3 of condescendence it is averred that in 1992 Mr McTear was diagnosed as suffering from inoperable lung cancer. "Said cancer was caused by his smoking." Dr Lewis was asked to report whether or not epidemiological data could
be used to draw conclusions about the cause of disease in any individual. It was his opinion that epidemiological data could not be used to draw conclusions about the cause of disease in any individual.
[5.775] Dr Lewis stated that epidemiology dealt with populations and aggregate data which combined the features of many individuals. It did not deal with the individuals themselves or their specific situations. The number of new cases of any given disease occurring within a population in a year, for example, might be one in 10,000 persons. This was called an incidence rate, which defined the risk of disease in a population. The word "risk" in this context meant a predictor of what occurred in a population in terms of any type of event. The incidence rate was the number of new events for any given period of time, usually one year, occurring within the base population, which needed to be well defined. The definition might be in terms of geographical area, age or sex, for example. The expression "incidence among exposed" related to the number of events occurring within a population having a specific attribute, for example hypertension or exposure to medication. The word "exposure" included endogenous as well as exogenous factors. The expression "incidence among unexposed" related to the number of new events occurring in a population of unexposed individuals in a year. The expression "prevalence of an exposure" was the proportion of individuals exposed to a risk factor in a defined population at a given time. These expressions were used by Dr Lewis in Table 1 in his report, entitled "Incidence and prevalence in epidemiologic studies". Table 1 was as follows:
|
Incidence the number or rate of new events (numerator) occurring in a defined population (denominator) in a given time (i.e. a year)
|
I p = |
new events general population |
||
|
Incidence among exposed the rate of new events occurring in a population of exposed individuals in a year |
I e = |
new events exposed population |
a a + b |
|
|
Incidence among unexposed the number of new events occurring in a population of unexposed individuals in a year |
I u = |
new events unexposed population |
c c + d |
|
|
Prevalence of an exposure The proportion of individuals exposed to a risk factor in a defined population at a given time |
Pe = |
exposed individuals general population |
||
He said that the incidence rate was an aggregate figure from millions of different individuals and told us little about the individual risk of contracting the disease in question. For example, if the incidence rate of the disease within a population in one year was one in 10,000 persons, on average each individual's risk of contracting the disease was 0.0001. Disease occurrence might be influenced by risk factors within populations. These were factors which could be consistently observed in studies to be associated with a specific health outcome. For example, hypertension was a risk factor for heart disease. Risk factors could be either endogenous, i.e. genetic, or exogenous, i.e. related to lifestyle, working environment, etc.
[5.776] Incidences of disease, Dr Lewis continued, differed within populations depending, among other things, upon age, gender, genetics, diet and other exposures. For heart disease occurrence, one could show the incidence of disease among those who had hypertension and compare it with the incidence of those who did not. If hypothetically the incidences were 100 per 100,000 among hypertensives (the exposed) and ten per 100,000 among people with normal blood pressure (the unexposed), the risk of hypertensives to incur heart disease would be ten times higher than that of non-hypertensives. This relationship was calculated by dividing the incidence rate among exposed by the incidence rate among unexposed (i.e. (100/100,000) / (10/100,000)) to show the rate ratio (ratio of two rates), which was also known as the relative risk. Dr Lewis referred to Table 2 in his report, entitled "Relative risk and risk difference". The table was in these terms:
|
Relative Risk The ratio of the incidence among the exposed and the incidence among the unexposed population. Its equivalent from a case-control study is called an odds ratio (OR). If the RR=1, there is no difference in risk, if RR>1, the exposure is associated with risk, if RR<1, the exposure is associated with benefit. |
RR = |
I e = I u |
a a + b |
|
|
c c + d |
||||
|
Risk Difference or Excess Risk |
RD = |
I e - I u |
||
The expression "odds ratio" was used in a case-control study, "relative risk" was used in a cohort context, in which a population was followed over time until the study was stopped and then the outcomes or events within the categories of exposed and unexposed individuals were counted; a cohort study was also sometimes referred to as a prospective or longitudinal study. A case-control study was based on the notion that diseases were usually rather rare events, so it would be impracticable to follow a population large enough for a long enough time to be able to accumulate enough events to enable a judgment or calculation on the relative risk to be made. So the fallback was to do a case control study in which, for example, cases were collected from hospitals and compared with a control group that was comparable to these cases in every respect except that they did not have the disease in question. The expression "risk difference" or "excess risk" was the difference in the incidence among exposed and unexposed.
[5.777] Estimates of relative risks were usually obtained through studies in which the occurrence of disease in exposed and unexposed populations was observed (in cohort or case control studies). The individuals observed generally chose their exposure themselves. This was a consideration that epidemiologists had to bear in mind, particularly in dealing with an exogenous exposure, because the investigator did not know for what reason anyone chose to become exposed to a certain medication, or to a certain lifestyle or habit, for example. The question that arose was whether there was any attribute or difference within the exposed group that determined their inclination towards exposure as compared with the unexposed group. This might be contrasted with clinical trials, where the exposure was chosen for the individual by an investigator. In observational studies, it was uncertain that those who had selected themselves for the exposure were comparable to those who had decided not to be exposed. This had been found in studies relating to hormone replacement therapy, where different results were obtained depending on whether the women had selected their exposure themselves or it had been randomly allocated to them. Further, estimates of risk were population-based figures. The individuals within these populations might belong to different groups or risk sets. The expression "risk set" denoted a group of individuals who shared attributes or a set of attributes, which had an association with outcomes.
[5.778] Asked what risk factors were associated with lung cancer, Dr Lewis said that they were smoking, cigarette smoking, socio-economic status, alcohol intake or abuse, age, gender, occupational factors, environmental factors, previous respiratory diseases and a variety of other associations. These included a family history of lung cancer, psychological factors, urban living, living in west central Scotland and poor diet. In his view, environmental tobacco smoke was not a risk factor for lung cancer.
[5.779] Dr Lewis then went on to discuss potential errors. He said that the information derived from epidemiological studies was based on population samples and was a statistical estimate. An incidence or a relative risk was a point estimate which was usually accompanied by some measure of variability, such as a standard deviation or a confidence interval. This standard deviation or confidence interval established the degree of random error. It indicated the likelihood of having obtained an incorrect result due to chance, i.e. "the confidence we can have in our results". Random error would occur in any situation in which multiple tests were done on an individual. For example, many laboratory tests on a single individual would show slightly different values for each sample, with the average value shown as a point estimate. Similarly, random error would occur in a population, in which the values of many individuals were combined.
[5.780] Dr Lewis then went on to discuss the topic of attributable fractions, also known as attributable risk percent. He said that although observational studies could not establish cause-and-effect relationships, epidemiologists might want to draw causal inferences from statistical associations observed in studies. An underlying premise of public health epidemiology was that disease incidences could be reduced by the removal or reduction of risk factors in the population. In order to estimate the quantity of disease removed from the population by the elimination of a risk factor, the concept of attributable fractions was developed. He was referred to Table 1.2 on p.17 of RCP 2000, entitled "Estimated number and percentage of deaths attributable to smoking by cause, UK 1997". He said that the figures shown under the heading "Deaths from disease estimated to be caused by smoking as % of all deaths from disease" appeared to be an example of the use of an attributable fraction.
[5.781] Whereas the relative risk (or odds ratio) compared disease occurrence within two groups (e.g. exposed and unexposed), the attributable fraction (AF), which was mathematically derived from the relative risk, calculated the fraction of cases in the population which was attributed to the exposure. There were two types of attributable fractions, one based only on the exposed population, and the other on the general population. In the hypothetical example given above, if the incidence of heart disease among hypertensives was 100 per 100,000 and among non-hypertensives ten per 100,000, the risk difference was ninety per 100,000, which was divided by the incidence among the exposed to yield 0.9 or 90%. Reference was made to Table 4 in Dr Lewis's report, entitled "Population risk measures". The Table was as follows:
|
Attributable Fraction (or risk percent) among exposed The proportion of cases within the exposed group attributable to the exposure
|
AF (exposed) |
= |
I e - I u I e |
|
Attributable Fraction (or risk percent) in the population The proportion of cases within the population attributable to the exposure |
AF (population) |
= |
I p - I u I p |
|
Attributable Fraction (or risk percent) among exposed A restatement of the attributable fraction among exposed based on relative risk. This forms the basis of the formula used for probability of causation (PC) or assigned share (AS) calculations |
AF (exposed) |
= |
RR - 1 RR |
This meant that 90% of the cases among the exposed were attributed to the exposure. Because attributable risk percent in the population included the whole population, of which only a fraction was exposed, the attributable risk percent of the population was lower than the attributable risk percent in the exposed population. On the hypothesis that 40% was the prevalence of hypertension in the general population, this implied that 78% of all heart disease occurring in the general population would not occur if hypertension were removed. Similar calculations for the attributable fraction of exposed could be made using relative risks (RR) or odds ratios (OR) derived from case control studies. An RR of 10 yielded the same 90% attributable fraction among the exposed as was calculated from the incidences.
[5.782] Dr Lewis continued by stating that population attributable fractions were introduced into epidemiology by Levin 1953 in reference to lung cancer risk and smoking. His estimates for the percentage of all lung cancers attributable to smoking and under the assumption that smokers, had they not become smokers, would have had the same lung cancer risk as non-smokers, covered a wide range from 56.5% to 92.5%. Because the accuracy and validity of these calculations were contingent on many factors, attributable fractions were rarely reported in the epidemiology literature. They were, however, seen in the public health literature. Referred again to RCP 2000, p.17, Dr Lewis said that it was not possible to tell whether the attributable fraction, as used in Table 1.2, was the attributable fraction among the exposed or the attributable fraction among the general population: he was unable to decide what exactly the attributable fraction characteristic here referred to.
[5.783] Dr Lewis then went on to discuss what he called "problem areas". He said that the problems associated with attributable fractions included technical, computational, statistical and interpretational issues. First, the apparent simplicity of the calculation itself obscured the complexity of the underlying human biology and study populations, which in turn were designed to reflect the condition within the real populations. The estimates used in these calculations were derived from observational studies, all of which were open to confounding, bias and variability. Confounding was an attribute or a factor within a population or a group which could be associated with both the exposure and the outcome that the study was dealing with. So, for example, age was a general confounder which one would control for in the design of the study. If one was not aware of the existence of a possible confounder one spoke of residual confounding, meaning that one postulated elements that were attributes and confounders, such as, for example, genetic predisposition and so on, which might have an influence on the study estimate and which might be removed if they were known. Bias was a different type of error, and was essentially a problem in study conduct. It was a systematic error which was either based on problems in obtaining the correct information, or on exposure, for example, or on outcome. So there were diagnostic issues and issues of questioning an individual about elements of the exposure, called information bias. And there were problems in selecting the correct study populations when exposed and unexposed populations were compared. That would be selection bias. Information bias might arise through the wording of questions or through the inability of members of the study group to remember things clearly. Over-diagnosis of a particular disease in the population could give rise to problems with diagnostic bias; so would under-diagnosis. Variability was often known as random error, meaning that, with any given population, there was a certain spread of values, which amounted to an average, and, depending on how broad that spread was, determined the variability. Similarly, the sample size determined the variability that there would be in the estimate that was finally achieved. The outcome of a study was the average of the group, but no member of the group might coincide with the average.
[5.784] Relative risk estimates from clinical trials or case control studies were difficult to extrapolate to the general population, because they dealt only with specific subsets. Attributable fractions were sensitive to the definition of exposure and to the definition and the diagnosis of the medical condition under observation. For example, Levin demonstrated how exposure definitions affected risk estimates by showing that a minor redefinition of exposure in Doll and Hill 1950, which lowered the proportion of smokers among cases from 99.7% to 98.7%, reduced the risk estimate from 13.8 to 4.0.
[5.785] In Levin 1953 at pp.536-7 there was discussion of Table 5 on p.535, entitled "Lung cancer and smoking: Indicated incidence and proportion attributable to smoking in four studies." The author wrote:
"The studies of Doll and Hill and of Wynder and Graham would indicate a much greater excess incidence [of lung cancer] among smokers as compared with non-smokers. The comparative figures in Table 5 are presented chiefly to point out the different conclusions as to relative magnitude of the possible effect of smoking which may be reached from different types of data."
Three measures of the association of smoking with lung cancer were tabulated in the last three columns of the table. The first gave the ratio of incidence among smokers compared to that in the general or total population. The second was the ratio of lung cancer incidence among smokers to that among non-smokers. The third was the indicated maximum proportion of lung cancer attributable to smoking. Dr Lewis regarded the third index as being the equivalent of attributable fraction in the population. The first line of the table took Doll and Hill's figures of 99.7% of smokers among lung cancer cases and 96% among controls. The first ratio derived from these figures was 1.04, the second was 13.8 and the third was 92.5.
[5.786] At p.536 Levin stated:
"[I]n studies of the association of some environmental factor (such as smoking) with disease, the apparent extent of the association may be greatly magnified by relatively small differences between the control and test groups, when the general prevalence of the environmental factor is high. Thus, in Table 5, if the percentage of smokers in the lung cancer group in line (1) were 98.7 instead of 99.7, the indicated figure for (r) [the second index] would be 4.0 instead of 13.8. This figure is dependent, in part, on the definition of a smoker. Thus, if persons who never smoked more than 1-4 cigarettes were excluded from the category of smokers, the figures for line (1) would be those in line (6), and (r) would be only 2.8."
The table showed that line (6) used Doll and Hill's figures, excluding those who smoked 1-4 cigarettes per day. On that basis, the percentage of cigarette smokers among lung cancer cases was 96.0 and among controls was 89.5, the first index was 1.07, the second was 2.8 and the third was 61.7.
[5.787] At pp.536-537 Levin stated of the third index:
"This index is based on the assumption that smokers, if they had not become smokers, would have had the same incidence of lung cancer as that found among non-smokers. This assumes that other etiological factors are equal in the two groups (an assumption which can be further tested)."
Dr Lewis took the expression "etiological factors" to be risk factors in some way also associated with lung cancer. He accepted that the assumption that "other etiological factors are equal in the two groups" was a valid statement, but said that the assumption was not based on fact because the author had not shown that the groups were comparable. There were a number of risk factors for lung cancer, of which smoking was only one, so that if one were to remove smoking as an exposure in the population, the other risk factors would remain. A minor redefinition of exposure also had profound effects on attributable fractions.
[5.788] Dr Lewis next referred to Hsieh and Walter 1988, in the introduction to which the authors stated:
"Relative and attributable risk are common measures of association between the risk of a health outcome and exposure to a postulated risk factor. Relative risk is defined as the ratio of the rate of the disease in persons exposed to that in persons not exposed. Attributable risk is defined as the proportion of cases associated with (or attributed to) the exposure. Relative risk measures the strength of the effect of the exposure; it does not take into account the prevalence of the exposure in the population."
Dr Lewis stated that, as in his formula and the Levin formula, the prevalence had to be fed in so as to reach a figure for attributable risk. Hsieh and Walter continued:
"Attributable risk takes into account both the strength of the effect of exposure and the prevalence of the risk factor. Its use as a measure of association is useful for priority setting and policy making in public health."
Dr Lewis commented that in the taking of public health decisions, one would be guided towards dealing with a very prevalent exposure, where there was a higher attributable fraction, rather than trying to deal with an exposure which was comparatively uncommon. Hsieh and Walter continued:
"Attributable risk is also known as the etiologic fraction, attributable fraction and attributable proportion. Since attributable risk has also been referred to as the risk difference, hereafter we adopt the term attributable fraction in this presentation."
[5.789] The second problem area discussed by Dr Lewis was confounding, bias and variability. He stated that there was a vast amount of literature on the computational issues arising from confounding, bias and variability and their effects on the estimates of attributable fractions. Attempts to deal with these issues were only moderately successful, because epidemiological data were not precise. These data were based on a small sample of the population. This small sample was then used to reflect the entire population. Although the sample was composed of individuals who could have many different traits, and although all these traits were summarised in one estimate, the sample could not reflect all the differences within the population.
[5.790] Dr Lewis referred to Greenland 1988. The author of this chapter in a book to which there were a number of contributors was from the Division of Epidemiology at UCLA School of Public Health in Los Angeles, California. At p.95 the author stated:
"In his contribution to this volume, Weed has put forth a 'Popperian' evaluation of Hill's criteria for assessing causal inference. Elsewhere, he has eloquently described and prescribed a Popperian approach for the general problem of causal inference in epidemiology. In both contributions, he criticizes the notion that either causal inference or rational decision-making should be approached with methods founded on concepts of judgment, belief, opinion or subjective probability. [...] I will here argue for the hypothesis that the conflict between Popperian and more judgment-oriented epidemiologists such as Susser arises from the failure of both sides to distinguish the needs of the branch of science called epidemiology from the needs of the branch of public health called epidemiology. I will further argue that the Popperian methods proposed thus far fail to meet all the needs of either decision-making or inference, and so as yet require epidemiologists to fall back on other methods. In other words, though Popperian methods may be necessary for good scientific or decision analysis, none of the Popperian approaches described so far constitutes a sufficient system for such analyses."
At p.96 the author stated:
"Epidemiology as a science is not inherently concerned with anyone's opinion about how things are, but only how things are. It is thus understandable that Popperian epidemiologists wish to identify as scientific only statements about nature (and then only certain types), and reject from science individual beliefs or opinions about such statements. It is equally understandable that a sound methodology for practicing the science of epidemiology will be found inadequate for practicing the public-health profession of epidemiology: public health is not a science, but a form of social activism, one whose benefits appear profound enough to society that it is institutionalized and heavily subsidized by governments."
Dr Lewis said that he recognised the dichotomy between the branch of science called epidemiology and the branch of public health called epidemiology, and agreed with the view that there was a difference between these in the decision-making process. He agreed with the sentiment that public health was not a science but a form of social activism, and with the author's view that a public health activist would be concerned with communicating his or her own opinions, evaluating the opinions of colleagues and influencing the opinions of governmental figures and the public. He recognised the process, referred to by the author, whereby public health activists, on the strength of a belief, would then wish to transmit that belief to governmental figures in order to influence governmental action.
[5.791] Dr Lewis went on to state that further, one estimate could not reflect all the differences within the sample. It was a mathematical approximation which described the relationship between a disease and a risk factor. Epidemiological studies were typically short compared to the time it took some diseases to develop, particularly cancers. It was therefore not clear whether the attributable fraction reflected an excess risk due to the exposure, or whether the exposure simply moved a condition which would occur anyway forward in time. Finally, he stated, studies were performed on different populations with different methods and did not always produce the same result. It was therefore left to the discretion of the researcher to decide which study and which estimates to use in the calculation of an attributable fraction.
[5.792] The third of Dr Lewis's problem areas was that of interpretational issues. He said that the interpretational issues relating to attributable fractions within a population were causality and disease prevention. Attributable fraction calculations also touched on issues of individual attribution. The calculation of an attributable fraction assumed a cause-and-effect relationship between a risk factor and a disease. It could not be taken from an attributable fraction that it was evidence of a cause-and-effect relationship, because that notion was already embedded in the concept of attributable fraction. Although they were called "attributable", attributable fractions were derived from statistical associations (relative risk measures), which did not establish causation.
[5.793] In this context, Dr Lewis referred to Rockhill et al. 1998. In this commentary the authors considered computational and conceptual issues relevant to population attributable fraction estimation that were infrequently discussed elsewhere, with illustrations from the breast cancer literature. At p.15 they explained that they used the term "population attributable fraction" in the sense used by Greenland and Robins, who distinguished between "excess fraction" (what epidemiologists usually estimated when they computed "population attributable risk" or "population attributable fraction") and "etiologic fraction", which was not estimable without strong biologic assumptions. Dr Lewis explained that such an assumption was that observed associations had a cause-and-effect relationship.
[5.794] At p.15 the authors stated:
"The population attributable fraction is most commonly defined as the proportional reduction in average disease risk over a specified time interval that would be achieved by eliminating the exposure(s) of interest from the population while distributions of other risk factors in the population remain unchanged. This also can be interpreted as the proportion of disease cases over a specified time that would be prevented following elimination of the exposures, assuming the exposures are causal."
At p.16 the authors stated:
"Perhaps the most important aspect of population attributable fraction estimation is correct interpretation and communication."
At p.17, in a discussion of breast cancer risk factors, the authors stated:
"The population attributable fraction does not address probability of causation for a specific case of disease, nor does its estimation enable epidemiologists to discriminate between those cases caused by, and those not caused by, the risk factors under consideration."
In their conclusion, on p.18, the authors stated:
"Many public health researchers are interested in evaluating the potential population impacts of identified risk factors. For some of these evaluations, estimation of the population attributable fraction is appropriate and valuable. The assumptions underlying valid population attributable fraction estimation include the following: a causal relationship between the risk factors and disease; the immediate attainment, among those formerly exposed, of the unexposed disease risk following elimination of the exposures; and independence of the considered risk factors from other factors that influence disease risk so that it is possible to conceive of changing the population distributions of the considered factors only. Such assumptions are often not justified. Those who present population attributable fractions have a duty to ensure that estimates are correctly computed and that their limited meaning is correctly communicated, given the interest among researchers, clinicians, and the public in quantitative figures that attempt to summarize the state of etiologic knowledge about a disease."
Dr Lewis said that he agreed with all these statements.
[5.795] He went on to add that attributable fractions attempted to predict what would happen in the population if a risk factor were removed. That is, if the attributable fraction was 90% for a given risk factor and a disease, the model presumed that the removal of that risk factor would reduce the incidence of the disease by 90%. In many instances, however, as shown by studies exploring the effects of reducing blood pressure, cholesterol and cigarette smoking, the reduction in disease had been far lower than predicted by attributable risk percent calculations. The reduction of the risk factor had in general failed to reduce disease by the predicted amount.
[5.796] Dr Lewis then turned to the topic of individual causation. If the attributable fraction addressed the attributability of disease to a risk factor on a population level, what was the relationship, if any, between the attributable fraction and the individual case? Attributable fractions did not purport to deal with the individual case. Nor could they distinguish between those cases attributed to, and those not attributed to, the risk factors under consideration (Rockhill et al. 1998). However, attributable fractions had been used to develop the concept of "probability of causation", thereby seeking to move from the population level to the individual level. Probability of causation was now generally referred to as "assigned shares". One problem with assigned shares was exposure specificity. If only a single exposure was considered, high specificity must be assumed. This meant that a direct link was implied between a risk factor or an attribute and disease occurrence. Dr Lewis said, however, that most diseases were rare both in exposed and unexposed. Lung cancer, for example, was a rare disease in non-smokers, and somewhat less rare in smokers. An association for exposure to smoking, for example, which was observed to elevate annual risk from one in 10,000 in the unexposed to ten in 10,000 in the exposed, was not highly specific.
[5.797] Under reference to a diagram, Dr Lewis explained that any population that was being researched or observed in terms of a particular exposure and a particular outcome would fall into four groups: a group that was exposed and had the outcome; a group that was exposed and did not have the outcome; a group that was unexposed and had the outcome; and a group that was unexposed and did not have the outcome. In general the last group would be quite large and the first group would be quite small because most people were healthy and disease was rare in the population. He said that it was not possible to know whether group membership proved a causal connection between outcome or lack of outcome and exposure or lack of exposure.
[5.798] Dr Lewis went on to discuss heterogeneity and group memberships. He said that populations were composed of unique individuals who differed in many aspects. The assumption for individual attributability of population rates was that the likelihood of disease (or individual susceptibility) was identical for each member of an unexposed cohort or population. This was patently never the case, because it assumed that the background hazard for disease was the same in all subjects: all individuals were assumed to have precisely the same exposures. Population estimates were based on group data, an average of many individuals, which could apply only to the group and not to any particular individual. If blood pressure was measured in many individuals, no single individual need have the group's average blood pressure. Each population estimate of this sort was a group or aggregation artefact which could contain patterns and associations that did not apply to homogeneous subgroups or to individuals.
[5.799] In populations, he said, we had a baseline rate of disease occurring for which a putative causal agent was not known, and this was often called idiopathic. The background risk posed by this undetermined causal factor could not be defined, because it was unknown. When the baseline rate was compared to the rate observed among the exposed, an excess might be detected. Epidemiologists assumed that the excess in the population was due to the exposure. If this was applied to the individual, then we were again assuming that the exposure rather than the unknown agent was the cause of disease in the exposed. In the case of an exposed individual with disease, however, it was never known whether this person belonged in the "excess" or the "baseline" group.
[5.800] By the same token, he went on to say, individuals could belong to a large number of groups. The identification of an individual by age, sex and exposure did not capture all attributes which might determine risk. People were black or white, rich or poor, employed or unemployed, had unique genetic profiles and engaged in various health and lifestyle habits. Each one of these groups constituted a different risk set and would have a different attributable fraction. Each individual, who might have different characteristics such as being white, poor and unemployed, thus belonged to multiple risk sets, and had multiple attributable fractions that might be applicable to him. Further, for any given disease, the sum of all attributable fractions for all factors would exceed 100%. An individual who ceased to belong to one risk set, by ceasing to be exposed to the relevant exposure, nonetheless might remain in other risk sets. For example, taking hypertension and high cholesterol levels, both of which were risk factors for heart disease, and assuming that the relative risk for each of them is 10 and that 40% of the population had each of these exposures, so that the attributable fraction for each would be 78%, the sum of these attributable fractions would be 156%. Dr Lewis said that thus the assigned share concept did not account for heterogeneity of exposures, heterogeneity of the groups studied in observational epidemiology, or the heterogeneity of the sampling situation within one study and of several studies; nor could it account for multiple group membership of single individuals. The concept of probability of causation or "assigned shares" ignored the existence of diversity and individual differences.
[5.801] Reference was made in this context to Rothman and Greenland 1998. At p.13 the authors stated:
"There is a tendency to think that the sum of the fractions of disease attributable to each of the causes of the disease should be 100%. For example, in their widely cited work, The Causes of Cancer, Doll and Peto (1981; Table 20) created a table giving their estimates of the fraction of all cancers caused by various agents; the total for the fractions was nearly 100%. Although they acknowledged that any case could be caused by more than one agent (which would mean that the attributable fractions would not sum to 100%), they referred to this situation as a 'difficulty' and an 'anomaly.' It is, however, neither a difficulty nor an anomaly, but simply a consequence of allowing for the fact that no event has a single agent as the cause. The fraction of disease that can be attributed to each of the causes of disease in all the causal mechanisms has no upper limit: For cancer or any disease, the upper limit of the total of the fraction of disease attributable to all the component causes of all the causal mechanisms that produce it is not 100% but infinity. Only the fraction of disease attributable to a single component cause cannot exceed 100%."
Dr Lewis said that he agreed with this passage.
[5.802] Dr Lewis was referred to Callum 1998, in the introduction to which, at p.1, it was stated:
"People who have never smoked cigarettes die from diseases that smoking can cause, and to that same extent some cigarette smokers too can die of the disease but not as a result of their smoking. The methods used to estimate the number of deaths caused by smoking are all based on a proportion of deaths caused by smoking, and cannot be traced back to individuals."
Dr Lewis said that he agreed with this statement.
[5.803] Dr Lewis went on to say that another major problem in applying epidemiological data to individuals surrounded the counterfactual condition, which was the proposition that in an alternate world, where the individual was not exposed, disease would not have occurred. This was a separate concept from that of attributable fraction. It assumed an alternate setting in which the same individual remained healthy because he or she was not exposed to the putative causal agent. This, however, could never be proved because the "alternate world" in which the individual was not exposed did not exist and could not be tested. This also entailed considerations of competing risks and of rare events, i.e. the comparison of rare events with even rarer events. The assertion that the exposure caused the disease in a specific individual implied that if there had been no exposure, the individual would not have had the disease. It was unclear, however, what the individual's "alternate world" would look like if the exposure were removed. For example, if a smoker of normal weight had not smoked in an alternate world, he might have gained weight, thereby placing himself in a different risk category. Any exposure might be linked to any number of other factors which might affect the risk category in an unpredictable fashion. The counterfactual conditional therefore implied that, if the exposure had not been present, some other element of risk would not have occurred, and asserted that the individual in question did not belong to the group of unexposed who would have incurred the disease.
[5.804] This, Dr Lewis said, posited a theory counter to fact. It also neglected the fact that, whether the individual was exposed or unexposed, the initial likelihood of developing most diseases was low (for example, the lifetime risk of developing lung cancer was approximately 1 in 100 for the unexposed (non-smokers) and 10 in 100 for the exposed (smokers)). (Lifetime risk should of course be contrasted with annual risk, and Dr Lewis said that to simplify matters he took a lifetime as being 100 years.) He said that these disease likelihoods or predictions accompanied the individual through life until the individual either got the disease or ended his life without getting the disease. If he died without getting the disease (i.e. from another cause), then his disease likelihood would be zero instead of 1 in 100 or 10 in 100. If he did get the disease, his disease likelihood would be 1, although previously it had been 1 in 100 (i.e. 0.01) or, if exposed to the risk factor, 10 in 100 (i.e. 0.1). Whether an individual was exposed or not exposed, the likelihood of disease occurring was close to zero, with the unexposed likelihood slightly closer to zero than the exposed. Dr Lewis illustrated this by reference to a diagram, to demonstrate that for the unexposed group the prior likelihood of getting the disease was always 1 in 100, and for the exposed group the likelihood was 10 in 100, both of which were proportionately far removed from the end likelihood of having the disease. He said that this likelihood turned to 1 immediately for both and unexposed once disease had occurred. It was clearly not justified to argue from the fact that disease did occur that it occurred due to the exposure without taking into account the actual likelihood of disease occurrence. In both instances, the risk of developing lung cancer was very low, and it was unlikely that disease would occur in any individual, but the likelihood was not zero. Even if the exposure was withdrawn in an alternate world, the individual still might develop the disease. Conversely, on an individual level for any case of lung cancer we could, he said, posit ninety-nine "alternate worlds" in which lung cancer would not have occurred to a person who was not exposed, and nine "alternate worlds" in which lung cancer would not have occurred to a person who was exposed. Non-exposure to a specific substance was but one of the many circumstances which could have produced a situation in which no lung cancer occurred. In short, because the initial risk of lung cancer was low in any case, the likelihood that, if circumstances had been different, lung cancer would not have occurred, was high.
[5.805] Dr Lewis's conclusions were expressed in these terms:
"Epidemiological evidence cannot be used to make statements about individual causation. Firstly, the information provided in observational epidemiology is generally such that it can neither confirm [n]or refute a causal relationship, particularly when the exposure in question is not specifically associated with a certain condition. Epidemiology produces information on associations which may be used for public health decision-making and individual guidance, but it cannot provide information on the likelihood that an exposure produced an individual's condition. Secondly, the population attributable risk is a measure for populations only and does not imply a likelihood of disease occurrence within an individual, contingent upon that individual's exposure. Thirdly, the fact that cases and non-cases can emerge both from the unexposed and the exposed groups shows that the likelihood of the individual occurrence cannot be reliably predicted from his or her exposure group membership alone. Fourthly, the group estimates obscure the underlying heterogeneity of the population, so that it is entirely possible that other group memberships besides exposure, like genetic profile, socioeconomic status, workplace, diet and other exposures make a major contribution to disease occurrence. Furthermore, the question of using epidemiological data for individual causation raises the problem of identifying a particular individual who was harmed by the exposure. Models such as the assigned share concept, derived from attributable fractions, have attempted to deal with this, but suffer from the limitations mentioned above. Additionally, the attempt to identify exposure as the sole cause of disease in an individual produces a statement counter to fact in that it implies that the individual would have remained healthy if the exposure had not occurred. This is not provable and cannot be derived from epidemiological data."
[5.806] Dr Lewis was invited to comment on a number of publications. In Robins and Greenland 1989 the authors offered a mathematical definition for the probability of causation that formalised the legal and ordinary-language meaning of the term. They showed that, under this definition, even the average probability of causation among exposed cases was not identifiable from epidemiologic data. This was because the probability of causation depended both on the unknown mechanisms by which exposure affected disease risk and competing risks, and on the unknown degree of heterogeneity in the background disease risk of the exposed population. Dr Lewis was shown passages in the text of the article in which this summary was developed, and said that he agreed with them. In particular, he agreed with a passage at pp.1134-1135 in which the authors stated that the average probability of causation among exposed individuals developing disease could not be estimated from epidemiologic data without resorting to non-identifiable biologic assumptions. Furthermore, even in the absence of competing risks, the average probability of causation among all exposed cases could not be estimated. "This is because these probabilities of causation depend on (i) the unknown mechanisms by which exposure affects disease risk and competing risks, (ii) the unknown degree of heterogeneity in the background risks of disease, and (iii) the unknown degree of dependence between risk of disease and competing risks."
[5.807] Dr Lewis was next asked to consider Colby et al. 1995. At p.93 it was stated:
"There are some variations in the histologic subtypes among smokers and non-smokers (Table 7-4), with squamous cell carcinoma and small cell carcinoma showing the highest association with smoking."
Table 7-4 was headed "Histologic subtypes of carcinomas in smokers and nonsmokers". For those with squamous cell carcinoma the table showed that out of 2926 cases of squamous cell carcinoma, 98% were smokers and 2% were nonsmokers. The table was said to have been modified from reference 9, Rohwedder and Weatherbee 1974. After considering the text of this article, Dr Lewis confirmed that it did not appear to be the correct reference. The correct reference appeared to be reference 10, Rosenow and Carr 1979.
[5.808] Table 2 at p.235 of that paper listed the prevalence of the five most common lung cancers in men and women, smokers and never-smokers, as they had presented at the Mayo Clinic, Rochester, Minnesota. Table 7-4 in Colby et al. could be seen to be a modification of this table. The table in Rosenow and Carr 1979 showed that out of a total of 992 epidermoid (i.e. squamous cell) carcinomas there were 899 in men, 892 of whom were described as smokers and seven as never-smokers, and ninety-three in women, eighty of whom were described as smokers and thirteen as never-smokers. At p.233 the authors stated:
"The Mayo Lung Project of the Mayo Foundation is conducting a controlled prospective lung cancer screening program in males 45 years and older who smoke at least 20 cigarettes per day."
Dr Lewis commented that there were two pieces of information. The first was that there was a high risk project which used only exposed males, and the second was that the Mayo Clinic also captured other lung cancer cases. It was not said how these entered into the figures in the table. No definition was given of the expressions "smoker" and "never-smoker", though by inference a smoker could be taken as someone who had ever smoked, even for a short period at some earlier time in his or her life, i.e. an "ever-smoker". If, by way of illustration, and on the assumption that there was no statistical association between smoking and lung cancer, 98% of the population from which the lung cancer patients were drawn were ever-smokers, one would expect 98% of cases of squamous cell lung cancer among ever-smokers. In Dr Lewis's view there was insufficient information about the population sampled, and the general population from which it was drawn, to enable conclusions to be drawn about the relationship found between ever-smoking and epidermoid cancer. There was no indication that the authors addressed the question of whether there was a statistical association between epidermoid cancer and any other feature to be found in the members of the group, that is to say, it did not appear that they had controlled for confounding in any way.
[5.809] Returning to the passage in Colby et al. 1995, quoted above, Dr Lewis commented that the authors did not provide a statistical test. If they intended to imply that squamous cell carcinoma of the lung was statistically associated with smoking, they would not be entitled to say so on the evidence of Rosenow and Carr 1979. It was unknown where the population discussed in that paper was sampled from. Most likely, most of the information was based on the high risk exposed group, so that there was an over-sampling of individuals with the exposure and it could not be known what a random sample of lung cancer patients would look like.
[5.810] In Van Rossum et al. 2000 the authors followed up a group of British male civil servants aged 40-69 years for twenty-five years to establish whether or not differences in their socio-economic structure had some influence on the mortality experience of this cohort. They concluded that differentials in mortality persisted at older ages for almost all causes of death. Table 1 on p.179 showed aged-adjusted mortality rates per 1000 person years (number of deaths) by employment grade and mortality rate ratios for "other" grade versus administrative. The table included figures for malignant neoplasm of the lung. Dr Lewis explained that these showed a distinct gradient in the mortality experience, from lowest mortality from neoplasm of the lung in the highest socio-economic stratum, to highest mortality of neoplasm of the lung in the lowest socio-economic stratum, the overall estimate of the influence of socio-economic status being 4.08. This, he said, indicated that there was an independent influence of socio-economic status on the mortality experience from malignant neoplasm of the lung (and also, as appeared from the table, from many other causes of mortality), independent of the factors that were adjusted for, such as smoking.
[5.811] In Hart et al. 2001, which related to a study investigating differences in lung cancer mortality risk between social classes in two study groups, one from the general population of Renfrew and Paisley in the west of Scotland and the other from workers employed in the Central Clydeside conurbation, also in the west of Scotland, the authors concluded that there was a difference in lung cancer risk between social classes, in addition to the effect of smoking. This could be explained by poor lung health, deprivation and poor socio-economic conditions throughout life. Dr Lewis said that by "difference" the authors meant a statistically significant difference, and that their conclusion was consistent with that in Van Rossum et al. 2000.
[5.812] Dr Lewis was then asked to comment on Dong and Hemminki 2001. The Swedish Family-Cancer Database, with a study population of 5,520,756 offspring and their parents from 2,112,616 nuclear families, was used to systematically estimate the effects of parental and sibling cancers on the cancer risks in the individuals born after 1934. The article started with the statement that it was well documented that there was familial aggregation of cancer at practically all sites. Dr Lewis said that, in summary, the authors found, specifically for lung cancer, and looking only at parental determinants, a risk ratio of around 5 for the association of two members of a family having lung cancer, and of 13 for the association of both a parent and a sibling having lung cancer.
[5.813] In Samet et al. 1996 the authors used data from a population-based case-control study to examine the relationship between personal and family illness history and lung cancer risk. They stated that the collection of detailed information on cigarette smoking and the use of multiple logistic regression facilitated the control of the effects of smoking when the illness variables were assessed. Dr Lewis explained that this was a statistical technique which adjusted for confounding factors determined in the study, of which smoking was one. The authors found that a history of chronic bronchitis or emphysema was associated with a doubling of lung cancer risk and that if at least one parent had lung cancer the odds radio was increased fivefold. Dr Lewis referred to figures given in Table 5 on p.467, and said that these increases in risk, as shown on the table, were statistically significant.
[5.814] In Yang et al. 1999 the authors concluded that their results suggested the presence of a high-risk gene contributing to early-onset lung cancer in a population where the probands (i.e. study subjects) were nonsmokers. Dr Lewis explained that the authors had estimated the attributable fraction due to the genetic makeup first as the attributable fraction due to an environmental exposure, and postulated that the cause of lung cancer in younger probands was "a rare autosomal co-dominant gene".
[5.815] In NCI 1978 the contributors provided estimates of the fraction of cancer in the United States related to occupational factors. At p.1 they stated that if recent evidence was considered and if the full consequences of occupational exposures in the present and recent past were taken into account, estimates of the fraction of cancer incidence in the United States attributable to occupational exposure in the present and in the foreseeable future of at least 20% appeared much more reasonable than previous, lower estimates, and might even be conservative.
[5.816] Carnow 1978 stated at p.17 that there would appear to be a general consensus that cancer, particularly lung cancer, had multifactorial aetiology which included cigarette smoking, urban air pollution and occupational exposure. Separating out these factors and assessing their quantitative impact on health was critical. The author concluded at p.20:
"I noted at the onset of this discussion that any association found in the presence of such an overwhelming factor as cigarettes must be seriously considered and explained. All of the major prospective sampling studies carried out in the United States and in other countries examined by us, in addition to the regression studies we carried out, revealed a significant relationship between urbanization and lung cancer incidence, and, where benzo[a]pyrene levels were measured, an even greater relationship to levels of benzo[a]pyrene. These relationships hold, even when one standardizes for cigarette smoking, as was done in all of these studies, suggesting that air pollution is, indeed, a significant factor in the etiology of lung cancer."
[5.817] In ACC 1981 at p.13 a table was given for age-adjusted death rates per 100,000 population for selected sites for forty-two countries, derived from statistics for the years 1976 and 1977. For lung cancer in males the rate for Scotland was 108.5, which made Scotland the highest ranking country. For Ireland the rate was 61.8 and its rank was sixteenth, and for Japan the rate was 28.3 and its rank was thirty-first. Dr Lewis added the comment that at that time Japan was known to have one of the highest smoking prevalences in the world.
[5.818] In Gillis et al. 1988a male lung cancer cases and age- and sex-matched controls were interviewed between 1976 and 1981 in a case-control study of cigarette smoking habits and lung cancer in Glasgow and the west of Scotland, "an area with the highest recorded incidence in the world". The authors said that the relative risk of lung cancer increased significantly for smokers whose consumption was below twenty cigarettes per day, but did not rise significantly in those who smoked more than twenty cigarettes per day. The male lung cancer rates during the period of the study were 158 per 100,000 in the Glasgow area and 156 per 100,000 in the Renfrew/Paisley area, about 50% greater than the whole of Scotland rate in ACC 1981. At pp.41-42 the authors stated:
"The major finding in this study is the steep increase in the relative risk of lung cancer observed in West of Scotland smokers with an average consumption of 1-14 and 15-24 cigarettes daily, compared with the small increase in relative risk in smokers with a higher average daily consumption [...]. This is at variance with the majority of the literature which describes a steady increase in relative risk above an average consumption of 20 cigarettes daily. [...]
Thus, in an area of exceptionally high lung cancer incidence, there is a lack of increase of the relative risk of lung cancer at higher levels of cigarette smoking. We have been unable to explain this observation on the basis of confounding bias or artefact.
The low level of relative risk found at all levels of cigarette consumption coupled with the small increase in relative risk observed at the highest levels of smoking represent a paradox for an area with such a very high rate of lung cancer.
The existence of a higher than average proportion of heavy smokers in the West of Scotland population would not seem therefore sufficient by itself to be responsible for the high lung cancer rate. Thus the question of additional susceptibility to lung cancer in the local population is raised."
[5.819] The authors stated that the accompanying paper, Gillis et al. 1988b, attempted to place their findings in the context of a prospective cohort study of west of Scotland men with known smoking habits followed for ten and a half years. At p.46 the authors stated that the main reason for presenting the results derived from the cohort study was to establish whether the unusual finding of a flattening in the dose-response relation of cigarette smoking and lung cancer observed at the higher levels of cigarette consumption in the case-control study in the West of Scotland could be substantiated. The availability of results from both a case-control and a cohort study conducted in the same geographical area during the same period removed many of the biases inherent in the former. The fact that both studies produced dose-response relations which flattened at the higher levels of cigarette consumption suggested that the observation was not an artefact. After further discussion, the authors stated, at p.47:
"It thus appears that it is not just the West of Scotland smoker who is at an increased level of risk compared with his equal smoking counterpart elsewhere but also the West of Scotland non-smoker who may also experience a higher than expected lung cancer risk.
The flattening of the dose-response relation seen in the case-control study has not been explained by any of the smoking characteristics examined. The fact that both the size and shape of the relation can be reproduced in the cohort study leads us to believe that this is a genuine representation of the smoking and lung cancer relation in the West of Scotland.
What this might mean in terms of the aetiology of lung cancer in the West of Scotland is not known. [...]"
Dr Lewis commented that a study such as this demonstrated the need to bear in mind that geographical location might affect results, as might also membership of a risk set by reference to occupation, such as previous military service or membership of the medical profession.
[5.820] Kissen 1963b reported on a study on male lung cancer patients and non-cancer patients in three hospital chest units. The location of these units was not specified, but since the author was a research associate in psychosomatic medicine at the Southern General Hospital, Glasgow, it might be taken that they were in the west of Scotland. The author found that among the lung cancer patients as opposed to the non-cancer controls there was a lower incidence of certain childhood behaviour disorders and a high rate of concealment, or bottling up, of emotional difficulties. He stated that these findings appeared to be independent of cigarette smoking status, and of tumour histology; nor was there any apparent relationship with lack of aggression or lack of sociability. In other words, he stated, the findings supported the hypothesis that lung cancer patients had a poor outlet for emotional discharge and that they tended to conceal or bottle up their emotional difficulties. The significance of childhood behaviour disorders was that they represented early outlets for emotional discharge. In other words, the statement that lung cancer patients, in contrast to non-cancer patients, had poor capacity for emotional discharge, was supported by clinical observation in the first instance, and two subsequent independent objective measures.
[5.821] Dr Lewis said that Eysenck 1965 showed that there was an association between personality and lung cancer. At p.105 the author stated that his general conclusion from studies of personality, social behaviour and body build was that cigarette smokers on the whole tended to show extraverted behaviour patterns and that this relationship was a quantitative one, i.e. the greater the number of cigarettes smoked the greater the degree of extraversion of the smokers concerned. Dr Lewis said that Eysenck was the most prominent, and perhaps the first, person working in his field to suggest a relationship between personality and lung cancer and that in common with Kissen he found that persons with lung cancer differed in personality type from controls.
[5.822] Dr Lewis said that Horne and Picard 1979 explored the psychosocial situation of individuals who had not yet had any diagnosis of cancer, looking at stability functions, such as childhood stability, job stability and marriage stability, and composed from these a scale which was then set out as a predictor of the underlying condition. The authors stated in their conclusion at p.512:
"Smoking is well established as a risk factor for both coronary heart disease and lung cancer. In this study, psychosocial factors were one to two times as important as smoking history in predicting diagnoses of cancer.
Thus, the present data suggest that just as there are psychological risk factors for coronary heart disease [...], there may well be psychosocial risk factors for lung cancer that could be utilized in programs to prevent or detect the illness."
Dr Lewis explained that this meant that psychosocial factors were a stronger predictor of lung cancer than was smoking history in the study group.
[5.823] Knekt et al. 1996 reported on a prospective study of the association between depressiveness and subsequent incidence of lung cancer. The study population were included in the nationally representative Mini-Finland Health Survey. The authors reported that after a fourteen year observation period the relative risk of lung cancer between depressive persons and individuals with a normal depressiveness score was 3.32. Dr Lewis said that the confidence intervals showed that this was a statistically significant result and that this was an adjusted risk estimate which was not influenced by potential confounding factors such as age, education, geographic area and smoking.
[5.824] Alavanja et al. 1992 reported that pre-existing respiratory disease among lifetime non-smokers and previous cigarette smoking among former smokers who had stopped for at least fifteen years were significantly linked to lung cancer in a population-based case-control study of non-smoking women. Asthma, emphysema, pneumonia and tuberculosis appeared meaningfully related to lung cancer. Overall, a history of non-malignant respiratory disease might explain the cause of nearly 16% of the incident cases of lung cancer among non-smokers. Dr Lewis said that it was possible from these results to postulate that a history of non-malignant respiratory disease might explain the cause of some percentage of the incident cases of lung cancer among smokers also.
[5.825] Bandera et al. 2001 reviewed the epidemiological evidence for association of alcohol and lung cancer. The authors stated:
"The studies reviewed here provide some indication that alcohol and particularly beer intake may increase lung cancer risk after controlling for cigarette smoking. Although the evidence is not conclusive, it warrants additional consideration of alcohol as a risk factor in lung cancer etiology, independent of cigarette smoking."
The authors also commented that diet was an important factor to consider when evaluating the relationship between alcohol and lung cancer, because consumption of alcoholic beverages might displace other elements of the diet, particularly for heavy drinkers, and the effects of ethanol (i.e. alcohol) on appetite, digestion and absorption of nutrients were well documented. Diets high in fat and low in fruit consumption had been observed among heavy alcohol drinkers and beer drinkers. Also, smokers tended to consume more high-fat foods and less fruit and vegetables. Dr Lewis said that he accepted the authors' view that the current epidemiological evidence was suggestive of an increased lung cancer risk associated with drinking alcohol, although it remained to be established whether this was a group phenomenon or whether the link between alcohol and lung cancer was direct.
[5.826] Pollack et al. 1984 was a prospective cohort study of the relation between alcohol consumption and the subsequent occurrence of the five most frequent cancers in Japanese men in Hawaii. The analysis, which was adjusted for the effects of age and cigarette smoking, revealed inter alia a significant positive relation between alcohol consumption and lung cancer incidences, accounted for primarily by an increased risk among subjects who consumed larger amounts of wine or whiskey. The authors reported that among the subjects who drank more than forty ounces of alcohol per month, a larger proportion were current cigarette smokers, but even after adjustment for this factor, the apparent relation between alcohol consumption and lung cancer incidence remained significant. In addition, even when several categories for the number of cigarettes smoked were used as co-variates, the relation still held.
[5.827] In Darby et al. 2001 the authors (one of whom was Sir Richard Doll) examined the relationship between diet and lung cancer in a case-control study of cases of lung cancer and population controls in south-west England in which subjects were interviewed personally about their smoking habits and their consumption of foods and supplements rich in retinol or carotene. The authors reported at p.734:
"In our study, even after adjusting for smoking, for social class, and for all the other dietary factors simultaneously, we still observed statistically significant and apparently independent associations of lung cancer risk with pre-formed retinol (increase in risk with increased intake), and with fish liver oil, vitamin pills, carrots, and tomato sauce (decreases in risk with increase in intake). These specific associations have all been observed in at least some previous studies."
At p.735 they concluded that there was at least one as yet unidentified factor that was causally related to lung cancer risk and of considerable importance in terms of attributable risk, and that any such factor, or factors, were probably dietary, although not necessarily so.
[5.828] Finally, Herman and Rao 1971 investigated differences in lung cancer diagnosis among the Jews and non-Jews of Pittsburgh, Pennsylvania in the period 1953 to 1967. Dr Lewis said that they found that there were differences by religion in that there appeared to be a lower lung cancer risk among Jewish people. He said that this exemplified ethnic differences.
Cross-examination of Dr Michael Lewis
[5.829] In cross-examination, Dr Lewis said that he had been approached as an epidemiologist by lawyers acting for ITL. He had given a presentation to members of ITL in about 2001, and had provided a formal paper. He had come from Germany to give evidence. He did not know why no British epidemiologist was giving evidence. Asked whether there was "any truth in the story that the tobacco industry could not find a British epidemiologist to come and give evidence", he said he did not know. He accepted that it was generally agreed among doctors and scientists in the United Kingdom that tobacco smoking might cause lung cancer and that the evidence for this came almost entirely from epidemiological studies. He agreed that smoking consistently showed a strong association with lung cancer. On the other hand, Dong and Hemminki 2001 showed that family clustering had similarly strong associations.
[5.830] Asked about the statement by Bandera et al. 2001 that "cigarette smoking is undoubtedly the most important risk factor for lung cancer", he said that if this meant that it was the one with the strongest association found, then he agreed. Asked about the association between alcohol consumption and lung cancer, he agreed that there were studies that showed that there was no association. This was in part because it was difficult to understand, as there might be a group phenomenon of individuals who combined the two factors of smoking and alcohol consumption. Asked about various risk factors, such as diet, geographical location and socio-economic class, he said that where two or more risk factors were found together, they might or might not have some kind of combined effect. Groups with two or more risk factors might be sharing confounding associations. "We can simply observe that there are groups in the population that are at increased risk of lung cancer and that these groups combine certain factors which are themselves independently associated. But this does not allow us any statement concerning the mechanics of this." The observed association between lung cancer, smoking and exposure to asbestos could be regarded as showing a statistical synergism.
[5.831] Dr Lewis said that his evidence about the reduction of the risk factor, for example cigarette smoking, was not related to the fact that it took years for the death rates among ex-smokers to reduce to the same level as those of non-smokers. He was asked to comment on this passage in RCP 1971, p.54, para.4.11:
"Those who give up smoking cigarettes are much less likely to get cancer of the lung than those who continue; the excess risk decreases rapidly after stopping. In British doctors [...] who had stopped smoking the risk fell within five years to about one half of that of continuing smokers and after fifteen years was only three times the very small risk of non-smokers (Figure 4.3)."
He was also shown this figure, the caption to which contained the statement:
"Continuing smokers have a greatly increased risk of lung cancer but this risk is halved in those who had stopped between one and five years, and in those who had stopped fifteen years is down to about twice that of non-smokers."
He said that he accepted what was shown there, in the context that this constituted results from an observational study, as distinct from what he had said about the prediction that removal of exposure resulted in a reduction in risk. He was not suggesting that if a person stopped smoking, the risk would disappear. This was an observational approach contingent on the population under investigation and, within the context of this population, there could be seen to be a decline in risk associated with those who did not continue to smoke rather than those who did. This did not allow a distinction to be made between the characteristics of smokers and non-smokers. That was why, when looking at the attributable risk calculations, what was being dealt with was a prediction which needed to be examined by experiment. The studies he referred to in that regard took an experimental approach: certain groups were randomised to exposure and non-exposure, and therefore differences in the observational groups were equalised. This allowed the observers to focus on attributes and exposures such as blood pressure, cholesterol and cigarette smoking, but in these studies the kind of risk reduction that had been shown in the observational studies had not been shown. A statement that 90% of lung cancer cases were caused by smoking assumed causation, this was the assumption that underlay the attributable fraction calculation, which again was unproven until put to the test, and in his opinion it had not been.
[5.832] Dr Lewis said he was aware of Doll et al. 1994. Counsel asked him in the first place about the abstract. This indicated that the subjects were 34,439 British male doctors who replied to a postal questionnaire in 1951, of whom 10,000 had died during the first twenty years and another 10,000 had died during the second twenty years. Of the results, the abstract stated:
"Excess mortality associated with smoking was about twice as extreme during the second half of the study as it had been during the first half. The death rate ratios during 1971 - 91 (comparing continuing cigarette smokers with life-long non-smokers) were approximately threefold at ages 45 - 64 and twofold at ages 65 - 84. The excess mortality was chiefly from diseases that can be caused by smoking. [...]"
The conclusion, also in the abstract, was:
"Results from the first 20 years of this study, and of other studies at that time, substantially underestimated the hazards of long term use of tobacco. It now seems that about half of all regular cigarette smokers will eventually be killed by their habit."
Asked whether he accepted the results put forward by the authors, Dr Lewis said he accepted that they showed an increase of the association in the second half of the study. He did not understand the reason for that. He did not see any basis that he could reconstruct for the statement that half of all regular cigarette smokers would eventually be killed by their habit. He did not agree that the results from the first twenty years substantially underestimated the hazards of long term use of tobacco. He agreed that this particular study showed a dose relationship between the number of cigarettes smoked per day and the likelihood of developing lung cancer, and also a relationship between the length of time someone smoked and the likelihood of developing lung cancer.
[5.833] Table III in Doll et al. 1994, on p.903, was said to show the rates for seventeen types of cancer. The authors stated:
"Three types (cancers of the upper respiratory sites, lung, and oesophagus) were particularly closely related [to smoking], with the mortality in heavy cigarette smokers at least 15 times that in non-smokers [...]."
Dr Lewis commented that the table showed an analysis that he was not familiar with. They were performing a standardised test for trend which did not allow him to establish whether or not it was indeed fifteen times. In other words, they were running the test across all categories of smoking and establishing a standardised test for trend, and this was a rather unusual form of analysis. It did not give a risk ratio, which meant that he could not reconstruct from the table whether the figure of 15 was correct.
[5.834] At p.906 the authors stated, under the heading "Actuarial survival by smoking habit":
"As the observations have been made over such a long time, it has been possible to follow many men into their 10th - and a few even into their 11th - decade of life. It has, therefore, been possible to calculate "actuarial" survival curves for different categories of smoker not only throughout middle age but also into old age, with results that are reasonably reliable up to at least 85 years of age. The results from 35 years of age are shown for non-smokers and continuing cigarette smokers in figure 1, and for non-smokers and continuing light, moderate, and heavy cigarette smokers in figure 2. The most notable differences are in the proportions who die between 35 and 69 years of age, which vary from 20% in non-smokers to 41% in cigarette smokers as a whole and to 50% in those who smoke 25 or more cigarettes a day."
Dr Lewis said that he had seen the figures (i.e. graphs) and in all honesty it had always been difficult for him to understand this paper, simply because he did not see the numbers behind it. He saw the figures, he did not know what the numbers were behind them and he did not see a table that gave him the numbers, stratified for example for age groups in the way that was laid out in the figures. He could see that the trend described by the authors was there, but he was unable to know where the figures were actually coming from in terms of the real data. He had read the article many times, but it simply did not supply him with the numbers that he required in order to actually reconstruct the figures. Counsel asked whether the authors were putting forward false numbers. Dr Lewis said that was not what he was saying; he simply did not see the numbers, as would be the norm for a paper such as this, to inform the scientific reader of the numbers that were in each of these strata. They had gone straight to informing him of the fact that this was what the results were. It was traditional for scientific papers to provide the numbers so that one could satisfy oneself that the results were indeed correct.
[5.835] Dr Lewis was asked about a passage on p.908, under the heading "Discussion: causation, confounding, and chance". The authors wrote:
"In general, the results in this study for specific causes of death accord with those in many other studies. Reinforced by other evidence, they helped lead more than 30 years ago to the conclusion that the associations observed between smoking habit and mortality are chiefly causal in character (see also the more recent reviews by the US Surgeon General and the International Agency for Research on Cancer.) In most of the causal associations, smoking seems to act synergistically with other aetiological agents such as consumption of alcohol; various aspects of diet; level of blood pressure, blood lipids, or other cardiovascular risk factors; or exposure to asbestos, radon, or possibly some infective factors."
Dr Lewis said of the words "smoking seems to act synergistically with other aetiological agents" that the authors were drawing conclusions from an observational study in which they observed populations in which they noted, on the basis of other studies, that there was some statistical synergism of other shared attributes which influenced the association with lung cancer. He said that going from a population study to this statement was going beyond the evidence.
[5.836] The authors concluded the paragraph: "The quantitative effect of smoking will, therefore, vary with variation in the prevalence of these other agents." Dr Lewis commented that other factors could affect the association. The authors continued:
"In some instances smoking and other aetiological agents may also may be confounded, as with the consumption of alcohol, various aspects of personality, and perhaps some aspects of diet. Indeed, for a few causes of death the confounding between smoking and other factors may account for virtually all the observed association with smoking, in which case none of the excess mortality from those causes should be attributed to the habit. The only plausible examples of important associations that might be due to this, however, are those that involve cirrhosis, suicide, and other traumatic causes, for which the associations with smoking may be secondary to associations with alcohol and personality."
Asked whether he agreed with this paragraph, Dr Lewis said that it seemed speculative; it was part of the discussion, but he did not see where it was determined from the authors' data. The expression "plausible examples" appeared to him to be a part of what was going through the authors' minds, but not something which was actually obtained from the studies they had conducted.
[5.837] Later on the same page, the authors wrote:
"The attribution of some of the excess mortality in cigarette smokers to confounding does not, however, necessarily mean that the overall effect of smoking on all cause mortality is less than that observed, as the factors with which tobacco is confounded may also have other effects which tend to reduce mortality."
Dr Lewis said that he agreed with this observation inasmuch as, if it was reduced to population terms again, then indeed it might be so, that there were confounders that could act either way for that association.
[5.838] The authors continued:
"This is probably the case with alcohol, the consumption of which (at least in moderation) is associated with a reduced risk of ischaemic heart disease. If, as now seems likely, the consumption of a few units of alcohol a day can reduce the age specific mortality from ischaemic heart disease, then the confounding of cigarette consumption with the consumption of alcohol would mean that the total effect of cigarette smoking on ischaemic heart disease mortality is slightly more than that hitherto reported."
Dr Lewis said that they were positing this as an example, but it again implied that an association with one thing offset an association with another. This was a discussive aspect of the paper which had not been shown in any study that he knew of. It had not been confirmed one way or the other that low-level alcohol consumption in a smoking population resulted in an amelioration of overall mortality from myocardial infarction associated with cigarette smoking. He agreed with the statement at p.909 that two massive cohort studies carried out by the American Cancer Society, one in the 1960s and one in the 1980s, showed a similar increase over time in the excess mortality in cigarette smokers.
[5.839] Dr Lewis was next asked to consider passages in Callum 1998 at pp.37-38. The author stated that large social class differentials in cigarette smoking were described earlier in the paper, and that social class differentials in smoking "imply social class differentials in the percentage of deaths caused by smoking." She went on:
"Social class gradients in mortality are a well-documented and long-standing feature of UK mortality, the differential widening during the 1970s and 1980s."
Dr Lewis said that he accepted this statement, and the figures adduced in support of it.
[5.840] The author continued:
"Ideally deaths due to smoking would be estimated separately by social class and then amalgamated to provide a figure unaffected by social class differentials in mortality independent of smoking. Comprehensive, consistent and reliable data for such an exercise are unfortunately not available."
Dr Lewis said he thought that this was correct.
[5.841] The author then went on to say that the data that were available permitted a limited but illustrative exercise to be undertaken. Percentages and numbers of deaths attributable to smoking by social class for each disease and for men and women were estimated. The total obtained by adding separate social class estimates was then compared with the estimate for all classes combined. In the case of lung cancer a second estimate by social class was produced. This was said to subsume different mortality risks of smoking for each social class, with United States relative risks by educational level (derived from a study known as CPS II) acting as a crude proxy for social class. The author stated, at p.38:
"In conclusion, the figures provide no evidence of significant bias in the UK estimate of deaths caused by smoking associated by social class differentials in cigarette smoking and mortality. If anything they suggest that the estimate derived for all classes combined slightly understates deaths from smoking."
Dr Lewis said that he did not accept that this was demonstrated by the figures. He felt that it was rather a contorted exercise, inasmuch as the author used data from CPS II and transposed them to the United Kingdom. It was known from studies carried out in the United Kingdom that all showed an independent association of mortality with social class. So he could not confirm the author's conclusions, which were based on so many assumptions surrounding the attributable fractions, that data from CPS II apply to the United Kingdom, and so forth. He did not see the evidentiary aspect for the statement that the figures provided no evidence of significant bias arising from social class differentials. Studies in the United Kingdom did indeed show that social class was a risk factor. This study took estimates from the studies that showed that and proposed that certain fractions of mortality were attributable to social class. This introduced a different aspect: it went from the study estimate to the notion that, if social class differences were removed, this would be the result. The Whitehall Study (discussed in Van Rossum et al. 2000) was very well able to distinguish populations in different social classes, but he did not think that it was possible, on a population basis, to make such a distinction in the approach adopted by Callum.
[5.842] Dr Lewis said that he was aware in a general way of statistics given in MacAskill et al. 2002 that there was a marked social class gradient in Scotland, there being higher concentrations for smokers in more disadvantaged social groups. He was asked about the statistics used in Dong and Hemminki 2001, Knekt et al. 1996, Carnow 1978, Kissen 1963b and Gillis et al. 1988b; he explained how it could be seen from the statistics provided in these papers how the authors reached their conclusions. Counsel asked him whether the statistics were based on a small number of cases of lung cancer in each instance, and whether it was not very safe to rely on such small numbers for epidemiological studies. Dr Lewis said that in any group lung cancer was a rare disease, which is why estimates were used, and designated as such; the larger the numbers, the more stable the estimates.
[5.843] Dr Lewis agreed that a strong association had been shown between cigarette smoking and lung cancer, but he disagreed that the conclusion could be drawn from this that it was highly probable that smoking caused lung cancer in any individual. This was because it was necessary to look at all the other attributes and features which might also contribute or be associated with the emergence of that type of carcinoma. These could not be ruled out. The condition was one that in any event had low incidence, so there was a low likelihood of its occurring under any condition. It would not be scientific, where a disease occurred in association with more than one attribute, to conclude that it was causally connected with one of those attributes, even where the association with it was stronger.
Re-examination of Dr Michael Lewis
[5.844] In re-examination Dr Lewis said that he was aware that Professor MacRae, a biostatistician with a great deal of epidemiological experience who worked at the University of Surrey, had been giving advice in this case, but died during 2002. Dr Lewis had become involved thereafter in active preparations for giving evidence himself.
[5.845] He was asked about RCP 1971, p.55, Figure 4.3. He explained that in the observational context, reflected in this figure, the groups of individuals had self-selected themselves for the exposure by deciding to become smokers and had deselected themselves from the exposure over time. There were a core group of individuals who decided to continue to smoke. So all this was a selection phenomenon, occurring in this observational study, which did not allow it to be distinguished whether the effect being seen was a group phenomenon based on the type of individual that made a decision, or whether it actually was an exposure phenomenon. In the case of an interventional-type study or quasi-experimental study, the exposure was randomly allocated and the groups selected for "special intervention" and "usual care" were followed over time. Because this was done in a randomised way, it was possible to establish whether or not an effect could be seen based on the reduction of the exposure, independent from the characteristics of the individuals. MRFIT was a large study in the United States, with a sample of about 300,000 men. From these a group of high risk individuals were taken and were evenly distributed in "intervention" and "usual care" groups, of 12,000 each. Over a period of about twenty or thirty years multiple risk intervention was carried out by singling one randomly allocated group out for special intervention, focusing on reduction of cardiovascular risk factors, including smoking, and then looking to see whether at the end there was any reduction in overall mortality and specific mortality, such as cardiovascular mortality, but also lung cancer mortality. They found no difference in the lung cancer mortality, nor in overall mortality, when they compared the "special intervention" and the "usual care" groups, despite a good success in reducing all risk factor levels, including the smoking prevalence in the intervention group. This raised questions about the causal connection between the exposure and the disease outcome. It showed at least that the attributable fraction calculation was not valid because, under a quasi-experimental setting, no effect was seen of reducing the exposure in a randomly allocated group.
[5.846] Finally, Dr Lewis said that he was aware that there were studies that indicated that squamous cell lung cancer was found in never-smokers as well as in smokers: the table in Rosenow and Carr 1979, even though it was not accurate in terms of the population distribution, showed never-smokers with squamous cell carcinoma.
[5.847] Professor Charles Platz, aged 66, was a retired surgical pathologist. After completing his education, which included the degree of MD from the University of Chicago and postgraduate posts held there, he held a series of academic appointments. Between 1967 and 1975 he held successively the posts of Instructor, Assistant Professor and Associate Professor in Pathology at the University of Chicago. From 1975 to 2000 he held posts as Associate Professor and thereafter Professor of Pathology at the University of Iowa College of Medicine, before becoming Professor Emeritus in Pathology there in 2000. Alongside his academic appointments, from 1970 to 1975 he was Associate Director in Surgical Pathology at the University of Chicago. From 1975 to 1976 he was Director in Surgical Pathology at University of Iowa Hospitals and Clinics. From 1977 to 1995 he was Consultant in Pathology at the Veterans Administration Hospital, Iowa City, and from 1980 to 1992 Consultant in Pathology at the Veterans Administration Hospital in Des Moines, both Iowa. From 1982 to 1987 he was Director and from 1987 to 1994 Co-Director in Surgical Pathology at University of Iowa Hospitals and Clinics. From April 2002 he had been Consultant in the Iowa State Cancer Registry at the University of Iowa. He had also recently held some locum tenens appointments.
[5.848] His CV included a list of the conference presentations he had made, many of them on various aspects of human pathology, and a list of numerous publications of which he was author or co-author, many or most of them relating to aspects of various types of cancer. His area of research interest was the pathology of neoplastic diseases. He explained that these included benign and malignant cellular proliferations, the latter otherwise designated as cancer. Some of his research had involved work relating to lung cancer. It was a regular feature of his clinical practice over the years to examine lung tissue for the purposes of considering the diagnosis of lung cancer. He had given a number of invited lectures, relating to a variety of cancers.
[5.849] Professor Platz explained that University of Iowa Hospitals and Clinics was a large tertiary care centre. It was originally set up to care for indigent patients, but they were no longer as large a part of the patient care as used to be the case. The institution had become a fairly widely known research and treatment institution throughout the Mid-West. About half the patients were funded by insurance. Some of them were funded by Medicare, a state programme in the United States. A small percentage paid their own way. Veterans Administration (VA) hospitals were set up to care for veterans of the armed services, originally veterans of the First World War and thereafter the Second World War. To qualify for treatment, patients required to have served a specified amount of time in the armed forces. For a time all veterans were given free medical care, but thereafter limits had been introduced on the disorders or diseases which would be cared for by VA hospitals. There were also income restrictions on patients, so that the majority of the VA administration hospital patients were low-income to indigent.
[5.850] Professor Platz was asked about Randall et al. 1987. The authors of a previous study had concluded that there were virtually no meaningful differences in the use of health services between veterans and nonveterans, which suggested that healthcare planning within the VA might proceed similarly to healthcare planning for the civilian population. At p.1099 Randall et al. stated:
"Our own analyses of 1979 NHIS [National Health Interview Survey] data for male veterans suggest that VA health planning should proceed along significantly different lines than community health planning. In particular, although veterans and nonveterans may show no meaningful differences in the aggregate in the use of health services, male veterans who choose to use the VA system differ markedly in demographic, economic and health characteristics from male veterans who choose to use community health care facilities. These results indicate that VA healthcare planning must account for the distinct differences between its user population and the veteran population in general."
[5.851] After discussing the data, the authors turned to the results presented in Table 2. They said at p.1102 that these suggested that the VA patient had lower education attainment than his counterpart who used community hospitals, had less annual income than his community counterpart, was more likely to be retired because of health reasons than his community counterpart, was more likely to be out of the labour force, was more likely to live alone, was more likely to be unable to perform his usual activity because of limitations as a result of chronic conditions and was more likely to rate his health status as poor than his counterpart who used community hospitals. They went on to say:
"The above results show that the veteran who receives care in VA hospitals is quite different from his counterpart who seeks care in community facilities. The observed differences in education, annual income, labor force participation, and family structure suggest that the veteran who seeks care from the VA may have very different health care needs than the veteran who seeks care from the community. The results concerning limitations on activity and overall health status tend to confirm these other suggestive findings. [...]
The health problems of the poor and disadvantaged tend to be more numerous and more complex than those of the middle and upper income groups. Poor nutrition, inferior housing, inadequate sanitation, and the physical and psychological stresses of unemployment and deprivation all interact to intensify the health problems of the poor. The poor and less educated have higher death rates than others. For people between 45 and 64 years of age, the prevalence of chronic conditions such as arthritis, diabetes, hearing and visual impairments, heart conditions, and hypertension remains two to three times higher for those with low incomes than for others. [...]
The finding that veterans and nonveterans use hospital services at similar rates [...] does not justify the conclusion that VA and community health planning can proceed along similar lines. Instead, VA planning needs to recognise that VA facilities do not treat a representative sample of all veterans, but rather treat a unique subset of veterans that have different demographic, economic and health characteristics than veterans who seek care from the community."
Professor Platz said that the findings of Randall et al. 1987 were consistent with his own experience. He added that nutritional status was of relevance for the maintenance of the bronchial epithelium because vitamin A was involved in its repair and a deficiency in that vitamin could have implications for its maintenance.
[5.852] During the course of his evidence Professor Platz was asked to examine a biopsy slide which had been prepared from tissue taken from Mr McTear by means of the bronchoscope during the course of the procedure carried out on 10 June 1992. He used a microscope with a video attachment which enabled the slide to be seen on screens in court. He was able to indicate features of the sample as he gave his evidence. He explained that in examining such a slide he was looking for a disruption in the organisation of the normal bronchial epithelium by neoplastic epithelial cells. He described how part of the sample showed normal organisation of basal and columnar cells, with cilia on the surface, and part showed clusters of carcinoma cells. The other features which he looked for were mitotic figures, which were cells in the process of division. He indicated one such figure. He said that what he looked for in squamous carcinoma were structures called intercellular bridges and a process of maturation called keratinisation. The intercellular bridges were difficult to find, and because of this and the absence of keratinisation the squamous cell carcinoma should be described as poorly differentiated. Squamous metaplasias in the bronchus sometimes had keratin and sometimes they did not. Squamous metaplasia was a very common finding as a response to any kind of noxious stimulus or inflammation, it accompanied infections and was also part of a repair process after injury.
[5.853] Having considered and rejected the diagnosis of adenocarcinoma, Professor Platz expressed the opinion that the slide showed a sample of squamous cell bronchocarcinoma. He said that such a condition did not differ histologically as between smokers and non-smokers, as between drinkers and non-drinkers, as between persons suffering from anxiety, stress or depression and those with a different sort of personality, and those with a family history of cancer and those without. It was not possible to tell from the slide (and the other slides, which he had also examined before giving evidence) what caused Mr McTear's disease.
[5.854] Professor Platz was asked to comment on Auerbach et al. 1957 and Auerbach et al. 1961. He said that Dr Auerbach was a pathologist in the VA system and spent most of his time at the Veterans Administration Hospital in East Orange, New Jersey. He also become Professor of Pathology at New York Medical College, while remaining Consultant in Pathology at Veterans Administration Hospital. The 1957 paper reported on slides prepared from the tracheobronchial trees of 117 patients who had died and been autopsied at the Veterans Administration Hospital, East Orange, New Jersey. The material was studied only after smoking histories of the patients had been obtained. All patients dying of lung cancer were included. Those dying of other causes were included only if an adequate smoking history could be obtained. In the 1961 paper the authors stated at p.253:
"In 1957 two of us [...], with several collaborators, presented a report of progress on a study of the relative frequency of carcinoma in situ, squamous metaplasia, stratification and basal-cell hyperplasia in the lungs of nonsmokers and light and heavy smokers of cigarettes. The study was carried out on the lungs of 117 male patients dying with and without lung cancer by means of step sections of the entire tracheobronchial tree, totaling 208 sections per case. [...] All the changes mentioned above were found to be more frequent in the lungs of cigarette smokers than in those of nonsmokers, but the results of greatest interest concerned carcinoma in situ. [...]
These figures seemed to show a very definite linkage between cigarette smoking and carcinoma in situ. But they were challenged because it was claimed that we were including cases of squamous metaplasia in the category of carcinoma in situ and that our results should be re-examined on that basis. We became convinced that there was some justification for this criticism. We discovered that in the interpretation of sections of bronchial epithelium from several children our tabulations had included large recordings of carcinoma in situ. On careful re-examination it was found that the lesions observed in these cases were very different from the carcinomas in situ found in the bronchial epithelium of men who died of lung cancer. It seemed important, therefore, to restudy the whole question."
Professor Platz explained that "carcinoma in situ" referred to a change that occurs in the epithelium, out of which it was thought that invasive carcinoma would ultimately develop. Carcinoma in situ was not invading and therefore not yet capable of metastasis. The expression "carcinoma in situ" in the paper was applied to all bronchial epithelial lesions composed entirely of atypical cells and lacking cilia. The authors said that they did not mean to imply that, given sufficient time, all such lesions would inevitably become invasive.
[5.855] At p.254 of the 1961 paper the authors stated:
"Ever since 1954 the tracheobronchial tree has been dissected out of the lungs of every person examined at autopsy at the Veterans Hospital, East Orange, New Jersey. More recently, a similar arrangement was made with 11 other hospitals, most of which are in upstate New York. All this material is imbedded in paraffin and is available for study."
Altogether 402 white males, including the 117 subjects in the 1957 paper, were included in the 1961 study.
[5.856] Professor Platz commented that nothing was said about the eleven other hospitals referred to in the paper, so it was not possible to tell how many subjects, in addition to the 117 in the progress report, came from VA hospitals. It was not apparent when the arrangement with the eleven other hospitals first came into effect. The authors did not analyse whether or not the incidence of changes in the lungs differed as between patients at the VA hospital in East Orange and those at other hospitals. In very general terms the patients at a VA hospital were a selected population. It was not known in the case of such patients whether there had been exposure to toxic gas, for example, as part of their service commitment. The morphological responses might be related to irritants of more than one sort. Moreover, the VA hospital population had some features associated with low socio-economic status and with possibly inadequate nutrition, which might play a role in the disease processes of the patients, as many of them were likely to have pulmonary problems of one sort or another in the form of infection. Soldiers who served in the trenches in the First World War might have been exposed to poison gas. While the authors had addressed the relative frequency of various changes in the lungs of nonsmokers and light and heavy smokers of cigarettes, they had not addressed whether or not the incidence of lesions varied with nutritional status, socio-economic status, a family history of lung cancer or a propensity to anxiety, stress or depression. So far as Professor Platz was aware no study had been carried out along these lines addressing these questions. Nor had a study been carried out addressing whether or not the incidence of such lesions varied with the consumption of alcohol. He understood that there was an association between heavy drinking and malnutrition.
[5.857] At p.267 the authors stated:
"In this study, we found hyperplasia, destruction of cilia (or destruction of ciliated cells) and the occurrence of atypical cells all to increase greatly with the amount of cigarette smoking. We cannot escape the conclusion that a direct causal relation exists between the degree of exposure to cigarette smoke and the frequency and degree of these changes in bronchial epithelium. [...]
In our opinion the histologic evidence from this study greatly strengthens the already overwhelming body of epidemiologic evidence that cigarette smoking is a major factor in the causation of bronchogenic carcinoma."
Professor Platz said that he was unable to agree with this conclusion because the other possible factors involved in causing the changes observed in the study had not been evaluated. The authors looked at one factor and not at any others. They looked at a factor that would support the statistical evidence, referred to by them as epidemiologic evidence, and while that might be the case, Professor Platz said that he certainly did not know that it was.
[5.858] On the question how an autopsy came to be performed in any particular case, Professor Platz said that in his own experience an autopsy was carried out either with permission of the family, or if there was a potential legal issue, or if the body was unclaimed. This latter factor might select the autopsy population towards indigent individuals living in chronic care facilities.
Cross-examination of Professor Charles Platz
[5.859] In cross-examination Professor Platz says that he did not know that cigarette smoking could cause lung cancer. He was aware that in the United Kingdom the great majority of scientists and doctors and in the United States every cigarette company and the United States Government agreed that cigarette smoking could cause lung cancer. He said he was not convinced. Asked whether that put him "out on a limb", he said that it left him in the minority.
[5.860] He was then taken through Auerbach et al. 1961 in more detail than had been put to him in the course of his evidence in chief. Since, however, he did not in any way disagree with the factual findings of the authors, I do not find it necessary to repeat all of this detail here. It is sufficient to quote from pp.266-7:
"The three principal types of epithelial changes recorded were as follows: increase in number of cell rows; loss of cilia; and presence of atypical cells. [...] Taken singly, each of the three variables increased rapidly with the amount of cigarette smoking. This was true both of the number of sections showing a particular change and of the degree of the change. The proportion of sections with these changes was of the same order of magnitude or somewhat greater in men who died of lung cancer than in heavy smokers who died of other diseases.
In sections for nonsmokers such lesions as were found more often than not showed only one of the three types of change (that is, 3 or more rows of cells but cilia present and no atypical cells, cilia absent but no increase in thickness and no atypical cells, a few atypical cells but cilia present and less than 3 cell rows). Among nonsmokers there were no lesions composed entirely of atypical cells with cilia absent, extremely few lesions with cilia absent and some atypical cells and few lesions with 3 or more cell rows with some atypical cells and cilia present.
In sections from men who smoked 2 packages or more of cigarettes a day and in sections from men who died of lung cancer atypical cells were found in every lesion with 3 or more cell rows and in every lesion with cilia absent. Among these subjects, atypical cells were almost always present, even in areas with neither hyperplasia nor loss of cilia. Both in lesions with cilia present and in lesions with cilia absent the average proportion of atypical cells increased somewhat with the increasing number of cell rows.
The most striking lesions (aside from invasive carcinoma) were those composed entirely of atypical cells with cilia absent. The great majority of these were 5 or more cell rows in average depth. In other words, they exhibited all three types of change (that is, increase in number of cell rows, loss of cilia and the maximum proportion of atypical cells). None were found among men who never smoked regularly, and very few were found among light smokers, but such lesions were found in 4.3 per cent of sections from men who smoked 1 to 2 packages a day, 11.4 per cent of sections from those who smoked 2 or more packages a day and 15.0 per cent of sections from those who died of lung cancer."
Professor Platz, as I have said, did not take issue with these findings.
[5.861] He also agreed with this passage at p.267:
"It is known that cigarette smoke contains irritants, toxic substances and agents that are carcinogenic to the skin of laboratory animals. The application of irritants to epithelial tissue usually produces hyperplasia; toxic substances can destroy cells, and carcinogenic agents produce atypical changes in cells."
On the basis of the study Professor Platz was of the view that a strong association was established between the degree of exposure to cigarette smoke and the frequency and degree of the changes found in bronchial epithelium. He said that it was a question of judgment whether this constituted a causal relationship. He was not aware of any study of the other risk factors for lung cancer which showed the same graded changes to the bronchial epithelium.
Ü Þ