England and Wales Court of Appeal (Civil Division) Decisions

IN THE SUPREME COURT OF JUDICATURE
COURT OF APPEAL (CIVIL DIVISION)
ON APPEAL FROM CHANCERY DIVISION
MR JUSTICE NEUBERGER

		Royal Courts of Justice Strand, London, WC2A 2LL
		31st July 2002

1. This is the judgment of the court.
2. These appeals are against orders of Neuberger J in three sets of proceedings, two of them being brought by companies seeking declarations of non-infringement and revocation of European Patent UK No. 0148605 and the third being an action alleging infringement of that patent by its registered proprietor Kirin-Amgen Incorporated and their exclusive licensees. We will refer to them as Amgen. The practical effect of those actions was that Amgen claimed relief for infringement and the appellants Transkaryotic Therapies Inc. and Hoechst Marion Roussel Limited denied infringement and counterclaimed for revocation. We will refer to them as TKT.
3. There were other proceedings against what may be conveniently called the Roche parties, but after judgment Amgen and those parties settled their differences.
4. The judge concluded that claim 19 and its dependent claims were invalid upon the ground of insufficiency. Otherwise he rejected the attack upon the patent. He held that TKT infringed claim 26. What we have called the main appeal is concerned with those conclusions. TKT contend that the patent is invalid and not infringed. Amgen contend that the judge should have rejected the attack on claim 19 and its dependent claims. They also support the findings of the judge in their favour on additional grounds to those contained in that part of the judge’s judgment that was in their favour.
5. After judgment the judge had to decide a number of disputes relating to the order that should be made. To do that he directed that there should be a hearing starting on 3rd May 2001. At that hearing it was not possible to finalise the order, but the judge made an interim order dated 16th May 2001 which included a declaration that claim 19 and its dependent claims were invalid, but the other claims were valid. He also ordered that Amgen should serve on TKT an application notice annexing a copy of the specification of the patent in the form in which it was proposed to amend it. The order gave directions for resolution of the amendment proceedings and to enable the court to decide whether relief should be granted under section 63 of the Patents Act 1977. The order also adjourned to a later date other disputes. Pursuant to that order, on 18th May 2001 Amgen served an application to amend the patent so as to delete claim 19 and its dependent claims.
6. The other disputes came back before the court and were resolved by the judge in his judgment and order of 8th August 2001. The appeal against part of that order raises distinct issues.
7. On 10th and 11th September there was another hearing to decide the form of the injunction, the issues on delivery up and the terms upon which the injunction would be stayed. They were resolved by the judge in his judgment and order of 13th September 2001.
8. There followed in February 2002 the hearing of the application to amend. In his judgment of 21st March 2002, the judge decided to give Amgen permission to amend by deleting claims 19 to 25. The order made on 18th April 2002 gave permission to amend the patent and also contained a proviso that Amgen “need not secure such amendments from the Comptroller of Patents pending an appeal on the issue of the validity …”. It also declared that the specification of the patent was framed in good faith and with reasonable skill and knowledge. The result being that the requirements of section 63 of the 1977 Act were satisfied and relief by way of damages and costs became possible. The order went on to deal with damages or an account and costs. TKT appeal against the declaration that the patent was framed in good faith and with reasonable skill and knowledge.

The Main Appeal

9. The patent was filed on 12th December 1984. It is entitled “Production of erythropoietin” which we will refer to as EPO. EPO is a protein that functions in the blood to regulate the production of red blood cells. They transport oxygen from the lungs into all the body tissues. Under normal circumstances only minute quantities of EPO are produced because red blood cells last for about 120 days in the circulation. In humans EPO is produced by certain cells in the kidney. When the level of oxygen in the blood drops, those kidney cells respond by producing more EPO and releasing it into the blood. Circulating EPO then binds to other proteins on the surface of certain cells in the bone marrow which divide and ultimately become red blood cells. When the oxygen carrying capacity of the blood returns to normal EPO production by the kidney is reduced.
10. The amount of EPO present in the blood is minuscule. Although EPO can be detected in blood, it was never purified from this source. In 1977 EPO was reported by Miyake et al. to have been purified in small amounts from urine from patients with aplastic anaemia. In fact only a few milligrams of urinary EPO were obtained from some 660 gallons of urine.
11. Amgen did not discover EPO, but the work carried out by Dr Lin (the inventor) and his aides, which forms the basis of the patent, identified the amino acid sequence of EPO, and identified and located the nucleotide sequences that coded for human EPO. From that it was possible, by using known genetic engineering techniques, to produce substantial amounts of EPO.
12. The skilled addressee of the patent would have been familiar with the basics of DNA and the techniques of recombinant technology. They have been set out in a number of judgments e.g. Genentech Inc’s Patent [1989] RPC 147 at 158, but for convenience we adopt the clear explanation given by the judge of that technology as applied to the discovery and production of EPO.

“41. EPO is a protein. Proteins are molecules which carry out many vital biochemical functions. They consist of one or more chains of polypeptides. (EPO only has one polypeptide chain, so I shall ignore proteins with more than one chain). A polypeptide is a chain of amino acids. There are twenty different amino acids, and some of them are more common than others. Each amino acid has a carboxyl group (-COOH) at one end and an amino group (-NH₂) at the other end. Each amino acid can be written:

H

|

NH₂—C—COOH

|

R

where R is its specific constituent, which varies from one amino acid to another. This specific constituent is known as the side chain of the amino acid, because, when the amino acid is in a polypeptide, this constituent is, in effect, a branch off the polypeptide chain.

42. In a polypeptide, amino acids are linked by a peptide bond, which is formed with the loss of an -OH from the carboxyl group of one amino acid and an -H from the amino group of the other acid, i.e. with the effective loss of a water molecule. Thus, the simplest polypeptide (i.e. one with two amino acids) has one peptide bond and may be written:

H O H

| || |

NH₂—C—C—N—C—COOH

| | |

R₁ H R₂

Strictly each amino acid component in a polypeptide or protein is referred to as an amino acid residue, but a residue is sometimes called simply an amino acid.

43. Many side chains are subjected to modification during the process of their formation. An important and relevant example of such modification is the addition of saccharides (or sugar residues), a process known as glycosylation.

44. While glycosylation does not normally occur in bacteria, it does occur in higher organisms. Glycosylation can involve addition of a single sugar residue, or a substantial number of sugar residues. Such residues can substantially vary in size, structure, and composition one from the other. There are two common types of glycosylation, N-glycosylation and O-glycosylation. The reference to N and O is a reference to the atom (nitrogen and oxygen respectively) to which the carbohydrate residue (known as the “glycan”) attaches. A glycosylated protein is known as a glycoprotein. EPO is in a glycoprotein.

45. In a polypeptide, the amino acid chain is treated as starting at the amino end of the first amino acid, known as the N-terminus or NH₂-terminus, and ending with a carboxyl group of the last amino acid, known as the C-terminus or COOH-terminus. A polypeptide only becomes a protein once it is folded into the correct shape, often with cross linking between cysteine amino acids known as disulphide bridges. If a protein loses its correct shape, it ceases to be effective as a protein: it is denatured. A protein also has some internal sequences - known as active sites - which are more important for its biological functions than other such sequences.

46. Although a given protein has a specific sequence of amino acids in its polypeptide chain, a change, or indeed a number of changes, in the identity of the amino acids in a protein may not destroy its effectiveness. A single change might reduce or even destroy a protein’s effectiveness; another change or other changes might have no effect on, or actually improve, its effectiveness; sometimes a change can result in side effects. Variants of protein where amino acid residues have been changed (or even removed) without the effectiveness of the protein being lost are also known as analogues of the protein.

47. In general the greater the number of changes in the amino acids in a given protein the more likely it would be that its effectiveness will be diminished or will cease. Further, as I have mentioned, some of the amino acid sequences in a protein were of greater importance than others. Accordingly, changes, even a number of changes, in the amino acids in less significant locations might have little, if any effect; whereas a single change in a more sensitive area, in particular the protein’s active site or (if more than one) its active sites, could destroy the effectiveness of the protein. The characteristics of amino acids vary, so a change from one amino acid to another with similar characteristics might be expected to be less significant than a change to an amino acid with different characteristics.

The genetic code

48. In order to make (or “express”) a protein, the amino acids have to be assembled in the correct sequence. The sequence for a particular protein is specified by a gene which is present in the genetic material of a cell - i.e. within the DNA in the cell. The “genetic code” is the relationship between a sequence of genetic material and the amino acid sequence of the corresponding protein.

49. All living organisms are composed of cells. Eukaryotic cells have a nucleus containing the cell’s genetic material; the nucleus is surrounded by a cytoplasm which is in turn bordered by the cell membrane. Prokaryotic cells have a single cellular compartment bounded by a cell membrane and a cell wall. Mammals (including humans) plants and fungi are made up of eukaryotic cells; bacteria are made up of prokaryotic cells.

50. The genetic material in a cell is made from deoxyribonucleic acid, DNA, which consists of two strands of complexes called nucleotides. Each nucleotide includes a nitrogenous base, a phosphate and a sugar. DNA contains four different types of nucleotide, each of which is characterised by a different nitrogenous base, namely Adenine, Guanine, Cytosine and Thymine, abbreviated respectively to A, G, C and T. The nucleotide bases in DNA are linked by bonds between a carbon atom of one nucleotide, known as a 3’-carbon, and another carbon atom, known as the 5’-carbon, of the neighbouring nucleotide. By convention a strand of DNA begins with the 5’end and finishes with the 3’end. The 5’end is “upstream” (to the left of the DNA sequence as written out) and the 3’end is “downstream” (to the right).

51. In its normal state, DNA consists of two strands of nucleotides in the well-known “double helix” configuration. Each of the two strands of DNA is complementary - i.e. linked to the other strand - by the invariable rule that A pairs with T, and G pairs with C. In eukaryotic cells, the double helix DNA is divided into a number of different chromosomes, which are compact structures, in which the DNA is complexed with many proteins.

52. Cells divide and duplicate their genetic material. Accordingly, on division of a cell, each DNA molecule becomes two identical molecules each carrying the same information as was in the original. This process, known as replication, involves the two strands of DNA separating so that each strand acts as a template, with new nucleotides being added one by one on the basis already mentioned, namely A pairing with T, and G with C.

53. The order of the nucleotides in a particular segment of DNA - called a gene - provides the blueprint for the “expression” of a particular protein. In effect, the DNA blueprint is a succession of code words, each consisting of a set of three nucleotides, known as codons, in the gene. Each codon (i.e. each set of three nucleotides) specifies a particular amino acid. With three exceptions, a particular codon instructs the cellular machinery to add a particular amino acid to a growing polypeptide chain. The three exceptions are “stop codons” which instruct that machinery where to stop the manufacture of the protein.

54. As there are twenty different amino acids and four different nucleotide bases, once one excludes the three stop codons, there are 61 possible permutations of bases to encode 20 amino acids. Two amino acids are encoded by only one group of three bases, i.e. by only one codon. Some amino acids are encoded by as many as six different codons. The majority of amino acids are encoded by two or four codons.

55. This feature of the genetic code, namely that (with two exceptions) each amino acid can be encoded by more than one codon, is known as degeneracy. As a result of degeneracy, once a protein’s amino acid sequence is known, many (normally many millions or billions) of different possible gene sequences could theoretically encode that protein. However, only very few may in fact do so.

Expression of proteins

56. The cells of a given organism almost all contain the same DNA, namely the same genetic material which, according to the evidence, can encode for tens of thousands of different proteins (“the genome”). However, each particular type of cell produces only a few of the proteins encoded by the DNA. Thus virtually all human cells contain DNA encoding for insulin, but only pancreatic cells actually produce that protein. In eukaryotic cells, as I have mentioned, DNA is divided into a number of different chromosomes, in which the DNA is combined with certain proteins. DNA sequences in a particular cell may, depending on their chromosomal regions, be “open” or “closed”. If “open” (like the insulin gene in pancreatic cells), the sequence is capable of expression; if “closed”, it is not. The EPO gene is included in the DNA in Chromosome 7.

57. When a cell responsible for the production of a particular protein detects the need for that protein, the relevant segment of DNA that codes for that protein (i.e. the relevant gene) produces a substance related to DNA, called ribonucleic acid, RNA. This is made from a slightly different nucleic acid from DNA, and consists of a single strand rather than a double strand. Further, in place of one of DNA’s bases, Thymine, it has a different base, Uracil: it contains U instead of T. The RNA manufactured in the cell is known as mRNA - i.e. Messenger RNA. mRNA is made by a process called transcription. Transcription involves an enzyme called RNA polymerase travelling along the relevant gene making the single stranded RNA with bases complementary to the bases on the DNA. Thus, A is reproduced as U (there being no T in RNA), G is reproduced as C, T is reproduced as A, and C is reproduced as G.

58. In higher organisms (i.e. those with eukaryotic cells), structural genes have non-coding sequences of bases, called “introns”, between the portions, called “exons”, which correspond to the portions of the mRNA which encode the protein. The introns are effectively edited out before translation of the mRNA, leaving only those regions of codons in the mRNA which are protein encoding, i.e. the exons. This process, which involves the removal of the segments which do not encode for the protein (most notably the introns) and joining the encoding exons together, is known as splicing. The splicing process is determined by sequences near the exon-intron junctions and the intron-exon junctions in the RNA; these sequences are known respectively as splice donor sites and splice acceptor sites.

59. Once the transcription of the mRNA is complete, its code must be “translated” to the amino acid sequence. For this purpose, the RNA is released from the cell cytoplasm (in the case of a eukaryotic cell). Translation takes place in a complex known as the ribosome, a microscopic structure in the cell. The ribosome effectively reads off the base sequence of the whole of the mRNA (starting at the 5’end with a start codon) and translates it into polypeptide by use of another type of RNA called transfer RNA, tRNA. This translation ends when a stop codon is reached. The tRNA contains the appropriate codons for the amino acids of the particular protein, and is joined by specific enzymes to the amino acid which corresponds to the particular codon. These amino acids are expressed and then joined by peptide bonds, with the assistance of an enzyme called peptidyl transferase.

60. Genes from higher organisms also have a non-coding regulatory region before the encoding regions (i.e. upstream of the first encoding region) and another non-encoding region downstream of the last exon. The start of the gene, i.e. where the RNA polymerase starts transcribing, is known as the cap site. The regulatory region controls the transcriptional activity of the gene. The promoter is that portion of the regulatory region which initiates the transcription and is upstream of the cap site. The promoter is thus upstream of the encoding region of the gene, and it provides a binding site for RNA polymerase, the enzyme which transcribes the gene into mRNA. In fact, there are sometimes two promoter sites in a particular gene. Immediately before the start of the protein encoding region is a series of nucleotides known as the Kozak sequence, which, in effect, helps the ribosome to recognise when translation should start. Dr Kozak found that certain sequences worked better than others for this purpose.

61. The encoding region of the gene which actually transcribes is also called the structural gene or the reading frame. Failure to translate the mRNA precisely from the correct reading frame (i.e. the correct triplet of nucleotides) results in failure to produce the desired protein. The first codon, known as the initiation codon, of the first encoding exon in a gene is almost always ATG (or AUG in the corresponding RNA). Immediately after the last encoding exon of the gene is the terminator sequence which instructs the RNA polymerase to stop copying the DNA. There then follows the downstream non-coding region, which almost always ends with a number of consecutive adenines - the so called “poly A tail”.

62. Chemical modifications may occur to proteins immediately after, or even during, their expression. Thus, after expression, mammalian proteins are normally glycosylated - i.e. sugar or saccharide sub-units are chemically attached to some side chains of the protein. Glycosylation can enable a protein to persist longer in the bloodstream. Further, after (or during) translation, certain amino acids may be cleaved from the protein. In many cases, the first several amino acids at the N-terminus of the protein serve as a chemical signal directing the protein into the first of a series of membrane-enclosed compartments within the cell. These initial amino acids are often referred to as a leader peptide. The leader peptide, which marks the protein for secretion from the cell, causes the protein to enter the secretory pathway, which is a series of membrane-enclosed compartments within the cytoplasm of the cell.

63. The first of the compartments within the secretory pathway is called the rough endoplasmic reticulum (“RER”). As a protein produced with a leader peptide emerges from the ribosome, it directs the ribosome to bind to the RER so that the growing protein is inserted through the membrane of the RER into its internal space. The leader peptide sequence may then be cleaved from the protein (which therefore becomes “mature”), and the protein remains in the central cavity of the RER. It is also within the RER that initial glycosylation occurs. Then, the protein moves to another array of membrane-enclosed compartments, called the Golgi apparatus (“the Golgi”), where certain modifications to the sugar chains take place. Finally, the protein is secreted by the cell and travels through the body to act on other cells.

….

71. II. RECOMBINANT PRODUCTION OF PROTEINS

From the 1920s, some proteins were obtained by purifying them from human or animal tissues or from bodily fluids. During the 1970s, recombinant DNA technology was developed. This is a technique which involves isolating or synthesising the gene which codes for the desired protein, combining the gene with other (“vector”) DNA, inserting this “recombinant” DNA into a “host cell”, which then expresses the protein.

72. One reason for preparing recombinant DNA is to produce - or to “express” - large quantities of the protein encoded by a particular gene. In order to express a particular protein using recombinant DNA, one needs enough information regarding the gene encoding the protein, so that additional vector DNA segments can be recombined with the gene to effect protein expression. This requires the relevant gene to be isolated, identified and characterised. The difficulty of this exercise will be appreciated from the fact that the amount of DNA present in each cell of a human (and most higher animals) amounts to around 6,000,000,000 nucleotides in two complementary strands of DNA (i.e. 3,000,000,000 “base pairs”). This is known as the genome, and the evidence in this case suggested that it comprises over 50,000 genes (although very recently reported research appears to suggest that this may be somewhat on the high side).

73. A frequent starting point is the selection of a suitable library of genetic information and a set of “probes” with which to screen a “library” consisting of strands of DNA. There are two normal types of libraries that can be used for these purposes: genomic DNA libraries and complementary DNA ("cDNA") libraries. A library consists of fragments of DNA resulting from cleaving the DNA at certain points. The identity of the fragments of DNA will depend of course on the source of the DNA.

74. Ideally, a genomic library should contain a complete set of all the DNA sequences present in the genome, i.e. it will contain all the genetic information present in the cells from which the library was made. However, there is no guarantee that every part of the genome will find its way into the library. A genomic library is stored as pieces of the DNA in a suitable set of carriers. Complementary DNA is so called because it is the complement of the mRNA. Unlike genomic DNA, cDNA does not contain introns or promoters and rarely contains repetitive sequences. Further, only a sub-set of the genomic sequences is expressed in any one cell-type or tissue, and a cDNA library can only contain DNA sequences derived from reverse transcription of mRNA of these expressed sequences.

75. As at the priority date, a genomic DNA library of repute was the Lawn library disclosed in 1978 by Lawn et al in Cell 15:1174. Its use was described in the standard work in the field, the so-called Maniatis Manual (Molecular Cloning, A Laboratory Manual, by Maniatis, Fritsch and Sambrook 1982, especially in Chapter 7). While it was not clear that the Lawn library contained all the genes, there was a high probability that the gene one was searching for would be present. By 1983 the technique of Okayama and Berg (Mol.Cell Biol 2(2):161) was also widely known and accepted as a method for constructing cDNA libraries.

76. Researchers construct probes in order to screen libraries based on information concerning the actual or putative amino acid sequence of the protein of interest. Probes are typically fairly short sequences of DNA nucleotides, called oligonucleotides, made synthetically. They are constructed with bases which are complementary to the bases of the targeted gene according to the base pairing rules (A with T, C with G) so that the probes will bind - or hybridise - to the gene in the same way that the two complementary strands of the DNA molecule bind to each other in the cell. “Screening” is the process of finding a desired clone from within a population in a library. Such a screening exercise is likely to be futile if the probes are based on unreliable amino acid sequencing of the relevant part of the protein of interest.

77. A significant problem with genomic libraries in 1983 was that there was no index or map leading researchers to a desired clone or gene: as explained above, each gene comprises a tiny fraction of the overall DNA, or genome. Another problem that compounded the difficulty of using such libraries is that the DNA was fragmented in a random manner. Thus, while part of the sequence of a gene of interest may be contained in a single fragment of DNA, the complete sequence may very well be split between two or even more fragments, and those fragments might well contain sequences which were not part of the gene concerned.

78. If the targeted gene (or a significant part of it) has previously been isolated, identified and sequenced, the probing exercise is normally relatively easy. One designs a probe which is an exact complement to that gene (or the part). Otherwise knowledge of the amino acid sequence of at least a fragment of the targeted protein is essential for determining the constitution of the probe. Armed with that knowledge, probes may be designed and used to screen the library in an attempt to find the gene which codes for that sequence. Probes are designed by determining which codons code for the amino acid sequence which is being targeted. However, this process is greatly complicated by the existence of degeneracy.

79. The targeted sequence will inevitably include amino acids which are coded for by more than one codon so, in order to be sure of having a sequence which is an exact match to the gene of interest, one has to screen the library with a set of probes containing multiple nucleotide sequences each of which will be different. A "fully degenerate" set of probes covers all possible sequences coding for the targeted amino acid sequence. As a matter of logic, the longer the amino acid sequence selected for targeting, the greater the number of probes required. The probes are normally "labelled" with a radioactive atom so they can be traced.

80. The library, which is to be screened, is first "plated out" so that one can distinguish individual clones among the potentially millions of clones in the library. The DNA in the individual clones is denatured, i.e., the double-stranded helix structure is disrupted, making it single-stranded. The probes are then reacted with the denatured colonies or plaques in the two types of libraries. The plates are washed to remove the surplus probe material. They are then inspected to see whether any of the probes have stuck ("hybridised") to any parts of the library - i.e. whether there are any "positives". Positives may be missed if the background "noise" is too high. If a probe hybridises to one of the clones in the library, then the clone is sequenced to determine whether it contains part or all of the gene of interest. The ability to confirm the sequence is limited by the information known concerning the gene and/or the corresponding protein.

81. Screening a genomic DNA library is complicated by the fact that the probe for the selected amino acid sequence may target parts of the exon portions of the gene that are interrupted by an intron. Such a probe, described as "spanning an intron", might not hybridise to the clones in the library; thus it would not enable identification of the desired gene. Until the gene has been cloned (i.e. isolated, identified and sequenced), the existence, number and location of any introns are usually unknown. Screening a cDNA library also has problems. For the library to be useful, the cell or tissue from which it has been obtained needs to have been expressing a sufficient amount of the mRNA of interest at the time when the cell or tissue was harvested. If the cell or tissue was not expressing sufficient amounts of the mRNA of interest, or was expressing no such mRNA, there will not have been enough or any "message", hence there will be insufficient or nothing appropriate from which the complementary DNA can be derived.

82. “False positives" can result where a probe matches or hybridises to a sequence outside of the region of interest. Given the size of the DNA libraries (especially genomic DNA libraries) and the fact that there are only four bases in DNA, it is possible that a selected sequence may be repeated several times in the genome. The longer the probe the less likely that will be. However, longer probes may show certain sequences which would also lead to false positives.

83. The conditions in which the probing takes place can vary. The more stringent the conditions the more difficult it is for the probes to hybridise with the DNA strands. Accordingly, high stringency conditions have the advantage of reducing the potential for false positives: however, they have the disadvantage of increasing the risk of there being no positives. The stringency of the hybridisation conditions is normally related to the temperature and the salt concentration of the solution in which the hybridisation exercise is being carried out. A technique often used is to increase the stringency gradually in a particular experiment, so that a number of probes hybridise initially, and thereafter, as the stringency increases, the number of hybridising probes decreases, ending ideally with only one probe hybridising, namely the gene one is looking for.

84. After hybridising, the resultant material is often subjected to “washing” which involves subjecting it to somewhat higher stringency conditions in a solution. This is with a view to washing off all or some false positives which still remain. Once a particular gene has been isolated, identified and characterised, there are various ways in which that information can be utilised to form recombinant DNA, that will enable a cell that contains the gene to produce greater quantities of the desired protein.

85. While the basic processes of transcription and translation are common to all cells, not all types of cell are appropriate as host cells. Bacteria cells cannot deal with introns and therefore cannot deal with genomic DNA (as opposed to cDNA). Further, bacteria do not glycosylate proteins; accordingly, proteins, such as EPO, which require glycosylation in order to be fully effective, need to be produced in eukaryotic cells. Even in different eukaryotic cells, the same protein may be glycosylated differently.

86. In 1984 the types of mammalian cells which were thought to be suitable host cells was limited. Those in conventional use included the Chinese Hamster Ovary (“CHO”) cell, the baby hamster kidney (“BHK”) cell, and the COS monkey (“COS”) cell, as well as certain types of human cells.

87. Recombinant DNA techniques

There are a number of techniques for introducing DNA into a cell (which, in 1983, would normally have been a bacterium or a yeast), with a view to expressing a gene which it would not ordinarily express. The process of introducing DNA into a cell is called transformation or transfection. If a transformed cell is to express quantities of a desired gene, the relevant DNA which has been introduced must replicate as the cell replicates. Isolated fragments of DNA do not, in general, replicate in a bacterium; this is why one inserts the relevant DNA into another piece of DNA (the vector) which replicates in the cell. Suitable vectors are the DNAs of bacterial viruses or bacteriophages (“phages”) which naturally infect bacteria and replicate within them. Alternatively, bacterial plasmids (circular pieces of DNA capable of self-replication) can be used.

88. In order to manufacture this recombinant DNA, the vector DNA and the DNA of interest are first both cleaved. The two sets of fragments are then incubated to join the fragments to form recombinant DNA. These recombinant DNA techniques were and are often carried out, in E.coli cells. A single transformed phage cell gives rise to a plaque. Within a particular plaque each phage particle is genetically identical. Such a group of genetically identical organisms is called a clone, and the process of creating a colony of cells which all contain the same inserted DNA is called gene cloning.

89. Two principle techniques in gene cloning are (a) cutting (cleavage) and joining (ligation) of nucleic acids, and (b) synthesising oligo- and polynucleotides.

90. The discovery of restriction enzymes rendered possible the cleavage of a double-stranded DNA molecule into discrete gene fragments, thereby enabling gene manipulation. Restriction enzymes catalyse the breaking of a specific type of bond between adjacent nucleotides at particular sites (“restriction sites”) within a DNA molecule. They are sometimes called restriction endonucleases, distinguishing them from exonucleases, which catalyse the breaking of the bond between the last nucleotide in a nucleic acid chain and the remainder of the nucleic acid.

91. Restriction endonucleases are highly specific. Most of them are of so called Type II; they cleave DNA molecules only in regions where particular sequences (usually of from 4 to 6 nucleotides) are present. The resultant DNA fragments can associate by hydrogen bonding with other DNA fragments. If produced by some enzymes, they can have overlapping 5' and/or 3’ends; for this reason the fragments are said to have “sticky ends”. DNA fragments from diverse sources can be joined by means of such cohesive ends. Other enzymes make even cuts giving rise to so-called “blunt-ended” fragments with no cohesive ends at all. Some enzymes recognise tetranucleotide sequences, while others recognise longer sequences, and this has an effect on the average fragment length produced. HindII, apparently the first Type II restriction endonuclease to be discovered, is an example of an enzyme which recognises more than one sequence.

92. If a restriction endonuclease is used to produce fragments with sticky ends, these fragments will be capable of associating with fragments produced by the action of the same restriction endonuclease on a different DNA molecule. They will then be able to ligate. When the sticky ends associate, the join has “nicks” in the backbone of the DNA chains a few base pairs apart in opposite strands. The E.coli bacterium has an enzyme, DNA ligase, which can be used to repair these nicks in the sugar-phosphate backbone of DNA chains to form an intact double-strand.

93. A problem in the development of cloning techniques was the dependence on the availability of restriction endonuclease cleavage sites in suitable positions. Although vector molecules were developed with restriction sites at specific and useful locations, the position of sites in the DNA to be cloned was effectively random. In 1976, a method was developed to enable any double stranded DNA molecule generated by a particular restriction enzyme to be inserted into a cloning vector at a non-matching restriction endonuclease site. This involved the chemical synthesis of a short oligonucleotide containing the desired specific restriction endonuclease recognition site, known as a linker. Such linkers could be joined by DNA ligase onto the DNA to be inserted and the resulting molecule cut with a restriction enzyme to generate sticky ends. The molecule could then be inserted into the matching restriction site in the cloning vector.

94. As I have mentioned, in the regulatory region upstream of the coding sequence of a gene there is a promoter. In gene cloning, an artificial promoter may be inserted into the regulatory region (or even upstream of that region). In 1983, specific viral promoters which were generally effective for production of recombinant DNA in E.coli were well known; indeed, they were commercially available. It was generally considered that insertion of the new promoter should occur as near the initiation codon (ATG) as possible, and that it should be accompanied by removal of the natural promoter.

95. The resultant “recombinant” DNA is inserted - or “transfected” - into the host cell where it integrates with the native DNA essentially at random. In a eukaryotic cell, the chromosomal structure around the recombinant gene will influence whether it expresses, and, if so to what extent, just as with a naturally occurring gene. By 1983, it was known that the level of protein expression depended on the number of copies of the relevant gene present in the cell and the expression level of those genes. There were also techniques available to select cells which contained multiple - or “amplified” - copies of a particular gene.

96. Indeed, DNA amplification can occur spontaneously in many lines of mammalian cells. When it occurs as a desirable event, those cells that contain amplified copies of DNA can be selected by various means from those that do not. One means of selection of cells that have amplified DNA involves the use of the gene that directs the synthesis of dihydrofolate reductase (“DHFR”), a protein normally produced in cells because it is required by the cell to synthesise the nucleotides for DNA. DHFR can be blocked by a cancer drug called methotrexate (“MTX”); this inhibition will kill the cell unless the cell can avoid the inhibition. One way the cell can avoid the inhibition by MTX is to produce more DHFR, so that it escapes the MTX inhibition.

97. Amplification normally confers no advantage to the cell, and it is not stable. However, in the presence of MTX, amplification of the DHFR gene means that the amplified cells can produce more DHFR, and so are better able to survive. The cells in which no DHFR gene amplification occurs (“DHFR-cells”) die. Therefore, in the presence of MTX, the cells that have spontaneously amplified the DHFR gene (“DHFR+ cells”) are selected for survival. The more DHFR made by the cell, the higher the concentration of MTX in which the cells can grow.

98. During amplification, the amplified DNA segment usually includes more than the DHFR gene. If another gene is within the amplified region of DNA, it will therefore be amplified with the DHFR gene. Such gene segments that are amplified along with a selectable gene, such as DHFR, are called passenger genes. To take advantage of the ability of cells to amplify passenger genes, a DHFR gene is linked recombinantly to the gene which is to be amplified. Cells containing this recombinant DNA are then treated with MTX, resulting in some cells surviving due to the presence of amplified DHFR genes. In addition to increased numbers of the DHFR gene, these cells also contain proportionately increased numbers of the passenger gene. As a result, these amplified cells will make increased amounts of the protein encoded by the passenger gene.

99. Synthesising DNA developed during the late 1970's, when several independent laboratories pioneered oligonucleotide synthesis. Various scientists reported the synthesis of specific genes in learned journals. At that time, chemical synthesis of oligonucleotides was carried out by solution-phase chemistry using a phosphodiester or phosphotriester method The solid-phase method then took over using phosphotriester or phosphoramidite chemistries, which although still a manual process, was a labour-reduced process. On the basis of the evidence, synthesis of a polynucleotide with up to 400 or 500 bases was possible in 1984, but only a limited number of groups of scientists could reliably achieve it.”

The Patent

13. The specification states at the outset that the invention “relates generally to the manipulation of genetic materials and, more particularly, to recombinant procedures making possible the production of polypeptides possessing part or all of the primary structural conformation [of EPO]”.
14. The specification continues with a summary of the background. That summary describes transcription and how the focus had for the last decade been on the attempt to manufacture industrially and pharmaceutically significant substances using organisms. The art, the specification states, is rich in patent and literature publications relating to “recombinant DNA” methodologies for the isolation, purification and amplification of genetic materials for use in transformation of selected host mechanisms. The specification examines such procedures by reference to certain articles. It concludes at page 4 line 34:

“There ... continues to exist a need in the art for improved methods for effecting the rapid and efficient isolation of cDNA clones in instances where little is known of the amino acid sequence of the polypeptide coded for and where “enriched” tissue sources of mRNA are not readily available for use in constructing cDNA libraries. Such improved methods would be especially useful if they were applicable to isolating mammalian genomic clones where sparse information is available concerning amino acid sequences of the polypeptide codes for the gene sought.”

15. The second section in the specification is headed “Erythropoietin As a Polypeptide of Interest”. It contains a description of EPO and its importance in the body and states “Because erythropoietin is essential in the process of red blood cell formation, the hormone has potential useful application in both the diagnosis and the treatment of blood disorders …”. The specification then points out the importance of EPO and goes on to explain what attempts had been made to obtain it. Prior attempts to isolate EPO in good yield from urine or plasma are said to have proven relatively unsuccessful. Other unsuccessful methods attempted were production of hybrid lymphoblastoid cells, immunological procedures and attempts to employ cell fusion and hybridization techniques to develop monoclonal antibodies to EPO for use in the isolation and quantitative detection of EPO. The specification states at page 6 line 57:

“While substantial efforts appear to have been made in attempted isolation of DNA sequences coding for human and other mammalian species erythropoietin, none appear to have been successful. This is due principally to the scarcity of tissue sources, especially human tissue sources, enriched in mRNA such as would allow for construction of a cDNA library from which a DNA sequence coding for erythropoietin might be isolated by conventional techniques. Further, so little is known of the continuous sequence of amino acid residues of erythropoietin that it is not possible to construct, e.g., long polynucleotide probes readily capable of reliable use in DNA/DNA hybridisation screening of cDNA and especially genomic DNA libraries. ...It is estimated that the human gene for erythropoietin may appear as a "single copy gene" within the human genome and, in any event, the genetic material coding for human erythropoietin is likely to constitute less than 0.00005% of total human genomic DNA which would be present in a genomic library.

To date, the most successful of known reported attempts at recombinant-related methods to provide DNA sequences suitable for use in microbial expression of isolatable quantities of mammalian erythropoietin have fallen far short of the goal.”

16. The specification proceeds to explain how the patentees achieved that goal. It starts by setting out the claims. It continues at page 9 line 31:

“Vertebrate (e.g., COS-1 and CHO) cells provided by the present invention comprise the first cells ever available which can be propagated in vitro continuously and which upon growth in culture are capable of producing in the medium of their growth in excess of 100U (preferably in excess of 500U and most preferably in excess of 1,000 to 5,000U) of erythropoietin per 10⁶ cells in 48 hours as determined by radioimmunoassay.

Also provided by the present invention are synthetic polypeptides wholly or partially duplicative of continuous sequences of erythropoietin amino acid residues which are herein for the first time elucidated. These sequences, by virtue of sharing primary or tertiary structural and conformational characteristics with naturally-occurring erthyropoietin may possess biological activity and/or immunological properties in common with the naturally-occurring product such that they may be employed as biologically active or immunological substitutes for erythropoietin in therapeutic and immunological processes.

Illustrating the present invention are cloned DNA sequences of monkey and human species origins and polypeptide sequences suitably deduced therefrom which represent, respectively, the primary structural conformation of erythropoietins of monkey and human species origin.”

17. The specification then points out that having elucidated the EPO sequence, the invention provides for the total and/or partial manufacture of the DNA sequence coding for EPO, including such advantageous characteristics as incorporation of codons “preferred” for expression by selected certain hosts, provision of sites for cleavage and additional initial, terminal or intermediate DNA sequences.
18. There follows a detailed description of the invention with the aid of examples. Examples 1 to 3 describe the isolation of EPO from human and monkey cells and the way that the gene was sequenced. The results are set out in Table V (monkey) and Table VI (human). The sequences disclosed include the leader and promoter sequences and sites for potential glycosylation. There follow examples showing the initial EPO expression using COS-1 cells. The best result reported was 302 mU/ml for a five day growth sample and 567 mu/ml for a seven day growth sample. Expression using Chinese hamster ovary (CHO) DHFR- cells gave a better result. A 30 day sample contained 3089 ± 129 U/ml. Other examples describe methods of synthesis of EPO using the sequence in Tables V and VI. However the specification makes it clear that the examples are not meant to be limiting on the scope of the invention which is said to embrace other EPO products such as polypeptide analogues of EPO in addition to naturally occurring allelic forms of mature EPO. It states at page 47 line 27:

“In addition to naturally-occurring allelic forms of mature EPO, the present invention also embraces other “EPO products” such as polypeptide analogs of EPO and fragments of “mature” EPO.”

19. It continues at page 47 line 53:

“….the cloned DNA sequences described herein which encode human and monkey EPO polypeptides are conspicuously valuable for the information which they provide concerning the amino acid sequence of mammalian erythropoietin which has heretofore been unavailable despite decades of analytical processing of isolates of naturally-occurring products.”

20. Other uses and benefits of the sequence are given at page 48 line 3.

“DNA sequences of the invention are also conspicuously suitable materials for use as labelled probes in isolating EPO and related protein encoding cDNA and genomic DNA sequences of mammalian species other than human and monkey species herein specifically illustrated. The extent to which DNA sequences of the invention will have use in various alternative methods of protein synthesis (e.g., in insect cells) or in genetic therapy in humans and other mammals cannot yet be calculated. DNA sequences of the invention are expected to be useful in developing transgenic mammalian species which may serve as eukaryotic "hosts" for production of erythropoietin and erythropoietin products in quantity. See, generally, Palmiter et al., Science, 222 (4625), 809-814 (1983).

Viewed in this light, therefore, the specific disclosures of the illustrative examples are clearly not intended to be limiting upon the scope of the present invention and numerous modifications and variations are expected to occur to those skilled in the art. As one example, while DNA sequences provided by the illustrative examples include monkey cDNA and genomic DNA sequences, because this application provides amino acid sequence information essential to manufacture of DNA sequence, the invention also comprehends such manufactured DNA sequences as may be constructed based on knowledge of EPO amino acid sequences. These may code for EPO (as in Example 12) as well as for EPO fragments and EPO polypeptide analogs (i.e., "EPO Products") which may share one or more biological properties of naturally-occurring EPO but not share others (or possess others to different degrees).

In a like manner, while the above examples illustrate the invention of microbial expression of EPO products in the context of mammalian cell expression of DNA inserted in a hybrid vector of bacterial plasmid and viral genomic origins, a wide variety of expression systems are within the contemplation of the invention. Conspicuously comprehended are expression systems involving vectors of homogeneous origins applied to a variety of bacterial, yeast and mammalian cells in culture as well as to expression systems not involving vectors (such as calcium phosphate transfection of cells). In this regard, it will be understood that expression of, e.g., monkey origin DNA in monkey host cells in culture and human host cells in culture, actually constitute instances of "exogenous" DNA expression inasmuch as the EPO DNA whose high level expression is sought would not have its origins in the genome of the host. Expression systems of the invention further contemplate these practices resulting in cytoplasmic formation of EPO products and accumulation of glycosylated and non-glycosylated EPO products in host cell cytoplasm or membranes (e.g. accumulation in bacterial periplasmic spaces) or in culture medium supernatants as above illustrated, or in rather uncommon systems such as P.aeruginosa expression systems (described in Gray, et al.. Biotechnology, 2, pp. 161-165 (1984).”

21. The relevant claims are as follows:

“1. A DNA sequence for use in securing expression in a prokaryotic or eukaryotic host cell of a polypeptide product having at least part of the primary structural confirmation [sic] of that of erythropoietin to allow possession of the biological property of causing bone marrow cells to increase production of reticulocytes and red blood cells and to increase hemoglobin [sic] synthesis or iron uptake, said DNA sequence selected from the group consisting of:

(a) the DNA sequences set out in Tables V and VI or their complementary strands;

(b) DNA sequences which hybridise under stringent conditions to the protein coding regions of the DNA sequences defined in (a) or fragments thereof; and

(c) DNA sequences which, but for the degeneracy of the genetic code, would hybridise to the DNA sequences defined in (a) and (b).

….

19. A recombinant polypeptide having part or all of the primary structural conformation of human or monkey erythropoietin as set forth in Table VI or Table V or any allelic variant or derivative thereof possessing the biological property of causing bone marrow cells to increase production of reticulocytes and red blood cells to increase hemoglobin synthesis or iron uptake and characterized by being the product of eukaryotic expression of an exogenous DNA sequence and which has higher molecular weight by SDS-PAGE from erythropoietin isolated from urinary sources.

20. A glycoprotein polypeptide according to claim 19 having an average carbohydrate composition which differs from that of human erythropoietin isolated from urinary sources.

26. A polypeptide product of the expression in a eukaryotic host cell of a DNA sequence according to any of Claims 1, 2, 3, 5, 6 and 7.

27. A process for production of a polypeptide having at least part of the primary structural conformation of erythropoietin to allow possession of the biological property of causing bone marrow cells to increase production of reticulocytes and red blood cells and to increase hemoglobin synthesis or iron uptake, which process is characterized by culturing under suitable nutrient conditions a prokaryotic or eukaryotic host cell transformed or transfected with a DNA sequence according to any of Claims 1, 2, 3, 5, 6 and 7 in a manner allowing the host cell to express said polypeptide; and optionally isolating the desired polypeptide product of the expression of the DNA sequence.

28. A process according to Claim 27, characterised by culturing a host cell of any one of Claims 12 to 16 [which refer back to Claim 1].

29. A process according to Claim 27 or 28 for production of a polypeptide of any one of Claims 19 to 23 and 26.

30. A pharmaceutical composition comprising a polypeptide produced in accordance with the process of Claim 27, 28 or 29 and a pharmaceutically acceptable diluent, adjuvant or carrier.

…”

22. There is no dispute that in 1984 that the isolation of the EPO gene needed skilled work involving invention. It was at the heart of the invention as patented. That the judge stated in a number of paragraphs. It is sufficient to cite just one.

“623. … The essence of the invention embodied in 605, its contribution to the art or what one might call its inventive concept, is the disclosure encapsulated in Table VI, which contains the whole of the encoding sequences, the whole of the intervening introns, and a large proportion of the upstream and downstream sequences. It enabled that to be done which was previously impossible, namely the production of EPO in accordance with biotechnological methods, as they existed at the relevant date as they would have been expected to develop and improve over the ensuing years. ...”

Construction

23. This appeal is concerned with the way that the work of the inventor is patented. The form of the claims reflects the requirements of an invention under the European Patent Convention (EPC) which are reflected in section 1 of the Patents Act 1977. It is not possible to obtain a patent for the discovery of a gene which by definition is found in the human body. Thus construction of the claims needs to be approached with that in mind. The position has now been made clear. The Directive 98/44/EC provides in Article 5:

“1. The human body, at the various stages of its formation and development, and the simple discovery of one of its elements, including the sequence or partial sequence of a gene, cannot constitute patentable inventions.

2. An element isolated from the human body or otherwise produced by means of a technical process, including the sequence or partial sequence of a gene, may constitute a patentable invention, even if the structure of that element is identical to that of a natural element.

3. The industrial application of a sequence or a partial sequence of a gene must be disclosed in the patent application.”

24. The judge construed the ambit of the claims before deciding the issues of infringement and validity. That approach was consistent with practice as the ambit of the claim must be the same whatever the issue. With the assistance of his judgment, we have decided, with one exception, to decide the issues of construction in the context where they arise. When doing that we will apply the Protocol on Interpretation and the guidance given in Wheatley v Drillsafe Ltd [2001] RPC 133.
25. The exception concerns claims 19 and 26. We will concentrate on claim 26. The judge construed it as limited to products made by a process involving what he called a process of expression of a claim 1 sequence. Mr Kitchin QC, who appeared with Mr Meade and Miss Lane for TKT, submitted that in doing so the judge fell into error. He submitted that the claim monopolised products which had the features of products produced by the process of expression of a claim 1 sequence, but it was not limited to products made by such a process.
26. Mr Kitchin’s submissions started from a consideration of the EPC which is reflected in the Patents Act 1977. To be valid a claimed invention must be new in the sense that it must not have been made available to the public (sections 1 and 2 of the Act). The claims must be clear and concise (section 14). Infringement is defined as doing an act when either the invention is a product or a process (section 60). From that background Mr Kitchin proceeded to a number of decisions decided by the Technical Boards of the European Patent Office which, he submitted, established that a product-by-process claim can only be novel if the product itself was novel. Novelty, he submitted, could not be conferred by the process limitations themselves. That position was, he submitted, consistent with an established policy behind the patent system that a person should not monopolise old articles. That did not prevent patentees from obtaining protection for new and improved ways of making old products. He submitted that the law in this country should be the same as that applied in the European Patent Office and in any case the view expressed by the Office was consistent with the Act and the way it should be applied.
27. In our view it is not right to lump all claims which contain a process feature into a category called product-by-process claims. A patentee has to have clear and concise claims, but otherwise is not restricted by the 1977 Act or the Convention as to the way that he defines his monopoly. It is also important to bear in mind that a claim defines the ambit of the monopoly. Everything within the claim is monopolised and everything without is disclaimed. Thus a patentee can define the monopoly claimed so as to disclaim products made by a particular process or only disclaim products which do not have the features of products made by a particular process. The two types of claim can be loosely called product-by-process claims, but to do so is likely to hide the difference between the two.
28. Against that background we turn to claim 26. We agree with the judge that the words “of the expression in a eukaryotic host cell of a DNA sequence according to any of Claims 1, 2, 3, 5, 6 and 7” limit the claim to a polypeptide product which is the product of what we have called the expression of a claim 1 sequence. It follows that products which do not come from a process of such expression are not as a matter of language within the claim. Anybody can make them.
29. There is no need to examine in detail all the decisions of the Technical Boards of the European Patent Office in the bundles before us as they all apply the reasoning in International Flavors & Fragrances Inc T150/82 [1984] OJEPO 309. In that case claim 9 claimed “A product whenever produced by the process of claim 1”. The Technical Board stated:

“7. Inventions fall either into the category of products, e.g. articles, devices or materials, of processes, e.g. methods of preparing a product, or using an article, or obtaining a result. Nevertheless, the invention defined in the claims for products or for processes must all be novel, inventive and industrially applicable according to Article 52(1). Whilst a process may well be novel and deserves full protection in view of its inventiveness, the same may not be true for its product if that is known or obvious in the light of the state of the art. Notwithstanding this, the special protection provided by Article 64(2) EPC extends even to products which are not themselves inventions. According to the submissions of the appellants, the protection provided by “product-by-process” claims should go beyond the limits of “direct products” in Article 64(2) and ought to be equal to that enjoyed by products which are claimed per se, with no restriction to the details of their preparation. This, irrespective of the fact that the product protected in this manner may not represent an invention at all, as such.

…

8. The Guidelines for Examination in the EPO (C.III 4.7b) allows claims for products defined in terms of a process of manufacture provided the products themselves fulfil the requirements for patentability. This may well be the only way to define certain natural products or macromolecular materials of unidentified or complex composition which have not yet been defined structurally. Nevertheless before such claims are allowable their patentability as products must be established since such definition is in lieu of the normal definition by structure.

…

10. An earlier decision of the Board already established that “the effect of a process manifests itself in the result, i.e. in the product in chemical cases together with all its internal characteristics and the consequences of its history of origin, e.g. quality, yield and economic value” (“Gelation/Exxon” T119/82** 12.12.1983). Although problems may be recognised in processes known in the state of the art which are then removed by appropriate modifications or by an altogether different approach, the effect of such measures en route ultimately manifests itself in the technical and economic characteristics of the product, the real purpose of the exercise. Whilst some features of such end-effects may be drawn into the definition of the process for reasons of clarity and of conciseness, the product is in consequence of the invention, without being the invention itself, which is rather the novel interaction represented by the process in such cases. Any attempt to claim the in itself non-inventive product by means of product-by-process claims is claiming the mere effects instead. Whilst reliance on the provisions of Article 64(2) EPC may nevertheless provide protection beyond the invention in processes leading to known or patentable products alike, this should not be afforded for both kinds of product themselves on the same footing, irrespective of their character. This must therefore be rejected as unjustified and contrary to the requirements of Articles 52(1) and 84 EPC. The Board takes the view that in order to minimise uncertainty, the form for a claim to a patentable product as such defined in terms of a process of manufacture (i.e. “product-by-process claims”), should be reserved for cases where the product cannot be satisfactorily defined by reference to its composition, structure or some other testable parameters.

11. The Board has seriously considered the well known fact that both “omnibus” and “product-by-process” claims were commonly admitted in the United Kingdom, one of the Member States of the Convention. Nevertheless, it is also important to note that in no other Member State have they gained acceptance beyond a manner of claiming structurally undefinable product inventions, and there appears to be no room under the Articles or Rules of the Convention to admit such claims on the basis of practice in a single Contracting State. Since the appeal is unsuccessful as regards the issues under consideration, the refund of the appeal fee must be rejected.”

30. The Board accepted that it is permissible to have a claim to a product defined in terms of a process of manufacture, but state that such claims should only be granted in cases when the product cannot be satisfactorily defined by reference to its composition, structure or other testable parameter. That is a rule of practice which is not the concern of the national courts. The European Patent Office is the authority which grants the patents and it operates according to its Rules and Guidelines.
31. It seems that the Office concluded that claim 26 fell within the type of case where the product could not be satisfactorily defined by its features. It may or may not have been right; that is not the concern of this Court. The claim includes within it polypeptide products made using the process of expression of a claim 1 sequence and excludes those that are not the product of such a process. Such a claim is consistent with Article 64(2) of the EPC which allows patents to “extend to the products directly obtained by such process”. If patents can extend to the products directly obtained by a process, then there can be no reason why a claim should not be drafted which does just that. That is reflected in section 60(1)(c) which covers acts where the invention is a process. Thus if the patent claims an inventive process, protection is provided to the patentee against unauthorised disposal of products obtained directly by means of the process, in effect products of the process.
32. The conclusion that claim 26 is limited to products of the process of expression of a claim 1 sequence is consistent with the teaching of the specification when read as a whole. The crux of the invention is the provision of the DNA sequence so that it can be manipulated and used to express EPO. Thus the construction of claim 26 which we believe to be correct serves to distinguish the polypeptide product of claim 26 from what is acknowledged to be prior art, namely uEPO. The construction sought by TKT would not, unless an analysis of the two products showed different characteristics. Thus the preferred construction leads to greater certainty as to the width of the monopoly, whereas the contrary construction requires a comparison of the products of the particular process with any process used by the potential infringer. It follows that the literal construction is that favoured by the Protocol.
33. Of course a product claim must, to be valid, be new. But there is no reason why the limitation of a claim to products produced by a process could not impart novelty. In fact that is contemplated by Article 64(2) of the Convention and section 60(1)(c). If a person invents a new method of extracting gold from rock, he can obtain a claim to the process and as Article 64(2) makes clear, he can also monopolise the gold when produced directly by the process.
34. Although not relevant to the construction of the claim, we draw comfort from the Directive which allows claims to biological elements “isolated … or otherwise produced by means of a technical process even if the structure of that element is identical to that of a natural element”. Clearly the Directive envisages that claims can be validly directed to polypeptides produced by a process.
35. Claim 19 should be construed in the same way. It claims a recombinant polypeptide “… characterized by being the product of eucaryotic expression of an exogenous DNA sequence …”. It is therefore limited to products produced by such expression.

Infringement

36. TKT use a process called gene activation. It was accurately described by the judge in this way:

“595. In summary terms, TKT’s activity involves gene activation technology, which I have briefly explained when considering common general knowledge, and in particular homologous recombination. In most human cells, the EPO gene (like almost any other gene) is “switched off” by a negative regulatory element (“NRE”). TKT introduce into the genome of a human cell, upstream of the EPO gene disclosed in Table VI of 605, a nucleotide sequence which effectively overrides the NRE, and “switches on” the EPO gene. This nucleotide sequence is a DNA “targeting construct” which does not include the encoding region. This construct contains a powerful viral promoter, called the CMV promoter, which is more effective than the SV40 promoter referred to in the 605 patent, and in particular in Examples 7A and 7B. Thus, after this treatment, TKT’s cell line contains endogenous EPO DNA so far as the coding regions are concerned, but it also contains an exogenous promoter construct inserted upstream of that endogenous DNA.

596. What I have just described is a somewhat over-simplified version of TKT’s technology whereby they produce their product, known as GA-EPO. It involves the process known as homologous recombination, which I have already briefly mentioned. As explained by Dr Brenner, it is “the process whereby two DNA sequences which have a region of identity can exchange material at that region.” As he explained, it is a natural process used by all living systems to generate new combinations of genetic material. Targeting of mammalian genes using homologous recombination, i.e. adapting this natural process to enable genes to express proteins artificially, was first disclosed in 1985 by Smithies et al in Nature 317:230. Before that time, and therefore before the relevant date for the purpose of the 605 patent, (as I have already indicated) it was a procedure which had not been used.

597. It is necessary to describe TKT’s process in a little more detail. The targeting construct is a piece of DNA which is constructed outside the cell in which the EPO is to be produced. The targeting construct can conveniently be divided up into seven sequences. First, a so-called targeting sequence of some 3,000 nucleotides upstream of the EPO gene (-5787 to -2482). Secondly, a DHFR gene, whose function is the same as in Amgen’s technology. Thirdly, a CMV gene promoter for activation of the endogenous coding regions. Fourthly, a sequence encoding the first three and a third codons of the human growth hormone (“hGH”) peptide. Fifthly, a selectable marker gene, neomycin phosphotransferase (“neo”) to allow selection of stably transfected mammalian cells. Sixthly, an EPO splice donor site, to enable correct processing of the primary transcript following transcription. Seventhly, another targeting sequence of some 1800 nucleotides upstream of the EPO gene (-2481 to -675).

598. This targeting construct consisting of these seven components is known as pREPO 22. TKT’s technology involves introducing it into a tumour cell line known as HT-1080. As the targeting construct contains sequences at each end of more than 1500 nucleotides in length which are precisely homologous to sequences upstream of the natural EPO gene, the genome will recombine with these sequences by homologous recombination. Ideally, therefore, the targeting construct is effectively inserted into the genome a few hundreds of bases upstream of the EPO encoding sequences in chromosome 7. In other words, the targeting construct is inserted just upstream of the beginning of the sequence disclosed in Table VI of the 605 patent. The insertion is effected by the DNA in the first and seventh components being wholly homologous with these endogenous upstream sequences and therefore combining with them. The “switching on” of the endogenous DNA encoding sequences is achieved by the CMV promoter. The function of the other components of the construct has already been briefly described.

599. Although the targeting construct is intended to be inserted into chromosome 7 (being the chromosome in which the EPO gene is located), homologous recombination in mammalian cells is not efficient or reliable, and there will be insertions at sites in other chromosomes. The neo can be used to separate out those cells which have taken up the targeting construct, and TKT then screens those transfected cells for EPO production, because it will only be those cells where the construct has been inserted upstream of the EPO coding sequences where this will occur. Thereafter, TKT treat the cells with MTX, a standard procedure which I have already described and, indeed, is part of the 605 patent’s teaching, given that it was a standard technique in 1983.

600. As I have already explained, the cell will amplify the inserted DHFR gene, and, because of its proximity thereto, the EPO gene will similarly multiply (again, this is described in the 605 patent, in Example 10). The normal type of cell used in this procedure was the DuX-B11 CHO cells, also referred to in Example 10. Only those cells which contain sufficient copies of the DHFR gene, and, therefore, also multiple copies of the EPO gene, will survive the MTX treatment. In TKT’s technology, those cells were designated as R223 cells.

601. After homologous recombination of the exogenous pREPO 22 construct and the endogenous DNA, the CMV promoter and hGH fragment are integrated into the genomic DNA upstream of the natural promoter of the EPO gene. Subsequent thereto, the following steps occur in the R223 cell line:

1. Transcription of the EPO gene is initiated by the CMV promoter and proceeds along to the hGH and EPO splice donor site sequences upstream of the natural promoter, and along the five exons of the endogenous EPO gene, thereby producing the primary transcript;

2. The primary transcript is then spliced resulting in a variant of natural EPO mRNA. The splicing involves the exogenous hGH sequence being spliced to the next available downstream splice acceptor site. This is so designed as to be the sequence flanking the second exon of the endogenous EPO gene. Accordingly, the first encoding exon is not present in TKT’s EPO mRNA: it is spliced out. It is not required, because it is part of the leader peptide (the rest of the leader peptide being the upstream part of the second encoding exon) which, as I have explained, is cleaved at an early stage, resulting in relatively “mature” EPO;

3. Translation of the EPO mRNA produces immature EPO whose leader peptide sequence is a hybrid between the first three and a third codons translated from the hGH fragment and the first twenty two and two thirds codons translated from the second encoding exon of the endogenous gene;

4. Finally, the leader peptide is removed, and mature EPO is secreted from the cell.”

37. Mr Waugh QC, who appeared with Mr Watson QC and Mr Birss for Amgen, explained in more detail how the amplification referred to in paragraph 600 of the judgment took place. Upon methotrexate amplification of the transfected cell, the gene will be amplified and new copies of the new gene are made. On average there will be 14.6 copies of a new EPO DNA sequence hooked up to a new CMV promoter, and the whole gene itself amplifies so as to enhance the level of expression. Thus new gene copies are produced. In that sense they are newly made.
38. Amgen pleaded infringement of claims 19 and 26. For the reasons given above both claims are limited to products produced by the claimed process.
39. It is convenient first to consider claim 26 which claims a polypeptide product of the expression in a eucaryotic host cell of a DNA sequence according to at least claim 1. Thus it is necessary to turn to claim 1 to find out what is the relevant DNA sequence.
40. The judge construed the opening words of claim 1 as requiring the DNA sequence to be suitable for use in securing expression in a host cell which encodes EPO or certain analogues of EPO. He concluded that the host cell referred to had to be host to the sequence in the sense that the sequence had to be exogenous to the host cell. In essence the sequence had to be isolated so as to be suitable for use in the host cell.
41. Amgen submitted that the judge’s construction of claim 1 was wrong. They accepted that the words “for use in” meant “suitable for use in”. Thus the DNA sequence had to be suitable for use in securing expression. However they submitted that the judge had read words into the claim. The host cells referred to in the claim could include cells which were host to exogenous EPO DNA, but were not limited to such a case. They therefore included cells which contained the DNA sequence. The host cell had to be a cell which hosted the DNA sequence during expression. That was done in the TKT process. Put shortly, the HT-1080 TKT cells were hosts to the DNA sequence. Amplification took place to produce new copies of the gene which expressed EPO.
42. We believe the judge was right to construe claim 1 in the way that he did for the reasons he gave. As a matter of language, the “host cell” must be “host” to the DNA sequence. That means that the sequence must be exogenous. That is consistent with the disclosure in the specification when read as a whole and we can see no reason why it would be appropriate to give the words a wider meaning so as to cover a process, such as that used by TKT, where the sequence is already in the cell, particularly as such a process was not envisaged when the patent was sought in 1984. It is therefore unlikely that the patentee intended his claim to have that wider width. It follows that the TKT product does not fall within the words of claim 26 when literally construed.
43. The Protocol states that the ambit of the claimed monopoly should not be ascertained by literal construction of the claim. Claims have to be construed so as to give fair protection to the patentee and reasonable certainty to the public.
44. Having concluded that claim 26, as appendant to claim 1, did not literally cover the TKT product the judge rightly went on to consider whether there was infringement adopting the principles of construction laid down in the Protocol. He concluded that there was infringement. When doing so he used as an aid the Protocol questions referred to in Wheatley.
45. The judge identified the variant as being, that in TKT’s process the EPO encoding sequence had not been isolated and the encoding DNA is not in a “host cell”. He realised that those features were, at least to an extent, different ways of putting the same point. He concluded that that variant did not have a material effect on the way that the Amgen invention worked. He said:

“623. My broader reason for rejecting Mr Kitchin’s contention in this connection is, at least to an extent, substantially along the same lines as my reasoning for favouring Amgen’s arguments on the issues of Biogen insufficiency and breadth of claim. The essence of the invention embodied in 605, its contribution to the art or what one might call its inventive concept, is the disclosure encapsulated in Table VI, which contains the whole of the encoding sequences, the whole of the intervening introns, and a large proportion of the upstream and downstream sequences. It enabled that to be done which was previously impossible, namely the production of EPO in accordance with biotechnological methods, as they existed at the relevant date as they would have been expected to develop and improve over the ensuing years. It seems to me that, in a fast developing technology such as that involved here, it would have been inconceivable to the notional reader of the 605 patent at the relevant date that there would not be significant developments and changes in the technology of genetic engineering over the life of the 605 patent. What TKT have done is to use a new technique, homologous recombination, to achieve EPO expression by the natural EPO-encoding sequences.

624. Indeed, this is consistent with the evidence of Professor Proudfoot who said there was nothing new about “an SV40 promoter, the gene of interest, DHFR for amplification placed in a CHO cell, that was all standard in 1984”. He accepted that “the basic invention was the isolation and sequencing of... the EPO gene followed by the disclosure of a route to its expression.” He said that TKT “certainly uses newer technology to achieve that result”. While he immediately went on to say that he did not think that was the only difference, it appeared to me that the other differences he identified were either inherent in TKT’s technology (e.g. the ability to target the promoter) or were features admitted on behalf of TKT not to assist on the issue of infringement (e.g. the stronger promoter). In other words, the only significant variant is the use of newer technology, which did not exist in 1984

625. The teaching of the 605 patent and TKT’s technology involve many of the same essential features. They employ the same EPO encoding sequences; they involve expressing the same EPO artificially; the do so in a eukaryotic cell; they employ an exogenous promoter; the biochemical/chemical way in which they express the EPO is substantially the same. Neither could be achieved without the essential disclosure - the contribution to the art - of the 605 patent itself (as I shall explain in a little more detail below).”

46. The judge then answered the second Protocol question, namely “Would it have been obvious to a person skilled in the art that the variant would work in the same way?” He concluded in his main judgment and in his judgment when the matter was reopened that the question should be answered in the affirmative. In essence he concluded that once the TKT process had been explained to the skilled person so that he/she realised the process would work, it would be obvious to the skilled person that it worked in the same way.
47. The judge then concluded that there was nothing which would have led the skilled person to understand that the patentee intended to confine claim 1 to the strict literal meaning of the claim.
48. Amgen supported the reasoning and conclusion of the judge, whereas TKT submitted that the conclusion was wrong in law and in fact.
49. We have come to the conclusion that the judge was wrong to conclude that the TKT product infringed claim 26. We believe that his conclusion was the result of a number of errors. First, we cannot accept his conclusion as to what the variant was. He correctly identified the reason why the TKT process did not fall within the literal construction of claim 1, but to leave it at that hid the real difference between the invention and the TKT technology. The claim 1 DNA sequence had to be suitable for use in securing expression in a host cell in the sense that it must be isolated for use in a host cell. TKT use an endogenous sequence which is activated by introduction of the construct.
50. The first Protocol question requires the court to answer the question – Does the variant have a material effect on the way the invention works? As was pointed out in Improver Corporation v Remington Consumer Products Limited [1990] FSR 181 at 191 the answer to that question depends upon the level of generality used to describe the way the invention works. The wider the description the easier it is to fit the variant into the description. The right approach is to describe the working of the invention at the level of generality with which it is described in the claim of the patent.
51. Amgen supported the conclusion of the judge that in claim 1 the patentee was enabling the production of EPO. The essence of the invention was the sequence which enabled production of EPO using biological methods. According to Mr Waugh the invention of claim 1 was “the DNA sequence suitable for expressing EPO in a host cell” and claim 26 was “EPO produced in a host cell with the DNA sequence of claim 1” and the level of generality was consistent with the invention.
52. There can be no doubt that at the heart of the invention was the discovery and sequencing of the gene that produced EPO. That work of Dr Lin enabled recombinant EPO to be produced and was also part of the essential knowledge that was required before the TKT process could be carried out. However that gene sequence was not claimed as the invention we expect because a claim to the sequence per se would not be patentable. Thus the claim is to a DNA sequence which has been made suitable for use in a host cell to produce EPO. In effect the claim is to an exogenous DNA sequence suitable for expressing EPO when introduced to a host cell. The variant is very different. The DNA sequence is endogenous. It is not suitable for expressing EPO until after introduction of the construct. When so viewed there can be only one answer to the first Protocol question which is “Yes”. There are real differences between an isolated DNA sequence which is suitable for use in a host cell and a DNA sequence in a cell which needs activation.
53. Having regard to the way we have answered the first Protocol question, it is not necessary to go on and consider the second question on an assumption that the answer to the first question should be in the negative. It is also difficult to do so. However a considerable length of time was spent considering the judge’s approach on this question and we will shortly state our view upon that assumption.
54. After judgment had been handed down, TKT wrote to the judge suggesting that on his findings he should have answered the second Protocol question in their favour. They submitted that it was impermissible, having regard to the decision of this Court in American Home Products Corporation v Novartis Pharmaceuticals UK Ltd [2001] RPC 159, to assume that the skilled person would be told that the alleged infringing process would work.
55. American Home Products was a case where the basis of the invention was the discovery that rapamycin, a known product, acted as an immunosuppressant. The variant was an analogue of rapamycin and the Court held that it would not have been obvious to the skilled person that the particular analogue would work as an immunosuppressant. Thus the Court answered the second Protocol question in the negative.
56. As was pointed out in Improver, the second question supposes that the skilled person is told both of the variant and the invention and is then asked whether the variant would obviously work in the same way. In Improver, Hoffman J held that it was obvious that “any rod” which had the relevant qualities in sufficient degree and did not have other deficits, such as over-heating or falling to bits, would in principle work in the same way and that the rubber rod plainly belonged in that class. It would seem that the skilled person needed to be told of the construction of the alleged infringement and he then had to decide whether it obviously worked in the same way. If he believed that it would not work, then by definition it could not work in the same way. That approach is, we believe consistent with American Home Products where it was assumed that the skilled person knew of the analogue and then had to decide whether it worked in the same way, namely whether it would be an immunosuppressant.
57. The judge when answering this question failed to keep in mind that the second Protocol question is designed to secure a reasonable degree of certainty for third parties (see Wheatley and Sara Lee Household & Body Care Ltd v Johnson Wax Limited [2001] EWCA Civ 1609). He also started from such a broad definition of the way that the invention worked that the differences were excluded from consideration. He also wrongly assumed that the skilled person would know that the TKT process was an equivalent process for the production of EPO.
58. Having regard to the way that we have answered the first question, the answer to the second question must also be in the negative. However for this part of the judgment we are assuming that the variant works in the same way as the invention. Even so, we do not believe that that would have been obvious to the skilled person in 1984. The evidence was to the effect that in 1984 the skilled person would not have expected the variant to work at all. If so, it could not have been obvious that it would work in the same way.
59. If we had answered the first two Protocol questions in the affirmative, we would have answered the third question in Amgen’s favour.
60. Having answered the Protocol question in a way that indicates a conclusion of non-infringement, it is appropriate to stand back and consider whether that conclusion is consistent with the broad principles laid down in the Protocol. We believe it is. As we have said, the basis of the invention is the discovery and sequencing of the EPO gene. The patentee could not monopolise the gene per se as that existed in nature. The patentee therefore monopolised the DNA sequence encoding for DNA when isolated and in that respect was suitable for use to express EPO in a host cell. As of 1984 such a monopoly would have seemed to give fair protection. To seek to monopolise use of the sequence when not isolated by inserting a construct into a human cell would provide a monopoly not properly supported by the description in the specification. We also believe that third parties could reasonably expect that if they did not use a DNA sequence for insertion into a host cell, there would be no infringement.
61. We have concentrated on claim 26 as appendant to claim 1: but for essentially the same reasons claim 19 is not infringed. That claim is limited to a recombinant polypeptide “… being the product of a eucaryotic expression of an exogenous DNA sequence …”. Clearly the DNA sequence must be that which can express EPO.
62. Mr Waugh submitted that the claim covered the TKT process which used an exogenous CMV promoter together with the exogenous hGH leader sequence fragment coupled with the EPO coding regions. Since the exogenous hGH and CMV protein were expressed, TKT’s EPO was the product of the expression of exogenous DNA. We disagree. A polypeptide of the TKT process is not the product of expression of an exogenous DNA sequence. The promoter, the leader sequence, the DHFR gene, and the targetting sequences make up the basic elements of the construct. The relevant polypeptide is not the expression of that construct which is the only exogenous sequence used.
63. We have not felt it necessary to resolve the dispute as to the ambit of the word “recombinant” in claim 19 as we have already concluded that the TKT product does not fall within the literal meaning of claim 19. However the judge’s conclusion as to the meaning of recombinant would appear to be correct. Further, resort to the Protocol cannot assist Amgen. Similar reasons to those expressed in relation to claim 26 apply. Claim 19 is not infringed.
64. We conclude that the pleaded allegations of infringement fail.

Validity

65. Novelty – TKT alleged that at least claim 26 of the patent lacked novelty as it had been anticipated by the pre-1984 isolation of uEPO. That submission was based upon TKT’s construction of claim 26 to the effect that claim 26 covered products the same as those that would be expressed using a claim 1 DNA sequence. As we have held that the judge was right to reject that construction, the allegation must fail. That follows from the accepted fact that nobody prior to 1984 had produced a polypeptide product using an isolated EPO encoding DNA sequence. In any case, it could not be right to construe the claim to cover uEPO which the specification acknowledges to have been isolated prior to 1984.
66. Insufficiency – TKT attacked at least claims 1, 19 and 26 on the ground of insufficiency. They mounted a separate and different attack of insufficiency against claim 19 and its dependent claims. It was that attack which succeeded. We will deal with the different attack on claim 19 separately.
67. Section 72(1)(c) provides for revocation of a patent if it “does not disclose the invention clearly enough and completely enough for it to be performed by a person skilled in the art.”
68. The first ground of insufficiency was put forward as a squeeze. If the claims of the patent covered the TKT process and products produced by it, then the claims were insufficient as the specification did not enable the TKT process to be performed.
69. That submission cannot succeed as we have concluded that the TKT process and products produced by it do not fall within the claims. In any case we are not convinced that the basis of the submission is sound. The law contemplates that patents will not lack sufficiency even though the claims cover inventive improvements. If the law was otherwise there would be no room for patents which disclosed a principle of general application unless the specification described how to carry out later inventions using the principle.
70. The second ground of insufficiency depends upon the width of the claimed DNA sequence. Group 1(a) of claim 1 is confined to Tables V and VI and 1(b) to those which hybridise under stringent conditions to the appropriate regions and 1(c) extends to those, which but for degeneration, would hybridise. It is not in dispute that the result is that claim 26 as dependent on claim 1 covers millions, Mr Kitchin says billions, of products of analogues which will express EPO. TKT submitted that there is nothing in the specification which provides assistance to the skilled person to find the analogues which will work. Thus, it was submitted, the patent was insufficient.
71. The judge concluded that there were two types of insufficiency which he called classic insufficiency and Biogen insufficiency. In so doing he fell into error. Section 72(1)(c) provides just one ground of invalidity. To put it broadly, the specification must enable the invention to be performed. The standard of enablement has been elucidated in such cases as Valensi v British Radio Corporation [1973] RPC 337 and Mentor Corporation v Hollister Incorporation [1993] RPC 7). However prior to Biogen it was thought that the invention was sufficiently disclosed in the specification if it enabled performance of a single embodiment. That was held to be wrong. As Lord Hoffman said at page 48:

“The specification must enable the invention to be performed to the full extent of the monopoly claimed. If the invention discloses a principle capable of general application, the claims may be in correspondingly general terms. The patentee need not show that he has proved its application in every individual instance. On the other hand, if the claims include a number of discrete methods or products, the patentee must enable the invention to be performed in respect of each of them.”

72. The judge came to consider the attack based on analogues in paragraph 490 of his judgment. He said:

“490. As to analogues, Professor Proudfoot’s unchallenged evidence was that the specific structures and activities of the proteins coded for are not predictable. EPO has 165 amino acids, and a change of one amino acid to another specific amino acid could be deleterious, beneficial, or make no difference; the number of different permutations involved in changing any two of the 165 amino acids runs into millions. Once one contemplates the possibility of changing, say, up to ten of the amino acids (and the evidence I heard suggested, albeit not specifically, that over ten changes in a protein with 165 residues may well not affect functionality, particularly if the changes were to residues not in the active sites of the protein), the permutations are, almost literally, approaching the infinite. Over and above this, the patent purports to cover deletions and additions. It seems to me that investigation as to which analogues (and therefore which encoding sequences) fall within Claim 1 would involve work of a routine nature, but it could not possibly be said that it would take a reasonable time. This is nicely encapsulated in an observation of Professor Wall:

“I know scientists who have spent a career taking a protein and changing every amino acid and seeing what its biological function is. They have long boring careers in my opinion.”

The last adjective in that passage indicates the routine nature of the work, and the penultimate adjective shows that the time involved would not be reasonable.

491. The 605 patent gives very little, if any, significant help on this aspect. There is nothing to indicate which amino acids might be changed and/or which amino acids could probably not be changed in the EPO sequence contained in Table VI. If the patent had revealed the three dimensional structure of EPO, and, perhaps even more, if it had revealed which of the internal sequences constituted the active sites, that would have given some assistance, possibly substantial assistance, to the reader. As I have mentioned, amino acid residues in the active sites of a protein are generally more likely to be essential to the functioning of the protein than residues outside the active sites.

492. The evidence of Dr Browne of Amgen, indicated that Amgen had started an analogue programme, but it was not given high priority, compared, for instance, with expanding the effort on the development of mammalian cell lines producing higher levels of EPO. He also said that Amgen knew very little about the three dimensional structure of EPO or its active sites. He accepted that the investigations necessary to give any real guidance as to which amino acid residues in EPO could be varied without the resultant polypeptide losing its EPO-like characteristics would have involved “a research programme”, the very thing which has been said to give rise to classic insufficiency.”

73. There was evidence which enabled the judge to come to the conclusions he did in the paragraphs we have quoted. As to the group 1(b) of claim 1, Professor Proudfoot said in his first report, “The specific structures and activities of the proteins coded by those sequences are not predictable.” He made a similar statement in respect of group 1(c). His overall view was that a huge effort would be needed to determine when a sequence was covered by the claim and when it was not. However he accepted that “once the EPO gene had been cloned, it would have been a matter of course for the skilled worker in 1983/4 to apply each of the standard techniques described to that cloned gene. Far from being notable achievements of the patent, these techniques were routine and had been carried out on a wide variety of cloned genes or cDNAs by 1984/5.”
74. Mr Kitchin submitted that this case was similar to American Home Products where it had been held that the patent was insufficient because the skilled person could not predict which analogue would have the intended activity.
75. In American Home Products the claim covered a large number of analogues. The judge found that it was almost certain that many of them would not exhibit immunosuppressant activity which was the basis of the invention. That finding, together with the findings that it was not possible to predict whether any analogue would have that activity and that an undue burden was required to discover whether they would work, was the basis for the conclusion that the claim of the width contended for by the patentee would have been insufficient.
76. We asked Mr Kitchin during his reply speech to direct the Court to the evidence that established that one or more analogues within claim 1 would not express EPO. He referred the Court to the evidence which was accurately summarised in the paragraphs of the judgment we have quoted. There was no evidence that an analogue within the groups of claim 1 would not work. The evidence was directed to predictability.
77. Mr Waugh drew to our attention that the hybridisation of group 1(b) had to be under stringent conditions which narrowed the range considerably. He relied on the absence of evidence of any analogue which had been produced by anybody which did not have activity. The investigations that had taken place by Amgen were designed to discover an analogue that had better activity.
78. In our view the judge was right to conclude that the specification disclosed a principle capable of general application. It follows that Amgen were entitled to a claim in correspondingly general terms. To obtain the grant of the patent Amgen did not need to show that they had proved its application in every individual instance. In any case the question of support for a claim was for the European Patent Office not this Court. To establish this ground of insufficiency TKT needed to prove that at least one DNA sequence of groups (a), (b) and (c) of claim 1 would not be suitable for expressing EPO. As there was no such evidence TKT have not established this ground of insufficiency.
79. The third allegation of insufficiency turns upon what the specification says is the invention which has to be described clearly enough and completely enough to be performed.
80. TKT submitted that the essence of the invention was that it provided means of obtaining EPO in greater quantities than the prior art and in particular at high levels. They drew attention to passages in the specification which pointed out the deficiencies of the prior art methods and said that prior attempts to obtain EPO in good yield had failed. At page 6 line 57 it said:

“While substantial efforts appear to have been made in attempted isolation of DNA sequences coding for human and other mammalian species erythropoietin, none appear to have been successful. This is due principally to the scarcity of tissue sources, especially human tissue sources, enriched in mRNA such as would allow for construction of a cDNA library from which a DNA sequence coding for erythropoietin might be isolated by conventional techniques. Further, so little is known of the continuous sequence of amino acid residues of erythropoietin that it is not possible to construct, e.g., long polynucleotide probes readily capable of reliable use in DNA/DNA hybridisation screening of cDNA and especially genomic DNA libraries. ...It is estimated that the human gene for erythropoietin may appear as a "single copy gene" within the human genome and, in any event, the genetic material coding for human erythropoietin is likely to constitute less than 0.00005% of total human genomic DNA which would be present in a genomic library.

To date, the most successful of known reported attempts at recombinant-related methods to provide DNA sequences suitable for use in microbial expression of isolatable quantities of mammalian erythropoietin have fallen far short of the goal.”

81. It continued at page 9 line 32:

“Vertebrate (e.g., COS-1 and CHO) cells provided by the present invention comprise the first cells ever available which can be propagated in vitro continuously and which upon growth in culture are capable of producing in the medium of their growth in excess of 100U (preferably in excess of 500U and most preferably in excess of 1,000 to 5,000U) of erythropoietin per 10⁶ cells in 48 hours as determined by radioimmunoassay.”

82. The judge accepted the submission that the experiment in the patent which produced EPO at the level promised on page 9 of the patent used amplified DHFR- CHO cells and that Amgen had failed to obtain the level in unenhanced CHO and COS and in human cells. The judge went on to review the evidence and concluded that high level expression was not enabled except with DHFR enhanced cells. Even so he concluded that the specification was not Biogen insufficient.
83. We believe that the instinct of the judge was correct in that he came to the right conclusion, but we cannot accept his reasoning. The step which led to the invention was the discovery and sequencing of the EPO gene. The specification envisages high level expression, but it is wrong to categorise the “invention” with which section 72(1)(c) is concerned as being limited to high level expression. The basic invention is the DNA sequence of claim 1 which produces EPO whether or not it is done badly to produce low level expression or carried out using techniques not contemplated in 1984 which would produce very high levels of expression. The invention is not limited to high level expression. That being so, there is no basis for this allegation of insufficiency and the appeal must fail on this point.
84. Claim 19 – The attack on this claim concerned the limitation of the claim to the products of the expression of a sequence which had a “higher molecular weight by SDS-PAGE from erythropoietin isolated from urinary sources”.
85. Having analysed the evidence the judge concluded:

“480. In light of these conclusions, it appears to me that Claim 19 is incapable of being infringed. According to the teaching of the 605 patent (in Example 10 on page 31), and indeed the evidence I have heard, any difference in apparent molecular weight is small. It appears to me that the variations in apparent molecular weight between different batches of urinary EPO, coupled with the fact that it is clear that many recombinant EPOs do not satisfy the test, would put the skilled addressee seeking to discover whether his product was within Claim 19, and seeking to discover this in a reasonable way, in an unsatisfactory, indeed, an impossible, position.”

86. He went on to conclude that the patent was insufficient. He said:

“486. The clarity of the disclosure required by Section 72(1)(c) is not merely conceptual clarity. If it were, then, as I say, Amgen would have no problem with insufficiency on Claim 19. The clarity and completeness have to be such as to enable the skilled person to “perform” the invention. The word “perform”, in this context, carries the implication of work which is non-inventive, routine, not prolonged and not involving research. Even if the skilled addressee finds his product appears to satisfy Claim 19 when run against one sample of uEPO on SDS-PAGE, he might still have to check it against any number of other EPOs. There is no teaching in the patent to suggest that he would have to do this. Further, that he might have to do it was not within common general knowledge. To obtain a number of uEPOs from different sources (whether to find out which has the highest apparent molecular weight or whether to run each of them against his product) would involve the addressee in prolonged work; it would not be routine. Further, it is an exercise which he could never be sure he had completed, unless and until he found that he was not within the Claim (because he identified urinary EPO which had the same or a higher apparent molecular weight than his product). His investigation could be never ending, unless it established non-infringement. His research, even if it ended with a conclusive result, could be time consuming and difficult. His research would not be routine, let alone taught, by the patent.”

87. Section 14(3) requires the specification of an application to disclose the invention in a manner which is clear enough and complete enough for the invention to be performed. That is equivalent to section 72(1)(c). Section 14(5) is as follows:

“14.(5) The claim or claims shall-

(a) define the matter for which the applicant seeks protection;

(b) be clear and concise;

(c) be supported by the description; and

(d) relate to one invention or to a group of inventions which are so linked as to form a single inventive concept.”

88. Section 72 which contains the grounds of revocation has no equivalent to section 14(5). Thus clarity, conciseness and support are matters to be considered by the Patent Office granting the patent, but lack of clarity, conciseness and support are not grounds for revocation. That is also the position under the EPC. It is therefore important, when considering an attack based on section 72(1)(c), to look at the disclosure of the invention in the specification and then decide whether it is enabling. The fact that a claim is not clear or is not supported by the specification is likely to be irrelevant.
89. Upon the findings of fact made by the judge claim 19 lacked clarity in that there was no standard uEPO against which the recombinant polypeptide could be tested using SDS-PAGE. Even so, we do not believe that that means that the specification does not disclose the invention clearly enough and completely enough for it to be performed. Nobody suggested that a skilled person could not carry out the SDS-PAGE test nor that uEPO could not be obtained. Thus the test could readily be performed without undue effort. The difficulty as found by the judge resided in the clarity of the claim in that the test did not produce a standard result.
90. We cannot accept the reasoning of the judge in paragraph 486. He concluded that the skilled person would be able to carry out the SDS-PAGE test with uEPO, but held that the patent was insufficient because of the variety of uEPOs that he could use. He envisaged that the research to obtain a conclusive result could be time consuming and difficult.
91. It cannot be a requirement of the specification that the test be carried out more than once. Certainly the specification does not make that suggestion. Section 72(1)(c) is concerned with enablement. The SDS-PAGE test could be carried out without undue effort. The claim is enabled across its width. In our view TKT’s allegation is not a ground of insufficiency, it is an attack of lack of clarity dressed up to look like insufficiency. We can see no reason to stretch 72(1)(c) to seek to cover issues of lack of clarity of claiming as patentees will not be able to establish infringement of unclear claims.
92. Even if it be assumed that the allegation raises a ground of insufficiency, we believe that the judge approached that allegation incorrectly. He construed claim 19 as covering the use of any uEPO for the SDS-PAGE test. He said at paragraph 453:

“453. … When considering the construction of Claim 19, I reached the conclusions that:

1. The closing words of Claim 19 require that there was a detectable overall difference between the performance of the two types of EPO on SDS-PAGE;

2. The uEPO need only be isolated to the extent necessary to carry out the SDS-PAGE exercise: in particular, it need not be completely pure;

3. The uEPO need not be from a pooled source;

4. The uEPO could be isolated in accordance with any technique referred to in the patent (or within the common general knowledge at the relevant time) with or without any workshop modification.”

93. Sufficiency cannot be approached as a pure matter of construction. It has to be judged as of the date of the application, in this case 1984 (see Biogen at page 54). The specification could therefore be sufficient if the claimed difference in molecular weight was apparent using the uEPO available in 1984 even though the result might be different with post-1984 uEPO.
94. I will come to the factual evidence, but when considering the ambit of a comparative test, such as that set out in claim 19, the court should bear in mind the oft quoted passages from the speech of Lord Shaw in British Thomson-Houston Company Ltd v Corona Lamp Works Ltd [1922] 39 RPC 49. In that case the claim required the filament of tungsten used in the lamp to be “of large diameter”. The Court of Appeal accepted that the word “large” was ambiguous in that it did not enable the skilled person to decide when the claim was infringed. That was rejected by the House of Lords. Lord Shaw said at page 89 line 30:

“It is easy, whenever language of generality and relativity applicable to size, measure, weight or the like have been employed, to put puzzles which might indicate that, at certain points, you are at that thin strip of mechanical territory in which, on the one side, there might be failure, and, on the other, success. But the reply to such criticism is that that is not how practical men work. They work to achieve success, and if, adopting the broad lines laid down in a specification, they do not find any real difficulty in achieving success, this may not conclude the matter, but it goes far to show that the vagueness of the specification has not mislead them – practical people who were not seeking for failure but for success – had not caused the invention to fail to reach their mind, but has, on the contrary, for all practical purposes, guided them, and that without difficulty, towards the success which prior to that no experiment nor invention had achieved.

I think there is no rule, whether of benevolent or malevolent construction, which should apply to patent specifications. A specification must take its rank among all ordinary documents which are submitted to a reader for his guidance or instruction, and a reader ordinarily intelligent and versed in the subject-matter. Such a reader must be supposed to bring his stock of intelligence and knowledge to bear upon the document, not unduly to struggle with it, but anyhow to make the best of it; if, as the result, he understands what the invention is, can produce the object and achieve the manufacture by the help of the written and drawn page, then the subject-matter of the invention cannot fail on the head of vagueness.”

95. Lord Shaw went on to apply those principles and concluded at page 92 line 23:

“And as to the facts of the present case, there is no one, from beginning to end of it, who denies that the description in the Specification could be followed so as to produce the effect so much and so long desired. The one lamp-maker in the case is Mr. Swinburne, and he says, in effect, and says truly, that the invention produced a revolution, that as for working it out its explanation was plain and simple, and that, with the Specification in hand, the article embodying the invention could be made. I attach no weight to the suggestion in argument that you may extend your experiments of large and small in such a way as to come to a point at which the merit evaporates, because in all such cases, in which there is an ambit of the application of a principle, there must, or there may, come a point which is, as I have said, the outer edge of that ambit. The answer to which is, there is nothing intricate in the experiment: keep in view what you want to achieve, apply the principle reasonably and you will have success. But do not blame the patentees for vagueness if he accepts the common-sense rule that in applying an invention within its successful ambit it is expected that those operating the manufacture will be honestly looking, not to failure, but for success in the range in which the principle is applied.”

96. For the purpose of this case the crucial considerations are: first that the onus is upon TKT to establish that the test is insufficient; second that the question of insufficiency has to be judged as of 1984; third it has to be decided by the court through the eyes of the skilled person; fourth the skilled person is deemed to be seeking success rather than failure; fifth lawyers can often think up puzzles at the edge of a claim, but skilled persons are concerned with practicalities not puzzles.
97. Mr Watson for Amgen submitted that the test had to be carried out using 1984 uEPO which could, for example, be made using the four prior published documents named in the patent or uEPO which was available on the market at that time. He must be right in that submission unless the post-1984 uEPO used is shown to have the same molecular weight as uEPO prepared according to the standards available in 1984. The contrary would mean that the claim could be judged as clear in 1984 and ambiguous in say 1990.
98. The first line of attack was based on experiments carried out by TKT on uEPO that had been supplied by Amgen for the purpose of these proceedings. That uEPO was made by Dr Strickland of Amgen using the method described in Miyake, the paper referred to in the specification. Miyake had used the 17mM fraction which contained over 80% of the protein to carry out further steps to isolate the pure uEPO. When Dr Strickland repeated Miyake, he found that 85% of the protein was in the 30mM fraction. He went on to carry out the remaining steps of purification, as described in Miyake, on both the 17mM and the 30mM fraction. TKT used in their SDS-PAGE experiments the 17mM Strickland fraction. It only contained 9.9% of the protein and was four times less pure than the 30mM fraction. If the 30mM fraction had been used, it would have shown that the uEPO was of lower molecular weight as required by claim 19.
99. There was considerable disagreement as to the results of the TKT experiments and Mr Watson showed us photographs of the gels and the evidence as to the criticisms of the technique used by TKT and of the conclusions they drew. That showed there was room for dispute as to what, if anything, was established. But we conclude that the judge was entitled to come to the findings of fact he did and this Court should not interfere with them. He said:

“467. Nonetheless, Dr Strickland proceeded to seek to isolate the uEPO by the subsequent steps in the Miyake procedure from each of those two fractions. He then ran the 30mM fraction on SDS-PAGE against the GA-EPO made by TKT, which, according to Amgen infringes 605. In accordance with his conclusion, it seems to me that the urinary EPO (the “30mM uEPO”) migrated a little faster than the GA-EPO. Accordingly, at least as against the 30mM uEPO, GA-EPO had a somewhat higher apparent molecular weight, and this was also found to be the case by TKT in its experiments.

468. However, the uEPO obtained from the 17mM fraction (the “17mM uEPO”) performed rather differently as against GA-EPO, according to TKT’s SDS-PAGE experiments, as analysed by Professor Matsudaira, with whose view I agree. It seems to me that those experiments show TKT’s GA-EPO migrating to the same point on SDS-PAGE as the 17mM uEPO, thereby suggesting that they enjoyed the same apparent molecular weight.”

100. The judge went on, as we believe he was entitled to, to reject the criticism of Amgen that TKT used a polyclonal antibody. He then came to the next criticism which was that it was wrong to take into account the result using the 17mM fraction which contained a small amount of the protein. He said:

“471. The second criticism of TKT’s experiments relating to the 17mM uEPO is, in my judgment, more formidable. At first sight, at any rate, it would be almost perverse to take the 17mM uEPO, as opposed to 30mM uEPO. The great majority of the uEPO was in the 30mM fraction, and, indeed, the concentration of EPO in the 17mM fraction was actually lower after the fractionation than before. On the face of it, the natural course, if one had to choose between the fractions, would be obvious: one would go for the 30mM fraction. Sir John Walker expressed that view in robust terms.

472. However, I do not think the point is quite that simple. The teaching of Miyake in his paper (op. cit.) involves taking the 17mM fraction, and not the 30mM fraction. Not surprisingly, that is because, as his paper reports, Miyake obtained a much greater concentration of EPO in the 17mM fraction than in the 30mM fraction (unlike Dr Strickland). I have accepted Amgen’s construction of the closing words of Claim 19, in the sense that it refers to isolation in accordance with methods referred to in the patent itself (or within the common general knowledge). It seems to be tacit common ground that the most satisfactory or common of these methods is that reported by Miyake. There is thus obvious force in the contention that a person in Dr Strickland’s position seeking to follow the teaching of Miyake in order to apply the test laid down at the end of Claim 19, would find himself in a state of some uncertainty: should he follow the 17mM fraction, in accordance with the literal teaching of Miyake, or should he follow the 30mM fraction, in accordance with what might be said to be the spirit of Miyake?

473. Although alternative extremes were adopted by witnesses, it seems to me that the correct approach was that of Professor Matsudaira, whose evidence was to the following effect. He saw no reason why the skilled addressee should have to choose between the two fractions: given the dilemma, he would follow both. However, he accepted that, if one had to choose between the two fractions, one would “follow the protein”, to use Mr Watson’s expression, and therefore one would go with the 30mM fraction. In my judgment, a careful person carrying out the work, as Dr Strickland did, would have thought the safest course was to continue the Miyake process in relation to both fractions. I draw support for that conclusion from the fact that this is what Dr Strickland of Amgen actually did. It is suggested on behalf of Amgen that he merely did this because he was carrying out his work in the context of the present litigation, and, in particular, on the instruction of the lawyers acting for Amgen. There was no reliable evidence which even starts to support this contention, in my view. Indeed, there is contemporary documentary evidence to suggest the contrary. There is no record of any relevant discussions with any lawyer (whether in this country or the United States); no lawyer was called to give evidence on the point; Dr Strickland could not recall any relevant discussions with lawyers. Following both fractions seems to have resulted from discussions he had with Dr Egrie. The two doctors were relatively disinterested and relatively skilled scientists (albeit working for Amgen) and not lawyers. I believe my conclusion receives a little support from the fact that urinary EPO is so very difficult to obtain and therefore so scarce. That would be an additional practical reason for not discarding the 17mM fraction.”

101. It is in paragraph 473 that we believe the judge went wrong. He was right to conclude that “it would be almost perverse to take the 17mM uEPO, as opposed to 30mM uEPO.” The evidence was solidly in favour of following the protein which by definition produced the purest sample. However, the judge adopted the approach of Professor Matsudaira to the effect that “He saw no reason why the skilled addressee should have to choose between the two fractions: given the dilemma, he would follow both.” We accept that an expert witness such as Professor Matusdaira would see no reason why he should have to choose, but the true attitude is that contained in the speech of Lord Shaw. The skilled person is deemed to be looking for success and is not concerned with puzzles at the edge of the claim. Seeking success the skilled person would follow the protein. That, Professor Matsudaira made clear in his answer to a question from the judge:

“Mr JUSTICE NEUBERGER: I have the picture. What you are saying really is this, if you do the Strickland experiment following Miyake you would be faced with a dilemma; either you would follow the protein or, if you were strictly going to follow Miyake’s method, you would take 17. You would be faced with a dilemma so you do both but if you had to choose you would follow the protein, probably?

A. I would follow the protein.”

102. The judge drew comfort from what Dr Strickland actually did, but that was no indication of what the skilled person, who was seeking success not failure, would do. Dr Strickland’s repeat of Miyake was carried out for the purpose of patent proceedings. It was not evidence as to what the skilled person would do seeking to produce uEPO to carry out the SDS-PAGE test.
103. Mr Kitchin submitted that the judge was entitled to come to the conclusion that it was reasonable to follow both the 17mM and the 30mM fraction produced in Dr Strickland’s report of Miyake. He may be right if the protein is looked at with the knowledge available today and if the idea is to obtain information. But the task of the skilled person was to carry out SDS-PAGE using the uEPO contemplated by the patentee. Such a skilled person would, when aiming for success, have used uEPO purified to the extent and in the way suggested in the papers referred to in the patent or uEPO that was available in 1984.
104. Professor Matsudaira said that he believed that the 17mM fraction was sufficiently pure to do the test. Perhaps it was, but it gave a different reading to the 30mM fraction. The question is – what would the skilled person, seeking success, do? Would he have bothered about the amount of protein present? If he was seeking success and trying to avoid failure, he would have taken the best fraction. He would go for the protein as did Miyake. That is the effect of the evidence as a whole. If in any case he did do both, he would immediately realise that the test in claim 19 needed to be carried out with the best sample, the 30mM fraction. To suggest as Mr Kitchin did that it was sufficient to run the SDS-PAGE test upon any isolate of any purity which was sufficient to do the test disregards the correct attitude required of the skilled person. The specification teaches that when recombinant polypeptide is compared with uEPO using SDS-PAGE the recombinant will be of higher molecular weight. The patentee could only have made the comparison with 1984 material. Such uEPO was purified according to 1984 techniques to a particular standard of purity. Such techniques included pooling of urine from different sources to avoid a rogue sample.
105. The judge also relied on two papers which were referred to as Imai and Inoue, published in 1990 and 1994 respectively. He said:

“474. There was further evidence about the performance of urinary EPO according to papers which were published after the relevant date, in particular, Imai et al. in Eur. J. Biochem. 194:457 (1990) (“Imai”) and Inoue et al. in Biol. Pharm. Bull. 17(2):180 (1994) (“Inoue”). Inoue reported that if uEPO is purified from urine without the use of phenol (which was a treatment included in the process taught by Miyake) then it migrated more slowly on SDS-PAGE than if phenol was used. Inoue assessed the difference as showing a difference in apparent molecular weight in the region of 2kDa. Inoue also demonstrated that the in vivo performance of the uEPO isolated without the use of phenol was somewhat greater than that of the uEPO where phenol was used. A similar result was vouchsafed by Imai, but he also reported that uEPO prepared without the use of phenol treatment had the same mobility on SDS-PAGE as CHO rEPO.

…

478. There is no reason to doubt the conclusions of Imai and Inoue. Those conclusions are particularly significant because the uEPO produced without the use of phenol has a reported apparent molecular weight 2kDa greater than the uEPO isolated strictly in accordance with the teaching of Miyake, and that sort of difference is similar to, or greater than, any difference in apparent molecular weight recorded between rEPOs and uEPOs in the experiments reported in published papers, and conducted by Dr Egrie and by Dr Strickland. In particular, as I have mentioned, Kung and Goldwasser, upon whom Amgen rely, suggest that CHO rEPO has an apparent molecular weight 1.2kDa greater than uEPO purified in accordance with the teaching of Miyake. I accept that it can be dangerous and unsafe to compare the results of two independent experiments in this way, particularly when they are only contained in reported papers (albeit in respected peer-reviewed journals). However, it would seem from these two papers that, while CHO rEPO has a higher apparent molecular weight than uEPO prepared strictly in accordance with the teaching of Miyake, it has approximately the same, and if anything a lower, apparent molecular weight than uEPO isolated in accordance with the teaching of Miyake subject to a small, almost trivial, modification.”

106. We accept the findings of fact in those paragraphs, but they do not appear to be relevant to the issue. First, the papers were published after 1984 and therefore do not establish what the skilled person seeking success not failure would have done in 1984. Second, the procedures used to purify the proteins were substantially different to those contained in the 1984 papers. For example phenol was not used. Miyake used 7 steps whereas Imai used 6, of which only one was the same as Miyake. Inoue used two purification protocols which contained different steps to Miyake. Importantly there was no evidence that the differences between the protocols of Miyake, Imai and Inoue would in 1984 have been thought irrelevant, nor that the skilled person in 1984 seeking to carry out the test in claim 19 in the way intended by the patentee would have thought of adopting the Imai or Inoue protocols.
107. The judge also relied on experiments carried out by Amgen on three types of uEPO. The first was Lot 82. This was, according to the judge, uEPO isolated substantially in accordance with the teaching of Miyake. We accept that finding, but it was prepared by Dr Strickland using urine from a single source and using a modified protocol.
108. The second was called Goldwasser uEPO which was uEPO isolated from pooled urinary sources by Dr Goldwasser substantially in accordance with the teaching of Miyake.
109. The third was a sample of uEPO purchased by Amgen for a company called Alpha Therapeutics. Nothing was known of the method of purification nor whether other samples were available for purchase by others. Therefore it was not established to be a uEPO available to the public in 1984. The judge held that the result of Amgen’s experiments were:

“460. The results of the Amgen experiments, at least as interpreted by Dr Egrie at the time, were:

1. Lot 82 uEPO had a higher apparent molecular weight than Goldwasser uEPO;

2. COS rEPO had the same apparent molecular weight as Goldwasser uEPO;

3. CHO rEPO had the same apparent molecular weight as Lot 82 uEPO;

4. Alpha Therapeutics uEPO had the same apparent molecular weight as Lot 82 uEPO.”

110. Although Mr Watson drew to our attention the criticisms of Amgen’s experts as to the experiments, mainly those of Professor Cummings, we believe that the judge was entitled to make those findings. In our view the crucial matter is their relevance to the issues of lack of clarity and insufficiency.
111. The judge regarded the differences between the protocols as obvious modifications. He said:

“477. The experiments carried out by Dr Egrie establish the basic facts as I have summarised them, by reference to Goldwasser uEPO and Lot 82 uEPO, one of which was prepared in accordance with Miyake’s teaching, and the other (subject to the urine having been obtained from the single source) also used Miyake’s method subject to a minor workshop modification. The work carried out by Dr Strickland and TKT in connection with these proceedings complicates matters further, in that it indicates, at least to my mind, that following the teaching of Miyake can result in urinary EPO being taken from two different fractions at one of the stages of the purification process, and the two lots of uEPO having different apparent molecular weights, and therefore producing inconsistent results as against a sample of recombinant EPO with regard to the test in the closing words of Claim 19. The position is further complicated by the reports of Imai and Inoue: both of them showed that a minor workshop modification to the Miyake process results in the uEPO thereby produced having a significantly higher apparent molecular weight by SDS-PAGE than uEPO isolated strictly in accordance with the teaching of Miyake, the modification being the non-use of phenol at one of the seven stages taught by Miyake. Although I heard evidence and arguments as to the reason and effect of using phenol, I do not consider that it is necessary for me to go into that aspect: Miyake used phenol, and because it was suspected that phenol might have an effect on the ultimate isolated product, Imai and Inoue dispensed with it. As its use was unnecessary, it seems to me, standing at the relevant date, it would have been an obvious modification in the eyes of the skilled addressee to omit the phenol.”

112. We are not sure what the judge meant by “an obvious modification”. However there was no evidence that any modification would be one that the skilled person would or could have made in 1984 if he was seeking success in carrying out the SDS-PAGE test in claim 19. It is only those modifications that would be relevant when considering clarity of claim 19.
113. As to Lot 82, it was not prior art uEPO in 1984. It was not prepared from pooled urine and a wheatgerm agglutinin agarose was substituted for the G100 column of Miyake; also different buffers were used. In these circumstances, it was not established that a skilled person seeking success would have produced a sample the same as Lot 82. The evidence did not establish why such a person might change the column and the buffers nor why he would not use pooled urine.
114. The modifications referred to by the judge appear to have produced a result different to that produced using the 30mM fraction of Miyake. They therefore had an influence. Whether or not they were “obvious” is irrelevant. The issue requires a decision as to whether they were modifications that the skilled person would have made to Miyake in 1984, when seeking success. There was no evidence that they were.
115. We believe that the judge was right when he said in paragraph 481:

“481. In my judgment, it would be unreasonable to expect the skilled addressee seeking to discover whether his product is within Claim 19, to carry out the exercise required of him by the closing words of that Claim by running his product against more than one sample of urinary EPO… .”

116. However we are of the view that he was in error when he went on to say: “The closing words of Claim 19 suggest that the draftsman believed that urinary EPOs have effectively the same apparent molecular weight, irrespective of the method by which they were isolated.” The draftsman only knew of the uEPO that had been isolated by 1984 using 1984 techniques. The skilled reader of the specification would be able, using the teaching of the specification to produce such uEPO, and seeking success he could arrive at an answer. For our part we do not believe that TKT established that claim 19 lacked clarity nor that SDS-PAGE could not be performed. The only experiment applying best practice as perceived in 1984 used the 30mM fraction. That showed uEPO to have a lower molecular weight than recombinant EPO. A person who repeated Miyake would use the purest form which in the case of the Strickland repeat was the 30mM fraction.
117. The judge held claim 20 insufficient on additional grounds. He said:

“487. There is no specific teaching in the patent to the effect that some or all recombinant EPOs differ in their glycosylation characteristics from some or all EPO isolated from urinary sources. The nearest one gets to any such teaching is in the last two paragraphs of Example 10 (page 31, lines 10 to 22) which describes a “preliminary attempt... to characterise recombinant glycoprotein products from... COS... and CHO cell expressions... in comparison to human urinary isolates”. The disclosure that CHO rEPO had a “somewhat higher molecular weight” than COS rEPO, which in turn had a “slightly larger” molecular weight than pooled urinary EPO would be understood to suggest that the aggregate apparent molecular weight of the glycans on the two types of recombinant EPO was greater (albeit not by much) than on urinary EPO. Beyond that, however, the patent gives no guidance as to the nature of the difference in the “average carbohydrate composition” between the “glycoprotein polypeptide” and the “human erythropoeitin isolated from urinary sources” referred to in Claim 20. Furthermore, as I shall discuss in a little more detail when considering the issue of novelty, there is, despite Professor Cummings’s suggestions to the contrary in his written report, no satisfactory evidence, let alone specific teaching, even today, as to what differences exist in the glycosylation of some or all rEPOs and some or all uEPOs. Accordingly, I think Claim 20 is insufficient on that ground. In any event, it is insufficient for the reasons given in relation to Claim 19, given that it incorporates that Claim by reference.”

118. It seems that the judge held claim 20 insufficient because there was no unifying factor between the natural and recombinant proteins. Even if he is right, that would not provide a ground of insufficiency. Insufficiency is concerned with enablement not with clarity of the claim nor with sufficiency of support. The test required by this claim can be carried out without undue burden. This claim in our view stands or falls with claim 19.

Amendment of the Patent

119. The judge held claim 19 invalid and also, as a consequence, its dependent claims. During the discussion after trial it became apparent that he believed that the patent should be amended to deal with that finding. That being so, Amgen took the initiative and sought permission to amend under section 75 of the Act by deleting the invalid claims. After a hearing, the judge concluded that there was no objection to those amendments. The order drawn up and agreed between the parties gave Amgen “permission to amend” the patent by deleting claim 19 and its dependent claims. It contained a proviso that Amgen “need not secure such amendments from the Comptroller of Patents pending an appeal on the issue of validity of claim 19”; leave to appeal having already been given.
120. That procedure was mistaken. Permission to amend to delete claim 19 was given. Rule 53 of the Patents Rules 1995 states that where an order has been made allowing the proprietor of a patent to amend the specification the proprietor has to inform the Comptroller. That triggers the amendment which is recorded on the register. No time limit for informing the Comptroller appears in the rules, but that was remedied by the CPR Practice Direction 4.8 which requires the patentee to file the appropriate documents “forthwith”. It follows that the judge in his endeavour to safeguard Amgen’s position was led by the parties into making an incorrect order.
121. Amgen should have gone to the judge to seek relief under section 63 (see paragraph 134 below). The judge would then have directed that the patent be amended. He should then have stayed that order pending the appeal against the finding that claim 19 was invalid. That would have enabled Amgen to appeal both the finding of invalidity and the direction.
122. When this was drawn to Amgen’s notice, they sought to regularise the position by seeking leave out of time to appeal the judge’s order so as to vary it to state that Amgen were directed to amend the patent with the direction being stayed pending the appeal on the issue of validity of claim 19 and its dependent claims.
123. TKT did not object to the time for appealing being extended nor to the order sought being granted. We therefore extended the time for appealing and made the order sought thereby varying the judge’s order. That brings the order in line with what was intended by the judge and Amgen.
124. Having regard to the decision that claims 19 and 20 are valid, the order should vary the judge’s order and also set aside the direction to amend.

The Cells Appeal

125. At the hearing on 8th August 2001, Amgen sought relief in respect of infringement of claim 1 by use of TKT cells. According to the judge the information as to possible infringement of claim 1 came to Amgen’s notice in May 2000, but they may only have appreciated the significance of that information in around July 2000.
126. Amgen had not pleaded that claim 1 was infringed. They contended before the judge and also in this Court that they did not need to amend to plead infringement of claim 1 as they could on the enquiry as to damages recover in respect of infringement of claim 1. That they submitted was the practice; alternatively, relief could be granted under the plea of “further or other relief”. As an alternative to those submissions Amgen sought leave to amend to plead infringement of claim 1.
127. As the judge pointed out, Practice Direction, White Book 2002, vol. 2, 2D-7 at paragraph 6.1 makes it clear that a claimant’s particulars of infringement must show which claims are alleged to be infringed and give at least one instance of each type of infringement alleged. That reflects the established practice which enables the Court to decide all the issues of infringement. It follows that the judge was right to hold that Amgen needed to amend if they wished to recover relief in these proceedings for infringement of claim 1. General Tire and Rubber Company v Firestone Tyre and Rubber Company Ltd [1972] RPC 457 is not an authority to the contrary. The judge was also right to conclude that the request for further and other relief did not encompass an allegation that claim 1 had been infringed.
128. The judge went on to consider whether he should exercise his discretion and allow the amendment. Having set out the case advanced by Amgen, the judge concluded that he should refuse the amendment. He said:

“On this aspect of the case my mind has fluctuated, but I have come to the conclusion that it would not be right to give Amgen permission to amend in relation to the cells. First, it seems to me that there is a real possibility that TKT would be prejudiced in terms of the questions that might have been asked at trial, had this aspect been pleaded earlier. The issues raised with the experts were wide-ranging and difficult and it is by no means inconceivable that questions might have been put to the experts which would have cast a different light on some of the evidence, or even resulted in a difference to some of the conclusions I reached.

Secondly, I cannot understand why Amgen did not think it right to warn TKT of the fact that they would be making a claim in relation to the cells, having faced up to that issue in November. It seems to me that it was simply not fair not to warn TKT before trial, and then seek to raise the issue now; Amgen chose to take the risk of not doing so.

Thirdly, it appears to me that it will not be very prejudicial, and may not even be at all prejudicial, to Amgen if I refuse it permission to claim this relief. If Amgen can establish, on findings I have already made, that TKT's cells contain infringing GA-EPO, then, subject to any argument that can be raised by TKT, it seems to me that there would be a powerful case for appropriate relief being granted in relation to the cells. Even if Amgen could not establish that, it would be clear that use of the cells in this jurisdiction to produce GA-EPO would be a breach of the injunction which Amgen is already claiming.

In all the circumstances, therefore, - albeit with some hesitation - I have come to the conclusion that it would be right to refuse permission to amend in relation to the cells, even in the United Kingdom.

So far as domestic relief in relation to the use of relevant data is concerned, it seems to me that logic indicates the same conclusion. Assuming there is jurisdiction to grant such relief (which even though the relief is domestic, may be questionable) many of the arguments for refusing to exercise my discretion in favour of Amgen on the long arm relief apply. Further, the case for refusing to permit Amgen to claim such relief is stronger than in relation to domestic relief on cells. In particular, the factual basis for such a claim was known to Amgen before proceedings were even issued, as the potential for raising further facts investigations, issues and new defences open to TKT would be much more extensive. Again, Amgen would probably be protected by its claim for damages if there is a claim in relation to the use of the relevant data in the UK.

…

Conclusion. In conclusion, therefore, I am of the following view. … Secondly, the claim relating to cells in the UK, so far as it involves a fresh contention that they infringe claim 1, should be rejected. However, I emphasise that that should not prevent Amgen from being able to argue that, on the case as already pleaded, it is entitled to succeed in relation to the cells. …”

129. Mr Watson submitted that the judge had failed to give proper weight to the fact that TKT could not point to any prejudice to them if the amendment was allowed. That, he submitted, was particularly relevant as TKT must have realised that if claim 26 had been infringed, claim 1 would be infringed by acts taking place within the United Kingdom. Mr Watson also submitted that the judge had failed to take sufficient account of the failure of TKT to identify any instance of how TKT would have altered the way their case was presented if claim 1 had been alleged to have been infringed. There were, he submitted, properly explained reasons for the failure to plead infringement of claim 1 and for the delay since summer 2000.
130. We understand why the judge’s mind fluctuated, but we can find no fault in the way he exercised his discretion. It would not be right for this Court to come to a different conclusion. In particular it cannot be right that Amgen, who were seeking permission to amend their pleading should require TKT to demonstrate that they would have altered the way they would have presented their case. It must be enough that there was a serious possiblity that they would have done so. We therefore would dismiss the appeal on this point.

Good faith and Reasonable Skill and Knowledge

131. As the judge had held the patent to be partially valid, Amgen sought to amend it by deleting claim 19 and its dependent claims. They also sought relief in respect of the part held valid. That was possible under section 63 provided that Amgen established that the specification was framed in good faith and with reasonable skill and knowledge. Amgen sought to do just that. TKT called no witnesses, but hotly contested Amgen’s contentions. The battle lasted 8 days. In our view this resulted in an unwarranted expenditure of costs and unnecessary use of judicial time. We have no doubt that in future the judges sitting in the patents court will use case management techniques to restrict such disputes to 2 or 3 days at the most, even if that involves restrictions on the time allowed for cross-examination and oral argument.
132. The judge held that the amendments to the patent were allowable and that Amgen had established that the specification had been framed in good faith and with reasonable skill. To do so, he had to look both at the specification in the form that the patent was granted and in the form after amendment in the European Patent Office in 1999. That was necessary because of a number of sections in the 1977 Act.
133. Section 77(4) states that when a European Patent (UK) is amended in accordance with the EPC, the amendment shall have effect for the purposes of parts I and II of the Act as if the specification had been amended under the 1977 Act. Section 75(3) provides that an amendment under this section “shall have effect and be deemed always to have had effect from the grant of the patent.”
134. It follows that the amendments made in the European Patent Office in 1998 were deemed to have effect from the grant of the patent in 1990. However section 62(3) would apply in respect of any pre-1998 infringements as it provides:

“62(3) where an amendment of the specification of a patent has been allowed under any of the provisions of this Act, no damages shall be awarded in proceedings for an infringement of the patent committed before the decision to allow the amendment unless the court or the comptroller is satisfied that the specification of the patent as published was framed in good faith and with reasonable skill and knowledge.”

135. As Amgen claimed damages for infringement prior to the amendments in 1998 they needed to prove that the specification as published was framed in good faith and reasonable skill and knowledge.
136. Having found the patent partially valid, section 63 of the 1977 Act applied.

“(1) If the validity of a patent is put in issue in proceedings for infringement… and it is found that the patent is only partially valid, the court… may, subject to sub-section (2) below, grant relief in respect of that part of the patent which is found to be valid and infringed.

(2) Where in any such proceedings it is found that a patent is only partially valid, the court… shall not grant relief by way of damages, costs or expenses, except where the plaintiff … proves that the specification for the patent was framed in good faith and with reasonable skill and knowledge, and in that event the court… may grant relief in respect of that part of the patent which is valid and infringed, subject to the discretion of the court… as to costs or expenses and as to the date from which damages should be reckoned.”

137. The specification to be considered is the specification with the 1998 amendments as that is deemed to have effect from the date of grant.
138. The argument before the judge turned on passages in Example 10 as published and after amendment in 1989. As published the passages were as follows:

“A preliminary attempt was made to characterise recombinant glycoprotein products from conditioned medium of COS-1 and CHO cell expression of the human EPO gene in comparison to human urinary EPO isolates using both Western blot analysis and SDS-PAGE. These studies indicated that the CHO-produced EPO material had a somewhat higher molecular weight than the COS-1 expression product which, in turn, was slightly larger than the pooled source human urinary extract. All products were somewhat heterogeneous. Neuraminidase enzyme treatment to remove sialic acid resulted in COS-1 and CHO recombinant products of approximately equal molecular weight which were both nonetheless larger than the resulting asialo human urinary extract. Endoglycosidase F enzyme (EC 3.2.1) treatment of the recombinant CHO product and the urinary extract product (to totally remove carbohydrate from both) resulted in substantially homogeneous products having essentially identical molecular weight characteristics” (emphasis added).

Purified human urinary EPO and a recombinant, CHO cell-produced, EPO according to the invention were subjected to carbohydrate analysis according to the procedure of Ledeen, et al. Methods in Enzymology, 83(Part D), 139-191 (1982) as modified through use of the hydrolysis procedures of Nesser, et al., Anal.Biochem., 142, 58-67 (1984). Experimentally determined carbohydrate constitution values (expressed as molar ratios of carbohydrate in the product) for the urinary isolate were as follows: Hexoses, 1.73; N-acetylglucosamine, 1; N-acetylneuraminic acid, 0.93; Fucose, 0; and N-acetylgalactosamine, 0. Corresponding values for the recombinant product (derived from CHO pDSVL-gHuEPO 3-day culture media at 100 nM MTX) were as follows: Hexoses, 15.09; N-acetylglucosamine, 1; N-acetylneuraminic acid, 0.998; Fucose, 0; and N-acetylgalactosamine, 0. These findings are consistent with the Western blot and SDS-PAGE analysis described above” (emphasis added).

Glycoprotein products provided by the present invention are thus comprehensive of products having a primary structural conformation sufficiently duplicative of that of a naturally-occurring erythropoietin to allow possession of one or more of the biological properties thereof and having an average carbohydrate composition which differs from that of naturally-occurring erythropoietin.”

139. By amendment in 1998 the middle paragraph was deleted and claim 19 was inserted.
140. TKT contended that there were errors in the patent as published namely:

“a. The first paragraph of Example 10 described COS rEPO as having a “slightly larger” apparent molecular weight than urinary EPO, whereas there is not and never has been any evidence available to Amgen to support this;

b. The first paragraph of Example 10 described CHO rEPO (and indeed COS rEPO) as having a higher apparent molecular weight than urinary EPO, whereas Amgen knew that urinary EPO had a variable apparent molecular weight which was sometimes equal to that of CHO rEPO (and was sometimes equal to and sometimes greater than, that of COS rEPO);

c. The carbohydrate analysis of recombinant EPO as described in the second paragraph was demonstrably inaccurate.”

141. The judge reviewed the evidence as to whether there were errors and how they had occurred in considerable detail. He held that the passages in Example 10 did contain errors as alleged by TKT. As to error (a), he held that it was perpetrated by Mr Borun who was the patent attorney who drafted the specification and that it had been carried through when amendments were sought in the European Patent Office.
142. As to error (b), the judge held this was an error, but that it was excusable. He said:

“137. In all these circumstances, I am of the view that the presence of error (b) in the Patent as approved by the Appeal Board is attributable to Amgen in the sense that they did not draw the attention of the European Patent Office to the performance of CHO rEPO as against Lot 82 uEPO and Therapeutics uEPO, but that their failure to do so cannot be criticised. An applicant for the Patent, using proper skill and care, could reasonably have decided, as indeed the European Patent Office itself decided, that these types of urinary EPO were irrelevant even for the purpose of comparison with recombinant EPO in the context of the experiments described in the first paragraph of Example 10, and, indeed, in light of the qualification to Claim 19 as expressed in Auxiliary Request 11. Even if Amgen had taken a different view from the European Patent Office, they were entitled to take advantage of the considered view expressed in the letter of 25^th August 1994, unless it had been apparent to Amgen that the Rapporteur was under a misapprehension or was not aware of some relevant fact. There is no suggestion of that.”

143. The facts as to error (c) were accurately set out by the judge. He said:

“140. Error (c) can be dealt with more shortly. Not long before the filing of the fourth US Patent application and the application before the European Patent Office, Amgen arranged for samples of recombinant EPO made in accordance with the teaching of the Patent to be sent to a Dr Yu of Yale University. This was because, according to Dr Lin, there was no one in his laboratory “with expertise in determining the carbohydrate constitution of glycoproteins”. Dr Yu analysed samples of urinary EPO, and of recombinant EPO made in accordance with the teaching of the Patent, and reported to Amgen that the results of his experiments were as recorded in the second paragraph of Example 10. These results were reported shortly before the filing of the fourth US Patent application and the application for the grant of the Patent to the European Patent Office. Because no one at Amgen had expertise in the carbohydrate composition of glycoproteins, the fact that Dr Yu’s reported results were inaccurate was not appreciated by anyone at Amgen, or indeed by Dr Borun. It appears that it was only during the Interference Proceedings, namely sometime in 1990 or 1991, that Amgen appreciated the error. During the appeal process before the Appeal Board, but well before the hearing in September 1994, Amgen conceded that the experiment reported in Example 10 was inaccurate, and indeed the Appeal Board described it as admittedly “wrong and unreliable” in its decision in November 1994.”

144. The judge went on:

“141. So far as I am aware, there has never been any explanation as to how Dr Yu can have arrived at or recorded what are now admitted to be plainly wrong figures, or, to be more precise, one egregiously wrong figure (namely the figure of 15.09 for Hexoses in relation to recombinant EPO) in his analyses. Given that nobody within Amgen had the requisite expertise, and there is no basis for criticising Mr Borun in this connection, it seems to me that it cannot be said that any employee or representative of Amgen demonstrated a want of reasonable skill and knowledge in connection with error (c).”

145. TKT submitted that the fact that the carbohydrate experiments were conducted by Dr Yu was irrelevant. The judge rejected that submission. He said:

“144. These arguments have obvious attraction. However, I have come to the conclusion that they should be rejected. I am ultimately concerned with the question of whether the inclusion of error (c) involved a want of reasonable skill or knowledge on the part of Amgen and this would include Mr Borun as the individual responsible for drafting the Patent. It was positively creditable for Amgen to have taken the view that none of their employees had the skill to carry out the carbohydrate analysis in question, and no criticism can, or has, been made of them in having contracted out that exercise to an apparently competent expert, Dr Yu. It is not suggested, other than by reference to the notional addressee of the Patent, that his mistake should have been obvious to Amgen or Mr Borun.

145. There is no warrant, in my view, for equating all the components of skill and knowledge to be expected of the notional reader of a Patent to the actual writer of the Patent, even accepting that the writer under section 63(2) is treated as extending to the patentee, its employees and its patent agent. There is nothing to suggest a departure from reality when asking whether the patentee was guilty of want of reasonable skill and knowledge under section 63(2). The Statutory language and the authorities appear to require the court to take the patentee and his patent agent as it finds them, albeit that one obviously is entitled to expect them to be honest and to expect the patent agent to be “properly instructed” and professionally competent (see the guidance given in Chiron (No. 7) at [1994] FSR 458 at 467 to 468).

146. In these circumstances, I am of the view that error (c) did not result from want of reasonable skill or knowledge on the part of Amgen.”

146. The judge then came to consider whether Amgen had established the specification had been drafted in good faith. He reviewed the evidence and made a number of criticisms of Mr Borun. His conclusion was:

“165. In all these circumstances, albeit not without some hesitation, I have come to the conclusion that Amgen have established that there was no lack of good faith on their part, or on the part of any of their agents, when including error (a) in the first paragraph of Example 10 in 1984, or in supporting the same in 1994, or in supporting and drafting Claim 19 in its various forms, in 1984 or 1994. A finding of lack of good faith would involve a serious criticism of Mr Borun. While the allegation only has to be made out on the balance of probabilities, I must bear in mind the gravity of the allegation and also the substantial lapse of time and the inevitable fading of memories. The onus is on Amgen to establish good faith, and they have discharged it.”

147. Mr Kitchin submitted that the judge had failed to implement the principle that the onus was on Amgen. We can find no basis for that submission. Mr Kitchin then took us through the facts as found by the judge with the aim of persuading us that he came to the wrong conclusion. Predictably Amgen in their skeleton argument supported the conclusion reached by the judge.
148. Whether a specification was drafted in good faith is a question best decided by the trial judge. In this case the judge had the advantage of seeing the relevant witnesses giving evidence during prolonged cross-examination. He did not accept all the evidence of Mr Borun who was the critical witness. Even so, he held that the specification as drafted and as amended was framed in good faith. Unless the judge has gone clearly wrong, it would not be right for this Court to come to a contrary conclusion. He did not fall into such an error. We therefore reject the appeal against his conclusion that Amgen had established good faith.
149. We also believe that TKT’s submission that the judge wrongly held that the patent was framed with reasonable skill and knowledge should be rejected. We believe that the judge was in substance right for the reasons he gave which we incorporate into our judgment in full:

“180. Turning to the more specific points which favour the non-engagement of section 63(2) in the present case, it seems to me that they are as follows. First, the Patent is a very long and detailed document, in which only three errors have been identified, and only one of those errors is attributable to want of reasonable skill and knowledge. Secondly, HMR did not challenge the view expressed by Mr Bannerman, a patent agent of 30 years experience, and the current president of the General Legal Commission on Biotechnology Patents. Having read the Patent as granted originally by the European Patent Office and the Opposition Division, the decisions in 1994 and 1998 of the Appeal Board, and my judgment of 11^th April 2001, he gave evidence as to various aspects relating to these applications, and concluded:

“[W]e have been looking in detail at one claim and one short passage in a long and complex specification. In my view, the skill in drafting overall was of the highest order and to end up with only the present areas of concern is to my mind, highly creditable.”

181. No reason was advanced to challenge the impartiality of Mr Bannerman or the reasonableness of this conclusion; indeed, the conclusion was not challenged. Obviously, that does not bind the court, nor does it mean that it is not open to HMR to invite the court to take a different view. However, having considered the whole of the specification of the Patent in no little detail over a period of more than four weeks last year, and having considered parts of it in relation to the present applications, my view is that, taken as a whole, the specification was well and competently drafted.

182. Thirdly, if one focuses on error (a), it appears to me that, taken on its own, it did not have a particularly wide ranging effect. Whether one considers the claim which is now Claim 19 in the form and context of the specification as drafted in the original application, or as granted by the Examining and Opposition Divisions, or as proposed to be modified by the Appeal Board, it appears to me that it could and would have been granted in the same form as it was, and indeed would have been valid, if the only objection to the claim was the inaccuracy contained in error (a). That is because, on this hypothesis, while one would treat the contents of error (a) as notionally excised from Example 10, error (b) would remain, as would, at least until it was effectively withdrawn in 1994, error (c). So long as error (c) was present, there was a basis upon which Amgen could perfectly properly contend that recombinant EPO could be distinguished from prior art EPO on the grounds of carbohydrate content. Even after error (c), and with it the whole of the second paragraph of Example 10, had been excised from the Patent, it seems to me that error (b) would still have justified granting the claim which is now Claim 19, in its various forms. That is because error (b) appeared to establish that a particular recombinant EPO, namely CHO rEPO, had a higher apparent molecular weight by SDS-PAGE than the sole type of urinary EPO which, at least in the opinion of the European Patent Office, represented prior art EPO, namely Goldwasser uEPO.

183. In my judgment, if Amgen had conceded that error (a) should be discharged because it was wrong, the Examining Division, the Opposition Division and the Appeal Board would each have nonetheless been prepared to grant the claim which has now become Claim 19 in the same form as they respectively thought was acceptable. This is because they would have concluded that, on the evidence available, at least one way of performing the invention claimed by the Patent produced a glycoprotein which could be distinguished on SDS-PAGE from the prior art. The only disadvantage to Amgen would have been that it would have been clear from their concession that, at least on the basis of their experiments, it would not have appeared that COS rEPO was within the ambit of the Claim. However, it is relevant in this connection to mention again that the commercially valuable process as described in the specification of the Patent involved manufacturing CHO rEPO, and not COS rEPO.

184. Fourthly, if error (a) had been the only error in the Patent, there would have been no need for Amgen to apply to amend the Patent. If error (b) had not been a mistake, I would simply have concluded that CHO rEPO, and any other recombinant EPO which fell within the ambit of the Patent, would fall within Claim 19, unless it had the same apparent molecular weight as, or a lower apparent molecular weight than, Goldwasser uEPO. Ironically, it is only because the Appeal Board took a different view from me on product-by-process claims, and I took a different view from the Appeal Board as to whether Lot 82 uEPO and Therapeutics uEPO constituted prior art, that Amgen need to amend the Patent and an argument can be run against them under section 63(2).

185. In this connection, however, the important point is that the reason I found Claim 19 invalid was not because of the mistake made in error (a), but because of the mistake which in my view underlies error (b). That mistake was that, at least in my opinion, the apparent molecular weight by SDS-PAGE of urinary EPOs was variable to such a significant extent that, at least on the basis of the evidence as it seemed to me, it would be impossible to be confident whether a particular recombinant EPO did in fact have a higher apparent molecular weight than any prior art urinary EPO. One might test one’s product against, say, six or seven different samples of urinary EPO, and find that it appeared to fall within Claim 19, but there would always be the possibility of there being another source of urinary EPO which manifested the same apparent molecular weight as one’s product. That problem underlies error (b), but not error (a).

186. As I hope has been made clear, errors (b) and (c), although mistakes, are not attributable to any want of reasonable skill and knowledge on the part of Amgen. Error (b) is plainly based on a view which a reasonably skilled and knowledgeable person could take, and indeed mirrors the considered view taken by the European Patent Office in its letter from the Opposition to Amgen. Error (c) resulted from a mistake made by an expert in whom Amgen had reasonably placed reliance.

187. Fifthly, it does not seem to me that the Opponents before the Appeal Board, or indeed the Appeal Board, itself were relevantly misled. In their decision of 21^st November 1994, the Appeal Board refer to the “consensus” among the Opponents that “the claims to the polypeptides lacked novelty” because “given an EPO preparation, it was impossible to establish whether it was uEPO or rEPO on the basis of the sugar composition”. The Appeal Board then referred to the evidence of a number of different expert witnesses on behalf of the Opponents, including Dr Conradt. They went on to say that the contents of the first paragraph had “not been shown to be wrong” and stated “that recombinant EPO produced from COS-1 and CHO cell expression had a higher molecular weight than uEPO”. The Appeal Board then said that Claim 19 in the form in which it had been originally accepted by the Examining Division and the Opponent Division “suggests merely that products different from uEPO can be obtained, not that there is any recognisable advantage in doing so”.

188. The Appeal Board then expressed the view that the evidence “reveals that… differences [between recombinant EPOs and urinary EPOs] can be attributed only to the particular cases under investigation and cannot be generalised to rEPO as a class”. Consistently with this view, they went on to state that Amgen had been unable to show that any of the alleged “distinguishing features for rEPO is a “universal” one for the whole class of rEPO” and therefore these alleged differences between rEPO and uEPO, which included the molecular weight difference, were regarded by the Appeal Board “as not reliable”. The Appeal Board expanded this view in these terms:

“[T]here appears… to be no certainty of getting a particular rEPO glycosylation pattern. The glycosylation patent for uEPO would also appear to depend on the time of day, and physiological status of the patient from whom it is obtained. rEPO thus appears to share with uEPO the characteristic that the carbohydrate composition is to a considerable degree a matter of chance.”

They also went on to make express reference to Amgen’s submissions to the FDA.

189. I think that two points can fairly be made about these observations. First, it is clear that the Opponents were making the point that there was much evidence to suggest that many recombinant EPOs had the same apparent molecular weight as urinary EPO. That view is reinforced if one turns to the evidence of Dr Conradt to the Appeal Board. That evidence suggested in clear terms that Amgen’s own experiments showed that COS rEPO had the same apparent molecular weight as prior art urinary EPO, and indeed Dr Conradt cited the published papers to which I have referred, as supporting that contention. Secondly, it does not appear that the Appeal Board was in any doubt as to the variability of the carbohydrate content, and hence the apparent molecular weight, of different types of recombinant EPO and urinary EPO, although of course they did not consider that the conclusions embodied in error (a) or error (b) were wrong.

190. In these circumstances, it is perhaps a little difficult to understand why the Appeal Board nonetheless allowed Claim 19 in the form proposed in Auxiliary Request 11. The Appeal Board expressed its reason shortly in these terms:

“The basis for this restriction was simply given by reference to the first paragraph of Example 10.”

As already mentioned, it seems to me that if error (a) had been excised from Example 10, but error (b) had remained, the Appeal Board would have had no good reason for reaching a different conclusion from that which they reached.

191. Bearing in mind these various factors, I have reached the conclusion that Amgen have established that the Patent was drafted with reasonable skill and knowledge. Although there was one error attributable lack of reasonable skill and knowledge, the specification as a whole was well drafted, and, taken on its own, the error concerned would not have led to the Patent being granted by the various sections of the European Patent Office (including the Appeal Board) in a form different from that which it was granted, and, indeed, it was not error (a), or the reason for my finding that it was a mistake, which resulted in my finding that Claim 19 was invalid.”

150. Mr Kitchin submitted that the judge had erred in considering the specification as a whole. He submitted that if part of it had not been drafted with reasonable skill and knowledge, then the patentee could not discharge the burden of proof. That cannot be right. Sections 62 and 63 require the specification to be framed with reasonable skill and knowledge. That requires the court to consider any mistake in the context of the whole specification, making due allowance for any difficulty that the draftsman has and of course the importance of the passage said to be mistaken.
151. Mr Kitchin also submitted that the judge was wrong to conclude that claim 19 would have been granted even if mistake (a) had been corrected and also wrong to take that conclusion into account. He relied on three reasons. First, it was not pleaded. Second, the conclusion was irrelevant and third it was wrong. The pleading point should have been taken at an early stage before the judge and in any case should not be determinative. The other submissions are wrong. The information as to COS r EPO was not an essential support for claim 19 and therefore there could not have been a good reason for the European Patent Office to reject claim 19 if the COS rEPO information had been omitted. The judge was entitled to reach that conclusion. We have no doubt it was a proper matter to take into account when considering the skill of the draftsman.
152. The judge in paragraph 186 of his judgment (cited in paragraph 149 above) appears to have excused error (c) because it resulted from a mistake made by Dr Yu, a renowned expert. That Mr Kitchin criticised. He submitted that the judge had failed to take account of the guidance in Chiron Corporation v Organon Teknika Ltd (No.7) [1994] FSR 458. He submitted, rightly in our view, that what was important was that the specification should be framed with reasonable skill and knowledge. That must encompass the knowledge of the person which formed the basis of the information in the specification. It mattered not whether that knowledge was obtained using work done by a contractor. The issue had to be answered objectively and a patentee who used a contractor was not in any better or worse position than a patentee who did not. Thus where a specification contained an obvious mistake due to work done by a contractor however skilled, the patentee had to shoulder the burden of establishing that the specification was framed with reasonable skill and knowledge.
153. TKT did not call any evidence on the effect of error (c). Even so, we are prepared to accept that it was an error which had not been fully explained and with hindsight seemed an obvious error. However the specification has to be looked at as a whole. Error (c) has no consequence in the context of the issues raised in the action. We conclude that the judge was entitled to decide that “the specification as a whole was well drafted” and that Amgen did discharge the onus on them in that respect.

Conclusion

154. We pay tribute to the way that the judge grappled with the difficult technical issues and made clear findings of fact. That greatly alleviated the task of this Court.
155. For the reasons given we have come to the conclusion that the patent is valid, but not infringed. Having regard to the effect of that conclusion, it will be necessary, after time for reflection, to hear counsel as to the appropriate order to be made.