Mr Justice Warren :
Introduction
- This is an appeal from a decision of the Hearing Officer, Mrs S E Chalmers, ("the HO") dated 25 January 2012 ("the Decision") to refuse the applications ("the Applications") of the Appellant ("PDI") under sections 1(1)(b) and 1(2)(c) Patents Act 1977, based on lack of inventive step, the mental act exclusion and the computer program exclusion.
- PDI argues that the HO identified the wrong inventive concept in her decision, and accordingly erred in her finding that the Applications lack inventive step. PDI also appeals against the finding that the Applications were for excluded matter.
- The Comptroller supports the decision of the HO on her characterisation of the inventive concept, and the finding of obviousness which followed, for the reasons she gave. Further, by way of Respondent's Notice, the Comptroller argues that the inventive concept put forward by PDI lacks support in the specification, contrary to section 14(5)(c) Patents Act 1977, and/or amounts to added matter, contrary to section 76(2) Patents Act 1977 Mr Mitcheson, who appears for the Comptroller, argues that this is further reason to uphold the finding of the Hearing Officer. The Comptroller also supports the HO's decision on excluded matter on the basis that the Applications amount to excluded matter.
- The invention relates to the field of genetics. It is directed towards a screening method for differentiating between copy number variants ("CNV"s – repeated sections of DNA) which are associated with a particular condition or phenotype, and CNVs which are present in the population at large and do not appear to cause disease.
The Nature of the Appeal
- Mr Mitcheson reminds me of the correct approach to an appeal of this nature. The starting point is that the appeal is a review, not a rehearing. Although he accepts that in many cases there may be little difference between the two standards, he draws attention to the observations of Lord Hoffmann in Designers Guild [2001] FSR 11 at [29], and also those of Robert Walker LJ in Reef [2003] RPC 5 at [17]-[30] and particularly [26]-[28]. I do not repeat those well-known passages. In relation to the evaluation of and conclusion on the primary facts, there is no single standard appropriate to every case, but the appellate Court should have regard in particular to
"the nature of the evaluation required, the standing and experience of the fact-finding judge or tribunal, and the extent to which the judge or tribunal had to assess oral evidence" (see Reef at [26]).
- I adopt that approach bearing in mind the experience of the HO in the multifactorial decision which she made.
The Background
- Dr Broughton (who appears on behalf of PDI) puts the background to PDI's case in this way:
a. It has long been known that changes in copy number of part of the genome, on a gross scale, can be responsible for a wide range of phenotypes. Dr Broughton explained (although this was not strictly in evidence) that this is the basis of karyotype analysis, where the genetic copy number changes observable involve segments of the DNA that are large enough to be observable using microscopy, generally in the range of 5 to 10 Mb. At the other end of the size scale, changes in individual nucleotides within the genome are also known to be capable of being causative of phenotypes.
b. The invention relates to CNVs that fall between these two size extremes. CNVs detected in the methods of the invention are sections of DNA which are duplicated or deleted, and which are of a size too small to be observed using standard microscopy techniques. However, CNVs are significantly larger than the single nucleotide base mutations referred to in the previous paragraph. In particular, the CNVs that are detected in the invention are obtained using techniques capable of achieving a resolution of 30 to 50 Kb.
c. The invention (or rather, I would say, the first step in the invention) provides for the detection of CNVs at the claimed resolution – ie a genome wide screen for CNVs. The prior art describes methods for the genome wide detection of CNVs at this size resolution. Most developed in the prior art was the technique of Comparative Genomic Hybridisation (CGH). However, other techniques, emerging from "next generation" DNA sequencing are also capable of providing for such genome wide studies.
d. It was known prior to the invention that CNVs could be causative of or associated with a phenotype. It was also known that not all CNV's are causative of an observable phenotype in an individual. It was therefore postulated that the genome contains normal polymorphism in copy number. In the search for a CNV that is causative of the phenotype of interest, normal polymorphic CNVs are a distraction. It is normally not possible easily to determine whether a CNV that is identified in a screen is causative or merely an irrelevant polymorphic CNV. Since only a single CNV can be expected to be causative in any individual, and non-causative CNVs may be very much more prevalent, the causative CNV may be swamped in a sea of normal but "noisy" CNV data.
- Dr Broughton then goes on to say that identification of the CNVs present in the genome of an individual, at a certain size resolution, is not the end of the endeavour. It is necessary, from amongst the data obtained from, for example a CGH screen, to identify which (if any) of the CNVs detected is causative of the phenotype in question. If this final step can be achieved it leads to the identification, not only of the cause of the phenotype in the individual under investigation, but to the identification of a genetic locus which, when disrupted by any mutation, can cause the phenotype in other individuals. Identification of such a genetic locus can therefore lead to diagnostic tests based not only upon copy number variation but upon any cause of disruption at that locus, and additionally can lead to lines of investigation into the underlying biochemistry of the phenotype and eg possible therapeutic approaches where the phenotype is a disease or condition.
The Claim
- Claim 1 of the Application (which I will refer in this judgment as Claim 1) directed towards human subjects reads:
"1. A method of identifying a relevant copy number variant for a phenotype comprising:
a) identifying a set of copy number variants in a genome wide screen of a human subject with said phenotype, using a method capable of achieving a resolution of 30kb to 50kb;
b) providing genome wide copy number variant frequency data from a population of at least 1000 human individuals without said phenotype obtained using a consistent method capable of achieving a resolution of 30kb to 50kb and comparing said set of copy number variants of step (a) to that data; and
c) determining a copy number variant which is present in said subject but not present in said genome wide copy number variant frequency data as being relevant to the phenotype."
- Claim 1 of the Application directed towards non-human subjects is similar but with reference in the opening words to a phenotype in a non-human subject and reference to non-human in place of human in paragraphs a) and b).
The Specification
- Focusing on the Application directed at human subjects, it is important to note the references which are to found to the size of the reference population used to ascertain the relevant CNV frequency data. Within the section headed "Summary of the Invention" are to be found the following in relation to different embodiments:
i) [0018] in one embodiment "a large population subjects comprising of thousands or at least 10,000 subjects" and in another embodiment "very large groups comprising of thousands or at least 10,000 subjects".
ii) [0020] in a further embodiment "very large groups comprising of thousands or at least 10,000 subjects" again.
iii) [0021] in one embodiment, comparison with CNVs "in at least 1,000 subjects" and in another embodiment, data from "at least 1,000, 5,000, 10,000, 25,000 or 50,000 subjects". I should note that in this embodiment, a method is provided "to determine the relevance of a copy number variant in a subject….".
iv) [0022] I refer to this, not because of a reference to the number within a group of subjects, but because Dr Broughton relies on it as an example of the specification teaching that the goal is to identify a CNV of relevance to a phenotype. "In one aspect, the KMTs [see below at the end of this paragraph] allow for a determination of whether one or more copy number variation in a subject, test subject or patient is associated with a condition or disease".
v) [0025] one aspect of the invention is directed to accessing a set of data representing frequencies of one or more CNVs "in at least 1,000, 5,000, 10,000, 25,000 or 50,000 subjects".
vi) [0026], [0029] and [0032] similarly contain reference to 1,000, 5,000, 10,000, 25,000 or 50,000 subjects.
And within the section headed "Detailed Description of the Invention" are found the following:
vii) [0060] refers to one embodiment where the use of a KMT permitting improved accuracy for diagnostic purposes:
"This tool allows a comparison of results against a compilation of genetic information derived from thousands of individuals to focus the analysis and permit one to distinguish between diseases and/or conditions and disregard normal variations in the genome."
viii) [0063] In one embodiment, the relevant pool is "at least 10,000 individuals (e.g., normals, not displaying a targeted/identified phenotypic effect)". The paragraph goes on:
"Previous studies which have shown, for example, that the frequency of 47,XXY was 1:1000 – thus there were only 10 such individuals ascertained in a cohort of 10,000 newborns. As such, a database of only 1,000 individuals would have resulted in a significant possibility that this well known and important chromosome abnormality would either not have been detected or detected only once….."
This paragraph contains, as Mr Mitcheson notes, the only reference to a database of 1,000 individuals but even here it is clear that the intended database comprised at least 10,000 individuals.
ix) [00153] Again, this is relied on by Dr Broughton in relation to the goal. It says: "In one aspect of the invention, the KMTs are utilised in a method of identifying the relevance of a copy number variant in a subject."
[The KMT is a "knowledge management tool for cytogeneticists to rationally interpret genomic data, including array CGH (aCGH) data in patents": see [0060].]
- Paragraphs [0065]-[0068] provide "a sample calculation illustrating how an interpretation of the significance of copy number changes in a disease state would proceed". Table 1 in [0066] on page 15 is introduced by the following:
"A given copy number variant is present in a proportion, p<1, of individuals with a given phenotype. A comparison of that figure to the KMT, shows that none in the normal cohort possess this change. What follows is a statistical analysis for different values of (p) and the database size (n):"
- [0067] then states that, as the
"non-limiting example above demonstrates, where the frequency of the copy number change in a disease cohort is 0.001 (1/1,000), a database of at least 5,000 is sufficient to identify whether the variation is "normal" or linked to disease. As the size of a database is reduced, the likelihood that the copy number change is not present is high (i.e., for a size of 100, it is over 0.9 etc). Of course this is but one example, and depending on the frequency of one or more variations the required number of individuals in a given database becomes correspondingly smaller."
- It is clear from this, as Mr Mitcheson submits, that the detection of CNVs of different frequencies/proportions would require different sized databases (the example provided demonstrating that a database of at least 5,000 individuals would be required for a CNV present in a frequency of 1/1000 in a disease cohort). However, he submits that there is nothing here which suggests that a database of 1,000 would comprise all polymorphic CNVs present in a population of humans when using a method capable of achieving a resolution of 30kb-50kb. The size of the database required must depend on the frequency of the CNV in the population and the resolution of the screen being carried out. I agree.
- The Specification contains a number of examples starting on p 57. Example 2 is headed "Array-Based Comparative Genomic Hybridization". This provides for a database of CNVs based upon 10,000 normal healthy individuals. [242] states
"Results: The output data from the preceding aCGH analysis of 10,000 individuals will provide a database of normal copy number variations which can be used in the KMTs of the present invention to provide genomic assessments for a test subject or patient or groups of the same, in diagnosing disease or designing therapeutics"
- Example 3 is headed "Comparative Analysis". It provides in [00244] that, once the reference database is compiled, algorithms (similar to those previous described) can be utilised to generate copy number data for test subjects as from the cohorts used to create the normal variation database. Building on Example 2 it is provided that "such analysis will allow for direct comparison of the results on a given individual with the normal expected variation in 10,000 individuals of the same ethnic group". Then at [00253] one finds this:
"In summary the databases will comprise cohorts of 10,000 normal individuals from differing ethnic groups, using complementary but different platforms...".
- Mr Mitcheson's pithy summary is this: Overall, it is clear from the specification that there are a large number of CNVs present in the population, with only relatively few of these being causative of a certain disease or condition. The specification allows for the comparison of the CNVs from an individual with a database of CNVs from a number of individuals. But the number of individuals required for each comparison is unknown.
- It is worth mentioning here for completeness the divisional application relating to non-human subjects. It is part of the Common General Knowledge ("CGK") that different animals have different sized genomes. Mr Mitcheson submits that, notwithstanding the fact that there is nothing in the specification that would indicate that polymorphic CNVs can be associated with certain phenotypes in non-human subjects, given the variance in the size of the genome, as well as the varying frequency of the particular CNV in issue, the size of the database required would be different for both human and non-human subjects.
Prior Art
- Notwithstanding that identification of the inventive concept comes before a comparison of that concept with the prior art, I have found it helpful to consider the prior art at this stage of my judgment in order to set the context for the eventual identification of the inventive step, if any. There are four items of prior art, three of which (D1 – D3) were cited by the HO and one additional item (D4) cited by PDI but not mentioned by her.
i) D1. Pinkel and Albertson: "Comparative Genomic Hybridisation", Annu. Rev. Genomics Hum. Genet. 2005, 6:331-354
ii) D2. Wilson et al: "DNA copy-number analysis…", Human Molecular Genetics, 2006, Vol 15 No 5, 743-749.
iii) D3. Vissers et al (2005): "Identification of disease genes by whole genome CGH arrays", Human Molecular Genetics, 2006, Vol. 14, No 2, R215-R223.
iv) D4. Vissers et al (2003): "Array-Based Comparative Genomic Hybridization for the Genomewide Detection of Submicroscopic Chromosomal Abnormalities", Am. J. Hum. Genet. 73: 1261-1270, 2003. This was not relied on by the HO, nor did Mr Mitcheson rely on it. I say no more about it.
- D1, Pinkel and Albertson, is a review document discussing advances in the technique of CGH, and its application to the identification of CNVs. Relevant parts of the document include the following:
i) The Abstract refers to variations in copy numbers and the aberrations which can occur. "Detecting these aberrations, and interpreting them within the context of broader knowledge, facilitates identification of critical genes and pathways involved in biological processes and diseases and provides clinically relevant information".
ii) At p 333, first full paragraph, it is said that the complexities of the genomic DNA and of the DNA in the array elements significantly affect signal intensities and thus play a dominant role in determining the genomic resolution of different array CGH technologies, going on to say
"For example, copy number information from genomes such as bacteria and yeast (28, 100) is easier to obtain than from mammalian genomes, which are 100 to 1000 times larger, because the concentration of each portion of the genome in the hybridization is corresponding higher."
iii) At p 338 the beginning of the first paragraph under the heading "Variation in Normal Genomes" explains:
"Differences in gene structure and variability in gene families produce DNA dosage polymorphisms in the human genome (14,35). A comprehensive understanding of these normal variations is of intrinsic biological interest and is essential for proper interpretation of array CGH data and its relation to phenotype."
iv) In the first full paragraph on p 339, it is explained:
"Although CGH has established the prevalence of copy number polymorphisms in the human genome, the picture of this normal variation is incomplete. In results reported to date, measurement noise has restricted detection to polymorphisms that involve genomic sections of many kilobases or larger, genome coverage has been far from comprehensive, and the population has not been adequately sampled. The published studies also raise questions concerning their mutual consistency……….Understanding the copy number polymorphisms that are detectable by a particular array CGH technique is important so that normal variations are not falsely associated with disease, and conversely, to determine if some so-called normal variation may underlie phenotypic characteristics such as disease susceptibility (89).".
v) In discussing experiments then underway by a number of groups looking at dosage abnormalities in developmental genetics, it is stated (see the end of the top paragraph on p 343):
"As the effective resolution of array CGH techniques increases, one expects an increasing rate of discovery of medically important dosage aberrations. However, as indicated above, interpreting the primary data will become more complex due to the need to better understand the "normal" polymorphisms in the genome."
vi) Further down the same page reference is made to array CGH with ~1Mb resolution arrays.
- D2, Wilson et al, is a research paper where the authors sought to determine genes involved in bipolar disorder and schizophrenia by identifying CNVs. Relevant parts of the document include the following:
i) Identification in the Abstract of the use of aCGH screening at a resolution of approximately 1.4 Mbp.
ii) In the Results section (starting on p 744 col 1) it is explained that the authors used BAC aCGH with 1.4 Mbp resolution to assay DNA copy number in 105 post-mortem brain samples (n=35 each schizophrenia, bipolar disorder and controls). Positive findings for copy-number aberrations were verified and extended to a replication sample using quantitative real-time PCR. In the initial screening phase, each of the 105 samples was hybridized against a commercially available reference sample composed of pooled male genomic DNA from six individuals. The phase of analysis which followed included a comparison with two previous studies which allowed previously identified polymorphic copy-number differences to be recognised and subsequently disregarded. Seven polymorphic loci were removed from the preliminary screen leaving 27 psychosis-specific loci with a total of 80 aberrations (64 deletions, 16 insertions).
iii) Given what the authors described as the poor validation rate from the preliminary automated process, they modified their strategy. In order to improve the probability of finding relevant genes with potential copy-number aberrations, a more stringent manual analysis was undertaken. This involved selection of data from the 18 affected samples. The 18 array results from that set were screened for certain BAC clones. The resulting list of BAC clones was then surveyed for the presence of brain-expressed genes and any BAC clone containing a brain-expressed gene was then used to query a specified database of all CGH results from the 105 sample set (referred to above) to find
"those clones in which possible copy-number aberrations were present either in affected samples only, or in which affected samples predominated. A final list of 14 BAC clones was produced by this method. Twelve of these clones contained putative copy-number aberrations in affected samples only. The remaining two clones…..contained putative copy-number aberrations in five affected samples and one control sample, and four affected samples and two control samples respectively. These 14 loci were tested using qCPR and three of the 14 loci were verified…… These three validated aberrations were from the set of 12 putative copy-number differences seen in affected individuals but not in controls."
iv) It can be seen, therefore, that only 3 of the putative CNVs were verified as being true CNVs; and that it was noted that the 3 verified CNVs were not present in the control individuals. Dr Broughton draws attention to the fact that there is no disclosure of any individual in which more than a single CNV was identified.
- D3, Vissers et al (2005), is a review article including some original work from the authors' laboratory. Relevant parts of the document include the following:
i) The abstract refers to array CGH (at ~100kb resolution).
ii) It is stated (see p R216 col 2):
"When compared with conventional karyotyping, array CGH provides a higher resolution, a higher dynamic range and better possibilities for automation. In addition, it allows for direct linking of copy number alterations to known genomic sequences…. The increase in data obtained through these high-density arrays requires standardized storage systems as well as thorough statistical tools for normalization and automated detection of genomic copy number alterations… Pilot studies using 1 Mb resolution genome-wide BAC arrays… have recently indicated that causative microdeletions and/or duplications are present in ~10% of patients with unexplained mental retardation and congenital malformations……. It is important to note that these studies also identified submicroscopic copy number alterations that have no direct phenotypic consequences…… These alterations represent a novel class of polymorphisms within the human genome, termed large-scale copy number variations or copy number polymorphisms, whose exact frequency in different ethnic groups remains to be established. It is essential to rule out such submicroscopic variation by studying parental samples and/or independent normal controls before drawing any firm conclusion on whether an aneusomic segment [a CNV] is causative for the disease under investigation."
iii) In the section headed "Conclusions and Future Prospects" on p.219 it is noted that rapid developments in current microarray technologies will lead to a significant increase in the number of elements to be tested. Reliable copy genome screening of most if not all exons present within the human genome will soon become possible. Then comes the following:
"On the basis of published data for X-linked diseases and for some comprehensively studied inherited cancer genes……, the overall percentage of gross deletions involving one or more whole exons may account for up to 15% of all mutations. Assuming an average of 10% whole exon deletions or duplications in monogenic diseases, one would have a 65% chance of identifying any diseased gene among 10 unrelated patients, and nearly 90% chance of identifying the causative gene if 20 such patients were available. This suggests that a further development of methods for gene dosage measurement will result in a general strategy for disease gene identification that is applicable to individual patients."
- I will come to what the HO had to say about the prior art when considering step 3 of the structured approach to inventive step to which I now turn.
Inventive step
- I turn now to the structured approach to inventive step explained in Windsurfing International Inc. v Tabur Marine (Great Britain) Ltd [1985] RPC 59 and Pozzoli SPA v BDMO SA [2007] EWCA Civ 588 ("Pozzoli"). It must, however, always be borne in mind that the Pozzoli is a guide and not legislation. The issue is always whether there has been an inventive step within sections 1(1)(b) and 3 Patents Act 1977 which itself turns on whether the alleged inventive step is "obvious to a person skilled in the art". The approach involves the following steps:
(1) (a) Identify the notional "person skilled in the art"
(b) Identify the relevant common general knowledge of that person;
(2) Identify the inventive concept of the claim in question or if that cannot readily be done, construe it;
(3) Identify what, if any, differences exist between the matter cited as forming part of the "state of the art" and the inventive concept of the claim or the claim as construed;
(4) Viewed without any knowledge of the alleged invention as claimed, do those differences constitute steps which would have been obvious to the person skilled in the art or do they require any degree of invention?
Person skilled in the art and common general knowledge
- There is no dispute about the identity of the "person skilled in the art". It is, as the HO explained in [13] of the Decision, a team comprising molecular biologists and cytogeneticists. Nor is there any dispute about the common general knowledge ("CGK") of the person skilled in the art. It includes: the fact that the DNA copy number will modify gene expression and function; that copy number variation is found among the normal population and that variation may also be associated with various diseases/phenotypes; molecular biological techniques such as polymerase chain reaction (PCR) and array comparative hybridization (aCGHA) and basic statistical methods to enable assessment of the significance of result. The assessment leads to the identification of any CNVs present.
Inventive Concept
- There is, however, a serious dispute about the inventive concept of the claims in question. In resolving that dispute, assistance is to be obtained from certain passages in the judgment of Jacob LJ in Pozzoli:
"17. What now becomes stage (2), identifying the inventive concept, also needs some elaboration. As I pointed out in Unilever Plc v Chefaro Proprietaries Ltd [1994] R.P.C. 567 at 580:
"It is the inventive concept of the claim in question which must be considered, not some generalised concept to be derived from the specification as a whole. Different claims can, and generally will, have different inventive concepts. The first stage of identification of the concept is likely to be a question of construction: what does the claim mean? It might be thought there is no second stage—the concept is what the claim covers and that is that. But that is too wooden and not what courts, applying Windsurfing stage one, have done. It is too wooden because if one merely construes the claim one does not distinguish between portions which matter and portions which, although limitations on the ambit of the claim, do not. One is trying to identify the essence of the claim in this exercise."
18. So what one is seeking to do is to strip out unnecessary verbiage, to do what Mummery L.J. described as make a précis.
19. In some cases the parties cannot agree on what the concept is. If one is not careful such a disagreement can develop into an unnecessary satellite debate. In the end what matters is/are the difference(s) between what is claimed and the prior art. It is those differences which form the "step" to be considered at stage (4). So if a disagreement about the inventive concept of a claim starts getting too involved, the sensible way to proceed is to forget it and simply to work on the features of the claim."
- Jacob LJ came back to this in his judgment in Actavis v Novartis [2010] FSR 18 at [19]:
19. I would only add an extra word about step 2—identifying the inventive concept. It originally comes from Oliver L.J.'s formulation of the approach in Windsurfing International Inc v Tabur Marine (Great Britain) Ltd [1985] R.P.C. 59 at 73. Strictly, the only thing that matters is what is claimed—as Lord Hoffmann said in Conor Medsystems Inc v Angiotech Pharmaceuticals Inc [2008] UKHL 49; [2008] RPC 28 at [19]:
"The patentee is entitled to have the question of obviousness determined by reference to his claim and not to some vague paraphrase based upon the extent of his disclosure in the description."
- The provisions of section 125(1) Patents Act 1977 should also be borne in mind in providing that "an invention" shall, unless the context otherwise requires,
"be taken to be that specified in a claim of the specification …… as interpreted by the description and any drawings contained in that specification….."
- Mr Mitcheson submits that the analysis of the inventive concept must be based on the wording of the claim itself. The passages from Pozzoli and Actavis v Novartis demonstrate that it is the inventive concept of the claim which is crucial not what is contained in the specification. He submits that section 125 is entirely consistent with that approach. The reference by Jacob LJ to the possibility that the first stage (construction/what does the claim mean) might be too wooden is directed at the exclusion of irrelevant material ("excessive verbiage") resulting in the sort of précis indicated by Mummery LJ. It is not directed at admission of material contained in the specification in order to extend the scope of the inventive concept which is to be derived from the claim; but this is precisely what Mr Mitcheson says PDI seeks to do on this appeal.
- I agree with Mr Mitcheson that the correct approach is as he submits; it is the approach adopted by the HO who based her analysis on the wording of the claim. It is still necessary, of course, to construe the claim in order to arrive at the inventive concept based on that claim.
- In reaching her conclusion on inventive concept, the HO referred to the way in which the examiner, as recorded in [15] of the Decision, identified the inventive concept. It was a method for identifying a CNV that is relevant for a phenotype by identifying CNVs in a subject using a method capable of achieving a resolution of 30kb to 50kb and comparing these variants to the results of copy number variations in a "normal" population of at least 1000 individuals without the phenotype. She recorded Dr Broughton's disagreement with that assessment of the inventive concept, stating this summary of the inventive concept in this way:
"the use of a reference set of at least 1000 individuals without a phenotype in order to eliminate all normal polymorphic CNVs from a set of CNVs obtained in a high resolution genome wide screen from an individual in order to identify a CNV associated with a phenotype in that individual i.e. you can use a reference set which is large enough to exclude polymorphic CNVs, and that reference set is 1000 individuals."
- Then, after referring to Dr Broughton's citation from the speech of Lord Walker in Generics (UK) Limited and others v H Lundbeck A/S S [2009] UKHL 12, [2009] RPC 13 to the effect that "inventive concept" is concerned with the identification of the core (or kernel, or essence) of the invention, she went on to express her view that the core of the invention was
"the use of a reference set of CNVs from a population of 1000 individuals without a particular phenotype in order to eliminate "normal" CNVs identified as being present in a high resolution screen in an individual with a particular phenotype through a comparison of the two sets of data, thereby identifying a CNV in the individual which is relevant to the particular phenotype."
- On this appeal, Dr Broughton submits that the Hearing Officer has defined the inventive concept too narrowly. He says that she failed, at least explicitly, to included the "important and inherent feature that the reference set of CNVs is chosen to be of an adequate size to contain substantially all polymorphic CNVs present in the population". She has also failed to acknowledge that the purpose of the invention "is to eliminate a large majority of polymorphic CNVs (so that the CNV of interests, i.e. the "needle in the haystack") can be determined". In the light of those failings, he now states the inventive concept slightly differently from the way in which the HO recorded him as having stated it, as follows:
"The use of a reference set of CNVs from a population of 1,000 individuals without a particular phenotype, said reference set including substantially all normal polymorphic CNVs found in the population, in order to eliminate a large majority of polymorphic CNVs identified as being present in the set of CNVs by high resolution screening in an individual with the said phenotype, through a comparison of the two sets of data, such that the CNV which is relevant to the phenotype in that individual can be determined."
- As I understand that formulation, the reference to "a large majority of polymorphic CNVs…..said phenotype" is a reference to a large majority of the entire field of polymorphic CNVs which exist in the whole population, some of which may be found in the individual. In other words, the inventive concept envisages a reference set including substantially all polymorphic CNVs; a CNV which is comprised within that substantial proportion of all polymorphic CNVs will be eliminated from the set of CNVs found in the phenotypical subject. The result will be that a large majority of polymorphic CNVs actually present in the phenotypical subject will be eliminated. But it will be a large majority rather than all such CNVs because the individual may possess a CNV which is polymorphic although not found in the reference set (which only comprises substantially all polymorphic CNVs).
- I note at this point two common features of all of these formulations of the inventive concept and Claim 1 itself. The first is the reference to a population and, in relation to that population, to a figure of 1,000: the claim itself refers to "at least 1000 human individuals" as do the formulations of the examiner, the HO herself and Dr Broughton as recorded in [16] of the Decision. The formulation which he now propounds refers to 1000 individuals.
- The second is that all of them refer to a single target subject with a given phenotype and not to a group of persons with that phenotype. It is not part of the inventive concept to carry out a comparison of more than one person with the phenotype under consideration with a control group, an exercise which might, in some cases, also help in the identification of a CNV consisting of a particular phenotype.
- I further note that, whilst the HO did not express the purpose of the invention as being to eliminate a large majority of polymorphic CNVs (so that the CNV of interest, ie the "needle in the haystack", can be determined), she did acknowledge that the purpose was to eliminate the normal CNVs found in the high resolution screen in the phenotypic individual (those normal CNVs being the ones found in the reference set).
- If it is right, as I have held it to be, that the inventive concept of Claim1 is to be derived from the claim alone (properly construed), then Mr Mitcheson submits that the HO's formulation is to be preferred to that of Dr Broughton. He submits that neither of the features on which Dr Broughton relies is to be found, expressly or implicitly, in Claim 1. There is nothing in the claim taken by itself which suggests that the reference set is to be chosen with a view to containing substantially all polymorphic CNVs found in the population. Still less is there anything to suggest that a reference set of 1000 would be sufficient. I agree with that. The meaning of Claim 1 as a matter of ordinary language is clear. Accordingly, if the inventive concept is to be derived from the claim alone, as in my judgment is the correct approach, the HO was right in her identification of the inventive concept.
- Would the position be any different if it were permissible to discover the inventive concept by adding to the claim material derived from the specification? Mr Mitcheson submits that an analysis of the specification demonstrates that there is no proper basis for the assertion that 1,000 human individuals is of "adequate size to contain substantially all polymorphic CNVs present in the population", nor that the comparator claimed will "eliminate a large majority of polymorphic CNVs" based on a resolution of 30kb to 50kb. He refers to various parts of the specification which I have described above. He notes in particular (i) that, whilst there are various references throughout the specification to the size of the reference pool needed to produce the required result, generally the number of individuals to be used as part of the reference population ranges from 100-50,000 and (ii) that the only specific reference to a database of 1,000 individuals is the one I have mentioned at paragraph 11 viii) above but even that is directed at a database of normals to be tested of 10,000 individuals. He also refers to Examples 2 and 3 to demonstrate that the reference sets under consideration comprise 10,000 individuals.
- Overall, he submits that it is clear from the specification that there are a large number of CNVs present in the population, with only relatively few of these being causative of a certain disease or condition. The specification allows for the comparison of the CNVs from an individual with a database of CNVs from a number of individuals. But the number of individuals required for each comparison is unknown. In that context he refers to paragraphs [0065]-[0068] and to the Table 1 on page 15. I have already dealt with Mr Mitcheson's submission on these paragraphs at paragraph 13 above, expressing my agreement with him.
- This last point, about the number of individuals being unknown, is highlighted when one considers the divisional claim in relation to non-human subjects. It is part of the CGK that different animals have different sized genomes. Given the variance in the size of the genome, as well as the varying frequency of the particular CNV in issue, the size of the database required would be different in relation to human and non-human subjects.
- Dr Broughton does not accept Mr Mitcheson's construction of Claim 1. He maintains that the claim, properly construed, does contain both of the features in dispute. There are two steps in his argument.
- Firstly, he says that it is apparent from the teaching of the application document that the goal is to identify a CNV of relevance to a phenotype (see for example paragraphs [0021] and [0022] as well as paragraph [00153] of the application specification). It is further apparent, and it is common ground, that when using high resolution screening methods, more CNVs would be identified than in the lower resolution screens of the prior art. It is also made clear in the specification, as I accept, that genome wide analyses used in the invention may reveal multiple CNVs. Since only a single CNV would be expected to be associated with a phenotype in any one individual, it is therefore apparent that it would be necessary to eliminate a majority of polymorphic CNVs identified in that individual in order to identify the CNV of relevance. I do not know whether it is accepted by Mr Mitcheson that it would be expected that only a single CNV would be expected to be associated with a phenotype but, for present purposes, I will assume that that is to be expected.
- Secondly, the skilled person, having read and understood this goal, would understand that it would be necessary for the reference set to be of an adequate size to contain substantially all polymorphic CNVs present in the population. If this was not the case, then the method would not operate to eliminate substantially all polymorphic CNVs. The consequence would be that, from the pool of CNVs identified in an individual, only some of the benign CNVs could be eliminated, thereby not allowing for the identification of the relevant CNV.
- As to the first of those steps, some consideration needs to be given to the meaning of "relevant" within the meaning of the opening words of Claim 1 (which refers to identifying a "relevant" CNV for a phenotype) and of "relevance" in the paragraphs mentioned by Dr Broughton. Ultimately, what Dr Broughton would no doubt say is really of interest is to find a single non-polymorphic CNV in the individual with the phenotype. Although it is not the case that the phenotype is necessarily caused by the single, non-polymorphic, CNV found, that CNV is nonetheless "relevant" for that phenotype in the context of the specification and the claim. However, on the basis that only a single CNV would be expected to be associated with a phenotype, if anything is apparent it is not, so it seems to me, that it would be necessary to eliminate a majority of polymorphic CNVs identified in that individual; rather, it is that it would be necessary to eliminate all of them if the invention is to identify a single relevant CNV.
- That is not to say that eliminating a majority would not be helpful. One might, for instance, eliminate all but two CNVs, a result which might well provide useful information to a researcher: the elimination of a large majority (whatever that may mean) of CNVs puts one on the road to the destination of identifying the relevant, potentially causative, CNV. But it does not get one to that destination. It is obvious, if I may use that word other than as a term of art, that no particular size of population (other than perhaps the whole world) can guarantee to disclose all polymorphic CNVs; and thus it will never be possible to guarantee that a CNV in a subject with a phenotype is a relevant CNV by use of the method of the claimed invention. Indeed, this is apparent from the specification: see for instance paragraph 12 above.
- This leads on to the second step of the argument. Dr Broughton says that it is apparent to the skilled person that it would be necessary for the reference set to be of an adequate size to contain substantially all [my emphasis] polymorphic CNVs present in the population. In saying that, he recognises the obvious point that a larger reference set may contain more polymorphic CNVs than a smaller set: the probability of capturing all polymorphic CNVs increases as the size of the reference set increases and it must, in my judgment, be accepted that the skilled person would understand that obvious point.
- There is, however, nothing in the teaching of the application which gives any indication of what the size a reference set would need to be in order to produce the elimination of any particular percentage of polymorphic CNVs. Nor is there any teaching that a reference set of 1,000 would result in the elimination of a large majority – query 90%, 99.9% – of polymorphic CNVs since there is no teaching that a reference set of 1,000 would contain any particular percentage of all polymorphic CNVs. Although this might be seen as going principally to support under section 14(5)(c), I see it going as well to identification of the inventive concept.
- It seems to me, in the light of those considerations, that the highest, from PDI's perspective, at which the inventive concept to be derived from the specification can be put is the elimination of CNVs by a comparison of the data derived from a screen in relation to the phenotypic subject with the data derived from the control group, coupled with the (obvious) idea that the larger the control group, the larger the number of polymorphic CNVs likely to be disclosed and, therefore, the greater the number of polymorphic CNVs which can be eliminated from the CNVs found in the subject.
- Adding those features derived from the specification to Claim 1 itself, the inventive concept at which one arrives is indistinguishable from the inventive concept as stated by the HO.
- In the light of the above discussion, I reject Dr Broughton's formulation of the inventive concept. I do not consider that the HO was in error on this point.
Inventive step
- Before coming to step 4 of the Pozzoli approach, I wish to address what Mr Mitcheson refers to as the arbitrary nature of the selection of a control set of 1,000 individuals. To put the point in context, Dr Broughton's approach is to identify the problem and then to explain that the invention is a solution to the problem. Put very briefly he says this:
i) The problem identified by him arises this way:
a) It is desirable to use higher resolution scanning because this identifies more CNVs and thus produces a greater chance of finding a CNV which is causative of a phenotype. The CNV may simply be too small to find with lower resolution systems.
b) The revelation of more CNVs leads, however, to the problem of what he describes as large background noise. In other words, there are many CNVs found but it is not known which, if any, of them is relevant.
c) This was not, he submits, a problem in the prior art because the number of CNVs discovered in any particular individual using previously available techniques was small.
ii) The solution is provided by the invention which is to look at 1,000 individuals who do not have the phenotype and to compare the CNVs in the phenotypic individual under investigation with the results derived from the control set. The conclusion to be arrived at is the polymorphic CNVs in the control set are not causative but that a CNV found in the individual but not in the control set may be causative.
- In this context, Dr Broughton submits that the selection of 1,000 individuals is not in any sense arbitrary. He says that this is merely a preferred embodiment which is mentioned repeatedly throughout the specification. He says that the inventors have found (although I note that this is not reported in the specification) that a reference set of 1,000 individuals provides for a successful result for almost all individuals tested since only very rare polymorphic CNVs are not represented. But since the skilled person would understand that even a set of 1,000 individuals cannot be guaranteed to include absolutely every polymorphic CNV, the specification describes further embodiments with larger reference sets.
- Further, it is said that the skilled person, carrying out the method described in the specification is taught that, in general, a database of 100 individuals is adequate, but would expect 1,000 to be better. Accordingly, although PDI believes there is a valuable and inventive contribution even at the level of 100 individuals, in these applications it has chosen to claim a preferred embodiment of even greater value.
- The invention is, it is said, a new way of investigating CNVs. The control set of 1,000 individuals is not arbitrary but is an optimisation of where the invention works best.
- Mr Mitcheson submits, on the contrary, that there is nothing in the specification that places any significance upon the choice of having a database comprising at least 1,000 individuals: the claimed selection of a database of 1,000 individuals for a CNV resolution of 30kb to 50kb is arbitrary. He also submits that there is no basis for claiming that a database of this size will solve the problem alleged to be solved by the Applications: there is no basis of claiming that the choice of 1,000 individuals would provide a database that would allow for all polymorphic CNVs in a population to be represented.
- If that is correct, then it follows, according to him, that the selection of a database of 1,000 individuals in the claim cannot contribute to the inventive step. He relies on what Jacob LJ said in Actavis v Novartis at [36] to [37]:
"36. Another aspect of obviousness which is not readily answered by the PSA is illustrated by the five-and-a-quarter inch plate paradox. This runs like this. Suppose the patent claim is for a plate of diameter five-and-a-quarter inches. And suppose no one can find a plate of that particular diameter in the prior art. Then (a) it is novel and (b) it is non-obvious for there is no particular reason to choose that diameter. The conclusion, that the plate is patentable, is so absurd that it cannot be so.
37. What then is the answer to the paradox? It is this: the five-and-a-quarter inch limitation is purely arbitrary and non-technical. It solves no problem and advances the art not at all. It is not inventive. And although "inventive step" is defined as being one which is not obvious, one must always remember the purpose of that definition—to define what is inventive. That which is not inventive by any criteria is not made so by the definition. Trivial limitations, such as specifying the plate diameter, or painting a known machine blue for no technical reason are treated as obvious because they are not inventive."
- Although that was said in the context of the "Problem-Solution-Approach", it seems to me to be relevant, too, to whatever approach to obviousness is taken (including the Pozzoli approach) if that approach lays down a formulaic test the application of which leads to an absurd result.
- Mr Mitcheson also relies on the decision of Kitchin J in Abbott Laboratories Ltd v Evysio Medical Devices ULC [2008] EWHC 800 (Ch), [2008] RPC 23, a case concerning coronary stents. The Judge said this at [181]:
"… there is no invention in stipulating a feature which is arbitrary and serves no useful purpose. It has long been established that a patent cannot be used to prevent a person from doing what is merely an obvious extension of what has been done or was known in the art before the priority date. The public are entitled to make obvious products using obvious and ordinary techniques. The selection of a number of these products by reference to an arbitrary parameter which has no technical significance does not involve an inventive step and does not create a patentable invention. It involves no technical ingenuity and solves no technical problem."
- Mr Mitcheson submits that the selection of a database of 1,000 individuals in the present case is equivalent to a five-and-a-quarter inch plate. There is no justification in the specification for such a database contributing anything technical or inventive to the claim, and no technical problem has been solved by its inclusion in the claim.
- In addressing that submission, one must not lose sight of the paragraph of Jacob LJ's judgment following the one just quoted where he said that the PSA does not assist in providing an answer to the paradox: "This is for the simple reason that there is no problem and so no solution to it". But if there is a problem, there may be a solution, and if there is a solution, whether it involves an inventive step must be ascertained according to the established tests. And this is as true when applying Pozzoli as when applying the PSA.
- Further, Kitchin J was addressing arbitrary features in the context of an extension of the prior art. If there is an entirely new inventive concept to be found in a claim, the inclusion of a feature which may appear to be arbitrary, in the sense that it is not supported by the specification by experimental data, does not mean that there is no inventive step. The inclusion of that feature may be necessary to distinguish something which may work from something which clearly does not. Take the present case, for example, where the skilled person would not expect a reference set of 1 person to be of any use at all but where there is some number (be it 1,000, 10,000 or 50,000) where the skilled person would recognise that the comparison could be of real assistance in identifying a causative CNV. The adoption of a figure of 1,000 would not detract from the inventive step. Further, unless the skilled person had reason to doubt that the selected figure (ie 1,000) would result in information of any practical use, there would be nothing in a suggestion that the invention lacked support given the contents of the specification. The absence of experimental data would not matter given that there has been no identification of any technical prejudice to overcome.
- In contrast, if the only difference between the inventive concept of the claim and the prior art (or an obvious extension of the prior art) is a feature which is arbitrary, then there is no inventive step. Taking the present case again, if the inventive concept of Claim 1 apart from the specification of the size of the control group was apparent from (or was an obvious extension of) the prior art, then the additional feature that the control group should be at least 1,000 individuals does not add anything inventive.
- It follows that Mr Mitcheson's argument in relation to the arbitrary nature of the 1,000 number for the reference set only gets him home if the underlying idea of the invention was apparent from, or obvious over, the prior art.
Differences between prior art and inventive concept
- It seems to me, therefore, that the question which now needs to be addressed is whether the use of a reference set of individuals without a particular phenotype (as an idea rather than related to any particular size of reference set) in order to eliminate "normal" CNVs identified as being present in a high resolution screen in an individual with a particular phenotype through a comparison of the two sets of data is obvious. If it is not obvious, then there is an inventive step under step 4 of Pozzoli.
- As to that, Mr Mitcheson says that all that is claimed is the notion that sample size must be increased as resolution is increased in order to differentiate normal subjects from those with the phenotype of interest. That concept was, he says, obvious from the prior art. Whilst that is one thing that is claimed, I do not think that is all that is claimed. There is potentially a prior point which is whether the prior art reveals a comparison at all for the purpose of identifying a CNV.
- This is the point described at greater length by the HO at [22] of the Decision, namely, "whether or not the prior art disclosed or suggested a solution to the problem of eliminating polymorphic CNVs present in the normal population from genomic screens in order to identify the particular CNV associated with a particular phenotype in an individual, or to identify what [Dr Broughton] called the "needle in the haystack".
- The HO referred to the prior art (other than D4) at [20] and [21] of the Decision. She referred to the examiner's identification of the differences between the prior art and the claimed invention: the present method uses a technique capable of achieving a resolution of 30kb to 50kb whereas the prior art methods describe resolutions of ~1Mb (referring to D1 and D2 respectively)) or 100kb – 1 Mb (referring to D3) and that the number of "normal'" subjects without the phenotype is at least 1,000 compared to 35 in D2 and a "reference pool" in D3.
- It is, I consider, useful to set out what the HO said in [21] of the Decision about Dr Broughton's submissions:
i) She recorded that Dr Broughton did not appear to dispute the differences between the cited prior art and the claimed invention identified by the examiner. However, he emphasised that D1 did not identify that it was possible to use a reference set to eliminate polymorphic CNVs, instead merely identifying that there is a problem with polymorphic CNVs being present. He makes the same point to me.
ii) She also recorded that he accepted that the use of high resolution screens was suggested, a proposition which Dr Broughton accepts before me but subject to the point that this was coupled with a warning in D1 that to do so would make the analysis more complex.
iii) She recorded his submission that D3 merely identified that there is a problem with polymorphic CNVs being present and that these CNVs needed to be excluded. That remains his submission before me, saying that no solution is offered to the problem.
iv) Dr Broughton noted, she observed, that D3 also failed to disclose the use of high resolution genome screens but agreed that it proposed methods which compare measured variations to parental samples or independent normal controls.
v) Finally, she recorded Dr Broughton's explanation that D2 utilised a small reference set to eliminate some more common CNVs which were not related to the brain (the tissue being investigated) but failed to disclose that a larger reference set could be used to eliminate all polymorphic CNVs, a position he maintains before me.
- The HO said nothing more about the differences in this section of the Decision.
- Dr Broughton now submits that the HO made three errors of law:
i) Ignoring the inventive concept identified by her, she repeated the examiner's identification of the differences between the cited prior art and, as he puts it in his skeleton argument, "the stark wording of the claim". There was no attempt to identify the differences between the cited prior art and the "inventive concept", as required by step (2) of Pozzoli.
ii) Secondly, the HO identified the differences between the invention and the prior art as a whole. There was no attempt by her to address each cited document separately to determine the differences from the invention, and then to determine whether the differences would have been supplemented by the skilled addressee's CGK. Moreover, there has been no explanation of why the cited prior art documents would have been combined in the search for a solution to the problem being solved.
iii) Thirdly, the Hearing Officer's analysis was incorrectly influenced by the benefit of hindsight, possibly as a result of an incorrect application of the Pozzoli approach.
- Taking those in turn, Dr Broughton submits that, by ignoring the inventive concept, in favour of the "stark wording" of the claims, the relevance of the thought processes and understanding of the authors of the prior art, and the person skilled in the art was lost from the analysis of inventive step. This legal error has led to the purpose of steps taken in the prior art and in the invention being incorrectly equated (in terms of their intended function), by the HO.
- I do not accept that submission. In the light of my findings about inventive concept, I do not think that she did anything other than to address the question of the differences on a correct view of the inventive concept.
- Nor do I do think that the criticism behind the second alleged error is entirely fair. What the HO did in [20] of the decision was to identify the differences between the cited prior art and the invention, differences which she recorded in [21] and which, according to her, Dr Broughton did not appear to dispute. The differences were as set out above. One difference was a difference in the resolution under methods of the claimed invention and the resolutions under the prior art methods. However, the resolutions in each piece of prior art are described and it can be seen clearly what the difference is in relation to each piece of prior art viewed on its own. The other difference related to the number of normal subjects being at least 1,000 in the claimed invention compared to 35 in D2 and a reference pool in D3. Again, the difference between the claimed invention and each individual piece of prior art is identified. The remainder of the asserted second error of law really goes not to the differences themselves, but to the way in which those differences were used in carrying out step 4 under the Pozzoli approach.
- Those identified differences are not, however, the only matters of relevance. As appears from [21] of the Decision, Dr Broughton made other observations on what the prior art did and did not identify as compared with the claimed invention. I have listed these matters in paragraph 71 above.
- As to the third alleged error, hindsight, Dr Broughton relies, at least to some extent, on an incorrect application of the Pozzoli approach.
- Dr Broughton has carried out a helpful analysis of what he now sees as the relevant differences between the cited prior art individually, and the inventive concept which I can deal with as follows, setting out Mr Mitcheson's observations.
- D1, Pinkel and Albertson – Dr Broughton's analysis:
i) Although this discusses the potential of high resolution screening of individuals, and recognises that this will increase the number of CNVs identified (and in turn that that will mean that interpreting the primary data will become more complex) there is no reduction to practice of the suggestion and no discussion or recognition of the scale of the increased complexity that should be anticipated. There is no teaching that high resolution screening would obscure the determination of relevant CNVs from amongst a large majority of polymorphic CNVs. The document does not, therefore, teach or describe a process for determining a CNV associated with a phenotype from amongst a set of CNVs identified at high resolution. Other than in the very general terms noted above, the document simply did not address the issue of eliminating polymorphic CNVs from a set to arrive at a causative CNV of interest.
ii) Because of that, the document further differs from the invention in that it does not suggest the use of a reference set containing substantially all polymorphic CNVs in order to eliminate those polymorphic CNVs from a high resolution set.
iii) Since the document did not suggest such an approach, axiomatically, it also did not suggest that the reference set should contain CNVs from at least 1,000 individuals.
- Since I have rejected Dr Broughton's analysis of the inventive concepts, paragraphs ii) and iii) above fall away. In addition, Mr Mitcheson submits that the skilled person would appreciate that the problem identified was that "more results lead to more noise": the skilled person would, he says, appreciate that a concept of comparing the result for a subject with the results for a group.
- D2, Wilson et al. - Dr Broughton's analysis:
i) This differs from the claimed invention in that low resolution screening was used to identify CNVs in individuals with a phenotype. It contains no disclosure of the use of high resolution screening to identify a set of CNVs, and because low resolution techniques were used there is no individual in which more than a single CNV is identified. Accordingly, D2 further differs from the invention in that there is no disclosure of an attempt to eliminate a large majority of polymorphic CNVs from a set of CNVs identified in an individual (at high resolution). Moreover, since there was no attempt to eliminate such CNVs there was no attempt to do so by comparison to a reference set which includes substantially all polymorphic CNVs.
ii) Dr Broughton accepts, of course, that D2 does discuss that the 3 CNVs identified were not present in 35 "normal controls" (ie individuals without the phenotypes under investigation).
iii) However, there is no indication, he says, that the authors believed that the 35 controls would have been adequate to contain substantially all polymorphic CNVs, and in view of the incomplete knowledge about the "picture of this normal variation" and the lack of understanding of the "exact frequency in different ethnic groups", the authors could not have known about the potential coverage of polymorphic CNVs in 35 individuals. There is no teaching, therefore, in the use of these controls, of the idea of using the type of reference set employed in the invention.
- Again, Dr Broughton's argument falls away insofar as it relies on the two features which I have held do not form part of the inventive concept. Further Mr Mitcheson does not agree with this analysis of D2. He accepts that the initial experiment did not get anywhere much, and that the authors then adopted a different approach. But he points out, correctly, that they were using of sample of normal people to eliminate CNVs causing abnormality. A comparison was made of the subjects showing a phenotype with normal persons, with the normal results being discarded to establish the abnormal. Although the paper shows the use of mixed techniques, it does show the use of controls. The HO was, he says right, in what she said at [21] of the Decision.
- Vissers et al (2005) – Dr Broughton's analysis:
i) This differs from the claimed invention in that low resolution screening was used to identify CNVs in individuals with a phenotype. Because, in the actual work reported, low resolution techniques were used there is no individual in which more than a single CNV is identified. Accordingly, it is said that this document further differs from the invention in that there is no disclosure of an attempt to eliminate polymorphic CNVs from a set of CNVs identified in an individual (at high resolution). Moreover, since there was no attempt to eliminate such CNVs there was no attempt to do so by comparison to a reference set which includes substantially all polymorphic CNVs.
ii) Dr Broughton observes that the document does state that: "It is essential to rule out such submicroscopic variation by studying parental samples and/or independent normal controls…" However he says that the closest the authors come, in practice, to ruling out submicroscopic variation (polymorphic CNVs) is to compare identified CNVs to those from parental samples. Parental samples (where, for any individual, there can of course only be a maximum of two) could not be expected to include substantially all polymorphic CNVs, and their use is not therefore the same as, or functionally equivalent to, the use of the reference set in the claimed invention. Rather, and as stated by D3, their use is to determine whether CNVs have arisen de novo or were inherited. CNVs inherited from an unaffected parent are not expected to be causative of the phenotype of the affected child.
iii) There is no development or explanation of the suggestion to use "independent normal controls". Dr Broughton therefore submits that this stand-alone statement simply does not disclose the use of a reference set containing substantially all polymorphic CNVs.
- Mr Mitcheson submits that the statement in relation to the need for independent normal controls is important and cannot simply be dismissed. What the statement tells the reader is that normal CNVs are to be ruled out or eliminated. That, it is obvious he would say, can be done by taking samples to establish what is normal.
- Dr Broughton submits that the differences which he has identified were not obvious over any of the cited documents taken separately with or without CGK. Further, care must be taken in combining the cited documents. But even if they are taken together and with CGK, the claimed invention was still not obvious. He has carried out a detailed analysis in support of his submission in relation to each document.
- In doing so, he has, of course, identified the differences on which he relies by reference to his formulation of the inventive approach. Since I have rejected his formulation, I consider that parts, at least, of his analysis of obviousness is flawed. In particular, in relation to each piece of cited prior art, reliance is placed on an absence of any suggestion of the use of a reference set containing substantially all polymorphic CNV's in order to eliminate those polymorphic CNVs from the high resolution set.
- Since Mr Mitcheson relies on the Decision, I find it helpful at this stage, rather than considering the detail of Dr Broughton's analysis, to see what the HO had to say about this aspect of her case when addressing what flowed from her (correct) identification of the inventive concept taking into account the observations made by Mr Mitcheson which I have already mentioned in relation to each piece of cited prior art. Having done that, I will consider how it is said by Dr Broughton that she has fallen into error. I start with what she records Dr Broughton as having accepted (and which remains his position as I understand it):
i) The skilled person was aware of high resolution screening methods from the prior art and of the fact that such high resolution screens may identify more CNVs that lower resolution screens (referring to D1).
ii) The skilled person would know that polymorphic CNVs present in the normal population would complicate the analysis of results from high resolution DNA screening methods.
iii) The prior art disclosed methods which could be used to eliminate some common CNVs, for instance by comparing results to the parental genome or to normal controls seen in D3 or by using a "small" population of controls as in D2.
iv) The high resolution screen in itself was not inventive.
v) Selecting the level of resolution of 30kb to 50kb leads to the problem which the claimed invention is said to solve ie you will identify many more CNVs at this resolution, and thus you are left with the issue of finding the "needle in the haystack".
vi) I add in relation to that last feature, that, in [21] (as I have already recorded) she noted that Dr Broughton himself had acknowledged that D1 identified that there was a problem with polymorphic CNVs being present as did D3. Since a problem was identified in D1, that problem must be taken as a known feature in assessing the obviousness of a claimed invention over D1: the invention is not something which identifies, before going on to solve, a hitherto unappreciated problem. The same applies, mutatis mutandis, to the problem identified in D3. The problem identified in D3 is reflected in particular by the passage already included in the quote at paragraph 22ii) above:
"It is essential to rule out such submicroscopic variation by studying parental samples and/or independent normal controls before drawing any firm conclusion on whether an aneusomic segment [a CNV] is causative for the disease under investigation."
- The HO then identified, in [24] of the Decision, the arguments presented by Dr Broughton:
i) The skilled person had no knowledge of the number of CNVs present in the normal population.
ii) Accordingly, they would not conclude that a reference set could be used to eliminate normal polymorphic CNVs or have any idea how big such a reference set would need to be.
iii) Contrary to the examiner's suggestion, the inventors had not merely increased the number of controls to increase statistical significance.
iv) Given those factors, the skilled person would have no motivation to derive a method as claimed and it would not be obvious to try as the skilled person would not know that they could succeed.
- Before moving on to later paragraphs in the Decision, I note, in relation to subparagraph i) of the preceding paragraph, that that there is nothing in the application either to say how many CNVs there are in the population and so there is no indication of how large a control group should be. It has to form part of Dr Broughton's case in support of his version of the inventive concept that the skilled person, without any experimental information in the specification, would appreciate a control group of 1,000 to be sufficiently large to encompass a large majority of polymorphic CNVs. If that is so, subparagraph ii) would not be the case. Even if the skilled person would not pick on 1,000, he or she might pick on, say, 5,000. But then the point of principle is made namely that once the concept of a comparison was present in the mind of the skilled person, the skilled person could be confident of selecting a reference population which would lead to success.
- In [25] the HO noted that the lowest number for the sample population was 100 "an order of magnitude lower than the number of individuals specified in the claim". That figure, she noted was only around three times as many as the number used in the experiments found in the prior art (eg D2). She noted that the description failed to include any specific examples which demonstrated the use of the claimed 1,000 individuals in the reference population used in the comparison step.
- In the light of what she had said in the preceding paragraphs, she identified, in [27] the question for her as being whether the use of the control population of 1,000 individuals without the phenotype to eliminate "normal" CNVs in order to identify a CNV which is relevant to the particular phenotype, was inventive. Then in [28] she effectively repeated as being well known from the prior art the matters which she had recorded Dr Broughton as accepting.
- Then came the meat of her decision, where, in [29] she concluded that, in the light of the knowledge from the prior art, the skilled person would have had ample motivation to try increasing the number of individuals in the normal control population when using higher resolution screening methods. It would be obvious to them that this would filter out more of the polymorphic CNVs associated with the normal population and help them to identify CNVs which were relevant to a particular phenotype in an individual. As Mr Mitcheson puts it, it cannot be inventive to seek to increase the size of the sample of the normal population to achieve this: it is basic experimental methodology and statistics.
- The HO held that the reference in the claim to using data from 1,000 individuals in the comparison step did not add an inventive step to the claimed method. Although the prior art did not specifically suggest using control populations containing 1,000 individuals (she described D2 as referring to the use of a control population of 35 CNVs to exclude common CNVs), she repeated that the skilled person would be motivated to try increasing the number of individuals in the normal control population. In that regard, she noted the lower limit of 100 persons in the application as originally filed, rather than the 1,000 in the application before her. She again noted the absence of any specific examples demonstrating the use of 1,000 individuals, concluding that the claimed population of 1,000 individuals had not been shown to be any more advantageous than the other reference population sizes specified in the description.
- Dr Broughton submits that, in her discussion of inventive step, the HO was wrong to take account (as she did in [25] and [29] of the Decision) of size of the reference set (100) in the application as originally filed. He says that the HO's reliance on this in support of a finding of lack of inventive step is erroneous.
- Firstly, he says it is the invention, as claimed, that should have been examined, rather than the content of the specification. Section 1(b) Patents Act 1977 requires that it is the "invention" that should involve an inventive step, and Section 125(1) Patents Act 1977 defines the invention as "that specified in a claim". I do not disagree with that proposition.
- It does not follow, however, that the HO should have ignored the size of the reference set in the original application. Her reference to the size of that set in [25] was made in considering the arguments which she had identified in [24]. And her reference to the size of that set in [30] was made in the context of the motivation of the skilled person to increase the number of individuals in the normal control population. In both instances, it was coupled with the observation concerning the lack of any examples which demonstrated the use of the claimed 1,000 individuals in the reference set. What she drew from this was that the claimed population of 1,000 has not been shown to be any more advantageous than the other reference population sizes specified in the description. I do not consider it was wrong of her to refer to the reference set of 100 individuals in reaching that conclusion.
- Secondly, Dr Broughton submits that the Hearing Officer's reference to 35 controls in D2 is misplaced since those controls served a different purpose from the reference set of the claimed invention. He says that the purpose was not clear from D2 itself, but what can be said is that the 35 controls were not intended as a reference set that included all polymorphic CNVs used to exclude polymorphic CNVs from a set of CNVs identified in an individual. That may be true: but nor does the specification in the present case identify that as the purpose of the reference set. And for reasons already given at length, the inventive concept, whether identified by reference to the claims alone or by reference to the entire specification, does not include either of the features (i) that the reference set of CNVs is chosen to be of an adequate size to contain substantially all polymorphic CNVs present in the population or (ii) the elimination a large majority of polymorphic CNVs (so that the CNV of interest ie the "needle in the haystack" can be determined). What D2 does show, in my view and in agreement with Mr Mitcheson, is the elimination of some CNVs (Dr Broughton would say the very commonest, but I do not think anything turns on that) by a comparison of data derived from a control group. But that, coupled with the idea that the larger the control group, the larger the number of polymorphic CNVs likely to be disclosed (and, therefore, the greater the number of polymorphic CNVs which can be eliminated from the CNVs found in the subject) is precisely the inventive concept which, in my view, the claims display. Indeed, as Dr Broughton puts it in his skeleton argument
"…The Hearing Officer's failure to determine the difference between D2 and the inventive concept (rather than between D2 and the stark wording of the claims), together with an unallowable application of hindsight, resulted in this important distinction [ie between the purpose of the control in D2 and the purpose of reference set in the claimed invention] being missed from the analysis."
- That point largely falls away, however, when it is accepted, as in my judgment it must be, that the stark wording of the claim is the real indicator of the inventive concept.
- I reject Dr Broughton's criticisms of the HO's approach. I consider that she reached the correct conclusion for the reasons given. In any case, the prior art disclosed methods which could be used to eliminated some common CNV as identified by the HO at [23] of the Decision. Mr Mitcheson refers in particular to the D2. He describes this as the use of a sample of 35 controls as a comparator for a study with resolution of 1.4mb a description with which, in spite of Dr Broughton's characterisation of this experiment, I agree. I also agree with Mr Mitcheson that it would have be obvious to the skilled person that using a higher resolution method, which would generate more results, would necessarily require a larger number of controls.
- Further, in D3, higher resolution screens (at 100kb so not as high resolution as those contemplated by the application) are specifically mentioned and the problem of ruling out normal polymorphisms is said to be "essential". That could only be done by comparing a sufficient number of parental samples or independent normal controls.
- Given that the inventive concept does not, in my judgment, include either of the two features contended for by Dr Broughton (ie a reference set containing substantially all polymorphic CNVs and the elimination of a large majority of polymorphic CNVs), the real issue, to my mind, is whether the concept of making the comparisons described in the claims was obvious in the light of the activities and concepts explained in the prior art – and obvious without the benefit of hindsight. Obvious, in this context, means obvious to the unimaginative person skilled in the art. I consider that the concept was obvious. Given that conclusion, the inclusion in the claim of a reference set of 1,000 individuals is not inventive for the reasons which I have already given. In that regard, I agree with what the HO said in [30] of the Decision.
Conclusion on inventive step
- The claimed invention is obvious over the prior art and CGK.
Excluded matter
- The HO considered patentability in [32] to [43] of the Decision. She held that the contribution of the invention lay solely in excluded fields and that there was no technical contribution. Mr Mitcheson seeks to uphold her decision and reasoning. Dr Broughton says her approach was legally flawed and her answer incorrect.
- Given my conclusion in relation to obviousness, it is not necessary, in order to dispose of this appeal, to decide whether the HO was also right in her conclusions in relation to excluded matter. However, since her decision would have been enough to dispose of the appeal to her even if she had reached a different conclusion on inventive step, I consider that I ought to say something about this aspect of the case. I would comment, however, that if she had reached a different conclusion on inventive step, she might well have reached a different conclusion on excluded matter. This is because some of the factors which go to obviousness (and in particular the identification of the inventive concept) would also go to excluded matter (and in particular the identification of the actual contribution in accordance with the test which I describe below).
- Given my conclusions on obviousness, it follows, in my judgment, for the reasons which I give below, that the claimed invention of Claim 1 falls within excluded matter. I do not, however, propose to address, even for completeness, what the position might be in relation to excluded matter if my conclusion on obviousness is wrong. I cannot sensibly do so without knowing, and understanding, why my conclusion is wrong. Accordingly, what I do have to say is of little consequence since if I am right on obviousness that concludes the appeal but if I am wrong, the issue of excluded matter remains undecided by me.
The Law
- Section 1(2) of the 1977 Patents Act provides relevantly as follows:
"It is hereby declared that the following (amongst other things) are not inventions for the purposes of this Act, that is to say, anything which consists of
:…
(c) a scheme, rule or method for performing a mental act, playing a game or doing business, or a program for a computer;
…..
but the foregoing provision shall prevent anything from being treated as an invention for the purpose of this Act only to the extent that a patent or application for a patent relates to that thing as such."
- These provisions, and the case law of the UK courts and the EPO, were comprehensively reviewed by the Court of Appeal in Aerotel Ltd v. Telco Holdings; Macrossan's Application [2007] RPC 7 ("Aerotel"). Both the computer program and business method exclusions were relied upon. The HO correctly identified the four-part test endorsed by the Court of Appeal:
i) properly construe the claim;
ii) identify the actual contribution (what the inventor has really added to human knowledge, as a matter of substance not form);
iii) ask whether the contribution falls solely within the excluded subject matter;
iv) check whether the contribution is actually technical in nature.
- Dr Broughton relies on Halliburton's Applications [2011] EWHC 2508 ("Halliburton") (a decision of HH Judge Birss QC sitting in this Division) in relation to the mental act exclusion, submitting that the HO has incorrectly applied that decision. That decision shows that a narrow scope is to be afforded to the mental act exclusion. Thus, at [63] the judge identified the purpose of the exclusion as being to make sure that patent claims cannot be performed by purely mental means and that is all. The exclusion will not apply if there are appropriate non-mental limitations in the claim. Accordingly, a method which is capable of being performed mentally but which is only claimed by, for instance, the operation of a computer will not fall within the mental act exclusion but it may then fall foul of the exclusion of a computer program as such.
The Decision
- I find it helpful to summarise the HO's reasoning. As to construction, I consider that the HO was correct in her construction of Claim 1.
- As to step 2, she addressed the question (formulated by Jacob LJ in Aerotel) "What has the inventor really added to human knowledge?" She concluded that what had been added to the stock of human knowledge was
"the comparison step involving the use of the reference set comprising data from a normal population of at least 1000 individuals without a particular phenotype to eliminate normal polymorphic CNVs from a set of CNVs identified in an individual with a phenotype, such that a CNV associated with the particular phenotype can be identified."
- Coming to steps 3 and 4 (which she unsurprisingly took together), she accepted that the scope of the mental act exclusion was narrow, following Halliburton. She referred to Claim 2 (a dependent claim under which step b) of the method is computer implemented) which suggested to her mind that the actual contribution which she had identified could be performed mentally.
- In relation to the computer program exclusion, she considered [00201] to [00207] of the specification together with Figure 4 being referred to, observing that there was no doubt that the KMTs used to carry out the comparison step involved the use of a computer program to analyse the data. Whilst accepting that the invention was technical in the broadest sense that it involved a computer, the enquiry "is whether the contribution relates solely to excluded matter and whether it is technical or not". She formed the clear view that the contribution made by the invention did relate to excluded matter as such and did not have a relevant technical effect.
- Her conclusion overall was that the contribution lay wholly in excluded fields and that there was no technical contribution. The inventions claimed in each application were excluded from patentability as a method for performing a mental act and as a computer program as such.
The rival contentions
- Dr Broughton disagrees with the HO's assessment of the contribution. The HO recorded his identification of the contribution as being:
"the provision of a reference set that is adequate to eliminate polymorphic CNVs from a set of CNVs identified in an individual in a high resolution screen such that the "needle in the haystack" single CNV of interest can be isolated from the large number of irrelevant polymorphs, enabling the use of such high resolution techniques for that endeavour (to identify CNVs associated with the phenotype). In the appellant's opinion, the actual contribution was technical."
- According to Dr Broughton, the HO has ignored what the method of the invention actually provides to users. She has identified a contribution that was tantamount to a restatement of what was perceived by her as the "new" part of the claim.
- Thus, prior to the invention, there was no method available for identifying CNVs in an individual at high resolution, and subsequently filtering those CNVs so as to exclude the large majority of polymorphic CNVs, thereby finding the "needle in the haystack" CNV of interest. The invention provides such a method, and does so in an inventive way. It is the ability to take advantage of emerging high resolution screening, by providing a method for the elimination of an overwhelming majority of polymorphic CNVs which are identified in such a screen, which the inventor has really added to human knowledge. The HO was wrong to conclude that, since high resolution screening had already been proposed in the prior art, PDI's formulation of the actual contribution was wrong. The applicant does not contend that the theoretical process of screening at high resolution is the contribution. The contribution is a way in which such high resolution screening can be made practicable in view of the presence of polymorphic CNVs.
- Dr Broughton criticises the HO for having (so it seems) concentrated solely on step b) of the claim, in order to determine whether the contribution falls solely within the excluded matter. Although she makes reference to eliminating normal polymorphic CNVs, and to identifying a CNV associated with the particular phenotype, the problem of conducting that elimination to achieve that result is not put into the context of a high resolution screen (where polymorphic CNVs overwhelmingly predominate). By losing that context she has failed fully to understand that the overall purpose of the claim, and not just the new way in which the claim operates, forms the actual contribution to be judged in step 3 of the Aerotel test.
- And so Dr Broughton says that, when assessed correctly, the actual contribution does not lie solely in the excluded subject matter. It is not solely a method of performing a mental act to be able to find, in an individual, a CNV associated with a phenotype, from amongst an overwhelming number of polymorphic CNVs of no interest. Whilst the method may employ a computer program, and may employ, at points, the requirement to conduct mental acts, the question which must be asked is not whether the invention employs computer programs or mental acts, but whether the "actual contribution" falls solely within these (or other) excluded matters. His answer to that question is that it does not.
- The fourth step of the Aerotel test requires a check on whether the actual or alleged contribution is actually technical in nature. Dr Broughton says that it is certainly technical to provide useful information about the genetic cause of a phenotype in an individual by screening out an overwhelming number of non-causative genetic markers.
- Dr Broughton concludes that the claims cannot be performed by merely mental means and the non-mental parts of the claims are not themselves excluded subject matter. As in Halliburton, the claimed method cannot be performed by purely mental means, and that is the end of the matter.
- Mr Mitcheson submits that the HO was right for the reasons which she gave. Thus, the contribution is the idea of using a comparison with 1,000 normal subjects in order to eliminate normal polymorphic CNVs, as the HO stated in [38] of the Decision. Even if, following Halliburton, the entire claim could not be said to amount to a mental act (which he accepts is what the HO appears to have to concluded), the idea of using 1,000 subjects is a mental act, and the remaining features, namely the actual comparison which follows, are carried out using a computer. He relies, as did the HO, on [00201] to [00207] of the application, and Figure 4. Therefore, applying stage 3 of the Aerotel test, he submits that the contribution falls solely within excluded subject matter.
- In addition, Mr Mitcheson submits that the contribution is not technical in nature. However, he adds that this is so not least because it is arbitrary and lacks invention. In relation to Dr Broughton's submission that it is technical to provide useful information about the genetic cause of a phenotype in an individual by screening out an overwhelming number of non-causative genetic markers, he submits that the mere provision of information does not provide a technical step, even if PDI is right on the issue of construction.
Discussion
- In Aerotel, at [26] it is explained that the question to be asked under the "technical effect approach" (which is the foundation of the 4-stage test later set out) is whether the claim makes a technical contribution to the known art; the rider adds that "novel or inventive purely excluded matter" does not count as a technical contribution. There is no reason to think that "novel" and "inventive" are to be read as meaning anything different from the ordinary concepts of novelty and inventiveness or that the "known art" is anything different from the prior art together with the CGK. At [27] of his judgment in Halliburton, Judge Birss QC reiterated that at the heart of the law is the consistent principle that an inventor must make a contribution to the art (that is to say the invention must be new and not obvious) and that contribution must be technical in nature (susceptible of industrial application and not within an excluded area).
- Although it is only at stage 4 of the Aerotel test that mention is made of technical contribution, it is nonetheless the case that the check must be made. Further, the reference to the technical nature of the contribution informs what it is that is being sought at stage 2, that is to say identification of the actual contribution. The wording of the rider to the "technical effect approach" indicates that that approach is concerned with identifying a contribution which is either novel or non-obvious. In the case of a novel or non-obvious contribution within excluded matter, the rider expressly states that it does not count as a technical contribution (and it cannot, of course, possibly be contended that an obvious contribution in an excluded field is a technical contribution). This flows through to the Aerotel test either at stage 3 or at stage 4.
- In the case of a contribution in a non-excluded field, the position is different. A novel or non-obvious contribution in a non-excluded field is a technical contribution so that the invention will not fall within excluded matter under the technical effect approach. Again, that flows through to the Aerotel test at stage 3 with a check at stage 4. But what is the position where the actual contribution has two elements, one which falls within excluded matter and one which, although not falling within excluded matter, is obvious? Is the obvious contribution which is not within excluded matter a technical contribution for the purposes of the "technical effect approach" and, applying the Aerotel test, does the obvious contribution take the case out of section 1(2)? In my judgment, the obvious contribution is not a technical contribution: a contribution which does not fall within an excluded field will only be a "technical contribution" if it is either novel or non-obvious. Further, the obvious contribution does not take the case out of section 1(2).
- The claimed invention of Claim 1 at least includes a mental act and may also include a computer program as such. The claimed invention is thus potentially within the scope of section 1(2) since there is a contribution within those excluded matters. I have, however, already held that the claimed invention is obvious. It follows that the contributions to the claimed invention outside the excluded matters are obvious and are not, therefore, technical contributions. The contributions to excluded matters (albeit also obvious contributions) bring the case within section 1(2) and there is no technical contribution in a non–excluded field to take it out again.
- In my judgment, therefore, the claimed inventions are excluded from patentability, falling within the exclusion of mental acts and the exclusion of the computer programs as such.
- I would like, notwithstanding that conclusion and the reasons for it, to say something about Dr Broughton's arguments. His argument starts with his construction of Claim 1 and what it is that the inventor has added to the sum of human knowledge. I have rejected his construction from which it follows that I take a different view of what the inventor has added to the sum of human knowledge. Had I accepted his construction, I might have taken a different view in relation to obviousness and held Claim 1 to be inventive. In those circumstances, his argument that the contribution was not solely within excluded matter would have had considerable force. But starting where I do, with a different construction and a finding of obviousness, Dr Broughton's arguments do not succeed.
- I would also make one point in relation to Halliburton. I do not think that Judge Birss was saying that, in any case where an invention involves an activity outside excluded matter, the invention is outside section 1(2). What he said in [63] was to the effect that the mental act exclusion was to be narrowly construed and that its purpose is to make sure that the patent claims cannot be performed by purely mental acts. In the present case, it may be – and Mr Mitcheson appears to accept that this is so – that the entire claim does not amount to a mental act because implementation of the invention involves a computer. In Halliburton, Judge Birss considered whether on the facts the computer program exclusion applied, concluding that it clearly did not. In the present case, it is perhaps not clear whether the computer program exclusion would apply to the computer-aided comparison used to implement step c) of Claim 1 although I incline to the view that it would. Further, I am also attracted by Mr Mitcheson's submission that the mere provision of information is not, as a matter of principle, a technical step. I do not need to decide either of those points since, having held that the claimed invention of Claim 1 is obvious, the contributions afforded by the computer program and by the provision of information cannot, in any event, be technical contributions.
Other matters
- It is, strictly, unnecessary to consider the other objections made by Mr Mitcheson to the applications. For completeness, however, I address them briefly.
Lack of support
- If I am wrong in my conclusion on obviousness, that can, I think, only be because the concept of the relevant comparison described in the Claims either cannot be found in the prior art or is not obvious over the prior art. In that case, I cannot reach the conclusion, on this appeal, that there is a lack of support in relation to the size of the reference set of 1,000 individuals. The HO did not address the point since it was not raised before her. Accordingly, I have no basis on which to say that the skilled person would regard a control set of 1,000 individuals in the same way as he or she would regard a control set of 1 individual, that is to say as entirely useless. I do not consider that I can properly conclude that the claim lacks support where the claim does comprise a real inventive step (as it does under the hypothesis now under consideration). If I am right in my conclusion on obviousness, the issue of support does not arise.
Added matter
- I do not propose to say anything about this not entirely straightforward issue. Even if I were to find in favour of Mr Mitcheson's argument, it would not form an alternative basis for rejecting the applications but would result only in the matter being remitted to the IPO.
Disposition
- PDI's appeal is dismissed.
