Neurim Pharmaceuticals (1991) Limited & Anor. v Generics (UK) Limited & Anor.

In these proceedings the Claimants allege infringement by the Defendants (“Mylan”) of European Patent (UK) No. 3 103 443 (“EP443”). EP443 is a second medical use patent which (as unconditionally proposed to be amended) claims the use of a prolonged release formulation of melatonin in a 2 mg dose for improving the restorative quality of sleep in a patient aged 55 years or older suffering from primary insomnia characterised by non-restorative sleep (“NRS”). The First Claimant (“Neurim”) is the proprietor of EP443. The Second Claimant markets a product falling within the claims of EP443 under the trade mark Circadin in the United Kingdom pursuant to an exclusive licence granted by Neurim. Mylan market a generic version of Circadin. There is no dispute that Mylan are infringing EP443 if it is valid, but Mylan contend that it is invalid on the ground of insufficiency, or more specifically lack of plausibility.

EP443 is a divisional of European Patent (UK) No. 1 441 702 (“EP702”). Accordingly, both EP443 and EP702 have the same priority date, 14 August 2001, and EP443 is due to expire on the same date as EP702 would have expired had it not been revoked, 12 August 2022. There is no material difference between the claims of EP443 and those of EP702 (again, as proposed to be amended) although the wording differs slightly.

In previous proceedings brought by the Claimants against Mylan for infringement of EP702, Marcus Smith J dismissed a number of challenges to the validity of EP702 advanced by Mylan for the reasons given in his judgment dated 4 December 2020 [2020] EWHC 3270 (Pat) (“the December Judgment”). Shortly afterwards, however, EP702 was revoked as a result of Neurim’s withdrawal of an appeal against an order for revocation by the Opposition Division of the European Patent Office at the conclusion of a hearing before the Board of Appeal.

The procedural background to the present appeal is set out in a previous judgment of mine given on 29 March 2022 [2022] EWCA Civ 370 at [2]-[21]. For present purposes it suffices to say that Marcus Smith J rejected Mylan’s challenge to the validity of EP443 for the reasons given in the December Judgment as supplemented in a judgment dated 8 March 2022 [2022] EWHC 519 (Pat) (“the March Judgment”). Because most of the judge’s reasoning was given in the context of a judgment concerning EP702, but is equally applicable to EP443, I shall use the term “the Patent” to refer to both.

Mylan appeal with permission granted by this Court on 16 March 2022. It is convenient to note before proceeding further that, since then, the Claimants have offered a cross-undertaking in damages in the event that Mylan’s appeal is dismissed, but EP443 is subsequently revoked by the EPO. In the light of that offer, Mylan no longer pursues its application for a stay of the injunction granted by the judge pending the decision of the EPO (in practice, until expiry of EP443) if the appeal is dismissed. It is therefore unnecessary for this Court to consider the issue identified in my judgment on the Claimants’ appeal against two aspects of the judge’s order concerning EP702: [2022] EWCA Civ 359 at [55].

The judge found in the December Judgment that the skilled person to whom the Patent is directed is a sleep medicine clinician with significant expertise in primary insomnia and the guidelines used in the diagnosis of insomnia.

The judge received expert evidence from Professor Thomas Roth (called by the Claimants) and Professor Kevin Morgan (called by Mylan). The judge described Prof Roth in the December Judgment at [13(1)(f)] as “one of the foremost sleep medicine experts in the world” and considered that Prof Roth had done his absolute best to assist to court. The judge explained that Prof Morgan’s expertise was both more general than that of Prof Roth, in that his work covered human ageing generally, and more specific, in that the effect of ageing on sleep was an important aspect of his work. The judge rejected the Claimants’ submission that Prof Morgan was not appropriately qualified. The judge was, however, highly critical of Prof Morgan’s evidence in his first and second reports, describing them at [13(3)(h)] as “in critical respects, disingenuous documents … calculated, not to assist, but to mislead, the court”. Accordingly, the judge was “not confident” that he could “rely on Professor Morgan’s reports, save with a degree of caution and reserve that a judge would not normally attach to the report of an expert”.

The judge made detailed findings as to the common general knowledge of the skilled person in the December Judgment at [33]-[36] and [50]-[67]. The findings which are relevant for the purposes of the appeal may be summarised, using the judge’s headings, as follows.

The term “primary insomnia” derives from the Diagnostic and Statistical Manual of Mental Disorders, the 4th edition of which (“DSM-IV”) distinguishes between “primary sleep disorders” and “secondary sleep disorders”. In broad terms, secondary sleep disorders are linked to another medical or psychological condition, whereas primary sleep disorders are not.

According to DSM-IV, primary insomnia is indicated where the predominant complaint is one or more of the following:

The 10th revision of the International Classification of Mental and Behavioural Diseases (“ICD-10”) describes primary insomnia as “nonorganic insomnia”, but uses similar diagnostic criteria to those in DSM-IV, although ICD-10 uses the term “poor quality of sleep” to denote NRS. I shall concentrate on the DSM-IV terminology, as did the judge.

The three characteristics of primary insomnia referred to above can co-exist or exist separately. According to DSM-IV and ICD-10, the most prevalent complaints are difficulty initiating sleep and/or difficulty maintaining sleep.

As the judge found at [53]:

The skilled person would expect NRS to present either as a solitary indicator of primary insomnia or in conjunction with other diagnostic indicators (i.e. difficulty initiating sleep and maintaining sleep).

Primary insomnia characterised by NRS is a purely subjective phenomenon because only the individual in question can say whether their sleep has been restorative. At [62(2)], the Judge accepted as accurate the following explanation given in a book co-authored by Prof Morgan in 1999:

Objective methods for monitoring sleep such as polysomnography (electrophysiological monitoring of brain waves etc) and actigraphy (monitoring of the patient’s movements) are not used in the diagnosis of NRS. Instead, primary insomnia characterised by NRS is diagnosed by ascertaining how, subjectively, the patient considers their quality of sleep, typically by taking a history.

Questionnaires may be used to ascertain how, subjectively, the patient considers their quality of sleep. One such questionnaire is the Leeds Sleep Evaluation Questionnaire (“LSEQ”) which was published in Parrott and Hindmarch, “Factor analysis of a sleep evaluation questionnaire”, Psychological Medicine, 8, 325-329 (1978) (“Parrott and Hindmarch”).

As Parrott and Hindmarch explains, the LSEQ comprises 10 questions, grouped into four chronological areas: (1) the ease of getting to sleep (GTS): questions 1, 2 and 3; (2) the perceived quality of sleep (QOS): questions 4 and 5; (3) the ease of awakening from sleep (AFS): questions 6 and 7; and (4) the integrity of behaviour following wakefulness (BFW): questions 8, 9 and 10.

The LSEQ is as follows:

The judge made the following findings at [64] concerning the skilled person’s understanding of questions 4 and 5 (emphasis in the original, footnotes omitted):

In footnote 76 to [64(6)(b)] the judge stated:

There was no treatment for NRS in 2001. It was known that melatonin could be useful in the treatment of circadian rhythm disorders, but it was not thought to be useful in the treatment of primary insomnia.

The specification begins at[0001] by stating that the invention “relates to the use of melatonin for the manufacture of a medicament for treating primary insomnia (as defined by DSM-IV) or nonorganic insomnia (as defined by ICD-10) when characterized by non-restorative sleep”. The specification proceeds to outline the nature of primary insomnia at [0002] to [0005].

At [0006] the specification notes that the possibility of administering exogenous melatonin to ameliorate sleep disorders is the subject of many scientific papers, some of which are described briefly in [0007] to [0012], and at [0013] it notes that there appears to be little or no evidence from published articles that the administration of melatonin in the dosage contemplated by the invention would be likely to improve the restorative quality of sleep in subjects affected by primary insomnia characterised by NRS. At [0014] the specification states that the inventors have surprisingly found that melatonin improves the restorative quality of sleep in patients suffering from primary insomnia.

After a summary of the invention at [0015] to [0018], there is a detailed description of the invention at [0019] and following. At [0027] the invention is said to be illustrated by five Examples. Examples 1 and 4 are both “reference” examples, which the skilled person would appreciate do not purport to describe something falling within the claims. Example 5 concerns the preparation of the prolonged release melatonin formulation and is not relevant for present purposes.

Example 2 is described as follows:

The questions set out in [0032] correspond to questions 4 and 9 in the LSEQ. There is no reference in Example 2, or anywhere else in the specification, to any of the other questions in the LSEQ.

Example 3 is described as follows:

By contrast with the results in Example 2, there is no claim to statistical significance for the results in Example 3.

It is sufficient for the purposes of the appeal to refer to claim 1 of EP443 as proposed to be amended. This claim is in “Swiss” form, but there is a parallel claim in EPC 2000 form:

There is no dispute as to the applicable legal principles. Since the Patent is a patent for a second medical use of a known medicinal compound, the Patent must plausibly disclose the effect that it claims. The criterion for plausibility was stated by Lord Sumption in Warner-Lambert Co LLC v Generics (UK) Ltd [2018] UKSC 56, [2019] Bus LR 360 at [36]-[37] as follows:

Mylan contend that the Patent does not plausibly disclose the effect that it claims because of what has come to be called “the lay-patient argument”. As is common ground, the plausibility of the claimed effect depends on what is disclosed by Examples 2 and 3 read in the context of the specification as a whole. As is also common ground, the key data in both Examples are those generated by the responses to the question concerning “quality of sleep”.

For convenience, I shall set out the question and the criteria for answering it again:

Mylan contend that there is nothing in this question or the responsive criteria which bears any relation to NRS. Mylan say that this is hardly surprising given that the LSEQ, from which the question is taken, was published nearly ten years before NRS was recognised as a diagnostic indicator of primary insomnia in the third edition of the DSM in 1987.

Furthermore, Mylan contend that the responsive criteria are plainly capable of invoking the non-NRS indicators of primary insomnia. For example, a patient who has difficulty maintaining sleep may well describe the quality of their sleep as “more restless than usual”. Likewise, a patient whose sleep initiation is improved might well conclude that the quality of their sleep is “more restful than usual”.

Mylan do not dispute that the skilled person would attach a technical meaning to “quality of sleep” when used in the context of ICD-10, namely as a reference to NRS. Mylan contend, however, that that is irrelevant when it comes to the critical issue in relation to plausibility, namely how the skilled person would understand the data in the Examples. This depends on how the patients would have answered the question, which in turn depends on how the patients would understand the question. As is common ground, the patients in the Examples would have been laypeople, not skilled persons.

Mylan point out that the stated purpose of Examples 2 and 3 is to investigate the effect of melatonin on “subjectively assessed sleep quality”: see [0032] and [0036]. Consistently with this purpose, the specification explains that the data were generated by asking patients to mark the level of their “perceived quality of sleep”. There is no suggestion in [0032] or [0036] that the patients were instructed to answer the “quality of sleep” question in any particular way or with any special meaning in mind. Mylan argue that, in order to single out NRS, such instructions would have needed to be highly exacting. In particular, when assessing whether the quality of their sleep had been “more restless” or “more restful” than usual, the patients would need to be instructed to exclude from their consideration (i) how long it took to get to sleep and (ii) any periods of waking in the night and (iii) any other aspect of their sleep except its restorative quality.

Accordingly, Mylan argue, the data in the Patent could only render the claimed effect plausible if it was common general knowledge that patients were so instructed in studies of the kind described in Example 2 and Example 3. Mylan point out that the judge made no finding that this was common general knowledge, nor was there any evidence before him to that effect.

Mylan submit that it follows that the responses to the “quality of sleep” question reported in Examples 2 and 3 cannot plausibly demonstrate that melatonin has an effect on NRS. Since the data themselves do not distinguish between NRS and other forms of insomnia, it is immaterial that the reader of the Patent is a skilled person who has a technical understanding of “quality of sleep” in the context of ICD-10.

Furthermore, Mylan argue that, even if the skilled person might think that the question was intended to be specific to NRS, the specification itself contradicts such a reading when it states at [0039]:

It is important to note two points before considering the judge’s reasoning. First, as Mylan accept, the lay-patient argument was not articulated very clearly by Mylan at trial. No doubt for that reason, the lay-patient argument was not squarely addressed by the judge in the December Judgment. It was, however, considered by the judge in the March Judgment on the basis of the evidence at trial and his findings in the December Judgment. Secondly, as Mylan also accept, the point about the last sentence of [0039] in the Patent was not advanced at trial. More specifically, it was not a point made by Prof Morgan, nor was it put to Prof Roth.

In the December Judgment the judge held that the Patent did plausibly disclose the effect that it claimed for the following reasons (so far as relevant, footnotes omitted):

In the March Judgment the judge rejected the lay-patient argument for the following reasons (original emphasis, footnotes omitted):

The judge’s decision that the Patent plausibly discloses the effect that it claims is an evaluative assessment. It follows that Mylan must establish that the judge made an error of principle in his assessment. The only error suggested is that the judge was wrong to reject the lay-patient argument.

Mylan make five main criticisms of the judge’s reasoning on this issue. First, Mylan criticise the judge’s reasoning in the December Judgment at [64(3)-(6)] (quoted in paragraph 20 above). Mylan argue that the judge was correct to distinguish between the way in which a skilled person would understand “quality of sleep” in the context of ICD-10 and the way in which a layperson would understand that expression, but failed to apply that distinction when considering how questions 4 and 5 of the LSEQ would be understood. The judge concentrated on how “the Skilled Person using the questionnaire” in 2001 would understand the questions, not on how the layperson answering the questions would understand them. Mylan go on to argue that the judge was therefore wrong to reject Prof Morgan’s evidence on this point in footnote 76 (quoted in paragraph 21 above).

Secondly, Mylan contend that, even if the judge was correct as to how questions 4 and 5 of the LSEQ would be understood in that context, he was wrong to treat that finding as applicable to the “quality of sleep” question in the Patent in the December Judgment at [112(3)]. Mylan point out that in [64(6)(c)] the judge relied upon the chronological sequence of the questions in the LSEQ, but as noted above there is no indication in the specification that any questions other than questions 4 and 9 were asked. Again, therefore, Mylan argue that the judge was wrong to reject Prof Morgan’s evidence on this point.

Thirdly, Mylan contend that the judge made the same errors in the March Judgment. Again, he focused on “the Skilled Person’s understanding of the terms of the questionnaire”, rather than upon how the layperson would understand the single “quality of sleep” question.

Fourthly, Mylan contend that the judge’s reasoning at [10] in the March Judgment (quoted in paragraph 43 above) involves assuming that the results support the conclusions drawn from them and therefore assuming that either the entire LSEQ was administered or the patients were instructed as to the meaning of “quality of sleep” when neither is disclosed. Mylan argue that this is to reduce plausibility to a mere test of good faith.

Fifthly, Mylan point out that, even in the March Judgment, the judge does not take into account the statement made in the last sentence of [0039].

I do not accept these arguments for the following reasons. First, while it is common ground, as noted above, that the patients in Examples 2 and 3 were laypeople and not skilled persons, I consider that the judge was right to focus on the purpose for which the “quality of sleep” question was being asked. Although Mylan are correct that [0032] and [0036] simply refer to “subjectively assessed sleep quality”, the skilled reader of the Patent reads those statements in their context in the specification as whole. Read in context, it would be clear to the skilled person that the purpose of Examples 2 and 3 is to demonstrate that melatonin is effective in enhancing the restorative value of sleep in primary insomnia patients i.e. treating NRS.

Secondly, I consider that the judge was also right to say that the skilled person would assume that the trials reported in Examples 2 and 3 were competently conducted unless there was some reason to think otherwise. Accordingly, the skilled person would presume, unless there was some reason to think otherwise, that the “quality of sleep” question had been administered in a way which was capable of producing meaningful data. This is not to reduce plausibility to a mere test of good faith, because plausibility depends on the reported results of the trials. Nor is it to assume that the results do support the conclusions drawn in the Patent, which is a separate question. Even if it is correct that the “quality of sleep” question is not without more specific to NRS, there are two ways in which it could have been administered so as to produce meaningful data.

The first is that the entirety of the LSEQ was administered, thereby ensuring that patients answered the questions in chronological sequence. This is inherently probable given that the use of the LSEQ was well established in the field and that questions 4 and 9 were asked. Moreover, Prof Morgan’s evidence was that, given the expense of conducting a trial of this size, he “would expect further data to have been collected from the 301 patients”. There would be nothing unusual in what is evidently a compressed report of the trials omitting some of the experimental details and results. The judge found, and was entitled to find, that, if all the questions in the LSEQ were asked in chronological sequence, the answers to questions 4 and 5 would be specific to NRS.

The second way in which this could have been done even if questions 4 and 9 were asked in isolation was to give the patients appropriate instructions, as Mylan accept could be done. Again, it would not be surprising if this was omitted from the report. Contrary to Mylan’s argument, this does not require a finding that it was common practice to educate patients in this way. It is sufficient that it is a possible way in which the Examples could have been competently performed in the absence of anything to indicate that they were not competently performed.

Thirdly, it is not necessarily correct that, even if administered in isolation and without further instructions, the “quality of sleep” question would not be specific to NRS. Prof Roth gave evidence in paragraph 7.20 of his third report that:

Fourthly, the judge was entitled to reject Prof Morgan’s evidence for the reasons he gave. In any event, the judge was entitled to prefer the evidence of Prof Roth, including the evidence I have quoted in the preceding paragraph.

Fifthly, Mylan’s best point is the point concerning the last sentence of [0039]. I agree with Mylan that the natural interpretation of this sentence is that set out in paragraph 40 above. But, read in that way, the skilled person would appreciate that there was an apparent contradiction between that statement and the remainder of the description of Examples 2 and 3. The question then is how the skilled reader would resolve that apparent contradiction. Given that it is a question of the skilled reader’s technical understanding, expert evidence is required before the court can draw a conclusion. For example, one possibility is that the skilled reader would consider that, notwithstanding the wording of the sentence in question, the conclusion was likely to have been informed by answers to questions 1-3 in the LSEQ. As noted above, Mylan failed either to adduce evidence from Prof Morgan about this or to put the point to Prof Roth. In those circumstances it is not open to Mylan to rely upon this point.

For the reasons given above I would dismiss the appeal.

I agree.

I also agree.