F Hoffman-La Roche Ltd v Cipla Ltd Expert Analysis


The long awaited Hoffman-La Roche v. Cipla judgement was finally passed last week marking the end of the first phase of a key battle between Big Pharma and the Indian generic industry. Although there are pending suits before the Delhi High Court which have been filed by Roche against several other generics, most prominently NATCO, Glenmark and Dr. Reddy’s, it is expected that the finding of non-infringement against Cipla will pre-determine the fate of those lawsuits as well. Shwetasree Majumder dissects the 275 page judgement and analyses the various issues argued and the court’s ruling.

By Shwetasree Majumder


Opening Remarks


The Hoffman La Roche/Cipla case was played out in far greater detail in the public domain than in the courts, right from the start. With all the attention it garnered (mostly unsavoury), particularly on the alleged Indian judicial climate turning anti-innovator, it won’t be an exaggeration if I say that thousands across the globe were waiting to see which way the wind blows with this crucial decision. And that is probably where this decision scores the highest. It panders to none, sticks to the facts and the applicable law and comes as a breath of fresh air in steering clear of access-pricing-patent linkages which have been known to dangerously dilute core patent law principles in the past.


In fact Justice Manmohan Singh’s judgement of 07.09.2012 exonerating Cipla’s generic version (Erlocip) of Roche’s cancer drug Tarceva from patent infringement while upholding the validity of Roche’s cancer drug patent (IN ‘774) can claim several firsts. Perhaps the most prominent of these, is that it can legitimately claim to be India’s first post-trial judicial application of mind in this specific field to a case of such magnitude. The speedy conclusion of the trial in this case is also worthy of mention – four years since the day it was filed (including the time taken for the hearing and decision on an intervening appeal from the interim order), several volumes of evidence and dozens of hearings later, we have this 275 page judgement – no mean feat!


The plot and premise of the dispute is straightforward – Roche was granted Indian Patent No. IN ‘774 in February 2007, under which per Claim 1, they had patent rights over the Erlotinib Hydrochloride (EH) molecule which has demonstrated breakthrough capabilities as an Epidermal Growth Factor Receptor (EGFR) inhibitor which spiked survival benefit in non-small cell lung cancer (NSLC) patients. Based on media intelligence declaring Cipla’s intention to launch a generic version of Roche’s drug in January 2008, Roche moved the Delhi High Court seeking to injunct Cipla from marketing Erlocip. Cipla counterclaimed, arguing that Roche’s patent was invalid.


The judgement (which marks the conclusion of the battle at only the first instance) deals with two key issues: first, whether the manufacture of Erlocip infringes Roche’s IN ‘774 patent and second, whether Roche’s IN ‘774 patent ought to be revoked as being invalid.


1.       The Validity Question


Taking the second issue first, Cipla’s counterclaim contended that IN ‘774 contains no inventive step and is not novel for two main reasons:


1. That the closest prior art to IN ‘774 is Example 51 of EP ‘226 (which was admitted as prior art in the IN ‘774 specification) which was based on extremely similar structures. They pointed to the fact that the IN ‘774 specification is silent on how it is an advancement on Example 51 and that more than one example in IN ‘774 is specifically disclosed in EP ‘226, which supports the conclusion that IN ‘774 was a mere combination and did not involve an inventive step.


2. That the only material distinction between IN ‘774 and Example 51 of EP ‘226 is the substitution of a methyl group with ethynyl at the third meta (3’) position which is obvious to a person skilled in the art on account of at least five prior art teachings which suggest that methyl and ethynyl can be used interchangeably with no categorical loss of efficacy.


In response, Roche argued that comparative efficacy in this case had been mapped according to inhibitory concentration (which term can be loosely understood as the concentration of a substance that causes a defined inhibition of a system), in which respect IN ‘774 is closest to a prior art reference EP ‘851 and is also inventive over it. Consequently, they saw no reason why EP ‘226 (and Example 51 specifically, which does not display an inhibitory concentration value within even the top five such values of compounds disclosed by EP ‘226) should be treated as the point of departure in the inventive step analysis. Indeed, EP ‘226 teaches a methyl/ethynyl substitution at the 6, 7- position and not at the 3’ position and there are several demonstrable differences in bond angle, bond length and bond strength and the type of reaction with EGFR kinase between methyl and ethynyl group that make the substitution ostensibly inventive.


Justice Singh relies on Roughton et al’s The Modern Law of Patents to create a matrix of material facts required to be shown by a defendant to prove obviousness in a revocation inquiry:


a. The selection of the impugned invention is taken from the examples of the known prior art.

b. That the selected invention is not far removed from the known range illustrated in the example. Rather, the same is closer to the known range.

c. That the selection area is not on the basis of any purpose of the inventor and is merely an arbitrary picking up the compound (para 66).


The Learned Judge holds that the first condition is fulfilled by Cipla but that the other two are not demonstrated by positive evidence. He holds that the deviation could have been shown by ‘clinical’ demonstration (even via deposition) and that there was no deposition on Cipla’s behalf pointed at establishing the arbitrariness of selection (The deposition on Cipla’s behalf stated that there was a possibility of the desired result by a methyl/ethynyl substitution but no guarantee). On the arbitrariness of selection, he holds that the non-arbitrary/purposeful character of the selection reflected in the invention is the dominant criterion and overrides any ‘inspiration’ taken from EP ‘226.


Apropos of the five patent specifications cited by Cipla to show that methyl/ethynyl substitution was well known in prior art, the Judge disqualifies them procedurally because they were filed after framing of issues and did not meet the requirement that they could not be filed on account of being outside Cipla’s knowledge or resources. Substantively, he states that four of the five patents pertained to non-quinazoline derivatives and the one that did – US ‘534 – quotes a 6, 7 methyl/ethynyl substitution which is inconsequential per se. Interestingly, it is in the context of this additional evidence which the court first discards on grounds of admissibility and then proceeds to cursorily refer to in order to satisfy its conscience, that we find the most succinct summing up of the Learned Judge’s view on novelty and Example 51. He observes at Para 104: “…the question is that why there would be an arbitrary adoption of example 51 and why the said plaintiff would apply and react the ethynyl only by replacing the methyl at the third position, when the as per the plaintiffs version, which is not disputed by the defendant, EP‘226 teaches to keep the methyl component stable and not variable.”


Apart from this, the judgement also interestingly invokes commercial success of the drug as an attending circumstance in establishing the purposefulness of the selection and defends this positive evidence standard as being necessary to avoid a slippery slope of judging competing products on the basis of superficial structural similarity. On an overall consideration of factors, Justice Singh holds that Cipla are unable to meet even a balance of probabilities to establish revocation.


2.       The Section 3(d) Challenge


The second limb of Cipla’s counterclaim – a Section 3(d) challenge to the validity of IN ‘774 – was born out of the fact that Roche had unsuccessfully applied for a ‘Polymorph B’ form of EH (IN ‘507), a claim which was rejected by the Controller of Patents in December 2008 with findings on evergreening, structural similarities between IN ‘774 and EP ‘226 and a lack of conclusive comparative clinical data to prove efficacy.


Nevertheless, Justice Singh rules that Cipla fail to meet the positive evidence onus to sustain a Section 3(d) challenge, since it required them to prove IN ‘774 is the ‘new form of an old substance’ (the ‘old substance’ being EP ‘226). He rules against Cipla on the grounds that none of their witnesses deposed to establish that EP ‘226 and IN ‘774 are the same substance and the fact that Example 51 of EP ‘226, through further reaction, can result in IN ‘774 is insufficient to establish ‘new form of an old substance’ unless proven to be contrary. Thus, seemingly against his initial desire to avoid importing presumptions of validity, Justice Singh holds that the reaction with a new reactant could create a new form of an old substance in this case but that it had not been proved by Cipla. Roche on the other hand had established sufficient evidence of difference in efficacy, sufficient certainly for it to be inferred that IN ‘774 was not hit by Section 3(d), though this was not done through deposition.


3.       The concealment/ non-disclosure challenge – observations on Section 8


Several subsidiary challenges to IN ‘774 were also raised by Cipla, which can all be classified as allegations of suppression and concealments by Roche. All of them barring one under Section 8 were rejected by the Court. Patent lawyer readers will agree with me that Section 8 which lays down disclosure requirements as to all corresponding foreign patent applications has been the single most problematic area of patent practice in India in recent times pursuant to a decision of the Delhi High Court in Chemtura Corporation v Union of India [(2009) 41 PTC 260] which laid down a strict threshold of disclosure and also warned that inadequacy of disclosure would be a ground to seek revocation of a patent.


The Roche v Cipla judgement does not refer to Chemtura but categorically does not revoke IN ‘774  despite holding on facts that “…the plaintiffs as patentee has not disclosed the information as required by the controller as per Section 8 of the Act which is evident upon from the examination report dated 22.8.2006 and the responses thereto which do not record the subsequent patent in US‘221 which ought to have been disclosed. Thus the ground of revocation under Section 64 (1) (m) is made out” (para 196). Despite holding that the ground under Section 64 (1) (m) is made out the Court chooses to exercise positive discretion in the plaintiff’s favour to not revoke the patent by inferring the existence of such discretion from the presence of the word “may” in the Section.


This observation will undoubtedly afford a breather to patentees who have been struggling with nightmares of being faced with revocation of substantively valid patents on account of inadequacy of disclosures under Section 8 when no specific prescription as to adequacy exists in the procedure.


4.       The Infringement Question


Once again using the “positive evidence” standard, Justice Singh correctly casts the onus to establish infringement on Roche. The principal confusion with regard to this issue is the interchangeable reference to EH and Tarceva as being covered by IN ‘774. The relief claimed however was for infringement of rights in the drug Tarceva as well as an injunction restraining the manufacturing and marketing Erlocip.  The Judge attributes this to Roche’s understanding at the time of filing of the suit that “the drug Tarceva corresponds with that of the patent and therefore the infringement was sought of the legal rights of the Tarceva (Erlotinib) medicine or drug” (para 209), an understanding he clearly disagrees with, and which ends up forming the substratum of his ruling on the infringement question.


Cipla’s defence to Roche’s infringement claim was based on the claim that IN ‘774 is for an admixture of Polymorphs A and B of EH but Tarceva is the stable form of Polymorph B, which corresponds to US ‘221 (and the rejected IN ‘507) and that since the same was not covered by ’IN 774 there can be no infringement.


The court had a lot to say on the shortcomings of Roche’s evidence on the issue of infringement. More specifically, it noted the absence of clinical tests to compare the constituents of Roche’s and Cipla’s drugs and specifically on whether Roche’s drug corresponded with the Claim 1 of IN ‘774. It noted that Roche merely led evidence (that too only from its own literature on the drug and not a scientific analysis of its constituents) to show that Cipla’s drug was also EH and claimed that manufacture of Polymorph B by Cipla was sufficient to trigger infringement of Claim 1 of IN ‘774. (Clearly, it flows in sequitur that any process involved in making Polymorph B would first involve the preparation of a combination of Polymorphs A and B; even US ‘221 states that EH in Polymorph B form results from re-crystallization of EH using different solvents and temperature conditions).


On this issue, Justice Singh places significant reliance on expert testimony of two kinds as adduced by Cipla – first, on X-ray diffraction tests which establish that Tarceva is Polymorph B alone and second, that the tablet form of EH does not follow directly from the claims in IN ‘774 (on account of the fact that further reactions of the product from IN ‘774 are required to produce Tarceva).


This leads the court to a construction of Claim 1 of IN ‘774 to analyse whether it subsumes Polymorph B of EH, which is admittedly Erlocip. On this issue the learned Judge goes on to adopt the Catnic Components approach to purposive claim construction as an aid which was extended to chemical compounds in Merck, and which advocates giving effect to ‘the real purpose for which the product was invented and explicitly involves a substituted judgement as to the real purpose of the product.


The other major substantive holding by Justice Singh on this issue is the adoption of Merck-style different thresholds for obviousness in revocation and in claim construction, based on the reason that an a fortiori determination of whether a variant can be subsumed within a broad patent claim is bound to be substantively different (and, presumably, lower) when compared with the obviousness standard in assessing the question of further working on an invention that could revoke it in the future.


Ironically, the final nails in Roche’s coffin are hammered in by the US ‘221 specification documents themselves (which clearly attest to Polymorph B as the sole efficacious commercial manifestation of EH) and by the fact that the very factum of Roche’s application for IN ‘507 knifes IN ‘774. On the latter issue, Justice Singh holds in para 260 (emphasis supplied):


In [the] absence of the explanation of the said role either as a major or minor reactants coupled with the fact that both in India as well as in [the] US, the [P]laintiffs have applied for the patent for the said process and product separately than the underlying compound, the purposive construction of the claim and the specification of IN ‘774 clearly indicates that the said [P]laintiffs did not intend to include the Polymorphic version B in the suit patent IN ‘774”.


The emphasised part makes it clear that if Roche had applied for either IN ‘774 or IN ‘507 in isolation or, indeed, IN ‘507 before IN ‘774, the Learned Judge would probably have found that Erlocip infringed Roche’s first patent. Thus even after an opinion that checks in at over 75,000 words, it is telling that the biggest battles in Indian patent law are being fought on wafer-thin margins.



Shwetasree Majumder is the proprietor of IP boutique firm Fidus Law Chambers and a patent litigator.