Idenix v Gilead - Patent profoundly invalid

If you heard the sound of groaning on Monday morning, it was probably this Kat's desk struggling to support the immensity of the latest offering of Mr Justice Arnold, adjudicating a leviathan struggle between Idenix and Gilead.  The judgment is available here on BAILII, and the case is properly called Idenix Pharmaceutical, Inc v Gilead Sciences, Inc & Ors [2014] EWHC 3916 (Pat) (01 December 2014).

The technical field is treatments for Hepatitis C; the specific area is nucleoside analogues having alleged activity against Hepatitis C Virus and other viruses of the Flaviviridae family.  Infringement by sofosbuvir (which Gilead markets under the trade mark Sovaldi) is swiftly established in 10 paragraphs at almost the end of a 621-paragraph whopper - Gilead is held to have infringed.  The real battle ground was validity.  Within this were a number of interesting and unusual aspects, all contributing to an overwhelming finding of invalidity of Idenix's patent - EP 1523489.

This moggy would like to begin with a number of personal observations.  He has noticed in life that first the trainee patent attorneys look young, then the qualified ones, then the policeman, and now the judges.  But for the first time he has noticed in a patent judgment expert witnesses - expert ones! - who graduated substantially after he did. In this case we have Dr Andrea Brancale receiving his PhD in 2001, and Dr Alistair Stewart receiving his DPhil in 2003.  That is in this millennium!  This Kat was happy to learn that Dr Stewart's DPhil supervisor was his very own internal examiner, Professor George Fleet.  (In the interests of full disclosure, this Kat also notes that Gilead has been a client of his at his former firm).

Novelty
The patent as granted was held to lack novelty over Gilead's own International Patent Application PCT/US2004/012472 ("the Pharmasset PCT" - Pharmasset having been acquired by Gilead in 2012).  What was interesting here was that this was a novelty only citation, and then only if it was entitled to its claimed priority - its priority date was earlier than the Idenix filing date (Idenix's own priority claim was not asserted) but its filing date later, and it was published after the Idenix filing date.

The US priority application was filed in the name of Mr Clark, who was employed by Pharmasset Georgia (and another inventor who was later removed); the PCT application was filed in the name of Pharmasset Barbados. Gilead claimed that Pharmasset Barbados was the successor in title to Mr Clark the inventor, and thus able validly to claim priority under the Paris Convention, by virtue of the employment contract of Mr Clark, and an agreement between the two Pharmasset companies.  This involved a complicated factual and legal inquiry, under US Federal and Georgia State law, into the effect of these agreements.

Gilead prevailed, the priority right was upheld, and under these circumstances lack of novelty of the claims as granted was conceded.  What troubles this Kat is that Gilead were only able to prove this because it was their prior application - what would have happened if this had belonged to a completely independent third party?  How would the validity of the priority right then have been established?

Inventive step
The claims as granted were held to lack inventive step as not contributing to the state of the art, because it was not plausible that substantially all the claimed compounds would have the necessary antiviral activity.  In fact, Idenix's expert Dr Brancale had 'expressed the opinion in paragraphs 201-202 of his first report that compounds of Formula (IX) would not have been thought likely to have antiviral activity where R1 and R2 were either "straight chained, branched or cyclic alkyl" or "benzyl, wherein the phenyl group is optionally substituted with one or more substituents".'  The judge was even more critical of the application as filed, calling it "stupendously broad" and agreeing with its characterisation by Gilead as a "land grab".

Even excluding the compounds for which lack of activity had been admitted, the judge considered that it was not plausible that substantially all the compounds covered by claim 1 as proposed to be amended would be effective against Flaviviridae.  A key point in this reasoning is that the patent itself did not state that the compounds disclosed and claimed therein had activity, only that 'the nucleosides of the invention "may inhibit Flaviviridae polymerase activity" and "can be screened" for such activity using known assays'.

Therefore the patent, as granted or proposed to be amended, was held to lack inventive step.

Sufficiency
For analogous reasons to the lack of inventive step, the patent was held to be invalid for insufficiency by excessive claim breadth.  This was pretty much to be expected, as the two grounds are closely connected.  More strikingly and unusually, the judge held that the specification did not enable the invention to be performed at all without undue burden, because it did not disclose how to make the compounds.  In the claimed structure there was an element that is rather difficult to make (Merpel explains for the chemists among our dear readers: quaternary substitution at the ribose 2' position, where the claimed structure required a fluoro and a methyl substituent) and which, as the evidence showed, Idenix's own scientists had struggled to make.  Experiments were adduced to show that a relatively accessible route to the compounds did achieve the required structure, albeit in low yield, but (as Merpel recalls seems always to be the way) they failed to achieve what they aimed to show.

As if the patent were not by now already dead enough, the judge turned to:

Added matter
The patent as granted was not held to involve added matter with respect to claim 1 - although the amendment as compared with the application as filed involved the selection of one subclass from an application of "stupendous breadth", it was  a class that was specified from the outset as being of particular interest.  Claim 4 as granted was however held to involve added matter, as it selected eight bases that were not disclosed as subclasses within the definition of purine or pyrimidine base in the Application.

More significantly, the proposed amendment to exclude the classes of compounds that Idenix's evidence showed would not plausibly have antiviral activity, while retaining those that their evidence showed would have such activity, was also held to involve added matter (notwithstanding that the judge had rejected plausibility for those classes as well.)  A bold suggestion by Idenix that it was not necessary for them amend the granted claims if they were partially invalid was also rejected.

So, the patent was ordered to be revoked.

Construction 

sofosbuvir

An interesting point that did not however make much difference in the end was the construction of the term "phosphate" in the claims.  The judge construed this to include masked phosphates, such as were to be found in the Gilead drug candidate sofosbuvir (masked by a phenol group and a phosphoramidate group, Merpel informs chemists).  A superficially attractive argument by Gilead that the situation was all square with Actavis v Lilly - that the claims had been amended during prosecution and therefore the terms should be construed narrowly - was rejected.  The judge recorded:

counsel pointed out that I had held in Actavis v Lilly at [112] that, in principle, a limitation made to a claim to avoid an objection of lack of clarity could be relied on as aid to construction. I adhere to that view, but with the benefit of the arguments of the present case I would add that such an amendment is less likely to be a useful aid to construction than a limitation to avoid an objection of lack of support, which was the primary focus of my discussion in Actavis v Lilly.

The IPKat is sure that many practitioners will welcome this clarification.

The IPKat is equally sure that more will be found to be said about this Arnoldian judgment.  It is a fun read, and nicely indexed so that the reader can easily find their way back the bit that they are trying to find.