Tuesday, 6 September 2016

Sofosbuvir revisited - a chemical puzzle

I last wrote about the blockbuster drug sofosbuvir in February 2015, around the time when a few oppositions were being filed against Gilead's European patent for the drug. The oppositions are currently going through the usual procedures at the EPO, and oral proceedings have been scheduled for 4-5 October 2016. From a quick look at the preliminary opinion of the opposition division, it looks like the patent will be upheld. With ten opponents and a huge amount of material on record, however, it still looks like a Herculean task for the patentee.
Chemical structure of sofosbivir (Wikipedia).

In the meantime, however, I noticed a recent article about a claim being made in the US that a patent owned by the University of Minnesota is infringed by the drug. The article does not mention the patent by number, but after doing a bit of simple searching I think it could be US6475985, titled "Nucleosides with antiviral and anticancer activity", which was granted to the University in 2002 (and was, incidentally, not cited as prior art in the EP oppositions), considerably before the earliest priority date for the sofusbuvir patent in 2007. Claim 1 of this patent reads as follows:

There are other independent claims that are directed to a chemical compound, rather than to a therepeutic method, but claim 1 does seem to have at least the use of sofosbuvir within its scope. It is, however, a bit of a complicated puzzle to figure out whether it does. Can anyone confirm this?


  1. Even if sofosbuvir would be within the scope of the claims of the earlier reference, as long as the sofosbuvir molecule is not exemplified in said reference, the molecule itself will be considered novel (selection invention). Note that the opposed patent only claims a single molecule, not a broader group (Markush).

    As Sofosbuvir made it to the market, the proprietor will almost certainly have experimental data that shows that sofosbuvir has certain benefits in comparison to compounds of a similar structure (i.e. those that were tested in parallel in the early phases of the research but that did not make it to the clinical trials). Thus, the particular compound is not only novel over the broad suggestion in the prior art, but also has a technical effect that made it more suitable for therapy. Thus, the requirements of inventive step are likely fulfilled.

    Finally, enablement shouldn't pose a problem in view of it being a successful drug.

    All the above with the disclaimer that I didn't review the file wrapper and oppositions. Prima facie, it seems hard to revoke a patent with single compound claims for a successful drug if there is no prior art reference that exemplifies the molecule per se.

    1. I think you might have the wrong end of the stick. You're right about sofosbuvir being a selection invention, but the question is whether it would infringe the earlier patent, not whether it is novel and inventive over it.

  2. Interesting article. Regarding the puzzle...

    Well, there is a typo in the allegedly infringed claim: "one or R12 and R13...".

    Assuming it should read "one of R12 and ...", then no the claim is not infringed unless by "equivalence" - because "one of R12 and R13 is azido, and the other is hydrogen, halo, hydroxy...(plus others)".

    Whereeas on sofosbuvir (as pictured) R12 is Flouro and R13 is "hydroxy". So no "azido".

    Thus, in the absence of infringement by "equivalence" - i don't see an action succeeding!!


  3. Dear Tufty,
    - The method of treatment claim 1 appears to be limited to cancer treatment while sofosbuvir is used in HCV treatment;
    - Formula V provides that one or (of?) R12 and R13 is azido (i.e. -N3): sofosbuvir does not have an azido group (a fortiori at these positions;
    - Formula V does not provide that the carbon to which R12 is attached may be substituted both by -CH3 AND -F.
    Therefore I would be of the opinion that claim 1 is not literally infringed by sofosbuvir (equivalents are another story).
    Best regards,

    1. Thanks for that. I wondered about the cancer treatment point as well. There are other claims that are not limited in the same way, but they didn't look to me like they could encompass sofosbuvir. Perhaps the infringement claim is trying to stretch things a bit too far.

  4. I am not sure whether the compound is within its scope (I didn't review in detail) - but as it specifies the inhibition of cancer cell growth, and Sofosbuvir is (I think) approved for HCV - then even if it is within the chemical genus described would it impact its approved use? I didn't see any claims to an anti-viral use despite the title of the patent, it may be that this was removed during prosecution?

    1. Sofosbuvir is indeed approved for treatment of HCV, so that would be a difference on the face of it. Perhaps I should have started with claim 19 instead.

  5. Claim 19 of the US patent is directed to a composition of matter, rather than a method of treatment, but defines the same formula V as in claim 1. The points of difference noticed by two commenters above would therefore also seems to apply.

  6. Indeed, there are several independent compound claims, but all seem to require either an azido at R12 or R13 or a different group at position R11. It seems hard to argue equivalence if throughout the application, the teaching is that one of R12 or R13 should be an azido.

    However, do note that the patent also contains claims in relation to antiviral therapy (claims 72 and 73).

    Finally, HCV is a frequent cause of liver cirrhosis and liver cancer. There are arguments to be made that treating HCV infections encompasses cancer prevention. Although I do agree that the wording of claim 1 (inhibiting cancer cell growth) does not seem to automatically follow from HCV treatment.