Wednesday, February 16, 2011

Market exclusivity, data exclusivity and S. 3921

There has been some confusion lately regarding the distinction between “data exclusivity" and "market exclusivity.” For most newly approved drugs, US law only provides "data exclusivity," five years for small molecule drugs and 12 years under the newly enacted biosimilar legislation. During the period of data exclusivity, generic competitors are prevented from relying on data generated and paid for by the innovator to secure FDA approval for a generic or biosimilar version of the innovator drug. This is not the same as market exclusivity, because a generic company is free to generate all of its data from scratch to obtain marketing approval for their generic version of the drug, although in practice is rarely if ever happens. True market exclusivity would completely preclude generic competition, while data exclusivity simply provides a period of time during which generic competitors are not permitted to rely on innovator's data to obtain marketing approval.

In contrast, the Orphan Drug Act provides a period of true market exclusivity for certain orphan drugs, which precludes FDA from approving a competing version of the same or highly similar drug product to treat the same orphan disease for a period of seven years. The key difference between marketing exclusivity and data exclusivity is that a competitor cannot circumvent marketing exclusivity by generating its own data and submitting a new application for FDA approval, it is an absolute bar to FDA approval of the same drug for the same indication.

Recently, members of Congress sent a letter to the FDA seeking to clarify this point. In the letter, signed by four Senators including Orrin Hatch and John Kerry, the Senators point out that in a recent notice published by FDA seeking comments on the newly enacted biosimilar legislation, FDA incorrectly states that the biosimilar legislation provides a 12 year period of “marketing exclusivity.” The letter quite correctly point out that the biosimilar legislation in fact that does not include any marketing exclusivity period, but rather a 12 year period of data exclusivity.

Perhaps FDA should be excused for the lapse. Reporters and commentators routinely refer to the data exclusivity provided for biologics as "market exclusivity." To see what I mean, just do a Google search for “Obama budget biologic marketing exclusivity,” and you will pull up numerous reports from leading news agencies and bloggers reporting that the Obama budget proposes reducing the market exclusivity period for biologics from 12 years to seven years. In fact, there is no market exclusivity period for biologics, only data exclusivity, they should all read the letter from the Senators and stop propagating the misperception that the biosimilar legislation provides a period of market exclusivity.

I suspect part of the reason people have taken to characterizing data exclusivity as market exclusivity is because it suits the purposes of those who would like to speed to market entry by generic competitors by lowering the barrier to entry of generic competition. When you characterize it as market exclusivity, it sounds like market competition has been completely blocked for 12 years, which is very different from a 12 year period in which competitors cannot free ride on data paid for by the innovator.

For example, in a report on biosimilar legislation prepared by the FTC in 2009, a report of which I've been highly critical, the FTC repeatedly refers to the data exclusivity provisions of biosimilar legislation as marketing exclusivity. FTC's mischaracterization of data exclusivity as marketing exclusivity might have been an attempt to strengthen its argument by implying that no biosimilar product could enter the market during the 12 year data exclusivity period, when in fact this is not the case. But this mischaracterization has spread, apparently leading to confusion even at FDA, and thus necessitating this letter from the Senators.

Related to this debate is the "Ethical Pathway Act of 2010," (S. 3921), a bill introduced into the Senate by Senator Sanders on September 29, 2010, which would effectively eliminate any data exclusivity period for both conventional drugs and biologics. Under this proposed legislation, as I read it, an innovator would be forced to share its clinical and animal data with a competitor, to be used by the competitor to obtain FDA approval for a generic or biosimilar product. The bill includes provisions for a "cost-sharing arrangement," pursuant to which the generic company would be required to compensate the innovator for some of the costs incurred in generating the data. If the innovator and generic company cannot agree on the amount, they are required to submit the matter to an arbitrator who is to determine a "reasonable and fair fee.”

The bill identifies specific factors to be considered in determining the "reasonable and fair fee," including "actual out-of-pocket costs of the applicable clinical investigations, the risk of the investigations, as reflected in the probabilities that similar investigations result in successful applications for marketing, any federal grants, tax credits, or other subsidies that reduce the net cost of the investigations, the expected share of the global market for the product involved, by the party seeking to rely upon the investigations for marketing approval, and the amount of time the holder or holders of the relevant applications for licenses has benefited from exclusive rights, and the cumulative revenue earned on the products that relied upon the regulatory tested at issue."

I think it is reasonable to predict that innovators are highly likely to be undercompensated by this imposed "cost-sharing arrangement." Essentially, after the innovator company has taken on a huge amount of risk bringing a drug to patients, and then establishing a market for the drug, this bill would allow a competitor to free ride off all of this risky investment by simply paying some fraction of the cost for clinical trials. As reported last Monday in Reuters, a survey has found that traditional small molecule drugs have only a 7% chance of making it from phase I through FDA approval. The success rate is even lower for some of the most important drugs-4.7% for cancer drugs, and 5.7% for cardiovascular drugs. There needs to be powerful financial incentives to invest huge amounts of money when there is such a low probability of a payoff, which I think helps explain why we are seeing such a dramatic drop-off in pharmaceutical R&D spending and in the rate of new drug approvals.

When generic competition hits the market, the innovator does not just lose a piece of the pie, the pie shrinks dramatically. Innovators need some substantial period of exclusivity on the market in order to bring in the profits that ultimately drive innovation. Patents play an important role in this regard, but patent protection is unpredictable and not always available, even for important and innovative life-saving drugs, and this is particularly true in the case of biologics.

To me, this is analogous to me going to the racetrack with a friend, and deciding to bet $1000 on a longshot horse. If the horse comes in, and I win $25,000, it is certainly not fair for my friend to wait until the race is won and then demand that I accept $500 (half the price of the bet) in exchange for half of the winnings ($12,500). But this is essentially what the "cost-sharing arrangement" under the proposed legislation would entail. In fact, it's even worse than that, because to take the analogy further, since generic competition reduces profits for everyone, we're not talking about sharing $25,000 anymore, but substantially less than $25,000.

While the legislation would hurt innovators, and ultimately innovation, it is premised upon a quite legitimate ethical concern, i.e., conducting clinical trials on human beings when we presumably already know the product is safe and effective raises substantial ethical issues. While I agree this is an entirely valid concern, I don't think it should be addressed in a manner that substantially reduces incentives for future innovation.

I would suggest that perhaps the best way to deal with these concerns is to provide drug innovators with an actual period of marketing exclusivity, independent of patent rights and independent of data exclusivity. This would give innovators a period of market exclusivity in which to recoup profits that incentivize innovation, without having to worry about patent challenges, and without subjecting human subjects to unnecessary clinical trials.

Thursday, February 10, 2011

Does disclosure of laundry list of protein variants provide adequate written description for later claim to specific protein?

The traditional written description requirement prevents patent applicants from amending their claims during patent prosecution to encompass subject matter not disclosed in the patent specification as filed. Clearly, a patent specification that specifically discloses (without claiming) a small number of protein sequences provides adequate written description to support adding a claim later that recites one of those protein sequences. But what if a patent specification effectively discloses all protein sequences, for example, by generically describing "a protein comprising between 10 and 1000 amino acids, wherein at each position in the sequence the amino acid can be any of the 20 genetically coded amino acids.” It defies common sense to think that such a broad, generic disclosure would be sufficient to satisfy the written description requirement with respect to a claim introduced by amendment specifically reciting any protein of this size, even though the disclosure does literally describe any possible protein of 10 to 1000 amino acids in length.

The question then becomes, at what point do we draw the line? In other words, to what extent is it possible for a patent applicant to satisfy the written description requirement with respect to a later added claim directed towards a specific protein sequence by simply disclosing in the specification as filed an astronomical "laundry list" of protein sequence variants that happens to include the later claimed proteinsequence? This is the question that was presented to a district court recently in the lawsuit brought by Novozymes against Danisco/Genencor, alleging infringement of a patent claiming certain alpha-amylase variants genetically engineered for improved thermostability. (Last September I reported on the district court's denial of a motion for preliminary injunction by Novozymes, available here).

The Federal Circuit (and its predecessor court the CCPA) has addressed this question in the context of traditional small molecules. A seminal case is In re Ruschig, a 1967 CCPA decision that held that the mere disclosure of a large genus of molecules is not necessarily sufficient to satisfy the written description requirement with respect to a later claim directed to some species falling within the genus. Under such circumstances, the Ruschig court held that in order to satisfy the written description requirement, the specification must provide "blaze marks" specifically pointing towards the later claimed species.

In 1996, the Federal Circuit applied the Ruschig "blaze mark standard” in Fujikawa v. Wattanasin, and held that the disclosure of a genus did not provide adequate written description of a later claimed sub-genus, pointing out that "just because a moiety is listed as one possible choice for one position does not mean there is [adequate] support for every species or sub-genus that chooses that moiety. Were this the case, a 'laundry list' disclosure of every possible moiety for every possible position would constitute a written description of every species in the genus. This cannot be because such a disclosure would not 'reasonably lead' those skilled in the art of any particular species."

In the Novozymes v. Danisco, the issue was raised in a motion for summary judgment filed by Danisco, arguing that a disclosure in Novozymes specification of a laundry list comprising an astronomical number of protein variants does not provide adequate written description support for a claim added later directed toward a small subset of those variants. The Novozymes patent specification, filed in 2000, disclosed variants of a specific alpha-amylase (the sequence of which was disclosed in the application) comprising an alteration at one or more positions selected from a list of 33 possible positions. The specification went on to define an alteration as either an insertion downstream of the amino acids which occupies position, a deletion of the amino acid which occupies a position, or a substitution of the amino acid which occupies the position with a different amino acid. So there are multiple variable parameters.

The 33 positions can be independently altered in three different ways. Any number of the 33 positions can be altered independently, and there are 20 different amino acids from which to choose in deletions or substitutions. Genencor did the math, and concluded that this amounted to a “staggering” 8.598 x 1042 different possibilities, a long laundry list indeed. Novozymes did not dispute this calculation. Furthermore, the specification does not specify what effect a specific mutation would have on thermostability, although one passage in the specification discusses "achieving altered stability, in particular improved stability (i.e., higher or lower), at especially high temperatures."

In 2009, nine years after the specification was filed, Novozymes added the claims at issue in the case, directed specifically toward a subset of the proteins on the laundry list that have a substitution at position 239 in amino acid sequence, resulting in increased thermostability. That is, the claim is limited to one of the 33 amino acid positions identified in the specification as filed (position 239), one of the three alterations (substitution), and one of the possible alterations in stability (increased thermostability).

The court discussed applicable precedents such as Ruschig and Fujikawa, but did not really seem to demonstrate a thorough understanding of the principles behind those decisions. Perhaps this is because, as the judge complained in her decision, "a general difficulty in this case has been the party's tendency to argue past each other." Nonetheless, she ultimately concluded that although "I still have doubts that the specification . . provides an adequate written description for the claims, I conclude that the defendants have not met their burden to prove by clear and convincing evidence that the [patent] is invalid as a matter of law." The decision was largely dictated by the cases procedural posture, which imposed a substantial burden of proof on Danisco in attempting to invalidate an issued patent on summary judgment as a matter of law.

The judge explicitly noted that "it is not without hesitation that I'm denying defendant's motion.” It seems clear that she was bothered by the fact that the originally filed specification discloses such an astronomical number of "possible inventions" without seeming to provide any "blaze marks" directing one towards the later claimed variants. But she went on to point out that even if the claims comply with the written description requirement, "to the extent the specification would require undue experimentation before a person of ordinary skill in the art could discover the claimed invention, that may suggest a lack of enablement rather than a problem with the written description." I think she might have a point there.

A copy of the decision is provided here. Thanks to Docket Navigator ( for making me aware of this case.