Thursday, April 24, 2008

PTO Issues Revised Written Description Guidelines, Further Muddying the Waters

On March 25, 2008, the PTO issued revised written description training materials to supersede and replace the interim written description guidelines training materials published in 1999. The 1999 guidelines were published in response to Regents of the University of California v. Eli Lilly, a controversial 1997 Federal Circuit decision which has been widely interpreted as establishing a new form of the written description requirement applicable to originally filed claims, and particularly targeting biotechnology inventions. As of 1999 Lilly was the only Federal Circuit decision applying this novel form of the written description doctrine. However, since that time the Federal Circuit has decided a number of cases involving what I refer to as the “Lilly written description requirement,” or LWD (to distinguish it from the traditional written description requirement, a doctrine which serves to police against patent applicants amending or adding new claims encompassing subject matter not adequately described in the specification as filed), including Enzo, Rochester, Noelle, and Wallach.

Both versions of the guidelines attempt to explain the application of LWD by means of a series of examples, nearly all arising out of biotechnology, including hypothetical claims directed to genes and other polynucleotides, antibodies and other proteins, and bioinformatic inventions. The 2008 guidelines include a number of new examples specifically directed towards the holdings in Wallach, Noelle and Rochester (Examples 5, 14 and 17, respectively), and also retain many of the original examples from the 1999 guidelines. Most notably, there are a several instances where the 2008 guidelines substantially diverge from the earlier version.

For instance, Example 6 from the 1999 guidelines concludes that a claim reciting an “isolated gene comprising SEQ ID NO:1,” wherein SEQ ID NO:1 represents the sequence of a novel “cDNA fragment,” is invalid for failing to comply with LWD. The 2008 guidelines do not include an example reciting an "isolated gene," but do include Example 4, which includes claims reciting an "isolated DNA comprising SEQ ID NO:16” and an "isolated nucleic acid comprising SEQ ID NO:1,” wherein the SEQ ID NO:’s represent EST sequences (essentially another word for cDNA fragments). In essence, the only difference between the claims is the substitution of the word DNA or nucleic acid for gene. Interestingly, the 2008 guidelines find that the claims reciting a DNA or nucleic acid are valid under LWD.

Why the opposite results with such very similar claim language? The 1999 guidelines explain that mere disclosure of the cDNA fragment is inadequate to describe the claimed genus of genes, because the regulatory elements and untranslated regions generally associated with genes, e.g., introns and promoter regions, are not described. In contrast, the 2008 guidelines conclude that because the presence of the recited cDNA fragments “defines the scope of the claim genus," and because "it is within the level of skill and knowledge to add any desired DNA sequence to either end of [a cDNA fragment] with no more than routine extermination," one skilled in the art would recognize that the applicant was in possession of the structural features shared by members of the claimed genus.

The 2008 guidelines specifically recognize that the nucleic acid and DNA genuses of Example 4 would include the full length open reading frame, fusion constructs and vectors comprising the cDNA fragment, and might even include the full-length genomic gene in cases where the cDNA fragment is derived from a single exon. Thus, under the new guidelines a short EST sequence representing only a fragment of a single exon is sufficient to claim the full length genomic sequence, in spite of the fact that the regulatory elements and untranslated regions required by the 1999 guidelines are not disclosed, to say nothing of the other non-disclosed exons.

Thus, with respect to claims to nucleic acids comprising an EST sequence, the 2008 guidelines provide for a substantial lowering of the bar to satisfying LWD. This interpretation of LWD is consistent with Ex parte Fisher, the BPAI decision that was the subject of the Federal Circuit’s in In re Fisher decision, which essentially found EST sequences invalid under the utility requirement. In Fisher, the examiner had rejected claims to EST sequences for failure to comply with LWD, noting that additions to the disclosed partial cDNA sequences would confer function not possessed by the originally disclosed fragments. The board reversed, however, finding that a claim encompassing any polynucleotide comprising a partial cDNA sequence did not raise written description issues. The board failed to explain its rationale for this determination. On appeal, the Federal Circuit limited its inquiry to the utility issue, and did not address the board’s questionable interpretation of LWD, which is now apparently embodied in the guidelines.

An even more direct conflict exists between Example 9 of the 1999 guidelines and Example 6 of the 2008 guidelines. Both involve claims directed towards an isolated nucleic acid that specifically hybridizes under highly stringent conditions to the complement of a disclosed sequence, wherein the nucleic acid encodes a protein having the functional activity of the protein encoded by the disclosed nucleic acid sequence. The 1999 guidelines find that this claim satisfies the written description requirement, while the 2008 guidelines finds essentially the same claim invalid. The 1999 guidelines found that a person of skill in the art would not expect substantial variation among species encompassed within the scope of the claims because the highly stringent hybridization conditions would be expected to yield structurally similar DNAs. In contrast, the 2008 guidelines conclude that those of ordinary skill in the art would not be able to identify without further testing which of the nucleic acids that satisfy the hybridization requirement would also encode a polypeptide having the recited functional attributes, based on the unpredictable relationship between protein structure and function.

Similarly, Example 14 of the 1999 guidelines found that a claim directed to protein variants sharing at least 95% sequence identity to a disclosed protein sequence and retaining the functional activity of the disclosed protein complied with LWD, while Example 10 of the 2008 guidelines finds the same claim to be invalid. Again, the 2008 guidelines focus on the unpredictability of the protein structure-function relationship, pointing out that one of skill in the art would be unable to identify which variants satisfying the 95% identity criteria would also retain the recited function.
Interestingly, the same Example 14 finds a broader claim, directed to any variant sharing 95% identity with the disclosed protein sequence, (i.e., not limited to functional variants, as was the case with the invalid claim) to comply with LWD. In other words, the guidelines would find that a subgenus claim is invalid for failing to comply with LWD, while a broader genus claim that encompasses the entire unpatentable subgenus does satisfy LWD. Thus, the guidelines teach that a patent applicant can overcome an LWD rejection to a percent identity claim limited to functional variants by simply broadening the scope of the claim to include non-functional variants.

In 2007 I published an article reviewing all the federal court and Board of Patent Appeals and Interferences decisions I could find which applied LWD. In the article, I noted that relatively few patent claims had actually been invalidated under the doctrine, that the courts and the board have routinely found very broad claims relating to genetic and biotechnology inventions to comply with LWD, and that in most (arguably all) of the cases where LWD has been used to invalidate a claim it has been applied in a manner that was explicitly or implicitly redundant with invalidation by means of the enablement requirement. My conclusion was that LWD was not playing any positive doctrinal role that could not be accomplished by means of the enablement requirement, particularly as it has been applied in cases such as Amgen v. Chugai.

However, since I conducted my study, the PTO appears to be heading in the opposite direction, seeking to reinvigorate LWD by carving out a role for the doctrine distinct from the enablement and utility requirements. I think this is a mistake, and this is reflected in the confused logic that runs throughout both versions of the guidelines. Although the guidelines speak in terms of a "possession” test for compliance with LWD, they fail to articulate a standard of possession that can be meaningfully distinguished from enablement. In explaining the outcomes arrived at in its examples, the 2008 guidelines repeatedly assert that one of skill in the art would (or would not) have concluded that the applicant was in possession of invention, without providing any guidance with respect to the nature of the general test for “possession.”

It seems clear that some of the examples in the 2008 guidelines found to comply with LWD would be held unpatentable on other grounds, such as lack of utility and/or enablement. For example, I suspect that the claim directed to 95% identical protein sequences without a functional limitation might be considered invalid for lack of enablement, which could resolve the anomaly that under the guidelines the claim satisfies LWD, while a subgenus claim limited to functional variants is invalid under LWD. What might really be useful would be a more holistic guidance document explaining how LWD should be applied in conjunction with the other 112 requirements, such as utility and enablement.