Patent Docs: What’s in a name? When the chemical structure of an API is wrong

Drug regulation and licensing (and the patents that claim active pharmaceutical ingredients) are based on accuracy, including accurate descriptions of the molecule, how it is formulated and (to the extent it is understood) how it works. But what happens when there is an error? After all, mistakes happen; there is even a book, entitled Human Error, that discusses how and why they happen. The Federal Circuit addressed the consequences of human error (or perhaps more accurately, instances where there was a less-than-perfect understanding of the chemical structure of a claimed invention) in a surprisingly lenient fashion in Cubist Pharmaceuticals, Inc. v. Hospira Inc. And the reasoning behind that answer harkens back to the time when the technology for determining structure was not as robust as it is now, and when biological molecules comprising chains of amino acids were not routinely defined (and disclosed) by their sequence but rather by how they were isolated or otherwise produced.

 

The decision arose from ANDA litigation involving daptomycin (sold by Cubist as Cubicin), an antibacterial agent originally developed by Eli Lilly & Co. where the patent on the molecule per se had expired long ago (2002). Hospira argued that the patent claims at issue were invalid because, as it turns out, the patent (as well as all the regulatory documents and the understanding in the art) was mistaken regarding the structure of the daptomycin itself. Specifically, it was understood that the molecule comprised an L-asparagine residue at a particular position in the chemical structure, but instead it was later found that this residue was a D-asparagine.

 

The original specification contained three descriptions of daptomycin. The first was as “an A-21978C cyclic peptide” prepared from “a group of closely related, acidic peptide antibiotics” described in U.S. Patent No. 4,208,403” and produced by fermentation of Streptomyces roseosporus. The second mode of describing daptomycin in the specification was by a code name assigned by Eli Lilly (the company that originally discovered the drug) that was known in the art as being specific for daptomycin. The third way was by an expressly disclosed chemical structure containing L-asparagine rather than D-asparagine.

 

“At the time the [patent] application was filed, and until well after that patent was issued, it was universally believed that the asparagine amino acid in daptomycin was the L-isomer of asparagine, as set forth in the structural diagram,” according to the court, and it was only years afterwards that the true stereochemistry of the molecule was discovered. Cubist availed itself of provisions in patent law that permits changes “of minor character” that do not broaden the scope of the claims, and Hospira challenged this characterization of the correction. The District Court disagreed, saying that the change was merely to the formula (i.e., the depiction of the compound’s structure) and not any change in the compound itself, and the Federal Circuit agreed saying that: “a chemical structure is ‘simply a means of describing a compound; it is not the invention itself'.” and further that “the specification as a whole must be considered” not the chemical structure alone. In addition, the patent specification taught that daptomycin is obtained through fermentation of Streptomyces roseosporus, which necessarily results in daptomycin, not the variant with the L-isomer of asparagine. The evidence at trial established that the L-isomer variant cannot be produced by fermentation and can only be produced synthetically.

 

The Federal Circuit’s decision in this case is reminiscent of the days before the biotechnology revolution provided purified proteins as the result of recombinant genetic methods wherein the amino acid sequence of the produced proteins were predicted from the cloned nucleic acids encoding them. Prior to that, purified proteins were isolated from natural sources (for example, insulin from animal pancreata obtained inter alia from slaughterhouses). Patent law recognized it to be sufficient for an applicant to disclose a reliable source of the desired protein and methods for isolating it in useful quantities. But a consequence of this regime, mandated by the technology available, was that any “purified” preparation of a protein from a natural source was likely not to be homogeneous.

 

It is well known that animals in an outbred population will have a plurality of alleles of a gene for almost any protein that have small variances in amino acid sequence. This population heterogeneity was recognized but not considered relevant to providing a disclosure that supported claims to the isolated protein preparation. In contrast, producing proteins by recombinant genetic technology imposed a homogeneity on the produced protein product that may by itself be sufficient to distinguish such preparations from preparations made from natural sources. But it also changed (some would contend, raised) the quanta of disclosure needed to satisfy the written description requirement so that undisclosed sequence differences became relevant to claim validity.

 

In this decision the Federal Circuit eschewed such sequence-dependent reasoning and relied instead on the totality of what the specification disclosed. The decision was based in large part on the identity of what S. roseosporus fermentation produced, and the irrelevance (for the panel) of the erroneous structure contained in the specification (and claims). This isn’t a right-or-wrong approach; rather, it relies (as many patents granted in the decades prior to biotechnology did) on identifying a reliable source of a natural product and an assessment of what the specification provided. Which, even in the biotechnology era, seems a refreshing application of sound patent law.