Mathematical Scrutiny: Genus Claims and Enablement Analysis at the Federal Circuit
The grant of a patent is a kind of social quid pro quo. We, the public, confer a monopoly right on the patentee for the duration of the patent, provided that “those ‘skilled in the art’ of the invention can read and understand the inventor’s contribution, and[,] after the patent expires[,]…make and use the invention themselves.”[1] This is meant to guard against the potential anticompetitive and constrictive effects of a patent with expansive claims, where such claims are not supported by “adequate teaching” or where the inventor is not “in possession of the inventive principle.”[2] Put differently, the patent provides an exclusionary right to the inventor, so long as the disclosure is “commensurate with the scope of protection sought by the claims.”[3] In modern patent practice, the need for adequate teaching—the enablement requirement—is satisfied if a person having ordinary skill in the art (“PHOSITA”) can “make and use the invention without undue experimentation.”[4] The determination of whether a disclosure would require “undue experimentation” rests on a multi-factor test set forth by the Federal Circuit.[5]
To complicate matters further, a particular invention may be utilized in several different ways, each of which is called an embodiment.[6] Often these embodiments are analogized to individual species, whereas the entire set constitutes a genus.[7] Moreover, the claims of a patent may exhibit both structural (e.g., “Composition comprising X”) and functional (e.g., “Composition comprising X, wherein X, when bound to Y, binds to at least one of {A, B, C,…O} and blocks Z”) limitations.[8] With respect to genus claims and “broad functional language,”[9] the Federal Circuit has found lack of enablement where undue experimentation would be required to “make and use the full scope of the claimed compounds.”[10] In other words, this type of language “raises the bar for enablement.”[11] As a result, many prospective patentees face a dilemma. If the claims are drafted too broadly without an appropriate level of disclosure, there is a risk of invalidation. Conversely, if they are drafted too narrowly, others may be able to “‘design around’ the patented invention by varying an insignificant part of the total structure [or function].”[12] This predicament is palpable in the context of chemical and biotechnology patents, where a compound may have millions or even billions of analogs.[13] Indeed, the multi-factor test discussed above arose in In re Wands, a seminal case involving “antibody technology.”[14]
Today, the issue of enablement is back in the news. In November 2021, Amgen filed a petition for a writ of certiorari, asking the Supreme Court to review a Federal Circuit decision that affirmed an invalidation of some of Amgen’s patent claims.[15] The patents in question “describe antibodies that purportedly bind to the PCSK9 protein and lower LDL levels by blocking PCSK9 from binding to LDL receptors.”[16] Back in October 2011, Amgen obtained a patent on an antibody with the foregoing parameters, and it did so by claiming a specific amino acid sequence.[17] The next month, Sanofi-Regeneron[18] obtained a patent on a different “anti-PCSK9 antibody” with a different amino acid sequence.[19] So far, so good. In the fall of 2014, however, Amgen was granted two additional patents (the “’165 and ’741 patents”), each containing claims defined “not by structure, but by meeting functional limitations.”[20] Thus, for example, claim 1 of the ‘165 patent reads as follows: “An isolated monoclonal antibody, wherein, when bound to PCSK9, the monoclonal antibody binds to at least one of the following residues [followed by a list of 15 amino acid residues], and wherein the monoclonal antibody blocks binding of PCSK9 to [LDL receptors].”[21] In a common specification, Amgen disclosed “amino acid sequences for twenty-six antibodies, including the antibody” claimed in the original 2011 patent.[22] Shortly after obtaining the ‘165 and ’741 patents, Amgen sued Sanofi-Regeneron for infringement, arguing that Sanofi-Regeneron’s PCKS9 inhibitor was covered by the “class of antibodies” claimed in Amgen’s patents.[23] Following two trials and extensive motions practice, Amgen found itself in a somewhat ironic position: the very claims it had been trying to enforce were held to be invalid.[24]
Although the Federal Circuit considered all of the Wands factors, for purposes of this post, I would like to focus briefly on one: “the quantity of experimentation necessary.”[25] This factor is tied closely to another common dichotomy in patent law, that of predictable and unpredictable arts.[26] Chemistry, particularly in nascent fields, falls squarely within the latter category.[27] Generally speaking, there is an inverse correlation between the predictability of an art and the quantity of experimentation required to “make and use” such art (i.e., more predictable art requires less experimentation).[28] Amgen and its supporters, including amici curiae consisting of large pharmaceutical companies and a group of intellectual property law professors, have framed the Federal Circuit’s decision as the latest evidence of a shift towards a more onerous enablement standard.[29] To quote from a recent academic article upon which Amgen and amici rely greatly, “the Federal Circuit seems to implicitly assume that the PHOSITA must test every species within the genus for enablement purposes.”[30] Apparently, this purported distortion of the enablement standard has elevated the importance of the “quantity of experimentation necessary,” along with an additional, interrelated Wands factor: the “breadth of the claims.”[31] Even if that is so, Amgen and amici seem to misinterpret the “full scope” gloss on enablement. In these cases, the Federal Circuit is looking at the probability that a PHOSITA will “discover [any] undisclosed claimed embodiments,”[32] not the prospect of “find[ing] every species that works to make and use the invention.”[33]
While the Wands test is not entirely quantitative, some math may help to clarify the issue.[34] In the specification, Amgen disclosed “3,000 antibodies that bind to PCSK9, which Amgen narrowed down to 85 that blocked the interaction between PCSK9 and LDL receptors by 90% or more.”[35] So, excluding the 3,000 antibodies identified by Amgen, let us assume that the (theoretical) set of all remaining antibody candidates is 3,000,000, and the number of PCSK9-binding antibodies therein is 30,000. Assume further that, with “well-known screening techniques” and “anchor antibodies” and all the guidance in the specification, a PHOSITA is able to reduce the size, or cardinality, of the set by one order of magnitude.[36] At this point, a PHOSITA would likely have to rely on “trial and error.”[37] With 30,000 viable antibodies in an adjusted set of 300,000, a PHOSITA would have a 1/10 chance of discovering an undisclosed antibody that binds to PCSK9. One final assumption: given the probability of identifying a suitable antibody (1/10), the probability that such an antibody will “block[ ] the interaction between PCSK9 and LDL receptors” is 100/3,000, or 1/30—a rounded figure based on the 85 antibodies identified by Amgen, out of a pool of 3,000. Thus, the probability of discovery is calculated by multiplying 1/10 and 1/30. The answer is 1/300, or .33%. Obviously this is a gross oversimplification, and my assumptions may be unwarranted or misguided. But if these odds approximate reality, it is easy to see why the Federal Circuit views such broad claims with some degree of skepticism. Perhaps the Supreme Court will be more sympathetic.
Footnotes