Caught in a Time Warp: The (In)validity of BRCA1 Oligonucleotide Claims

    By Kevin E. Noonan --

Albeit a bit anticlimactically, a study published in the journal Genomics assesses the patentability of one of the claims invalidated on Monday by Judge Robert W. Sweet of the Southern District of New York in Association of Molecular Pathology v. U.S. Patent and Trademark Office.  The study is interesting in that it illustrates the pitfalls in patents on oligonucleotides related to isolated genes, filed at a time prior to the elucidation of the human genome by the Human Genome Project.  Most relevant to the question of whether such claims are patentable are the many revelations from the HGP that overturned settled dogma based on a now-recognized unrealistic view of the randomness of genomic DNA sequences, as illustrated by the results reported in this study.

The research was performed by Thomas B. Kepler, from the Department of Biostatistics and Bioinformatics at Duke University, and the Center for Computational Immunology; Colin Crossman, a lawyer and owner of Memento Mori, LLC; and Robert Cook-Deegan from the Institute for Genome Science and Policy and the Sanford School of Public Policy, Duke University (Kepler et al., "Metastasizing patent claims on BRCA1").  According to these authors, one of the claims of U.S. Patent No. 5,747,282 seemed "exceptionally broad" -- this is claim 5 that depends on claim 1 (which is directed to an isolated DNA molecule encoding a BRCA1 polypeptide):

1.  An isolated DNA coding for a BRCA1 polypeptide, said polypeptide having the amino acid sequence set forth in SEQ ID NO:2.

5.  An isolated DNA having at least 15 nucleotides of the DNA of claim 1.

The authors estimate that "the human genome contains over one million oligonucleotides covered by this claim," and follow up this statement with the following calculations:

Accounting for bias in the usage of amino acids as reported, for example, in [7], the usage-weighted geometric mean codon degeneracy per amino acid is 3.107.  Therefore, the mean number of 15-mers encoding a polypeptide of length 5 chosen at random from a vertebrate proteome is 3.107⁵, about 290.  There are 5,575 15-mers in BRCA1, so, if we consider all of the nucleotide sequences that encode the BCRA1 protein, there are about 1.6 x 10⁶ 15-mers embodied by the claim.  There are 4¹⁵=1.07 x 10⁹ different 15-mers altogether, so the probability that a 15-mer chosen at random will be covered by the claim is p=1.6 x 10⁶ / 1.07 x 10⁹=0.0015 (roughly, 1 in 600 possible 15-mers).  A typical human gene (before RNA editing) contains 10,000 bases, so, if human genes were random strings of nucleotides, one would expect a human gene to contain an average of 15 15-mers claimed under the patent.

The predicted results of this analysis were found in the study.  The authors analyzed the nucleotide sequence of human chromosome 1 looking for only a subset of 15-mers encompassed by claim 5.  ("Computing time" was minimized by excluding two of the six degenerate codons for serine, leucine and arginine amino acids, a reduction said to only "slightly underestimate the degree of redundancy and breadth of claim 5.")  The authors reported finding 340,000 "matches" of the claimed sequences in the 250,000,000 nucleotides comprising chromosome 1.

The authors also examined 713 entries in GenBank representing complete coding sequences for human mRNAs deposited in 1994, reporting that 80% (568/713 contained at least one of the claimed 15-mers.

There is nothing incorrect about this analysis; however, it benefits from current knowledge and (in its implied conclusion of intentional overreaching) neglects to consider the state of the genomics art on August 12,1994, the earliest priority date of the '282 patent.  The calculation was much more simple (or perhaps naïve) then:  any particular 15-mer was expected to occur once in every 1.07 x 10⁹ nucleotides, and thus in a completely random genome the size of the human genome (haploid size of about 3 x 10⁹ nucleotides), to occur about 3 times; a 16-mer would be expected to occur four times less frequently, etc.  Thus, the results set forth in this paper were almost completely unexpected when approached from the outlook of the person of skill in the art in 1994.

As it turns out, the human genome (and most other genomes) are much more inhomogeneous than expected, and the effects of evolution and the relatedness of all organisms (as well as the conservation of motifs and functional domains between species) were equally unexpected.  Indeed, even the number of genes encoded in the human genome turns out to be much smaller (2- to 3-fold fewer genes) than was expected.  As the authors state:  "human genes are not random strings [of nucleotides.]"  Evident now, not so evident 16 years ago.

No one wants an invalid patent.  The in terrorem effect of such a patent is greatly exaggerated, particularly when its invalidity is so easily demonstrated.  Here, these claims, as well as similar claims in other patents, turn out to be sufficiently overbroad as to be easily invalidated.  It is unlikely that the University of Utah, the National Institutes of Health, or Myriad Genetics (all owners of this patent) wanted this result.

Nevertheless, the remainder of the paper discussed the policy implications of this invalid claim.  The authors correctly note that their results indicate that claim 5 was anticipated by the 15-mers known in the prior art as exemplified by the GenBank results.  Thus, "[i]f challenged by re-examination or in litigation, claim 5 may be deemed invalid due to readily identifiable prior art covered by the claim."  The paper cites U.S. Patent Examiner James Martinell, examining a 1991 expressed sequence tag patent application, for recognizing that certain 15-mers could be found in "many genes" and that it was impossible using then-current computer technology to search all known sequences for the 700,000 15-mers claimed in that patent.  Perhaps for that reason, there is no evidence in the prosecution history of the '282 patent that such a search was ever performed.

Despite this strong evidence that claim 5 of the '282 patent is invalid (even before the District Court's decision yesterday), the paper speculates on the scope and reach of the claim (including oligonucleotide primers for performing the polymerase chain reaction).  Regarding the effects of this claim (and Myriad's patents) on basic research, the authors recognize what many others have noted:  no appreciable effect (which belies the introductory sentence of this section that "[t]he effect of this claim on research is very difficult to assess"):

A PubMed search for the term "BRCA1" returned 7,107 articles.  This suggests a large body of research on the gene has been published in the technical literature.  Myriad has not enforced its patents against most research, with the exception of laboratories engaged in clinical research that entailed giving test results to individuals beyond their home institutions [11–13].  Any such research that entailed analysis of DNA molecules containing BRCA1 sequences in the United States very likely infringed this claim, however, so enforcement of this claim would have substantial impact on research.  (Claims to BRCA1 sequences are somewhat narrower in other English-language jurisdictions such as Canada, Australia and New Zealand, and a fortiori in Europe, where the claims that emerged from opposition proceedings were dramatically narrowed.)  There is a very narrow "research exemption" from infringement liability in the United States under common law, and a broader exemption for research that results in data contributed to the government for a regulated medical product or service [14].  Since laboratory-developed tests are not currently subject to Food and Drug Administration approval, however, this exemption may not apply.

The simplest conclusion about the effect of claim 5 and Myriad's other BRCA1 patents on research and clinical testing is that Myriad has only rarely enforced its patents in research, has vigorously enforced its patents against commercial genetic testing, and has selectively enforced its patents in clinical research.  It is also apparent that research on BRCA1 for the past 12 years has entailed massive pervasive infringement of this claim, even if the claim's scope were restricted to BRCA1 research.  Any such research in the United States was thus undertaken under risk of infringement liability and its associated uncertainty.  While Myriad has stated publicly that it has not enforced its patents against basic research [11,12,15], it has not stated it will not do so in the future, and therefore BRCA research in the United States continues only with Myriad's indulgence.

The authors are correct that there is no exemption from infringement liability for basic research, but they admit that Myriad has not enforced (or attempted to enforce) the '282 patent against any basic researchers.  The absence of the exemption may be the problem, not claim 5 of the '282 patent (in view of the results of this research).  Myriad's track record with regard to basic research is consistent with its public statements that it would not enforce its patents against basic research; there is very little else it can do.  And after yesterday's decision, basic or clinical researchers or commercial entities that practice the oligonucleotides recited in claim 5 no longer risk infringement liability.

Comments

[Devesh Srivastava]

Whether to enforce a patent claim is a business decision. To spend the money necessary to litigate a claim, there must be a monetary reward (i.e. lost profit or royalty stream).

In the case of academic researchers, and especially those at state institutions who may enjoy 11th amendment immunity, why would Myriad waste their resources to pursue legal action when the monetary return is not worth the expenditure?

[EG]

Kevin,

Good article. And thanks for addressing the important question: is Myriad's patented technology novel (and if it is) unobvious over the prior art?

Judge Sweet's opinion yesterday in invalidating Myriad's patents under 35 USC 101 amounts to a judicial enactment of the Becerra bill. That's "judicial activism" at its worst. We don't need that if we're to have any respect for our judicial system.

[Allison Williams Dobson]

Judicial activism with hootzpah!

[WE]

As Professor Andrew Chin showed, there's a way of defeating all oligonucleotide claims regardless of the outcome of the Myriad litigation -- just put the sequences on a CD-ROM and deposit it in a library. http://unclaw.com/chin/scholarship/artfulpriorart.pdf

[Keep It Real]

I'll never understand why you feel compelled to defend Myriad's greed, Kevin. They filed on a lot of garbage. Deal with it.

"Thus, the results set forth in this paper were almost completely unexpected when approached from the outlook of the person of skill in the art in 1994."

Really? I have no idea how you reach this conclusion. Note that there is no reason that searching the database circa 1994 for sequences that anticipate or render this claim obvious would need to be limited to deposited *human* sequences.

The claim was a bad joke when it was filed and it stinks now. It was plainly overreaching when it was filed. What I would like to know is: why did Myriad stop there? They should have filed claim 6 on "antibodies that bind to proteins encoded by claim 5." And then claim 7: "A DNA encoding an antibody of claim 5." Etc. etc. Maybe Myriad just ran out of paper?

Also, is it just me, or is claim 5 broader than claim 1, from which it depends? Is that not a problem?

[EG]

"I'll never understand why you feel compelled to defend Myriad's greed, Kevin."

KIR,

Why do you assume that Kevin is defending Myria's "greed"? The fact is both the ACLU, as well as Judge Sweet, are resorting to "judicial activism." What the ACLU and Judge Sweet wanted was in the Becerra bill which didn't pass. Why is it now right for them to grossly distort the patent law precedent, as well as mischaracterizing what Myriad's patents cover?

The fact is Sweet's decision is going to be trounced by the Federal Circuit. That's not because of "judicial fiat," but by following the patent law precedent, as well as construing Myriad's claims properly (which don't cover the genes and are not simply a "product of nature"). If you don't believe me, then just wait to see how Sweet's decision is handled by the Federal Circuit (and be prepared for a judicial version of a verbal "blood bath").

In fact, as I pointed out above, I'm perfectly fine if Myriad's patents go down because they lack novelty or are obvious over the prior art. In other words, judge the claimed invention on the merits, not based on some "moral imperative" that ignores, distorts, and mischaracterizes the relevant facts, as well as the relevant patent law precedent.

[Kevin E. Noonan]

Dear Keep:

"No one wants an invalid patent. The in terrorem effect of such a patent is greatly exaggerated, particularly when its invalidity is so easily demonstrated. Here, these claims, as well as similar claims in other patents, turn out to be sufficiently overbroad as to be easily invalidated. It is unlikely that the University of Utah, the National Institutes of Health, or Myriad Genetics (all owners of this patent) wanted this result."

So on this list of malfeasors we need to include the NIH and the University of Utah, correct? I think my explanation is more likely, that we know a lot more now than we did in 1994 and that information has consequences. (Of course, it would be interesting to look at claims for patents having been filed post-2000 to see how many of them have oligo claims).

And you noticed that your hypothetical claims are not in the patent, so you should give them credit for limited overreaching (which has come back to bite them after all). On the other hand, your comment on claim 5 versus claim 1 is right on.

If we have time maybe we will look at how many universities filed such oligo claims in the 1990s.

Thanks for the comment.

[Keep It Real]

KN: "So on this list of malfeasors we need to include the NIH and the University of Utah, correct?"

If they were involved in the prosecution: yes. And shame on the PTO for allowing such claims to issue. Maybe it was all the Ecstacy the kids were taking back then. On the other hand, they still issue a lot of nutty stuff but much less in the 1600 art unit, as far as I can tell.

"And you noticed that your hypothetical claims are not in the patent, so you should give them credit for limited overreaching"

Very funny.

[James Demers]

Are you suggesting that a claim can be overbroad today, in view of what's known today, but not be overbroad when filed in 1994? That has some curious consequences.

In any event, the fact that "any particular 15-mer" was expected to occur about 3 times in a random 3 Gbase genome doesn't quite get to the point: you have to multiply those 3 random occurrences by the 1.6 million 15-mers encompassed by claim 1. With roughly 5 million such 15mers scattered through the genome, it was a mathematical certainty (even in 1994) that a large fraction of known gene sequences would contain at least one of them. What was missing in 1994 wasn't knowledge -- what was missing was the computing power to do the search (Examiner Martinell's excellent point.) Claim 5 was clearly invalid, even in 1994.

Claim 1 is where the action is. How does one claim a gene, without leaving room for design-arounds, if this decision is upheld? Perhaps we should pay more attention to ensuring that the compositions and combinations actually used in diagnostic methods are well-claimed.

[Kevin E. Noonan]

Dear James:

What I was trying to counter was the implication that the patentees intentionally overreached. I think that the claim was much more reasonable in 1994 when less was known about the structure of the human genome.

I know there are 1.6 million 15-mers, but the thought was that each 15-mer would be present no more than 3 times (admittedly, the math works better for a 16-mer). I think it was not a mathematical certainty that any particular 15-mer would be detected in the genome anywhere other than in the BRCA1 gene; as it turns out, it is a biological reality that such oligos are found hundreds of thousands of times in chromosomal DNA from a chromosome without any related BRCA sequences.

In any event, claim 5 was invalidated, and in view of the data in the Kepler paper it is likely that these types of oligo claim would be valid only if such a search was performed and the uniqueness of the sequence confirmed.

Thanks for the comment.

[James Demers]

"[i]t was not a mathematical certainty that any particular 15-mer would be detected in the genome anywhere other than in the BRCA1 gene..."

You are correct. But it was 1.6 million times more likely that *at least one* of the 15-mers of claim 5 would be detected in another gene.

Whether the applicants *intentionally* overreached depends on whether or not they appreciated the difference, and on that score I'm in agreement with your conclusion. This was overreaching beyond sec. 112, and deep into sec. 102 territory -- and who intentionally files a claim guaranteed to be anticipated? (Indeed, we might be arguing different points - you're making a 112 argument, which I'm countering with the probability of a 102 issue.) At any rate, given its other problems, I think we can agree that not a lot of thought went into claim 5.

The PTO doesn't (yet) have the computing power to deliver a yes-or-no 102 verdict on such claims. They can do the statistics, but can an examiner reject a claim under 102 based on a 99.99% probability of anticipation? How many nines before a claim can be denied under 102? How many nines justify a 112 rejection?

[6]

"Are you suggesting that a claim can be overbroad today, in view of what's known today, but not be overbroad when filed in 1994? That has some curious consequences."

The only thing curious about it is that examination in this area has apparently been a joke for over 15 years.

"and who intentionally files a claim guaranteed to be anticipated? "

Quite a few applicants in my art.

"How many nines before a claim can be denied under 102? "

I think you need a bar over the nines before you can deny under 102. As you well know, .9bar=1.

http://en.wikipedia.org/wiki/0.999...

[Keep It Real]

"They can do the statistics, but can an examiner reject a claim under 102 based on a 99.99% probability of anticipation?"

Don't forget about obviousness. If you've ever tried to argue that an oligonucleotide of say, 30 nucleotides, which overlaps with another oligonucleotide by 28 bases is "non-obvious", you'll appreciate that it's not easy.

And it shouldn't be easy.

[Kevin E. Noonan]

Dear Keep:

Agreed.