Saturday, May 28, 2011

Therasense v. Becton Dickinson: Rethinking the Status of Patents as Technical Documents

Earlier this week the Federal Circuit issued its en banc decision in Therasense v. Becton Dickinson, which essentially made it more difficult to establish inequitable conduct. The decision has been reported and analyzed widely over the Internet, for example on the Patently-O and Patent Docs blogs. This post discusses one aspect of the case I find interesting, which is the problems that can occur when the PTO and courts unjustifiably assume that "legalese" statements appearing in patents constitute technical disclosures. To illustrate, let me begin by summarizing how Therasense (now Abbott Diabetes Care, and referred to in the decision and in this post simply as Abbott) got into trouble in the first place.

Abbott owns US patent number 4,545,382 (the ‘382 patent), which issued in 1985 and is directed towards a sensor comprising an electrode with enzyme bound to its external surface. In the specification, the ‘382 patent includes the following sentence (referred to herein as the "Sentence") which was the root cause of Abbott's alleged inequitable conduct: "Optionally, but preferably when being used on live blood, a protective membrane surrounds both the enzyme and the mediator layers, permeable to water and glucose molecules."

The Sentence refers to a protective membrane, employed to protect the electrode from being "fouled" by components of whole blood, such as red blood cells. As discussed later, it is Abbott's contention that at that time, as a practical matter, one of skill in the art would have considered it necessary to employ such a protective membrane when using the electrode in a whole blood. Nevertheless, as is typical practice in the drafting of patents, whoever drafted the patent (presumably a patent attorney or agent) identified the protective membrane as "optional but preferable" rather than mandatory. This is quite reasonable, in view of all the cases in which patentees have found their rights limited by unnecessarily limiting language appearing in the patent specification. The patent attorney prudently sought to avoid an unnecessarily restrictive interpretation of the scope of the claimed invention, in case it turned out to be possible to practice the invention without a protective membrane. When understood in this context, note that the "optional but preferable" language should not be interpreted as a technical disclosure that use of the electrode in live blood without a protective membrane would have been enabled at the time.

Abbott also prosecuted a counterpart to the ‘382 patent in Europe, containing the same "optionally, but preferably" language. The European Patent Office (EPO) cited as prior art against the application a German reference which required the use of a diffusion-limiting membrane. It is important to recognize that a diffusion-limiting membrane serves a very different purpose than a protective membrane. A diffusion-limiting membrane is used to slow the flow of glucose to the electrode, which was necessary for earlier electrodes that could not deal with a rapid influx of glucose, not to protect the electrode fouling.

In order to distinguish over the German reference, Abbott argued that the membrane mentioned in its patent application is a protective membrane, not a diffusion-limiting membrane. Abbott pointed out that the Sentence specifically notes that the protective membrane must be permeable to glucose molecules, and hence not diffusion-limiting as the membrane in the German reference. The arguments submitted by Abbott to the EPO characterized the protective membrane in their application as optional, parroting the language of the Sentence, but this was really tangential to the argument they were making, which was based on the distinction between protective and diffusion-limiting membranes. Unfortunately for Abbott, the briefs submitted to the EPO in 1994 and 1995 included representations that would come back to haunt them in the US, such as "[i]t is submitted that this disclosure [i.e., the Sentence] is unequivocally clear. The protective membrane is optional . . ..”

A few years later, in 1997, Abbott was prosecuting a different patent application in the US patent office that included claims directed to a sensor that did not require a protective membrane when used in whole blood. The ‘382 patent was cited by the PTO as prior art. To overcome the rejection, Abbott argued that one of skill in the art at the relevant time would have felt that the sensor disclosed in the ‘382 patent required the use of a membrane, notwithstanding the "optionally, but preferably" language appearing in the ‘382 patent specification.

In response to an examiner’s request for evidentiary support, Abbott submitted a declaration by Abbott's Director of Research and Development (Dr. Sanghera) stating that "one skilled in the art would not read [the Sentence] to teach that the use of a protective membrane with a whole blood sample is optionally or merely preferred." In submitting the affidavit, Abbott's patent attorney (Lawrence Pope) represented that "one skilled in the art would not . . . have read the "optionally, but preferably" language . . . as a technical teaching but rather mere patent phraseology." [Emphasis added]

In other words, Abbott argued that the "optionally, but preferably" language did not represent a technical disclosure of the state-of-the-art, but rather was an example of patent legalese inserted by the patent after to avoid unnecessarily limiting the scope of the patented invention. Note that there's nothing wrong with this, it is quite proper under some circumstances for an inventor to obtain patent claims that cover embodiments of her invention that employ after-arising technologies that were not part of the prior art at the time the patent application was filed.

In the declaration and accompanying argument apparently worked, and the application issued as US patent number 5,820,551, (the ‘551 patent), the patent that was ultimately held to be unenforceable for inequitable conduct by the district court based on Abbott's failure to disclose to the PTO the briefs filed to the EPO. The district court was evidently bothered by the appearance that in one forum, to suit its purposes, Abbott asserted that the term “optionally” is mere "patent phraseology" which in fact means "necessarily," while in a different forum it appeared to have argued that the same language is "unequivocally clear," i.e., "optionally" actually means "optionally."

In my view, the representations made to the US and European patent offices are not necessarily inconsistent. The problem is, the district court assumed that the "optional but preferable" language appearing in the patent is a technical disclosure, which it interpreted as a disclosure that one of skill in the art would have been able to use the electrode without a protective membrane. In fact, the language is better understood as patent legalese, or in the words of Abbott's attorney "patent phraseology," not intended as a technical disclosure but as language introduced into the specification to avoid unnecessarily limiting the scope of the patented invention.

The declaration and arguments made to the US PTO essentially made this point, and argued that one of skill in the art would have recognized this as patent legalese rather than a technical disclosure of the state-of-the-art. Before the EPO, Abbott does not appear to have argued that the state-of-the-art would have allowed the use of the electrode without a protective membrane, which is tangential to the argument it did make, which was based on the distinction between protective and diffusion-limiting membranes. Abbott's briefs to the EPO merely parroted the “optionally” language of the specification to make it clear that as a legal matter, Abbott did not intend to disclaim patent protection for electrodes used without a protective membrane, on the eventuality that this might become feasible someday.

In this case, Abbott was faced with a rejection based on patent legalese appearing in one of its own patents. But this sort of thing happens all the time, and often the patent cited as prior art is owned by a third party. In prosecuting patent applications, I have on numerous occasions been frustrated by the patent legalese appearing in patents asserted by examiners as prior art. These patents often include disclosure that is clearly wrong as a technical matter, and evidently was introduced into the patent specification by the patent drafter, but it can be difficult convincing a patent examiner that the disclosure appearing in a patent is inaccurate or not enabled. This is essentially the situation in which Abbott found itself, which required it to submit a declaration to the effect that a facially clear statement that use of a protective membrane is "optional” is inaccurate as a technical matter, and should be dismissed as mere patent phraseology.

Formally, the law treats a patent as a technical document. By statute, the patent specification is supposed to be directed towards a technical audience, i.e., the ubiquitous "person having ordinary skill in the art.” Federal Circuit case law attributes a higher degree of credibility to technical disclosure appearing in US patents than it does to other technical documents, such as peer reviewed journal articles. The Federal Circuit has held that the disclosure appearing in patents is presumptively enabled, apparently based on the rationale that the patent office has examined the patent for compliance with the enablement requirement. Other technical documents, such as articles appearing in technical journals, do not receive the same presumption of enablement.

I am quite skeptical with regard to this presumption favoring the credibility of patents over other technical documents. For one thing, patent examiners are charged with examining the enablement of patent claims, not the specification as a whole, which is a very different thing. The fact that a patent issues in no way implies that a patent examiner has reviewed all of the disclosure in a patent and found it to be technically accurate. As exemplified by the Abbott patent, much of the language appearing in patents is "patent phraseology," introduced by the patent practitioner drafting the patent to achieve certain legal objectives, such as broad scope of protection, or the ability to amend the claims a later date, rather than to provide the most accurate technical disclosure. Some patent attorneys might even disclose hypothetical embodiments simply to create prior art, essentially poisoning the well to prevent others from patenting around their invention. It is quite easy for a patent attorney to come up with hypothetical embodiments and describe them in the specification, and usually little downside to the practice (although it did come back to haunt Abbott in this case). It seems to me that the disclosure appearing in a peer-reviewed journal is much more likely to accurately describe the state-of-the-art than a patent, and that the Federal Circuit's presumption in favor of patents has gotten things backwards.

In fact, I would assert that patent law should recognize the reality that much of the disclosure appearing in patents is not technical in nature, but rather patent legalese introduced into the application by patent practitioners for purposes other than accurately describing the state of technology. In my experience, most scientists and other technical people are well aware of this, and recognize that much of what appears in patents is patent phraseology rather than an accurate and objective disclosure of technology. Patent law would do well if it would acknowledge that patents are often drafted for the primary purpose of furthering the legal and business objectives of the patent owner, not as true technical documents.

Wednesday, May 4, 2011

Billups-Rothenberg v. ARUP: Genetic Testing Patents Found to Be Anticipated and in Violation of Lilly Written Description Requirement

Robert Cook-Deegan and I filed an amicus brief in AMP v. US PTO (the Myriad gene patent case) essentially arguing that with respect to gene patents we should not throw out the baby with the bathwater. Some gene patents do appear overly broad and/or overly reaching, particularly in the context of genetic diagnostic testing, but these problematic patents are best addressed by means other than declaring all gene patents patent ineligible, which could have a substantial detrimental effect on biotechnology. For example, we pointed out that other doctrines of patentability, particularly the disclosure requirements of section 112 (written description and enablement) and the requirements of novelty and nonobviousness should be sufficient to address many if not all of the concerns associated with some gene patents. In Billups-Rothenberg v. ARUP, the Federal Circuit has just provided an illustration of what we were talking about, invalidating all of the gene patent claims that have been asserted against companies providing genetic diagnostic testing for mutations in the HFE gene, which are associated with hereditary hemochromatosis.

Gene patents have become quite controversial, based largely on a perception that they impose undue restrictions on the development and availability of genetic diagnostic testing. I have argued that the concerns have been overblown, in part because they assume that gene patent claims are extremely broad in scope (and thus difficult if not impossible to design around), when in fact if interpreted as broadly the claims would be highly susceptible to invalidation under a variety of patentability doctrines. Unfortunately, human gene patents have rarely been litigated in the context of genetic testing, there has been little case law to point to directly on point, and critics of gene patents have assumed the worst.

In fact, in a comprehensive survey of human gene patent litigation in the United States that I conducted in 2007, I found only a handful of lawsuits that had been filed asserting infringement of a human gene patent by a provider of genetic diagnostic testing services, and no substantive judicial decisions arising out of any of these lawsuits, owing to the fact that they all settled shortly after the lawsuit was filed. The Billups-Rothenberg lawsuit was filed shortly after I conducted my survey, and to my knowledge represents the only case in which the courts have actually reached a decision concerning an allegation that a human gene patent was infringed by genetic testing (in the Myriad case, Myriad has not alleged that any of the plaintiffs are infringing its patents)- it is certainly the first time the Federal Circuit has addressed the issue. Consistent with the point we were attempting to make in our amicus brief, the Federal Circuit invalidated the asserted human gene patents for violation of the written description requirement and for lack of novelty. This is a precedential opinion, and I think quite informative as to how the courts will address human gene patents that appear to claim more than was justified by the nature of the inventor's discovery.

Written Description
The asserted claims of US patent 5,674,681 were all held to be invalid for violation of the written description requirement. Claim 2 is representative:
2. A method to identify an individual having or predisposed to having hemochromatosis, comprising the steps of:
providing from the individual a sample containing a gene encoding a nonclassical MHC class I heavy chain and

detecting a mutation in said gene, which mutation results in the reduced ability of said heavy chain to associate with said β2 microglobulin, wherein the presence of said mutation identifies said individual as having or predisposed to having hemochromatosis.

As a sidenote, note that this claim explicitly requires providing a patient sample, and thus would only be infringed by an entity that obtained a physical DNA sample from a patient and analyzed it to detect a mutation in the gene. This sort of limitation appears in most gene patent claims, and is increasingly significant as we move toward whole genome sequencing, since it implies that these sorts of claims will not be infringed in situations where one entity sequences a patient's genome and another entity (for example a doctor) analyzes the genetic data disease-related mutations. On the other hand, the Federal Circuit's recent decision in Prometheus makes clear that a patent claim that could be infringed by merely analyzing genetic data, without physically manipulating DNA molecules, would be invalid based on patent ineligibility. This is part of the reason I believe that gene patents will become less of an issue as personal whole genome sequencing becomes established, and the analysis of DNA molecules becomes decoupled from the analysis of genetic information.

Like most written description decisions coming out of the Federal Circuit, the court is not entirely clear as to exactly why the claims failed the written description requirement, but I see two distinct bases before the court's decision. First, the court found that the specification does not demonstrate possession of the claimed subject matter, and that the inventors had "attempted to preempt the future before has arrived." In other words, the court invoked Lilly written description as a doctrinal "wildcard" (using the language of Judge Rader) to invalidate an "unworthy" patent claim, along the lines seen in cases such as Ariad and Rochester.

The court also based its decision on the failure of the patent to disclose the DNA sequence of the gene recited in the claims, citing Regents of UC v. Lilly and Fiers v. Revel. The patent discloses the general location of the gene in the human genome, but not its exact location, nor its DNA sequence, nor the sequence of any mutations.

The court made the important observation that "this case is like Regents and Fiers, in which the DNA sequences at issue were unknown in the art.” Prior to the Federal Circuit's 2008 decision in Carnegie Mellon v. Hoffman-La Roche, it was my opinion that the only way to reconcile Federal Circuit decisions which invalidated patent claims under Lilly written description for failure to provide DNA sequence information from decisions that upheld the validity of claims failing to provide DNA sequence information was the distinction between previously known and unknown DNA sequences. If the heart of the claimed invention involved identification of a novel DNA sequence (Fiers, Regents of UC, In re Wallach, In re Kubin (a rare precedential BPAI decision), a strict interpretation of Lilly written description has been applied requiring adequate disclosure of DNA sequence. On the other hand, in cases where the heart of the invention involves a novel recombination or modification of a previously known DNA sequence, the court has not required to disclosure of DNA sequence (Amgen v. HMR, Capon v. Eshhar, Invitrogen v. Clontech, and Falko_Gunter Falkner v. Inglis).

In Carnegie Mellon v. Hoffman-La Roche, Roche made the entirely legitimate argument that Lilly written description only requires disclosure of DNA sequence in cases where the invention involves the discovery of a new gene sequence. Since the heart of the claimed invention in that case involved a novel recombination of genetic elements, not the discovery of a new genetic sequence, Roche argued (correctly in my view) that a disclosure of DNA sequence was not necessary to satisfy Lilly written description. But the Federal Circuit rejected Roche's argument, and specifically held that the requirement for compliance with Lilly written description is the same regardless of whether or not an invention is based upon discovery of the new genetic sequence. I thought this was a poor decision, because without that distinction I don't know how you make sense of the disparate outcomes in the previous Federal Circuit written description decisions.

In Centocor, decided earlier this year, the Federal Circuit explicitly held that the requirement for compliance with Lilly written description does depend upon whether or not invention involves a newly discovered molecule. I would say this was the only way it could reconcile its decision to invalidate Centocor's antibody claim for failure to provide adequate disclosure of structure, while running into direct conflict with Noelle v. Lederman, which held that a broad claim directed to antibodies can comply with Lilly written description in the absence of any structural description. Of course, this seems irreconcilable with the court’s statement in Carnegie Mellon that it is irrelevant whether or not the invention involves discovery of a new molecule.

The decision in Billups-Rothenberg adopts the Centocor approach, by emphasizing the distinction between claims directed towards DNA sequences known in the art versus other claims directed towards DNA. I think this distinction is critical in any attempt to reconcile Federal Circuit decisions involving Lilly written description, with the caveat that the Carnegie Mellon University decision is an outlier.

The Billups-Rothenberg decision is I think also illustrates how courts can use the written description requirement to invalidate gene patent claims that seem to be overreaching. For example, even if the patent had disclosed the gene sequence, I can imagine a court invalidating a claim broadly directed towards methods of identifying mutations in the gene for failure to satisfy Lilly written description, because the specification fails to evidence "possession" of all the mutations that might be linked to disease, and because the specification fails to identify the sequence of specific mutations linked to disease.

The other patent asserted in the case, US patent number 6,355,425, was found to be invalid based on anticipation by one of the defendant’s patents, 6,025,130. The ‘425 patent claim the method for diagnosing hereditary hemochromatosis by determining the presence of a mutation in exon two of the HFE gene. The patent apparently covers diagnostic testing for the S65C mutation, which has been correlated with the hemochromatosis. The defendants test for this mutation.

The prior art ‘130 patent discloses the S65C mutation, but also states that presence of the mutation in patients "shows no increase in risk of acquiring [hereditary hemochromatosis] and thus may only be a polymorphic variant within the population." In other words, while the prior art patent apparently discloses testing for the S65C mutation, it also explicitly acknowledges that the inventors had not identified any basis to infer that identification of the mutation would be useful for diagnosing risk of hemochromatosis, and thus failed to disclose any practical utility for testing for the mutation.

Billups-Rothenberg argued that prior art disclosing only “a clinically insignificant polymorphism unrelated to disease” did not constitute an anticipatory reference to a claim directed towards a method of testing for the mutation in order to diagnose for disease risk. However, the court rejected this argument, and basically held that the mere disclosure of the mutation and a method of testing for it was sufficient, regardless of whether the reference disclosed a practical utility in testing for the mutation. The court also stressed that disclosure appearing in an issued patent is presumed to be enabled.

This aspect of the decision reminds me of In re Gleave, discussed in an earlier post. The approach adopted by the Federal Circuit in these decisions could have significant implications for the patentability of newly discovered genetic mutations. For example, suppose a researcher identifies a mutation in a known gene of clinical significance, such as correlation with disease. I can imagine a court extending the logic of Billups-Rothenberg and Gleave to conclude that a prior art patent disclosing the gene sequence, and perhaps stating that one could test the gene sequence for mutations, is anticipatory prior art, even though the patent does not disclose the mutations later discovered to be correlated with disease.