This week I am focusing on patent law, which is one of the more arcane and technical areas of pharmaceutical policy. A recent major decision by the Court of Appeals for the Federal Circuit (CAFC) in Bristol-Myers Squibb v. Teva Pharmaceuticals (BMS v. Teva) is rather remarkable in the degree to which it departs from prior decisions on the patentability of small molecules as the active ingredients in drugs. Chris Holman, a leading scholar in the intersection of intellectual property and the biotechnology and pharmaceutical industries, wrote a great article a few years ago arguing that a significant degree of unpredictability in patent law would substantially depress pharmaceutical innovation. Holman argues persuasively that the uncertainty as to whether or not a patent claim to a drug's active ingredient would be enforceable is, in essence, an additional cost burden on pharmaceutical research and development, and that this increasing cost burden is responsible for a decrease in the output of pharmaceutical research. Holman pointed to a long period of stagnation in the number of new drugs approved as evidence of the decreased output of pharmaceutical research. He uses two examples of Eli Lilly patents that had been invalidated as evidence of the unpredictability of patent law. Holman's analysis of the unpredictability problem centered on three different ways in which uncertainty is created:
the proliferation of loosely defined standards rather than bright line rules; unpredictability associated with long-delayed clarification of critical and identifiable ambiguities in patent law; and perhaps worst of all, unpredictability that occurs when courts adopt a new interpretation of legal doctrine and apply it retroactively, to the detriment of the investment-backed expectations of patent owners.
I think that Holman's article is an important contribution to the discussion of patents and pharmaceutical policy. In light of the June 12, 2014, decision of the CAFC in BMS v. Teva, I would like to add some additional thoughts on this important issue. Although I do not entirely agree with Holman's assertion that the uncertainty in patent protection has already been a very significant factor in the diminishing relative productivity of the pharmaceutical industry, Holman was absolutely correct in his assertion that the return on investment which is made possible by patent protection is of fundamental importance to investment in pharmaceutical research. In the long run, if uncertainty in patent law increases to the point where the CAFC regularly finds a significant percentage of active ingredient drug patents to be invalid, then it would be devastating to the pharmaceutical industry and the development of important new drugs. It does not yet appear to be the case. However, it is an issue that requires and deserves a significant amount of further research; and, if the BMS v. Teva decision is a harbinger of things to come, it might soon become a major issue.
So, what is "good" in patents from a pharmaceutical policy perspective? As Chris Holman's article states, predictability is good. However predictability in and of itself is not good if the result of predictability is a continuing stream of very similar drugs competing on the basis of marginal differences on endpoints other than real safety and efficacy (see my posts of April 12th, 18th, and 26th). "Good" from a pharmaceutical policy perspective would be interpretations of patent law that lead to both predictability and more innovation. In a 2008 article (Reach-Through Claims for Drug Target Patents: Rx for Pharmaceutical Policy, Nat. Biotech. 26:55-56), I argued that it would be good to provide a broad scope of patent protection for inventors who establish the therapeutic utility of a new biological target (such as a receptor or an enzyme which had not previously been a key feature in other drugs' mechanisms of action). My argument was that the biggest value-added step in pharmaceutical research is the validation of new targets for drugs and that providing greater protection when such research succeeded would lead to greater innovation and fewer me-too drugs.
What is "bad" in pharmaceutical policy? As Holman argued, unpredictability is bad; and, when it comes in the form of new interpretations of patent law applied retroactively, it can get "ugly." The CAFC decision in Ariad Pharmaceuticals v. Eli Lilly might well be an example of a new interpretation of patent law applied retroactively to the detriment of the inventors and their investors. That is not to say that the patent should have been upheld, but rather that from a predictability perspective it went badly wrong in finding the patent invalid because of a new interpretation of the written description requirement. The Ariad court held that a written description in a patent is insufficient to support a reach-through claim to a drug target even if the description enables others to practice the invention. The CAFC struck down Ariad's claim despite their holding in a prior case, U. of Rochester v. Searle, that a very similar reach-through claim was invalid only because it failed to enable others to practice the invention. In other words, the CAFC raised the bar, which is a form of unpredictability that is in Holman's terms "worst of all," if not downright "ugly."
Why was the outcome in BMS v. Teva particularly difficult to predict? The issue in the case centered on the obviousness of Entecavir, the active ingredient in BMS's Hepatitis B drug Baraclude. To translate the key patent term "obviousness" for non-patent lawyers, an invention must be "non-obvious" to a person who has ordinary skill and knowledge of the field of the invention, such as pharmaceutical chemistry, and who has studied the scientific and technical literature that relates to the invention. For a chemical compound, such as the active ingredient in BMS's drug, this means that the invention would not have been obvious both as to the precise structure of the molecule and its success as a drug with relatively low toxicity and its low susceptibility to resistance by the Hepatitis B virus.
In the case of BMS's drug the CAFC upheld a decision that the active ingredient was obvious despite the fact that it combines features from two previously known molecular structures (2'-CDG and Madhavan 30) that share most of their basic elements (a "carbocyclic" ring attached to a "guanine" ring).
For the chemists who read this, below are the four chemical structures discussed in the opinion. The non-chemists can just skip ahead to the conclusion. (The PDF displays the structures better than this html.)
Deoxyguanosine, shown top right, is a naturally occurring nucleoside that has no antiviral properties. 2'-CDG, lower left, has antiviral properties and is different than deoxyguanosine because it replaces the "O" oxygen atom in the top of the left-hand pyramid with a carbon (dotted circles in the diagram). Madhavan compound 30, shown bottom right, differs from both deoxyguanosine AND 2'-CDG because it has a double-bonded carbon at the top (5') position of the carbocyclic ring at the left of the chemical structure AND has no NH2 group at the bottom of the guanine ring on the right-hand side of the structure (a difference that escaped a dotted circle in the CAFC opinion for reasons that are not "obvious" to me). Madhavan compound 30 has antiviral activity. Thus BMS's drug differs from 2'-CDG in one position and from Madhavan compound 30 in a second position. While it may have been obvious to try combining features of both structures or, in CAFC terms, using 2'-CDG as a lead structure and adding a key feature of Madhavan compound 30, few experts would have predicted with confidence that the resulting drug would work as well as it did and, given its unexpected ability to avoid viral resistance, the patent on the drug would be held invalid on obviousness grounds.
Conclusion: As I have previously stated, I am generally opposed to a patent policy that results in the development of competing drugs with active ingredients and with no clear difference in their safety or effectiveness. BMS invested tens of millions of dollars based on the patentability of their drug's active ingredient. The decision by the CAFC clearly went counter to BMS's expectations and results in a significant problem of unpredictability in pharmaceutical patents. In the absence of a relatively clear standard for the patentability of small molecules, the CAFC's decision in BMS v. Teva only creates greater uncertainty without any real gain in pharmaceutical policy terms. Bright-line rules that clearly demarcate the standards for patentability are not always "good", but, unpredictability in patent law, as Holman persuasively argued, is always "bad." For Bristol-Myers Squibb unpredictability became downright "ugly."