Federal Circuit invalidates claims to fully human antibodies

On 23rd February 2011 the Federal Circuit held invalid for lack of written description a patent owned by Johnson & Johnson’s subsidiary Centocor Ortho Biotech in an appeal from a judgment that Abbott’s product Humira (adalimumab), a fully human monoclonal antibody specific to tumour necrosis factor used to treat rheumatoid arthritis and some other autoimmune diseases, infringed the patent (Centocor Ortho Biotech, Inc v Abbott Laboratories, 2011 WL 635291, No 2010-1144 (Fed Cir, 23rd Feb 2011)).

Centocor’s US Patent 7,070,775 was originally based on the discovery of murine and chimeric antibodies to TNF-α. The chimeric antibody was comprised of a murine variable region and human constant region, which made it less immunogenic than the murine antibody. However, because it still contained a murine variable region, it was more likely to elicit an immune response than a fully human antibody. Some eight years after the priority date, Centocor submitted claims to a fully human antibody. An illustrative claim, rewritten in independent form and shortened for clarity, was:

"An isolated recombinant anti-TNF-α antibody comprising a human constant region and human variable region, wherein said antibody (i) competitively inhibits binding of A2 to human TNF-α, and (ii) binds to a neutralizing epitope of human TNF-α in vivo with an affinity of at least 1x 108 liter/mole."

After a five-day trial in the Eastern District of Texas, the jury found that Abbott had wilfully infringed, rejected its argument that the asserted claims were invalid and awarded Centocor more than $1.67 billion in damages. The district court denied Abbott’s motion for judgment as a matter of law. Abbott appealed.

The Federal Circuit reversed, holding the claims invalid for lack of written description under 35 USC § 112. The specification of the ’775 patent detailed the characteristics of the chimeric antibody, including its ability to bind TNF-α with high affinity, neutralising activity and A2 specificity. It also identified the sequence of TNF-α and contained examples of making and using chimeric antibodies to TNF-α. The specification provided the amino acid sequence of a murine variable region of an antibody that had the desired characteristics of high affinity, neutralising activity and specificity, but did not illustrate making fully human antibodies with these characteristics. The Federal Circuit found the specification’s failure to illustrate a fully human antibody with the desired characteristics rendered the asserted claims a mere wish-list of properties that a fully human anti-TNF-α antibody should have (ie, high affinity, neutralising activity and A2 specificity).

The opinion recognised that the written description requirement does not, in all cases, demand working examples or an actual reduction to practice for a patent’s description to be found sufficient under 35 USC § 112. Responding to Centocor’s argument, the court acknowledged that Noelle v Lederman (355 F3d 1343 (Fed Cir 2004)) taught that disclosure of a well-characterised antigen would sometimes be sufficient to describe claims to antibodies to that antigen. However, it clarified that the adequacy of the description in such cases was premised on discovery of a new antigen to which antibodies were raised using routine methods. In the case at hand, by contrast, the antigen (TNF-α) was in the prior art and the claimed “invention” was a class of antibodies with desirable therapeutic properties that the applicants had never made.

The opinion also discussed the example in the US Patent and Trademark Office’s Written Description Guidelines, in which the full characterisation of an antigen was said to support claims to isolated antibodies capable of binding to that antigen, even without working examples of such antibodies. As with its discussion of Noelle, the court explained that this example assumed that the specification described a new antigen and that the production of antibodies to that antigen was routine. By contrast, the production of fully human antibodies was assuredly not routine as of the priority date of the ’775 patent, rendering the example in the guidelines of no help to Centocor.

The court’s distinction of Noelle and the Written Description Guidelines illustrates an often unstated interplay between written description and obviousness. Here, for example, the known role of TNF-α in certain autoimmune diseases and the known desirability of blocking TNF-α with a therapeutically acceptable monoclonal antibody would render obvious the idea of a fully human monoclonal antibody with high and specific affinity for TNF-α that binds in a neutralising manner. However, claiming that desired result in a patent is not the same as doing the work. Stated differently, claiming the solution to a recognised problem without having made a real contribution towards actually realising that solution – which, in this case, was the hard work of actually making the fully human antibodies – is not the type of activity that the patent laws are intended to promote and protect.

The court also noted that Centocor had not itself made fully human antibodies to TNF-α, instead waiting until after they had been made by Abbott before adding the asserted claims to a pending application. It did not rely on this fact for its holding that the claims were invalid for lack of written description, but the relative timing of the amendment and the creation of the accused infringing product – coupled with the applicants’ failure to themselves make the desired antibodies – did not make out a factually appealing case for Centocor.