Last week, OSI Pharmaceuticals learned that their cancer treatment patent would remain after an obviousness-based challenge on appeal to the Court of Appeals for the Federal Circuit. Finding that OSI’s cancer treatment claims could not have been obvious over prior art that did not provide a reasonable expectation of success for the claimed treatment, the court reversed a U.S. Patent and Trademark Office (USPTO) inter partes review decision tossing out OSI’s patent as unpatentable.

The patent at stake in OSI Pharmaceuticals v. Apotex covered treatment of non-small cell lung cancer (NSCLC) by administering erlotinib, an epidermal growth factor receptor (EGFR) inhibitor. While EGFR inhibitors have had long-recognized potential as antitumor agents, many EGFR inhibitors have not been effective for real treatments. Some EGFR inhibitors had poor pharmacokinetics, some had undesirable interactions with other drugs, some were toxic due to binding with healthy cells or other receptors. Cancer treatments, too, are unpredictable—drugs may work for some types of tumors, but not others. In particular, over 99.5% of NSCLC drugs that started Phase II clinical trials had ultimately failed.

Recognizing this landscape of difficulties in NSCLC treatment research, the court reviewed a decision from the USPTO Patent Trial and Appeal Board (PTAB) finding that OSI’s treatment was obvious over one main prior art reference (Schnur) over either of two second references: Gibbs, or an earlier form 10-K, which was a financial regulatory submission by OSI itself. Schnur teaches treatment of hyperproliferative diseases, such as cancers, with a category of compounds, listing numerous examples and noting a preference for erlotinib. It mentions several types of cancer for treatment, including lung cancer, but without specifically noting NSCLC.

The PTAB looked to the combination of Schnur with Gibbs or the OSI 10-K to try to show a reasonable expectation of success in using erlotinib to treat NSCLC. Gibbs mentions clinical trials for erlotinib and another EGFR inhibitor, and identifies NSCLC treatment, but on that point cites two references: one mentioning the other EGFR inhibitor for NSCLC, and a second mentioning erlotinib without mentioning NSCLC. In its 10-K, OSI had earlier disclosed that erlotinib targets several cancers including NSCLC, and that OSI had finished Phase I trials and would proceed with Phase II trials.

While the PTAB found OSI’s NSCLC treatment unpatentable as obvious over this information, the Federal Circuit disagreed. The court especially weighed the prior art references’ complete lack of any data showing efficacy of erlotinib for treating NSCLC in particular, further reiterating the unpredictability of cancer treatments and the high failure rate of drugs entering Phase II clinical trials for treating NSCLC. While the court emphasized that efficacy data would not be necessary for all obviousness cases, and that the reasonable expectation requirement for obviousness does not demand absolute predictability, it ultimately concluded that the Schnur, Gibbs, and 10-K prior art in such an unpredictable field did not sufficiently support the PTAB’s obviousness conclusion.

When applying for patent protection on treatment methods in unpredictable fields, applicants should be ready to point to this unpredictability in trying to overcome any obviousness rejections. This technique may be especially important where information about their own clinical trials may have become public before their application was filed.