Genentech, Inc. v. Sandoz, Inc.

Case Name: Genentech, Inc. v. Sandoz, Inc., Civ. No. 19-78-RGA (D. Del. Mar. 22, 2022) (Andrews, J.) 

Drug Product and Patent(s)-in-Suit: Esbriet® (pirfenidone); U.S. Patent Nos. 7,566,729 (“the ’729 patent”), 7,635,707 (“the ’707 patent”), 8,592,462 (“the ’462 patent”), 8,609,701 (“the ’701 patent”), 7,816,383 (“the ’383 patent”), and 8,013,002 (“the ’002 patent”). The ’729, ’707, ’462, and ’701 patents are referred to the Liver Function Test (LFT) patents, and the ’383 and ’002 patents are referred to as the Drug-Drug Interaction (DDI) patents.

Nature of the Case and Issue(s) Presented: Pirfenidone was a drug used to treat idiopathic pulmonary fibrosis (IPF), a chronic disease characterized by scarring (fibrosis) of the network of tissue that supports the air sacs of the lungs. It was primarily metabolized through the CYP1A2 enzyme pathway with contributions from other CYP enzymes. Pirfenidone had been first studied in 1973 and, in March of 2004, the FDA first approved pirfenidone in the US. The LFT patents claimed methods for administering pirfenidone to a patient that exhibited abnormal biomarkers of liver function in response to pirfenidone administration. The DDI patents claimed methods for avoiding adverse drug interactions with fluvoxamine and pirfenidone or other moderate to strong inhibitors of CYP enzymes. Sandoz sought FDA approval to market generic pirfenidone. In response, Genentech sued Sandoz for patent infringement. After a three-day bench trial, the court found that Sandoz did not induce infringement of either the LFT or DDI patents, that the LFT patents were invalid as obvious, and that the DDI patents were not invalid.

Why Sandoz Prevailed: Non-infringement of the LFT patents. Each LFT patent required the administration of pirfenidone to treat IPF, a patient exhibiting a Grade 2 elevation in one or more liver function test, and a dose modification. Only the dose-modification limitation factored into the infringement inquiry. The parties disputed whether Sandoz’ proposed label for its ANDA product showed its intent to induce others to infringe (Genentech’s position) or permitted certain infringing methods as well as non-infringing methods (Sandoz’ position). The court found that of the five bullet points in the label corresponding to dose modification, two—written in the active voice—were “directions” while three—using a passive voice—were merely options. Moreover, the court found that the proposed label recommended some dose-modification methods that were non-infringing. Because Sandoz’ label provided physicians with multiple dose-medication options, some infringing and some not, and “leaves it to the physician’s clinical judgment to determine how to modify the patient’s dosage,” Genentech had not shown that Sandoz’ label evinced a specific intent to induce infringement of the LFT patents.

Invalidity of the LFT patents. As of November 10, 2008, the priority date for the LFT patents, a POSA would have known that pirfenidone had efficacy for the treatment of IPF. In October 2008, the Japanese equivalent to the US-FDA-approved pirfenidone for the treatment of IPF published (i) a label for the drug containing detailed instructions for using pirfenidone to treat IPF and (ii) a report that summarized the efficacy and safety data supporting the Japanese regulatory authority’s approval of the drug. Additionally, the specific dose modifications claimed in the LFT patents would have been obvious over the prior art combined with known, standard medical practices. The prior art expressly disclosed continuing pirfenidone administration in patients exhibiting elevated liver enzymes and showed that it was well-known in the art that continuation of drug administration generally in patients with elevated liver enzymes was feasible, potentially following a temporary reduction or interruption in dosage. Next, the court found irrelevant Genentech’s evidence of secondary considerations. “Pirfenidone’s arduous road to approval” was not the inventive element of the claimed methods. The court similarly dispensed with Plaintiff’s arguments concerning skepticism, copying, and teaching away.

Non-infringement of the DDI patents. The DDI patents claimed treating an IPF patient with pirfenidone, treating the patient with fluvoxamine during or immediately prior to pirfenidone treatment, and a dose modification. Fluvoxamine treated depression. While the Sandoz label explicitly referenced fluvoxamine and potential drug-drug interactions with fluvoxamine, Sandoz argued that there was no evidence of likely direct infringement—a patient taking both pirfenidone and fluvoxamine—so there could not be induced infringement. The court agreed. “All three medical experts testified that in the seven years pirfenidone has been available for treatment of IPF in the US, none of them has had a single patient receive fluvoxamine before taking pirfenidone or receive both concurrently.” In any event, even if a patient were prescribed both pirfenidone and fluvoxamine, given the label, a physician would likely modify the dose by using a non-infringing treatment adjustment.

Validity of the DDI patents. The court found that Sandoz had not proven, by clear and convincing evidence, that a POSA would have expected to find a significant drug-drug interaction between pirfenidone and fluvoxamine. “That pirfenidone is primarily metabolized by CYP1A2 was a novel discovery by Intermune and was not known prior to the DDI patents’ priority date.” In view of: (i) the express teachings-away of the Japanese pirfenidone label and report; (ii) the fact that fluvoxamine was known to inhibit only between three and four of the five enzymes responsible for the metabolism of pirfenidone; (iii) the lack of an understanding at the time that any one of the five identified CYP enzymes was primarily responsible for the metabolism of pirfenidone; and (iv) the general understanding at the time that drug-drug interactions involving drugs metabolized via multiple enzymes were the exception rather than the expectation, the court found the DDI patents not invalid.