WV court denies Regeneron's motion for PI in Eylea litigation

In a recently unsealed opinion, a N.D.W.V. court denied Regeneron's motion for a preliminary injunction against Amgen seeking to enjoin Amgen's ABP 938, a biosimilar version of Eylea. Regeneron had previously been successful in obtaining preliminary injunctions in the same court against a number of other defendants. But Amgen's formulation differed from those defendants, resulting in a different outcome.

In short, the claims of the Regeneron's 865 patent separately listed "a vascular endothelial growth factor (VEGF) antagonist" and a "buffer." Amgen's formulation utilized the antagonist as a buffer, and did not contain any separate buffer, i.e., it is self-buffering. The district court, relying primarily on the Federal Circuit decision in Becton, Dickinson & Co. v. Tyco Healthcare Grp., LP, 616 F.3d 1249 (Fed. Cir. 2010) and cases citing it, construed the claims to require the VEGF antagonist and buffer to be separate components in the formulation. Because it was undisputed that Amgen's formulation lacked a separate buffer, the court found Regeneron was unlikely to be able to prove infringement. 

Becton and its progeny hold that the separate listing of elements creates a presumption that the components are distinct. Thus, the focus of the arguments was on whether the specification provided any suggestion that one component could satisfy multiple limitations. The court found none.

It noted, for example, that the claims not only separately listed the antagonist and the buffer, but dependent claims contained separate units of measurement for each component, with dependent claim 2 reciting an antagonist concentration of 40 mg/ml and dependent claim 7 reciting a concentration range of the buffer of 5-25nM. The court then addressed Regeneron's arguments that a POSA could easily convert between the two units of measurement, but rejected that argument because 40 mg/ml of the antagonist would be "substantially outside of the claimed concentration range of 5-25 nM recited for the 'buffer' in claim 7." The court concluded that such claim language "is rendered nonsensical unless the 'VEGF antagonist' and 'buffer' are interpreted as separate and distinct components." While I agree with the district court that the listing of separate measurements for the two components generally supports an understanding that they are separate components, the district court's logic seems to be conflating infringement of dependent claim 7 with the scope of claim 1. More specifically, the district court's logic seemed to be that because it would be impossible for this formulation to simultaneously meet both dependent claim 2 and dependent claim 7, the formulation could not be covered by independent claim 1. That logic escapes me. That in a particular example, a formulation might not meet the limitation of a dependent claim (let alone two dependent claims) does not render the claim  language in the independent claim nonsensical. Indeed, it is hardly surprising that some formulations would fall within the broader scope of the independent claim, but not within the narrower scope of dependent claims.

Regardless, there was other claim language supporting the district court's interpretation. It further noted that because claim 1 recited "wherein said VEGF antagonist fusion protein is glycosylated and comprises amino acids 27-457 of SEQ ID NO:4," the claims effectively required the VEGF antagonist to be the protein aflibercept, which would mean under Regeneron's theory the antagonist is always a buffer, rendering the buffer limitation superfluous. While Regeneron argued that the aflibercept may only act a buffer under certain conditions, the court noted that Regeneron had previously argued against other defendants (in the context of 112) that the buffer limitation was structural, not functional. 

The court then turned to the specification and found no support for any embodiment in which the antagonist and buffer were the same component. To the contrary, it found in all twenty two described embodiments, the buffer was a distinct component. The court found Regeneron's counterarguments unconvincing. Regeneron first argued that the 865 patent incorporated by reference a second patent, which disclosed that other components (not the antagonist or buffer) could have different functions than those specified in the 865 patent. Regeneron also argued the specification states that "all technical and scientific terms and phrases used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the invention belongs." Unsurprisingly, the district court was not persuaded that either passage suggested that the antagonist and buffer could be the same component. 

Though it noted that it didn't believe it was necessary to consider extrinsic evidence, the court did so "for completeness" and found it did not provide any evidence rebutting the Becton presumption. Interestingly, Regeneron attempted to rely on an Amgen patent application, filed slightly before the 865 patent (but not yet published) as prior art demonstrating a POSA's understanding that proteins could act as buffers. While questioning whether it was even proper to rely on "secret prior art" to show the ordinary meaning in the art, the court further found that Amgen's patent application further demonstrated that POSAs would not have understood the claims to cover self-buffering proteins. In sum, there was no evidence that a POSA would commonly assume that the protein antagonist could also serve as a buffer in Regeneron's claims. 

Finally, the court noted again that Regeneron's arguments conflicted with arguments Regeneron made in previous injunction motions against other defendants, where when defending against 112 invalidity arguments it argued that a buffer is a well known class of excipient.

Based on its construction, the court held that Regeneron was unlikely to show literal infringement, and that an injunction based on DOE was rare. It further noted that Regeneron's arguments appeared to entirely vitiate the buffer limitation. Again while noting such an analysis was unnecessary, the court stated its agreement with Amgen that the insubstantial differences test for DOE was more appropriate, and that it was likely a self-buffering composition was substantially different, as Amgen's biosimilar, if approved, would be the first buffer-free fusion protein formulation approved by FDA. It also found Regeneron was unlikely to succeed under a function-way-result analysis as well.

Harkening back to my last post, this again illustrates the importance of 112 arguments for defendants to box in a plaintiff's arguments—even if the 112 defense is not successful. The district court repeatedly cited the plaintiff's own 112 arguments against it. And had Regeneron succeeded in its claim construction argument, it seems like it would have had a clear 112 problem in demonstrating it was in possession of self-buffering formulations.