Monday, April 13, 2015

Do Biosimilar Manufacturers Have To Dance? A District Court Answers “No”

In 2010, Congress enacted the Biologics Price Competition and Innovation Act (BPCIA) as part of the Affordable Care Act. BPCIA is, in a broad sense, intended to be the analog of the Hatch-Waxman Act for biologic drugs. Hatch-Waxman provides a pathway for Food and Drug Administration (FDA) approval for small-molecule generic drugs. Vastly simplified, the Hatch-Waxman process comes down to this: if a follow-on (i.e., generic) manufacturer can make an identical copy of the branded drug molecule, it can obtain FDA approval to market the drug without the clinical trials that the drug’s originator had to go through to prove that the drug is safe and effective. This saves costs for the generic manufacturer and, once the generic goes on the market, lowers prices for consumers. Under Hatch-Waxman, a follow-on manufacturer’s act of filing a so-called Abbreviated New Drug Application is an act of patent infringement, and so the originator can try to keep generic drugs off the market using patent law. The Food, Drug, and Cosmetics Act also provides periods of market and data exclusivities even for originator drugs that are not covered by patents. But once those periods end and the generic manufacturer can prove chemical identity to the brand, the generic drug is good to go on the market as far as the FDA is concerned—and the only barrier left is the brand’s potential patent infringement claims.

While making identical copies of small-molecule drugs is relatively straightforward, making identical copies of so-called biologic drugs is anything but. As I and many others (for example, Bryan Liang) have explained, there are significant differences between small-molecule drugs and biologics. The former have well defined structures and can generally be synthesized in reproducible manner, but the latter are large and complex molecules—and often mixtures of molecules—whose precise structure might not be known even to the originator. In addition, the structure and activity of a biologic drug can be highly dependent on the process of its preparation. Because of these factors, the process for making follow-on versions of biologics is significantly more challenging and costly than for generic small molecule drugs. A recent article by Nicholson Price and Arti Rai (behind paywall) explains that “at least for complex biologics, the predominant view of the scientific community, shared by the FDA, is that fundamental knowledge has not advanced to the point where product characteristics can fully be identified and characterized through analysis of the final product.”

It may come as no surprise, then, that it took five years from the passage of the BPCIA for the FDA to approve its first follow-on biologic drug. The original drug, called filgastrim and marketed by Amgen under the brand name Neupogen, is a glycoprotein indicated for treatment of neutropenia. This disorder is characterized by a deficiency in the amount of a certain type of white blood cell critical for the body’s immune response against infections. Under the BPCIA, there are two forms of FDA approval that are available for the follow-on researcher. It can show that the follow-on product is “biosimilar” to the brand (or “reference”) product, and, with additional data, it can show that its product is “interchangeable.” The “interchangeable” standard would appear to be very difficult for biologics to meet because it is a challenge to produce an equivalent of something whose structure might not even be known, and in fact the FDA has not even issued a guidance on what it takes for a biologic to be interchangeable. The “biosimilar” standard is more permissive, but it comes with disadvantages for the follow-on manufacturer—for example, biosimilars are not subject to state mandatory “generic substitution” laws. Recent state legislation related to biosimilars is an evolving, fascinating area of law—with some states affirmatively prohibiting biosimilar substitution. And in Europe, where a biosimilars regime has existed for a number of years, biosimilar substitution has generally not been achieved. All this suggests that substitution may simply not be possible for biologic drugs—though some cost savings might still be obtainable if biosimilars are allowed to enter the market.

The first follow-on biologic licensed by the FDA, a filgastrim product that would be marketed by Sandoz under the name Zarxio, was approved as a biosimilar (rather than interchangeable) product. The originator drug (the one marketed as Neupogen) was approved back in 1991, which means that the four-year data exclusivity and twelve-year market exclusivity periods to which Amgen is entitled under the BPCIA have expired. Just as with generics, though, FDA approval and the end of the exclusivity period for the originator biologic does not mean that the follow-on drug can be marketed right away. There is also the question of patent protection for reference biologic products (which Yaniv Heled has critically described as “double-dipping”). The patent regime under the BPCIA differs from that under Hatch-Waxman in several important respects. One of the differences is the availability under the BPCIA of a procedure colloquially known as “patent dance,” codified at 42 U.S.C. § 262(l). One of the reasons for the existence of this procedure must surely be the absence of an “Orange Book” equivalent for biosimilars. The Orange Book includes patents that purportedly cover branded small-molecule drugs, which gives some notice to generic manufacturers regarding what patent rights they would have to contend with. In contrast, the BPCIA describes a complex process in which the would-be biosimilar manufacturer provides the originator with a copy of its FDA application (including a description of its manufacturing process), the originator responds with a list of patents that it thinks the biosimilar would infringe. After this initial exchange, the parties negotiate what patents would be litigated and whittle down the number of patents, if any, that would be at issue in the originator’s infringement suit.

This finally brings us to our case, Amgen v. Sandoz (see docket entry #105), which was recently decided by the Northern District of California. The question the case answers is a simple one to frame, and one that occurred to anyone from a middle-school student to a reluctant club-goer: “Do I have to dance?” This question arose because in this case Sandoz opted not to participate in the patent dance. In other words, Sandoz yelled at Amgen, “Come at me, bro!”—with all the patents you got. Presumably, Sandoz felt confident that it could ward off Amgen’s patent infringement claims and so looked to get the patent litigation started as soon as possible—and, one imagines, Sandoz did not want to share its manufacturing information with Amgen. Amgen sued, arguing that the BPCIA does not allow follow-on manufactures to opt out of the patent dance and sought for the district court to force Sandoz to participate in the BPCIA’s patent dance. The district court held, however, that the dance-off was optional.
The critical BPCIA language appears under subparagraph § 262(l)(1)(B), titled “In General.” This subparagraph provides:

“When a subsection (k) applicant [i.e., biosimilar applicant] submits an application under subsection (k), such applicant shall provide to the persons described in clause (ii) [i.e., the originator], subject to the terms of this paragraph, confidential access to the information required to be produced pursuant to paragraph (2) [i.e., the biosimilar applicant’s FDA application] and any other information that the subsection (k) applicant determines, in its sole discretion, to be appropriate . . . .” (emphasis added)

Paragraph § 262(l)(2), titled “Subsection (k) application information,” says the following:

“Not later than 20 days after the Secretary notifies the subsection (k) applicant that the application has been accepted for review, the subsection (k) applicant—(A) shall provide to the reference product sponsor a copy of the application submitted to the Secretary under subsection (k), and such other information that describes the process or processes used to manufacture the biological product that is the subject of such application . . .” (emphasis added)

And finally, under subparagraph § 262(l)(9)(C), titled “Subsection (k) application not provided,” the statute says:

“If a subsection (k) applicant fails to provide the application and information required under paragraph (2)(A), the reference product sponsor, but not the subsection (k) applicant, may bring an action under section 2201 of title 28 for a declaration of infringement, validity, or enforceability of any patent that claims the biological product or a use of the biological product.” (emphasis added)

Interestingly enough, Amgen’s claims arising from Sandoz’s purportedly illegal refusal to dance under § 262(l)(1) and (2) were styled as unlawful competition under the California Business and Professions Code, as well as common-law conversion based on “Sandoz’s use of Amgen’s FDA license for Neupogen in its biosimilarity” application. The Federal Circuit recently addressed the interaction of the California Business and Professions Code and federal food and drug law in a case called Allergan v. Athena Cosmetics, but the preemption issues that case dealt with need not concern us for now. Instead, assuming that the California law route for enforcing the federal patent dance provisions is not preempted, the district court considered whether federal law forced Sandoz to dance—and concluded that it did not.

The court acknowledged that § 262(l) repeatedly uses the mandatory world “shall”—particularly in paragraphs (l)(1) and (l)(2) reproduced in part above, and noted that, in some other parts of § 262, Congress used the permissive word “may.” But the court explained “that an action ‘shall’ be taken does not imply that it is mandatory in all contexts.” It referred to the language describing the consequences of failure, such as that in subparagraph (l)(9)(C) reproduced above, and determined that “these procedures are ‘required’ where the parties elect to take advantage of their benefits, and may be taken away when parties ‘fail’” to do so. The court reasoned that the biosimilar applicant can choose to take a more streamlined path to patent infringement litigation, and enjoy a temporary safe harbor from an infringement suit, by following the patent dance provision—or else be exposed to the possibility of being sued on a large number of patents right away.

The court summarized its reasoning as follows: “Congress intended merely to encourage use of the statute’s dispute resolution in favor of litigation, where practicable, with the carrot of a safe harbor for applicants who would otherwise remain vulnerable to suit.” It also found important that “[t]he statute contains no stick to force compliance in all instances.” The court concluded that “[b]ecause Sandoz’s actions did not violate the BPCIA, it has committed no unlawful or wrongful predicate to sustain Amgen’s claims” under the California Business and Professions Code. With regard to conversion, the court explained “the BPCIA expressly contemplates that a subsection (k) applicant will rely on the reference product’s license and other publicly available safety and efficacy information about the reference product.” It also found “unworkable” Amgen’s argument that “Congress intended for sponsors to resort to state laws to enforce mandatory provisions in a federal statute . . . in addition to exacting the consequences written expressly into the legislation itself”—the consequences being unfettered patent litigation. Indeed, the BPCIA also provides, and the court found important, that the follow-on manufacturer’s failure to share its biosimilar application and confidential manufacturing in formation is defined as an act of infringement under 35 U.S.C. § 271(e)(2)(C)(ii).

This is a fascinating case—not only given its practical importance, but also given the questions it raises regarding statutory interpretation and the allocation of power between the federal government and the states when it comes to enforcement of food and drug laws. On the one hand, the word “shall” quite clearly suggests mandatory compliance with the patent dance section. But on the other hand, it is curious that the BPCIA itself provides no explicit remedy for the failure to dance, other than the default channel of regular patent litigation unconstrained by the dance provision. Of course, even where no remedy is specifically provided, the courts have found an implied private right of action for violation of numerous federal laws—securities litigation comes to mind, and there are many other examples. But Amgen appears not to have raised the implied private right argument, and relied exclusively on state law claims. While the basis for the district court’s holding was not preemption, the court was surely influenced by what it perceived to be the oddity of enforcing federal law though the claims based on the California Business and Professions Code and state common law. Amgen did ask the FDA directly, via a Citizen Petition, to require Sandoz to dance. But the FDA recently denied that request, reasoning that the BPCIA “generally does not describe any FDA involvement in monitoring or enforcing the information exchange by creating a certification process or otherwise.”

An interesting question remains whether the outcome would be different for the originator if the implied private right of action claim is raised in a future case—or does the district court’s interpretation of the BPCIA, if affirmed on appeal (which is already under way), foreclose this claim too? The parties have requested expedited briefing at the Federal Circuit, so perhaps we will have some sort of an answer sooner rather than later. And perhaps there might even be a biosimilar on the market sometime soon.

Note: Cross-posted on the Center for Law and the Biosciences Blog