“A One-Carbon Poison Pill”: Enanta v. Pfizer and the “Typo” that Lost a Priority Date

“The one carbon left out of Enanta’s provisional application had unwittingly gifted Pfizer free passage to an ocean of opportunity. . . .”

I. Introduction

In Enanta Pharms., Inc. v. Pfizer, Inc., No. 25-1427 (Fed. Cir. June 23, 2026) (“Enanta, slip op.”), the Federal Circuit affirmed the district court’s summary judgment that Enanta’s U.S. Patent No. 11,358,953 (“the ’953 patent”) is invalid. In its summary judgment ruling, the district court found that the ’953 patent was anticipated by the active ingredient “nirmatrelvir” in the accused product itself, Pfizer’s blockbuster COVID drug Paxlovid. See Enanta Pharms., Inc. v. Pfizer, Inc., No. 22-cv-10967-DJC, 2024 WL 5203036 (D. Mass. Dec. 23, 2024) (“Enanta, D. Ct. Decision”).

That holding presents a paradox. The asserted patent was invalidated by the accused product; the patent law version of a self-inflicted wound.

A patent is invalidated by anticipation when a single prior-art reference discloses every element of the claimed invention before the patent’s priority date. Here, Pfizer indisputably first synthesized nirmatrelvir after Enanta filed the provisional application leading to issuance of the ’953 patent.

How can a molecule anticipate a patent whose priority application predates the molecule’s very existence?

The answer is that the ’953 patent is not entitled to its claimed priority date. The ’953 patent claimed priority to U.S. Provisional Patent Application No. 63/054,048 (“the ’048 provisional”), which was indeed filed before Pfizer first synthesized and published nirmatrelvir. But a later application in the same filing chain that issued as the ’953 patent was itself filed after. Both courts ruled, for different reasons, that the ’048 provisional did not convey priority to the ’953 patent.

Enanta’s problem boiled down to this. The ’048 provisional disclosed a compound encompassing nirmatrelvir except for a single substituent incorporating one more carbon in its alkyl group than Pfizer used in nirmatrelvir. The ’953 patent smartly subtracted that extra carbon from its specification and claims, which then dutifully read on nirmatrelvir. But Pfizer’s publication of nirmatrelvir predated those edits in the ’953 patent, effectively elevating nirmatrelvir’s status to that of a prior-art compound.

This sequence of events ensnared Enanta in a potentially self-defeating case strategy. Successfully proving nirmatrelvir’s infringement of the ’953 patent might simultaneously prove the patent’s invalidity. Enanta sued Pfizer anyway, asserting priority to the ’048 provisional which, but for an alleged obvious typographical error, included nirmatrelvir’s abbreviated alkyl substituent in its disclosure.

The district court was unconvinced. It found that the alleged “error” in the ’048 provisional was not so obvious and declined to judicially correct it. Without correction, the ’953 patent fell victim to the anticipating prior-art compound, nirmatrelvir. The Federal Circuit determined the district court misapplied the judicial correction caselaw but held Enanta lost anyway because the ’048 provisional lacked priority-conveying written description support for the ’953 patent’s nirmatrelvir-capturing claims.

This case challenged Enanta to navigate competing priorities that undergird patent law. On the one hand, patent disclosures provide the public fair notice of the patent’s scope. You can bet your company on it, to the extent of running an obscure drug candidate through costly clinical trials and FDA approval. On the other hand, patentees may act as their own lexicographers, defining ordinary terms in non-traditional ways; most often to avoid prior art and to focus a patent’s claims. The price of that leniency is judicial impatience with a patentee requesting relief from its own mistakes.

Enanta’s case failed to achieve escape velocity from those undergirding gravitational forces. These unspoken policies may have compelled the outcome of the case as much as the legal doctrines espoused in the courts’ published decisions. In the end, Enanta was confined to its voluntarily chosen words in a titans’ clash over a single carbon atom in a blockbuster drug.

To appreciate Enanta’s courtroom fall to earth, we must attend carefully to a timeline hastened by the peculiar events confronting the world, beginning in 2020.

II. Background

In early 2020, a novel coronavirus swept across the world faster than the institutions built to contain it. Within months it had closed borders and schools, filled hospitals, and ended the lives of millions globally; there was no vaccine, and no drug reliably treated the disease it caused. The world’s laboratories dedicated themselves to the emergency, compressing the ordinary, studious procession of drug discovery into a race measured in months rather than years.

A. The Race for the Protease

By the spring of 2021, much of the world’s chemical-research enterprise had spent the better part of a year converging on a single enzyme. SARS-CoV-2, the virus behind COVID-19, depends for its replication on a protein-cutting enzyme called “the main protease” (Mpro or 3CLpro), and that biological dependence made the protease a promising target for anyone hoping to stop the virus with a pill. Laboratories screened thousands of existing drugs against it; structural biologists mapped how candidate molecules settled into its active site; and a handful of companies set out to design a wholly new inhibitor from scratch. See Laura Howes, X-Ray Screening Finds Two Drugs That Could Help Treat COVID-19, Chem. & Eng. News (Apr. 8, 2021) (describing a screen of nearly 6,000 drug molecules against the main protease). The protease had become, in the field’s shorthand, the thing to inhibit.

Two entrants in that race filed papers at the Patent Office within weeks of one another in the summer of 2020, and those two filings would meet years later in this case. On July 20, 2020, Enanta filed the ’048 provisional, directed to compounds that inhibit coronavirus replication by blocking the main protease. Two days later, on July 22, 2020, chemists at Pfizer first synthesized a protease inhibitor of their own: a mere seven milligrams of a compound then designated internally as “PF-07321332” and later named “nirmatrelvir.” See Bethany Halford, Pfizer Unveils Its Oral SARS-CoV-2 Inhibitor, Chem. & Eng. News (Apr. 7, 2021) (first quantity synthesized in late July 2020). On September 3, 2020, Pfizer filed its own provisional application disclosing that compound. Enanta, D. Ct. Decision at 2–3. Neither company yet knew what the other had done; provisional applications are filed in confidence, and both sets of papers sat sealed at the Patent Office.

B. The Single Carbon

The commercial winner of that near photo finish of patent filings was determined by a single carbon atom. Nirmatrelvir, like the compounds the ’953 patent claims, hangs various chemical groups off a central molecular scaffold. One of those groups, at the position the patent labels “A,” bears an acyl amide substituent written in the patent’s shorthand as —NHC(O)—C₁-alkyl. The “C₁” forms the crux of the dispute here. It says the fragment at that point is an alkyl group that is only one carbon long. In nirmatrelvir, that fragment is a trifluoromethyl group, a single carbon bearing three fluorines, the heart of what chemists call a trifluoroacetamide. See Enanta, D. Ct. Decision at 3; Enanta, slip op. at 3.

One carbon. No more.

Both the ’048 provisional and the ’953 patent define the term “substituted” by reciting dozens of permissible substituents, a dense, carefully assembled catalog of optional chemical fragments. See Enanta, slip op. at 3. Buried in the definition’s catalog is the entry at issue here. In the ’953 patent, it reads “—NHC(O)—C₁-C₁₂-alkyl”: an alkyl fragment optionally ranging anywhere from one to twelve carbons long. Because the range begins at one carbon, it literally covers nirmatrelvir’s single carbon. But the ’048 provisional (the document filed in July 2020, the early date critical to Enanta’s case) disclosed something conspicuously different. Its corresponding entry read “—NHC(O)—C₂-C₁₂-alkyl”; it began counting at “two” carbons. Id. And a range that begins at two does not cover a fragment only one carbon long.

For Pfizer, that one-carbon gap was akin to open water. Nirmatrelvir’s fragment is a one-carbon —NHC(O)—C₁-alkyl, and Enanta’s earliest disclosure of —NHC(O)—C₂-C₁₂-alkyl does not include it. See Enanta, D. Ct. Decision at 20. Pfizer’s compound therefore sat just outside the boundary that Enanta’s ’048 provisional had drawn. One carbon clear of it. Nor was Pfizer navigating by Enanta’s charts. It built nirmatrelvir from scratch during the urgency of the pandemic, seeHalford, supra, all while Enanta’s provisional sat sealed and unseen at the Patent Office. The one carbon left out of Enanta’s provisional application had unwittingly gifted Pfizer free passage to an ocean of opportunity.

C. The Disclosure That Burst into Global View

For nine months, nirmatrelvir was Pfizer’s proprietary secret. That ended on April 6, 2021. At the American Chemical Society’s Spring 2021 national meeting, Dafydd Owen, a Pfizer director of medicinal chemistry, walked the audience through the design of the company’s oral antiviral candidate in a symposium of the Division of Medicinal Chemistry. It was no small audience. The meeting was virtual that year (the pandemic had pushed ACS online), so the session was not a mere gathering of specialists in a hotel ballroom but a presentation open to anyone, anywhere, who registered. And registration that spring surged by nearly half over the prior national meeting. See Am. Chem. Soc’y, 2021 Annual Report 2. The next morning, Chemical & Engineering News, the Society’s news magazine, carried Pfizer’s revelation to a global readership, complete with the molecule’s full chemical structure. See Halford, supra.

Printed in that structure, in its upper corner, was a trifluoromethyl group, a single carbon bearing three fluorines, bonded through a carbonyl to an amide nitrogen. In the patent shorthand that would later decide the case, it was an —NHC(O)—C₁-alkyl group: an amide built on one carbon. The very feature at the center of this dispute was, as of April 7, 2021, on public display before the world’s chemists.

Pfizer could afford that openness because it had already laid claim to nirmatrelvir. It had filed its own provisional application disclosing nirmatrelvir seven months earlier; the disclosure to the global audience strategically followed the patent paperwork, not the other way around. Enanta’s filings, by contrast, remained invisible. A provisional application is never published on its own, and the patent family that would claim its benefit did not publish until March 10, 2022, by which time Paxlovid had been authorized for emergency use and was already reaching patients. See 35 U.S.C. § 122(b); U.S. Patent Application Pub. No. 2022/0073499 (Mar. 10, 2022).

So on the day Pfizer’s structure traveled the world’s online video networks, Enanta’s ’048 provisional remained sealed with its single misaligned substituent, while Pfizer’s Paxlovid was well on its way to pharmacy shelves around the world.

D. The Post-Reveal Scramble

For at least nine months the two companies had chased the same protease in parallel, each blind to the other’s work. Enanta’s and Pfizer’s provisional applications, filed six weeks apart, sat sealed in the same Patent Office; neither company had access to the other’s filings. But Pfizer’s April 6, 2021 disclosure broke that symmetrical silence and handed Enanta a long-shot gamble. With its own patent filings sealed until March 2022, Enanta could unilaterally preview both Pfizer’s molecule and its early demonstrated progress into the clinic. And Pfizer’s molecule, with its single one-carbon substituent, sat enticingly within Enanta’s reach.

Before April 6, 2021, the —NHC(O)—C₂-C₁₂-alkyl range recited in Enanta’s ’048 provisional promised Enanta exclusive rights to the disclosed protease inhibitors doing their work with two to twelve carbons. And Enanta has been developing protease inhibitors of its own that presumably reflect the provisional’s disclosure. See, e.g., Press Release, Enanta Pharms., Inc., Enanta Pharmaceuticals Doses First Subject in a Phase 1 Clinical Study of EDP-235, its Oral 3CL Protease Inhibitor Specifically Designed for the Treatment and Prevention of COVID-19 (Feb. 16, 2022). If so, Enanta can be applauded for initiating a defensive patent strategy that someday may be used to exclude commercial knockoffs. But any such commercial prospects lie ahead in an uncertain future. Enanta’s current commercial prospects however would change dramatically if the ’048 provisional’s disclosed alkyl range could somehow be read to include —NHC(O)—C₁-alkyl.

By its own account, Enanta did not immediately recognize this awaiting opportunity. It was not until July 9, 2021, just over three months after Pfizer’s nirmatrelvir disclosure, that Enanta “realized” that the “C₂” reference in its provisional application was a “typographical” error. In Enanta’s telling, “C₁” was the correct and intended starting point of the alkyl group’s range. Enanta, slip op. at 3.

Immediately after that “realization,” Enanta and its patent counsel sprang into action. Ten days later, on July 19, 2021, Enanta filed a non-provisional application reciting an —NHC(O)—C₁-C₁₂-alkyl range. And there had been no time to waste. That new filing was made one day before the twelve-month window closed under 35 U.S.C. § 119(e) to preserve the ’048 provisional’s priority date. Id. That one-carbon edit weaponized an otherwise defensive chain of patent filings. When the final application in the chain issued as the ’953 patent on June 14, 2022, the patent boasted claims to a drug already sold under emergency use authorization to a world in desperate need. The ’953 patent offered Enanta extraordinary leverage.

The commercial stakes were staggering. Pfizer had paired nirmatrelvir with another active ingredient, ritonavir, to formulate Paxlovid. The FDA authorized the combination drug for emergency use on December 22, 2021, and full approval followed in May 2023. Paxlovid earned nearly $19 billion in 2022 alone. See Paxlovid (Nirmatrelvir and Ritonavir) FDA Approval History, Drugs.com (last visited June 2026); FDA Grants Full Approval to Paxlovid to Treat Covid in High-Risk Adults, NBC News (May 25, 2023).

With its retargeted patent finally in hand, Enanta sued Pfizer on June 21, 2022, asserting that Paxlovid infringed claims 1, 2, 5, and 9 of the ’953 patent, which had issued only seven days before. Because the patent claims were drafted in full view of Pfizer’s nirmatrelvir, it was an unusually formidable lawsuit. On the infringement question alone, it was clearly a winner. Yet that winning strength was simultaneously the lawsuit’s greatest weakness. To succeed, Enanta had to prove that nirmatrelvir, shown to the world over a year before the ’953 patent’s filing date, was not anticipating prior art. This would require Enanta to convince the court that “C₂” as written in the ’048 provisional was properly understood to mean “C₁” and that any ensuing confusion was merely the result of an obvious typographical error.

III. The Courts’ Analysis

Enanta’s case strategy had to clear two significant hurdles: the U.S. District Court for the District of Massachusetts and the U.S. Court of Appeals for the Federal Circuit. Enanta lost in both courts but for different reasons.

A. The District Court

1. The typo in Enanta’s case

Enanta’s typographical-error theory was no throwaway alternative argument. It was likely Enanta’s best pathway to a possible victory. The issue presented to the district court was whether the alleged typographical error could transport the ’953 patent’s priority back in time to the July 2020 filing date of the ’048 provisional.

A provisional application cannot be amended to say something it never said; once filed, the substance of its disclosure is fixed in time. Nor can a later application reach back and supply details the provisional failed to disclose; that is pejoratively referred to as adding “new matter,” which courts reflexively ignore for priority.

Here, the ’953 patent conveniently sports the case-winning expression “—NHC(O)—C₁-C₁₂-alkyl” in its specification and claims. But that expression is found nowhere in the ’048 provisional. To the casual observer, it looks like new matter. Enanta however insisted that the missing single-carbon alkyl is not new matter. According to Enanta, it was modestly sitting in the ’048 provisional all along, thinly veiled by an obvious typographical error. If Enanta could succeed on this argument, it just might win the case.

It is true that a court may read a patent as though an obvious error never existed, an act known as “judicial correction.” But that judicial authority is narrowly confined. It applies only where the error is obvious and its correction is “not subject to reasonable debate” on the face of the claims and specification, and where the prosecution history suggests no different reading. Novo Indus., L.P. v. Micro Molds Corp., 350 F.3d 1348, 1354 (Fed. Cir. 2003); see also In re Oda, 443 F.2d 1200, 1203-04 (CCPA 1971) (correcting obvious error does not introduce new matter). The strictness of that standard is not mere judicial fussiness; it necessarily protects competitors who must be able to trust a patentee’s own written words.

The question confronting the district court therefore was not whether Enanta’s reading of the ’048 provisional was plausible, but whether the alleged error was obvious and the necessity of its correction beyond reasonable debate. If the obviousness of an error and its correction are beyond reasonable debate, then the burden and responsibility of perceiving the error properly fall on competitors who review the patent filings. If not, then courts refuse to lift a finger to help, and the patentee is left in the predicament it created for itself.

2. The “alkyl” definition

The district court readily agreed with Enanta that the ’048 provisional presented some conspicuous inconsistencies. The foremost inconsistency is apparent on the face of the ’048 provisional’s definition of “alkyl,” which is copied below:

The term ‘alkyl’ as used herein, refers to saturated, straight- or branched-chain hydrocarbon radicals. ‘C₁-C₄ alkyl,’ ‘C₁-C₆ alkyl,’ ‘C₁-C₈ alkyl,’ ‘C₂-C₁₂ alkyl,’ ‘C₂-C₄ alkyl,’ or ‘C₃-C₆ alkyl,’ refer to alkyl groups containing from one to four, one to six, one to eight, one to twelve, 2 to 4 and 3 to 6 carbon atoms respectively.

Enanta, D. Ct. Decision at 9 (emphasis added).

As Enanta asserted, there is indeed “an internal inconsistency between the numerical and written descriptions of the alkyls” for the disputed carbon grouping. Id. at 17-18. Numerically, the group is expressed as “C₂-C₁₂ alkyl” but in words, it is written “one to twelve . . . carbon atoms.” Enanta noted further that “the written numbers ascend in order within the definition of alkyl (i.e. one to four, one to six, one to eight, one to twelve)” but the numerical presentation is not in order. Enanta therefore argued that “it would have been obvious to a POSA that ‘C₂-C₁₂ alkyl’ was incorrect.” Id. at 18.

The expression “C₂-C₁₂ alkyl” is indeed both out of numerical order and contradicted by the text that follows. It does appear to be an unwelcome guest in the “alkyl” definition. Given these glaring inconsistencies, one might expect Enanta itself to have discovered them during the ten days that transpired between its July 9, 2021 “realization” and its July 19, 2021 filing of the corrected non-provisional application leading to the ’953 patent.

Apparently, it did not. With a hint of incredulity, the district court commented, “It is undisputed, however, that the internal inconsistency in the definition of ‘alkyl’ in the ’048 Provisional was not corrected and remains in the ’953 Patent.” Id. The district court also found that the ’953 patent left undisturbed similar inconsistent numerical ordering in the ’048 provisional’s definitions of “alkenyl” and “alkynyl.” Id.

These inconsistencies, asserted at summary judgment as evidence of obvious mistakes, appear to have been no concern at all to Enanta when preparing and filing the ’953 patent and its preceding applications. Why then should they play any material role in the litigation? It seems Enanta’s attention had been turned elsewhere all along. Instead of correcting the definition of “alkyl,” Enanta focused solely on editing the specification’s definition of “substituted.”

3. The “substituted” definition

The original, pre-edited definition of “substituted” in the ’048 provisional, too lengthy to reproduce here, is where the disputed “—NHC(O)—C₂-C₁₂-alkyl” makes its home. And it has a great many neighbors. Spanning nearly a full page, the definition catalogs not only every alkyl substituent the provisional permits but a long parade of others as well: alkenyls, alkynyls, cycloalkyls, and the like.

As for the alkyl substituents, there are nineteen in total. Conspicuously, the first is a bare “C₁-C₁₂-alkyl,” its range beginning, plainly, at one carbon. The remaining eighteen are each written “C₂-C₁₂-alkyl,” and each is attached to the scaffold through a connecting group: —NH—C₂-C₁₂-alkyl, —O—C₂-C₁₂-alkyl, —C(O)—C₂-C₁₂-alkyl, —NHC(O)—C₂-C₁₂-alkyl, and fourteen more in similar fashion. Every one of those eighteen alkyls begins counting its carbons at two.

Enanta seized on the lone “C₁-C₁₂-alkyl” at the top of the list to argue that its presence “would have indicated to a POSA that the subsequent references [in the definition] to C₂-C₁₂ alkyl were ‘obviously’ erroneous.” Enanta, D. Ct. Decision at 19 n.8. Enanta’s expert opined that all subsequent “C₂-C₁₂-alkyls” should have been written “C₁-C₁₂-alkyl” like the first alkyl appearing at the top of the list. According to the expert, the origin of those repeated mistakes was “likely” a “copy paste error.” Id. at 19. So, to reverse those mistakes, Enanta did some more copying and pasting to change all eighteen “C₂-C₁₂-alkyl” substituents to “C₁-C₁₂-alkyl” substituents in the later applications that matured into the ’953 patent.

Knowing that Enanta changed all eighteen to C₁-C₁₂-alkyl, we are left to wonder how Enanta went about drafting the ’048 provisional’s original definition of “substituted” in the first place. Are we to believe all eighteen C₂-C₁₂-alkyl designations were made in error? In other words, Enanta didn’t make just one unfortunate typographical error when recording the disputed —NHC(O)—C₂-C₁₂-alkyl; it made the same mistake seventeen more times within the definition of “substituted.” This would mean that when proofreading the prefiling drafts of the ’048 provisional, Enanta somehow overlooked eighteen obvious typographical errors. And if the first alkyl in the list was written “C₁-C₁₂-alkyl,” then how did a “copy paste error” result in the remaining eighteen being written as “C₂-C₁₂-alkyl”? Or more astutely, outnumbered eighteen to one, doesn’t the single occurrence of “C₁-C₁₂-alkyl” look more like the outcast from the group than the other way around?

Enanta’s case theory does invite some perplexing questions. A cynic might entertain a simpler view of things; that Enanta drafted the definition of “substituted,” including its eighteen C₂-C₁₂-alkyl substituents, as originally intended. If so, then Enanta wrongly substituted eighteen substituents when drafting the ’953 patent, all while claiming priority to the ’048 provisional.

4. A not so Obvious Error

Ultimately, the district court required neither cynicism nor answers to any of these unanswered questions to rule in summary judgment against Enanta; its original intent was never at issue. As applied, the caselaw did not compel the court to decide whether the eighteen “mistaken” alkyl references were an accident or a choice. It questioned only whether the errors were objectively obvious and whether their correction was subject to reasonable debate. In the end, the undisputed inconsistencies Enanta kept alive in the ’953 patent’s definition of “alkyl” supplied ample raw material for just such a case-ending reasonable debate.

Enanta’s first misstep was made by its own expert who “admitt[ed] that a C₂-C₁₂ alkyl is a chemically possible combination. . . .” Enanta, D. Ct. Decision at 19. “Chemically you can have a C₂-C₁₂ alkyl,” he said. Id. So, according to Enanta’s expert, the disputed “—NHC(O)—C₂-C₁₂-alkyl” expression would not, on its own, stand out as an obvious error. It could be chemically correct as written. Following the expert’s lead, the district court concluded, “[T]his is not the type of error that courts have recognized as being so facially obvious to constitute a typographical error.” Id. But the absence of facial obviousness was not enough to decide the case. The court still had to determine if correcting the alleged error was subject to reasonable debate. For that inquiry, the district court had to look to the broader context.

The district court recognized that the disputed “—NHC(O)—C₂-C₁₂-alkyl” expression does not live alone in the ’048 provisional. It’s an alkyl, and the provisional provides its own definition of “alkyl” to inform the reader’s understanding of the alkyls appearing elsewhere within the document. What remained for the court to determine was whether Enanta’s “alkyl” definition could provide enough clarity to make correcting the disputed alkyl, from a “—C₂-C₁₂-alkyl” to a “—C₁-C₁₂-alkyl,” an imperative beyond reasonable debate.

The “alkyl” definition provided no such clarity at all. The district court ultimately determined that “the correction of the error remains subject to reasonable debate,” adding ominously, “particularly here. . . .” Id. at n.9 (emphasis added). In the district court’s telling, “reasonable debate” about the error’s correction “remains” for one self-inflicted reason, “[T]he prosecution history adds ambiguity because Enanta did not correct the internal inconsistency in the definition of ‘alkyl.’” Id.

The “alkyl” definition, as carried forward to the ’953 patent, defines not one but two relevant ranges of alkyls: “C₂-C₁₂alkyl” and “one to twelve . . . carbon atoms.” Id. at 10. The written definition fits snugly within Enanta’s case theory that “—NHC(O)—C₂-C₁₂-alkyl” can be understood to mean “—NHC(O)—C₁-C₁₂-alkyl” when read in context. But the numerical definition of “alkyl” underscores the literal reading of “—NHC(O)—C₂-C₁₂-alkyl” as recited in the original definition of “substituted.” Those conflicting readings render correction of the disputed alkyl expression to “—NHC(O)—C₁-C₁₂-alkyl” helplessly subject to reasonable debate, especially when Enanta’s own expert admitted that the alkyl’s expression, as originally written, is chemically unremarkable. See Enanta, D. Ct. Decision at 19 n.9 (citing Bella Summit LLC v. Gamebreaker, Inc., No. 21-cv-06007, 2022 WL 17882138, at *7-8 (C.D. Cal. Oct. 17, 2022) (reasonable debate found where error repeated multiple times and later corrected in one place but not another)).

Had the ’953 patent been drafted differently, Enanta just might have had a shot at winning before the district court. Instead, Enanta’s own patent drafting left the district court with no choice but to “conclude that Enanta’s changes in the ’953 Patent impermissibly broadened the scope such that the ’953 Patent cannot claim priority to the ’048 Provisional.” Id. at 20.

5. Anticipation

Following the collapse of its asserted priority claim to the ’048 provisional, the ’953 patent’s fate was effectively sealed. And here is the ultimate irony of the patent’s defeat. To prove anticipation in a typical patent case, Pfizer would bear the burden of proving by clear and convincing evidence that all elements of the asserted ’953 patent claims are found in a single prior art reference. Here, Enanta’s own infringement contentions carried that burden for Pfizer, proving that Pfizer’s nirmatrelvir anticipated all elements of the asserted claims. See Enanta, D. Ct. Decision at 22 (citing Gamevice, Inc. v. Nintendo Co., No. 18-cv-01942-RS, 2023 WL 7194871 (N.D. Cal. Oct. 31, 2023), and Empire Tech. Grp. Ltd. v. Light & Wonder, Inc., 705 F. Supp. 3d 1181 (D. Nev. 2023)). After losing the priority battle, Enanta’s success in proving Pfizer’s infringement compelled its loss in the invalidity war.

The district court therefore granted summary judgment of the patent’s invalidity. It saw no need to go further in its analysis and expressly declined to reach Pfizer’s alternative argument that the ’048 provisional independently failed the written-description requirement. Enanta, D. Ct. Decision at 20 n.11.

That issue was taken up later on appeal before the Federal Circuit.

B. The Federal Circuit

1. What this case is not

“We start with what this case is not,” the Federal Circuit announced after summarizing the factual and procedural background. Enanta, slip op. at 5. And in short fashion, the Federal Circuit made clear that despite Enanta’s framing of the issues and the district court’s summary judgment, this case is not about judicial correction of a patentee’s errors. Enanta, slip op. at 6. Without finding fault in the district court’s underlying factual determinations, the Federal Circuit explained that the district court misapplied the law of judicial correction.

First, the district court mistakenly looked to In re Oda for the proposition that obvious errors can be judicially corrected. See Enanta, D. Ct. Decision at 16-17 (citing Oda, 443 F.2d at 1203-04). The Federal Circuit clarified that Oda concerned only the Patent Office’s authority “to reissue patents with corrections pursuant to 35 U.S.C. § 251” where a person of ordinary skill in the art would consider the errors obvious. Enanta, slip op. at 6. The Federal Circuit tersely commented, “But this case is not before the Patent Office, and a change from C₂ to C₁ has not been shown here to be a correction of an obvious error.” Id.

Next, the Federal Circuit addressed the district court’s reliance on Novo for authority to judicially intervene where “the correction is not subject to reasonable debate.” See Enanta, D. Ct. Decision at 15 (citing Novo, 350 F.3d at 1354). Again, the district court had overlooked the specific context of its cited authority. The Federal Circuit explained that Novoauthorizes the district court to “‘correct an error in a patent by interpretation of the patent where no certificate of correction has been issued.’” Enanta, slip op. at 6 (quoting Novo, 350 F.3d at 1354 (emphasis added)). Here, Enanta did not seek correction of the ’953 patent itself but instead requested curative interpretations of the ’048 provisional. The focus was on the wrong document. As for “the existence of an error in the ’048 provisional,” the Federal Circuit expressed its satisfaction that any such alleged error “is certainly ‘subject to reasonable debate.’” Id.

In short, “neither Oda nor Novo applies here.” Id.

2. Priority as possession

Broadly speaking, the district court addressed the right underlying legal issue: whether the ’953 patent properly claimed priority to the filing date of the ’048 provisional. Where it strayed in its analytical path was in examining alleged errors in the ’048 provisional. On appeal, the Federal Circuit simply took the ’048 provisional at face value to determine whether it “provided written description support for —NHC(O)—C₁-alkyl disclosed and claimed in the ’953 patent.” Id. at 7. At stake in this new analysis was whether the ’953 patent was a rightful heir to the ’048 provisional’s priority date.

A patent inherits an earlier application’s priority date only if the earlier application satisfies the written-description requirement of 35 U.S.C. § 112 for the patent’s claims. This means that the priority application must describe the later claimed invention in enough detail that a skilled artisan can perceive that the inventor possessed the claimed invention as of the earlier filing date. Id. at 5–8 (citing Lockwood v. Am. Airlines, Inc., 107 F.3d 1565, 1571 (Fed. Cir. 1997)); see also New Railhead Mfg., LLC v. Vermeer Mfg. Co., 298 F.3d 1290, 1294 (Fed. Cir. 2002). After all, one cannot inherit something that was never in the decedent’s estate regardless of what else might be written in the will.

Measured against that standard, the ’048 provisional failed both to show possession and to convey priority to the ’953 patent for an extraordinarily simple reason. “C₂ is simply different from C₁,” the Federal Circuit wrote; a disclosure that recites an alkyl range beginning at two carbons does not convey possession of the omitted one-carbon alkyl. Enanta, slip op. at 8.

But the issue presented here involved much more than simply counting carbon atoms. The Federal Circuit invoked the perspective of a chemist to emphasize that the difference between one carbon and two is often no trifling matter at all. “The issue in this case,” the Federal Circuit said, “is akin to asking whether a disclosure of ethanol, a two-carbon alcohol regularly consumed by people, would provide adequate written description support for methanol, a one-carbon alcohol that is highly toxic to people.” Id. at 10. That stark “example,” the Federal Circuit observed, “illustrates why a disclosure of one chemical compound, or integer in this case, cannot necessarily be a disclosure of another, even one close by structurally.” Id.

The harsh reality is that “[t]he ’048 provisional did not disclose —NHC(O)—C₁-alkyl” in its definition of “substituted,” and the provisional therefore “provided no written description support for the ’953 patent.” The inventors could not possess an alkyl substituent they failed to disclose, and the ’048 provisional could not bequeath priority to an invention its words did not describe. Id. Having persisted in reading the ’048 provisional at face value, the Federal Circuit displayed little curiosity in the potentially enlarging “one to twelve . . . carbons” text found elsewhere in the definition of “alkyl.” Possession, it appears, requires accurate disclosure where one would expect to find it.

Although arriving from a different analytical path, the Federal Circuit nonetheless affirmed the district court’s “summary judgment that the ’953 patent claims were anticipated by Pfizer’s disclosure of nirmatrelvir.” Id. The Federal Circuit concluded its decision with a trace of reassurance, “We respect applicants’ statements in their specification that they invented what was specifically disclosed in the ’048 provisional, but similarly we conclude that they did not invent what they did not disclose.” Id. at 10.

IV. Conclusion

It is tempting to read this case as a story of utter defeat, but that mistakes the magnitude of what remains within Enanta’s grasp. The ’048 provisional was never a lottery ticket with just a single number. It discloses a Markush genus, a formula whose variables describe not one molecule but an astronomically large family of them, by fair computational reckoning more distinct structures than there are atoms in the universe. With a net cast so wide, missing nirmatrelvir is statistically cruel. But the doctrine that cost Enanta that single compound is the same one that secures the rest: “it did not invent what it did not disclose,” but it possesses, against the world, all that it did.

Enanta seems to understand this. It swung hard at a missed opportunity from the past, yet it appears not to be mourning its strike at the plate. It is already bringing other compounds forward, any one of which might still change a life. Perhaps that is the subtle lesson here, and not for patent lawyers alone. Missed chances and small regrets are the ordinary freight of an ambitious life, but the work we commit to the record remains ours to build on, and the best of what a laboratory, or a life, unfolds lies more often ahead of it than behind.