The New Owner Didn’t File the Patent. Here’s Who Actually Controls It.

Copyright © DrugPatentWatch. Originally published at https://www.drugpatentwatch.com/blog/

Pfizer has sold Xtandi since it bought Medivation for $14 billion in 2016. Pfizer’s name is not on any of the drug’s three core Orange Book patents. All three, filed with a May 13, 2005 priority date, are assigned to the Regents of the University of California [1]. UCLA gave Medivation, then a startup, “the right to seek patent protection under UC’s name,” while the university kept legal title for itself [2]. A decade later, when three cancer patients asked the federal government to force generic competition on the drug, the petition ran through Bayh-Dole march-in rights, a mechanism that only applies because the patent owner of record took federal research money. That mechanism does not care who sells the drug. It cares who owns the patent.

That gap between marketer and owner is not an edge case. It is the normal condition of a mature drug patent. A molecule gets discovered in a university lab or a small biotech, gets licensed or sold to a larger company, and the larger company sometimes gets acquired itself. Each step can change who controls the invention without changing whose name is stamped on the original filing. Reading a patent’s face, or a drug’s marketing materials, tells you who filed. It does not reliably tell you who now owns.

The Short Answer

A patent’s original applicant and its current legal owner are frequently different entities, and the gap widens the longer a drug has been on the market. Ownership changes through licensing (which does not transfer title), corporate acquisition (which does, but only if someone files the paperwork), and outright sale of a royalty stream (which changes who gets paid without changing who owns the patent at all). The U.S. Patent and Trademark Office (USPTO) records these changes in a public assignment database, but recordation is what the agency itself calls a “ministerial function”: the office does not verify that the transaction is real, valid, or complete [3]. The FDA’s Orange Book listing process works the same way. It relies on the applicant’s own attestation and does not independently confirm chain of title [4]. When that gap is not tracked correctly, it has ended patent infringement lawsuits before trial [5].

The Findings That Matter

  • Xtandi’s three core Orange Book patents (7,709,517, 8,183,274, and 9,126,941) remain assigned to the Regents of the University of California more than eight years after Pfizer acquired the drug’s commercial rights through its purchase of Medivation [1] [6].
  • UCLA has collected more than $1 billion in royalty income from Xtandi without ever taking on manufacturing, marketing, or the patent litigation that Pfizer and Astellas conduct in its name [7].
  • When Gilead Sciences acquired Pharmasset for roughly $11.2 billion in January 2012, the sofosbuvir patent portfolio was not reassigned to “Gilead Sciences.” It sits under a separate corporate entity, Gilead Pharmasset LLC, a name a keyword search for “Gilead” alone will often miss [8] [9].
  • Emory University sold its 20 percent royalty interest in the HIV drugs Emtriva and Truvada to Gilead and Royalty Pharma for $525 million in 2005, without transferring the underlying patents themselves [10].
  • In 2023, LifeArc, the UK charity that inherited a royalty stake in Keytruda from a 2007 research collaboration with Organon, sold part of that stake to the Canada Pension Plan Investment Board for $1.3 billion, sixteen years after the antibody program’s original corporate home had already been absorbed twice [11].
  • The Federal Circuit dismissed Abraxis Bioscience’s infringement suit over the anesthetic Naropin with prejudice because Abraxis could not prove it held legal title to the asserted patents on the day it filed the complaint, even though a later, backdated assignment said otherwise [5].
  • Under 35 U.S.C. 261, an unrecorded assignment can be legally void against a later buyer who has no notice of it, unless the assignment is recorded within three months of its date [12].

What “Original Applicant” Actually Means on a Patent

Every issued U.S. patent carries an assignee field, and most patent search tools, including Google Patents, surface it as “original assignee.” That single label hides two separate facts: who owned the invention at the moment of filing, and who owns it now. The two frequently diverge, and nothing on the face of an old patent tells you that they have.

The Inventor Isn’t Automatically the Owner

Patent law starts from a default rule that is easy to forget inside a large organization: an invention belongs first to the person who invented it. Employers, universities, and funding agencies only get ownership because the inventor signed something that gives it to them, not automatically because they paid for the lab. The Supreme Court restated this in stark terms in 2011, holding that the Bayh-Dole Act, which governs inventions made with federal research funding, “does not automatically vest title to federally funded inventions in federal contractors” [13]. Title moves by assignment, and an assignment is a specific legal act, not a byproduct of a grant or an employment relationship.

Employment Agreements vs. Automatic Assignment

The distinction matters because most employment and university invention agreements use future-tense language: the employee “agrees to assign” rights that arise later. That is a promise to sign something in the future, not a transfer that happens the moment the invention exists. A minority of agreements use present-tense, self-executing language instead, of the kind the Supreme Court quoted in Stanford v. Roche: an inventor who signs a document stating he “will assign and do[es] hereby assign” his rights has already transferred them, automatically, the instant the invention is conceived [14]. In that case, a Stanford researcher named Mark Holodniy had signed exactly this kind of self-executing agreement with Cetus Corporation, a company where he did short-term training, years before he signed Stanford’s own weaker, future-tense agreement. When Cetus’s PCR business was later sold to Roche, Roche inherited rights that predated, and beat, Stanford’s own claim to the same invention. Stanford owned the patents. It did not own enough of them to stop Roche from selling competing HIV test kits [14].

Original Assignee vs. Assignee of Record: Two Different Database Fields

The USPTO’s own recordkeeping distinguishes between the assignee named at filing and the current owner of record, tracked through its separate patent assignment database rather than the patent document itself. The Manual of Patent Examining Procedure describes assignment as the transfer of “all or part of its right, title and interest in a patent [or] patent application” from one party to another, a transaction that is recorded, not automatic, and that requires a written instrument [15]. A patent granted in 2005 to a small biotech and never touched again will show that biotech as both the original assignee and the current owner. A patent that changed hands through an acquisition in 2012 will show the original biotech in the patent’s own front-page metadata forever, because that metadata is fixed at grant, while the current owner only appears in the separately maintained assignment database, if someone bothered to record the transaction there [12] [15].

Why a Patent’s “Original Assignee” Label Can Mislead Years Later

This is where most keyword-based patent research goes wrong. Searching “Merck” or “Pfizer” in a patent database’s assignee field will surface everything currently or originally filed under that name, but it will systematically miss patents that were filed under an acquired company’s name and never formally reassigned in the USPTO’s separate database, even though the acquiring company now controls every commercial decision about the drug. The patent’s own bibliographic page keeps saying “Regents of the University of California” or “Organon” indefinitely. Only a separate assignment-database search, cross-referenced against corporate transaction history, closes that gap.

The Legal Mechanics Behind Patent Ownership

Patent ownership in the United States runs through a compact but consequential statute, chapter 26 of Title 35, and a set of regulations that describe exactly how, and how loosely, the government tracks it.

35 U.S.C. 261: Patents as Personal Property

The statute opens by declaring that patents “shall have the attributes of personal property,” assignable “by an instrument in writing” [12]. That single sentence is why a patent can be sold, licensed, mortgaged, or willed to heirs like any other asset a company owns. It is also why an oral agreement, an email describing an intent to transfer patent rights, or a corporate merger by itself is not sufficient on its own to move legal title. Something has to be signed.

Recordation Is a “Ministerial” Act, Not a Verification

The USPTO is required to “maintain a register of interests in patents and applications for patents” and to “record any document related thereto upon request” [12]. Federal regulation, 37 CFR 3.54, is explicit that recording an assignment document is “an administrative action and not a determination of the validity of the document or of the effect that the document has on the title” [16]. In plain terms, the USPTO will file whatever paperwork a party submits and will not check whether the underlying transaction was real, complete, or legally sufficient. Harvard Business School’s own guidance to researchers using the database repeats the agency’s own caution almost verbatim: the USPTO “does not verify the validity of the information,” and recordation “is a ministerial function” [17]. Two separate parties can, in principle, both file paperwork claiming to own the same patent, and the registry will record both, leaving the actual legal question to a court.

The Three-Month Race Against Subsequent Purchasers

The statute does attach one real consequence to recording promptly. An assignment “shall be void as against any subsequent purchaser or mortgagee for a valuable consideration, without notice, unless it is recorded in the Patent and Trademark Office within three months from its date or prior to the date of such subsequent purchase” [12]. This is a notice-race rule borrowed from real estate law: whoever records first, or within the three-month grace window, generally wins against a later buyer who did not know about the earlier deal. It is also the exact provision at the center of the Abraxis dispute discussed below, where a company’s failure to get clean, timely paperwork in place cost it the case.

Why This Matters for M&A Due Diligence

Because recordation is optional and self-reported, a company’s own internal records, not the USPTO database, is often the only reliable source for confirming that every patent in an acquired portfolio was actually, formally assigned at each step of its corporate history. Ropes & Gray’s own patent transaction guidance describes chain-of-title diligence as running “from initial inventors through to the current assignee-owner,” a process undertaken precisely because the public record cannot be assumed to be complete or current [18].

How the USPTO Assignment Search Database Works Today

The USPTO retired its two legacy lookup tools, Patent Assignment Search and Assignments on the Web, and replaced them with a single consolidated Assignment Search application inside its Assignment Center platform, effective September 27, 2025 [19]. The underlying database still covers recorded assignments from August 1980 forward; records before that date are held by the National Archives [20]. The database can be searched by patent number, by assignor, or by assignee name, which makes it possible to trace a specific patent’s ownership history directly, distinct from relying on the assignee field baked into the patent document itself.

Case Study: Xtandi and the License That Never Became a Deed

How a UCLA Chemistry Lab Produced a $5-Billion-a-Year Franchise

UCLA chemist Michael Jung and physician-scientist Charles Sawyers began the research that became enzalutamide at the university in the early 2000s, discovering a molecule that blocks androgen receptor signaling in prostate cancer cells [6] [7]. UCLA filed the resulting patent applications, and on August 12, 2005, licensed three of them to Medivation, then a small biopharmaceutical startup [6]. Medivation entered a global co-development and commercialization agreement with Astellas Pharma in October 2009 [6]. The FDA approved the resulting drug, Xtandi, in 2012. Pfizer acquired Medivation for $14 billion in 2016, inheriting Medivation’s license and its share of the Astellas collaboration [7] [9]. By 2021, US sales alone added nearly $1.2 billion to Pfizer’s revenue for the year, and combined global sales run to roughly $5 billion annually split between Pfizer and Astellas [8] [7].

Why the Patents Still Say “Regents of the University of California”

None of that changed who owns the patents. UCLA’s own governing board was explicit about the structure in a 2018 memo to its Health Services Committee: the university “retained title to enzalutamide and gave Medivation, a start-up company at the time, the right to seek patent protection under UC’s name,” so that when Pfizer later pursued patent protection for Xtandi in India, the memo describes Pfizer as acting “as the successor-in-rights” to a license, not as a patent owner in its own name [2]. Every enforcement action tracks that structure. When Astellas, Pfizer’s Medivation unit, and the Regents of the University of California sued Sun Pharmaceutical Industries in December 2022 over a proposed generic, all three were named plaintiffs, because none of the licensees alone had standing to sue without the record owner joining the case [8].

UCLA has collected more than $1 billion in royalty income from a drug it has never manufactured, marketed, or sold at retail, income generated entirely because its name, not Pfizer’s or Astellas’s, still sits on the patents [7].

What This Means for Bayh-Dole March-In Rights

Because federal research funding supported the original UCLA work, the patents are subject to the Bayh-Dole Act’s march-in provision, a mechanism that lets a federal funding agency force additional licensing of an invention under specified conditions, including a failure to make the invention available to the public on reasonable terms [21]. In November 2021, three prostate cancer patients petitioned the Department of Health and Human Services to exercise march-in rights on the Xtandi patents specifically because of that federal funding history, not because of anything Pfizer or Astellas had separately done as licensees [21] [22].

The March 21, 2023 Denial

The National Institutes of Health, acting on behalf of HHS Secretary Xavier Becerra, rejected the petition on March 21, 2023, reasoning that the drug was widely available on the market and that march-in rights were not an effective tool for addressing price alone [10] [22]. The agency’s reasoning is contested by health-policy researchers and members of Congress, who argue HHS did not directly answer the petitioners’ central claim that the drug’s US price violated the statute’s “reasonable terms” language [23]. What is not contested is the legal mechanism itself: march-in rights only exist here because the original patent owner, UCLA, took federal funding for the underlying research decades earlier. A generic drug’s patent thicket built entirely on private R&D dollars carries no equivalent lever.

What This Means for Standing in Generic Litigation

The Sun Pharmaceutical suit illustrates a practical consequence of the ownership structure beyond march-in exposure. A patent infringement suit requires the plaintiff to hold, or be a sufficiently exclusive licensee of, the patent being asserted. Because Medivation’s rights under the UCLA license are non-exclusive as to the record title itself, the university has to appear as a named party in every Xtandi patent suit for the case to proceed cleanly, adding a coordination step that a fully owned patent estate would not require.

Case Study: Keytruda and Two Corporate Mergers

From a Dutch Antibody Program to Merck’s Top Seller

Researchers Andrea van Elsas and Hans van Eenennaam began the antibody work that became pembrolizumab in 2003 at Organon, a Dutch pharmaceutical company, through a partnership with MRC Technology, the commercialization arm of the UK’s Medical Research Council [24] [11]. Organon was acquired by Schering-Plough in 2007 for roughly $14 billion, and Schering-Plough was itself acquired by Merck two years later, in 2009, for $41 billion [11]. Pembrolizumab, still an early-stage program at the time of the second deal, became Keytruda, one of the highest-grossing drugs in the world once Merck advanced it through clinical development and FDA approval [24].

Tracing the Chain: Organon to Schering-Plough to Merck

The corporate chain here differs from Xtandi’s in one important structural way. UCLA never sold the company that held its license; Medivation was acquired, but UCLA’s ownership stake in the patents themselves never moved. With Keytruda, the entity that actually employed the inventors and initially controlled the program was acquired twice over, first by Schering-Plough and then, as part of that combined company, by Merck [11]. Corporate acquisitions of this kind transfer legal control of the acquired company’s assets, including its patent applications, but only formally complete a patent’s ownership record if the resulting assignment paperwork is filed with the USPTO at each step. A biosimilar or generic developer researching Keytruda’s foundational antibody patents by searching “Merck” alone in an assignee field risks missing filings still recorded under Organon’s or Schering-Plough’s corporate name if that recordation step lagged behind the underlying M&A closing.

LifeArc’s Royalty Stake and the $1.3 Billion Sale to CPPIB

Ownership of the patents is not the only interest that survived two acquisitions. MRC Technology, renamed LifeArc in 2017, retained a royalty interest in Keytruda’s worldwide sales stemming from its original 2007 collaboration agreement with Organon, an interest that neither the Schering-Plough nor the Merck acquisition disturbed [11]. In 2023, the Canada Pension Plan Investment Board paid $1.3 billion for a partial stake in that royalty stream, a transaction between a Canadian pension manager and a UK medical research charity that has nothing to do with Merck’s ownership of the underlying patents [11].

Economic Interest Is Not Legal Title

LifeArc does not appear as an assignee on Keytruda’s patents, and CPPIB’s purchase did not change that. What CPPIB bought is a contractual right to a share of Merck’s future Keytruda sales, an economic interest layered entirely on top of, and legally separate from, the patent ownership itself. A researcher checking the USPTO assignment database for Keytruda’s patents will find no trace of LifeArc or CPPIB anywhere in it, because neither one owns, or has ever owned, any interest the statute defines as patent title.

Case Study: Sovaldi and the Subsidiary Nobody Searches For

Pharmasset, Emory, and the Origins of Sofosbuvir

Sofosbuvir, the active ingredient in Gilead’s hepatitis C drug Sovaldi, traces to research led by Emory University biochemistry professor Raymond Schinazi, who founded Pharmasset Inc. to commercialize the compound [25]. Pharmasset raised roughly $45 million in a 2007 IPO and spent an estimated $62.4 million of its own funds developing sofosbuvir through 2011, according to the company’s SEC filings [25]. Gilead Sciences acquired Pharmasset for approximately $11.2 billion in January 2012 [25] [26].

Why the Assignee of Record Is “Gilead Pharmasset LLC,” Not “Gilead Sciences”

The acquisition did not fold the sofosbuvir patent estate directly into “Gilead Sciences, Inc.” as the recorded assignee. A search of issued patents on Justia’s assignee index shows the sofosbuvir-related patent family, including formulations combining sofosbuvir with velpatasvir and voxilaprevir, assigned to a distinctly named subsidiary, Gilead Pharmasset LLC [9]. This is a common and legally unremarkable structure: acquiring companies frequently hold acquired IP inside a dedicated subsidiary for tax, liability, or internal accounting reasons. It is also a practical trap for anyone researching drug patent ownership by brand-name keyword search alone. A search for “Gilead Sciences” as an assignee will not reliably surface patents recorded under “Gilead Pharmasset LLC,” even though the parent company controls both entities completely.

A Practical Search Problem

The same pattern recurs across the industry. Acquired biotech and pharma patent estates routinely sit inside subsidiary entities that carry the target company’s original name, a merger subsidiary’s name, or a holding-company name invented purely for the transaction, rather than the parent’s public brand. Complete ownership research requires searching by patent number or by every known corporate name in a company’s acquisition history, not by the parent brand alone.

The Separate Emtriva and Truvada Royalty Sale to Royalty Pharma

Sofosbuvir was not the first Emory-originated compound Gilead came to control. Emtricitabine, the active ingredient in Gilead’s earlier HIV drugs Emtriva and, in combination, Truvada, was discovered by Emory researchers including Dennis Liotta and Raymond Schinazi and initially licensed to Triangle Pharmaceuticals, which Gilead acquired for $464 million in December 2002 [27]. Emory retained a 20 percent royalty interest in the resulting drugs even after the Triangle acquisition. In July 2005, the university sold that royalty interest to Gilead and the specialty finance firm Royalty Pharma for an upfront payment of $525 million [10].

A $525 Million Reminder That Legal Title and Cash Flow Are Different Assets

Emory’s 2005 deal did not transfer patent ownership; Emory had already licensed the underlying invention to Triangle, and Gilead had already acquired Triangle’s patent rights through the 2002 acquisition. What Emory sold in 2005 was a royalty stream contractually owed to it under the original licensing terms, a cash-flow right that is legally and structurally distinct from patent title itself. The same distinction separates a patent’s assignee of record from a royalty holder in nearly every academic-origin drug: the patent registry tracks who owns the invention, not who gets paid on it.

When Getting the Chain of Title Wrong Costs You the Case

Abraxis Bioscience v. Navinta: Standing Requires Title on Day One

Abraxis Bioscience marketed the anesthetic Naropin, protected in part by patent 4,870,086, covering ropivacaine hydrochloride monohydrate, and two related method patents [5]. Abraxis had acquired the patents from AstraZeneca’s UK entity through a 2006 Asset Purchase Agreement, but the actual written transfer of the specific patents at issue was executed by different AstraZeneca affiliates in March 2007, months after Abraxis had already filed its infringement suit against generic maker Navinta over a proposed ropivacaine ANDA [28] [29]. Abraxis argued that a later, backdated “nunc pro tunc” assignment, one explicitly written to take legal effect retroactively, cured the gap. The Federal Circuit disagreed. It held in November 2010 that “a plaintiff could not execute a nunc pro tunc assignment of the asserted patent to remedy the fact that the plaintiff lacked legal title to that patent at the time it filed the lawsuit,” reversing the district court and ordering the case dismissed with prejudice [30]. The Federal Circuit denied rehearing in March 2011, and the Supreme Court denied certiorari [29].

Why a Nunc Pro Tunc Assignment Couldn’t Fix It

The court’s reasoning tracks directly back to Article III standing doctrine, not just patent-specific rules: a plaintiff has to hold enforceable legal title to the patent at the moment it files suit, and no subsequent paperwork, however it is dated, can manufacture standing that did not exist on the filing date [31]. A company that believes it owns a patent because a deal has been agreed, signed in principle, or is in the process of closing does not yet own it for litigation purposes until the specific, written assignment instrument required by 35 U.S.C. 261 has actually been executed by the party that held title at that moment.

Stanford v. Roche: “Will Assign” vs. a Promise to Assign

The Supreme Court’s 2011 decision, discussed above for its holding on Bayh-Dole, doubles as the clearest illustration of how far a single sentence of contract drafting can move legal ownership. Stanford researcher Mark Holodniy had signed Stanford’s own invention agreement, using language that only promised a future assignment. He later signed Cetus Corporation’s confidentiality agreement, which stated he “will assign and do[es] hereby assign” his rights, language the Court treated as an immediate, self-executing transfer [14]. When Cetus’s business was later sold to Roche, Roche’s chain of title reached back to that self-executing clause and predated Stanford’s own weaker claim to the same invention [13] [14]. Stanford owned patents on the technology. It did not, in the end, own enough of the underlying rights to stop Roche from selling a competing product.

What This Means for ANDA Litigation Defendants

Both cases give generic and biosimilar challengers a specific, procedural line of attack that is independent of the underlying patent’s validity. A defendant facing a Paragraph IV suit or an ANDA infringement case can, in principle, investigate whether the plaintiff’s chain of title, not just its patent claims, is airtight as of the filing date, particularly in the aftermath of a recent acquisition, licensing renegotiation, or corporate restructuring involving the asserted patents.

How the FDA’s Orange Book Compounds the Confusion

Form FDA 3542 Lets a Patent Owner Fill It Out, But Not File It

Patents claiming a drug’s active ingredient, formulation, or an approved method of use are listed in the FDA’s Orange Book through a specific submission process. Within 30 days of NDA approval, the applicant must submit patent information on Form FDA 3542 for each qualifying patent [32]. FDA’s own guidance on the form is explicit that the patent owner can complete it, but “only an NDA holder can submit the forms directly to FDA under their NDAs” [33]. In practice, that means the entity legally listing a patent with FDA and the entity that legally owns that patent are frequently two different companies working from the same paperwork, exactly the licensor-licensee structure seen in the Xtandi case.

The FDA Doesn’t Verify Chain of Title Either

Nothing in FDA’s patent listing regulation, 21 CFR 314.53, requires the agency to independently confirm that the patent owner named on Form 3542 actually holds clean legal title to the patent at the time of listing [34]. The requirement is that the listed patent claim the drug substance, drug product, or an approved method of use, and that a person not licensed by the patent owner could reasonably be accused of infringement by making, using, or selling the drug [34]. The listing process, like the USPTO’s own assignment database, is a declarative system built on applicant certification, not an independent title search.

A Second Ministerial Registry

This creates a structural parallel between the two federal systems a pharmaceutical company has to satisfy. The USPTO records assignments without verifying them. FDA lists patents based on the NDA holder’s declaration without independently verifying who owns them. A gap or error in the underlying chain of title can sit undetected in both registries simultaneously, surfacing only when a generic challenger’s litigation team, or a court, actually tests it, as happened in Abraxis.

What Happens When the NDA Holder and Patent Owner Are Different Companies

The Xtandi structure again illustrates the pattern cleanly: Astellas and Pfizer’s Medivation unit hold the NDA and file the Orange Book patent listings, while the Regents of the University of California hold the underlying patents [1] [2]. FDA’s process accommodates this without friction, because the regulation contemplates a patent owner and an NDA applicant being different parties from the outset [33]. The complexity only becomes visible downstream, in litigation, in march-in petitions, and in any due diligence exercise that assumes the NDA holder and the patent owner are the same company because they usually are.

Four Types of Ownership Divergence: An Original Taxonomy

The cases above sort into four recurring patterns. None of these categories is officially recognized by USPTO or FDA; they are a way of organizing the recurring structures this analysis surfaced across real drug patent histories.

Type 1: License-Only Control

The original applicant keeps legal title permanently. A licensee, sometimes after being acquired itself, controls commercialization, litigation strategy, and pricing, but never becomes the patent’s assignee of record. Xtandi is the clearest example: UCLA has held title continuously since 2005, through Medivation’s entire corporate life and Pfizer’s acquisition of it [1] [2].

Type 2: Corporate-Successor Control

The original applicant company is itself absorbed into an acquirer, and legal title moves with the corporate transaction, but the assignment record can lag behind the deal, or route through a subsidiary name the public rarely encounters. Sovaldi’s patents sitting under “Gilead Pharmasset LLC” rather than “Gilead Sciences” is the cleanest illustration [9].

Type 3: Economic-Only Carve-Out

Legal title to the patent never moves. A separate contractual right to a share of future revenue is sold or monetized independently, to a party that never appears in the patent’s ownership record at all. Emory’s 2005 sale of its Emtriva and Truvada royalty stream, and LifeArc’s 2023 sale of part of its Keytruda royalty stake to CPPIB, both fit this pattern [10] [11].

Type 4: Defective or Contested Title

The paperwork intended to transfer ownership is incomplete, improperly timed, or contested, creating real litigation exposure independent of the patent’s underlying validity. Abraxis v. Navinta is the clearest cautionary example; Stanford v. Roche shows the same risk arising from contract drafting rather than a corporate transaction [5] [13].

Recognizing the Warning Signs

Recent M&A activity involving the patent, a licensor and licensee with separately negotiated litigation rights, a patent asserted shortly after an acquisition closes, and any assignment recorded close to, or after, a lawsuit’s filing date are the four practical signals worth checking before relying on a patent’s face-value ownership.

CaseOriginal ApplicantCurrent Legal OwnerTaxonomy Type
Xtandi (enzalutamide)Regents of the University of CaliforniaStill the Regents of the University of CaliforniaType 1: License-only control
Sovaldi (sofosbuvir)Pharmasset, Inc.Gilead Pharmasset LLCType 2: Corporate-successor control
Emtriva / Truvada royaltyEmory University (royalty interest)Royalty split between Gilead and Royalty Pharma; patents held by GileadType 3: Economic-only carve-out
Naropin (ropivacaine)AstraZeneca UK entitiesDisputed; Abraxis’s claim to own the patents on the suit’s filing date was rejectedType 4: Defective or contested title

How to Actually Trace Who Controls a Drug Patent

Start with the Face of the Patent, Not the Brand Name

Pull the actual patent number from the Orange Book listing, not the drug’s brand name, and check the original assignee field on the patent document itself. That field is fixed at grant and will never update, no matter how many times the drug changes commercial hands.

Cross-Check the USPTO Assignment Search Database

Search the patent number directly in the USPTO’s Assignment Search application, the consolidated tool that replaced the legacy Patent Assignment Search and Assignments on the Web systems in September 2025 [19]. This database, not the patent document, is where a recorded change of ownership actually appears.

Read the Litigation Caption, Not the Press Release

When a drug’s patents are being litigated, the parties named as plaintiffs are a reliable, court-tested indicator of who currently holds enforceable rights, because standing requirements force accurate disclosure in a way that marketing materials do not. A university, a small biotech, or a licensor’s name appearing alongside the marketing company in a case caption, as with UC Regents alongside Astellas and Pfizer’s Medivation unit, is a direct signal of a split ownership structure [8].

Watch for Subsidiary Shell Names

When a drug’s controlling company has been through an acquisition, search the assignment database under every known corporate name in that acquisition’s history, including subsidiary and merger-vehicle names, not just the current parent brand.

A Four-Step Reproducible Method

In order: identify the specific patent numbers from the Orange Book listing; check the original assignee on the patent’s own bibliographic record; search that patent number directly in the USPTO Assignment Search database for any recorded changes of ownership; and confirm the result against the parties actually named in any related litigation. Each step checks a different registry, because no single registry in this system is designed to give a complete answer on its own.

What This Means for Generic and Biosimilar Challengers

A Paragraph IV certification requires sending notice to the patent owner, not merely to the NDA holder, within a fixed statutory window [34]. Confirming the correct legal owner, rather than assuming it matches the NDA holder or the drug’s marketing company, reduces the risk of a notice defect that could delay or complicate a generic’s own ANDA timeline. Chain-of-title review is also, as Abraxis shows, a legitimate line of defense independent of invalidity or non-infringement arguments: a plaintiff’s standing can fail entirely if its own paperwork does not hold up.

What This Means for Licensing, M&A, and Due Diligence

An acquirer buying a drug company is not just buying the patents its target company holds directly. It is inheriting whatever licensing structure, royalty carve-outs, and assignment gaps already exist underneath those patents, some of which, as with UCLA’s continued title to Xtandi or LifeArc’s continued royalty interest in Keytruda, can persist for decades without ever converting into outright ownership. Confirming which patents are actually owned outright, as opposed to merely licensed exclusively, changes the real value and risk profile of the deal.

What This Means for Drug Pricing and March-In Advocacy

Bayh-Dole march-in rights only attach where federal funding supported the original invention and where the patent owner, not merely the drug’s marketer, remains identifiable and subject to the statute [21]. Advocacy aimed at a drug’s price by targeting its marketing company can miss the actual legal lever entirely if the underlying patents are still held, as with Xtandi, by an academic institution operating under a different set of institutional incentives than the company selling the drug.

Methodology

This analysis draws on primary sources wherever available: the text of 35 U.S.C. 261 and the Manual of Patent Examining Procedure directly from USPTO and Cornell Law’s Legal Information Institute; the Supreme Court’s own opinion in Stanford v. Roche and the Federal Circuit’s opinion in Abraxis v. Navinta, both read from the courts’ own published text; FDA’s own Form 3542 instructions and 21 CFR 314.53; and University of California Board of Regents meeting materials describing the Xtandi license terms in the university’s own words. Corporate transaction figures (acquisition prices, royalty sale amounts) are sourced to contemporaneous press coverage, SEC filings, and university press releases, and are presented as reported rather than independently recalculated. Patent assignee-of-record information for the Sovaldi portfolio was confirmed against Justia’s public patent assignee index rather than relying on secondary description alone.

Limitations

The USPTO Assignment Search database itself is self-reported and, as the agency states, unverified, meaning that any conclusion drawn from it about a patent’s current legal owner should be treated as a strong, but not absolute, public record. Complete certainty about a specific patent’s chain of title requires the actual underlying assignment instruments, which are not always publicly available in full even when a recordation entry exists.

Key Takeaways

  • A patent’s original assignee field is fixed at grant and never updates. Current ownership has to be checked separately, in the USPTO’s Assignment Search database.
  • Licensing a patent, no matter how exclusive or long-running, does not transfer legal title. Xtandi’s patents have remained with the Regents of the University of California through Medivation’s entire existence and Pfizer’s acquisition of it.
  • Corporate acquisitions can leave a patent portfolio recorded under a subsidiary’s name rather than the parent brand, as with Gilead Pharmasset LLC holding the Sovaldi patents.
  • Royalty interests and patent title are legally distinct assets that can be sold separately, as Emory’s 2005 Emtriva and Truvada royalty sale and LifeArc’s 2023 partial Keytruda royalty sale both demonstrate.
  • Neither the USPTO’s assignment registry nor FDA’s Orange Book listing process independently verifies chain of title; both rely on the submitting party’s own representations.
  • Getting the chain of title wrong carries real litigation consequences. Abraxis Bioscience’s infringement suit was dismissed with prejudice because it could not prove it held legal title on the day it filed the case.

Frequently Asked Questions

What is the difference between a patent’s original assignee and its current owner?
The original assignee is whoever the applicant named as owner at the time the patent was filed or granted, a field that is fixed permanently on the patent document. The current owner is tracked separately, in the USPTO’s Assignment Search database, and only reflects reality if every ownership change since filing was actually recorded there [12] [19].

Does licensing a patent change who owns it?
No. A license, even an exclusive, worldwide, decades-long license, grants rights to use or commercialize the invention. It does not transfer legal title, which requires a separate written assignment under 35 U.S.C. 261. Xtandi’s patents remain owned by the Regents of the University of California despite Medivation and then Pfizer holding an exclusive license since 2005 [12] [1].

Why do Xtandi’s patents still list the University of California as the owner?
UCLA structured its 2005 license with Medivation to keep patent title with the university while giving Medivation the right to pursue and enforce patent protection in the university’s name. That structure did not change when Pfizer later acquired Medivation, because Pfizer acquired Medivation’s license rights, not UCLA’s underlying patent ownership [2] [6].

What happens to a patent’s recorded ownership when the assignee company gets acquired?
Legal title generally transfers with the acquisition, but only if the acquiring party files the corresponding assignment paperwork with the USPTO. Until that filing happens, the patent’s public record can still show the acquired company, or a subsidiary entity created for the transaction, as the owner. Sofosbuvir’s patents, for example, are recorded under Gilead Pharmasset LLC rather than Gilead Sciences [9].

Can a company sue for patent infringement if it doesn’t yet legally own the patent?
No. Federal Circuit law requires a plaintiff to hold enforceable legal title to the asserted patent on the date it files suit. A later assignment, even one drafted to apply retroactively, cannot cure a standing defect that existed when the complaint was filed, as the Federal Circuit held in dismissing Abraxis Bioscience’s suit against Navinta [30] [31].

Does the USPTO verify who really owns a patent before recording an assignment?
No. Recordation is explicitly described in federal regulation as a ministerial act. The USPTO records whatever documentation a party submits without independently confirming the transaction’s validity or the submitting party’s actual right to make the transfer [16] [17].

Who is responsible for listing a patent’s current owner with the FDA’s Orange Book?
The NDA holder submits Form FDA 3542 to FDA, though a separate patent owner can complete the form for the NDA holder to file. FDA does not independently verify that the party identified as the patent owner actually holds clean legal title [33] [34].

What is a nunc pro tunc assignment, and why did it fail in Abraxis v. Navinta?
A nunc pro tunc assignment is a document executed at one point in time but written to take legal effect as of an earlier date. The Federal Circuit held that this kind of backdated fix cannot retroactively create the legal title a plaintiff needed to have already held on the date it filed suit, because Article III standing is assessed as of the filing date itself [30] [31].

How do Bayh-Dole march-in rights depend on who holds legal title to a patent?
March-in rights apply to inventions developed with federal research funding, and the mechanism runs through the funded institution that holds patent title, not through any company that later licenses or markets the resulting drug. Xtandi’s march-in petition targeted rights tied to UCLA’s federally funded original research, a lever that would not exist if UCLA had sold, rather than licensed, the patents to Medivation [21] [22].

How can I check who currently owns a specific drug patent?
Identify the patent number from the drug’s Orange Book listing, then search that number directly in the USPTO’s Assignment Search database, the consolidated tool that replaced the agency’s legacy assignment-search systems in September 2025. Cross-check the result against any parties named in related patent litigation, since litigation captions require accurate standing disclosures that marketing materials do not [19] [34].

References

  1. Knowledge Ecology International. (2022, December 2). Xtandi (INN enzalutamide) timeline. https://www.keionline.org/xtandi-timeline
  2. University of California, Office of the President. (2018, August 14). Memorandum to members of the Health Services Committee regarding Xtandi. https://regents.universityofcalifornia.edu/regmeet/aug18/h4.pdf
  3. 37 C.F.R. § 3.54.
  4. 21 C.F.R. § 314.53.
  5. Abraxis Bioscience, Inc. v. Navinta LLC, 625 F.3d 1359 (Fed. Cir. 2010).
  6. Zaveri, M. (2024, April 17). Universities are profiting from blocking drug-price reform. Jacobin. https://jacobin.com/2024/04/ucla-profit-drug-price-big-pharma
  7. LAmag. (2025, August 21). UCLA benefits handsomely from Xtandi cancer drug royalties amid controversy over drug pricing. https://lamag.com/health/ucla-benefits-handsomely-from-xtandi-cancer-drug-royalties-amid-controversy-over-drug-pricing/
  8. Bloomberg Law. (2022, December 19). Astellas, Pfizer sue Sun to block copies of Xtandi drug. https://news.bloomberglaw.com/ip-law/astellas-pfizer-sue-sun-to-block-copies-of-xtandi-cancer-drug
  9. Justia Patents. Patents assigned to Gilead Pharmasset LLC. https://patents.justia.com/assignee/gilead-pharmasset-llc
  10. Chaifetz, S., Chokshi, D. A., Rajkumar, R., Scales, D., & Benkler, Y. (2007). Closing the access gap for health innovations: An open licensing proposal for universities. Globalization and Health, 3(1). https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1797018/
  11. Genetic Engineering & Biotechnology News. (2023, June 5). Citing “strong sales,” pension manager buys partial Keytruda rights for $1.3B. https://www.genengnews.com/news/citing-strong-sales-pension-manager-buys-partial-keytruda-rights-for-1-3b/
  12. Board of Trustees of the Leland Stanford Junior University v. Roche Molecular Systems, Inc., 563 U.S. 776 (2011).
  13. 35 U.S.C. § 261.
  14. Board of Trustees of the Leland Stanford Junior University v. Roche Molecular Systems, Inc., 563 U.S. 776, 780 (2011) (Roberts, C.J.). https://www.law.cornell.edu/supct/html/09-1159.ZO.html
  15. U.S. Patent and Trademark Office. MPEP § 301: Ownership/assignability of patents and applications. https://www.uspto.gov/web/offices/pac/mpep/s301.html
  16. Ehrenfried, M. L. (2010, December). Federal Circuit law governs patent assignment interpretation and nunc pro tunc assignments cannot retroactively confer standing. Finnegan. https://www.finnegan.com/files/Upload/FCN_Dec10_5.html
  17. Harvard Business School, Baker Library. Patent search: US patent ownership. https://www.library.hbs.edu/services/help-center/patent-search-us-patent-ownership
  18. Ropes & Gray LLP. Best practices for diligencing chain-of-title for post-AIA patent applications with assignee-applicants. https://www.jdsupra.com/topics/assignments/patent-ownership/patent-applications
  19. U.S. Patent and Trademark Office. (2025, August 27). USPTO enhances assignment records search process. https://www.uspto.gov/subscription-center/2025/uspto-enhances-assignment-records-search-process
  20. U.S. Patent and Trademark Office. Patents assignments: Change & search ownership. https://www.uspto.gov/patents/maintain/patents-assignments-change-search-ownership
  21. 35 U.S.C. §§ 200-212 (Bayh-Dole Act).
  22. Love, J. (2023, March 28). How soon could President Biden enable generic competition to Xtandi? Very quickly, if there is the will. Petrie-Flom Center, Harvard Law School. https://petrieflom.law.harvard.edu/2023/03/28/how-soon-could-president-biden-enable-generic-competition-to-xtandi-very-quickly-if-there-is-the-will
  23. Diament, N. (2023, June 23). The obscure NIH official blocking lower drug prices. The American Prospect. https://prospect.org/health/2023-06-23-nih-official-blocking-lower-drug-prices/
  24. Labiotech. (2022, June 24). Meet the Dutch scientists who invented Keytruda, “the president’s drug.” https://www.labiotech.eu/interview/interview-keytruda-cancer-inventors/
  25. Sachs, J. D. (2020, July 7). The drug that is bankrupting America. https://www.jeffsachs.org/newspaper-articles/wy5hjz9tsj7syast4g7r3ja5pxn7fp
  26. BioPharma Dive. (2016, March 10). Gilead and Merck lock horns in court over sofosbuvir patent. https://www.biopharmadive.com/news/gilead-and-merck-lock-horns-in-court-over-sofosbuvir-patent/415408/
  27. Gilead Sciences, Inc. (2002, December 4). Gilead Sciences to acquire Triangle Pharmaceuticals for $464 million [Press release]. U.S. Securities and Exchange Commission EDGAR. https://www.sec.gov/Archives/edgar/data/1022622/000104746902007125/a2096276zex-99_a7.txt
  28. Studicata. Abraxis Bioscience, Inc. v. Navinta LLC – Case brief summary. https://www.studicata.com/case-briefs/case/abraxis-bioscience-inc-v-navinta-llc
  29. Lexology. (2010, December 31). Federal Circuit law governs patent assignment interpretation and nunc pro tunc assignments cannot retroactively confer standing. https://www.lexology.com/library/detail.aspx?g=db40d821-0257-44c2-a269-973186819804
  30. CourtListener. Abraxis Bioscience, Inc. v. Navinta LLC, 625 F.3d 1359. Free Law Project. https://www.courtlistener.com/opinion/179002/abraxis-bioscience-inc-v-navinta-llc/
  31. Harvard Journal of Law & Technology Digest. (2011, March 28). Abraxis Bioscience, Inc. v. Navinta LLC: Federal Circuit rules federal law trumps state law in interpretation of patent ownership rights. https://jolt.law.harvard.edu/digest/abraxis-bioscience-inc-v-navinta-llc
  32. U.S. Food and Drug Administration. (2019, May 20). Patent submission and listing requirements for new drug applications. Federal Register, 84(97), 22860. https://www.govinfo.gov/link/fr/84/22860
  33. U.S. Food and Drug Administration. What’s new with Forms FDA 3542 and 3542a [Presentation]. Office of Generic Drugs. https://www.fda.gov/media/132129/download
  34. 21 C.F.R. § 314.53; 21 C.F.R. § 314.95.

Make Better Decisions with DrugPatentWatch

» Start Your Free Trial Today «

Copyright © DrugPatentWatch. Originally published at
DrugPatentWatch - Transform Data into Market Domination