Context Is Paramount: Ex parte Chowdhury and the Proper Use of Markush Groups

Executive Summary: Ex parte Chowdhury confirms that whether alternatives form a proper Markush group depends on their technical unity in the context of the claimed invention, rather than on whether they are identical in every structural or functional respect. Context determines the relevant level of analysis, but a shared result stated at a high level of generality cannot alone transform an arbitrary list into a proper group. This article situates Chowdhury within the predecessor-court decisions addressing structural similarity, common use, and substitutability, while acknowledging that improper-grouping doctrine lacks an express statutory foundation and that the Federal Circuit has not comprehensively tested the Office’s contemporary MPEP framework. It also examines Federal Circuit authority concerning the practical consequences of Markush drafting, including closed-group construction, selection of multiple listed members, additional unlisted components, anticipation by one listed alternative, indefiniteness, and prosecution-history estoppel. Finally, the article explains that proper grouping does not establish written description or enablement under § 112(a) and offers practical guidance for defining technical unity, stating whether mixtures are intended, evaluating every listed member against the prior art, and preserving commercially meaningful claim scope.

Introduction

 

Markush claiming compresses a potentially unwieldy set of alternatives into a single limitation. The familiar formulation - "a member selected from the group consisting of A, B, and C" - appears most often in chemical, pharmaceutical, biotechnology, and materials claims, but its utility is not confined to those fields. Any claim that permits one of several defined components to perform a particular role can present the same drafting problem.

That convenience carries doctrinal consequences. The alternatives must possess sufficient unity to form a proper group. The language used to introduce the group may close the limitation to unlisted alternatives. Unless the claim or intrinsic evidence indicates otherwise, the group may exclude mixtures of listed members. And, for validity, a prior-art disclosure of just one member ordinarily satisfies the entire alternative limitation.

In Ex parte Chowdhury, Appeal No. 2025-002261 (P.T.A.B. Feb. 5, 2026) (informative), the Patent Trial and Appeal Board addressed the first of those questions: when do technically different alternatives constitute a proper Markush group? The Board's answer was emphatically contextual. The listed microRNAs did not need to perform identical biological functions. They needed to be substitutable for the function they performed in the claimed treatment methods - serving as quantifiable markers of radiation exposure and treatment efficacy.

The USPTO designated the decision informative on August 25, 2026, following a stakeholder nomination. Although informative rather than precedential, the designation makes the opinion particularly useful as examination guidance: the Office summarized the decision as holding that a group is proper when its members form a subgenus and the specification describes them as performing similar functions in the context of the invention. See USPTO, Following Stakeholder Nomination, USPTO Designates as Informative an Appeal Decision Addressing Markush Groupings (Aug. 25, 2026).

The decision is significant, but it should not be read in isolation. Chowdhury draws on binding holdings of the Court of Customs and Patent Appeals, particularly In re Harnisch and In re Jones. Modern Federal Circuit decisions address what follows after an applicant chooses Markush form: closure, mixtures, additional ingredients, anticipation, indefiniteness, infringement, and prosecution-history estoppel. Read together, these authorities offer a coherent lesson. A sound Markush group requires technical unity defined at the level of the claimed invention, while sound Markush drafting requires precision about exactly what the group includes and excludes.

The Threshold Question: Is the Group Proper?

 

A conventional Markush claim recites a list of alternatives, commonly introduced by "selected from the group consisting of." MPEP § 2117 explains that an improper grouping may exist when the members either lack a single structural similarity or do not share a common use. When a group appears as part of a combination or process, the alternatives must be substitutable for one another with the expectation that the same intended result will be achieved.

The last part of the test is easy to state but can be difficult to apply. At what level should the common result be defined? Two compounds may behave differently in many respects while acting interchangeably as solvents in a claimed reaction. Two biomarkers may regulate different biological pathways while serving interchangeably as indicators of the same condition. Two mechanical fasteners may have little structural resemblance while each performing the claimed fastening function.

The answer cannot be obtained by comparing the alternatives in the abstract. It depends on the role the alternatives play in the claim. As the Federal Circuit observed in Multilayer Stretch Cling Film Holdings, Inc. v. Berry Plastics Corp., the species in a Markush group are generally understood to be "alternatively usable for the purposes of the invention." 831 F.3d 1350, 1357 (Fed. Cir. 2016). The relevant comparison is therefore tied to the claimed purpose, not every characteristic the alternatives may possess.

Context determines the relevant level at which unity is evaluated; it does not dispense with the need for a technically reasoned class. A shared desired outcome, stated at a sufficiently high level of generality, cannot alone convert an arbitrary enumeration into a proper Markush group.

Ex parte Chowdhury: Different miRNAs, One Claimed Function

 

The application in Chowdhury concerned methods for assessing and treating radiation-induced damage. The specification explained that exposure to total-body irradiation produces dose-dependent changes in the serum levels of particular microRNAs, or miRNAs. Those changes correlate with radiation exposure, prognosis, and the efficacy of treatments intended to reduce radiation-induced injury.

Representative claim 60 required measuring a first serum level of one or more miRNAs selected from a closed list, administering a treatment, measuring the miRNAs again, and determining from an increase or decrease in specified miRNA levels whether the treatment was effective. Depending on that determination, the method required either administering additional doses of the first treatment or administering a different treatment. Ex parte Chowdhury, Appeal No. 2025-002261, at 2-4.

The examiner rejected the claims for improper Markush grouping. Each miRNA had a different nucleotide sequence, the examiner reasoned, and the fact that all miRNAs comprise nucleotides did not establish the required structural similarity. The examiner also concluded that the alternatives did not constitute a recognized chemical or art-recognized class because the prior art did not show that the miRNAs behaved in the same manner or could be substituted with the same intended result. Id. at 6-7.

The Board reversed. Its central observation was that the examiner had evaluated the wrong function. The claims were not directed to the miRNAs as compounds in isolation, nor did the claimed methods depend on each miRNA carrying out the same regulatory role. Within the claims, the miRNAs operated as markers. Their relevant common property was that exposure to radiation caused a measurable change in their serum levels and that the change could be used to assess radiation exposure or treatment efficacy.

The specification placed the alternatives within that common functional class. It identified the miRNAs as responsive to radiation exposure, described their quantification, and associated increases or decreases in their levels with the claimed treatment determination. The examiner likewise acknowledged that the claimed miRNAs were correlated with radiation exposure and that miRNAs are transcribed and processed in the same way. On that record, the Board found the alternatives sufficiently unified and substitutable for the claimed purpose. Id. at 8-10.

The Board illustrated the point with the MPEP's disposable-diaper example. Pressure-sensitive adhesive, hook-and-loop material, a snap, and a buckle are structurally divergent, but each may belong to an art-recognized class in a claim requiring a repositionable and refastenable fastener. What unifies the alternatives is not a close resemblance in every physical feature. It is the expectation that each will perform the same role in the claimed combination.

The analogy exposes the flaw in the examiner's reasoning. Studying the precise biological function of each miRNA might matter to a claim directed to gene regulation. It did not resolve whether the miRNAs were interchangeable as measurable radiation-response markers. Because individual miRNA activity was not the mechanism on which the claimed treatment method depended, differences in that activity did not defeat the group.

The Board did not abolish the requirements of structural similarity and common use. It instead required those requirements to be applied at the correct level. The miRNAs remained members of the same general molecular class, shared basic structural and processing characteristics, and were described as a subgenus defined by their radiation-responsive behavior. Their different nucleotide sequences did not outweigh that unity in the context of the claims.

The Examiner's Initial Burden Still Matters

 

The decision also reinforces the examiner's burden. Citing In re Oetiker, the Board reiterated that the Office bears the initial burden of presenting a prima facie case of unpatentability. 977 F.2d 1443, 1445 (Fed. Cir. 1992). An improper-grouping rejection therefore should do more than identify differences among the alternatives. It should explain why the asserted differences prevent the members from sharing a qualifying structure, common use, or expected substitutability in the claimed relationship.

For applicants, that allocation of the burden is important but not a reason to leave the technical record undeveloped. When substitutability is not apparent, the specification should identify the common function and the feature or art-recognized classification that links the alternatives. Technical literature, comparative data, or a declaration may be appropriate if the examiner disputes that members of the group would be expected to achieve the same claimed result.

The Predecessor-Court Foundation—and Its Limits

 

Although Chowdhury is an informative PTAB decision, its analysis draws on binding predecessor-court holdings. The Federal Circuit adopted holdings of the Court of Customs and Patent Appeals as binding precedent in South Corp. v. United States, 690 F.2d 1368, 1369 (Fed. Cir. 1982) (en banc).

The principal authority is In re Harnisch, 631 F.2d 716 (C.C.P.A. 1980). The claims in Harnisch recited coumarin compounds useful as dyes, some of which could also be used as intermediates to make other dyes. The court considered the compounds as wholes, found a shared structural relationship relevant to the claimed objective, and concluded that they belonged to a scientifically coherent subgenus. Id. at 722-23. The court described this relationship as a "unity of invention," drawing on In re Jones, 162 F.2d 479, 481-82 (C.C.P.A. 1947), where a shared tetralyl nucleus and common plant-growth-regulating function supported the grouping despite differences among the compounds.

In re Driscoll, 562 F.2d 1245, 1249 (C.C.P.A. 1977), supplies a complementary formulation: members of the group should be alternatively usable for the purposes of the invention. That principle ultimately reappeared in Multilayer Stretch and in the MPEP language applied by the Board in Chowdhury.

These cases do not establish that a common end result alone can unite any arbitrary list. The group still requires a reasoned technical basis. But they reject an approach that treats every difference among the alternatives as disqualifying. The question is whether the members form a coherent class in the claimed relationship. Chowdhury applies that principle to modern biotechnology: miRNAs with different sequences and different broader biological functions can still constitute a proper class when each is used as a radiation-responsive marker.

Two other CCPA decisions help define the procedural boundaries. In re Ruff, 256 F.2d 590, 598-99 (C.C.P.A. 1958), distinguished actual equivalence from prior-art equivalence. Evidence that alternatives are interchangeable may support an applicant's grouping, but it does not automatically prove that the prior art taught the equivalence for purposes of rejecting the claim. And In re Weber, 580 F.2d 455 (C.C.P.A. 1978), together with In re Haas, 580 F.2d 461 (C.C.P.A. 1978), rejected the use of claim rejection as a substitute for restriction practice merely because a single claim might encompass multiple inventions.

A doctrinal caveat is warranted. Harnisch rejected the notion of a single, freestanding "Markush doctrine" and described an evolving, not altogether consistent body of case law. It nevertheless held that the Board could derive improper-grouping rules from the cases even though that ground "does not have a specific statutory basis." 631 F.2d at 719-21. Weber held—and Haas applied—that § 121 cannot itself supply a basis for rejecting a single claim merely because it embraces multiple inventions. Weber expressly reserved and remanded a distinct case-law-based improper-grouping ground; Haas noted that no such separate ground was before it. Weber, 580 F.2d at 458-59; Haas, 580 F.2d at 464 & n.6. The Office's current MPEP § 2117 supplies the structural-similarity-and-common-use framework that Chowdhury applied as an appealable merits rejection. Chowdhury does not resolve how the Federal Circuit would address a modern challenge to the precise legal foundation, application, or outer limits of that formulation.

There appears to be no modern precedential Federal Circuit decision squarely reconsidering the precise grouping-propriety issue decided in Harnisch and Chowdhury. Most of the Federal Circuit's Markush decisions begin after that threshold has been crossed. They ask what a Markush group means and what legal consequences follow from its use.

Closure: The Price of "Consisting Of"

 

The starting point for modern claim construction is Abbott Laboratories v. Baxter Pharmaceutical Products, Inc., 334 F.3d 1274 (Fed. Cir. 2003). The claim required an amount of a Lewis-acid inhibitor "selected from the group consisting of" specified compounds. The Federal Circuit treated the conventional Markush language as closed. The recited group excluded unlisted alternatives, and, absent qualifying language, the reference to "a" member selected from the closed group did not encompass a mixture of multiple listed members. Id. at 1280-81.

Abbott offers a direct drafting lesson. If two or more listed members are intended, the claim should say so—for example, "at least one member selected from," "one or more members selected from," or, where blends matter, "and mixtures thereof." Those formulations address the selection or combination of multiple listed members; they do not by themselves permit an unlisted species to serve as the member satisfying the closed limitation. Whether additional unlisted components may coexist depends on the surrounding claim language and the intrinsic record. See Abbott, 334 F.3d at 1280-81; Amgen, 945 F.3d at 1378-80.

The contrast with Gillette Co. v. Energizer Holdings, Inc., 405 F.3d 1367 (Fed. Cir. 2005), is instructive. The razor claim in Gillette referred to a "group of first, second, and third blades" but did not use "consisting of" or otherwise signal a closed Markush group. The court held that the limitation did not exclude a fourth blade. Id. at 1371-73. A list is not closed merely because its items are grouped together; the words introducing the list matter.

The Federal Circuit refined Abbott in Multilayer Stretch. The court separated two related presumptions. First, conventional Markush language produces a very strong presumption that unlisted species are excluded from satisfying the limitation. Overcoming that presumption requires intrinsic evidence that unmistakably manifests a different meaning. Second, the presumption against mixtures of listed species is weaker. Whether a mixture of listed alternatives is permitted may depend on the claim language, specification, and prosecution history. 831 F.3d at 1358-62.

That distinction matters. A claim may remain closed to an unlisted resin while allowing a layer made from a blend of two listed resins. Treating closure and mixtures as the same question can produce either unduly narrow drafting or an unsupported infringement position.

Multilayer Stretch also exposes a dependency trap. A dependent claim attempted to add low-density polyethylene even though that material fell outside the independent claim's closed group. Because a dependent claim must include and further limit every element of the claim from which it depends, the court held the dependent claim invalid under 35 U.S.C. § 112(d). Id. at 1362. Every dependent claim that adds a species to a Markush limitation should therefore be checked against the exact boundaries of the parent group.

Embedded Groups, Mixtures, and Additional Components

 

Claims often combine an open transition for the overall composition with a closed Markush group for one element. Amgen Inc. v. Amneal Pharmaceuticals LLC, 945 F.3d 1368 (Fed. Cir. 2020), explains how those phrases coexist.

The claims in Amgen used "comprising" for the overall pharmaceutical composition but required "at least one binder selected from the group consisting of" listed binders and a similarly framed disintegrant limitation. The Federal Circuit held that the Markush language closed the identity of the ingredient used to satisfy each recited limitation. It did not, however, exclude every additional ingredient from the overall composition. Because the claim as a whole was open, an accused formulation could contain an unlisted ingredient - even another ingredient capable of acting as a binder or disintegrant - so long as a listed ingredient satisfied the recited Markush limitation. Id. at 1376-79.

Amgen does not render the embedded group meaningless. An unlisted material cannot itself satisfy the closed limitation. The decision instead distinguishes between what counts as the claimed element and what else may coexist with that element in an open claim. This is an especially important distinction for pharmaceutical and chemical formulations in which ingredients can serve more than one function.

Shire Development, LLC v. Watson Pharmaceuticals, Inc., 848 F.3d 981 (Fed. Cir. 2017), illustrates the other side of the boundary. The claims recited closed Markush limitations for matrix components. The accused product's outer matrix contained magnesium stearate, which was not listed. The court rejected the argument that specification language permitting optional excipients opened the specific matrix limitation. Because magnesium stearate was structurally and functionally related to the matrix, its presence prevented the accused matrix from literally satisfying the closed limitation. Id. at 984-86.

The result can differ when the additional item is wholly unrelated to the claimed combination. In Norian Corp. v. Stryker Corp., 363 F.3d 1321 (Fed. Cir. 2004), an additional spatula did not avoid infringement of a kit claim using "consisting of" because the spatula did not interact with or alter the claimed chemical components. Id. at 1331-32. Shire treated this as a narrow exception, not a general license to disregard additional matter.

The location of the Markush language also matters. In Abbott Laboratories v. Andrx Pharmaceuticals, Inc., 473 F.3d 1196 (Fed. Cir. 2007), the court cautioned against giving a list in the written description the automatic closing effect of a Markush limitation in a claim. Markush terminology is principally a claim-drafting convention. A specification's discussion of possible alternatives may inform claim meaning, but the mere presence of a list does not necessarily disclaim everything unlisted. Id. at 1210.

Validity: One Member May Be Enough

 

Closure can benefit a patentee by defining a finite set of alternatives, but the alternative structure has a significant validity consequence. If a claim permits A, B, or C, a prior-art disclosure of A ordinarily satisfies that limitation. The reference need not disclose all three alternatives.

Fresenius USA, Inc. v. Baxter International, Inc., 582 F.3d 1288 (Fed. Cir. 2009), is the principal published Federal Circuit authority. The disputed limitation required at least one unit selected from a list of alternatives. Because the prior art disclosed one of the listed units, the Markush limitation was met for anticipation. Id. at 1298.

The same rule operates through inherency. In Schering Corp. v. Geneva Pharmaceuticals, Inc., 339 F.3d 1373 (Fed. Cir. 2003), the claimed compounds were presented in Markush form. One listed compound was necessarily produced in the body as a metabolite of a prior-art drug. The inherent production of that one member anticipated the claim. Id. at 1380.

The Federal Circuit recently applied the same principle in Board of Regents of the University of Texas System v. Boston Scientific Corp., Nos. 2024-2062, 2024-2063, slip op. at 12-13 (Fed. Cir. July 27, 2026). The court applied Fresenius separately to two dependent claims: the prior art's disclosure of "drugs" satisfied the therapeutic-agent alternatives recited in claim 11, and its disclosure of polyglycolic acid satisfied the polymer alternatives further enumerated in claim 17.

The rule is limitation-specific. Disclosure of one listed member satisfies the alternative limitation, but the same anticipatory reference must still disclose, expressly or inherently, every other limitation of the claim.

This doctrine creates an asymmetric drafting risk. Adding alternatives can broaden literal scope, but every added member supplies another possible route to anticipation. A prior-art search and patentability analysis should therefore examine every member of a proposed group, not merely the preferred or commercially important species. If a problematic species is unnecessary to the commercial embodiment, its inclusion may provide little enforcement value while threatening the entire claim.

Indefiniteness: Overlap, Narrowing, and Open-Ended Groups

 

Markush drafting can also generate indefiniteness disputes, but overlap among members is not automatically fatal. In Eli Lilly & Co. v. Teva Parenteral Medicines, Inc., 845 F.3d 1357 (Fed. Cir. 2017), the relevant group included both vitamin B12 and cyanocobalamin, terms the record supported as synonymous. The Federal Circuit declined to manufacture uncertainty from that redundancy. The overlap did not prevent a skilled artisan from understanding the claim's scope. Id. at 1371.

Nor does a later limitation necessarily contradict an earlier group. In Maxell, Ltd. v. Amperex Technology Ltd., 94 F.4th 1369 (Fed. Cir. 2024), a claim first required at least one transition metal selected from cobalt, nickel, and manganese and later required cobalt to fall within a specified mole-percentage range. The district court treated the provisions as contradictory options. The Federal Circuit reversed, explaining that claim limitations operate cumulatively. The later requirement narrowed the universe permitted by the earlier Markush clause; it did not conflict with it. Id. at 1372-75.

Open-ended Markush formulations present a different concern. In the nonprecedential decision In re Kiely, No. 2022-1076, 2022 USPQ2d 532, slip op. at 5 (Fed. Cir. June 8, 2022), the court affirmed an indefiniteness rejection of language reciting a selection from "the group comprising" identified alternatives. Given both the breadth of variation among the specified alternatives and the use of "comprising" to define the list, the formulation failed to make clear what unlisted alternatives were encompassed and did not provide the required certainty on the record presented.

An earlier nonprecedential decision, Lexington Luminance LLC v. Amazon.com, Inc., No. 2014-1384, slip op. at 8-10 (Fed. Cir. Feb. 9, 2015), held a claim using a similar phrase not indefinite because, in context, the intrinsic record made the scope reasonably certain. Kiely arose on appellate review of an indefiniteness rejection during prosecution; Lexington Luminance reviewed a district-court judgment on the pleadings and expressly noted the issued patent's presumption of validity and the challenger's burden. The decisions are not necessarily irreconcilable. They arose from different procedural postures and different intrinsic records, and together illustrate that an open-ended alternative formulation will be evaluated in the context of the disclosure and the certainty it supplies. But Kiely, which the MPEP now cites, confirms the prosecution risk. If the drafter intends a closed group, "selected from the group consisting of" states that intention more clearly than "group comprising."

Prosecution History Can Make Closure Permanent

 

Markush amendments and arguments may narrow more than the literal words of the issued claim. In Merck & Co. v. Mylan Pharmaceuticals, Inc., 190 F.3d 1335 (Fed. Cir. 1999), the applicant narrowed a broader polymer genus to a particular species after a prior-art rejection and election. The court held that the patentee could not later use the doctrine of equivalents to recapture the surrendered polymer combination. Id. at 1339-41. Although prosecution-history estoppel must now be analyzed under the framework established in Festo, Merck remains a useful warning about the substantive surrender created by narrowing a Markush genus.

Norian Corp. v. Stryker Corp., 432 F.3d 1356 (Fed. Cir. 2005), provides a related—although not conventional Markush—example. The claim recited "a solution consisting of water and a sodium phosphate." In context, the court construed "a sodium phosphate" to require one type of sodium phosphate rather than a mixture of different sodium phosphates. The closed transition, the specification, and the prosecution history supported that construction. Because the limiting language resulted from a narrowing amendment, prosecution-history estoppel also barred Norian's doctrine-of-equivalents theory. Id. at 1357, 1359-63.

The same caution extends to arguments about absent ingredients. In the related closure decision Azurity Pharmaceuticals, Inc. v. Alkem Laboratories Ltd., 133 F.4th 1359 (Fed. Cir. 2025), the patentee had adopted whole-claim "consisting of" language and relied on the absence of propylene glycol to distinguish the prior art. Those choices supported a construction excluding an accused formulation containing propylene glycol. Although Azurity is not principally a Markush-group case, it demonstrates how closed language and prosecution argument can combine to produce a lasting disclaimer.

Applicants should therefore avoid describing an alternative, mixture, or additional ingredient as categorically excluded unless that exclusion is needed for patentability and commercially acceptable. A narrow amendment may secure allowance while simultaneously creating an infringement gap that cannot be repaired through claim construction or the doctrine of equivalents.

The Broader Significance of Chowdhury

 

The immediate importance of Chowdhury lies in biotechnology and diagnostics, where the most useful claimed class may be defined by a shared evidentiary role rather than identical biological activity. Biomarkers can be associated with different pathways and still be interchangeable inputs to a claimed diagnostic or treatment decision. Nucleic-acid sequences can differ while belonging to a coherent response signature. Binding agents can possess different detailed structures while serving the same claimed detection function.

The principle can extend beyond the life sciences. Sensor modalities may differ physically while each producing the measurement required by a control method. Machine-learning features or classifiers may differ computationally while each supplying the same claimed decision input. Alternative materials may possess different compositions while satisfying the same recited mechanical or thermal criterion. In each setting, however, the specification must do the work of defining the class and connecting its members to the claimed result.

That is the best reading of Chowdhury. The decision does not replace technical unity with functional aspiration. It requires the Office to evaluate unity at the level at which the invention uses the alternatives. The claimed miRNAs were not grouped because all miRNAs are broadly similar or because the applicant labeled them as alternatives. They were grouped because the disclosure identified a particular radiation-responsive subgenus, explained how its members were quantified, and used each member in the same treatment-evaluation relationship.

Conclusion

 

Markush practice presents two related inquiries. The first is whether the alternatives form a proper group. Ex parte Chowdhury, read with Harnisch, Jones, and Driscoll, teaches that structural similarity, common use, and substitutability must be evaluated in the claimed context. Differences irrelevant to the alternatives' claimed role should not obscure their technical unity.

The second is what follows from Markush form. Abbott, Multilayer Stretch, Amgen, Shire, Fresenius, Eli Lilly, and Maxell show how a few words can determine whether a claim excludes unlisted species, permits mixtures or additional components, survives prior art, or reaches an accused product. Prosecution amendments and arguments may make those consequences permanent.

The practical lesson is straightforward: define the alternatives' unity in the specification; separately confirm written-description and enablement support for the full group; state the intended claim closure; and test each member against the prior art and commercial embodiment. Markush language is compact, but its legal effects are not.