Function Over Form: The PTAB’s New Stance on Biomarker Patents and Its Hidden Risks

By Anthony D. Marino

The short version. On August 25, 2026, the Patent Trial and Appeal Board (PTAB or Board) took the rare step of designating Ex parte Chowdhury as an “informative decision” on the topic of Markush groupings. The Board held that a list of 15 structurally distinct microRNAs (miRNAs) was a proper Markush group because the miRNAs served a common function in the claimed method, even though they share no meaningful structure. For biotech applicants fighting improper Markush rejections on biomarker panels, this is a significant win at the the United States Patent and Trademark Office (USPTO or Office).

It is also a win with a short leash. The Federal Circuit has never reviewed the rejection the Board was interpreting, and in the doctrines the court has reviewed (written description, enablement, and eligibility), it has been moving in the opposite direction. The claim scope the USPTO is now willing to grant is scope litigators may struggle to defend.

1. Markush Groups and the Biomarker Problem

A Markush group is the workhorse of biotech claim drafting: a closed list of alternatives that lets a single claim cover multiple embodiments without filing a separate application for each. For a diagnostics company with a validated 15-marker panel, the Markush format is often the difference between one patent and 15.

The challenge for applicants is a rejection that the USPTO revived in 2018 after being dormant for 23 years. Now found in MPEP § 2117, the “improper Markush grouping” rejection allows an examiner to reject a claim if the listed alternatives don’t pass a two-part test: all members must share a single structural similarity and a common use. To meet the structural requirement, the members must belong to a recognized class of compounds that are expected to behave similarly for the purposes of the invention.

Note what the rejection is not. It has no statutory basis. It is a doctrine that was judicially approved rather than judicially created, which the Office assembled from CCPA dicta and its own 2011 Supplementary Examination Guidelines. In the eight years since its revival, no federal court has ever invalidated a claim on this ground.

That has not stopped examiners from deploying it aggressively against nucleic acid and polypeptide panels. The logic is seductive, easy to manipulate, and, for years, hard to rebut. Every miRNA and polypeptide has a different sequence, so every miRNA and polypeptide has a different chemical structure. Since these molecules do not share a single structural similarity, they cannot fit into a proper Markush group.

That is precisely the rejection Dana-Farber Cancer Institute and Albert Einstein College of Medicine faced across four years and multiple Office Actions in Application No. 17/005,548.

2. The Decision: Structure Recedes, Function Advances

The claims at issue recite a method of treating radiation-induced damage. The physician measures serum levels of one or more miRNAs selected from a group of 15, administers a countermeasure selected from a second group (a cytokine, potassium iodide, Prussian blue, DTPA, bone marrow transplantation, blood transfusion, or surgery), measures again, and escalates or switches treatment based on whether the marker levels moved.

The examiner conceded the common-use prong (the specification disclosed all 15 miRNAs as correlated with radiation exposure) but held the line on the differing structure of the miRNAs. On February 5, 2026, the PTAB reversed this rejection. The reasoning in the Board decision is worth reading closely, because it relocates the entire analysis and was deemed informative by the Director.

The Board analogized to the Manual of Patent Examining Procedure’s (MPEP) own disposable-diaper example: a claim reciting a fastener selected from a pressure-sensitive adhesive, a hook-and-loop structure, a snap, and a buckle is a proper Markush group. A buckle and a hook-and-loop closure are radically different objects from a structural perspective, but they group properly because they do the same job in the claimed invention.

When applied to miRNAs, that framing dissolves the examiner’s position. According to the panel, the listed miRNA species need not function as miRNAs, in the same way, to the same end, or at all, for the invention to work. The listed miRNA species need only be quantifiable to achieve the invention’s purpose. Structural divergence becomes irrelevant under this line of reasoning. Members that are “structurally different (in that they are composed of different nucleotide sequences and potentially perform differently as biomolecules)” are nonetheless “substitutable for this purpose and are common members of an art-recognized class.”

While the structural prong has not been abolished, it has been redefined in functional terms. Under MPEP § 2117.II.A, an “art-recognized class” is one whose members behave the same way in the context of the claimed invention. If the claimed context is measurement, then anything measurable in the same assay behaves the same way, and the structural inquiry collapses into the functional one. Since an informative designation is not a rule change, applicants can expect to still receive an improper Markush group rejection but now have a much better way to resolve it.

3. The Widening Gap Between What the Agency Grants and What the Court Sustains

The Federal Circuit has never reviewed an improper Markush grouping rejection. Not once since the Office revived it in 2018 has the court addressed this doctrine. The doctrine has developed entirely inside the agency, is reviewed only by the agency, and, because the only appealable outcome is an affirmance; applicants who win at the Board, as Dana-Farber did, have nothing to appeal. Meanwhile, in every adjacent doctrine that does reach the court, the trend runs the other way with regard to functional claiming.

A. Written Description and the Genus Problem

The Federal Circuit is actively cracking down on broad functional genus claims. In Amgen v. Sanofi, the court ruled that claims defined by function are invalid if the patent only describes a fraction of the claimed class. This trend continued in Duke University v. Sandoz (Nov. 18, 2025), where a $39 million verdict was overturned because the patent didn’t sufficiently describe the full scope of the claimed chemical genus. Many cases with similar facts all point to the same conclusion: the court wants to see structural similarities, even as the PTAB now says structure is secondary to function.

The picture is not uniformly adverse, and the nuance is where the value is. In Teva Pharmaceuticals International GmbH v. Eli Lilly & Co. (Apr. 16, 2026), the Federal Circuit upheld method-of-treatment claims reciting a genus of anti-CGRP antibodies, distinguishing Amgen on the ground that a method claim using a genus is not a composition claim to the genus. The § 112 inquiry, the court held, focuses on the claimed use. That is the same analytical move Chowdhury makes, and it is the strongest appellate support the Board’s reasoning has. The emerging line seems to depend on the claim type rather than the technology. Method claims are being given more functional latitude, while composition claims are not.

B. Section 101 and the Board’s Own Self-Inflicted Vulnerability

The Board reversed the improper-Markush rejection and the case moved forward. But what is almost never discussed is that the examiner had originally issued a § 101 rejection on the merits, and that rejection was ultimately withdrawn, meaning the record does not show the PTAB’s reasoning for why the claims survived eligibility scrutiny, only that they did.

What the Board did say about the nature of the invention is deeply problematic for a § 101 defense. The listed miRNA species need not function as miRNAs, in the same way, to the same end, or at all, for the invention to work. The listed miRNA species need only be quantifiable to achieve the invention’s purpose. This framing makes a § 101 challenge almost too easy: the claim is directed to measuring a natural thing and comparing it to a baseline. That is the Ariosa v. Sequenom fact pattern, and Ariosa says those claims are ineligible.

The Board’s holding that miRNAs “need not function as miRNAs … or at all” is a powerful way to defeat an improper-Markush objection, but it can also be seen as a confession that:

This fact pattern collapses back into the Ariosa framing: start with a natural thing (miRNA levels in serum), measure it using well-known techniques (the specification says so), compare to a control, decide on treatment.

Recent cases show courts saying “ineligible” to exactly these facts. In Nextcea Inc. v. Lipotype, Inc. (D. Mass. Feb. 6, 2026), the district court invalidated claims reciting:

The court in Nextcea found the claims directed to a natural phenomenon (the correlation between biomarker isoforms and drug toxicity), using only routine steps (well-known assays), and held that there was no inventive concept beyond the discovery of the natural law. Sound familiar? That is just Chowdhury in a different technological space.

A patent litigator defending the Chowdhury claims would likely face this sequence:

Responses to these arguments are all undercut by the Board’s reasoning:

The irony is that the Board won the Markush battle by conceding the § 101 war. By saying the miRNAs “need not function as miRNAs … or at all,” the Board took the most functional, measurement-focused characterization possible, which is exactly what kills diagnostic method claims under § 101. This is why the § 101 risk and the Markush win are in tension. The Board authorized the Markush grouping by treating the markers as fully interchangeable measurement proxies. To defend against a § 101 challenge, a litigator will need to argue that there is something inventive in how you use the markers, which undermines the interchangeability that permitted the Markush group.

4. Best Practices to Consider: Take the Scope, Hedge the Risk

The right strategy isn’t to simply claim narrowly. Instead, applicants should consider taking the broader claim the Office is willing to grant, while simultaneously building a record designed to survive judicial scrutiny.

A. Specification Drafting for Markush Strength and § 101 Durability

Draft for both in the specification. Chowdhury was won on the specification. In view of this, applicants may want to expressly describe their listed alternatives as a subgenus united by a common function tied to the invention’s purpose, and state that they are substitutable for that purpose. But do not stop there. Also recite whatever structural commonality exists: shared biogenesis or processing pathway, shared size range, membership in a named art-recognized class, shared chemical backbone. While the functional statement may win the Markush fight at the agency, the structural statement is what your § 112 defense will need in five years. They cost the same to include and are not in tension.

Disclose an inventive concept beyond the biomarker discovery itself. This is the critical gap between prosecution and litigation. Do not simply say “we discovered these 15 miRNAs correlate with radiation exposure.” Instead, consider saying:

This specification foundation accomplishes two things: (1) it gives examiners grounds to distinguish § 101 rejections during prosecution, and (2) it builds a record for trial showing the claim is about an inventive method, not just a discovered phenomenon.

B. Claim Drafting: Anchor the Inventive Concept

The Markush group must not be the entire claim. The Board’s reasoning worked here because the independent claim (Claim 60 in Chowdhury) also included inventive steps: determining miRNA levels, administering a treatment selected from a separate Markush group of known agents, re-measuring, and using the comparison to decide on next steps. The Board’s § 101 safe harbor came from the presence of these administering and decision-making steps, not from the markers alone.

Applicants should draft independent claims to emphasize the decision protocol, not just the measurement:

Weak: “A method of treating radiation-induced damage comprising: (a) determining a level of one or more miRNAs selected from the group consisting of [15 listed miRNAs]; (b) administering a treatment; (c) determining a second level of the miRNAs; and (d) determining treatment efficacy based on the change in levels.”

Stronger: “A method of treating radiation-induced damage comprising: (a) measuring baseline levels of one or more miRNAs selected from the group consisting of [15 listed miRNAs] using [specific assay]; (b) determining a radiation-dose estimate from the baseline levels using a [non-routine therapeutic algorithm]; (c) administering a treatment selected based on the dose estimate and the specific miRNA profile; (d) re-measuring miRNA levels after a predetermined interval; (e) calculating a treatment-response score based on the pattern and kinetics of miRNA level changes; and (f) escalating or modifying the treatment if the response score falls below a predetermined threshold.”

The second version claims an inventive decision protocol and specific algorithm, which makes it about the method itself, and goes beyond mere measurement.

C. Litigation-Ready Drafting

The framing that wins a Markush appeal (these molecules are interchangeable; only measurement matters) is the framing an eligibility challenger wants. But you cannot avoid this framing at the USPTO because the Board requires it to sustain the Markush group. What you can do is disable that framing for litigation by disclosing in the specification why this particular combination and protocol is not merely measuring a natural phenomenon. For example, the application in Chowdhury could have included:

While the individual miRNAs listed herein are each natural products of cellular response to radiation, their specific selection and combination in the claimed therapeutic protocol provides technical advantages unknown in the prior art. Specifically: (1) the simultaneous measurement of [these 15 miRNAs in particular] using [specific assay design] provides a more accurate radiation-dose estimate than prior single-marker approaches, such as [cite prior art]; (2) the kinetics of the [specified subset] of miRNA level changes in response to [specific therapeutic intervention] were not predictable and provide early indication of treatment efficacy before clinical symptoms manifest [cite supporting data]; and (3) the algorithm disclosed herein for converting measured miRNA levels into a therapeutic dosing recommendation involves non-routine computational steps and clinical judgment [describe the algorithm in detail].

This language provides a counterpoint to a § 101 challenger arguing that everything is routine and conventional while also giving a trial court language to use when distinguishing Ariosa and Mayo. Without something akin to this, your only defense to a § 101 attack will be procedural or conclusory. With it, you have factual support that can be expanded upon by a litigator.

*** Opinions expressed are those of the author and those quoted, and not necessarily the firm’s or their colleagues’.