India’s pharmaceutical patent regime has always sought to balance two competing objectives: encouraging genuine innovation and ensuring access to medicines. The Draft Guidelines for Examination of Patent Applications in the Field of Pharmaceuticals, 2026, issued by the Office of the Controller General of Patents, Designs and Trade Marks, seek to strengthen that balance through a more detailed and technically informed examination framework.
From 2014 to 2026: What the Draft Changes
The shift is less about new law than about specificity: the 2026 Draft tells an examiner what evidence to look for, not merely what the statute says.
| Aspect | 2014 | 2026 Draft |
| Structure | provision-led, claim types in Chapter 4 | Claim-type led: separate treatment of NCEs, formulations, combinations, salts, polymorphs, product-by-process claims and more with Annexure-I |
| Novelty | Anticipation principles, with a carve-out permitting combination of prior art for Markush claims | Seven-stage “Seven Stambhas Approach”; the carve-out deleted, mosaicing barred outright |
| Inventive step | A five-step objective method | Incorporates judicial reasoning on motivation to combine, “obvious to try,” and a caution against rigid formulaic tests |
| Markush | Checklist requiring all embodiments and a test for each | Same checklist, relaxed to best representatives; adds a process-enablement requirement |
| 3(d) | States the efficacy requirement | Distinguishes efficacy from usefulness, addresses bioavailability specifically, confirms incremental innovation is not barred outright |
| 3(e) | Aggregation exclusion | Illustrated with a data-driven test for synergy versus aggregation |
| 3(i) | Very narrow interpretation | Broad interpretation, backed by various case laws |
Markush Claims: Scope Tied to Disclosure
A Markush claim defines a chemical genus through a core scaffold with variable substituents, and the combinatorics are extreme: five positions with twenty options each yield over three million compounds, of which an applicant may have made a few dozen. The Draft addresses unity and sufficiency into one inquiry. It requires that all alternatives share a common property or activity and that either a significant structural element is common to all, or they belong to a class recognized in the art. It adds a definition absent from 2014: a common structure exists where compounds share the framework of the Markush formula, or where the shared feature is structurally distinctive against the prior art.
Where 2014 required disclosure of “all the possible embodiments” and a “test conducted for each embodiment”, 2026 asks for the “best representatives, as known to the applicant”. The sufficiency chapter shifts the same way, from data on “each compound claimed” to “representatives of each embodiment”. What 2026 adds in exchange is a requirement that at least one process enabling the whole claimed scope be disclosed.
For drafters, the message is that a genus claim’s outer boundary should track the boundary of what was actually made and rationalized, rather than the breadth the language of the claim happens to permit. Illustration Pharma-IE-19 makes the point that a claim defining R1 across six options but exemplified only where R1 is phenyl draws objections for both lack of support and insufficiency.
Novelty: The Seven Stambhas Approach
The Draft adopts the seven-stage framework from Telefonaktiebolaget LM Ericsson v Lava International Ltd, CS(COMM) 65/2016, decided 28 March 2024. The stages run from understanding the claims, through identifying and analysing prior art, separating explicit from implicit disclosure and assessing material differences across the whole claim scope, to verifying novelty against the specific combination of claimed elements and finally to documenting the analysis with a reasoned finding citing the specific prior-art passages relied on. That seventh stage converts novelty reasoning from a conclusion into an auditable chain, and may do more for the quality of examination reports than any substantive rule in the document.
The 2014 edition said combination was “generally not permitted”, then carved out an exception for Markush claims, where “combination of prior arts vis-à-vis novelty is a logical step taken by an examiner.” The 2026 Draft deletes that carve-out and bars mosaicing without qualification. The full combination of claimed features must be found in a single disclosure, expressly or by necessary implication. The Draft also identifies the asymmetry between generic and specific disclosure: a copper spring anticipates a metal spring, but not the reverse.
Inventive Step: Structure Without Rigidity
With respect to inventive step, the draft discusses both the four-step test laid down in Lava International v. Ericsson and the five-step framework adopted by the Division Bench of the Delhi High Court in F. Hoffmann-La Roche Ltd. v. Cipla Ltd. (2015). At the same time, it appropriately cautions, in light of the Delhi High Court’s 2026 decision in Sulzer Mixpac AG v. Assistant Controller of Patents, that such structured tests “cannot be regarded as commandments cast in stone.” The draft further incorporates the prohibition against ex-post facto analysis recognized in Avery Dennison v. Controller of Patents, as well as the September 2025 decision in Saint Gobain Glass France v. Assistant Controller of Patents, which clarifies that, unlike in the context of novelty, multiple prior-art references may be considered together for assessing inventive step, provided that the analysis is not driven by hindsight.
For pharmaceutical claims specifically, structural or functional similarity to a prior-art lead compound may itself supply a motivation to combine teachings, while a genuinely surprising result, a synergistic effect, or documented prejudice in the art against the claimed approach can support inventive step. On “obvious to try,” choosing one option from a small, predictable set of alternatives remains obvious even if the choice turns out to work unusually well, a good outcome does not retroactively make a routine choice inventive.
Section 3(d): Efficacy Versus Advantage
The Draft rests on the Supreme Court’s 2013 decision in Novartis AG v Union of India (AIR 2013 SC 1311), which arose from the rejection of a patent for the beta-crystalline form of imatinib mesylate (Glivec). The Court drew two lines that the Draft now operationalizes for examiners: efficacy is not the same as usefulness, and for a medicine, efficacy means therapeutic efficacy a demonstrable improvement in clinical effect, not in how easily the compound is made, formulated or stored. Further, properties inherent to a different form of compound, such as hygroscopicity to a polymorph, does not satisfy Section 3(d). Further, increased bioavailability must be linked to therapeutic efficacy by research data. At the same time, the Draft is explicit that incremental innovation is not barred as such: salts, polymorphs, particle-size variants and reformulations remain patentable where genuine, evidenced therapeutic improvement is shown.
Sections 3(e)
The Draft states a functional-interaction test and states that combination claims asserting synergy to be supported by comparative data against the individual ingredients administered separately. More significantly, Novozymes v Assistant Controller (Madras HC, 20 September 2023) holds that 3(e) is not confined to compositions of known ingredients, nor to independent claims. Its object is to exclude any composition claim for a substance that merely exhibits the aggregate properties of its constituents. Combination products built around a novel entity, and dependent claims are now exposed.
Sections 3(i)
The draft’s discussion of Section 3(i) is substantially based on the Madras High Court’s 2023 decision in The Chinese University of Hong Kong v. Assistant Controller of Patents. The decision draws an important distinction between diagnostic processes performed on the human or animal body, which may fall within the exclusion under Section 3(i), and diagnostic products, kits and instruments, which may remain patentable subject to the other requirements of patentability. The decision also clarifies that a screening test capable of identifying a disease or medical condition may constitute a diagnostic process irrespective of whether the individual being tested is symptomatic or asymptomatic. The Delhi High Court’s consolidated summary in Sequenom Inc. & Anr. v. The Controller of Patents, EMD Millipore Corporation v. Assistant Controller of Patents and Designs, and Natera Inc. & Anr. vs. Assistant Controller of Patents & Designs adds that products, kits and devices remain patentable, that negative diagnosis is caught, that software-only diagnostic tools must additionally clear Section 3(k), and that dosage regimens are excluded. The Draft also warns that a claim nominally directed to a combination falls under 3(i) where the invention in substance resides in sequential administration.
Critical Analysis
The Draft’s real contribution is consolidation rather than innovation: it takes principles that were scattered across the statute, Patent Office practice and a decade of case law, and states them in one place in language examiners can apply directly. The Markush treatment is the strongest example, tying claimed breadth to demonstrated disclosure protects the basic patent bargain and gives Sections 10(4) and 10(5) practical bite. The staged approaches to novelty and inventive step should make examination reports more consistent and more transparent, though the Delhi High Court’s warning in Sulzer Mixpac against rigid formulas is a useful corrective: these structures are aids to reasoning, not substitutes for judgment on the actual facts of an invention.
Section 3(d) remains the most delicate provision to apply well, and the nineteen calibrated examples help ensure “new form” does not collapse automatically into “unpatentable”. The Draft is right to insist on therapeutic efficacy rather than mere advantage, but examiners will need to resist collapsing “incremental” into “trivial”: improvements in stability, delivery, dosing and formulation can carry genuine clinical weight even when they do not fit a simple efficacy narrative, and the Supreme Court’s own instruction is that Section 3(d) does not exclude every incremental invention. A parallel risk runs through the Draft’s broader emphasis on experimental data: justified for claims of real breadth, but liable to become an excessive burden if examiners expect exhaustive proof for every member of a scientifically coherent genus rather than a representative, rationally extended sample, a standard that could deter legitimate research as easily as it screens out weak patents.
None of this changes India’s underlying statutory standards; it makes their application more demanding and more predictable. Applicants will need to back broad claims with proportionate disclosure and comparative evidence rather than broader drafting, and examiners have a considerably more detailed roadmap for technically complex pharmaceutical inventions. The outstanding question is one of implementation, not drafting: whether the Guidelines are applied with the technical expertise and case-by-case flexibility they assume, so that genuine incremental innovation is not swept aside along with the weak and repetitive patents the Draft is aimed at.