The Delhi High Court has cancelled a registered pharmaceutical trademark because the owner could not show that anyone was actually using the stem it relied on. In Dr. Reddy’s Laboratories Limited v. M/s Razenta Pharmaceuticals Private Limited, C.O. (COMM.IPD-TM) 122/2025, decided on 17 August 2026, Justice Jyoti Singh allowed a petition under Section 57 of the Trade Marks Act, 1999 and struck registration No. 5208898 — the mark DAPLOGIN, registered in Class 05 on 14 November 2021 — off the Register. The Registrar of Trade Marks was directed to rectify the Register within six weeks, “for the sake of purity of Register”.
Dr. Reddy’s had used DAPLO since 2020 for a dapagliflozin medicine for Type-2 Diabetes Mellitus. Razenta’s defence was the familiar one in pharmaceutical disputes: DAPLO is not coined at all, it is simply lifted from the molecule, so nobody can monopolise it. Razenta said it had taken ‘DAP’ and ‘LOZIN’ from dapagliflozin, swapped the Z for a G and added ‘GIN’. The court rejected that on both limbs. DAPLO as a whole, it held, is not the prefix, suffix, abbreviation or short name of dapagliflozin — it is a coined combination of ‘DAP’ and ‘LO’, and so neither publici juris nor generic. And DAPLOGIN subsumes all five letters of DAPLO, with ‘GIN’ insufficient to dispel confusion: DAP-LO-GIN is pronounced with DAPLO prominent, and in pharmaceuticals, the judgment notes, prefixes carry great identifying force because recall is largely phonetic. With the same API, the same indication and common trade channels, the registration was found to fall foul of Sections 9(1)(a), 9(2)(a) and 11(1)(b).
The contrast: BEVATAS survived the same doctrine
Read alone, that looks like a hard line against anything resembling a rival pharma mark. It is not. As World Trademark Review reported on 22 October, the opposite result landed in Sun Pharma Laboratories v Intas Pharmaceuticals, the appeal docketed as RFA(OS)(COMM) 10/2026, where Sun Pharma asserted BEVETEX (a paclitaxel product) against Intas’s BEVATAS (bevacizumab). A single judge found for Sun Pharma; a division bench rejected the infringement claim; and according to WTR the Supreme Court declined to interfere on 27 July 2026 in SLP(C) 23851/2026. There, ‘bev’ and ‘beva’ genuinely were common syllables for anti-cancer drugs, the two products differed in composition, therapeutic purpose, dosage and administration, trade dress and labelling differed, and — decisively — the products had coexisted from 2016 to 2026 with no recorded instance of confusion. Phonetic resemblance on its own did not carry the case.
What actually changed: the price of the “everyone uses the molecule” defence
The doctrine of pharmaceutical trademark similarity did not move; the evidentiary burden attached to it did, and that is the part that should change behaviour. Razenta pointed to other registered marks carrying DAP or DAPLO. The judgment treats that as close to worthless, following Pankaj Goel v Dabur India and GlaxoSmithKline v Horizon Bioceuticals: “common to register” is qualitatively different from “common to the trade” under Section 17(2)(b). A party asserting that a stem is common to the trade has to establish substantial use by the proprietors of the marks it cites, the extent of trade in products bearing them, and that those marks actually threaten the distinctiveness of the mark being asserted. Citing registrations — which may be unused, or may never reach the market — does not get there.
So if your Indian pharma strategy has ever rested on “the stem is in the public domain”, that position now has a bill attached. The evidence it needs is market evidence: sales and turnover for the cited brands, invoices, distributor and stockist records, CIMS or MIMS-type listings, prescription-audit data. That file has to exist before the counterstatement or the written statement, not after — it is not assembled in a week. Where it cannot be assembled, the defence should be priced as unavailable and the brand changed instead.
The mirror obligation falls on brand owners. A stem search across the Register is not a clearance opinion, and the two 2026 outcomes show precisely which question a trademark clearance search has to answer: is the proposed mark reducible to the INN, or does it only begin there? LETROZ from letrozole, TEMO from temozolomide and NIMULID from nimesulide all produced thin rights, and the DAPLOGIN court distinguished each of them for exactly that reason. DAPLO did not, because ‘LO’ is not in the molecule name. If the mark you are about to adopt collapses into the molecule when you strip the last syllable, expect to be unable to stop a close copy later — and build the brand on a coined element instead.
Three portfolio actions this implicates
Start a confusion log, because its absence and its presence both counted. Intas’s ten clean years of coexistence were evidence that defeated similarity. Dr. Reddy’s line of DAPLO variants was evidence that reinforced it — the court accepted that a purchaser could read DAPLOGIN as another extension from the same house. Pharmacist queries, misdispensing reports, returned orders and wrong-brand enquiries are worth capturing contemporaneously on both sides of that argument; reconstructing them under a trial timetable is how cases are lost. Note too that the DAPLOGIN court dismissed the “these are prescription-only drugs” answer at the outset, on Cadila Healthcare; in the BEVATAS matter the prescription channel only helped because it came bundled with different APIs, different packaging and a clean record.
Audit Class 5 for marks that are really just the molecule. Those registrations will not support a Section 57 petition or an injunction, however long you have held them. Knowing which of your marks are enforceable and which are decorative changes where enforcement budget goes.
Treat a cancellation order as a downstream data problem, not just a win. A six-week rectification direction removes the Register entry, and anything keyed to that registration number — licences, security interests, customs recordals, marketplace brand-registry enrolments, export dossiers — has to be re-papered on the other side. For the losing proprietor, the rights simply stop. Section 57 also runs independently of any infringement suit, so a registration can be cancelled while separate litigation is still live; portfolios tracked in a single trademark docketing system are the ones where that disconnect gets caught rather than discovered at renewal.
The two decisions are consistent with each other and with Cadila Healthcare. Heightened scrutiny of medicinal marks is an instruction to look harder at confusion, not a presumption that resemblance equals infringement. What separated the outcomes was the record: same molecule, same indication, same channels and a wholesale absorption of the earlier mark on one side; different molecules, different products and a decade of documented coexistence on the other. Similarity opened both enquiries. Evidence closed them.