arrow_back

Back

Inference is not disclosure: Empagliflozin's compound patent falls while its polymorph stands

20 November 2026
schedule

4 min read

Add as a preferred source on GoogleAdd as a preferred source on Google

In Dr Reddy's v Boehringer Ingelheim [2026] EWHC 2944 (Pat), Meade LJ revoked Boehringer's compound patent for empagliflozin (EP 131) and, with it, the SPC that would otherwise have run until 26 November 2029. Empagliflozin is the active ingredient in Boehringer's SGLT2 inhibitor, Jardiance. The Judge upheld Boehringer's patent to a crystalline form of the drug (EP 552). Dr. Reddy's had brought the action to clear the way for a generic launch, and Boehringer counterclaimed for infringement, obtaining interim relief in November 2025.

Key takeaways on plausibility  

  • The court reads what the specification discloses. Working out what the patentee probably did is "detective work" and doesn't count.
  • Experimental data is not essential and a qualitative statement that testing was done and worked can be enough. Saying nothing about testing is fatal.
  • Broad claims about a whole class of compounds carry little weight, and a claimed compound can't hide in a long list.
  • Warner-Lambert governs for now, but the uncertainty after G 2/21 may need to be addressed by the UK Supreme Court 

What the contribution had to be

Dr. Reddy's argued that neither specification made plausible the technical contribution on which the claims depend, so the claims were insufficient or obvious. Applying Warner-Lambert, Meade LJ required the application, read with the common general knowledge (CGK), to give "reasonable scientific grounds" for expecting that the invention "might well work". By trial, the contribution had narrowed to in vitro SGLT inhibition by the 15 compounds listed in [0054] of EP 131. Meade LJ held that, on the specific facts of the case, in vivo efficacy did not also need to be shown.

Why EP 131 failed

Boehringer did not need experimental data to succeed, since a qualitative statement or a sound rationale could suffice. Its difficulty was that nothing in the disclosure of EP 131 said any SGLT test had been conducted.

Boehringer asked the court to infer testing from the patent's descending levels of preference. Meade LJ treated that as "detective work" into what the patentee probably did, rather than a reading of what the document discloses. The ranking in [0054] of EP 131 could equally reflect medicinal chemistry theory, patent drafting strategy, or ease of synthesis. The specification also reported yields and mass spectra for some of the compounds, but no potency figures.

Boehringer's fallback argument, based on a glucose moiety shared with phlorizin and T-1095/T-1095A, found no real support.

Might the Supreme Court revisit plausibility?

The current UK test comes from the Supreme Court majority in Warner-Lambert, which adopted "ab initio plausibility". Under that test, the application read with the CGK must positively make it plausible that the invention achieves the claimed technical effect. The minority preferred a looser test of "ab initio implausibility" where a patent fails only if the application gives reason to doubt that the invention works.

The pressure to revisit the Warner-Lambert test comes from the EPO. Meade LJ identified the Enlarged Board's decision in G 2/21 as the second "landmark" of the past decade, alongside Warner-Lambert itself. The English Court of Appeal considered G 2/21 in Apixaban and Dapagliflozin and held that Warner-Lambert remains binding. In Dapagliflozin, it also concluded that other EPC states and the Technical Board of Appeal gave no reason to depart from Apixaban.

Meade LJ nonetheless commented on the Court of Appeal's conclusion in Dapagliflozin that courts applying the EPC have not yet agreed on what G 2/21 means. The Supreme Court has so far declined to reopen Warner-Lambert since release of G 2/21. Meade LJ said that whether to do so is entirely a matter for it, but that it "would be rational for it to do so...when the right case comes along". He therefore thought there was enough chance of the law changing during these proceedings to justify making contingent findings on facts going to plausibility.

Those findings show what is at stake. On Dr. Reddy's case, he found that the "strikingly unusual nature of the -OTHF moiety in empagliflozin", together with its size, polarity and likely conformation, gave positive reason to doubt SGLT activity. That would matter under an implausibility test. On Boehringer's case, he accepted in narrow terms that Boehringer's own tests had recorded IC50s below 50nM, while making no finding on their reliability. He was "comparatively more dubious" about which legal standard those facts would ever go to. 

Why EP 552 survived

SGLT utility for the crystal form was no more plausible than for the compound itself. However, the PXRD and DSC data, read with Professor Price's evidence, made it plausible that a substantially single, purer, and more stable form had been obtained.

On industrial application, Meade LJ held under Human Genome Sciences that this was enough. A stable form makes it easier to test a compound that the patent presents as a candidate SGLT inhibitor. In his words, it was "a narrow contribution of modest magnitude, but it is real not fanciful" and it is industrial in nature.

Inference is not disclosure: Empagliflozin's compound patent falls while its polymorph stands