Plausibility without data: lessons from Accord v Novartis [2026] EWHC 2127 (Pat)

The Patents Court has handed down an important judgement in respect of Entresto, Novartis’s blockbuster heart failure drug. In this article we look at how it develops the law on plausibility: a separate article looks at the attack on Entresto’s SPC.

Key takeaway

A patent specification can satisfy the plausibility threshold for a pharmaceutical combination claim through qualitative descriptions of experimental results, even without numerical data, provided the disclosure, read as a whole and in the context of patent drafting conventions, goes beyond a bare assertion of efficacy.

Background

On 13 August 2026, Meade J handed down judgment in Accord Healthcare Limited v Novartis AG, a clearing-the-way action concerning European Patent (UK) No. 1 467 728 B1 and its associated SPC, which protect Novartis's drug Entresto (a combination of valsartan and sacubitril; the latter substance is part of a class of drugs known as “NEP inhibitors”). The drug treats heart failure, and in some territories, hypertension. Accord sought to invalidate the patent on multiple grounds, including that the patent application failed to make plausible that the claimed combination would achieve a clinically useful result in hypertension. The plausibility issue was, in Meade J's words, “the critical issue in the action”.

The legal framework on plausibility

The court applied the well-established Warner-Lambert framework as affirmed by the Court of Appeal in Sandoz v Bristol-Myers Squibb [2023] EWCA Civ 472 (Apixaban CA) and Generics v AstraZeneca [2025] EWCA Civ 903 (Dapagliflozin CA). The standard is one of "ab initio plausibility". This means that the application, read together with the common general knowledge, must positively make it plausible that the invention will achieve the claimed technical effect. The test is "relatively undemanding" but is more than just a test of good faith.

Accord's attack

Accord advanced a multi-pronged challenge:

• The application contained no numerical results whatsoever — at most, qualitative descriptions that some combination had given positive results in tests.
• Large parts of the specification were merely prophetic — descriptions of what could be done, written in the present tense in accordance with patent drafting convention, and not reflecting work actually carried out.
• To the extent anything positive was asserted, it related to "a NEP inhibitor" generically and not specifically to sacubitril.
• Without data, a trivially small (and therefore clinically useless) effect could not be ruled out.
• The patent could be attacked for obviousness to experts in either hypertension or heart failure because the patent mentioned both conditions.

The court's reasoning

Meade J acknowledged this was "a close call" but ultimately found the plausibility threshold satisfied, emphasising the following:

• Application to be read as a whole. The judge cautioned against focusing on individual passages in isolation. Critical to the finding were: (i) a clear disclosure that some work had been done and results were in hand; (ii) multiple specific pointers to the valsartan/sacubitril combination throughout; and (iii) a statement that "representative studies" with that specific combination had been carried out, immediately preceding descriptions of hypertension animal experiments.
• Numerical results not required. The court held that nothing in Warner-Lambert mandates numerical data, and that requiring them would be inconsistent with the deliberately low threshold. A descriptive, yes/no assertion of success in a test can suffice, as was accepted in Illumina v Latvia [2021] EWHC 57.
• Patent drafting conventions matter. The court endorsed having regard to the conventions of patent specifications. Prophetic statements, “surprisingly it has been found…”, and present-tense descriptions of aspirational work all have specific well-understood meanings when determining what the skilled person would understand from the disclosure. This is consistent with the Court of Appeal's approach to the skilled person's understanding of patent drafting in Virgin Atlantic v Premium Aircraft Interiors. 
• Implausibility due to “trivially low results” is narrower than it appears. Accord argued that because clinical utility in hypertension requires a blood pressure reduction of at least 4–8 mmHg, and the application discloses no numbers, the combination might achieve only a trivially low (and therefore useless) effect. Meade J rejected this, holding that plausibility may be established by an animal model and does not require showing a particular numerical level needed in humans.
• Bare assertion versus supported disclosure. The judge drew a clear line between a mere assertion of efficacy (which would not suffice) and a disclosure indicating that actual experiments had been carried out with positive results, even if those results are expressed qualitatively. The latter crosses the threshold.

Practical implications

1. For patent drafters: The judgment offers some comfort that qualitative disclosures of experimental results can survive a plausibility challenge, but it was a close call. The safer course remains to include at least some data wherever possible. The court said the patentee “could have avoided any risk by being explicit and by including experimental details and data”. The judgment is not an invitation to file thin specifications.
2. For challengers: The judgment confirms that the plausibility inquiry is a disclosure test, not an exercise in inferring what the patentee likely did. At the same time, courts will read specifications holistically and with an understanding of drafting conventions, which limits the force of "no data, no plausibility" arguments where indicators of genuine experimental work are present.
3. For patent litigators: Although not a new principle, the judgment is a clear application of the Philips v Nintendo approach: where a product claim is unrestricted as to field of use, an attack can legitimately be mounted from the perspective of any field covered by the claim. The policy argument that Novartis could not insulate itself by trimming heart failure references from the specification while retaining broad product claims was particularly incisive.
4. For life sciences patent strategy: In combination with the court's findings on collocation and technical contribution (where Meade J held that an actual experimental result showing an advantage for a specific combination is a technical contribution, even without quantitative synergy data), the case reinforces the value of including at least some empirical disclosure even qualitative for the specific claimed combination, rather than relying purely on class-level reasoning.

Comment

Accord v Novartis sits within the developing line of plausibility cases (Warner-Lambert, Apixaban, Dapagliflozin) and adds a useful data point at the lower end of what can satisfy the threshold. Practitioners should note, however, that the judge's candid acknowledgment that his view "shifted back and forth during the trial" underscores how fact-sensitive these assessments are. The judgment is unlikely to be the last word on the boundary between a sufficient qualitative disclosure and a bare assertion, particularly given the limited role post-filing evidence has in establishing plausibility.