BMS v Sandoz and others (Gerechtshof Den Haag, 200.327.532/01, 2023)

Dhenne Avocats · 11 October 2026

Our analysis

Summary

In summary proceedings, The Hague Court of Appeal set aside the judgments that had refused BMS an injunction against generic apixaban (Eliquis) and ordered Sandoz, Stada c.s. and Teva c.s. to cease all infringement of Dutch SPC 300500. Applying G 2/21, it held that a proprietor may rely, for inventive step, on a technical effect proven after filing provided that the skilled person can derive from the application that the effect is encompassed by its technical teaching; the application need not contain proof of the effect or make it plausible (paras 6.6 to 6.9). The court expressly departed from the English test applied to the same patent and left a generic company that launched during the life of the right to bear the risk of having to withdraw.

Facts and procedure

EP 1 427 415 B1, derived from WO 03/026652 and expired on 16 September 2022, is the basic patent for SPC 300500, in force until 19 May 2026 (paras 3.3 to 3.5). During prosecution the claims were limited to apixaban and BMS filed in vitro data showing stronger factor Xa inhibition than the closest compounds of WO 00/39131 (paras 3.7 and 3.8). A first summary action against Sandoz had been dismissed on 10 May 2022 (para 3.12). After G 2/21, BMS sued Sandoz again, together with Stada c.s. and Teva c.s.; the preliminary relief judge of The Hague District Court dismissed the claims on 17 and 31 May 2023, finding that the application did not make the technical effect plausible (paras 2.1 and 4.4).

The applicable law

The court applied the problem and solution approach and point II of the order in G 2/21 (paras 3.20, 6.2 to 6.4). Dutch courts are not bound by the Enlarged Board of Appeal, but its decisions are regarded as highly authoritative and are generally followed in the interest of harmonisation, so the court applied the G 2/21 test (para 6.13).

Question

Does the G 2/21 test require the application to make the asserted technical effect plausible and, if not, is there a serious chance that the court on the merits will invalidate the apixaban patent for lack of inventive step?

Decision

The only requirement laid down in G 2/21 is that the effect be derivable from the application; the words « would derive » mean nothing else (paras 6.6 and 6.7). Point 77 of G 2/21, which concerns sufficiency of second medical use claims, is irreconcilable with a general plausibility requirement for inventive step (paras 6.8 and 6.9). Here the effect was stated as the aim of the application, the claim covered a single compound, no reason to doubt the effect was advanced and the application disclosed a test for measuring the Ki value (paras 6.15 to 6.18); the larger scale synthesis of apixaban (example 18) and claim 8 of the application, directed to it, identified it as the most promising candidate (paras 6.21 and 6.22). The French and Norwegian judgments on the merits upholding inventive step carry authority without binding the preliminary relief court (paras 6.24 and 6.31). The added matter and priority objections (Article 87 EPC) failed (paras 6.38 and 6.44). Urgency followed from price erosion (paras 6.49 to 6.51). Sandoz had launched knowing of the G 2/21 referral and bore the risk that materialised; the guarantees offered, limited in amount and duration, and Article 16 of the Charter did not prevail (paras 6.54, 6.55, 6.58 and 6.59). The court ordered cessation, removal from the G-standaard, disclosure of information and a recall, subject to penalty payments (paras 7.1 to 7.5).

Key points for practice

• According to the court, G 2/21 does not require the application to prove the effect or make it plausible: it is enough that the effect is derivable from it (paras 6.6 to 6.9).
• The court expressly distinguished this test from the one the English courts draw from Warner-Lambert (para 6.28).
• A generic company that launches before expiry bears the risk of withdrawal; a unilateral guarantee limited in amount and duration does not prevent an injunction (paras 6.55 and 6.58).
• Practical point: an application that states the effect as its aim, individualises the compound and discloses the test method allows later data to be relied upon, although test results may be required depending on the circumstances (para 6.14); an intervening Enlarged Board decision may justify a fresh summary action (para 6.61).

Relevance before the UPC

As the UPC assesses validity under Articles 56 and 138 EPC, this reading of G 2/21, set against the English approach, may be debated before it in relation to post-filed data.

Provisions applied

European Patent Convention

Art. 56; Art. 83; Art. 87

EU law

Charter of Fundamental Rights of the European Union, Articles 16 and 17(2)

National law

Dutch Code of Civil Procedure (Rv), Articles 233(3) and 1019h; Dutch Patents Act 1995, Article 70(11)

Case law cited

G 2/21; G 1/19; G 3/97; G 1/12; T 939/92; T 1329/04; T 235/13; Gerechtshof Den Haag, 7 November 2017, ECLI:NL:GHDHA:2017:4029 (Leo Pharma/Sandoz); Warner-Lambert v Generics [2018] UKSC 56; Sandoz v BMS [2022] EWHC 822 (Pat)

Related decisions

• G 2/21, EPO Enlarged Board of Appeal, 23 March 2023, decision interpreted and applied (paras 6.4 to 6.14).
• Sandoz v Bristol-Myers Squibb, Court of Appeal of England and Wales, 4 May 2023, same patent, invalidity upheld in the UK (paras 3.21 and 6.27).
• Warner-Lambert v Generics, UK Supreme Court, 14 November 2018, test held to differ from G 2/21 (paras 6.27 and 6.28).
• Johns Hopkins University, GDF-9, EPO, 28 June 2005, implausibility case distinguished (para 6.17).

Prepared by Dhenne Avocats from the text of the decision (rechtspraak.nl, official XML version from data.rechtspraak.nl), consulted on 11 October 2026. Only the official text is authoritative.