The central division of the Unified Patent Court sitting in Milan has largely dismissed a preliminary objection by Novo Nordisk and, in doing so, told European patent proprietors that walking away from a patent is not necessarily a way to walk away from a ruling on whether it was ever valid. The order, issued on 22 October 2026 in case UPC_CFI_2221/2026, arises from Sandoz’s revocation action against EP 3 689 365, a unitary patent covering a composition comprising semaglutide for the prevention or treatment of type 2 diabetes administered once weekly at a 1.0 mg parenteral dose. Semaglutide is the active ingredient in Ozempic, Wegovy and Rybelsus. According to JUVE Patent, Sandoz filed in Milan this summer on lack of inventive step and added matter, and bolted on three further requests aimed at what it calls the “divisional game” — withdrawing a patent shortly before an adverse validity ruling while keeping equivalent subject matter alive in a pending divisional. EP 365 has such a divisional pending.
The panel — presiding judge Andrea Postiglione with legally qualified judge Anna-Lena Klein and technically qualified judge Rudi Goedeweck — split the three requests. The first, an abstract declaration that a semaglutide composition was obvious at the priority date independently of any specific patent right, was held manifestly inadmissible: too broad, unanchored, and liable to affect third parties who had never been heard. The court expressly distinguished it from an English-style Arrow declaration, which is directed at an identified product or process. The second request survived. Framed as a declaration that any claim of a European or unitary patent entitled to priority from the EP 365 family, broad enough to cover claims 1 to 5 of the patent in suit, would lack inventive step, it was read as genuine Arrow-type relief against the re-emergence of the same protective perimeter through a divisional — binding, Mewburn Ellis notes, only as between the present parties. On the third, an order barring Novo Nordisk from withdrawing its approval of the text at the EPO until validity is finally decided, the court also accepted jurisdiction, grounding it in Article 32(1)(b) and (d) UPCA and the principle that a central division must be able to protect the effectiveness of its own proceedings. It reserved whether the conditions for such an interim measure — urgency, proportionality — are actually met.
The jurisdictional reasoning is the part that travels. The panel found the court holds an “inherent jurisdiction” implicit in Article 32 UPCA where that is necessary to secure the effective exercise of the competences the Agreement expressly confers, reading Article 32 in light of the Agreement’s object and purpose, Article 47 of the Charter of Fundamental Rights, and Article 42(2) UPCA, which requires the court’s procedures and remedies not to distort competition. It drew on Court of Appeal reasoning and on earlier UPC case law about anti-anti-suit injunctions. On divisionals themselves the court was careful: filings under Article 76 EPC and Rule 36 EPC serve a legitimate function, but strategic filings designed to maintain uncertainty for third parties and delay generic market entry may amount to an abuse of rights.
What actually changes
The withdrawal exit is no longer clean. The single most consequential line in the order is that if a proprietor withdraws approval of the text under Article 113(2) EPC mid-proceedings, the Milan division may decline to dispose of the action under Rule 360 of the Rules of Procedure and rule on validity anyway — with consequential effects on divisionals whose claims substantially correspond to the adjudicated subject matter. Rule 360 is being treated as a discretion, not an automatic off-switch. Any European defence plan that assumes letting the parent die kills the revocation action with it needs to be re-tested on that assumption, now, not after a hearing date is fixed.
The divisional is inside the blast radius. Request 6 shows the shape of the pleading to expect: not “revoke this patent” but “declare that nothing in this priority family, broad enough to cover these claims, is inventive.” Even limited to the parties, a reasoned first-instance finding on an entire claim perimeter is persuasive material in national proceedings and decisive input into a generic or biosimilar launch date. Proprietors running a continuation-style perimeter strategy in Europe should be asking what a declaration framed against the family rather than the patent would do to it.
Document why each divisional exists — contemporaneously. “Abuse of rights” is an evidential question, and the record that answers it is the prosecution file. A divisional traceable to unsearched subject matter, a Rule 164 constraint or a specific objection reads very differently from one whose only visible purpose is to keep a competitor guessing. That record is built during prosecution, not reconstructed during litigation. Commissioning a hard-nosed patent invalidation review of your own family before a challenger does it for you is the cheaper version of this exercise.
For challengers, the drafting lesson is anchoring. Request 5 and Request 6 asked for substantively overlapping outcomes; one was manifestly inadmissible and one was not. The difference was a named family, a priority date and an identified claim scope. Abstract invalidity declarations remain outside the UPC’s competence, as Bristows — acting for Novo Nordisk — emphasises. The evidence that makes a request of that specificity worth filing is prior art developed to litigation standard, which is why prior-art and litigation search work belongs at the pleading stage of a UPC revocation, not after it.
Watch the knock-on to your EPO timetable. EP 365 faces 18 oppositions at the EPO — Sandoz, Teva, Zentiva and Synthon among them, alongside several straw men — and the Opposition Division has agreed to accelerate because of the pending UPC case. For anyone in that bundle, deadlines just compressed. More broadly, a UPC revocation action is now a lever that can pull an opposition timetable forward, which cuts both ways depending on which side of the patent you sit.
One caution against over-reading it. The panel granted leave to appeal in its entirety while indicating it is not currently inclined to stay first-instance proceedings, so the Court of Appeal may yet revisit the inherent-jurisdiction foundation while Milan carries on. Nobody should restructure a European portfolio on first-instance reasoning alone. But the planning assumption that a quiet withdrawal at the EPO ends an inconvenient validity case is the one thing that should not survive this order.