The Unified Patent Court’s proportionality defence against injunctions is now concrete enough to plan around, and the plan is uncomfortable for defendants. In a guest analysis published by The IPKat on 1 November 2026, Niels Hölder and Andreas Hänsele of Carpmaels trace the line from the Court of Appeal’s decision in Meril v Edwards to this summer’s first-instance ruling in Hologic v Siemens, and the shape that emerges is a strong rule with one narrow exit. Article 63(1) UPCA says the court “may” grant an injunction; the Court of Appeal held in its decision of 25 November 2025 (UPC_CoA_464/2024 and consolidated appeals, Edwards Lifesciences v Meril, EP 3 646 825) that the proprietor has a claim to the injunction, so the court has no positive discretion to withhold it — only a power to deny it for reasons of proportionality under Article 3 of the Enforcement Directive (2004/48/EC), taking third-party interests into account.
Where those third parties are patients, the Court of Appeal fixed two limbs: an exception may be justified where the infringing embodiment is the sole available treatment method, or where it improves on available treatments so as to notably enhance patient care. Applying limb one, the appeal court kept Meril’s XL-sized transcatheter heart valves outside the injunction, because for some patients no alternative existed. It also rebuilt the mechanics of that carve-out: the Munich Local Division had been content for Edwards to run a “Medical Request Portal” issuing single-use licences, and the Court of Appeal refused to leave patient supply dependent on the patentee continuing to operate a portal or on its own internal assessment. As Bardehle Pagenberg and V.O. Patents & Trademarks both record, the carve-out now turns on an objective trigger — a notification of intended use by a physician confirming the XL device is the only available treatment option.
Then came the test of limb two. In Hologic v Siemens Healthineers (UPC_CFI_758/2024, with UPC_CFI_259/2025), the Düsseldorf Local Division’s decision of 10 June 2026 found Siemens’ MAMMOMAT B.brilliant 3D mammography system — and its marketed “flying focal spot” feature — to infringe EP 2 352 431 B1, and dismissed the revocation counterclaim. Siemens argued that an injunction would harm patients because its system was the only or materially better tool for early breast-cancer detection. The court rejected it, pointing to comparable tomosynthesis systems already on the market — including Siemens’ own MAMMOMAT Revelation, Fujifilm’s AMULET Innovality and Planmed’s Clarity 3D. On the IPKat authors’ reading of the decision, the division went further and treated advantages such as a faster, less painful examination at the same image quality with less radiation as still not establishing an exception, because the situation was not comparable to Meril v Edwards. The injunction covers Germany, France and the Netherlands, with recall of systems placed on those markets since 15 August 2018 and destruction of stock; Hologic has said customers must also be refunded the purchase price and recall logistics costs.
What this actually changes
Read the two decisions together and limb two is close to a dead letter in practice. If “faster, safer, less radiation, same image quality” does not clear the bar, a defendant planning to argue clinical superiority is planning to lose. The only limb with proven bite is the first, and it is not a product-level argument at all — it is a cohort-level one. Edwards v Meril was won on specific valve sizes for specific patients, not on the merits of the Meril valve. If you are the accused party, the work is to identify, before the oral hearing, the narrow set of indications, sizes or configurations for which no adequate alternative exists, and to evidence it with clinical sources rather than marketing claims. The burden is yours, and a vague assertion that nothing on the market matches you is exactly what failed in Düsseldorf.
Second, ask for the mechanism, not just the exception. The Court of Appeal’s objection to the Medical Request Portal is the most portable part of the ruling: a carve-out that depends on the patentee’s goodwill is not a carve-out. Draft the relief you want — physician notification, defined confirmation wording, no patentee discretion — into your own prayer, because the court will not design it for you after judgment.
Third, the UPC’s version of this defence is more forgiving on one point than the national law many European litigators are used to. The IPKat analysis notes that Meril had failed to win a German compulsory licence partly for insufficient efforts to obtain a licence — the Interferon-gamma, Raltegravir and Alirocumab line of Federal Court of Justice case law — and had even been characterised as an unwilling licensee, yet still obtained a UPC carve-out, because patients cannot control the defendant’s conduct. Compare the Regional Court of Düsseldorf in 2022 (4c O 18/21), which refused to carve out an HCV drug for patients with no alternative precisely because the defendant had not genuinely tried to licence. A broken negotiating record therefore does not automatically sink a UPC proportionality defence. That is not an invitation to skip the licence approach — an offer is still the cheaper route, and the better record — but it changes what the defence is worth once negotiations have already failed.
Fourth, a carve-out is not free. Proportionality limits the injunction, not the damages claim: the patentee can still be compensated for products exempted from the prohibition. The IPKat authors flag that the court may take the estimated scope of future infringing use into account when fixing provisional damages under Rule 119 RoP, and may in exceptional cases apply Article 68(4) UPCA by analogy to set reasonable compensation. In other words the realistic best case for a defendant with a genuine patient-access argument is a court-imposed royalty on a sliver of its sales — which is why the commercial calculation should be run before the defence is pleaded, not after.
Finally, note how limb one is decided: by what else is on the market. In Hologic, the patentee’s competitors — and Siemens’ own earlier system — were the reason the defence failed. That makes the competitive-alternatives map a litigation asset for both sides, and a pre-launch question rather than a trial question. For regulated products with long launch cycles and no quick redesign, the time to know which UPC states carry exposure, and whether a court would see your device as the only option for anyone, is while the product is still in planning: that is ordinary freedom-to-operate work. And because the proportionality exit is this narrow — Bardehle counts only two UPC cases in which a disproportionality defence has even partly succeeded, the other being the preliminary-injunction order in Valeo v Magna (UPC_CFI_368/2024, 31 October 2024) — attacking validity and claim scope remains the main defensive route, which is where a rigorous patent invalidation search programme earns its keep. Siemens lost the patent fight twice before the injunction was ever argued: its EPO opposition was rejected and, as the case record shows, the appeal failed before the UPC action was even filed.
One further operational detail from Hologic deserves attention from anyone selling software-defined hardware in Europe. Siemens argued a software update disabling the flying-focal-spot function made a recall unnecessary. The court held that a software fix must objectively ensure the infringing state cannot easily be restored, and since Siemens retained firmware control, it could re-enable the feature at any time. A toggle is not a design-around. If your non-infringing alternative is a setting, assume the court will treat it as reversible and order the recall anyway.