Nineteen Countries, One Shortcut: FPC Denies Regeneron/Bayer’s PI Bid Over Sandoz’s Afqlir®

Eylea®, the anti-VEGF blockbuster behind Regeneron and Bayer’s blockbuster ophthalmology franchise, has just had its first Swiss skirmish with biosimilar competition. Sandoz’s Afqlir®, launched across Europe from November 2025, prompted Regeneron, Bayer HealthCare and Bayer Consumer Care to seek a PI covering Switzerland and nineteen further European countries. The request failed — but only after the court found standing through a three-tier licensing chain, extended the concept of “offering” to mandatory stock-exchange disclosures, and then applied a new-ish approach to the DoE by tweaking the assessment of Q2.

Eylea® Franchise Under Biosimilar Assault

Eylea® and its higher-dose successor Eylea® HD form Regeneron’s best-selling drug after Dupixent — but this business is being squeezed from two directions at once. Combined US net sales of Eylea® and Eylea® HD fell from roughly USD 6.0 billion in 2024 to USD 4.4 billion in 2025, and by the first quarter of 2026 quarterly Eylea sales had dropped below USD 1 billion for the first time in eight years. Trade press attributes that erosion only partly to biosimilar entry: Roche’s rival anti-VEGF drug Vabysmo® (faricimab) has taken substantial share in its own right, while a wave of Eylea biosimilars — Sandoz’s Afqlir® among them — has added a second front of pressure since patent exclusivity lapsed. Afqlir® received European Commission marketing authorisation in November 2024, and Sandoz began its UK launch on 24 November 2025 — when also Regeneron’s Swiss SPC had just lapsed. Roll-outs in Germany, France and further markets were slated to follow through 2026.

The European litigation landscape on this patent family has been anything but uniform; ¶¶32-36. In September 2025, the Munich Regional Court granted Regeneron and Bayer a cross-border PI against Formycon covering some twenty-two European countries, based on a finding of infringement under the DoE of EP 2 364 691 — apparently the first cross-border patent injunction granted by a national EU court following the CJEU’s BSH/Electrolux ruling. Formycon and its commercialisation partner Klinge fared better in London: a UK High Court PI request by Regeneron and Bayer against their product FYB203 (marketed as Ahzantive®/Baiama®) failed in October 2025 ([2025] EWHC 2527 (Pat), and Regeneron/Bayer and Formycon/Klinge went on to settle all European, Latin American and Asia-Pacific claims in March 2026, clearing FYB203 for launch across those markets from May 2026. The judgment also records a Dutch injunction against Samsung Bioepis over its product Opuviz® (Rechtbank Den Haag, C/09/675547, 1 October 2025, ¶32). Sandoz’s own German entities faced a similar arc close to home: the judgment records (¶33) that the Munich Regional Court initially granted Regeneron an injunction against Hexal AG and Sandoz Deutschland GmbH over Afqlir® (7 O 15541/25), only for the Munich appeal court to stay enforcement for insufficient proof on two claim features (6 U 180/26), after which Regeneron and Bayer withdrew the application. (The FPC of Germany had previously held that a limited version of EP 691 was valid; 3 Ni 15/23.) Further corresponding litigation is cited in Vienna (¶34, 20 Cg 87/25k – 17; PI denied), Belgium (¶35, C/25/00034; PI denied).

Sandoz has already settled all US litigation with Regeneron, deferring its own US aflibercept launch (Enzeevu®) to Q4 2026. Switzerland was therefore one of the last live fronts specifically involving Sandoz’s Afqlir®, even as Regeneron/Bayer’s fight against other biosimilar makers was resolving elsewhere through a mix of injunctions and settlements.

Patent in Suit

The patent in suit is EP 2 364 691 B1, a divisional of EP 2 029 103 claiming priority from a US application of 16 June 2006, asserted here in its Swiss, Austrian, Belgian, Bulgarian, Czech, Danish, Dutch, Finnish, French, Greek, Hungarian, Irish, Italian, Polish, Portuguese, Romanian, Slovak, Slovenian, Spanish and Swedish national parts. See the EPO Register and the Swissreg for the Swiss part.

The claims are authoritative in English (¶40):

ANALYSIS

Bundling equivalence — and the Revival of the second question

The claimed formulation uses a sodium phosphate buffer and a dissociated tonicity salt (sodium or potassium chloride); Afqlir® uses a histidine buffer and no separate tonicity salt, relying instead on trehalose to perform both stabilizing and osmotic functions. The plaintiffs invoked equivalence for these interlinked substitutions. Swiss doctrine tests equivalence through the three-question Pemetrexed trias — objective same effect, obviousness of that same effect to the skilled person (“Auffindbarkeit”), and equal value (“Gleichwertigkeit”) — confirmed in BGE 143 III 666.

Where the case breaks new ground procedurally is in how the FPC applied that trias. Rather than assessing each substituted feature (buffer, tonicity agent, pH) in isolation, the court held (¶73) that where several exchanged features are interdependent, the trias must be applied to their totality — otherwise the cumulative divergence between the accused and the claimed formulation is understated by pretending, question by question, that everything else is unchanged. Applied that way, the court accepted objective same effect (¶74, resting heavily on Afqlir’s biosimilar approval status) but denied the second question: the skilled person, even with full knowledge of Afqlir’s composition, had no “begründete Erwartung” that swapping to a histidine buffer combined with removing the dissociated tonicity salt would preserve stability, given how differently the two buffers behave ionically at the relevant pH (¶75).

Following Urinalventil (BGE 142 III 772 E. 6.4), Swiss doctrine formulates Q2 ex post: the skilled person is deemed to already know that the accused embodiment achieves the claimed effect. Same in the U.K.: the UK Supreme Court in Actavis v Eli Lilly [2017] UKSC 48 deliberately reformulated the old, harder Improver Q2 for exactly this reason, holding that requiring the skilled addressee to work out for himself whether the variant would succeed “places too high a burden on the patentee,” and directing that Q2 be asked on the assumption that the addressee already knows the variant works. Switzerland and the post-2017 UK position are therefore aligned. Germany’s Schneidmesser test not so (BGH, Schneidmesser I/II, X ZR 168/00 and X ZR 135/01): it asks whether the skilled person, drawing on general knowledge and the patent’s own teaching, would be able to find (“auffinden”) the modification as achieving the same effect — a capability question anchored in the patent’s disclosure, without built-in foreknowledge that this specific accused variant succeeds. None of that made Q2 a live issue in Switzerland before. Quite the opposite: precisely because Swiss courts, like the post-2017 UK courts, hand the skilled person the answer before asking the question (I apologize for the simplification), a finding against obviousness has been vanishingly rare. Fn 90 concedes that in every prior leading Swiss case — Urinalventil, Pemetrexed, and Deferasirox (4A_273/2023) — same effect would have been found obvious even under Germany’s stricter, non-presupposing formulation. This is apparently the first Swiss decision in which the second question actually decides the case — and it does so under the lenient version of the test, which is what makes the finding notable.

The FPC seems aware of exactly what it is doing: it warns that a looser reading of Q2 would leave the second question without “jeden Anwendungsbereich” — any field of application at all (¶75):

Wo der Fachmann ohne begründete Erfolgserwartung experimentell tätig werden muss, um die Gleichwirkung festzustellen, ist diese nicht offensichtlich, wie das die zweite Äquivalenz-Frage verlangt. Würde man hier einen anderen Massstab anlegen, verlöre die zweite Frage jeden Anwendungsbereich.

In my mind, this is about as candid an acknowledgment as a court offers that a doctrinal test had drifted into rubber-stamp territory. What appears to have revived it here is the bundling move itself (¶73): aggregating the buffer switch, the loss of a dissociated tonicity salt, and trehalose’s double duty into a single combined divergence makes the “no reasoneable expectation” finding far more plausible than it would be run feature-by-feature, where each individual substitution might look unremarkable in isolation. The bundling is what made a genuine “Auffindbarkeit” failure possible for the first time, even under a test built for failure (again, apologies for the simplification).

Left unresolved is whether the ex post framing itself should hold going forward. (My vote: )

Nineteen countries, one shortcut

The plaintiffs sought injunctive relief not just for Switzerland but for nineteen further EPC states. And, indeed: good news is the FPC’s willingness to actually adjudicate infringement of the foreign national parts of a European patent. For that step, the FPC invoked the CJEU’s Solvay/Honeywell (C-616/10) and, more recently, its February 2025 BSH/Electrolux ruling (C-339/22) (¶7): a nullity defence raised against the foreign parts of a European patent does not divest the seized court of jurisdiction over the infringement claim itself. I could not identify a prior FPC decision that had previously granted, or come as close to granting, relief spanning this many foreign jurisdictions in one summary proceeding. Even though this cross-border PI would have been a Swiss novum (in particular on infringement under the DoE(!); I’ll come to that in a minute), I understand that the general principle was not even in dispute between the parties.

In my mind, the most interesting bit is the mechanism the court used to make nineteen-country relief administrable within a summary proceeding at all: because Art. 69 EPC and its Protocol on Interpretation are applied uniformly across EPC contracting states (are they?!), the court reasoned (¶8) that a glaubhaft-gemacht finding on infringement of the Swiss part extrapolates, in result, to the other national parts, without the need for nineteen separate national infringement analyses. (Tit for tat: The Munich court made a similar assumption in 7 O 16055/24 (¶138), too.)

That is a significant economy of adjudication — and it is what actually made this size of request triable at all on a summary timeline. Tellingly, the court expressly left open (¶16) the genuinely hard cross-border questions it never had to answer: whether Belgian, Greek, Irish, Polish and Czech law require the licence to be registered for the licensee to sue, and whether French, Finnish, Dutch, Romanian, Slovenian and Slovak law require the patent owner’s consent to a licensee’s action. Those questions remain untested, available for a future case that actually turns on them.

Ad hoc disclosures as “Anbieten” — and a holding company on the hook

The plaintiffs argued that Sandoz Group AG’s mandatory SIX ad hoc disclosures of 30 October 2025 and 24 November 2025 — announcing, respectively, the planned and then actual roll-out of Afqlir® — themselves constituted an offer under Art. 8 para. 2 PatA, even though Sandoz Group AG is a non-operating holding company with no employees. The FPC agreed (¶¶20–21). Ad hoc disclosures required under the SIX Listing Rules address the investing public, but their audience is not limited to investors; a disclosure that a generic’s market entry has occurred or is imminent lets potential customers know that some group company stands ready to supply it, which the court found sufficient to constitute an “Anbieten,” irrespective of whether the disclosing entity itself carries out any operational activity.

Sandoz AG, the group’s second-named defendant and the entity responsible for the content of the sandoz.com website on which those disclosures were published, was found liable on the same reasoning (¶23), and additionally as an aider and abettor (Beihilfe) through its ownership of the AFQLIR trademark: because regulatory rules require a biosimilar to be marketed under its approved name and forbid other products from using it, Sandoz AG’s licensing of the mark to the operating country subsidiaries was found to be a conditio sine qua non for the alleged infringement (¶24).

Licensee standing down the sub-licence chain

Regeneron licensed Bayer HealthCare on a co-exclusive/exclusive basis to develop and, respectively, offer or sell Eylea® outside the United States; Bayer HealthCare in turn granted Bayer Consumer Care an exclusive licence for the same territory, later extended to include the right to sue third parties directly. The defendants argued that only the entity actually exercising the exclusive right of use could sue, and that standing under Art. 75 PatG travels with that right down a single line rather than sitting with multiple tiers in parallel.

The FPC confirmed and extended its 2019 Klagerecht sui generis line (S2019_002): an exclusive licensee’s right to sue is an independent right, not a form of Prozessstandschaft exercised on the patentee’s behalf, and it exists alongside — not instead of — the patentee’s own right where the patentee retains a protectable interest (¶11). That principle, drawn originally from a two-tier relationship, was here applied to a three-tier chain: patent owner, exclusive licensee, and exclusive sub-licensee were all found independently entitled to sue in parallel, provided each shows its own protectable interest (¶14). The FPC also confirmed that where the alleged infringement touches the substantive scope of an exclusive licence at all, the licensee’s standing is not artificially confined to the narrower subset of acts (offering, selling) for which its licence was originally drafted (¶15).

“Ungefähr” doesn’t stretch — narrowing Deferasirox

Both asserted claims specify a pH of “about 5.8–7.0.” Construing that feature, the plaintiffs argued that “about” signalled something beyond a rounding tolerance and pointed to the ±0.1 pH variation reflected in the patent’s own preferred embodiments. The court rejected that reading (¶48): absent a clear definition in the description, “about” cannot be stretched by an arbitrary margin the specification never actually states, and the qualifier is best understood as preserving the patentee’s equivalence arguments rather than expanding the literal scope of the numeral itself. That’s interesting. Apparently, there is still room for an argument of infringement under the DoE in case of numerical ranges: the FPC expressly makes a distinctive cross-reference to the earlier Deferasirox case (O2021_004, O2021_005); fn 62. (I cannot readily see how that could possibly pass a fair assessment of Q3 of the trias: if “about” is understood to signal that DoE arguments are preserved, I wonder why a patentee should not have expanded the numerical range right from the outset. If he knew he wanted to cover a broader range, he could and should have done so. Literally.)

The operative range, after ordinary rounding convention, becomes 5.75–7.04 in the case at hand. That’s pretty much the EPO’s pragmatic approach in first instance proceedings (GL F-IV, 4.7.1), and I have some sympathy for it at least in summary proceedings.

The court also settled a measurement-temperature dispute in the defendants’ favour: because pH is temperature-dependent for a histidine buffer and the formulation is administered at room temperature, pH must be measured at 20–25°C (in line with the European Pharmakopöe) rather than at the 5°C storage temperature used in the patent’s own examples (¶47). On the facts, the plaintiffs’ Physiolution laboratory report — based on a small-volume surface-droplet measurement of Afqlir® samples rather than the standard immersion method of the European Pharmacopoeia — was found unreliable, and the defendants’ own certificates of analysis (reporting 5.7, rounding-consistent with a true value below 5.75) were preferred (¶¶66–68). Literal infringement of the pH feature therefore failed on the facts as well as on construction.

Costs

The court awarded the defendants 50 kCHF in necessary expenses for foreign legal opinions on top of the 50 kCHF tariff-based party compensation, despite the invoices submitted being unitemised. It reasoned that because infringement of a European patent’s national parts is assessed uniformly under Art. 69 EPC, foreign-law opinions addressed solely to that question are not “necessary” expenses recoverable under Art. 32 PatGG in conjunction with Art. 3 lit. a CR-FPC — only opinions on genuinely jurisdiction-specific issues, such as the licensee-standing formalities left open at ¶16, qualify (¶78). Given the unitemised invoices, the court estimated rather than calculated the recoverable portion, applying its own prior S2024_011 precedent that a plainly-exceeded threshold need not be proven to the last franc.

MW

—

BIBLIOGRAPHY

Case No. S2025_003 | Judgment of 17 July 2026 | ‘Aflibercept’

Panel of Judges:

• • Dr. Mark Schweizer
  • Dr. Tobias Bremi
  • Dr. Michael Kaufmann

Judge-rapporteur:

• • Dr. Tobias Bremi

Court Clerk:

• • Sven Bucher

Representative(s) of Plaintiffs:

• • Dr. Simon Holzer (MLL Legal AG)
  • Louisa Galbraith (MLL Legal AG)
  • Dr. Ulrike Ciesla (MLL Legal AG), assisting in patent matters

Representative(s) of Defendants:

• • Dr. Andri Hess (Homburger AG)
  • Julian Schwaller (Homburger AG)

—

PUBLICATION ON 

—

JUDGMENT

—

PATENT IN SUIT

—

BE ON THE KNOW

The post Nineteen Countries, One Shortcut: FPC Denies Regeneron/Bayer’s PI Bid Over Sandoz’s Afqlir® appeared first on FPC Review.