Non‑validated Biomarkers and Conference Posters Can Render a Dosing Regimen Obvious: The High Court’s Sandoz v Bayer (Rivaroxaban OD) Decision

Introduction

In Sandoz AG & Rowex Ltd v Bayer Intellectual Property GmbH [No. 1] [2025] IEHC 357, the Commercial (Intellectual Property & Technology) List of the High Court of Ireland invalidated the Irish designation of EP 1 845 961, a dosage regimen patent covering once‑daily administration of rivaroxaban (BAY 59‑7939) for thromboembolic disorders. The substance (rivaroxaban) was off‑patent; only the once‑daily regimen remained protected until January 2026.

The sole ground pursued was obviousness. The factual crux was whether, as of the priority date (31 January 2005), the skilled team would have considered once‑daily dosing obvious to try with a reasonable expectation of success in light of Phase I data disclosed by Bayer at ASH 2003: the “Kubitza” SAD/MAD posters and the “Harder” thrombin‑generation poster, together with their abstracts in Blood. The case unfolded against a backdrop of divergent foreign decisions and a 2021 EPO Technical Board of Appeal (TBA) ruling upholding the European patent.

The judgment is significant for four reasons:

It squarely applies the Actavis v ICOS “obvious to try with a reasonable expectation of success” framework to dosage patents involving high‑risk anticoagulants.
It clarifies that non‑validated pharmacodynamic biomarkers (assays) and conference posters can legitimately inform dose‑frequency selection at Phase II, even without established clinical correlation, consistent with FDA/EMA guidance.
It explains when a court may assume the skilled team can “routinely” reproduce preclinical and Phase I data to reach the state of the art, and rejects “data deficit” as a blanket obstacle to obviousness.
It details how Irish courts weigh foreign judgments under judicial comity, giving particular persuasive weight to the UK final decisions invalidating the same dosing claims while distinguishing the TBA and other EPC outcomes.

Summary of the Judgment

The Court declared claim 1 invalid for lack of inventive step. Applying the Pozzoli steps and the Actavis multi‑factor analysis, Mulcahy J held that:

The skilled team (a clinician and a clinical pharmacologist) at the priority date, equipped with common general knowledge, would have read the Kubitza Phase I SAD/MAD posters and the Harder thrombin‑generation poster together.
The Harder poster explicitly suggested once‑daily suitability (“some parameters … indicate a long‑lasting pharmacodynamic effect … suggesting suitability for a once‑daily dosing regimen”). That teaching would be seized upon by the skilled team and was not to be dismissed on the basis that the assays were “experimental” or not clinically correlated.
FDA exposure–response guidance (2003) and aligned EMA practice permitted use of non‑validated biomarkers to inform Phase II dose and dosing frequency selection; correlation with clinical outcomes is not a pre‑condition to progression.
Routine preclinical and standard Phase I steps can be assumed achievable by the skilled team to “get to” the prior art position; the “data deficit” argument cannot preclude obviousness where the missing steps are routine.
Reading the prior art as the skilled team would, once‑daily dosing was obvious to try with a reasonable expectation of success (safety and efficacy) in a phase II hip‑surgery prophylaxis study, notwithstanding the special caution required for anticoagulants.
Foreign decisions were considered under judicial comity. The High Court aligned with the UK final decisions (High Court and Court of Appeal) invalidating the claim and distinguished the TBA and other validations (which often did not consider the Harder poster or were not final).
Secondary evidence (Bayer’s internal development documents admitted under confidentiality) was secondary but corroborative: contemporaneous records showed the thrombin‑generation study materially drove the decision to add a once‑daily arm in Phase II.

Analysis

Precedents Cited and Their Influence

Merck, Sharp & Dohme v Clonmel Healthcare [2022] IESC 11: restated Irish obviousness principles; the skilled addressee/team is a legal construct, uninventive but with common general knowledge.
Inhale Therapeutic v Quadrant [2002] RPC 419 and Asahi v Macropharma [2002] EWCA Civ 466: the skilled reader comes without preconceptions but reads prior art with interest; real‑world prejudices and mindsets matter.
Dyson v Hoover [2002] RPC 22: commercial mindsets can shape the skilled person’s horizon.
Ranbaxy v Warner-Lambert [2009] 4 IR 584 and Re Glaxo Group [2009] IEHC 277: common general knowledge; Pozzoli four‑step test; strict avoidance of hindsight.
Nichia v Argos [2007] EWCA Civ 741 and MedImmune v Novartis [2012] EWCA Civ 1234: obviousness is multi‑factorial; “obvious to try” is helpful where there is a fair expectation of success, assessed holistically.
Actavis v ICOS [2020] 1 All ER 213 (UKSC): central framework for dosage patents—routine steps are not ignored; consider burden, cost, alternative paths, value judgments, expectations of success; appellate courts should not over‑intrude on evaluative findings.
Molnlycke v P&G [1994] RPC 49: role of secondary evidence—cannot replace primary expert evidence but can test plausibility.
Illumina v TDL Genetics [2020] RPC 9: how a skilled person reads short conference abstracts—cautiously, without forensic dissection, but with care and interest.
Gilead v Mylan [2021] IECA 22; Warner‑Lambert v Generics [2019] 3 All ER 95; Norton v Boehringer [2022] IECA 58: judicial comity—foreign outcomes are persuasive “reality checks,” but domestic courts must engage with evidence before them.

The Court synthesized these authorities to craft a principled yet pragmatic approach to dosage regimen obviousness where safety stakes are high (anticoagulants), prior art includes Phase I posters, and parties contest the probative value of non‑validated biomarkers.

Legal Reasoning

The Court proceeded through the Pozzoli framework, addressing the key disputes within an Actavis‑style multi‑factor analysis.

Skilled team and CGK. The team comprises a thrombosis clinician and a clinical pharmacologist. CGK included anticoagulant risks, PK/PD, steady state, half‑life as a starting point, and knowledge of routine assays (PT/aPTT) and research assays (e.g., ETP). It did not include any accepted correlation between thrombin‑generation assays and clinical outcomes for direct FXa inhibitors.
What the prior art taught.
- Kubitza SAD/MAD posters: rivaroxaban was safe/tolerated in healthy volunteers; demonstrated predictable PK/PD; reported close PK–PD correlation; half‑life 3–6 hours; Phase I MAD concluded suitability for twice‑daily dosing up to 30 mg.
- Harder poster: single 5 mg and 30 mg doses reduced thrombin generation; several assay parameters were sustained at 12 hours, with graphs showing effects at 24 hours for 30 mg; concluding bullet explicitly suggested “suitability for a once‑daily dosing regimen.” The authors overlapped with Kubitza’s and included Prof. Breddin, a leading thrombosis scientist.
Reading conference posters. Posters are read with care, not forensic suspicion; the skilled team expects honesty and competence, especially given authorship pedigree and cross‑references between the studies. The Court rejected a “scientific nihilism” that treats poster results as untrustworthy absent all raw data.
Non‑validated biomarkers can be probative at Phase II design. The Court found that, consistent with FDA 2003 Exposure‑Response guidance and EMA practice, non‑validated biomarkers may be used to select dose/frequency for Phase II where they rationally reflect drug action—even if no established clinical correlation exists. Requiring clinically validated biomarkers before Phase II would stall drug development. This undermined Bayer’s core “no correlation/no reliance” argument.
Routine steps assumption. For obviousness over prior art, courts may assume the skilled team could perform routine preclinical and standard Phase I work to reach the state of the art described, absent evidence to the contrary. The “data deficit” (lack of raw data in posters) did not negate obviousness where the missing steps were routine; nor could Bayer rely on unspecified internal datasets to defeat an obviousness case without calling factual witnesses.
Resolving half‑life inconsistencies. Kubitza reported 3–6 hours; Harder’s introductory text noted 9–12 hours. Such variability in early studies is not unusual. Both datasets were internally consistent in showing strong early effects with waning over time; the Harder graphs showed persistence to 24 hours on several parameters. The skilled team would not dismiss Harder’s half‑life statement simply because the graphs and the text were succinct; the identity of the authors and proximity of the studies mattered.
Safety and ethics. Anticoagulants demand caution (risk of bleeding if overdosed; thrombosis if underdosed). Yet Phase II trials always involve managed risk; a drug may be obvious to try with reasonable prospects without eliminating risk. The hip prophylaxis setting is high‑risk overall but uses stringent oversight; the skilled team could reasonably expect a safe and effective outcome for an exploratory once‑daily arm given the totality of Phase I evidence.
Secondary evidence as plausibility checks. Internal Bayer documents were properly treated as Molnlycke “secondary evidence.” They corroborated that the thrombin‑generation study materially motivated the addition of once‑daily arms in Phase II. The Court refused to infer an undisclosed “invention story” from unspecified data absent factual witnesses.
Expert evidence and weight. The Court valued cross‑examination; it preferred Sandoz’s experts, finding Bayer’s experts overly sceptical and inconsistent on key points (e.g., non‑validated biomarkers and responsibility for proposing dosing arms). A late change in one expert’s position was held to be inadequately explained.
Foreign judgments and comity. The Court engaged with numerous EPC and non‑EPC decisions. It found the UK judgments (final) particularly persuasive given identical prior art and the same expert witnesses. It distinguished the TBA and some national validations where the Harder poster was not considered or where an incorrect “no correlation/no reliance” premise was adopted. Comity supported, rather than undermined, invalidation.

On this analysis, once‑daily rivaroxaban was “obvious to try” with a reasonable expectation of success at the priority date. The once‑daily claim was therefore invalid for lack of inventive step.

Impact

The decision has important practical consequences for pharmaceutical patent litigation in Ireland and beyond:

Biomarker jurisprudence in dosage patents. The Court’s express reliance on FDA/EMA guidance and acceptance of non‑validated PD biomarkers to inform Phase II dosing regimens sets a clear Irish marker. Claimants cannot defeat obviousness simply by labelling biomarkers “experimental” or “uncorrelated” where they rationally capture pharmacology and would be used by the skilled team.
Conference posters can be decisive prior art. Short, non‑peer‑reviewed posters and abstracts will be read with care but not cynicism; author credibility and internal consistency matter. Explicit suggestions (e.g., “suitable for once‑daily dosing”) can make a regimen obvious when the rest of the dataset points in the same direction.
Routine steps doctrine clarified. Irish courts will assume the skilled team can perform routine preclinical/Phase I work to “get to” the state of the art. A curated “data deficit” argument is unlikely to succeed where the missing steps are standard and the prior art teaches the critical idea.
Comity and cross‑border coherence. Final UK decisions on the same patent claims were persuasive. TBA and other national validations received respectful but limited weight where the evidential foundation differed (e.g., lack of the Harder poster) or where the non‑correlation point was overemphasised.
Litigation and R&D hygiene. The judgment cautions patentees about public‑domain disclosures that “suggest” a regimen; even in Phase I, clear language can be used against them. Conversely, challengers should assemble a coherent prior‑art narrative including posters, abstracts, and assay‑based PD evidence.
Expert practice. The Court emphasised the value of cross‑examination and warned against extreme scepticism untethered to how the skilled team would in fact use early‑phase data. Parties should ensure expert positions align with regulatory science and development norms.
Immediate market context. The patent was due to expire in January 2026; this decision accelerates generic certainty for once‑daily rivaroxaban in Ireland, aligning with final UK and Swiss outcomes and with Australia, while acknowledging continuing divergence across the EPC.

Complex Concepts Simplified

Pharmacokinetics (PK) vs Pharmacodynamics (PD). PK is how the body handles a drug (absorption, distribution, metabolism, excretion); PD is what the drug does to the body (biological effect). A drug’s half‑life is a PK measure; thrombin‑generation inhibition is a PD effect.
Half‑life and dosing frequency. Half‑life is a starting point for dose frequency (often one or two half‑lives between doses), but PD effects sometimes outlast measurable plasma concentrations. When PD “outlives” PK, less frequent dosing may work.
Therapeutic window. The range of drug exposure that is effective without undue toxicity. Anticoagulants have two‑sided risks: too little (thrombosis), too much (bleeding). A new drug’s window is unknown pre‑Phase III; Phase II explores dose and frequency.
Assays and biomarkers (e.g., ETP, PITT, PICT). Laboratory methods that track aspects of clotting biology. They may be “non‑validated” (not clinically correlated) yet still guide Phase II design by showing pharmacological effects over time.
“Obvious to try” with a reasonable expectation of success. Not mere hope. The team must have a rational, science‑based prospect that the trial will be both safe and effective, assessed holistically with reference to prior art, CGK, cost, effort, and alternatives.
Secondary evidence. Internal development documents are not primary proof of obviousness, but courts can use them to test whether expert narratives are plausible and to corroborate what the skilled team would have done.
Judicial comity. Irish courts consider foreign outcomes on the same patent as persuasive “reality checks,” especially where fact patterns and evidence align, but ultimately decide based on the record before them.

Conclusion

Sandoz v Bayer is a carefully reasoned and influential restatement of the Irish law of obviousness in the context of dosage regimen patents for high‑risk therapeutics. The Court:

Applied Actavis v ICOS and Pozzoli to find that once‑daily rivaroxaban was obvious to try with a reasonable expectation of success, given the combined teaching of the Kubitza and Harder Phase I posters.
Clarified that non‑validated PD biomarkers and research‑grade assays can legitimately inform Phase II dose and frequency selection; clinical correlation is not a pre‑condition to reliance in the obviousness analysis.
Confirmed that courts may assume the skilled team’s ability to perform routine steps to reach the state of the art; a “data deficit” in posters is not fatal where the missing work is standard.
Engaged meaningfully with foreign decisions, aligning with final UK invalidations and distinguishing EPC validations where evidential foundations diverged.
Used secondary evidence correctly as corroboration, not as a substitute for primary expert testimony.

The judgment provides practical guidance for litigants and R&D teams alike: early‑phase disclosures that explicitly point to a dosing regimen can, in combination with consistent PD data and routine development steps, render dosage claims obvious. It also sets a regulatory‑science‑aware benchmark for assessing the “reasonable expectation of success” in pharmaceutical cases under Irish law. As such, it will likely shape how future dosage regimen patents—especially in areas reliant on biomarker‑rich early trials—are drafted, prosecuted, and litigated.