When the Scope Outruns the Evidence

In June 2026 the European Union published the first Joint Clinical Assessment report in its history. It concerns tovorafenib (Ojemda), a targeted therapy for pediatric low-grade glioma, and on April 30 it was endorsed by the Member State Coordination Group under Article 12(2) of the HTA Regulation, the step that makes the assessment final. That endorsement is of the assessment, not the medicine: the report carries no benefit verdict and binds no national decision. The assessor at Ireland's NCPE, the co-assessor at Germany's IQWiG, and the Coordination Group behind them have earned the milestone they mark. European cooperation on the clinical assessment of new medicines is no longer an aspiration. It is a published document.

‍

Germany's national benefit assessment for the same medicine, the AMNOG procedure, opened on May 15, two weeks after the European report was endorsed. What opened on May 15 is the country that co-wrote the European assessment beginning to decide whether its national procedure will honor that assessment or reassert its own.

‍

The assessment scope for Ojemda set eight PICOs, eight defined comparisons across three patient populations. The published report answers one of them.

‍

For six of the eight, the developer submitted no comparative data: in the report's plain phrasing, none was available. For a seventh, data were submitted and the assessors set them aside as insufficient. The one that remained, in the BRAF V600E population, entered as an unanchored matching-adjusted indirect comparison, carrying what the assessors call major uncertainties and built on an effective sample size of 5.81 to 14.26 across outcomes, a third to a half of an already small sample. Eight questions asked, one answered, and the report spends much of its length qualifying that answer.

‍

‍

The gap the ratio exposes

That ratio is not the most interesting part. What it points to is more consequential than another readiness checklist and has been under-addressed by much of the commentary so far. I have called this the operationalization gap: the distance between what a JCA scope demands and what an asset's evidence base can actually produce. EFPIA warned of it in late 2025, cautioning that a consolidated scope could impose "an evidence standard that cannot be satisfied." The first report makes that warning concrete, and the eight comparisons did not fail for a single reason.

‍

For six of them, the report states that no comparator data exist in the target population, anywhere, to build against. The pivotal trial is single-arm and non-comparative, so there is no randomized anchor and no connected network to draw on. For those six, this is not a gap better foresight could have filled. It is one the evidence environment cannot supply.

‍

The seventh is a different case, and the difference is worth holding onto. There, comparator evidence did exist, but it reached the assessors only as conference abstracts, from which neither the methods nor the population could be verified. That study was run by an academic group, not a competitor; its fuller evidence, in theory, could have been obtained through the kind of investigator collaboration teams arrange when they plan years ahead of a foreseeable assessment. The six mark the limit of the evidence environment. This seventh marks the limit of how the evidence was orchestrated.

‍

Some will read this as a story about methods being too hard. At a regulatory meeting in Lisbon in May 2026, Anne Willemsen, who co-chairs the Coordination Group's scoping subgroup, noted that developers can do a great deal with indirect comparisons, and where one is not possible, they can say so and justify it. The toolbox is wide, and rigor is the price of entry. The bottleneck is not which methods are allowed but whether the comparisons a scope demands can be generated at all. Execution still matters, and here some of it fell short: on the one comparison that survived, the report is unsparing, citing a sensitivity analysis with errors and inaccurate conclusions, a primary response analysis set aside as inappropriate, and patient data left out without adequate justification.

‍

The report takes on the strict, uncertainty-forward posture German methods are known for, and that tilt is no accident; it reflects Germany's weight in the system, and it means every evidence team with European ambitions is quietly choosing which methodological anchor its global value dossier is built on. That choice is worth making deliberately rather than by inheritance, because an anchor left unquestioned is still a bet, only an unpriced one. The question a sophisticated evidence leader should be able to answer on demand is simple to state and uncomfortable to sit with: which anchor are we built on, and have we priced the alternative?

‍

The report exposes a downstream tension too: even a comparison that can be generated may not survive in a form a national payer can use. The one that did was allowed, then heavily discounted, its estimates marked as not to be read as causal. Aballéa and colleagues recently argued that Europe's guidance on indirect comparisons is rigorous to the point of being restrictive, and that rigor at the European level can become a relevance problem at the point of national decision. The first report is that argument observed. Carried far enough, the stringency has its own cost: at the limit, developers withdraw from or decline to launch a market rather than accept a price that does not reflect value.

‍

‍

Why a gap does not stay in Europe

The publication itself changes the strategic picture. The seven unanswered comparisons are not a private matter between a developer and one agency. They sit in a public report, on a public portal, beside the developer's dossier, which the Coordination Group has classified as not commercially confidential at the level of clinical methods and results. Anyone can read which questions went unanswered, and why.

‍

That public visibility is why an evidence package now must hold on more than one front at once. Nationally, it must stand on its own, as Germany is already deciding whether the European findings "can" be referred to, as its federal committee allows, or "must" be used, as its industry argues. At the European level, it must stay usable if those findings are honored at all. The Commission calls the new system "clearer and more predictable"; a first report that answers one of eight questions is a fair measure of how far that promise still must travel. And it must anticipate a third front that runs outside Europe entirely.

‍

A comparison skipped or unfilled does not stay local once it is recorded as a public gap. Payers and authorities well beyond the Union read these reports, and a gap in a JCA can resurface far from where it was set. The clearest case is the United States, where private plans and value assessors increasingly cite international evidence reviews and nothing requires them to read a European gap charitably. The exercise is European. Its reach is not.

‍

A quieter thread also deserves naming. The entire patient voice in Europe's first Joint Clinical Assessment was one carer, invited to submit written comments alongside a single clinical expert. That is the structure as built, and it is worth seeing clearly before the field decides it is enough. How patient-relevant evidence holds its weight as it crosses from a European report into national decisions is already live, and whether that input stays written comment or becomes something more deliberately structured is part of what is being decided. I will return to it in a later piece.

‍

The exercise is European. Its reach is not.

‍

‍

What comes next, and who will be ready

For now, I read this less as a finish line than as the first data point in something we get to watch in real time. Germany co-assessed the report, and its methods are the spine the assessment rests on, which gives what comes next a particular edge: the country that helped write the European assessment is also the first to rule on it. Germany's benefit assessment will tell us whether "can refer" behaves like "must," with a decision on clinical benefit expected in November. Over a longer horizon we will learn what stays in Europe and what travels. Ojemda is an orphan medicine; the first non-orphan case will be the harder, cleaner test. And more reports are due within months.

‍

One more thing shapes who will be ready, and it sits upstream of the analytics. Speaking at the 2026 RAPS Euro Convergence, and reported on the record in the trade press, the market-access consultant Thomas Ecker called JCA preparation "a multi-year project" that "cannot be done over the weekend": deciding who owns the work, who runs the statistics, who signs off, and how conflicts are resolved. Guidelines, he noted, tell only "half of the truth"; what decides the outcome is how they are used in practice. For an emerging biotech without deep in-house evidence methods, that is the most acute exposure of all, because even strong analytics have no operational home until those questions are answered first.

‍

What now separates the developers who are ready from those who are not, is not how much evidence they hold, but whether they build evidence architecture for the system that now exists or keep administering anchors set for the one it replaced. One reads the environment and builds ahead of what it will ask for; the other waits to be asked. That distance is exactly the work that this first report made impossible to ignore.