The Melanoma Result

On 19 August 2026, Merck and Moderna announced that INTerpath-001 — a randomised, double-blind, placebo- and active-comparator-controlled Phase 3 trial in 1,137 patients — met its primary endpoint of recurrence-free survival and its key secondary endpoint of distant metastasis-free survival at an interim analysis.

That is a real and significant result. A double-blind trial of that size is the strongest instrument this field has, and it agreed with the smaller study that preceded it.

The release contains no hazard ratio, no confidence interval, no p-value and no percentage describing the size of that effect. It does contain hazard ratios — 0.51 for recurrence, 0.411 for distant metastasis — but those are the five-year results of KEYNOTE-942, the 157-patient trial that preceded this one, and the release says so. Of the 1,137-patient trial it is announcing, the release states that both endpoints were met and gives no figure for either. No numbers have been released since. The companies say These data will be presented at an upcoming international medical meeting and shared with regulatory authorities. Announcing topline results ahead of a conference presentation is routine, and this piece is not an accusation that anything improper happened. It is about what a reader can and cannot conclude in the interval.

Why the blinding row matters

KEYNOTE-942 was open-label: patients and doctors both knew which arm they were in, and the primary endpoint was judged by those same investigators. That is a genuine weakness. A doctor who knows a patient received an experimental therapy may, entirely unconsciously, scan and interpret differently.

The Phase 3 substantially reduces that problem. It is double-blind and placebo-controlled, so nobody assessing a recurrence knows which treatment produced it. It is the more trustworthy trial by a wide margin — seven times larger, and blinded against the direction the earlier trial’s assessment bias could run in. Not against investigator assessment itself, which both trials keep, as the next paragraph sets out. Which is precisely why it is frustrating that its numbers are the ones we do not have.

One detail has had little attention. Merck's own description of the Phase 3 defines its primary endpoint as recurrence "as assessed by the investigator" rather than by central adjudication. That leaves clinical judgement in the endpoint, though the double-blind design substantially protects against it running in one direction: an assessor who does not know the arm cannot favour it. The key secondary endpoint, DMFS, is investigator-assessed too. One caveat the figures on this page raise themselves: intismeran is an intramuscular injection that caused injection-site pain in 59.6% of patients and chills in 51.0% in the earlier trial; the Phase 3 comparator is a saline injection. Where a therapy is that much more reactogenic than its placebo, patients and their doctors can sometimes infer the arm — and the primary endpoint is investigator-assessed. Blinding on paper is not always blinding in the room. Central adjudication is the standard answer, and this trial does not use it. An independent review committee would move that judgement to blinded central assessors rather than remove it, and adjuvant melanoma therapies have been approved on investigator-assessed recurrence before. It is a difference worth knowing about, not a flaw.

And one thing almost nothing we read mentioned

Of the coverage we hold, two specialist outlets touched it, and they disagree. Morning Glory Sciences — we give its argument on its merits rather than on its authority — the Phase 2b population was stage IIIB–IV, the Phase 3 adds node-negative disease, and “absolute recurrence risk in that group is lower, so the same hazard ratio delivers a smaller absolute benefit.” Pharmacy Times makes the opposite case about the same patients: resected stage IIB or IIC melanoma “can face risks of recurrence and melanoma-specific mortality similar to those observed in stage III disease.” Neither reading appears in any of the general coverage we hold.

Which of them is right cannot be settled from anything published. It would take the Phase 3’s own results broken down by stage, and there are none. Morning Glory says as much in the same article, and it is the more important half of its argument: Because the Phase 3 hazard ratios have not been disclosed, those 157-patient figures cannot be placed alongside the 1,137-patient result. They belong to the Phase 2b population, not to this one. That is not a rhetorical point. We looked: no hazard ratio, interval or p-value for this trial appears in either company release, in any of the specialist or general coverage we hold, or in the trial’s own registry record, which as of 2 September 2026 carries no posted results at all and gives an estimated primary completion date of 26 October 2029 (NCT05933577). The August announcement was a prespecified interim look — both company releases say so. The registry does not: its record has not been updated since 24 September 2025, so it carries nothing about the announcement either way.

One hazard ratio does appear under the trial’s name, and it is this article’s subject happening in public. On announcement day a melanoma oncologist posted — in a roundup we hold — Exciting announcement today from Phase3 INTerpath001 followed by RFS HR=0.51 and DMFS HR=0.41. Those are the Phase 2b’s figures. The outlet that logged the post said so itself, warning on the same page that the Hazard ratios circulating on announcement day are these Phase 2b figures. Nobody released a Phase 3 effect size; a Phase 3 effect size circulated anyway.

So putting the two sets of figures side by side would mislead, and here is exactly how. The trials do not share a population: KEYNOTE-942 enrolled stage IIIB–IV, the Phase 3 enrolled stage IIB–IV, widening it downward — adding IIB, IIC and IIIA below the earlier trial's floor of IIIB. The upper end did not move: stage IV was in both. They do not share a design: the earlier trial was open-label, this one is double-blind with the outcomes assessor masked too. They do not share a statistical standing: the Phase 2b’s later analyses are, in its own words, descriptive only, against a one-sided alpha of 0.10, while the Phase 3 crossed a prespecified interim threshold. And most simply, there is nothing to put beside them: the Phase 3 has released no effect size, so quoting 0.51 next to “met its endpoints” is not a comparison but a substitution — the reader supplies the missing number from the smaller, older, differently-designed trial. The structure produces that effect even where every outlet attributes correctly, which is what the outlet-by-outlet check found.

Those newly-included patients start at lower risk of recurrence, which leaves less room for absolute benefit, and Pharmacy Times’ counter is that their risk may be closer to stage III than the staging suggests. Whether the effect holds up across that wider range is a real open question, and exactly the kind of thing a subgroup breakdown would answer. Merck and Moderna have not published one.

Two scores, not one, because these are two questions and one number cannot answer both. Is the effect real scores 3.94; how large is it scores 1.0. The gap between them is the story. Rigor scores 5 — a double-blind, placebo-controlled, 1,137-patient Phase 3 is as good as trial design gets. Data support scores 1 — the rubric's anchor for 1 is purely qualitative assertion with no numeric support, which is what a statement that two endpoints were met, with no figure for either, is. Reproducibility scores 4 because the earlier Phase 2b and the much larger Phase 3 point in the same direction on RFS and DMFS, but no Phase 3 effect size has been released, so the Phase 2b magnitude has not been reproduced. Consensus scores 4 because the held sources agree on the directional Phase 3 result while the missing effect size leaves no numerical magnitude to agree on. Recency scores 5 because the central evidence is the August 2026 Phase 3 announcement, supported by 2026 longer-term follow-up of the earlier trial.

So the study is excellent and the evidence released about it is almost nothing. Those are different things, and until 3 September this scorecard averaged them into a single number that was wrong about both halves — understating the trial and overstating what is known about the size of what it found. Readers saw 3.4, which was the correct total of the working printed beneath it — but that working weighted two of the six dimensions differently from the rubric, which puts the same six scores at 3.35. That is recorded below. An outside reviewer said so and the scorecard was split. Only one of the six dimensions, data support, asks whether an effect size exists at all, so the magnitude score rests on that dimension alone; the rubric says so rather than hiding it. Source quality scores 3 rather than 5 for the same reason: the claim currently rests on a corporate press release, which the rubric ranks as industry analysis, not the peer-reviewed publication it will eventually become.

This is what a headline cannot do. “Landmark trial succeeds” and “real: 3.94, size: 1.0” describe the same event, and only one of them tells you the numbers are still missing.