Merck and Moderna clear the Phase 3 bar on a personalized cancer vaccine, minus the numbers

Merck and Moderna said Wednesday that a personalized mRNA cancer vaccine, added to an existing treatment, slowed the return of melanoma and its spread to other parts of the body in a late-stage clinical trial ^[1]. That is the first randomized Phase 3 result aimed at definitively proving that neoantigen vaccines work as cancer treatments ^[2], and the companies did not immediately release the detailed data behind it ^[3].

The specifics that were disclosed matter. The vaccine, intismeran, was combined with Merck's Keytruda in adjuvant melanoma, meaning the patients' disease had already been surgically removed ^[4]. Adjuvant is the friendliest setting for this idea: low or no measurable tumor burden, an immune system not yet exhausted by bulk disease, and an endpoint built around whether and when cancer comes back rather than whether a mass shrinks. The reported effect matches that design, a delay in recurrence and in spread to distant sites ^[1].

Context explains why this is being treated as a platform result rather than one drug's win. Neoantigen vaccines, which are built against mutations specific to an individual patient's tumor, have long been seen as having potential in cancer ^[6], and in 2024 early data on this vaccine served as a turning point for a field that had been floundering ^[5]. According to STAT, if the results hold up they could herald a new and powerful approach in oncology and stand as further evidence for mRNA as a platform for both conventional vaccines and therapeutics ^[7]. That last point is the commercial subtext: the same manufacturing logic that produced Covid shots is now being pointed at bespoke, one-patient-at-a-time products.

What has not been supplied is everything an operator would need to size the result. A statement that a trial met its endpoint carries direction but no magnitude, and no detailed data has been released ^[3]. Absent a hazard ratio, a follow-up duration, a curve shape, and a safety table, "slowed the return" is compatible with an effect that changes clinical practice and with one that does not survive contact with longer follow-up. The 2024 data that revived the field was early data ^[5], and early signals in immuno-oncology have a history of shrinking.

The second unresolved question is industrial rather than statistical. A personalized vaccine is a manufacturing commitment as much as a biological one, because each course is specified by one patient's tumor ^[4], and a positive Phase 3 in adjuvant melanoma converts that from a research workflow into a scheduling and logistics problem tied to a surgery date.

Worth watching: the full dataset, including effect size, follow-up time, and adverse events, which is what will determine whether this is a durable benefit or a modest delay ^[3]; whether the same approach reads out in tumor types other than melanoma, since the platform claim rests on generality rather than one indication ^[7]; and how quickly individualized doses can be produced after surgery, given that the tested regimen depends on adjuvant timing ^[4].