Merck and Moderna published results from a study of 1,137 patients with high-risk melanoma after surgical tumor removal. The combination of standard Keytruda immunotherapy and a personalized mRNA vaccine cut the risk of recurrence and distant metastases more effectively than baseline therapy.
How AI picks the targets
AI's role here is central. There's no one-size-fits-all drug, since every tumor carries its own unique set of mutations. Here's how it works: doctors sequence the tumor cells and the patient's blood, and Moderna's machine learning algorithm analyzes the data and predicts up to 34 targets most likely to trigger an immune response.
The mRNA vaccine gets printed to match that individual list. In effect, phase 3 was a test of how good the algorithm's predictions actually are. And the model keeps learning — it's continuously retrained on new clinical data, sharpening how it picks targets.
What phase 2 showed
Exact phase 3 efficacy numbers aren't out yet — those will be presented later at a medical conference. For now, the benchmark is the phase 2 results: back then, the five-year risk of recurrence or death dropped by 49%. Worth noting the sample was small, though — just 157 people.
Market reaction
The market reacted exactly how you'd expect: Moderna's stock nearly tripled in a single day. The same AI-driven target-selection approach is now being tested against lung, kidney, and bladder cancer.