AI Helped Build a Personalized Cancer Vaccine. It Just Passed Phase 3

Merck and Moderna announced positive topline results on August 19 from a Phase 3 trial testing a personalized cancer vaccine, formally known as an individualized neoantigen therapy (INT), alongside the checkpoint drug KEYTRUDA in melanoma patients whose tumors had already been surgically removed. The companies say it is the first positive Phase 3 trial of an INT and of any mRNA-based cancer therapy.

The therapy at the center of the trial, intismeran autogene, is built differently from a standard drug. Instead of one fixed formula, an algorithm helps select each patient's specific targets, based on their own tumor.

How The Algorithm Selects The Vaccine's Targets

Making intismeran starts with a sample of a patient's tumor and blood, sequenced to map its mutations. Moderna says a series of AI algorithms then reviews those mutations and predicts which resulting proteins, known as neoantigens, are most likely to trigger an immune response. Up to 34 of the top-ranked neoantigens are encoded onto a single strand of mRNA, wrapped in a lipid nanoparticle, and given to the patient.

The vaccine plays a different role from checkpoint drugs alone, says Diwakar Davar, vice chief of academic affairs at UPMC Hillman Cancer Center's melanoma program. "The drug works by augmenting DCs which aren't licensed by checkpoint blockade," Davar said to AIM Media House.

KEYTRUDA works by blocking PD-1 signaling, which increases the immune system's ability to detect and fight tumor cells that it can already recognize. Davar's point is that the vaccine adds something upstream of that: it activates dendritic cells, the immune system's scouts, to help generate new tumor-specific T-cell responses in the first place.

What The Trial Showed, And What Is Still Unknown

INTerpath-001 enrolled 1,137 patients with completely resected Stage IIB to IV melanoma, none of whom had received prior systemic treatment. Patients were randomized two to one to receive either the vaccine plus KEYTRUDA or KEYTRUDA alone.

At a planned interim analysis, the combination met its primary endpoint of recurrence-free survival and a key secondary endpoint of distant metastasis-free survival, both measured against KEYTRUDA alone, with no new safety signals reported. The announcement reported topline results only. Detailed efficacy figures, including hazard ratios, have not yet been released.

The result builds on five-year follow-up data from the earlier Phase 2b KEYNOTE-942 trial, presented at the 2026 ASCO Annual Meeting, which showed the same combination cut the risk of recurrence or death by 49% and the risk of distant metastasis or death by 59%, compared with KEYTRUDA alone. Merck and Moderna say they plan to present full Phase 3 data at an upcoming medical meeting and have begun discussions with regulators about a filing.

The result could shift how the disease is treated, says Ryan Sullivan, a physician at Massachusetts General Hospital and a co-author on the KEYNOTE-942 study. "Assuming the data looks as great as it sounds in the press release, this will change the way we treat patients with resected melanoma," Sullivan said to AIM Media House. "Further, the impact of a positive phase 3 trials of a cancer vaccine has the potential to catalyze the field to leverage this and other approaches to improve cancer therapy more broadly."

Beyond Melanoma

Intismeran is one of nine trials currently running in Moderna and Merck's INTerpath program, which includes studies in lung, bladder and kidney cancers using the same individualized neoantigen approach. Efficacy outside melanoma remains unproven, since only the melanoma trial has reported results so far.

Intismeran is not the only AI-linked compound to clear Phase 3 this year. Zasocitinib, a psoriasis pill that Takeda acquired in 2023 from Nimbus Therapeutics, a company co-founded by the computational drug design firm Schrödinger, met its primary and secondary endpoints in two Phase 3 trials for plaque psoriasis in December 2025, with more than half of patients treated reaching clear or almost clear skin by week 16, according to Takeda. The two results, in unrelated diseases and built on different computational approaches, are separate signs that AI-linked drug candidates are starting to clear the Phase 3 bar this year.

For patients, the significance goes beyond the trial data itself, says Stephanie Kauffman, CEO of the Melanoma Research Alliance. "This is an important milestone for melanoma patients and for a field that has spent decades working toward more precise, personalized approaches to cancer treatment," Kauffman told AIM Media House. "The idea that we may be able to use the unique characteristics of an individual patient's tumor to train their own immune system to help prevent melanoma from returning is incredibly powerful."

"We look forward to seeing the full data and the regulatory review, but these results offer real reason for optimism and underscore how advances in immunotherapy, genomics and technology are continuing to change what may be possible for melanoma patients and possibly for other cancer indications in the future," she says.