The question that has shadowed every personalized cancer vaccine trial for the past decade is not whether the approach can work — it is whether the work lasts. At the Society for Immunotherapy of Cancer’s (SITC) 41st Annual Meeting in Phoenix today, Danish biotech Evaxion A/S is presenting data that pushes that question into new territory: not just that responses are durable, but how the body’s immune architecture looks after three years of that durability.

The oral presentation — Session 105b, Abstract #502, 12:30 PM Mountain Time, Room 301BC at the Phoenix Convention Center — covers EVX-01, Evaxion’s personalized peptide cancer vaccine designed entirely by its AI-Immunology™ computational platform. The clinical data backs a 75% objective response rate (ORR) in 16 patients with advanced, unresectable melanoma treated in combination with Merck’s pembrolizumab (KEYTRUDA®). That figure matters because pembrolizumab alone historically produces ORRs in the range of 33–45% in first-line metastatic melanoma, making the 75% combined result a clinically meaningful signal — even given the trial’s small size. At two years, 92% of responders showed no disease progression and four patients achieved complete responses.

But today’s data goes further. Senior Scientist Mads Lausen of Evaxion will present systemic immunomodulatory findings that move beyond counting vaccine-specific T-cells and examine whether EVX-01 is reshaping the immune system at a broader level — cytokine dynamics, immune-cell subset changes, and systemic biomarkers that could illuminate why responses persist.

“Right at a moment when the entire field is asking whether personalized cancer vaccines have finally arrived,” said one observer, the timing is deliberate. At the same SITC meeting, Dana-Farber Cancer Institute’s Dr. Catherine J. Wu — one of the architects of neoantigen vaccine science — is delivering the 2026 Smalley Award Lecture, titled “Getting Off the Rollercoaster: Personalized Cancer Vaccines Come of Age.” That title is not incidental. For years, clinical results in this space swung between promise and disappointment. The 2026 SITC program reflects a consensus that the roller coaster may finally be leveling out.

How the AI Designs the Vaccine

EVX-01 is not an off-the-shelf drug. It is built from scratch for each patient using Evaxion’s AI-Immunology™ platform, which works through a chain of machine-learning decisions before a single peptide is manufactured.

First, the patient’s tumor DNA is sequenced to identify somatic mutations — the genetic errors that cancer cells accumulate. Each mutation creates a mutant protein sequence not found in normal tissue. Those mutant sequences are potential neoantigens: molecular flags that, in theory, the immune system should recognize as foreign. The practical challenge is that not every mutant sequence gets presented on the cell surface in a form T-cells can see. Presentation depends on the patient’s specific HLA type, the protein binding groove that “shows” peptides to circulating immune cells — and that HLA type is different for every patient.

Evaxion’s platform models HLA-peptide binding probability for each candidate neoantigen, assigns a prediction score, and selects the top-ranked targets for the patient’s custom vaccine. The validation of that selection logic is the Phase 2 trial’s most technically important finding: an 89% neoantigen recognition rate across all enrolled patients, combined with a statistically significant correlation between AI prediction score and T-cell response magnitude (p = 0.00013). In plain terms, the neoantigens the AI ranked highest were the ones that actually triggered the strongest immune reactions. That correlation is not guaranteed — it is the platform’s core value proposition, and the data supports it.

Three-Year Durability: What the Extension Reveals

The clinical milestone that has drawn the most attention is the three-year follow-up, first presented in poster form at the European Society for Medical Oncology (ESMO) Congress in Madrid in October 2026. Among all responding patients assessed at the three-year mark — including a subset of six who entered a one-year monotherapy extension phase, receiving EVX-01 without pembrolizumab — no relapses were observed. A majority of responders showed continued deepening of their tumor reduction, and 15 of 16 enrolled patients achieved measurable lesion reduction.

The monotherapy extension data is the mechanistic crux. Checkpoint inhibitors like pembrolizumab work by releasing the immune system’s brakes — they block the PD-1 pathway that tumor cells exploit to suppress immune activity. When pembrolizumab is stopped, that suppression mechanism can reassert itself. The fact that responding EVX-01 patients maintained their remissions after PD-1 blockade ended is consistent with one hypothesis: that the vaccine-primed T-cells established durable immune memory that sustains anti-tumor surveillance independently of ongoing checkpoint inhibition. Confirming that hypothesis is precisely what today’s systemic immunomodulation data is designed to address.

Conference Trajectory and Competitive Context

Evaxion’s EVX-01 program has been building a conference record. Two-year efficacy data was presented in a mini-oral session at ESMO 2025 in Berlin in October 2025, and three-year data appeared as a poster at ESMO 2026 in Madrid last month. Receiving an oral presentation slot twice at major oncology conferences — including today’s selection for SITC’s Rapid Oral Clinical series — signals sustained peer-validated interest in the data package.

The direct competitive comparison in the neoantigen vaccine space is Moderna’s individualized neoantigen therapy, mRNA-4157 (V940), which is now in Phase 3 trials with Merck. The two programs target different patient populations and use different delivery mechanisms — Moderna’s candidate is an mRNA-based vaccine administered post-surgery to prevent recurrence in resected high-risk melanoma (the adjuvant setting), while EVX-01 is a peptide-based vaccine targeting patients with active, unresectable metastatic disease (the first-line metastatic setting). The clinical populations are different enough that head-to-head comparison is not straightforward, but the two programs together represent the leading edge of AI-informed neoantigen vaccine development.

Beyond Melanoma: EVX-04 and EVX-05

Evaxion is using EVX-01’s melanoma results as proof of platform rather than a standalone program. Two pipeline candidates extend the AI-Immunology approach into harder targets.

EVX-04 targets acute myeloid leukemia (AML) and was the subject of an oral presentation at ASH 2025 in Orlando, where preclinical T-cell data demonstrated that AI-designed vaccine candidates elicited strong immune responses.

EVX-05 is Evaxion’s most technically ambitious program: an off-the-shelf glioblastoma vaccine that bypasses the neoantigen framework entirely. Rather than targeting patient-specific mutations, EVX-05 targets endogenous retroviral elements (ERVs) — remnants of ancient retroviral infections that account for roughly 8% of the human genome and are normally silenced but become aberrantly expressed in some cancers. Because glioblastoma carries very low tumor mutational burden (limiting neoantigen-based approaches), ERVs offer a different antigen source that could enable shared, non-personalized vaccine design. Proof-of-concept data from a collaboration with Duke University was presented at AACR 2026 in San Diego in April 2026. Evaxion’s AI-Immunology™ platform received the 2026 Prix Galien UK Award for Best Digital Health Solution.

Small Trial, Large Questions

Precision requires acknowledging what the EVX-01 data is not. A Phase 2 trial with 16 patients is a compelling signal — not a definitive proof. It is open-label (no placebo arm) and single-arm (no untreated control group), which means the incremental contribution of the vaccine itself, beyond pembrolizumab alone, cannot be formally quantified without a randomized comparison. The 75% ORR is striking relative to historical pembrolizumab monotherapy benchmarks, but historical comparisons carry significant caveats in oncology, where patient selection, center volume, and treatment sequencing all affect outcomes.

A larger randomized trial is what the field needs and what investors will watch for. Evaxion’s Q2 2026 financial results confirm cash runway into the second half of 2027, and the company’s quarterly financial results for Q3 2026 are scheduled for November 19. Business development discussions — the route to a Phase 3 partnership — are the logical next chapter if today’s SITC presentation strengthens the data package’s reception.

Why Today’s Session Matters

For a TechTimes audience, the most significant dimension of EVX-01’s story is not the clinical result alone — it is what the AI-Immunology™ platform’s neoantigen selection accuracy (89% recognition rate, p = 0.00013 score-to-response correlation) suggests about the future architecture of cancer treatment. Personalized medicine has been a promise in oncology for a decade. What Evaxion is presenting today is a platform that takes a patient’s unique tumor fingerprint, identifies the molecular targets most likely to draw a T-cell response, and then validates those targets in a living immune system years later.

The systemic immunomodulation data at SITC 2026 is the next layer: not just “did the T-cells respond,” but “did the immune system change in a way that explains long-term survival.” If the data shows that EVX-01 reshapes immune architecture at a system level — not merely producing tumor-specific T-cells but reprogramming how the immune environment operates — it would add biological depth to a clinical story that already has three years of durability behind it.

That is what an AI-designed vaccine is supposed to do. Today in Phoenix, Evaxion is presenting the evidence that it may.

SITC 2026 Annual Meeting runs November 4–8, 2026, at the Phoenix Convention Center. Evaxion’s oral presentation (Abstract #502, Session 105b) takes place today, Friday, November 6, 12:30–1:30 PM MT, Room 301BC. A companion poster presentation is scheduled for Saturday, November 7, 12:30–1:30 PM MT.