Oncolytic Vaccinia Virus VET3-TGI Shows 71% Response Rate in Checkpoint-Refractory Skin Cancers

At the 2026 Society for Immunotherapy of Cancer Annual Meeting in Phoenix, researchers presented early results for an engineered virus injected straight into tumors. The agent, VET3-TGI, was given alone or with the checkpoint inhibitor atezolizumab (Tecentriq) in the phase 1/1b STEALTH-001 trial (NCT06444815). The activity was concentrated in patients with skin cancers that had stopped responding to immunotherapy, and no dose-limiting toxicities were reported.

Among 12 efficacy-evaluable patients treated intratumorally, the objective response rate by blinded independent central review was 42%. That comprised one complete response and four partial responses, and the disease control rate was 75%. In the cutaneous subgroup of seven patients, the response rate was 71%. Three patients had Merkel cell carcinoma: one achieved a complete response and two achieved partial responses. Three had melanoma: two had partial responses and one had progressive disease as best response. The one patient with cutaneous squamous cell carcinoma progressed.

The median duration of response had not been reached. At the September 2026 data cutoff, every responder was still on study treatment, for periods of 10 to 19 months. Responses showed up in lesions that were injected and in lesions that were not.

That second point drew the most emphasis from lead author Yana G. Najjar, MD, associate professor of medicine and director of the Clinical and Translational Research Center at UPMC Hillman Cancer Center in Pittsburgh. She told OncLive that responses in non-injected lesions have been consistent, and called that the holy grail of intratumoral therapies. She said the virus adds evidence that its therapeutic payloads are working.

VET3-TGI is a Western Reserve vaccinia virus modified to express CXCR3, which lets it attach to lymphocytes and travel to tumors while avoiding neutralization. It also carries two payloads: a TGF-β inhibitor that originated at UPMC Hillman Cancer Center, and interleukin-12. Najjar noted that patients with advanced skin cancers who progress on checkpoint blockade have few options.

The open-label, first-in-human trial began with a 3+3 dose escalation in a mixed solid tumor population, testing monotherapy from 1 × 10⁸ to 3 × 10⁹ plaque-forming units with a 28-day window for dose-limiting toxicity. Investigators are now enrolling a monotherapy expansion limited to cutaneous cancers and a combination cohort pairing the virus, at or below the top monotherapy dose, with atezolizumab. Injections were given on days 1, 8, 15 and 22, followed by optional boosters as often as every two weeks. Atezolizumab was dosed at 840 mg every two weeks for four doses, then 1680 mg every four weeks. Primary end points were safety, dose-limiting toxicities and the recommended phase 2 dose.

Twenty patients had received intratumoral VET3-TGI, 16 on monotherapy and four on the combination. The cutaneous patients were heavily pretreated, with a median of three prior lines of therapy (range 1 to 9). All had received an anti–PD-(L)1 agent and were refractory. Of them, 42.9% had also had a CTLA-4 inhibitor plus anti–PD-(L)1, 57.1% had stage IV disease, 85.7% had lactate dehydrogenase above the upper limit of normal, and 57.1% had visceral metastases.

Paired biopsies showed more immune activity in all five evaluable injected lesions and all three evaluable non-injected lesions. In injected lesions, cytotoxic CD8+GRB+ T cells rose a median 7.2-fold and proliferating CD8+Ki67+ T cells 10.4-fold at days 15 to 22. In non-injected lesions, the day 15 increases were 2.8-fold and 3.1-fold. Investigators called these descriptive findings a rationale for durable responses, consistent with turning immunologically cold tumors hot.

Treatment-related adverse events were common but mostly low grade. Cytokine release syndrome occurred in 50% of patients, all grade 1 (20%) or grade 2 (30%). Chills and nausea each affected 40%, vomiting 35%, and decreased lymphocyte count 25%. Grade 3 events included lymphocyte count decrease (10%) and, at 5% each, nausea, vomiting, anemia, hyponatremia and raised alanine aminotransferase. Grade 4 events were decreased lymphocyte count (10%) and decreased white blood cell count (5%). There were no treatment-related deaths. Najjar said the lymphopenia improved over time and the profile was in line with oncolytic viruses.

Enrollment of cutaneous patients continues to shape a phase 2 design, and the intravenous arms of STEALTH-001 will be reported separately.