Intratumoral IL-12 mRNA Shrinks Distant Lesions in PD-1–Resistant Melanoma

A 33% confirmed objective response rate in PD-1–resistant melanoma — the trial's predefined Simon two-stage success threshold — anchors the first-in-human phase 1/2 data for STX-001, a lipid nanoparticle (LNP)–encapsulated, IL-12–expressing self-replicating mRNA injected directly into tumors. Presenters reported the findings, from trial NCT06249048, at the 2026 SITC Annual Meeting in Phoenix (November 4–8; abstract 823), and the responses were not confined to the injected lesion.

That is the headline for anyone treating advanced melanoma after checkpoint failure: tumor regression occurred in both injected and non-injected lesions, which lead author Georgina V. Long, AO, PhD, MBBS, medical director of Melanoma Institute Australia, characterized as a clear abscopal effect. "We want to induce enhanced antigen presentation [and] interferon-γ programs in the injected lesion, and then increased T cells that then travel to the non-injected lesions, and that's certainly what we see," Long said at the presentation.

What does the response data show?

Among 36 evaluable patients across all dose levels, more than 40% experienced tumor regression, with activity at every dose tested. In the 100-µg melanoma expansion cohort (n = 15), the confirmed ORR per RECIST was 33% (5/15), and more than half of that cohort had some degree of tumor shrinkage. Overall, 48.7% of patients (19/39) remained on study at least 6 months. Median duration of response was 8.7 months, with the cutoff hit when the last patient reached 6 months and many responses still ongoing.

One case stands out. A 72-year-old woman whose melanoma had progressed on PD-1, CTLA-4, and LAG-3 blockade received 30 µg STX-001, showed apparent progression at week 12, then achieved a complete metabolic response by week 30.

How was the trial built?

Eligible patients had advanced solid tumors, ECOG 0–1, adequate organ function, and at least one injectable, biopsiable lesion. Melanoma patients had to meet SITC criteria for confirmed PD-1 resistance. Investigators dosed STX-001 intratumorally once every 3 weeks, reinjecting the same lesion when possible. Dose escalation used a Bayesian optimal interval design — 30, 100, 300, and 900 µg — with a 10-µg cohort added later for optimization, and the 100-µg cohort expanded via Simon two-stage design. Primary endpoints were safety and dose selection; pharmacokinetics were reported previously at ASCO 2025.

The 39 enrolled patients had a median age of 60 (range 23–83), 51.3% were female, and 64.1% had ECOG 0. They'd seen a median of 3 prior therapy lines (range 0–9), including 2 lines of immunotherapy (range 0–7). Melanoma dominated at 84.6%; breast cancer, head and neck cancer, and breast angiosarcoma made up the rest. In the 100-µg melanoma cohort, all 15 patients had prior anti–PD-1 therapy, 53% had added anti–CTLA-4, and 40% had added anti–LAG-3.

What did safety show?

All patients had AEs; grade 3 or higher AEs hit 53.8% overall but 41.2% in the 100-µg cohort. Treatment-related AEs appeared in 97.4% overall (38.5% grade 3+), versus 94.1% and 23.5% at 100 µg. Long noted the grade 3 events were mostly transient blood count changes consistent with IFN-γ activity. Only 3 patients (7.7%) discontinued for TRAEs — none in the 100-µg cohort. CRS occurred in 35.9% overall (one grade 3+ case) and 29.4% at 100 µg with no grade 3+ events. One 900-µg patient developed sub–grade 3 ICANS.

Why did it work?

Paired RNA sequencing of injected lesions showed significant IFN-γ program upregulation in responders (n = 15) but not nonresponders (n = 13). Multiplex immunofluorescence told the mechanistic story: benefit-group patients showed median fold increases of 3.8x in CD8+ T cells, 5.4x in Th1 CD4+ helpers, 5.6x in Tc1 CD8+ effectors, and 6.2x in CD86+ antigen-presenting cells — versus roughly flat changes in non-benefit patients. Crucially, non-injected lesions in benefit patients showed 5.2x CD8+ and 4.5x Th1/Tc1 increases, consistent with primed T cells trafficking to distant disease. Investigators cautioned the data capture system didn't record non-injected non-target lesions, so waterfall plots undercount distant shrinkage.

The 100-µg dose now carries the program forward, with translational data pointing to IFN-γ signaling as the response biomarker to watch in larger melanoma studies.