Why This Study Matters
STAT3 promotes a pro-oncogenic tumor microenvironment. PEMDA-HN is described by the authors as the first randomized trial of a STAT3-targeting antisense oligonucleotide (danvatirsen) combined with pembrolizumab, tested in first-line, PD-L1-positive (combined positive score [CPS] ≥1) recurrent or metastatic head and neck squamous cell carcinoma (R/M HNSCC). The authors note that therapeutic progress beyond single-agent PD-1 inhibitors or chemoimmunotherapy in this setting has been limited. Pembrolizumab is the only immunotherapy approved as monotherapy in the first-line, biomarker-positive R/M HNSCC setting.
69 patients randomized; 66 treated (45 vs 21); 22 centers in 3 countries; 2:1 randomization; danvatirsen 3 mg/kg IV weekly plus pembrolizumab 200 mg IV every 3 weeks.
Combination danvatirsen + pembrolizumab did not improve ORR compared to pembrolizumab alone in first-line treatment of R/M-HNSCC.
— Study abstract (Conclusions)
Study Design
This was a multicenter, randomized, open-label, controlled phase 2 study at 22 centers in 3 countries (United States, United Kingdom, South Korea). Patients were randomized 2:1 by a centrally managed interactive response technology system and stratified by CPS (1–19 vs ≥20). Treatment ran in 21-day cycles until progression, death, unacceptable toxicity, or withdrawal. The design was a 2-arm Bayesian optimal phase 2 (BOP2) design, optimized under H0: ORR 0.15 vs 0.43 for the combination, with 83% power at 1-sided 5% alpha.
Arms: (1) danvatirsen 3 mg/kg IV on Days 1, 3 and 5 of Cycle 1 Week 1, then weekly, plus pembrolizumab 200 mg IV every 3 weeks; (2) pembrolizumab 200 mg IV every 3 weeks alone. The primary endpoint was confirmed ORR (partial + complete response, RECIST v1.1). Secondary endpoints included safety, duration of response, disease control rate, and progression-free survival.
Funding: Flamingo Therapeutics (NCI grant UG1CA233247 to NFS). Data cutoff: 19 Jun 2025.
Patient Population
Eligible patients had previously untreated R/M HNSCC with CPS ≥1, ECOG performance status 0 or 1, and measurable disease. Between 2 January 2024 and 12 May 2025, 69 patients were randomized (45 to the combination, 24 to pembrolizumab alone); 3 patients assigned to pembrolizumab alone withdrew before treatment, leaving 66 treated. Mean age was 64.7 years, 77.3% were male, 66.7% had ECOG 1, 24.2% were HPV positive, and 71.2% had CPS ≥20. The study was stopped after planned and unplanned interim analyses; enrollment closed on 01 June 2025.
Primary Endpoint Results
| Endpoint | Danvatirsen + Pembrolizumab (N = 45) | Pembrolizumab (N = 21) |
|---|---|---|
| ORR, n (%) | 7 (15.6) | 3 (14.3) |
| ORR 90% CI | 7.5, 27.2 | 4, 32.9 |
| Risk difference (90% CI) | 1.3 (−14.1, 16.7); Fisher's exact 2-tailed P = 1.0000 | Between-arm comparison |
| Median DOR, months (95% CI) | 6.36 (5.68, not estimable) | Not estimable |
| Median PFS, months (95% CI) | 3.52 (1.48, 6.83) | 2.92 (1.41, not estimable) |
Subgroup Analyses
| Subgroup | Combination ORR, n (%) [90% CI] | Pembrolizumab ORR, n (%) [90% CI] | Risk difference (90% CI) |
|---|---|---|---|
| CPS ≥20 | 6 (18.2) [8.2, 32.8] | 3 (21.4) [6.1, 46.6] | −3.2 (−24.4, 17.9) |
| CPS <20 | 1 (8.3) [0.4, 33.9] | 0 [0, 34.8] | 8.3 (−4.8, 21.5) |
Per the planned futility rule, enrollment in the CPS 1–19 subgroup was discontinued because ≤1 of 10 patients receiving the combination responded within 12 weeks. A second, unplanned, efficacy-only interim analysis in the CPS ≥20 group reported ORRs of 22.22% (combination) and 20.00% (pembrolizumab alone) before the study was stopped. The authors state the small sample size precluded deeper analysis by HPV or other biomarkers.
Safety Profile
Treatment-emergent adverse events (TEAEs) occurred in 97.8% of the combination arm vs 85.7% of the pembrolizumab arm. Serious AEs occurred in 25 patients (55.6%) vs 6 (28.6%). Eight patients in the combination arm and 4 in the pembrolizumab arm had TEAEs leading to death; one of the 8 (Klebsiella pneumonia) was considered possibly related to danvatirsen by the investigator but unrelated by the Safety Review Committee. Treatment-related SAEs in more than 1 patient were hepatitis (2 vs 0) and ALT increased (3 vs 0). Tumor hemorrhage was observed in 6 patients and was not associated with thrombocytopenia.
| TEAE (preferred term) | Combination (N = 45) n (%) | Pembrolizumab (N = 21) n (%) |
|---|---|---|
| Any TEAE | 44 (97.8) | 18 (85.7) |
| ALT increased | 19 (42.2) | 0 |
| AST increased | 16 (35.6) | 0 |
| Fatigue | 17 (37.8) | 8 (38.1) |
| Anemia | 14 (31.1) | 4 (19.0) |
| Nausea | 10 (22.2) | 1 (4.8) |
| Constipation | 10 (22.2) | 2 (9.5) |
| Diarrhea | 9 (20.0) | 4 (19.0) |
| Platelet count decreased | 8 (17.8) | 0 |
| Pneumonia | 6 (13.3) | 1 (4.8) |
Interpretation and Broader Context
The authors conclude that danvatirsen plus pembrolizumab was not better than pembrolizumab alone in this phase 2 randomized study. They caution that it cannot be concluded whether the negative result reflects the antisense-oligonucleotide platform or ineffectiveness of STAT3 targeting in this setting: only 3 patients volunteered for optional paired biopsies, results were inconclusive, and the PBMC assessment was omitted after early termination. They support continued, more refined evaluation of STAT3-directed approaches with novel combinations and biomarker-driven patient selection.
Limitations
The authors report a small sample size, an open-label design, potential site variability in RECIST and local CPS assessments, imbalances within the CPS ≥20 subgroup, and 3 patients on the control arm who did not receive study drug. They also note that the open-label design may have led to under-reporting of familiar adverse events in the control arm. Only 3 patients volunteered for optional paired biopsies and the PBMC assessment was omitted after early termination, so the trial could not establish why the combination did not improve response.