Pembrolizumab combined with nab-paclitaxel and platinum in recurrent/metastatic HNSCC: efficacy, safety, and survival predictive model
ESMO Open, 2026
Tang L., Gao R., Yao J., Zhang W., Zhang Y., Zhang Y., Lu H., An C., Gui L.
| Disease area | Application area | Sample type | Products |
|---|---|---|---|
Oncology Immunotherapy | Pathophysiology | Plasma | Olink Target 96 |
Abstract
Background
Recurrent/metastatic head and neck squamous cell carcinoma (R/M HNSCC) remains a therapeutic challenge. This study evaluated the triplet combination of pembrolizumab, nab-paclitaxel, and platinum in R/M HNSCC and developed a practical predictive model for treatment stratification.
Patients and methods
In this single-arm phase 2 study (NCT04857164), treatment-naïve patients with R/M HNSCC received pembrolizumab (200 mg), nab-paclitaxel (260 mg/m2), and cisplatin (75 mg/m2) [or carboplatin area under the curve (AUC) 5] once every 3 weeks for six cycles, followed by pembrolizumab maintenance. Additionally, we developed a clinical parameter-based random forest predictive model and conducted exploratory flow cytometry analysis and plasma proteomic profiling.
Results
From April 2021 to February 2025, a total of 148 patients with R/M HNSCC were enrolled. At a median follow-up of 23.0 months [95% confidence interval (CI) 19.4–29.6], objective response rate was 64.2% (95/148) and disease control rate was 93.2% (138/148). Median progression-free survival (PFS) was 12.7 months [95% confidence interval (CI) 10.2–14.7) and median overall survival (OS) was 21.8 months (95% CI 18.9–35.7). Hypothyroidism was the most common immune-related adverse event (27%, 40/148), with 39 grade 1-2 and one grade 3 case. The predictive model based on seven clinical parameters achieved excellent predictive accuracy, with time-dependent AUCs of 0.826-0.906 (training) and 0.753-0.919 (test) for PFS prediction, and 0.682-0.799 (training) and 0.784-0.908 (test) for OS. Higher model scores delineated a high-risk phenotype marked by nutritional depletion and chronic inflammation activation, mechanistically explaining the associated poor outcomes.
Conclusions
This study demonstrated encouraging efficacy and a manageable safety profile of pembrolizumab–nab-paclitaxel–platinum combination in R/M HNSCC. Furthermore, we established a predictive model using routine clinically accessible parameters, offering a practical and cost-effective tool for treatment stratification.