Abstract 5795: Dynamic proteomic biomarkers of survival outcomes with anti-PD-1 immunotherapy in head and neck squamous cell carcinoma

Abstract Background: Despite advances in anti-programmed death (PD)-1 based immunotherapy in recurrent/metastatic (R/M) head and neck squamous cell carcinoma (HNSCC), challenges remain in identifying who will benefit from these therapies. Current tissue-based biomarkers have many limitations. Due to its accessibility and ease of longitudinal monitoring, there is growing interest in blood-based dynamic biomarkers for immunotherapy outcomes in R/M HNSCC. Methods: In a prospective multicenter study of patients with R/M HNSCC (n=38) receiving anti-PD-1 based therapy, peripheral blood was collected pre-treatment and post-treatment (cycle 2 day 1). A total of 83 serum-derived cytokine biomarkers were evaluated with the Olink Target 96 Immuno-Oncology panel. Associations of pre-treatment, post-treatment, and the change in cytokine levels (post minus pre) were assessed with overall survival (OS) and progression-free survival (PFS). Unsupervised clustering analysis identified subgroups based on cytokine profiles. Further analyses were conducted using univariable and multivariable Cox proportional hazards and Cox Elastic Net machine learning models. Results: Among the 83 cytokines, 49 (59%) exhibited a statistically significant within-patient increase from pre- to post-treatment. Further, several cytokines were significantly associated with OS and PFS and 10 associated with OS were shared in both pre- and post-treatment analyses. Generally, increased cytokine levels were associated with poorer outcomes. In analyses of cytokine level change (post minus pre), seven cytokines were significantly associated (FDR0.05) with OS (i.e., VEGFA, HGF, PGF, CCL19, CCL23, MCP3, TNFRSF12A) and three with PFS (i.e., NOS3, MMP12, and CCL23) after adjustment for potential confounders. An unsupervised clustering analysis of patients based on the change in the levels of all 83 cytokines uncovered two subgroups that differed significantly with respect to OS and PFS even after adjustment of potential confounders. A prognostic cytokine score was derived using machine learning analysis using elastic net regularization. Not only were cytokine scores significantly associated with platelet count and, in some cases, neutrophil count, they also provided additional prognostic value beyond established prognostic factors. Conclusion: Dynamics of serum cytokines related to blood platelets during anti-PD-1-based therapy may serve as prognostic biomarkers in patients with R/M HNSCC. Citation Format: Md Saiful Islam Sai, Kartik Sehgal, Ze Zhang, Rondi Butler, John Wiencke, Annette Molinaro, Juliana Miller, Hannah Stolrow, Keisuke Shirai, Lucas Salas, Devin Kosetler, Robert Haddad, Brock C. Christensen, Karl Kelsey. Dynamic proteomic biomarkers of survival outcomes with anti-PD-1 immunotherapy in head and neck squamous cell carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 5795.