Abstract A012: Stratifying immune subtypes in real-world HNSCC cohort to identify those with better outcomes to standard treatment regimens

Abstract Head and neck squamous cell carcinoma (HNSCC) is a heterogeneous disease characterized by low survival rates and increasing incidence, highlighting the need for better patient stratification to understand the host immune mechanisms driving varied responses1. From 1,141 HNSCC patients from a real-world cohort (naïve, treated, recurrent, and metastatic), we identified 5 immune subtypes (C1-C5) using non-negative matrix factorization on the top 20% variant immune genes. We also identified two hypoxia subtypes (low and high) and four traditional subtypes (basal, mesenchymal, atypical, and classical) using hypoxia and most variant genes signatures, then related them to immune subtypes. Pathways enrichment in subtypes were assessed using GSEA. Subtypes were associated with clinical parameters (Fisher's exact test, FDR corrected). Tertiary lymphoid structures (TLS) presence was identified using H&E-stained slides. For 188 metastatic first-line treated patients (M1L) receiving Chemo (Carboplatin), Immune Checkpoint Blockade (ICB) (pembrolizumab) or Combo (Chemo+ICB), baseline biopsies were available. M1L subgroups were associated with median progression-free survival (mPFS) using a multivariate CoxPH model. The immune subtypes identified exhibited a distinctive mix of hypoxia and traditional subtypes but predominantly represented one. C2 and C3 were mostly atypical and low hypoxia; C4 was mesenchymal and high hypoxia; and C5 was basal and high hypoxia. C1 uniquely exhibited a mix of classical and atypical subtypes with no distinct hypoxia profile. Pathways related to immune responses were enriched in C2, cytokine production in C4 and Inflammatory responses in C2 and C4. The immune subtypes in HNSCC patients were predictive of clinical characteristics. Statistically significant correlations were found between subtypes and clinical parameters (e.g. stage 4 patients highest in C3; FDR P-value <0.01)). Correlations with outcomes (P-value <0.05) were less pronounced. Patients with outcome information showed that C3 and C5 subtypes had the highest number of progressive disease cases, while responders and stable disease patients were evenly distributed across all subtypes. Immune subtypes and TLS in M1L patients were predictive of treatment-specific progression and survival. M1L patients in C1 receiving Chemo or Combo had better mPFS (8.9 and 10.68 months) than C5 (3.67 months and 4.6 months) due to less hypoxia-induced immunosuppression. C2 had better mPFS with ICB alone (10.88 months). Moreover, ICB-treated patients with TLS at baseline had mPFS of 10.9 months, compared to 5.9 months without TLS. Although HPV status was related to all subtypes, there was no difference in mPFS between the HPV+ and HPV-M1L patients. Overall, we report that considering immune subtypes and TLS for stratifying patients in clinical trials would be instrumental in tailoring treatment plans and improving clinical outcomes for HNSCC patients. References: 1. Chen Z, John J and Wang JH. Frontiers in Oncology, 2022:12 Citation Format: Indu Khatri, Merzu Belete, Meijian Guan, Merel Jongmans, Susan Sun, Anantharaman Muthuswamy, Lauren Brady, Han Si, Sri Sridhar, Brandon W Higgs, Iris CRM Kolder. Stratifying immune subtypes in real-world HNSCC cohort to identify those with better outcomes to standard treatment regimens [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Tumor Immunology and Immunotherapy; 2024 Oct 18-21; Boston, MA. Philadelphia (PA): AACR; Cancer Immunol Res 2024;12(10 Suppl):Abstract nr A012.