Introduction: Although often presented as one disease for genomic studies, head and neck squamous cell carcinoma (HNSCC) is treated according to clinical factors such as anatomic subsite, and tumor stage. Oral cavity squamous cell carcinoma (OCSCC) comprises 1/3 of all HNSCC and is predominantly HPV(-). Depending upon clinical stage, OCSCC treatment involves surgical resection +/- neck dissection, followed by radiation +/- chemotherapy. Our group and others previously described four mRNA expression patterns (classical, atypical, basal, and mesenchymal), each with unique genomic features and prognosis. Here, we further examine the clinical utility of gene expression subtyping in HNSCC and introduce the potential for predictive applications in HPV(-) HNSCC according to clinically relevant subgroups including oral cavity cancer.

Experimental Procedures: Multiple independent HNSCC datasets were obtained from public repositories, totaling 562 patients, including from MD Andersen (MDA) (GSE41117) and the Cancer Genome Atlas (TCGA) sourced from the Genome Data Commons. Cases were included for further analysis if they had N stage and overall survival values. Samples were assigned molecular subtypes (basal, mesenchymal, atypical or classical) using a reduced gene set version of the classifier reported earlier. HPV status was determined by HPV gene expression. The clinical endpoint was overall survival at censured at 36 months. Kaplan-Meier (KM) plots and logrank tests were used to investigate associations between clinical variables and survival. Cox models were used to adjust for potential confounders. All statistics were performed using the survival package in R.

Results: Of the 418 training patients from the TCGA dataset that met analysis criteria, nearly 20% presented with stage I and II tumors. In the clinically relevant subgroup of node(-) OCSCC patients, mesenchymal subtype was associated with worse survival (HR 2.4, p=0.021), offering a novel and potentially actionable biomarker in otherwise early-stage and low risk disease. Associations in the MDA validation cohort confirmed the association. Node(-) non-mesenchymal OCSCC patients had far better survival (no deaths observed) compared to node(-) mesenchymal and all node(+) patients which had similarly poor survival. Differential responses as a function of radiation will be presented.

Summary and Conclusions: This study confirms that the mesenchymal subtype is associated with worse outcomes across all cases of HNSCC. However, we demonstrate that mesenchymal subtype is associated with poor survival, even in the clinically relevant setting of early-stage, node(-) OCSCC treated with surgical resection. These findings highlight the potential value of gene expression subtyping as an adjunct to pathology for deciding which treatment option is best suited for patients with HNSCC.

Citation Format: D. Neil Neil Hayes, MD, MS, MPH, Greg Mayhew, Josh Uronis, Jose Zevallos. Prognostic and predictive applications from mesenchymal gene expression subtype analysis for early-stage, HPV(-) head and neck squamous cell carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2142.