Abstract A01: HPV+ oral cancer PDXs identify prognostic utility for E2F target gene expression and mutation burden

Barriers to advancing therapy for HPV-related (HPV+) head and neck squamous cell carcinomas (HNSCCs) include a paucity of preclinical models and difficulty prospectively identifying the minority of cases with lethal potential. This study therefore sought to establish HPV+ patient-derived models with high genetic fidelity and use them to help identify aggressive molecular subtypes of HPV+ HNSCC. Using whole-exome sequence and viral transcript analysis, nine stably passaged patient-derived xenogafts (PDXs) were found to avoid typical artifacts in HPV+ cell lines, which include loss of canonical PIK3CA mutants, gains in 3q copy number, and increased viral integration. Notably, the models contained markedly increased rates of alteration in Notch pathway genes and in H3K4 methyltransferases than would be predicted by The Cancer Genome Atlas (TCGA). Although phenotypic correlates of these alterations were not overtly apparent in the models, analysis of their growth features uncovered strong positive associations of tumor mutational burden (TMB) with in vivo growth rate and in vitro organoid propagation capacity. This observation led to identification of a novel association in TCGA between TMB and both reduced local progression and survival of early HPV-negative HNSCCs, which was further validated in esophageal squamous cell carcinomas. Although TMB failed to be prognostic in the HPV+ HNSCCs, insight into high-risk HPV+ molecular features was provided by a PDX from an unusual case that suffered lethal relapse shortly after definitive therapy. Reduced viral E7 and p16INK4A expression was identified both in this model and in early lethal HPV+ cases in TCGA as well as those in a second HPV+ HNSCC cohort (JHU). The lower E7 and p16INK4A levels here suggested a diminished imprint of viral oncogene function upon cell cycle regulation in lethal cases. Hierarchical clustering of the cohorts using a broad E2F target gene expression panel revealed a distinctive pattern of E2F target dysregulation in the clusters that captured the early lethal cases. A subset of ten transcripts optimally predicted survival independent of stage in both TCGA and JHU cohorts. This analysis of the largest panel of HPV+ HNSCC PDXs described to date identified robust models for PIK3CA activation, Notch pathway dysregulation, and altered H3K4 methylation in this disease. At the same time, it defined novel clinical correlates of TMB and E2F target expression pattern that have potential utility in guiding risk stratification, biomarker development, and trial design. Citation Format: Nicole D. Facompre, Pavithra Rajagopalan, Alexander T. Pearson, Kathleen T. Montone, Claire D. James, Frederico O. Gleber-Netto, Hiroshi Nakagawa, Elizabeth A. White, Anil K. Rustgi, Joseph A. Califano, Curtis R. Pickering, Bradley Windle, Iaian M. Morgan, Phyllis A. Gimotty, Devraj Basu. HPV+ oral cancer PDXs identify prognostic utility for E2F target gene expression and mutation burden [abstract]. In: Proceedings of the AACR-AHNS Head and Neck Cancer Conference: Optimizing Survival and Quality of Life through Basic, Clinical, and Translational Research; 2019 Apr 29-30; Austin, TX. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(12_Suppl_2):Abstract nr A01.