Analysis of high dimensional genomic dataset to reveal mechanistic insights on divergent outcomes of head and neck cancer using system biology modeling coupled with experimental validation

Project Summary/Abstract In response to PAR-23-133, we propose to use high-dimensional datasets of human head and neck squamous cell carcinomas (HNSCC) available in public databases to identify new mechanisms of HNSCC pathogenesis and new markers associated with clinical outcome. HNSCC is a prevalent cancer in the US and worldwide (1). HNSCC is associated with carcinogens like tobacco usage (HPV-) or human papillomavirus (HPV+). Compared to HPV- HNSCC, HPV+ HNSCC generally exhibits better prognosis and responds to therapy favorably. However, it is less well-understood what underlying mechanisms differentiate the divergent outcome of HPV+ vs. HPV- HNSCC, and what specific effects of viral factors have on the phenotypes of tumor cells vs. tumor immune microenvironment (TIME). A better mechanistic understanding of these questions will inform design of effective personalized therapy, especially combinatorial strategies, and reveal what factors cause the differential responses of HPV- vs. HPV+ HNSCC to treatment, and how treatment strategies can be better tailored for different subtypes of HNSCC. In this application, we propose to analyze existing HNSCC datasets using novel methodology to uncover mechanistic insights on divergent clinical outcome of HPV+ vs. HPV- HNSCC. HPV+ and HPV- HNSCC differ substantially in their mutational and transcriptomic profiles as well as in their TIMEs. For instance, HPV+ HNSCC has a higher level of tumor-infiltrating B cells than HPV- counterpart (2); however, key differences between HPV- and HPV+ TIMEs need further investigation. Additionally, HPV+ HNSCC harbors TP53 mutations in only 3% of tumors and overall has far fewer mutations than its HPV- counterpart (3). In contrast, HPV- HNSCC contains a high level of genetic heterogeneity including mutations/amplifications of oncogenes (e.g., PIK3CA), or both loss-of-function mutations and potential gain-of-function mutations in multiple genes (e.g., TP53, CDKN2A, CASP8 and CREBBP/EP300) (3-13). An increasing amount of publicly available single-cell genomics data of HNSCC provides unique and cost-effective opportunities to investigate mechanistic questions of HNSCC pathogenesis. In this proposal, we will employ a Dynamo approach (14) recently published by MPI Dr. Xing's group to analyze single-cell genomic data within the framework of systems biology modeling, beyond traditional statistics-based informatics analyses (e.g., differentially expressed gene (DEG) analysis). We will also validate computational predictions through further experiments and patient data analyses.