Integrating RNA-Seq and AI-derived Digital Pathomics for Multimodal Survival Prediction in Oral Cancer - PROJECT SUMMARY/ABSTRACT Oral squamous cell carcinoma (OSCC) is an aggressive cancer that affects more than 389,000 people worldwide each year with 188,000 deaths. Despite advances in surgical and adjuvant therapies, survival remains poor, with tumor stage and clinicopathologic features providing only limited prognostic accuracy. The lack of reliable biomarkers underscores the need for robust prognostic biomarkers to guide treatment and improve survival. The goal of this proposal is to develop and validate a multi-modal survival prediction model that integrates RNA sequencing (RNA-Seq), artificial intelligence (AI)-driven digital pathomics, and clinicopathologic variables. This project will benefit from a prospectively collected biobank of OSCC patients at Loma Linda University that includes fresh-frozen tissues, noninvasive brush swabs, saliva, blood, and hematoxylin and eosin (H&E) slides. Our study proceeds with the following three aims. In Aim 1, we will identify RNA-Seq signatures predictive of 2- year survival and recurrence and perform functional pathway analysis to uncover targetable mechanisms of OSCC progression. In Aim 2, we will define AI-driven digital pathomic biomarkers by extracting deep learning features from H&E slides and correlating them with molecular profiles and clinical outcomes. In Aim 3, we will focus on developing a multi-modal OSCC mortality risk score by integrating RNA-Seq, digital pathomics, and clinical variables, validating the model in an independent prospective cohort. Additionally, we will assess the feasibility of using noninvasive brush swabs for transcriptomic profiling. The successful completion of this work holds the promise to establish a validated, personalized risk prediction tool for OSCC patients. The mentorship and advisory team for this training grant is comprised of individuals who have expert knowledge in all aspects of translational and clinical research, specifically in the areas of cancer, bioinformatics, AI, and medical computational analysis, all of which are critical for the proposed research. The career development plan is carefully structured, providing comprehensive training in both technical and professional skills through a series of courses, national meetings, and meetings with the mentorship team. Upon completion of the K99/R00, the principal investigator will be well-prepared to establish a career as an independent investigator and pursue R01- level funding.
