PROJECT SUMMARY Despite the dismal five-year overall survival rate, a moderate response rate to treatments, and one of the highest suicide rates among cancer patients, human papillomavirus-negative head and neck squamous cell carcinomas (HNCs) are curable if diagnosed early. Oral epithelial dysplasias (OEDs) and oral lichen planus (OLP) are potentially premalignant lesions that offer a window for disease eradication. The current standard of care for these precursor lesions involves H.&E. histologic grading and long-term clinical follow-ups. Most of OEDs and OLP do not progress to cancer. However, a significant challenge is that it is impossible to maintain high-frequency follow-ups for every patient with OED or OLP. Emerging adjunct clinical technologies often evaluate diagnostic success based on their power to detect “high-grade” OEDs. However, the WHO histologic grading of OED has little, if any, prognostic value in determining the transformation risks. In addition, the histologic grading of OEDs has low inter-observer and intra-observer consistency with the kappa-values and strength of agreement rated slight-to-poor. As a result, the American Dental Association has not recommended any adjunct diagnostic modalities for OED/OLP. Before we can deploy impactful early detection technologies, we must improve our understanding of the biology of high-risk OEDs. We first learn from decades of clinical observations. During the clinical examinations, erythematous color change and induration warrant a biopsy. These features indicate early inflammatory and mechanical changes in the microenvironment of initiating HNCs. Thus, we generated high-fidelity, genetically engineered mouse models to recapitulate these immune and mechanical alterations over the course of HNC initiation. These models are uniquely poised to establish the high-risk markers due to their 100% malignant transformation rate in the oral mucosa. Through robust longitudinal monitoring, we have uncovered an initial set of immunometabolic markers whose signals emerge before the HNC histology appears. This program will discover a comprehensive set of high-risk features and employ advanced machine learning to generate a weighted risk score, which will be validated through our extensive collections of low-risk leukoplakia and transformed OED/OLP human specimens. To support the robust on-slide technology, we also developed an optical biopsy tool, approved by the Food and Drug Administration, to perform non-invasive monitoring of molecular markers at a microscopic resolution below oral mucosal surfaces. This milestone-driven program will leverage the strengths of precision in high-fidelity modeling for transforming OEDs, the extensive translational resources, a cutting-edge optical biopsy platform, and single-cell technologies to extend the human senses in conventional histology and clinical examination of OED to unprecedented molecular levels. This integrated effort will inform transformative on-slide and optical biopsy ancillary tools to capture high-risk OEDs at the earliest phase for HNC prevention.

PUBLIC HEALTH RELEVANCE Oral cavity cancers are preceded by precursor lesions and are preventable if diagnosed early. However, it is challenging to maintain high-frequency clinical follow-ups for every patient with these very common precursor lesions, most of which will never become cancers, and the current pathology grading is often inconsistent and does not predict cancer risks. Thus, this program will leverage high-fidelity pre-clinical modeling, human specimens, precision engineering, and single-cell technologies to develop transformative on- slide and optical biopsy tools for the earliest detection of high-risk pre-malignant oral cavity lesions.