Data science approaches to classify aggressive tissue phenotypes and predict disease-free survival in HPV-negative head and neck squamous cell carcinoma (HNSCC)

Data science approaches to classify aggressive tissue phenotypes that impact survival in HPV- negative head and neck squamous cell carcinoma (HNSCC) Head and neck squamous cell carcinoma (HNSCC) is a potentially fatal disease with a reported 5-year overall survival of 64.5 percent. Despite decades of research into the molecular pathogenesis of HNSCC, researchers have yet to identify reliable prognostic factors to implement into clinical practice to guide treatment decisions beyond conventional TNM (tumor, node metastasis) staging. Efforts to correlate gene expression with aggressive histopathologic phenotypes, such as nodal disease and perineural invasion, have intensified with the increasing availability of sequencing data. However, singular tumor markers such as TP53 mutational status have not proven statistically significant in predicting recurrence or survival. Rather, clinical studies suggest that differences in histopathologic factors may explain differences in survival among patients within the same TNM stage. Consequently, there is a pressing need to elucidate genetic differences between indolent and more aggressive tissue phenotypes in HNSCC. Furthermore, the molecular pathways driving these aggressive tissue phenotypes HNSCC remain inadequately understood, and their presence is analyzed through visual examination alone, a method prone to imprecision and potential diagnostic oversights. To this end, a more precise evaluation method based on molecular data could enhance the detection of adverse histopathologic features that may lead to recurrence and decreased survival. This project aims to delineate molecular variations within tumors based on distinct histopathologic features and employ machine learning techniques to construct a predictive model using molecular data. This model would offer clinicians a more objective means of identifying adverse prognostic tissue phenotypes, potentially leading to improved stratification of patients into low-risk and high-risk groups for disease progression. Secondly, our findings will shed light on the underlying molecular pathways driving different histologic phenotypes that can open new avenues for targeted therapeutic interventions. Using existing data repositories from TCGA and DBGap as well as a multi-institutional cohort of cases (Rutgers, Indiana, Columbia), a key feature of this project is to apply machine learning methods on large-scale molecular data to develop an algorithm that can accurately predict the presence of aggressive disease. Dr. Yingci Liu will lead this research initiative under the K08 award proposal, with the goal of developing expertise in computational genomics and machine learning to establish an independent translational research program in computational genomics and head and neck cancer. Dr. Liu will receive support from a robust, multidisciplinary mentoring team consisting of experts in oncology, machine learning, and head and neck cancer, which includes Dr. Shridar Ganesan, Dr. Antonina Mitrofanova, and Dr. Flora Momen-Heravi. Project Number: 7K08DE034852-02 | Fiscal Year: 2026 | NIH Institute/Center: National Institute of Dental and Craniofacial Research (NIDCR) | Principal Investigator: Yingci Liu-Swetz | Institution: UNIVERSITY OF PENNSYLVANIA, PHILADELPHIA, PA | Award Amount: $168,610 | Activity Code: K08 | Study Section: National Institute of Dental and Craniofacial Research Special Grants Review Committee[DSR] View on NIH RePORTER: https://reporter.nih.gov/project-details/11505841