1119 - Pre-Treatment Dental Findings and Clinically Significant Weight Loss during Head and Neck Radiation Therapy
Presenter(s)
O. K. Peng1, J. W. Chan2, M. V. Elia3, K. Kim1, S. S. Yom2, and J. C. Hong2; 1University of California, San Francisco School of Dentistry, San Francisco, CA, 2University of California, San Francisco, Department of Radiation Oncology, San Francisco, CA, 3UCSF + UC Berkeley Joint Program in Computational Precision Health, San Francisco, CA, San Francisco, CA
Purpose/Objective(s): Weight loss =10% within any 6-month periods in head and neck cancer (HNC) patients represents clinically significant malnutrition associated with feeding tube dependency and mortality. Pre-treatment dental evaluations are standard of care for HNC patients undergoing radiation therapy (RT), yet dental findings are not integrated into nutritional risk models. We hypothesize that poor pre-treatment oral health contributes to complications that worsen nutritional outcomes. Using structured electronic health record dental data, we evaluated whether pre-treatment dental characteristics are associated with clinically significant weight loss (=10%) during and after HNC RT.
Materials/Methods: We conducted a retrospective cohort study of patients diagnosed with HNC (ICD-10: C00–C14, C30–C32, C41.0–C41.1, C76.0, C77.0) who underwent RT between March 2022 and April 2025 and had a pre-RT new-patient dental evaluation at a single academic medical center. Dental features included prevalence of teeth with caries, bleeding on probing, plaque, and mobility, and number of missing teeth. Clinical features included total dose, planned fractions, baseline weight, age, sex, retreatment status, and treatment site. The endpoint was =10% weight loss from baseline during RT through 90 days post-RT. Unpenalized logistic regression (LR) with dental features only, LASSO with clinical features only, and LASSO with combined features were assessed via nested cross-validation (5-fold outer, 3-fold inner for tuning), grouped by patient and stratified by outcome.
Results: The cohort included 221 RT courses across 212 patients. Weight loss of =10% occurred in 85 courses (38.5%; 85 patients, 40.1%). The combined model achieved a mean cross-validated AUC of 0.758 (95% CI: 0.708–0.809), compared to 0.750 (0.677–0.823) for clinical features alone and 0.587 (0.543–0.630) for dental features alone. Caries prevalence (OR 1.772 per 10 percentage-point (pp) increase, p=0.007) and plaque prevalence (OR 1.289 per 10 pp increase, p=0.027) were significantly associated with =10% weight loss. Among patients who lost =10%, only 17/76 (22.4%) recovered to within 5% of baseline by 12 months post-RT.
Conclusion: To our knowledge, this is the first study leveraging structured dental EHR data to predict nutritional outcomes during HNC RT. Pre-treatment caries and plaque prevalence were significantly associated with clinically significant weight loss during HNC RT, though may correlate with other clinical characteristics. Most affected patients did not recover weight within one year. Overall, these findings suggest that poor oral health correlates with nutritional decline during RT. Incorporating routinely collected dental findings may improve early identification of patients who would benefit from proactive nutritional support.