IMPACT OF HEAD AND NECK RADIATION THERAPY ON LONG-TERM NEUROCOGNITIVE FUNCTION

Abstract

Head and neck cancer patients undergoing radiotherapy are increasingly surviving longer, but often at the cost of neurocognitive complications that impair their quality of life. This study aimed to quantify the prevalence, predictors, and biological correlates of radiation-induced cognitive impairment in this patient population. A mixed-methods approach was employed, involving retrospective clinical data analysis, neurocognitive assessments, biomarker profiling, and machine learning modeling.As detailed in the results, significant cognitive impairments were observed post-radiotherapy, particularly in processing speed and attention. Radiation dose to the hippocampus and frontal lobes showed strong negative correlations with memory and executive function scores (Table 4). Logistic regression revealed hippocampal dose and comorbidities as independent predictors of cognitive decline (Table 5). Systemic inflammation also emerged as a contributory factor (Table 6). Among machine learning models, XGBoost achieved the highest accuracy (85%) and AUC (0.91) in predicting neurocognitive impairment (Table 7). Visualizations supported these findings, with Fig. 1 showing domain-specific cognitive decline and Fig. 2 confirming the dose-response relationship. These findings highlight the importance of incorporating neurocognitive protection and predictive modeling into standard radiation therapy planning for head and neck cancers. This integration can enable clinicians to balance therapeutic efficacy with quality-of-life preservation.