Topkan's CARWL Index Efficiently Predicts the Radiation-Induced Tooth Loss Rates in Radically Treated Locally Advanced Nasopharyngeal Cancer Patients

Purpose To assess the usefulness of the novel CARWL index in predicting radiation-induced tooth loss (RITL) rates in locally advanced nasopharyngeal cancer (LA-NPC) patients undergoing concurrent chemoradiotherapy (C-CRT). Methods The study retrospectively examined data from 323 LA-NPC patients. The patients were divided into two groups based on cutoff values for CAR and weight loss (WL). The ideal cutoff for RITL was 3.0 g/dL [AUC: 83.0%, sensitivity: 83.6%, specificity: 81.4%, J-index: 0.650]. CARWL index was created by combining pretreatment CAR and WL status (WL <= 5.0% vs > 5.0%, resulting in four groups: Group 1: CAR < 3.0 and WL <= 5.0%, Group 2: CAR < 3.0 and WL > 5.0%, Group 3: CAR >= 3.0 and WL <= 5.0%, and Group 4: CAR > 3.0 and WL > 5.0%. Results RITL was diagnosed in 67.2% of patients. Since the RITL rates of Groups 2 and 3 were statistically indistinguishable, we combined them and created the three-tiered CARWL score groups: CARWL-0: CAR < 3.0 and WL <= 5.0%; CARWL-1: CAR < 3.0 and WL > 5.0%, or CAR >= 3.0 and WL <= 5.0%; and CARWL-2: CAR > 3.0 and WL > 5.0%. Comparative analysis revealed that the RITL rates gradually and significantly increased from CARWL-0 to CARWL-2 score groups (49.4% vs 64.7% vs 83.0%; P <0.001) despite similar baseline disease and patient characteristics. Results of the multivariate analysis showed that higher CARWL score groups were independent and significant predictors of increased RITL rates (p < 0.001). Conclusion Present results suggest that the novel CARWL index is a reliable biomarker for predicting RITL incidence in LA-NPC patients.