Predictive Score for Recurrence after Complete Resection in Locally Advanced Oral Cavity Squamous Cell Carcinoma.

e17536 Background: Early progression, usually defined disease-free interval (DFI) less than six months after completing adjuvant platinum-based chemoradiotherapy (CRT), has very poor outcome for oral cavity squamous cell carcinoma (OCSCC). But there are no biomarkers to predict such early progression. Methods: Locally advanced OCSCC patients, after complete surgical resection and followed-up platinum-based adjuvant CRT, were retrospectively enrolled from Chung Shan Medical University Hospital (CSMUH, training cohort) and Taipei Veterans General Hospital (TPE-VGH, validating cohort) in Taiwan. Clinicopathologic variables of patients with DFI < or ≥ 6 months were compared by using the χ2 test. The Cox proportional hazards model was applied to identify independent factors for DFI. Survival was estimated using the Kaplan-Meier method and log-rank test. Results: A total of 350 high-risk OCSCC patients were enrolled, including 146 patients in training cohort and 204 in validating cohort. In multivariate Cox regression, pN > 0, extracapular spread, and depth of invasion ( ≥ 1cm) were independent factors for DFI in training cohort. If each factor scored one point, the scoring system could effectively predict early progression that sensitivity/specificity/area under curve (AUC) of training and validating cohort were 57.7%/91.2%/0.771 and 58.1%/83.9%/0.730, respectively (the cutoff level ≤ 2 or > 2). DFI between lower- (score 0–2) and high- (score 3) risk groups were also significantly different in both training (median DFI, 59.6 vs. 4.5 months, P < 0.001) and validating cohorts (NA vs. 9.3 months, P < 0.001). Conclusions: The established score system was effective to predict early progression after adjuvant CRT for locally advanced OCSCC.