A time-adapted risk stratification framework guided by longitudinal post-radiotherapy EBV DNA clearance for high-risk nasopharyngeal carcinoma patients.

PURPOSE: Patients with nasopharyngeal carcinoma (NPC) exhibiting persistent EBV DNA positivity throughout treatment (a “D-D-D” trajectory) represent a high-risk group with undefined management strategies. This study aimed to define critical post-radiotherapy timepoints to establish a precision surveillance strategy for these patients. METHODS: In a longitudinal cohort of non-metastatic “D-D-D” NPC patients, plasma Epstein-Barr virus (EBV) DNA was quantified at 1.5, 3, 6, 9, and 12 months post-radiotherapy. A time-dependent Cox model was employed to evaluate the dynamic prognostic value of EBV DNA status. Model performance was rigorously compared using Harrell’s C-index, the Akaike information criterion, and net reclassification improvement. Classification metrics, including sensitivity, specificity, and predictive values, were analyzed to identify clinically actionable decision points. RESULTS: Among 295 patients, early clearance by 1.5 months (10.5% of the cohort) conferred favorable survival, comparable to the “D-D-U” subgroup from our previous analysis. The 12-month EBV DNA status emerged as the strongest independent predictor of disease progression (adjusted HR = 2.804, 95% CI: 2.273–3.460, P < 0.001), offering high specificity (98.6%) but delayed intervention. In contrast, the 1.5-month timepoint, while highly sensitive (95.0%), generated an unactionable alert rate (89.0%). The 3-month assessment optimally balanced these parameters, maintaining substantial sensitivity (76.7%) within a clinically actionable alert rate (65.5%). CONCLUSIONS: We propose a time-adapted risk stratification framework wherein the 3-month assessment identifies patients for intensified surveillance, while persistent EBV DNA at 12 months identifies a population at high risk of progression. The clinical benefit of timepoint-guided interventions warrants prospective randomized investigation. CLINICAL TRIAL NUMBER: Not applicable