Establishment and validation of a prognostic nomogram to predict early metastasis in nasopharyngeal carcinoma patients within six months after radiotherapy and to guide intensive treatment.

PURPOSE:This study aimed to establish an effective prognostic nomogram to predict the risk of early metastasis (EM) in nasopharyngeal carcinoma (NPC) patients, as a guide for intensive treatment. MATERIALS AND METHODS:A total of 9021 patients with biopsy-confirmed NPC at our institute were enrolled in this study between December 2006 to December 2016. We randomized these patients using a proportion of 2/3 and 1/3 and selected 6044 and 2977 patients as the training and validation cohorts, respectively. All patients received radiotherapy with or without chemotherapy. Univariate and multivariate logistical regressions were used to identify independent risk factors. The nomogram's predictive value was evaluated by concordance indexes (C-indexes), calibration curves, probability density functions (PDFs), and clinical utility curves (CUCs). ROC analysis using Delong test was used to compare efficiency between the nomogram and other risk factors. RESULTS:In total, 174 (2.9%) and 81 (2.7%) patients in training and validation cohorts, respectively, had EM. Pretreatment plasma EBV DNA, N stage, LDH, ALP, BMI, and sex were independent predictive factors of EM. The C-indexes of nomogram were 0.756 (95% CI = 0.719-0.793) and 0.766 (95% CI = 0.720-0.813), in the training and validation cohorts, respectively. The C-index of the nomogram was significantly superior to any one of independent factors. According to the PDFs and CUCs and considering the balance of the true positive EM patients and true positive non-EM patients, we chose 5.0% as a threshold probability for clinical decision-making, which could distinguish about 85% and 48% of non-EM and EM patients, respectively. CONCLUSION:Our nomogram had good accuracy in predicting EM incidence, and a 5.0% threshold was appropriate for clinical decision-making.