Abstract
OBJECTIVE: The study was designed to construct and validate a nomogram for predicting overall survival (OS) of male breast cancer (MBC) patients with infiltrating duct carcinoma (IDC). METHODS: The cohort was selected from the Surveillance, Epidemiology, and End Results (SEER) database between January 1, 2004 and December 31, 2013. Univariate and multivariate Cox proportional hazard (PH) regression models were performed. A nomogram was developed based on the significant prognostic indicators of OS. The discriminatory and predictive capacities of nomogram were assessed by Harrell's concordance index (C-index), calibration plots, area under the curve (AUC) and the decision curve analysis (DCA). RESULTS: The median and maximal survival time of 1862 eligible patients were 49 and 131 months, respectively. Multivariate analysis showed that age (P < 0.0001), marital status (P = 0.002), T stage (P < 0.0001), N stage (P = 0.021), M stage (P < 0.0001), progesterone receptor (PR) (P = 0.046), human epidermal growth factor receptor-2 (HER2) (P = 0.009), and chemotherapy (P = 0.003) were independent prognostic indicators of IDC of MBC. The eight variables were then combined to construct a 3-and 5-year nomogram. The C-indexes of the nomogram were0.740 (95% confidence interval [CI] [0.709-0.771]) and 0.718 (95% CI [0.672-0.764]) for the internal validation and external validation, respectively. A better discriminatory capacity was observed in the nomogram compared with the SEER summary stage (P < 0.001) and AJCC TNM staging systems (6th edition; P < 0.001) with respect to OS prediction. Good consistency was detected between the nomogram prediction and actual findings, as indicated by calibration curves. The AUC for 3-and 5-year OS was 0.739 (95% CI [0.693-0.786]) and 0.764 (95% CI [0.725-0.803]) in the training cohort and 0.737 (95% CI [0.671-0.803]) and 0.735 (95% CI [0.678-0.793]) in the validation cohort, respectively. The DCA demonstrated that the survival nomogram was clinically useful. CONCLUSIONS: The nomogram was able to more accurately predict 3-and 5-year OS of MBC patients with IDC histology than were existing models.