Objective  To predict the prognosis of patients with invasive micropapillary carcinoma (IMPC) of the breast (IMPC-B), and compare the pathological characteristics and prognostic factors of patients with IMPC-B in China and the United States.

  Methods  This retrospective analysis included 83 cases of IMPC-B diagnosed between July 2006 and July 2015 at Xijing Hospital, The Fourth Military Medical University, China. In addition, the clinical data of 415 patients diagnosed with IMPC-B between March 2010 and March 2015 were obtained from the Surveillance, Epidemiology, and End-Results (SEER) database of the American National Cancer Institute. The clinical and pathological characteristics of the Chinese and American patients were compared, and independent risk factors for overall survival (OS) and cancer-specific survival (CSS) were analyzed using univariate and multivariate Cox proportional regression models and the Fine-Gray competing risk model. Based on the results, nomograms were developed for predicting OS and CSS. The effectiveness of the nomogram models was evaluated by internal and external validation, and the clinical benefits and application value of the models were evaluated using clinical decision curve analysis.

  Results  There were significant differences in age, tumor site, operation method, tumor status (for example, whether it was the primary tumor), and T-stage between the modeling set and validation set (P<0.05). Univariate and multivariate analyses using Cox proportional regression models and the Fine-Gray competing risk model showed that age, N-stage, M-stage, and molecular type were independent prognostic factors in patients with IMPC-B (P<0.05). Nomogram models were developed using the above variables. For the models for OS and CSS, the C-indexes of the modeling set were 0.85 and 0.79, those of the validation set were 0.72 and 0.70, and those for internal validation were 0.81 and 0.74, respectively. The calibration curve analysis showed that the model predictions for OS and CSS were consistent with the actual values, and the clinical decision curve analysis showed that the models had clinical utility.

  Conclusions  The developed nomograms could accurately predict the prognosis of patients with IMPC-B, providinga scientific basis for clinical diagnosis and treatment.