Objective To explore the prognostic risk factors of patients with lung metastasis of breast cancer and construct nomogram model for survival prediction.

Methods Datum were collected from the database of the Surveillance, Epidemiology, and End Results (SEER database). A total of 5 334 patients diagnosed as lung metastasis of breast cancer from 2010 to 2015 were screened and divided into the modeling set with 3 556 cases and the validation set with 1 778 cases according to the ratio of 2 to 1. Multivariate Cox regression model was used to analyze the factors related to overall survival (OS), and Fine-Gray competitive risk model was used to evaluate the factors related to breast cancer specific survival (BCSS).

Results There were no significant differences in terms of age, race, gender, primary tumor location, differentiation and grading, T stage, N stage, pathological type, accepting operation or not, accepting radiotherapy and chemotherapy or not, the number of regional lymph nodes, other distant metastasis sites (bone, brain, liver), molecular subtype, number of malignant tumors, marital status and insurance status between the modeling set and the validation set (P>0.05). Multivariate Cox regression analysis showed that age, tumor grading, T stage, pathological tissue type, operation, bone metastasis, brain metastasis, liver metastasis and molecular typing were correlated with the prognosis of patients with lung metastasis of breast cancer (P < 0.05). The receiver operating characteristic (ROC) curve and the calibration curve of 3-year survival were used for internal and external validations of the prediction model of overall survival, and the results showed that the area under the ROC curve (AUC) was 0.738 for internal validation and 0.746 for external validation, which indicated that the prediction model was highly discriminative and accurate. Fine-Gray competitive risk model was applied to analyze the factors affecting the cancer specific survival rate of patients with lung metastasis of breast cancer, and the results showed that age, differentiation degree, T stage, surgery, chemotherapy, bone metastasis, brain metastasis, liver metastasis and molecular subtype were independent factors affecting the specific prognosis of breast cancer. According to the analysis results of Fine-Gray competitive risk model based on the modeling set, the ROC curve and the 3-year survival calibration curve were used for internal and external validation of the cancer specific survival rate model, and the AUC of the ROC curve was 0.722 for internal validation and 0.708 for external validation, which indicated that the prediction model was highly discriminative and accurate.

Conclusion Based on the selected risk factors related to the prognosis of patients with lung metastasis of breast cancer, we establish a prediction nomogram with good accuracy, which provides an effective method for prediction of individual survival.