Abstract
Background Male breast cancer(MBC)is clinically rare,but its annual incidence has increased in recent years and the prognosis is poor.Objective To develop a clinical prediction model for prognosis in male breast cancer patients and externally validate it using data from MBC patients at our institution.Methods Patients with MBC from the SEER database from 1990 to 2019 were selected and randomly divided into a training set and a validation set at a ratio of 7:3.The overall 5-year death of MBC patients was used as the outcome indicator.Lasso regression,univariate,and multivariate Cox proportional hazards models were employed to screen for predictive features.The discriminative ability and calibration of the clinical prediction model were evaluated using receiver operating characteristic(ROC)curves,calibration curves,and external validation analysis.A nomogram was constructed to visualize the clinical prediction model.Results This study included 1 614 patients from the SEER database as the study subjects and divided them into a training set(n=1 129)and a validation set(n=485).Follow-up results showed that 274 died in the training set and 111 patients died in the validation set.LASSO regression identified 9 clinical features:age at diagnosis,histological grade,T stage,N stage,M stage,bone metastasis status,surgical intervention status,estrogen receptor(ER)status,and progesterone receptor(PR)status.Univariate Cox proportional hazards model analysis revealed that all variables were significantly associated with risk of death in MBC(P<0.05).Multivariate Cox proportional hazards model analysis demonstrated that age,tumor grade,T stage,N stage,M stage,and bone metastasis were independent risk factors for risk of death.At the same time,surgical intervention,ER positivity,and PR positivity served as protective factors(P<0.05).In the development set,the AUC values for the 1-,3-,and 5-year risk of death prediction models were 0.895,0.853,and 0.829,respectively,with a C-index of 0.811.In the validation set,the corresponding AUC values were 0.858,0.858,and 0.848,with a C-index of 0.798.External validation was performed on 44 patients with MBC in our hospital.The AUC for 5-year risk of death was 0.731,and the 10-year risk of death was 0.797.Conclusion This study establishes a prediction model based on diagnostic age,histological grading,T staging,N staging,M staging,bone metastasis status,surgical intervention status,ER status,and PR status,which has high predictive value for risk of death in MBC patients and provides a reliable tool for clinical doctors.关键词
男性/乳腺癌/临床预测模型/机器学习/列线图/预后/死亡Key words
male/breast cancer/clinical prediction model/machine learning/nomogram/prognosis/death分类
医药卫生
