Establish and analyze the predictive model of early stage brain metastases in patients with breast cancer

Abstract

To investigate the risk factors of cerebral metastasis of breast cancer and to provide guidance for the early diagnosis and treatment of brain metastases. Clinical data of postoperative patients with breast cancer were collected in our hospital from 2005 to 2009. All the patients were divided into two groups, with or without brain metastasis. The risk factors of brain metastases of patients with breast cancer were analyzed by the logistic regression. Eight hundred and twenty four early postoperative patients with breast cancer were enrolled. The median follow-up time was 68 months and 199 cases had brain metastasis. The univariate logistic regression results showed that higher grade of tumor, <35 years, premenopausal, clinical stage Ⅲ, HER-2 positive and ER negative, no adjuvant chemotherapy, no adjuvant normal therapy were the risk factors of brain metastasis. The multivariate logistic regression result showed that higher grade of tumor, <35 years, premenopausal, clinical stage Ⅲ, HER-2 positive, ER negative were the risk factors. The mathematical model was used to predict the probability of occurrence of brain metastasis. ROC curve were draw by the value of predict probability as test variable and the brain metastasis status as state variables. The AUC value of this predictive model was 0.743±0.018. The specificity and sensitivity are relative high. Age<35 years, premenopausal, clinical stage Ⅲ, HER-2 positive and ER negative were independent risk factors for brain metastases. This model has the predictive value for the occurrence of brain metastases from breast cancer.