乳腺癌脑转移预测模型的研究

The predictive model of brain metastases of breast cancer

目的探讨乳腺癌患者发生脑转移的危险因素和利用Logistic回归模型预测乳腺癌脑转移概率。方法对165例乳腺癌患者的年龄、分期、雌激素受体（ER）、孕激素受体（PR）及人类表皮生长因子受体-2（Her-2）评分、月经状态、是否进行辅助化疗、远处第一转移灶、无转移生存期进行分析，建立乳腺癌脑转移预测模型。结果确诊年龄〈35岁、腋淋巴结转移数≥4个、ER阴性、Her-2阳性是乳腺癌发生脑转移的独立危险因素（P〈0．05）。获得的预测模型的预测能力为0．783±0．037，预测脑转移发生的灵敏度、特异度可分别达0．786和0．765。结论临床可利用该模型预测乳腺癌患者脑转移的概率，且预测概率≥0．75。

Objective To investigate the risk factors of brain metastases of breast cancer and pre- dictive ability of the predictive model of brain metastases in breast cancer. Methods All the 165 breast cancer patients were analyzed for age, stage, estrogen receptor （ER）, progesterone receptor （PR）, hu- man epidermal growth factor receptor-2 （Her-2） , number of metastasic axillary menstrual status, adjuvant chemotherapy regimen, first sites of metastasis and metastasis-free survival. Results Younger than 35 years old at diagnosis, more than or equal to four metastatic axillary nodes, negative estrogen receptor and postive Herb-2 were independent risk factors associated with brain metastases of breast cancer. The predic- tive ability of the predictive model, which was measured as area under the curve, was 0. 783 ± 0. 037, and the sensitivity and specificity for predicting the occurrence of brain metastases were 0. 786 and 0. 765 re- spectively. Conclusion The predictive model can be used for predicting the occurrence of brain metastases of breast cancer.