结直肠癌伴同时性肝转移的特征性术前临床指标及预测模型

目的：探讨结直肠癌伴同时性肝转移的特征性临床术前指标，并期基于此些指标建立预测模型。方法：1052例结直肠癌患者中以伴或不伴同时性肝转移为分组，使用信息熵增益方法筛选出特征性术前临床指标，并且使用 Logistic 回归方法建立预测模型；在150例患者（独立样本）中进行模型测试。结果：筛选出CEA，CA50，CA199，肿瘤部位（结肠/直肠），原发肿瘤最大径为特征性术前临床指标。以术前血清CEA及CA50值建立模型，敏感性0.82，特异性0.7342，准确率0.7423，ROC曲线下面积AUC为0.838；模型测试表现：ROC曲线下面积AUC为0.9046。结论：CEA，CA50，CA199，肿瘤部位（直肠/结肠），原发肿瘤肿瘤最大径等术前临床指标与结直肠癌伴同时性肝转移有关，以术前血清CEA及CA50值建立Logistic回归模型对其诊断有一定的帮助。

Characteristic preoperative clinical indexes and prediction model for colorectal cancer with synchronous liver metastasis

Objective： To discuss characteristic preoperative clinical indexes for colorectal cancer with synchronous liver metastasis and establish prediction model based on these indexes. Methods： 1052 colorectal cancer patients were divided into two groups, with or without synchronous liver metastasis. Information entropy method was used to gain preoperative clinical characteristic indexes and Logistic regression method was used to establish prediction model. The model was tested in an independent sample of 150 colorectal cancer patients. Results： CEA, CA50, CA199, tumor location （colon/rectum） and maximum diameter of primary tumor were selected as the characteristic preoperative clinical indexes. Preoperative serum CEA and CA50 were used to set up prediction model with a sensitivity of 0.82, specificity of 0.7342 and accuracy of 0.7423, the AUC was 0.838. Model test performance showed the AUC was 0.9046. Conclusion： CEA, CA50, CA199, tumor location （colon/rectum） and maximum diameter of primary tumor were related with colorectal cancer with synchronous liver metastasis. Logistic regression model based on CEA and CA50 may be helpful for the diagnosis of synchronous liver metastasis in colorectal cancer patients.