Logistic回归和ROC曲线分析多种肿瘤标志物在肝癌诊断中的应用

目的应用Logistic回归和ROC曲线分析肿瘤标志物在肝癌中的诊断。方法收集原发性肝癌63例,良性肝病58例,正常对照76例,用ABBOTT AXSYM化学发光分析仪测定CEA、AFP、CA125和CA199,绘制ROC曲线并计算曲线下面积(AUC)评价各指标的诊断价值。结果肝癌-正常对照组,4项联合诊断肝癌的AUC(0.973)高于各单项检查的AUC(P0.05);肝癌-肝病组,CA125被剔除(P=0.28),CEA、AFP、CA199入选回归方程,CEA、AFP、CA199联检的AUC(0.909)高于AFP的AUC(0.814)(P0.05),2组AFP的AUC均高于其他各单项的AUC(P0.05)。结论 AFP在原发性肝癌的诊断中有较好的应用价值,AFP、CEA、CA199联检可提高诊断效能,为判别良、恶性肝病提供参考,而CA125在判定肝病良、恶性时价值不大。

Assessment of diagnostic value of various tumors markers for primary hepatic carcinoma with logistic regression and ROC curve

Objective To explore the diagnostic value of CEA,AFP,CA125,CA199 and the combined examination of them for primary hepatic carcinoma(PHC) with a model of logistic regression and ROC curve.Methods Serum CEA,AFP,CA125 and CA199 levels were measured with ABBOTT AXSYM in 63 cases with PHC,58 cases with benign disease of liver and 76 normal control cases.Applied logistic regression,plotted ROC curve and calculated the area under ROC curve(AUC) to assess the diagnostic value of each index.Results In PHC-control group,the AUC of combination of CEA,AFP,CA125,CA199 was larger than any AUC of them alone(P<0.05).In PHC-benign disease of liver group,Logistic regression excluding CA125(P=0.28),AFP,CEA and CA199 were selected into Logistic regression equation,the AUC(0.909) of combination of CEA,AFP and CA199 was larger than the AUC(0.814) of AFP(P<0.05),the AUC of AFP was larger than single index in both two groups(P<0.05).Conclusion AFP has a higher value in the diagnosis of PHC,the combined examination of AFP,CEA and CA199 for PHC is superior to any single index examination and it can give more information for differential diagnosis,while CA125 is less helpful.