论文部分内容阅读
目的:应用Logistic回归和ROC曲线探讨CEA、CA15.3和细胞角蛋白19片段(CYFRA21-1)对良恶性胸腔积液的诊断价值。方法:对明确诊断的胸腔积液患者93例(良性51例,恶性42例)同时采血和抽取胸液,采用化学发光法测定CEA、CA15.3和CYFRA21-1的浓度,计算特异性90%时的曲线下面积(AUC)、敏感性和阴性预测值。结果:良恶性胸水患者胸液CEA、CA15.3和CYFRA21-1的浓度两者之间差别有统计学意义(P<0.01)。胸液/血清比值的诊断效率并不优于胸液中单个肿瘤标志物浓度测定。单项检测中,CEA敏感性最高(59.3%),CYFRA21-1的曲线下面积最大(0.91)。CEA、CYFRA21-1和CA15.3联合检测时,敏感性增高到85.1%,阴性预测值增加到84.0%,AUC增加到0.91。结论:基于Logistic回归的ROC曲线分析简单有效,适用于多指标联合诊断试验的分析评价。在良恶性胸腔积液的鉴别诊断中,CEA、CYFRA21-1和CA15.3的组合有较好的诊断价值。
Objective: To investigate the diagnostic value of CEA, CA15.3 and cytokeratin 19 (CYFRA21-1) in the diagnosis of benign and malignant pleural effusion by Logistic regression and ROC curve. Methods: Ninety-three patients with benign pleural effusion (benign 51 and malignant 42) were collected blood and drawn pleural effusion simultaneously. The concentrations of CEA, CA15.3 and CYFRA21-1 were determined by chemiluminescence method, and the specificity of 90% Area under the curve (AUC), sensitivity and negative predictive value. Results: There was a significant difference between CEA, CA15.3 and CYFRA21-1 levels in benign and malignant pleural effusion (P <0.01). The diagnostic efficiency of pleural fluid / serum ratio is not superior to the single tumor marker concentration in pleural fluid. The highest sensitivity (59.3%) for CEA and the highest area under the curve for CYFRA21-1 (0.91) were found in the individual tests. CEA, CYFRA21-1 and CA15.3 combined detection, the sensitivity increased to 85.1%, the negative predictive value increased to 84.0%, AUC increased to 0.91. Conclusion: ROC curve analysis based on Logistic regression is simple and effective and is suitable for the analysis and evaluation of multi-index combined diagnostic test. In the differential diagnosis of benign and malignant pleural effusion, CEA, CYFRA21-1 and CA15.3 combination has a good diagnostic value.