摘要
目的探讨采用Logistic回归模型联合的肿瘤标志物在胃癌诊断中的价值。方法收集来我院就诊的原发性胃癌患者183例和对照者120例,所有入组者均需检测CEA、CA19-9、CA72-4和CA242四种肿瘤标志物。将所有的胃癌患者和对照者都分为两部分,一部分采用前进法Logistic回归将肿瘤标志物联合在一起,并采用受试者工作特征(ROC)曲线评价组合对胃癌的诊断价值。另一部分用于对其组合的诊断效率验证。结果四种肿瘤标志物都被选入了回归模型,其组合的ROC曲线下面积为0.844,灵敏度为72%,正确率为76.4%。诊断胃癌的ROC曲线下面积为0.819(95%的可信区间为0.696~0.942),灵敏度为78.8%,特异性为85%,正确率为81.1%。结论四种肿瘤标志物联合的方法可以很大地提高对胃癌的诊断效率。经验证预测概率值的方程具有很高的正确率,检查者可以通过此方程方便直观地得出胃癌的预测值。
Objective To explore the diagnostic value of logistic regression model combined tumor markers in gastric cancer.Methods 183 cases of primary gastric cancer and 120 cases of benign controls were collected from our hospital.Four tumor markers including CEA,CA19-9,CA72-4 and CA242 were detected in all the patients.The received gastric cancer patients and controls were divided into two parts for the model development.One part used forward logistic regression to combine tumor markers,and ROC curve was used to evaluate the diagnostic value of combination for gastric cancer.The other part was used to verify the diagnostic accuracy of its combination.Results The four tumor markers were combined by logistic regression.The area under the receiver operation characteristic(ROC)curve was 0.844,with the sensitivity 72% and the accuracy 76.4%.When we tried to verify the efficiency of this equation in the diagnosis of gastric cancer,the area under ROC curve was 0.819(95% confidence interval was 0.696-0.942),with the sensitivity 78.8%,the specificity 85%,and the accuracy81.1%.Conclusion The combination of four tumor markers can greatly improve the diagnostic efficiency of gastric cancer.This model has proved to be very accurate,and the examiner can easily and intuitively obtain the predictive value of gastric cancer through this model.
作者
王志刚
陈丁莉
李守霞
冯丽娟
张彦青
WANG Zhi-gang;CHEN Ding-li;LI Shou-xia;FENG Li-juan;ZHANG Yan-qin(Department of Laboratory,Handan Central Hospital,Handan 056008,China)
出处
《标记免疫分析与临床》
CAS
2019年第11期1898-1902,共5页
Labeled Immunoassays and Clinical Medicine
