摘要
目的建立一种基于实验室指标和CT、超声检查指标的无创模型来预测乙肝后肝硬化食管胃静脉曲张破裂出血的风险,尝试替代有创检查。方法回顾我院2010年7月至2017年7月乙肝肝硬化患者3 643例,其中资料完整的肝硬化食管胃静脉曲张破裂出血的患者412例,明确诊断为肝硬化但未发生食管胃静脉曲张的患者340例,分为研究组(静脉曲张出血)与对照组(无静脉曲张),采用倾向性评分匹配(PSA)进行1∶1配对,结果纳入研究组与对照组各248例,共计496例。采用倾向值匹配法,用SPSS 24. 0统计软件,建立2个Logistic风险预测模型,并用未纳入模型的164例静脉曲张出血及92例无静脉曲张患者进行预测验证,同时对模型的稳定性进行进一步验证。结果食管胃静脉曲张破裂出血风险的显著相关参数包括总胆红素(TBIL)、血小板计数(PLT)、红细胞计数(RBC)、门脉直径、脾脏长度、脾脏宽度;建立了2个风险预测模型:基于实验室指标和腹部超声指标的P1、基于实验室指标和腹部CT指标的P2。P1 ROC曲线下面积为0. 970(0. 956～0. 985),预测ROC的面积为0. 938(0. 906～0. 970)(P <0. 001),P2 ROC曲线下面积为0. 967(0. 950～0. 984),预测ROC的面积为0. 946(0. 916～0. 975)(P <0. 001)。两模型对于预测的敏感度和特异度以及总体预测的准确率都在90%以上,且模型稳定性好。结论本研究建立的无创模型在预测肝硬化食管胃静脉曲张破裂出血方面准确性高,作为一种非侵入性的测量方法,在进一步改进后可替代侵入性的方法。
Objective To establish 2 noninvasive models based on laboratory serum indexes,enhanced CT or ultrasonographic parameters to predict hemorrhagic risk of gastroesophageal varices in the patients with post-hepatitis B cirrhosis in order to replace the invasive methods. Methods A total of 3 643 patients with hepatitis B cirrhosis admitted in our hospital between July 2010 and July 2017 were enrolled and retrospectively analyzed. Among them,there were 412 cases with complete data of gastroesophageal variceal hemorrhage and 340 cases with a clear diagnosis of cirrhosis,but no gastroesophageal varices. According to the propensity score analysis（ PSA）,these patients were divided into the research group（ varicose bleeding）and the control group（ no varicose veins） at a ratio of 1∶ 1,and eventually,248 cases were included in each group,with a total of 496 cases. Two Logistic risk prediction models were established by using the dip matching method and SPSS 24. 0 statistical software,and were further verified in 164 patients with variceal bleeding and 92 patients without varicose veins who were not enrolled in the study. Results Risk factors of esophageal and gastric variceal bleeding included total bilirubin（ TBIL）,platelet（ PLT）,red bold cell count（ RBC）,portal vein diameter,and spleen length and width. There were 2 risk prediction models established：model P1 based on the laboratory indicators combined with the abdominal ultrasound indicators,and model P2 based on the laboratory indicators combined with the abdominal CT indicators. The area under the ROC curve of the P1 model was 0. 970（ 0. 956 ～ 0. 985）,and the predicted area of the ROC was 0. 938（ 0. 906 ～0. 970）（ P〈0. 001）. The area under the ROC curve of the P2 model was 0. 967（ 0. 950 ～ 0. 984）,and the predicted ROC area was 0. 946（ 0. 916 ～ 0. 975）（ P〈0. 001）. The sensitivity and specificity of the 2 models for prediction and the overall prediction accuracy were all 〉90%,with sound stablity. Conclusion
作者
徐征国
刘恩
赵泳冰
杨仕明
XU Zhengguo;LIU En;ZHAO Yongbing;YANG Shiming(Department of Gastroenterology,Second Affiliated Hospital,Army Medical University（Third Military Medical University）,Chongqing,400038,China)
出处
《第三军医大学学报》
CAS
CSCD
北大核心
2018年第21期1978-1984,共7页
Journal of Third Military Medical University
