吲哚菁绿15 min滞留率和肝脏硬度对预测病毒性肝炎失代偿事件发生的初步研究

Comparative study of indocyanine green retention rate at 15 minutes and liver stiffness measurement in predicting the occurrence of decompensated events in viral hepatitis

目的探讨吲哚菁绿15 min滞留率(indocyanine green retention rate at 15 minutes,ICGR15)和肝脏硬度(liver stiffness measurement,LSM)对预测病毒性肝炎失代偿发生的价值。方法回顾性分析2015年12月至2017年12月入住福建医科大学孟超肝胆医院,进行过ICGR15和LSM检查的435例病毒性肝炎患者的临床资料,探索ICGR15与LSM的相关性。提取患者首次入院时的性别、年龄、ICGR15、LSM、白蛋白、血小板等基本信息和生化指标。同时记录患者抗病毒药物的使用情况(药物种类、用药时间)并进行为期3年的随访,依据随访期间是否发生肝硬化并发症,将患者分为失代偿组与代偿组。首先进行单因素分析,比较两组之间各个指标的差异性,之后将单因素分析存在显著性差异或可能起到独立预测作用的指标纳入Cox回归,探索哪些指标可以独立预测失代偿事件的发生,并绘制ROC曲线及决策曲线,比较各个独立危险因素的预测价值,是否存在显著性差异。结果纳入研究的患者435例,其中失代偿组39例(8.97%),代偿组396例(91.03%),ICGR15与LSM的相关性为0.31(P<0.001);单因素分析显示,失代偿组与代偿组以下指标存在显著性差异:LSM[(2.4±0.5)m/s vs(1.9±0.5)m/s,t=3.342,P=0.001]、ICGR15[(29.5±17.0)%vs(17.5±13.3)%,t=3.079,P=0.002]、年龄[(55.9±7.7)岁vs(46.7±12.6)岁,t=2.600,P=0.010]、PLT[(109.5±71.7)×10^(9)L^(-1)vs(149.6±52.7)×10^(9)L^(-1),t=2.602,P=0.010]。Cox分析显示以下指标进入回归方程,可作为失代偿事件发生的独立预测因素,LSM:HR=4.541(95%CI:1.089~7.941)(P=0.038);ICGR15:HR=1.521(95%CI:1.082~2.364)(P=0.035)。LSM、ICGR15及两者结合AUC分别为0.752(95%CI:0.614~0.890)、0.743(95%CI:0.624~0.862)、0.842(95%CI:0.713~0.971),曲线下面积差异无统计学意义(P>0.05)。决策曲线也提示三者无差异。结论LSM与ICGR15存在相关性,两者均可作为病毒性肝炎失代偿事件发生的独立预测因子,两者预测效率无差异,将两者结合并不能显著提高预测效率。

Objective To compared the value of indocyanine green retention rate at 15 minutes(ICGR15)and liver stiffness measurement(LSM)in predicting decompensation of viral hepatitis.Methods Clinical data of 435 patients with viral hepatitis admitted to Mengchao Hepatobiliary Hospital of Fujian Medical University from Dec.2015 to Dec.2017 who underwent ICGR15 and LSM examination were retrospective analyzed to explore the correlation between ICGR15 and LSM.At the same time,the basic information and biochemical indexes such as gender,age,ICGR15,LSM,albumin and platelet were extracted from the patients at the first admission.These patients were followed up for 3 years,and they were divided into decompensated group and compensated group according to whether cirrhosis complications occurred or not.At the same time,the use of antiviral drugs(drug type and medication time)was recorded.Firstly,the univariate analysis was carried on,and the differences of indicators between two groups were compared.Then,indicators with significant differences in univariate analysis or that may play an independent role in predicting decompensation were included in Cox regression to explore which indicators could independently predict the occurrence of decompensations.And ROC curve and decision curve were drawn to compare the predictive value of each independent risk factor,and whether there were significant differences between them.Results A total of 435 patients were included in this study,including 39 patients(8.97%)in the decompensated group and 396 patients(91.03%)in the compensation group.The correlation between ICGR15 and LSM was 0.31(P<0.001).Univariate analysis showed significant differences in the following indicators between the compensation group and the decompensated group:LSM[(2.4±0.5)m/s vs(1.9±0.5)m/s,t=3.342,P=0.001],ICGR15[(29.5±17.0)%vs(17.5±13.3)%,t=3.079,P=0.002],age[(55.9±7.7)years old vs(46.7±12.6)years old,t=2.600,P=0.01],PLT[(109.5±71.7)×10^(9)L^(-1)vs(149.6±52.7)×10^(9)L^(-1),t=2.602,P=0.010].Cox analysis showed that the following indicators were included in the regression equation and could be used as independent predictors of decompensated events,LSM:HR=4.541(95%CI:1.089-7.941)(P=0.038);ICGR15:HR=1.521(95%CI:1.082-2.364)(P=0.035).The AUC of LSM,ICGR15 and the combination of them were 0.752(95%CI:0.614-0.890),0.743(95%CI:0.624-0.862),0.842(95%CI:0.713-0.971),and there was no significant difference in the area under the curve(AUC)(P>0.05).The decision curve also suggested that there was no significant difference of them.Conclusion There is a correlation between LSM and ICGR15,both of them could be used as independent predictors of viral hepatitis decompensatory events.There is no significant difference in the prediction efficiency between the two,and the combination of them can not significantly improve the prediction efficiency.