慢性乙型肝炎患者肝硬化无创诊断模型的验证

Validation of a noninvasive diagnostic model for liver cirrhosis in patients with chronic hepatitis B

目的验证已建立的肝硬化预测模型——铜蓝蛋白（CP）模型，评价其对肝硬化的诊断价值。方法收集2011年7月至2013年2月福建医科大学附属第一医院肝病中心确诊的慢性乙型肝炎（CHB）患者155例，作为验证组，同步进行肝活组织病理学、CP及生物化学指标检查。对前期由148例CHB患者构建的CP模型（模型组）进行内部验证，并与aspartate aminotranSferase／platelet ratio index（APRI）、globulin／plateletmodel（GP）、aspartate aminotransferase-／platelet／T-glutamyltransferase／α—fetoproteinindex（APGA）、FItb4积分和age／plateletindex（AP）分别进行正态性Z检验的两两比较。结果将CP模型应用于验证组，判断肝硬化的曲线下面积（AUC）为o．83，灵敏度为77．9％，特异度为83．3％，阳性预测值为46．1％，阴性预测值为95．4％，验证组与模型组的受试者工作特征（ROC）AUC分别为0．83和0．87，差异无统计学意义（Z=1．056，P=0．291），CP模型在两组的诊断价值相近。与APRl、GP、APGA、FIB4积分和AP模型比较，CP模型与肝硬化的相关性最高（r=0．554，P〈O．01）；而APRI与肝硬化无相关性（r=0．095，P〉O．05）。GP、AP、FIB-4积分、APGA和APRI模型用于预测肝硬化的AUC分别为0．75、0．74、0．73、0．62和0．56，与CP模型比较，差异有统计学意义（z=1．91、2．11、2．25、4．61和4．02，均P〈o．05）。结论CP模型可较好地预测早期肝硬化，并在一定程度上替代肝组织学检查。

Objective To validate the diagnostic value of the established noninvasive model eeruloplasmin （CP） model for predicting liver cirrhosis in patients with chronic hepatitis B （CHB）. Methods Liver biopsies and blood tests including routine biochemical tests and serum CP measurement were performed in 155 consecutive patients with CHB in Liver Center, First Affiliated Hospital, Fujian Medical University between July 2011 and February 2013 as the validation group. The diagnostic value of CP model, which was established from 148 CHB patients as a model group, was validated internally in the validation group, and was compared with five previously published noninvasive indices by Z test, i. e, the aspartate aminotransferase/platelet ratio index （APRI）, globulin/platelet model （ GP ）, aspartate aminotransferase/platelet/7 glutamyl transferase/a- fetoprotein index （APGA）, FIB-4 index and age/platelet index （AP）. Results Area under the receiving operating characteristic curve （AUC） of CP model to identify patients with liver cirrhosis in the validation group was 0.83, with 77.9＋ sensitivity, 83.3~ specificity, 46.1＋ positive predictivevalue （PPV） and 95.4% negative predictive value （NPV）. The diagnostic value of CP model for the validation group and the model group was comparable, with AUC of 0.83 and 0.87, respectively （Z= 1. 056, P=0. 291）. By Spearman rank correlation analysis, CP model was significantly correlated with cirrhosis （r=0. 554, P〈0.01）. Compared with five available models （APRI, GP, APGA, FIB- 4 index and AP）, CP model showed a highest correlation coefficient, while APRI was not correlated with liver cirrhosis （r= 0. 095, P〉0. 05）. The AUC of the five models （GP, AP, FIB4 index, APGA and APRI） were 0.75, 0.74, 0.73, 0.62 and 0.56, respectively, which were all significantly lower than that of the CP model （Z=I. 91, 2.11, 2.25, 4.61 and 4.02,, respectively; all P〈0.05）. Conclusion The CP model can be used to predict early liver cirrhosis and may reduce the need for liver biopsy among patients with CHB infection.