Abstract: Background　In recent years，intensive studies in chronic hepatitis B have yielded many noninvasive predictive models to substitute liver biopsy for fibrosis assessment and follow-up.However，the biomarkers required in some noninvasive models are limited in clinical use due to challenges in clinical measurement.Meanwhile，the diagnostic value of the models is also lack of large-scale clinical verification.Objective　To compare the values of six noninvasive fibrotic models of APRI，FIB-4，Forns，HB-F，APGA，and Hui in predicting significant hepatic fibrosis and cirrhosis in chronic hepatitis B patients，to acquire a simple feasible model with the highest diagnostic efficacy for clinical popularization.Methods　Two hundred and twenty-one patients with chronic hepatitis B recruited from the Affiliated Hospital of Southwest Medical University during June 2013 to December 2018 were analyzed retrospectively.All of them underwent liver biopsy，and had routine blood，biochemical and prothrombin time tests within four subsequent weeks.According to the Metavir scoring system，the degree of liver fibrosis of the patients was classified into three groups：F01（n=78，Metavir score=0-1），F23（n=98，Metavir score=2-3），and F4（n=45，Metavir score=4）.Significant fibrosis was defined as Metavir score≥2，cirrhosis was defined as Metavir score=4.Then the scores of six models of noninvasive index（APRI，FIB-4，Forns，HB-F，APGA and Hui ） were calculated.The area under the ROC curve（AUC） analyses of these models in diagnosing significant fibrosis and cirrhosis were performed in patients with ALT ≥ 2ULN or ALT< 2ULN，respectively.Results　 There was significant difference in the scores of six models among the three groups（P<0.05）.The scores of all the models were found to be positively with the severity of liver fibrosis（P<0.05）.For patients with ALT ≥ 2ULN，the AUCs of six models to predict significant fibrosis were 0.836 for APRI，0.738 for FIB-4，0.807 for Forns，0.836 for HB-F，0.912 for APGA，and 0.819 for Hui.The AUCs of APRI，FIB-4，Forns，HB-F，APGA and Hui for predicting cirrhosis were 0.732，0.705，0.789，0.784，0.811 and 0.863，respectively.For patients with ALT< 2ULN，the AUCs of APRI，FIB-4，Forns，HB-F，APGA and Hui were 0.790，0.811，0.825，0.768，0.820，0.787，respectively for predicting significant fibrosis，and 0.815，0.872，0.892，0.857，0.863，0.881 for assessing cirrhosis，respectively.The value of APGA in predicting significant fibrosis in patients with ALT ≥2ULN：when the cutoff value was determined as >0.955，39 of the 42 cases with a APGA score >0.955 were identified with significant fibrosis（Metavir score≥2）；when the cutoff value was determined as >1.005，32 cases with a APGA score >1.005 had significant fibrosis（Metavir score≥2）.The value of APGA in predicting cirrhosis in patients with ALT ≥2ULN：when the cutoff value was determined as >0.955，13 of the 42 cases with a APGA score >0.955 were identified with cirrhosis（Metavir score=4）；when the cutoff value was determined as >1.265，6 of the 7 cases with a APGA score >1.265 had cirrhosis（Metavir score=4）.The value of APGA in predicting significant fibrosis in patients with ALT<2ULN：when the cutoff value was determined as >0.815，93 of the 132 cases with a APGA score >0.955 were identified with significant fibrosis（Metavir score≥2）；when the cutoff value was determined as >1.005，45 of the 47 cases with a APGA score >1.005 were identified with significant fibrosis（Metavir score≥2）.The value of APGA in predicting cirrhosis in patients with ALT<2ULN：when the cutoff value was determined as >0.935，28 of the 70 cases with a APGA score >0.935 were identified with cirrhosis（Metavir score=4）；when the cutoff value was determined as >1.155，3 of the 16 cases with a APGA score >1.155 were diagnosed with cirrhosis（Metavir score=4）.Conclusion　The APGA model has a higher efficacy in diagnosing both significant hepatic fibrosis and cirrhosis for chronic hepatitis B patients with ALT ≥ 2ULN or ALT< 2ULN.Moreover，the measurement of APGA is simple and easy，so the model deserves extensive clinical promotion.

Key words: Liver cirrhosis, Hepatitis B, Biological markers, Diagnosis, Fibrosis, Noninvasive evaluation model of liver fibrosis

摘要： 背景　随着近年对慢性乙型病毒性肝炎（简称慢性乙肝）的深入研究，建立了许多非侵入性预测模型评估慢性乙肝患者的肝纤维化程度，以此替代肝组织活检进行肝纤维化评估和随访，但一些无创模型中所需的生物标志物在临床检测比较困难，临床使用受限；同时其诊断效果亦缺乏大规模临床验证。目的　探讨APRI、FIB-4、Forns、HB-F、APGA、Hui 6种模型对慢性乙肝患者肝纤维化及肝硬化的诊断价值，以期选择一种诊断效能高且简单易行的模型用以临床推广。方法　回顾2013年6月—2018年12月在西南医科大学附属医院治疗的慢性乙肝患者221例的临床资料，患者均曾行肝穿刺活检，且在肝穿刺活检4周内行血常规、生化及凝血检查。肝纤维化分期参照Metavir评分标准评定，并据此分为F01组（Metavir评分为0~1分）、F23组（Metavir评分为2~3分）和F4组（Metavir评分为4分），分别为78例、98例和45例。肝显著纤维化定义为Metavir评分≥2分，肝硬化定义为Metavir评分4分。计算APRI、FIB-4、Forns、HB-F、APGA、Hui评分。分别在丙氨酸氨基转移酶（ALT）≥2倍正常上限ULN及ALT<2倍ULN时，绘制6种模型诊断肝显著纤维化及肝硬化的受试者工作特征曲线（ROC曲线）并计算ROC曲线下面积（AUC）。结果　三组患者6种模型评分比较，差异有统计学意义（P<0.05）。6种模型评分与肝纤维化严重程度呈正相关（P<0.05）。ALT≥2倍ULN时，APRI、FIB-4、Forns、HB-F、APGA、Hui诊断肝显著纤维化的AUC分别是0.836、0.738、0.807、0.836、0.912、0.819，诊断肝硬化的AUC分别是0.732、0.705、0.789、0.784、0.811、0.863。ALT<2倍ULN时，APRI、FIB-4、Forns、HBF、APGA、Hui诊断肝显著纤维化的AUC分别是0.790、0.811、0.825、0.768、0.820、0.787，诊断肝硬化的AUC分别是0.815、0.872、0.892、0.857、0.863、0.881。APGA对ALT≥2倍ULN患者肝显著纤维化的诊断价值：当截断值>0.955分时，42例患者的APGA评分高于0.955分，其中39例Metavir评分≥2分；当截断值>1.005分时，32例患者的APGA评分高于1.005分，Metavir评分均≥2分。APGA对ALT≥2倍ULN患者肝硬化的诊断价值：当截断值>0.955分，42例患者的APGA评分高于0.955分，其中13例Metavir评分=4分；当截断值>1.265分时，7例患者的APGA评分高于1.265分，其中6例Metavir评分=4分。APGA对ALT<2倍ULN患者肝显著纤维化的诊断价值：当截断值>0.815分时，132例患者的APGA评分高于0.955分，其中93例Metavir评分≥2分；当截断值>1.005分时，47例患者的APGA评分高于1.005分，其中45例Metavir评分≥2分。APGA对ALT<2倍ULN患者肝硬化的诊断价值：当截断值>0.935分，70例患者的APGA评分高于0.935分，其中28例Metavir评分=4分；当截断值>1.155分时，16例患者的APGA评分高于1.155分，其中13例Metavir评分=4分。结论　APGA无创肝纤维化模型无论在ALT≥2倍ULN或ALT<2倍ULN时，对慢性乙肝患者肝显著纤维化及肝硬化均具有较高的诊断效能，且该模型涉及的指标简单易得，适合临床推广。

关键词: 肝硬化, 乙型肝炎, 生物学标记, 诊断, 纤维化, 无创肝纤维化模型