简易无创模型预测慢性乙型肝炎肝脏病理状态的评价  

Appraisement of simple non-invasive models for predicting pathological status of liver in patients with chronic hepatitis B

作　　者：张占卿[1] 陆伟[1] 史连国[2] 冯艳玲[2] 

机构地区：[1]复旦大学附属公共卫生临床中心肝病科,上海201508 [2]复旦大学附属公共卫生临床中心病理科,上海201508

出　　处：《同济大学学报（医学版）》2008年第5期59-63,共5页Journal of Tongji University(Medical Science)

摘　　要：目的评价简易无创模型AAR(AST-to-ALT ratio)、APRI(AST-to-platelet ratio index)、SPRI(Spleen-to-platelet ratio index)、API(age-platelet index)、ASPRI(age-spleen-to-platelet ratio index)预测慢性乙型肝炎肝脏病理状态的实践效能。方法慢性乙型肝炎170例,其中病理分级G1、G2、G3、G4的患者分别为28例、87例、49例、6例,病理分期S0、S1、S2、S3、S4的患者分别为6例、32例、43例、57例、32例。简易无创模型参照原始文献的描述构建。统计分析采用SPSS 13.0软件。简易无创模型与病理分级、分期之间的相关分析采用Spearman相关法。简易无创模型在不同病理分级和分期之间的比较采用单因素方差分析。简易无创模型预测慢性乙型肝炎肝脏病理状态的评价采用Bayes逐步判别分析。结果APRI、SPRI、ASPRI与病理分级呈显著正相关(P<0.05),AAR、APRI、SPRI、API、ASPRI与病理分期呈显著正相关(P<0.01)。AAR、SPRI、API、ASPRI在不同病理分级之间均无显著性差异(P>0.05),APRI在不同病理分级之间有显著性差异(P<0.01)。AAR、APRI、SPRI、API、ASPRI在不同病理分期之间均有显著性差异(P<0.01)。符合纳入变量标准、进入预测不同病理分级的Bayes逐步判别分析的简易无创模型只有APRI,进入预测不同病理分期的Bayes逐步判别分析的简易无创模型只有AAR和API。Fisher’s判别函数预测G1、G2、G3、G4的正确率分别为82.14%、20.69%、16.33%、33.33%,预测S0、S1、S2、S3、S4的正确率分别为16.67%、37.50%、23.26%、43.86%、50.00%。结论基于APRI构建的Fisher’s判别函数对预测轻微炎症活动度(G1)有较大价值,基于AAR和API构建的Fisher’s判别函数对预测严重纤维化程度(S3和S4)有一定价值。Objective To appraise the clinical efficacy of simple non-invasive models of AST-to-ALT ratio （AAR）, AST-to-platelet ratio index （APRI）, Spleen-to-platelet ratio index （SPRI）, age-platelet index （ API）, age-spleen-to-platelet ratio index （ASPRI） for predicting pathological status of liver in patients with chronic hepatitis B. Methods One hundred and seventy patients with chronic hepatitis B were retrospectively studied, in which 28 patients, 87 patients, 49 patients, 6 patients were judged pathologically as G1, G2, G3, G4 respectively ; and 6 patients, 32 patients, 43 patients, 57 patients, 32 patients were judged pathologically as SO, S1, S2, S3,S4 respectively. The simple non-invasive models were calculated as described originally.SPSS 13.0 was used for statistical analyses. One-way ANOVA was used for comparisons the simple noninvasive models among different pathological grading and staging. Bayes stepwise discriminant analysis was used for appraising the clinical efficacy of simple non-invasive models for predicting pathological status of liver in patients with chronic hepatitis B. Results APRI, SPRI, ASPRI were positively correlated with pathological grading significantly （P 〈 0.05 ）, AAR, APRI, SPRI, API, ASPRI were positively correlated with pathological staging significantly （P 〈 0. 01 ）. AAR, SPRI, API, ASPRI had no significantly differences among different pathological grading （ P 〉 0.05 ）, APRI had significantly differences among different pathological grading （ P 〈 0.01 ） ; AAR, APRI, SPRI, API, ASPRI had significantly differences among different pathological staging （P 〈 0.01 ）. Of all the simple non-invasive models, only APRI accorded with the entry criterion （P 〈 0. 05 ） for Bayes stepwise discriminant analysis for predicting different pathological grading, and only AAR and API accorded with the entry criterion （ P 〈 0.05 ） for Bayes stepwise discriminant analysis for predicting different pathological staging. The correct proportions of t

关 键 词：无创模型 慢性乙型肝炎 病理分级 病理分期 Bayes判别分析 

分 类 号：R512[医药卫生—内科学]