請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/53618
標題: | B型肝炎帶原者肝臟酵素異常之縱貫性分析 Longitudinal Analysis on Episodes of Elevated Liver Enzymes in Hepatitis B |
作者: | Siao-Ning Huang 黃筱寗 |
指導教授: | 于明暉(Ming-Whei Yu) |
關鍵字: | aspartate aminotransferase (AST),alanine aminotransferase(ALT),B型肝炎帶原者, Aspartate aminotransferase (AST),alanine aminotransferase(ALT),hepatitis B carrier, |
出版年 : | 2015 |
學位: | 碩士 |
摘要: | 背景:肝臟酵素aspartate aminotransferase (AST)及alanine aminotransferase (ALT)是臨床上廣為用於定期追蹤的肝炎標記。研究已指出AST或ALT異常和末期肝病有關,肝細胞癌也包括在其中。然而目前對於AST或ALT異常定義之理想切點值仍有質疑。本研究針對慢性B型肝炎感染者,利用縱式分析評估不同切點定義之肝臟酵素異常值對於未來長期持續異常和肝細胞癌罹病風險的預測能力。 材料與方法:本研究使用縱式世代研究資料,族群來自1989-1992年間在台灣公保健診中心定期接受健康檢查的2,665名HBsAg陽性男性個案,並且持續從1994年追蹤到2014年。總計有12,754時點的AST測量值及14,550時點的ALT測量值資料。定義切點25及40 U/L作為肝臟酵素異常值,以logistic regression model計算相對危險性(odds ratio; OR)以及95%信賴區間,並且使用ROC (receiver operating characteristic)曲線下之面積(AUC)評估模式的預測能力。 結果:以切點為25 U/L,在追蹤19年期間,AST異常盛行率介於37.8%-57.3%,而相對應ALT異常盛行率介於48.3%-59.7%;以切點為40 U/L,AST異常盛行率介於8.4%-13.7%,而其相對應在追蹤22年期間的ALT異常盛行率介於15.3%-23.8%。基線肝臟酵素異常者在其追蹤十年期間,肝臟酵素會持續異常的相對危險性為2-4倍,且OR值會隨著追蹤期間的增加而下降,AST由4.52至2.55,而ALT由3.65至2.53。不論切點,多時點對於持續肝臟酵素活性異常的能力都要比單時點來的佳(單時點預測能力:對於AST,AUC=0.52-0.72,對於ALT,AUC=0.53-0.78;多時點預測能力:對於AST,AUC=0.72-0.84,對於ALT,AUC=0.72-0.84)。而在肝細胞癌的風險預測也有相同發現,使用基線ALT測量值,其AUC為0.74,當加入5年多時點ALT測量值時AUC為0.78,其顯著增加預測的辨別力(p=0.0105)。 結論:五年期間的多時點資料,提升對於長期肝臟酵素持續異常和肝細胞癌之罹病風險的預測能力。未來有必要利用多時點資料建立罹病風險預測模式。 Background: Aspartate aminotransferase (AST) and alanine aminotransferase (ALT) are widely adopted as clinical indicators of hepatitis during routine checkups. Elevated AST or ALT is associated with advanced liver diseases including hepatocellular carcinoma (HCC). However, there is still uncertainty in the optimal definition of cut-off values for elevated liver enzymes. This study aimed to evaluate the impact of different cut-off values of liver enzymes on subsequent long-term measurements of liver enzymes, and their predictive abilities for the risk of HCC during chronic hepatitis B virus (HBV) infection. Methods: We used a longitudinal cohort study database, which includes a total of 2,665 hepatitis B surface antigen (HBsAg)-positive male who were recruited during routine physical examination from government employee central clinics between 1989 and 1992, and followed from 1994 to 2014. There are 12,754 time-points measurements for AST and 14,550 time-points measurements for ALT. Cut-off values of 25 and 40 U/L were used to define elevated liver enzymes. Logistic regression model was used to determined odds ratios (ORs) and 95% confidence intervals (CIs), and derive the area under the receiver operating characteristic curves (AUCs). Results: When using 25 U/L cutoff, the prevalence of elevated AST during 19 years of follow-up was 37.8%-57.3%, and the corresponding prevalence for elevated ALT was 48.3%-59.7%. For 40 U/L as cutoff, the prevalence of elevated AST was 8.4%-13.7%, and the corresponding prevalence for elevated ALT was 15.3%-23.8% during 22 years of follow-up. Elevated liver enzyme at baseline was associated with 2- to 4-fold increase in risk for subsequent elevation of liver enzyme during following 10 years, with the ORs declining over time from 4.52 to 2.55 for AST; and 3.65 to 2.53 for AST. Regardless of cutoff values, incorporation of multiple time-points liver enzyme measurements into regression model significantly improved predictive abilities (for AST: AUC=0.52-0.72; for ALT: AUC=0.53-0.78) for liver enzyme abnormalities in subsequent years, as compared to single time-point measurement (for AST: AUC=0.72-0.84; for ALT: AUC=0.72-0.84). Similar associations were observed for predicting HCC. When using baseline ALT as measurement, the AUC was 0.74. While for measurements spanning over 5 years, the predictive power in terms of AUC was 0.78 that significantly improved discrimination performance (p=0.0105). Conclusion: Multiple time-points measurements of liver enzymes in 5 years increase long-term predictive abilities for persistent liver abnormalities and HCC. The multiple time-point data should be considered for constructing longitudinal predictive algorithm for population stratification for HCC risk. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/53618 |
全文授權: | 有償授權 |
顯示於系所單位: | 流行病學與預防醫學研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-104-1.pdf 目前未授權公開取用 | 1.56 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。