Development and Validation of a Non-invasive Model to Predict Liver Histological Lesions in Chronic Hepatitis B Patients With Persistently Normal Alanine Aminotransferase and Detectable Viremia

Abstract

BackgroundA critical and controversial issue is whether antiviral therapy should be recommended in chronic hepatitis B virus (HBV) infection patients with persistently normal alanine aminotransferase (PNALT) and detectable HBV DNA. The study aimed to develop a non-invasive model for predicting significant liver histological changes (SLHC), which is the histological indication for antiviral therapy in chronic hepatitis B (CHB) patients with PNALT and detectable HBV DNA.Methods398 chronic HBV infection patients with PNALT and detectable HBV DNA who underwent liver biopsy were divided into the estimation set (n = 256) and validation set (n = 142). A multivariate logistic regression model was developed to predict SLHC in the estimation set, and the diagnostic performance was further validated in the validation set.Results132 patients (33.2%) with PNALT and detectable HBV DNA had SLHC. Aspartate aminotransferase (AST), cholinesterase (ChE), and liver stiffness measurement (LSM) were identified as the independent predictors of SLHC. The AUROC of the SLHC index, which combined AST, ChE, and LSM, was 0.824 and 0.816 in the estimation and validation set, respectively, for the prediction of SLHC. Applying the SLHC index ≤ 0.15, the presence of SLHC could be excluded with high negative predictive value in the estimation set (93.2%) and in the validation set (90.2%). Applying the SLHC index ≥ 0.55, the presence of SLHC could be considered with high positive predictive value in the estimation set (79.2%) and in the validation set (76.5%).ConclusionThe SLHC index provides a high accuracy in predicting liver histological indication for antiviral therapy in CHB patients with PNALT and detectable HBV DNA.

A critical review of diagnostic and prognostic markers of chronic hepatitis B infection

“…Numerous studies have demonstrated the accuracy with which non-invasive models, such as the FIB-4 and APRI indices [ 45 ], the ALT, Age, PLT, and LS model [ 48 ], the AAR, API, APRI, KING, Zeugma scores, and Hudu’s score [ 49 , 50 ], can evaluate liver histology and forecast substantial liver injury. It has been discovered that these non-invasive models exhibit remarkable accuracy, with values of the area under the curve ranging from 0.595 to 0.870 [ 51 , 52 ]. They can assist in determining whether antiviral therapy is necessary without requiring an invasive biopsy of the liver, which carries some risk.…”

“…The conventional approach to determining the prognosis of CHB infection involves liver biopsy; however, there are several drawbacks to this method, including cost, potential problems, and invasiveness [ 45 , 48 , 49 ]. An alternate method that has benefits including becoming kinder to the body, safer, and more economical is to use non-invasive diagnostics [ 51 , 52 ]. To forecast liver histological alterations and fibrosis in CHB patients, several non-invasive models were created utilising various indicators [ 53 ], [ 54 ], [ 55 ].…”