A novel diagnostic algorithm to predict significant liver inflammation in chronic hepatitis B virus infection patients with detectable HBV DNA and persistently normal alanine transaminase

Abstract

Significant liver inflammation might be found in 20–34% of chronic hepatitis B virus (HBV) infection patients with detectable HBV DNA and persistently normal alanine transaminase (ALT) (PNALT). We aimed to develop a diagnostic algorithm to predict significant liver inflammation in these specific patients. Using liver biopsy as the gold standard, we developed a novel, simple diagnostic algorithm to predict significant liver inflammation in a training set of 365 chronic HBV infection patients with detectable HBV DNA and PNALT, and validated the diagnostic accuracy in a validation set of 164 similar patients. The novel algorithm (AAGP) attributed to age, ALT, gamma-glutamyl transpeptidase (GGT), and platelet count was developed. In the training set, the area under the receiver operating characteristic curve (AUROC) of AAGP was higher than that of ALT and aspartate transaminase (AST), to diagnose significant liver inflammation (0.77, 0.67, and 0.59, respectively, p < 0.001). In the validation set, the AUROC of AAGP was also higher than ALT and AST (0.75, 0.61, and 0.54, respectively, p < 0.001). Using AAGP ≥2, the sensitivity and negative predictive value (NPV) was 91% and 93%, respectively, to diagnose significant liver inflammation. Using AAGP ≥8, the specificity and NPV was 91% and 86%, respectively, for significant liver inflammation. In conclusion, the AAGP algorithm is a novel, simple, user-friendly algorithm for the diagnosis of significant liver inflammation in chronic HBV infection patients with detectable HBV DNA and PNALT.