A novel liver stiffness measurement-based prediction model for cirrhosis in hepatitis B patients

Abstract

BACKGROUNDS/AIMS: While liver stiffness measurement (LSM) predicts histological cirrhosis accurately, complementary methods are needed for better performance. Furthermore, alanine aminotransferase (ALT) influences LSM, making it necessary to modify its use in patients with high ALT levels. We developed a new LSM-based prediction model for cirrhosis and estimated the thresholds for different ALT levels.

METHODS: From 2008 to 2009, we prospectively enrolled 330 consecutive patients who were diagnosed with chronic hepatitis B (CHB) and underwent a liver biopsy and LSM on the same day. For detection of cirrhosis, we performed univariate and multivariate analyses, using the chi(2)-test/t-test and logistic regression respectively. Thereafter, a prediction model was derived from multivariate predictors.

RESULTS: In multivariate analyses of patients with and without cirrhosis, we found significant differences in the LSM, spleen diameter and platelet count. Then, we developed an LSM-spleen diameter to platelet ratio index (LSPI): (LSM x spleen diameter/platelet count) x 100. The area under the receiver operating curve was 0.956, significantly higher than LSM alone (0.919, P=0.032). We suggested different thresholds in patients with ALT< or = upper limit of normal (ULN) (normal-ALT group, 164 patients) and ALT>ULN (high-ALT group, 166 patients). In the normal-ALT group, LSPI thresholds of 38 and 62 provided 95.7% negative predictive value (NPV) and a 95.5% PPV (positive predictive value), while in the high-ALT group, thresholds of 42 and 94 yielded 95.1% NPV and 96.4% PPV respectively. Therefore, liver biopsy could be avoided in 76.7% of the subjects.

CONCLUSIONS: LSPI is a useful, non-invasive tool that can replace liver biopsy in the assessment of liver fibrosis in the majority of CHB patients