Improving transient elastography performance for detecting hepatitis B cirrhosis

BackgroundTransient elastography is a well-established method for detecting cirrhosis.AimTo improve the performance of transient elastography in detecting hepatitis B cirrhosis by alanine aminotransferase (ALT)-stratified cutoffs, bilirubin normalization and transient elastography-based algorithms.MethodsA total of 315 compensated patients were analysed following liver biopsies, transient elastography, ultrasonography and blood tests.ResultsThe area under the receiver operating characteristics (ROC) curve of transient elastography for predicting cirrhosis was 0.88 (95% confidence interval 0.84–0.92). The cutoffs to exclude and confirm cirrhosis were 10.4 kPa and 17.3 kPa in patients with ALT <5 × upper limit of normal range, 13.7 kPa and 25.0 kPa in ALT ≥5 × upper limit of normal range, respectively. With ALT-stratified cutoffs, 68.6% of patients did not require liver biopsies. Areas under the ROC curve in patients with normal or abnormal bilirubin was 0.90(0.85–0.95) and 0.84(0.77–0.92), respectively. In patients with normal bilirubin, the cutoffs for excluding and confirming cirrhosis were 10.6 kPa and 16.9 kPa, respectively. By transient elastography screening, 78.3% of patients with normal bilirubin would not need a liver biopsy. Areas under the ROC curves between transient elastography and transient elastography-based algorithm including transient elastography–splenomegaly–platelet index [0.90(0.86–0.94)] and ultrasonic score-transient elastography index [0.91(0.86–0.96)] were not significantly different.ConclusionsAmongst ALT-stratified cutoffs, bilirubin normalization and transient elastography-based algorithm, bilirubin normalization was especially important for improving performance of transient elastography for compensated hepatitis B cirrhosis detection.