Diagnostic performance of non-invasive fibrosis markers for chronic hepatitis B in sub-Saharan Africa: a Bayesian individual patient data meta-analysis

Objective In sub-Saharan Africa, hepatitis B is the principal cause of liver disease. Non-invasive biomarkers of liver fibrosis are needed to identify patients requiring antiviral treatment. We assessed aspartate aminotransferase-to-platelet ratio index (APRI), gamma-glutamyl transferase-to-platelet ratio (GPR) and FIB-4 to diagnose significant fibrosis and cirrhosis in an individual patient data (IPD) meta-analysis. Design In total, 3,549 patients from 12 cohorts of HBsAg positive individuals in 8 sub-Saharan African countries were included. Transient elastography was used as a reference test for cirrhosis (>12.2 kPa), excluding patients who were pregnant, had hepatitis C, D, or HIV co-infection, were on hepatitis B therapy, or had acute hepatitis. A bivariate Bayesian IPD model was fitted with patient-level covariates and study-level random effects. Results APRI and GPR had the best discriminant performance (area under receiver operating curve 0.81 and 0.82) relative to FIB-4 (0.77) for cirrhosis. The World Health Organization (WHO) recommended APRI threshold of [≥]2.0 was associated with a sensitivity and specificity (95% credible interval) of 16.5% (12.5-20.5) and 99.5% (99.2-99.7) for cirrhosis. For APRI, we identified an optimised rule-in threshold for cirrhosis (cut-off 0.65) with a sensitivity and specificity of 56.2% (50.5-62.2) and 90.0% (89.0-91.0), and an optimised rule-out threshold (cut-off 0.36) with a sensitivity and specificity of 80.6% (76.1-85.1) and 64.3% (62.8-65.8). Conclusions The WHO recommended APRI threshold of 2.0 is too high to diagnose cirrhosis in sub-Saharan Africa. We identified new and optimised rule-in and rule-out thresholds for cirrhosis, with direct consequences for treatment guidelines in this setting.