Algorithm of Golgi protein 73 and liver stiffness accurately diagnoses significant fibrosis in chronic HBV infection.

BACKGROUND & AIMS: Serum Golgi protein 73 (GP73) is a potential biomarker for fibrosis assessment. We aimed to develop an algorithm based on GP73 and liver stiffness (LS) for further improvement of accuracy for significant fibrosis in patients with antiviral-naïve chronic hepatitis B virus (HBV) infection. METHODS: Diagnostic accuracy evaluation of GP73 and development of GP73-LS algorithm was performed in training cohort (n = 267) with an independent cohort (n = 133) for validation. RESULTS: A stepwise increasing pattern of serum GP73 was observed across fibrosis stages in patients with antiviral-naïve chronic HBV infection. Serum GP73 significantly correlated (rho = 0.48, P < .001) with fibrosis stage and was an independent predictor for the presence of significant fibrosis (OR, 95%CI: 1.02, 1.01-1.03, per increase in 1 ng/mL, P < .001). Both LS (AUROC, 95%CI: 0.82, 0.77-0.87, accuracy: 74.7%) and GP73 (AUROC, 95%CI: 0.76, 0.71-0.82, accuracy: 71.5%) well-predicted significant fibrosis and outperformed APRI (AUROC, 95%CI: 0.69, 0.63-0.76, accuracy: 66%) and FIB-4 (AUROC, 95%CI: 0.66, 0.60-0.73, accuracy: 63.6%). Using GP73-LS algorithm, GP73 < 63 in agreement with LS < 8.5 provided accuracy of 81.7% to excluded significant fibrosis. GP73 ≥ 63 in agreement with LS ≥ 8.5 provided accuracy of 93.3% to confirm significant fibrosis. Almost 64% or 68% of patients in the training or validation cohort could be accurately classified. CONCLUSIONS: Serum GP73 is a robust biomarker for significant fibrosis diagnosis. GP73-LS algorithm provided better diagnostic accuracy than currently available approaches. More than 60% antiviral naïve CHB patients could use this algorithm without resorting to liver biopsy.