Development and validation of a novel circulating fibroblast activation protein - based predictive model to improve fibrosis risk stratification in metabolic liver disease population

Abstract

Objective: Metabolic fatty liver disease drives chronic liver injury leading to fibrosis. This study aimed to establish a model utilising serum circulating fibroblast activation protein (cFAP) to diagnose advanced fibrosis in patients with fatty liver disease. Design: Two retrospective cohorts recruited from tertiary hepatology clinics were studied as training (n=160) and external validation cohorts (n=342), with prevalence of histologic advanced fibrosis (F3/F4) of 20% and 11%, respectively. A marker of activated mesenchymal fibrogenic cells, cFAP, was measured using our single-step enzyme assay. A predictive model, FAP Index, containing age, type 2 diabetes, alanine transaminase and ordinal cFAP was developed using logistic regression. Diagnostic accuracy of FAP Index was assessed on a single and then sequential basis. Results: FAP Index AUROC was 0.875 (95% CI 0.813-0.938) in the training cohort and 0.841 (95% CI 0.776-0.906) in the validation cohort. Low cut-off -1.68 (Sensitivity 80.0%, negative predictive value 95.5%) and high cut-off +0.953 values (Specificity 97.7%, positive predictive value 88.9%) excluded and diagnosed advanced fibrosis, respectively. In the validation cohort, FAP Index then FIB-4 reduced indeterminate results by one-third compared to FIB-4 alone. Whereas FAP-Index followed by NFS (NAFLD Fibrosis Score) resulted in a reduction of indeterminate results by 70% compared to NFS alone. Conclusion: FAP Index is a novel, rapid, robust, inexpensive diagnostic tool for advanced fibrosis in metabolic fatty liver disease. Applying FAP Index followed by FIB-4 or NFS facilitates accurate risk-stratification of patients by greatly reducing the frequency of indeterminate results compared to FIB-4 or NFS alone, without compromising negative predictive value.