Evaluation of new serum markers as predictors of liver fibrosis in patients with chronic liver disease irrespective of the aetiology

Background Identification of non-invasive serum fibrosis markers in liver disease is challenging. We aimed to identify a new marker panel for the identification of significant fibrosis in a large cohort of patients with chronic liver disease with mixed aetiology. Patients and methods Patients with any chronic liver disease were included in the study. Transient elastography (TE) was employed to phenotype the patients. Significant fibrosis was defined by TE results >7.5kPa and no-fibrosis by TE results <7.5 kPa. Serum levels of hsTNT, sFLT-1, PLGF, NT-proBNP, GDF-15, adiponectin, BMP-7, endoglin and HGF were determined and ROC analysis was applied to determine the predictive values for the prediction of fibrosis. Uni- and multivariate regression analyses were employed to identify independent predictors of significant fibrosis. Results Overall, 919 patients were included (median age 50y (17–88), m: f: 648: 402; fibrosis group: n=404, no-fibrosis: n=515). The predominant liver diseases were chronic HCV infection (n=585; 55.5%), NASH (n=105; 10.5%) and alcoholic liver disease (n=105; 9.9%). All tested markers, except adiponectin, were significantly correlated with unclassified TE results (p for all tests <0.001). In a ROC analsis GDF-15 and HGF as well as PLGF achieved the highest AUC (>0.75, p<0.001). Multivariate regression analysis including age and the new serum markers revealed that only PLGF, endoglin and HGF were independent predictors of significant fibrosis in our cohort. Subgroup analysis in patients with chronic HCV infection showed that patients with PLGF levels >20.8 pg/ml had the highest risk to present with significant fibrosis (RR=7.84 ; 95% C.I.: 5.12–12.00; p<0.001). Conclusions Our analysis identified a novel panel of new serum markers with promising predictive values for the identification of patients with significant fibrosis. However, further analysis is needed to clarify the predictive properties in clinical practice.