Aims: Our aim was to develop an accurate, non-invasive, blood-test-based method for identifying the main characteristics of liver fibrosis in non-alcoholic fatty liver disease (NAFLD).
Methods: Fibrosis was staged according to NASH-CRN and Metavir systems in 226 patients with NAFLD. A fully automated algorithm measured the fractal dimension (FD) and the area of fibrosis (AOF). Independent predictors of diagnostic targets were determined using bootstrap methods.
Results: (i) Development. Significant fibrosis defined by NASH-CRN F ≥2 was diagnosed by weight, glycaemia, aspartate aminotransferase (AST), alanine aminotransferase (ALT) and prothrombin index [area under the receiver operating characteristic (AUROC)=0.867]; significant fibrosis defined by Metavir F ≥2 was diagnosed by weight, age, glycaemia, AST, ALT, ferritin and platelets (FibroMeter AUROC=0.941, P<0.005). AOF was estimated by the combination of hyaluronic acid, glycaemia, AST, ALT, platelets and prothrombin index ((a) R(2) =0.530), while FD was estimated by hyaluronic acid, glycaemia, AST/ALT, weight and platelets ((a) R(2) =0.529). (ii) Evaluation. Although NASH-CRN was a better system for fibrosis staging, Metavir staging was a better reference for blood test. Thus, the patient rate with predictive values ≥90% by tests was 97.3% with Metavir reference vs. 66.5% with NASH-CRN reference (P<10(-3)). FibroMeter showed a significantly higher AUROC than the NAFLD fibrosis score for significant fibrosis, but not for severe fibrosis or cirrhosis, with both staging systems. Relationships between fibrosis lesions were well reflected by blood tests, e.g., the correlation between histological area and FD of fibrosis (r(s) =0.971, P<10(-3)) was well reflected by the relationship between respective blood tests (r(s) =0.852, P<10(-3)).
Conclusions: Different characteristics of fibrosis in NAFLD can be diagnosed and quantified by blood tests with excellent accuracy.
© 2010 John Wiley & Sons A/S.