RESUMO
OBJECTIVES: to evaluate the performance of magnetic resonance elastography (MRE) to stage liver fibrosis in patients with histologically confirmed nonalcoholic fatty liver disease (NAFLD) and to assess the impact of potential confounding factors in MRE diagnostic accuracy. The secondary objective was to compare MRE with other non-invasive methods for staging fibrosis such as transient elastography (TE) and non-invasive scores (APRI and FIB-4). METHODS: sixty-five histologically confirmed NAFLD patients were prospectively enrolled at the Hospital Universitario Virgen del Rocío (Seville, Spain). Liver stiffness was measured by MRE, TE and non-invasive scores (APRI and FIB-4). Fibrosis was assessed by liver biopsy using the steatosis, activity and fibrosis (SAF) score. Patients were classified into three groups according to the consistency between MRE and histopathological findings: underestimation, concordance and overestimation groups. Areas under the ROC curve (AUROC) and diagnostic performance were evaluated. RESULTS: the area under the ROC curve (AUROC) of MRE in advanced fibrosis (≥ F3) was 0.90 (0.82-0.97), while TE AUROC was 0.82 (0.72-0.93) (p = 0.22) and lower for the non-invasive test (FIB-4 0.67 and APRI 0.62). Inflammatory activity, steatosis grade and higher levels of liver biochemistry appeared to overestimate MRE results in the univariate analysis, but only gamma-glutamyl transferase (GGT) was statistically significant in the multivariate analysis (p < 0.01). Age, sex, body mass index (BMI), weight, diabetes mellitus (DM), high blood pressure (HBP), platelets or lipidic profile did not affect MRE accuracy. CONCLUSIONS: MRE is an effective and non-invasive method for detecting and staging liver fibrosis in NAFLD patients. MRE is more accurate than TE and allows the study of liver anatomy. Histological inflammation and surrogate biomarkers of inflammation can overestimate liver stiffness, but only GGT was statistically significant in the multivariate analysis. Important features of NAFLD patients such as obesity, DM, or lipidic profile did not affect MRE accuracy.