Using a 3% Proton Density Fat Fraction as a Cut-Off Value Increases Sensitivity of Detection of Hepatic Steatosis, Based on Results From Histopathology Analysis

Patrik Nasr, Mikael F. Forsgren, Simone Ignatova, Nils Dahlström, Gunnar Cedersund, Olof Dahlqvist Leinhard, Bengt Norén, Mattias Ekstedt, Peter Lundberg, Stergios Kechagias

It is possible to estimate hepatic triglyceride content by calculating the proton density fat fraction (PDFF), using proton magnetic resonance spectroscopy ((1)H-MRS), instead of collecting and analyzing liver biopsies to detect steatosis. However, the current PDFF cut-off value (5%) used to define steatosis by magnetic resonance was derived from studies that did not use histopathology as the reference standard. We performed a prospective study to determine the accuracy of (1)H-MRS PDFF in measurement of steatosis using histopathology analysis as the standard. We collected clinical, serologic, (1)H-MRS PDFF, and liver biopsy data from 94 adult patients with increased levels of liver enzymes (6 months or more) referred to the Department of Gastroenterology and Hepatology at Linköping University Hospital in Sweden from 2007 through 2014. Steatosis was graded using the conventional histopathology method and fat content was quantified in biopsy samples using stereological point counts (SPCs). We correlated (1)H-MRS PDFF findings with SPCs (r = 0.92; P <.001). (1)H-MRS PDFF results correlated with histopathology results (ρ = 0.87; P <.001), and SPCs correlated with histopathology results (ρ = 0.88; P <.001). All 25 subjects with PDFF values of 5.0% or more had steatosis based on histopathology findings (100% specificity for PDFF). However, of 69 subjects with PDFF values below 5.0% (negative result), 22 were determined to have steatosis based on histopathology findings (53% sensitivity for PDFF). Reducing the PDFF cut-off value to 3.0% identified patients with steatosis with 100% specificity and 79% sensitivity; a PDFF cut-off value of 2.0% identified patients with steatosis with 94% specificity and 87% sensitivity. These findings might be used to improve non-invasive detection of steatosis.