Quantification of Hepatic Steatosis: A Comparison of Computed Tomography and Magnetic Resonance Indices in Candidates for Living Liver Donation

Abstract

To evaluate, in a group of candidates for liver donation, the role of unenhanced computed tomography (CT) and magnetic resonance (MR) as noninvasive means to measure hepatic steatosis (HS). Sixty-one consecutive candidates underwent CT and MR evaluation for liver donation within 3 weeks of liver biopsy. On CT, three methods of HS quantification were evaluated: the measurement of hepatic attenuation (CT L), the ratio of hepatic attenuation to splenic attenuation (CT L/S), and the difference between the hepatic and splenic attenuation (CT L-S). On MR, HS was reported in terms of fat signal fraction (FSF) using in-phase/opposed-phase and fat/non-fat- saturated images, with and without normalization with the spleen (T1W IP/OP FSF, T1W IP/OP FSF spleen and T2W ± FS FSF, TW2 ± FS FSF spleen). The accuracy of each imaging index in the diagnosis of HS, according to various thresholds, was assessed using receiver operating characteristic analysis. On biopsy, 35 donors showed no significant HS (<5%); the remaining 26 showed HS ranging from 5% to 40%. With all CT and MR indices, there was a trend toward increasing diagnostic accuracy as the threshold levels of HS increased. When comparing all the indices, TW2 ± FS FSF(spl) showed higher accuracy at threshold levels of 5% and 10% of steatosis but without reaching statistical significance. In candidates for living donation, MR and CT indices are similar in estimating liver-fat content; however, MR with T2W ± FS FSF(spl) sequences shows higher accuracy when low threshold levels of steatosis (≤5% and ≤10% HS) are selected.