Diagnostic performance and accuracy of 3-D spoiled gradient-dual-echo MRI with water- and fat-signal separation in liver-fat quantification: comparison to liver biopsy

Abstract

Objectives: To prospectively evaluate a 3-dimensional spoiled gradient-dual-echo (3D SPGR-DE) magnetic resonance imaging (MRI) sequence for the qualitative and quantitative analysis of liver fat content (LFC) in patients with the suspicion of fatty liver disease using histopathology as the standard of reference.

Materials and methods: Thirty-four adult patients (15 women; mean age, 67 +/- 13 years) underwent hepatic 1.5-Tesla 3D SPGR-DE MRI including in-/out-of-phase (IP/OP) and fat-only sequences prior to hepatic surgery with biopsy. Histopathological analyses of total and macrovesicular LFC could be made from biopsies of 39 segments in 23 patients. Two radiologists independently classified steatosis visually using a 4-point scale for IP/OP and fat-only images. Additionally, fat signal fractions (FSF) were calculated from signal intensities on IP/OP (FSF(ip/op)) and fat-only images (FSF(fat-only)). Pearson correlation analysis and Student t test were used to study the relationship between the FSF and LFC as determined by histopathology. The accuracy of MRI in detecting pathologically elevated LFC was assessed by receiver operating characteristic analysis.

Results: Histopathology revealed steatosis in 29/39 (74%) segments in 15/23 patients (65%), with a total LFC ranging from <5% to 90%. Inter-reader agreement for visual steatosis grading was moderate (k = 0.53) for IP/OP images and good (k = 0.68) for fat-only images. FSF calculated from IP/OP and fat-only images significantly correlated with LFC from histopathology (both, P < 0.0001). Mean FSF(fat-only) and FSF(ip/op) values significantly differed from total LFC (both, P < 0.0001), whereas mean FSF(fat-only) showed no significant differences to macrovesicular LFC (P = 0.46). Both FSF(fat-only) and FSF(ip/op) performed accurately in discriminating between normal LFC and elevated LFC according to histopathology with good diagnostic accuracy (AUC: 0.85; 95% CI: 0.73-0.89 vs. AUC 0.90, 95% CI: 0.7-1.00).

Conclusions: FSF(fat-only) and FSF(ip/op) derived from 3D SPGR-DE MRI mostly reflect the macrovesicular as opposed to the total LFC from histopathology, whereas both discriminate healthy and fatty liver. Analysis of fat-only images improves interreader-agreement for visual liver steatosis grading.