Quantitative determination of liver triglyceride levels with 3T ¹H-MR spectroscopy in mice with moderately elevated liver fat content.

To diagnose hepatic steatosis with noninvasive magnetic resonance (MR)-based measurements, threshold values of liver fat percentages are used. However, these differ between studies. Consequently, the choice of threshold values influences diagnostic accuracy, especially in subjects with borderline hepatic steatosis. In this study, we compared (1)H-MR spectroscopy (MRS) and biochemically determined liver fat content in mice with moderately elevated fat content and studied the diagnostic accuracy of (1)H-MRS using two literature-based threshold values.Fifty mice were divided into three groups: 21 C57Bl/6OlaHSD (B6) mice on a high-fat diet, 20 B6 mice on a control diet, and 9 LDLr-/- mice on a high-fat high-cholesterol diet. (1)H-MRS was performed using multi-echo STEAM at 3T to derive a fat mass fraction ((1)H-MRS fat content). Biochemical fat content was determined from liver homogenates. Correlation and agreement were assessed with the Pearson correlation coefficient and the Bland-Altman analysis and diagnostic accuracy by calculating sensitivity, specificity, and positive and negative predictive values.All mice were pooled to form a single cohort. Mean (±standard deviation) biochemical fat content was 32.2 (±13.9) mg/g. Mean (1)H-MRS fat content did not differ at 30.2 (±12.0) mg/g (P = .13). Correlation r was 0.74 (P < .0001). Bland-Altman analysis indicated that (1)H-MRS fat content underestimated biochemical fat content by 2.1 mg/g. The diagnostic accuracy of (1)H-MRS depended to a great extent on the chosen reference threshold value.(1)H-MRS measurement of moderately elevated liver fat content in mice correlated substantially with biochemical fat content measurement. Contrary to earlier studies, diagnostic accuracy of (1)H-MRS fat content in borderline liver fat content appears limited.