Purpose: To assess the accuracy of noninvasive 3.0 T (1)H-magnetic resonance spectroscopy ((1)H-MRS) in an experimental steatosis model for the discrimination of clinically relevant macrovesicular steatosis degrees and to evaluate three different (1)H-MR spectrum-based fat quantification methods.
Materials and methods: Steatosis was induced in rats by a methionine/choline-deficient diet for 0-5 weeks. (1)H-MRS measurements of hepatic fat content were compared with histopathological and biochemical steatosis degree. In (1)H-MR spectra, areas under the curve (AUC) of fat (1.3 ppm), water (4.7 ppm), total fat (0.5-5.3 ppm), and total spectrum peaks (0.5-5.3 ppm) were determined and hepatic fat content calculated as follows: [AUC(total fat peaks)/AUC(total peaks)], [AUC(fat)/AUC(fat) + (AUC(water)/0.7)], and [AUC(fat)/AUC(water)].
Results: A significant correlation was found between (1)H-MRS and macrovesicular steatosis (r = 0.932, P < 0.0001) and between (1)H-MRS and total fatty acids (r = 0.935, P < 0.0001). (1)H-MRS accurately distinguished mild from moderate and moderate from severe steatosis. Calculations using [AUC(fat)/AUC(water)] ratio in severe steatotic livers resulted in higher hepatic fat percentages as compared to the other methods due to a decrease in hepatic water content.
Conclusion: (1)H-MRS quantification of hepatic fat content showed high correlations with histological and biochemical steatosis determination in an experimental steatosis rat model and accurately discriminated between clinically relevant steatosis degrees. These results encourage further application of (1)H-MRS in patients for accurate steatosis assessment.