Measuring liver triglyceride content in mice: non-invasive magnetic resonance methods as an alternative to histopathology

Quantitative assessment of liver fat is highly relevant to preclinical liver research and should ideally be performed non-invasively. This study aimed to compare three non-invasive Magnetic Resonance (MR) and two histopathological methods against the reference standard of biochemically determined liver triglyceride content (LTC). A total of 50 mice [21 C57Bl/6OlaHsd mice (C57Bl/6), nine low-density lipoprotein (LDL) receptor knock-out -/- (LDL -/-) mice and 20 C57BL/6 mice] received either a high-fat, high-fat-high-cholesterol or control diet, respectively. Mice were examined 4, 8 or 12 weeks into the diet using MR [H-1-MR Spectroscopy, Proton Density Fat Fraction (PDFF), mDixon] and histopathological methods (visual scoring or digital image analysis (DIA) of Oil-Red-O (ORO) stained liver sections). Correlations [Pearson's coefficient (r)] were studied with respect to LTC. Microvesicular steatosis was seen in 42/50 mice. H-1-MRS values showed normal to moderately elevated liver fat content. Visual scoring and DIA of ORO-sections correlated moderately with LTC at r = 0.59 and r = 0.49 (P < 0.001), respectively. H-1-MRS, PDFF and mDixon correlated significantly better, at r = 0.74, r = 0.75 and r = 0.82, respectively. Non-invasively determined MR measures of normal to moderately elevated liver fat in mice had a higher correlation with LTC than invasive histopathological measures. Where available, MR is the preferred method for fat quantification.