Comparing magnetic resonance liver fat fraction measurements with histology in fibrosis: the difference between proton density fat fraction and tissue mass fat fraction

OBJECTIVE: Magnetic resonance spectroscopy (MRS) provides a powerful method of measuring fat fraction. However, previous studies have shown that MRS results give lower values compared with visual estimates from biopsies in fibrotic livers. This study investigated these discrepancies and considered whether a tissue water content correction, as assessed by MRI relaxometry, could provide better agreement. MATERIALS AND METHODS: 110 patients were scanned in a 1.5 T Philips scanner and biopsies were obtained. Multiple echo MRS (30 × 30 × 30 mm volume) was used to determine Proton Density Fat Fraction (PDFF). Biopsies were assessed by visual assessment for fibrosis and steatosis grading. Digital image analysis (DIA) was also used to quantify fat fraction within tissue samples. T(1) relaxation times were then used to estimate tissue water content to correct PDFF for confounding factors. RESULTS: PDFF values across the four visually assessed steatosis grades were significantly less in the higher fibrosis group (F3–F4) compared to the lower fibrosis group (F0–F2). The slope of the linear regression of PDFF vs DIA fat fraction was ~ 1 in the low fibrosis group and 0.77 in the high fibrosis group. Correcting for water content based on T(1) increased the gradient but it did not reach unity. DISCUSSION: In fibrotic livers, PDFF underestimated fat fraction compared to DIA methods. Values were improved by applying a water content correction, but fat fractions were still underestimated. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s10334-022-01052-0.