Introduction Approximately 12 500 liver biopsies are performed annually in England and Wales. Liver biopsies are associated with sampling errors, observer variability and complications. In contrast, use of ultrasound and MR elastography may be limited by inter-operator variability, cost and access. Fat, iron accumulation and fibrosis are critical contributors to the pathogenesis of major chronic liver disease.

Aim To correlate quantitative multimodal MRI parameters with histological grading of fat, iron accumulation and fibrosis.

Methods MRI relaxation time data (T₁, T₂ and T₂*) were acquired using a novel Echo Planar Imaging technique with a respiratory-triggered (r.t.) rapid acquisition method, repeated multiple times with different contrast to generate accurate voxel-by-voxel (3×3×8 mm³ voxel) relaxation time maps of the liver tissue. 1H MR spectra were acquired (r.t.) using a multiple echo PRESS acquisition which allowed for individual T₂ correction to the spectrum for accurate quantification of the fat fraction in a 30×30×30mm³ voxel.

Results Sixty two subjects (range 23–72 years) were included in the study based on the adequacy of liver biopsy. Liver biopsies were (mean ±SD) 28.7±7.3 mm long and 0.96±0.2 mm wide. Histology revealed non-alcoholic fatty liver in 50%, chronic viral hepatitis (HCV and HBV) in 24%, alcoholic liver disease in 11%, normal in 11% and haemochromatosis in 3%. There was significant correlation between MR measures of fat fraction and percentage of fat on histology (r =0.81; p<0.001), T₂* with hepatic iron content (r = − 0.65, p<0.001) and T₁ with percentage of fibrosis on histology (r=0.48, p<0.001). Performance characteristic of T₁ MR parameter in the detection of different histological stages of fibrosis is listed in table 1.

Conclusion In chronic liver diseases, MR measures of fat fraction, hepatic iron content and fibrosis correlate well with histological grading. Quantitative MR techniques allow accurate distinction of different stages of fibrosis on liver biopsy. If these are replicated in independent cohorts, these techniques can be translated into clinical applications.