Background/aims: Currently the diagnosis and severity of hepatic steatosis can be established accurately only by liver biopsy. Previous small studies found that steatosis measured by magnetic resonance spectroscopy (MRS) and imaging (MRI) correlated with histological assessment of liver triglyceride content. However, the accuracy of MRS/MRI for grading the severity of steatosis has not been addressed. The aims of this study were (1) to determine whether MRS and MRI can discriminate grades of steatosis in a large cohort of consecutive patients with a wide spectrum of liver disease aetiology and severity (2) to evaluate the effect of hepatic fibrosis, inflammation and iron on quantitation of intrahepatocellular lipid (IHCL) by these techniques.
Methods: Ninety-four sequential patients who underwent percutaneous liver biopsy or liver resection had MRS and MRI (Dixon in phase/out of phase (Dixon IP/OP) and with/without fat saturation (+/-FS) images) to determine IHCL. Histology was used as the reference standard.
Results: Close relationships were observed between the percentage of steatosis estimated by histology and MRS/MRI (r(s)=0.88 p<0.001 for all techniques). However, separate equations were required for the percentage of steatosis to avoid underestimation by imaging for patients with moderate or advanced fibrosis. All techniques had good diagnostic accuracy for mild steatosis (AUROC > or =0.87) as well as moderate/severe steatosis (AUROC > or =0.89). Hepatic inflammation and mild iron deposition (Perls' grade 1 and 2) did not interfere with estimation of steatosis by imaging.
Conclusions: MRS and MRI had good accuracy for grading the severity of steatosis in subjects with liver disease, provided that stage of fibrosis was considered.