Background: Liver iron and fat are often co-deposited, synergistically aggravating the progression of chronic liver disease. Accurate determination of liver iron and fat content is helpful for patient management.To assess the accuracy of hepatic iron/fat decomposition using dual-energy computed tomography (DECT) for simultaneously quantifying hepatic iron and fat when both are present.Methods: Sixty-eight New Zealand rabbits on a high-fat/cholesterol diet plus iron injections were used to establish a model of coexisting hepatic iron/fat. Abdominal imaging was performed using dual-source DECT.The iron and fat fractions (Iron-CT and Fat-CT , respectively) were calculated using a 3-material decomposition algorithm. The spectroscopic liver iron concentration (LIC) grading (normal, mild, moderate, severe, and massive iron overload) and the histopathological fat fraction (Fat-ref) grading (normal, mild, moderate, severe steatosis) were used as references. Correlations between the DECT parameters and the references were analyzed. Hepatic iron/fat quantification equations were established and validated. Analysis of covariance was used to assess the influence of fat on iron measurements and vice versa.Results: Iron-CT highly correlated with LIC (r=0.94, P<0.001), and Fat-CT highly correlated with Fatref (r=0.88, P<0.001). Both the Iron-CT -and Fat-CT -derived LIC and fat fraction showed good agreement with spectroscopy/histology. The linear relationship between Iron-CT and spectroscopic LIC was not affected by the grade of hepatic fat (F=1.93, P=0.16). The linear relationship between Fat-CT and Fatref was unaffected by hepatic iron grades from normal to severe (F=0.18, P=0.91). However, with massive iron overload [>15.0 mg Fe/g (270 μmol/g)] the regression began to deviate, causing fat underestimation (F=5.50, P=0.04).Conclusions: Our DECT-based iron/fat decomposition algorithm accurately measured hepatic iron and fat when both were present in a rabbit model. Hepatic fat may be underestimated when there is massive iron overload.

No abstract