Unenhanced CT for Assessment of Macrovesicular Hepatic Steatosis in Living Liver Donors: Comparison of Visual Grading with Liver Attenuation Index

Abstract

To retrospectively compare the accuracy of visual grading and the liver attenuation index in the computed tomographic (CT) diagnosis of 30% or higher macrovesicular steatosis in living hepatic donors, by using histologic analysis as the reference standard. Institutional review board approval was obtained with waiver of informed consent. Of 703 consecutive hepatic donor candidates, 24 patients (22 men and two women; mean age +/- standard deviation, 36.3 years +/- 9.7) who had 30% or higher macrovesicular steatosis at histologic analysis and same-day CT with subsequent needle biopsy in the right hepatic lobe (at least two samples per patient) were evaluated. An age- and sex-matched control group of 24 subjects included those who had less than 30% macrovesicular steatosis but otherwise met the same criteria as the patient group. A diagnostically difficult setting was made by selecting those with the highest degree of macrovesicular steatosis when there were multiple control subjects matched for a particular subject in the patient group. Two independent radiologists assessed steatosis of the right hepatic lobe by using two methods: a five-point visual grading system that used attenuation comparison between the liver and hepatic vessels and the liver attenuation index (CT(L-S)), defined as hepatic attenuation minus splenic attenuation and calculated with region of interest measurements of hepatic attenuation. Interobserver agreement was assessed. Accuracy in the diagnosis of 30% or higher macrovesicular steatosis was compared by using a multireader, multicase receiver operating characteristic (ROC) analysis. For visual grading, kappa = 0.905 (95% confidence interval [CI]: 0.834, 0.976). Intraclass correlation coefficient for CT(L-S) was 0.962 (95% CI: 0.893, 0.983). The area under the ROC curve of visual grading and CT(L-S) were 0.927 (95% CI: 0.822, 1) and 0.929 (95% CI: 0.874, 0.983), respectively, indicating no statistically significant difference (P = .975). Both visual grading and CT(L-S) are highly reliable and similarly accurate in the diagnosis of 30% or higher macrovesicular steatosis in living hepatic donor candidates.