The utility of radiological imaging in nonalcoholic fatty liver disease

This prospective study evaluates the role of radiological modalities in establishing the diagnosis of nonalcoholic steatohepatitis (NASH).

Consecutive patients with biopsy-proven nonalcoholic fatty liver disease (NAFLD) were enrolled (2000-2001). Patients with other liver diseases and significant alcohol consumption (>20 g/day) were excluded. Clinicodemographic data were gathered at the time of liver biopsy. Each biopsy specimen was assessed by a hepatopathologist. Each patient underwent a limited abdominal ultrasonography (US), computerized tomography (CT), and magnetic resonance imaging (MRI). Films were interpreted by a radiologist who used a predetermined radiological protocol. Each radiological study was reread by the same radiologist and a second radiologist.

Patients with NASH had greater aspartate aminotransferase levels (P = 0.03), greater ferritin levels (P = 0.05), more hepatocyte ballooning (P < 0.0001), and more fibrosis (P = 0.002). None of the radiological features distinguished between NASH and other types of NAFLD. No radiological modality detected the presence of hepatocyte ballooning, Mallory's hyaline, or fibrosis, which are important features in the diagnosis of NASH. The presence of >33% fat on liver biopsy was optimal for detecting steatosis on radiological imaging.

Differences between NASH and nonprogressive NAFLD were not apparent with any radiological modality. Of the pathologic features important for establishing the diagnosis of NASH, only the severity of steatosis was reflected in these radiological modalities. Good intraobserver agreement was evident for each modality (US, CT, and MRI) that was superior to interobserver agreement.