Computed tomography features of nonalcoholic steatohepatitis with histopathologic correlation

Objective: This study was conducted to describe the computed tomography (CT) features of nonalcoholic steatohepatitis (NASH) and to evaluate if the CT features could be used to diagnose and stage NASH. Methods: From 1994 until 2004, pathology records revealed 68 patients with NASH. Of these, 12 patients underwent CT scans before (n = 6), on the same day as (n = 3), or after (n = 3) a liver biopsy. Using the same database, 9 patients with steatosis alone evaluated with a CT scan before (n = 2), on the same day as (n = 3), or after (n = 4) the liver biopsy were selected as a control group. Two radiologists measured liver attenuation (compared with spleen) and assessed the pattern of steatosis, craniocaudal liver span, caudate-to-right lobe ratio, preportal space distance, and presence of porta hepatis lymph nodes and ascites. Biopsy specimens were assessed by a pathologist, and the degree of necroinflammatory activity, steatosis, and fibrosis was determined. Histopathologic and CT findings were compared between patients with NASH and patients with steatosis alone using the Mann-Whitney U test and Fisher exact test. Results: In patients with NASH, the mean liver-to-spleen attenuation ratio was 0.66 (range: 0.1-1.1). Steatosis was diffuse (n = 9), geographic or nonlobar (n = 2), or diffuse with an area of focal sparing (n = 1). The liver craniocaudal span varied from 17.5 to 25.5 cm (mean = 21.4 cm), and hepatomegaly was present in 11 (91.7%) patients. The caudate-to-right-lobe ratio (mean = 0.43) and preportal space (mean = 4.5 mm) were normal in all cases. Porta hepatis lymph nodes were present in 7 (58.3%) patients; their mean dimensions were 16 mm x 11 min. Ascites was absent in all patients. On histopathology, the degree of necroinflammatory activity was mild (n = 9), moderate (n = 1), or severe (n = 2). The degree of steatosis was 33% to 66% (n = 5) or >67% (n 7). All but 3 patients had fibrosis; 6 had focal nonbridging fibrosis, I had multifocal nonbridging fibrosis, and 2 had bridging fibrosis. There was a significant correlation between the degree of steatosis on pathologic examination and the liver-to-spleen attenuation ratio on CT (P = 0.048). The severity of inflammation and stage of fibrosis on pathologic examination did not correlate with the CT features. Among patients with steatosis alone, the mean liver-to-spleen attenuation ratio was 0.80 (range: 0.3-1.2); the craniocaudal liver span varied from 12 to 20 cm (mean = 16 cm); hepatomegaly was present in 2 (22.2%) patients; the caudate-to-right lobe ratio was normal in all patients, with a mean of 0.36 (range: 0.22-0.47); the preportal space distance was enlarged in 2 cases (mean = 7.5 mm, range: 1-16 mm); porta hepatis lymph nodes were present in 7 (77.8%) patients, and their mean dimensions were 11 mm x 8 mm (large axis range: 6-19 mm, short axis range: 4-14 mm); and no patient had ascites. There was a significant difference in the craniocaudal liver span between patients with NASH (mean = 21 cm) and patients with steatosis (mean = 16 cm) (P < 0.05). The caudate-to-right-lobe ratio was also significantly different between patients with NASH (mean = 0.43) and patients with steatosis (mean 0.36) (P < 0.05). There were no significant differences in liver-to-spleen attenuation ratios, measurements of preportal space, or the presence of porta hepatic lymph nodes. Conclusion: The CT features of NASH include steatosis, hepatomegaly, and porta hepatis lymph nodes, and the liver-to-spleen attenuation ratio correlated with the degree of steatosis on histopathology. Patients with NASH had a greater liver span and increased caudate-to-right-lobe-ratio compared with patients with steatosis alone.