Ultrasound diagnosis of fatty liver in patients with chronic liver disease: a retrospective observational study.

OBJECTIVES: Hepatic ultrasound (US) is readily available and physicians usually trust the results of an US report suggesting fatty liver, but there are conflicting reports on its accuracy, especially in patients with chronic liver disease (CLD). Therefore, we retrospectively examined liver biopsies in patients with CLD and compared the histologic results to the hepatic US findings. METHODS: Liver biopsies were graded for fat (grades 0 to 3), inflammation (grades 0 to 4), and fibrosis (stages 0 to 4) in 131 patients with CLD (89% had chronic hepatitis C). Hepatic US interpretations were grouped into 3 categories-"normal," "fatty liver," and "nonspecific." A secondary analysis was performed using 3 sonographic categories based on the echogenicity: normal, "increased echogenicity," and "heterogenous." The US results were then compared with the liver biopsy results. RESULTS: A normal US report was associated with many false negatives, as 25% of these patients had fat (grades 1 to 3) on biopsy; furthermore, 46% had "significant fibrosis" (stages 2 to 4) or "significant inflammation" (grades 2 to 4). A "fatty liver" interpretation correctly identified fat on biopsy in 36.4% and "significant fat" (grades 2 to 3) in 11.4%, but 66% had significant fibrosis or significant inflammation. An US with increased echogenicity correctly identified fat in 43.5% and significant fat in 19.4%, but 69.4% had significant fibrosis or significant inflammation. The sensitivity of an US ranged from 11.4% to 88.2% and the specificity ranged from 40.4% to 86.2%, depending on the degree of steatosis on biopsy and the sonographic interpretation being considered. CONCLUSIONS: US is inaccurate for diagnosing hepatic steatosis in patients with CLD. Echogenic abnormalities are more likely to be the result of fibrosis or inflammation in this setting.