Severe Liver Fibrosis or Cirrhosis: Accuracy of US for Detection—Analysis of 300 Cases

PURPOSE: To determine the accuracy of various ultrasonographic (US) signs for assessment of the degree of liver fibrosis, with histologic results as reference standard. MATERIALS AND METHODS: Three hundred consecutive asymptomatic patients with at least 6 months of increased levels of aspartate aminotransferase and/or alanine aminotransferase underwent liver US and biopsy. The estimated pretest probability of severe fibrosis or cirrhosis was 35%. Three US parameters were investigated: liver surface nodularity, caudate lobe hypertrophy, and pattern of hepatic venous blood flow. US results were compared with histologic results obtained after liver biopsy, which constituted the reference standard for diagnosis of severe fibrosis or cirrhosis. The degree of fibrosis was graded according to METAVIR criteria, with stages 3 and 4 considered together. Data were analyzed with kappa and chi2 statistics. Sensitivity, specificity, positive and negative likelihood ratios, and posttest probability were calculated for each US sign. RESULTS: In 107 (36%) patients with severe fibrosis (n = 34) or cirrhosis (n = 73), liver surface nodularity had the highest diagnostic accuracy, with specificity of 95% and positive and negative likelihood ratios 11.6 and 0.51, respectively. When liver surface nodularity was considered alone, posttest probability of severe fibrosis or cirrhosis increased from 35% to 86%. When caudate lobe hypertrophy and hepatic venous blood flow were also taken into account, posttest probability increased by only 2% (ie, to 88%). CONCLUSION: US determination of liver surface nodularity is an accurate method for identifying the subset of asymptomatic patients with severe liver fibrosis or cirrhosis, which indicates a worse prognosis.