Diagnosis of cirrhosis by spiral computed tomography: a case-control study with feature analysis and assessment of interobserver agreement.

PURPOSE: To determine the accuracy and interobserver agreement of spiral computed tomography (CT) in the diagnosis of cirrhosis.
MATERIALS AND METHODS: We retrospectively identified 126 patients who underwent spiral CT at our institution and who had a contemporaneous histopathologic confirmation of cirrhosis (n = 67) or clinical and biochemical evidence of a normal liver (n = 59). Two experienced readers independently recorded the overall likelihood of cirrhosis and the presence or absence of hepatic and extrahepatic findings of cirrhosis and portal hypertension on a 5-point scale from 1 (definitely absent) to 5 (definitely present/severe).
RESULTS: Receiver operating characteristic curve and kappa statistic analyses showed that the overall likelihood of cirrhosis was the most accurate and objective observation, with an area under the curve (AUC) of 0.97 for reader 1 and 0.90 for reader 2 and a kappa value of 0.70. Individual findings that were accurate and objective were diaphragmatic surface nodularity (AUC = 0.95 and 0.88 for readers 1 and 2, respectively, kappa = 0.75), global or segmental volume loss (AUC = 0.95 and 0.87 for readers 1 and 2, respectively, kappa = 0.70), and superior diaphragmatic adenopathy (AUC = 0.85 for both readers, kappa = 0.78). Of note, portal vein diameter was not significantly different between normal and cirrhotic patients as measured by either reader (P = 0.54 and 0.65).
CONCLUSION: Spiral CT demonstrates high accuracy and interobserver agreement in the diagnosis of cirrhosis, suggesting CT may be a supplementary diagnostic test in patients who have contraindications to biopsy or have equivocal biopsy findings.