Background/aims/methods: The diagnosis of cirrhosis is currently based on percutaneous liver biopsy, although this procedure may give rise to false negative results. This prospective study blindly investigates the accuracy of an ultrasonographic score, derived from liver, spleen and portal vein features, in predicting the final diagnosis in 212 patients with compensated chronic liver disease undergoing percutaneous liver biopsy.
Results: Taking biopsy as the standard, the ultrasonographic score differed significantly between chronic hepatitis (39+/-33) and cirrhosis (100+/-35) (p<0.0001). Discriminant analysis with stepwise forward selection of the variables identified liver surface nodularity and portal flow velocity as independently associated with the diagnosis of cirrhosis (p<0.005), and a score based on these two variables correctly identified cirrhosis in 82.2% of cases. One or both of these abnormalities were also found in 27/32 patients who were diagnosed as having cirrhosis at ultrasound, but were not cirrhotic histologically. Eight of these 32 cases developed signs of decompensated liver disease and/or portal hypertension in the subsequent 6-month follow-up, thus supporting the diagnosis of cirrhosis.
Conclusions: Our data suggest that ultrasound is accurate in predicting the final diagnosis in patients with compensated chronic liver disease and may identify cirrhosis even in the absence of a typical histopathological pattern. However, neither percutaneous liver biopsy nor ultrasonography can be assumed to be the definitive criterion for the diagnosis of compensated cirrhosis.