Accuracy of Routine Clinical Ultrasound for Staging of Liver Fibrosis

Objective: To determine the diagnostic accuracy of routine clinical ultrasound in the staging of liver fibrosis in chronic viral hepatitis. Materials and Methods: A retrospective evaluation of the ultrasound images of 156 patients with chronic viral hepatitis who underwent liver biopsy was performed. Two radiologists in consensus, blind to the biopsy results and clinical details, evaluated the ultrasound images for liver fibrosis. The readers specifically assessed three features - surface nodularity, liver edge, and parenchymal echotexture - with scores of 0 to 3 (0 = normal, 1 = mild, 2 = moderate, 3 = severe). Accuracies of each sonographic feature for the detection of mild fibrosis and above (>= F1), significant fibrosis (>= F2), severe fibrosis (>= F3), and cirrhosis (F4) were determined with histopathology as the reference standard. Results: Fibrosis was present in 99 patients (F1=34, F2=20, F3=22, and F4=23) and absent in 57 patients. The sensitivities for the detection of significant fibrosis with surface nodularity, liver edge, and parenchymal echotexture were 57%, 15%, and 41%, respectively. The accuracies for the detection of >= F1, >= F2, >= F3, and F4 stages were 50.5%, 59%, 59%, and 65% for liver surface, 51%, 53%, 54%, and 55% for liver edge, and 58%, 59%, 63%, and 63% for parenchyma echotexture, respectively. The combined scores from all three features had accuracies of 56%, 59%, 62%, and 66% for the detection of >= F1, >= F2, >= F3, and F4, respectively. Conclusion: Routine clinical ultrasound is a not a sensitive predictor of early fibrosis in chronic viral hepatitis. Surface nodularity is the most sensitive sonographic feature for the detection of significant fibrosis and routine clinical ultrasound is the most useful for the detection of cirrhosis.