Improved diagnostic accuracy of blood tests for severe fibrosis and cirrhosis in chronic hepatitis C

Objective Blood tests are usually designed to identify significant fibrosis. We evaluated their diagnostic accuracy, and how to increase it, for the clinically important targets of severe fibrosis and cirrhosis. Methods The accuracy for severe fibrosis or cirrhosis of four blood tests was evaluated based on Metavir staging in 1056 patients with chronic hepatitis C recruited in five independent hospitals. Results Using original scores, an original diagnostic target (significant fibrosis) and best diagnostic cutoff, the correct classification rates in severe fibrosis and cirrhosis stages were, respectively: FibroMeter: 90.1, 100%, Fibrotest: 78.2, 95.1%, Hepascore: 73.8, 94.9%, aspartate aminotransferase to platelet ratio index (APRI): 71.4, 88.0% (P<0.003, P=0.004, respectively, between tests). The corresponding area under the receiver operating characteristics were FibroMeter: 0.885, 0.907, Fibrotest: 0.837, 0.882, Hepascore: 0.834, 0.896, APRI: 0.822, 0.841 (P<0.003, respectively). Observed 100% negative predictive values for severe fibrosis and cirrhosis were, respectively, FibroMeter: 15.4, 47.5%, Fibrotest: 3.6, 31.9%, Hepascore: 0.3, 24.6%, APRI: 1.4, 5.3% of patients (P<0.003, respectively, between tests). By calculating a specific test for cirrhosis, including the FibroMeter markers, the correct classification (93.0%) was significantly higher for the cirrhosis diagnosis compared with the original FibroMeter (90.9%, P=0.005). This specific test provided a 100% positive predictive value for cirrhosis diagnosis versus 88% for original FibroMeter. Conclusion Using the most accurate original test, cirrhosis can be excluded in 475% of patients and is correctly diagnosed, as significant fibrosis, in 100% of patients. A specific test for cirrhosis provides a significant gain in diagnostic accuracy to 93% and in positive predictive value to 100% compared with the original test. Eur J Gastroenterol Hepatol 21:28-38 (C) 2009 Wolters Kluwer Health I Lippincott Williams & Wilkins.