Diagnostic accuracy, reproducibility and robustness of fibrosis blood tests in chronic hepatitis C: a meta-analysis with individual data.
Résumé
OBJECTIVES: To evaluate the diagnostic accuracy of liver fibrosis tests and its influencing factors in a meta-analysis with individual data. DESIGN AND METHODS: Four independent centers provided four blood tests and Metavir staging from 825 patients with chronic hepatitis C. RESULTS: FibroMeter AUROC (0.840) for significant fibrosis was superior to those of Fibrotest (0.803, p=0.049), APRI (0.789, p=0.001) and Hepascore (0.781, p<0.001). The misclassification rate was lower for FibroMeter (23%) than for Fibrotest and Hepascore (both 28%, p<0.001). The variation in the diagnostic cut-offs of tests among centers, reflecting the overall reproducibility, was: FibroMeter: 4.2%, APRI: 24.0%, Fibrotest: 24.2%, Hepascore: 35.0%. Accordingly, the proportion of patients diagnosed with significant fibrosis changed: FibroMeter: 0.8%, Hepascore: 2.4% (p=0.02 vs FibroMeter), Fibrotest: 5.8% (p<10(-3)), APRI: 18.2% (p<10(-3)). CONCLUSIONS: This study on clinical applicability shows significant differences in diagnostic accuracy, inter-center reproducibility, and robustness of biomarkers to changes in population characteristics between blood tests.