BACKGROUND/AIMS: We evaluated the test performance profile (TPP) of blood tests of liver fibrosis. METHODS: Three hundred and fifty-six patients with C chronic hepatitis were included in two centers. Metavir staging of liver specimens by two independent pathologists and the following tests were evaluated: Fibrotest (FT), APRI, FibroMeter (FM), and Hepascore (HS). RESULTS: Metavir stages were: F0: 4%, F1: 55%, F2: 26%, F3: 11%, and F4: 4%. The AUROCs were not significantly different, respectively, FT, FM, APRI, HS: >or=F2: 0.79, 0.78, 0.76, >or=0.76; F3: 0.81, 0.85, 0.81, 0.81; and F4: 0.86, 0.94, 0.92, 0.89. The TPP relies on the paired comparison of blood-test misclassification based on liver specimen, e.g. FT vs FM, respectively: F0+1: 18 vs 28% (p=0.0003), >or=F2: 43 vs 31% (p=0.004). There was no center effect. CONCLUSIONS: In those populations, the four blood tests had a similar performance for significant fibrosis (F>or=2), lying in the lower range of published results which is attributable to a low >or=F2 prevalence, and for >or=F3 and F4. However, FM and FT had performance profiles significantly different as a function of fibrosis stages or diagnostic target (fibrosis cut-off). This has to be considered during the interpretation process. Moreover, the performance should be reported with different diagnostic targets.
BACKGROUND/AIMS: We evaluated the test performance profile (TPP) of blood tests of liver fibrosis. METHODS: Three hundred and fifty-six patients with C chronic hepatitis were included in two centers. Metavir staging of liver specimens by two independent pathologists and the following tests were evaluated: Fibrotest (FT), APRI, FibroMeter (FM), and Hepascore (HS). RESULTS: Metavir stages were: F0: 4%, F1: 55%, F2: 26%, F3: 11%, and F4: 4%. The AUROCs were not significantly different, respectively, FT, FM, APRI, HS: >or=F2: 0.79, 0.78, 0.76, >or=0.76; F3: 0.81, 0.85, 0.81, 0.81; and F4: 0.86, 0.94, 0.92, 0.89. The TPP relies on the paired comparison of blood-test misclassification based on liver specimen, e.g. FT vs FM, respectively: F0+1: 18 vs 28% (p=0.0003), >or=F2: 43 vs 31% (p=0.004). There was no center effect. CONCLUSIONS: In those populations, the four blood tests had a similar performance for significant fibrosis (F>or=2), lying in the lower range of published results which is attributable to a low >or=F2 prevalence, and for >or=F3 and F4. However, FM and FT had performance profiles significantly different as a function of fibrosis stages or diagnostic target (fibrosis cut-off). This has to be considered during the interpretation process. Moreover, the performance should be reported with different diagnostic targets.
Authors: Katherine E McGoogan; P Brian Smith; Steve S Choi; Wallace Berman; Ravi Jhaveri Journal: J Pediatr Gastroenterol Nutr Date: 2010-03 Impact factor: 2.839
Authors: Papa Saliou Mbaye; Anna Sarr; Jean-Marie Sire; Marie-Louise Evra; Adama Ba; Jean Daveiga; Aboubakry Diallo; Fatou Fall; Loic Chartier; François Simon; Muriel Vray Journal: PLoS One Date: 2011-07-25 Impact factor: 3.240