In the present study, the APRI and the FI predicted the presence of F≥2 with acceptable reliability. Thus, the APRI predicted the presence of F≥2 with 91% certainty, and misclassified

15% of the patients with scores ≥1.5, who showed only portal fibrosis in the LB. The FI predicted the presence of F≥2 with 89% certainty, and 19% of the patients with scores ≥6.9 showed F1. These errors of classification are not relevant as all misclassified patients had fibrosis in the LB. Therefore, 22% of the patients would benefit from not undergoing an LB, if the indexes were used as an aid for making decisions regarding anti-HCV therapy. Screening patients with indeterminate LGK-974 in vivo results

for the APRI, i.e. those with APRI<1.5, with the FI can increase the proportion of correctly Caspase activity assay classified patients to 30%. Hence, the percentage of patients with fibrosis ≥F2 correctly identified in this study was 10% lower than in the validation studies in HIV/HCV-coinfected patients, where one-third or more of the patients were correctly classified using each index [9–16] and 40% by applying a sequential combination of the APRI and the FI [9]. In previous studies on the APRI and the FI in patients with HIV and HCV coinfection, the AUROC of the APRI to predict F≥2 has ranged from 0.66 to 0.85 [9–17]. Glutathione peroxidase The PPV of APRI>1.5 to diagnose F≥2 has lain between 82% and 97% [9,13,15], after excluding extreme values [14,16]. For the FI, the AUROC values to predict F≥2 were between 0.59 and 0.77 [9–13], and the PPV of FI >6.9 to detect F≥2 was >90% [9,13]. Thus, the diagnostic yield of the APRI and the FI was lower in the present study than in previous validation studies in HIV/HCV-coinfected patients. The better results obtained in previous studies can be explained by the design of those studies. All of them were carried out in tertiary care centres [9–17]. One of the validation studies was a subanalysis of a randomized clinical

trial [12]. It is probable that the study populations were highly selected in those studies. LB was used as a reference for the diagnosis of fibrosis in those reports, as in the present study. However, liver biopsies were reviewed at each participating centre and/or centrally by expert pathologists in the validation studies [9–17]. In contrast, we collected the information that was available on liver fibrosis classification at each centre, provided that liver fibrosis was staged following the METAVIR score. Thus, the quality of the reference for liver fibrosis was poorer in the present study compared with the validation studies. Liver fibrosis staging in biopsies can be inaccurate because of sampling variability. This is an issue that is difficult to control.