DOI: /hiv British HIV Association HIV Medicine (2014), 15, ORIGINAL RESEARCH

Transcription

1 DOI: /hiv ORIGINAL RESEARCH Comparison of transient elastography (FibroScan), FibroTest, APRI and two algorithms combining these non-invasive tests for liver fibrosis staging in HIV/HCV coinfected patients: ANRS CO13 HEPAVIH and FIBROSTIC collaboration L Castera, 1 M Winnock, 2 E Pambrun, 2 V Paradis, 3 P Perez, 4 M-A Loko, 2 J Asselineau, 4 F Dabis, 2 F Degos 1 and D Salmon 5 1 Department of Hepatology, CRB3 INSERM U773, Beaujon Hospital, AP-HP, Clichy, France, 2 INSERM U897/ISPED Victor Segalen Bordeaux University, Bordeaux, France, 3 Department of Pathology, Beaujon Hospital, AP-HP, Clichy, France, 4 Clinical Epidemiology Unit and CIC-EC7, Victor Segalen Bordeaux University, Bordeaux, France and 5 Department of Infectious and Tropical Diseases, Cochin Hospital, AP-HP, Paris, France Objectives Combining noninvasive tests increases diagnostic accuracy for staging liver fibrosis in hepatitis C virus (HCV)-infected patients, but this strategy remains to be validated in HIV/HCV coinfection. We compared the performances of transient elastography (TE), Fibrotest (FT), the aspartate aminotransferase-to-platelet ratio index (APRI) and two algorithms combining TE and FT (Castera) or APRI and FT (SAFE) in HIV/HCV coinfection. Methods One hundred and sixteen HIV/HCV-coinfected patients (64% male; median age 44 years) enrolled in two French multicentre studies (the HEPAVIH cohort and FIBROSTIC) for whom TE, FT and APRI data were available were included in the study. Diagnostic accuracies for significant fibrosis (METAVIR F 2) and cirrhosis (F4) were evaluated by measuring the area under the receiver-operating characteristic curve (AUROC) and calculating percentages of correctly classified (CC) patients, taking liver biopsy as a reference. Results For F 2, both TE and FT (AUROC = 0.87 and 0.85, respectively) had a better diagnostic performance than APRI (AUROC = 0.71; P < 0.005). Although the percentage of CC patients was significantly higher with Castera s algorithm than with SAFE (61.2% vs. 31.9%, respectively; P < ), this percentage was lower than that for TE (80.2%; P < ) or FT (73.3%; P < ) taken separately. For F4, TE (AUROC = 0.92) had a better performance than FT (AUROC = 0.78; P = 0.005) or APRI (AUROC = 0.73; P = 0.025). Although the percentage of CC patients was significantly higher with the SAFE algorithm than with Castera s (76.7% vs. 68.1%, respectively; P < 0.050), it was still lower than that for TE (85.3%; P < 0.033). Conclusions In HIV/HCV-coinfected patients, TE and FT have a similar diagnostic accuracy for significant fibrosis, whereas for cirrhosis TE has the best accuracy. The use of the SAFE and Castera algorithms does not seem to improve diagnostic performance. Keywords: FibroScan, HIV/hepatitis C virus coinfection, liver biopsy, liver fibrosis, noninvasive, serum markers, transient elastography Accepted 17 July 2013 Correspondence: Dr Laurent Castera, Service d Hépatologie, Hôpital Beaujon, 100 boulevard du Général Leclerc, Clichy, France. Tel: ; fax: ; laurent.castera@bjn.aphp.fr 30

2 Algorithms combining noninvasive tests in HIV/HCV-infected patients 31 Introduction Among the 35 million people currently living with HIV world-wide, around one-fourth are coinfected with hepatitis C virus (HCV) [1]. HIV-positive individuals with chronic hepatitis C show a faster progression of liver fibrosis [2,3]. On average, nearly half of patients have developed liver cirrhosis after 25 years of HCV infection. Assessment of liver fibrosis is thus of critical importance in HIV/HCV-coinfected patients not only for prognosis but also for antiviral therapy indications. Indeed, two endpoints are clinically relevant: (i) the presence of significant fibrosis, which is the hallmark of progressive disease and an indication for antiviral treatment; (ii) the presence of cirrhosis, which is an indication for specific monitoring of complications related to portal hypertension and to the increased risk of developing hepatocellular carcinoma [4]. Liver biopsy (LB) is classically considered the gold standard for staging fibrosis [5], although it has several limitations: it is an invasive and painful procedure [6 8], with rare but potentially life-threatening complications [9], and prone to sampling errors [10 12]. Thus, many patients are reluctant to undergo LB, especially HIV/HCV-coinfected patients who may be discouraged from initiating anti-hcv treatment for this reason. These limitations have stimulated the search for noninvasive approaches [13 15]. A variety of methods have been proposed for the noninvasive assessment of fibrosis in patients with HCV monoinfection. These include the use of serum markers, ranging from those employed in simple routine laboratory tests, such as the aspartate aminotransferase-to-platelet ratio index (APRI) [16], to more complex scores, such as the Fibrotest (FT) [17], and more recently measurement of liver stiffness by transient elastography (TE) [18,19]. Algorithms combining different noninvasive tests, such as TE and FT (Castera) [18] or FT and APRI (SAFE) [20,21], have been proposed and suggested to improve diagnostic accuracy in HCV-monoinfected patients [22]. However, no information is available regarding the performance of such algorithms in HIV/HCV-coinfected patients. The aim of this study was to compare, taking LB as a reference, the diagnostic performances for significant fibrosis and cirrhosis of TE, FT and APRI, as well as two algorithms combining these noninvasive tests, in HIV/ HCV-coinfected patients. Patients and methods Patients The study population consisted of 116 patients with HIV/ HCV coinfection, selected from two French multicentre studies (the Agence Nationale de Recherche sur le SIDA (ANRS) CO13 HEPAVIH cohort and FIBROSTIC; see Appendix), who underwent noninvasive tests for fibrosis evaluation (TE, FT and APRI) within a maximum of 1 year from LB. ANRS CO13 HEPAVIH is a nationwide French cohort, supported by the National Agency for AIDS and Viral Hepatitis, which recruited 1175 HIV/HCV-coinfected individuals from 17 centres between 2006 and 2008 [23]. The FIBROSTIC study is a multicentre prospective crosssectional diagnostic accuracy study conducted in 23 French centres, which included 1839 patients with chronic viral hepatitis who underwent TE, tests for biomarkers and LB, of whom 110 had HIV/HCV coinfection [24]. A total of 245 patients met the inclusion criteria. Of these, 129 patients could not be included in analyses for the following reasons: missing data (n = 30), antiviral HCV treatment initiated between noninvasive tests and LB (n = 38), coinfection with hepatitis B virus (n = 15), decompensated cirrhosis (n = 4), liver transplantation (n = 35), hepatocellular carcinoma (n = 1) or LB unavailable for central reading (n = 6). Finally, data for 116 patients were analysed (84 patients from the ANRS CO13 HEPAVIH cohort and 32 from the FIBROSTIC study). Liver stiffness measurement Liver stiffness measurements were performed using TE (FibroScan ; Echosens, Paris, France). Details of the technical background and examination procedure have been given previously [25]. Ten successful measurements were performed on each patient. The success rate was calculated as the number of validated measurements divided by the total number of measurements. The results were expressed in kilopascals (kpa). The median value of successful measurements was considered representative of the liver stiffness in a given patient, according to the manufacturer s recommendations [26]. Serum fibrosis scores Biological parameters (aspartate aminotransferase, alanine aminotransferase, γ-glutamyl-transpeptidase, total bilirubin, α2-macroglobulin, apolipoprotein A1, haptoglobin and platelet count) used to calculate FT and APRI were determined. The FT score was purchased from the Biopredictive website ( The APRI was calculated as follows: aspartate transaminase ( upper limit of normal) 100/platelet count (109/L). Cut-offs for APRI and FT were taken from the original publications [16,17]. Liver histology and staging of liver fibrosis LBs were performed by senior operators at each site. Biopsy specimens were fixed in formalin and embedded in paraffin.

3 32 L Castera et al. (a) (b) HIV/HCV-coinfected patients (n = 116) HIV/HCV-coinfected patients (n = 116) APRI (n = 116) FIBROSCAN + FIBROTEST (n = 116) <0.50 (n = 37) (n = 56) >1.50 (n = 23) Disagree Agree FIBROTEST (n = 56) FS 7.1 kpa and FT 0.48 (n = 12) FS<7.1 kpa and FT >0.48 (n = 24) FS<7.1 kpa and FT 0.48 (n = 35) FS 7.1 kpa and FT >0.48 (n = 45) <0.48 (n = 22) >0.48 (n = 34) LIVER BIOPSY NEEDED (n = 59) Significant fibrosis present No need for liver biopsy (n = 57) LIVER BIOPSY NEEDED (n = 36 ) Significant fibrosis absent or present No need for liver biopsy (n = 80) Fig. 1 SAFE (a) or Castera (b) algorithms for significant fibrosis ( F2 by METAVIR score). Each figure reports the cut-offs used for each noninvasive test, and the number of patients for each branch when the algorithm was applied to the 116 HIV HCV coinfected patients. APRI, aspartate aminotransferase-to-platelet ratio index; FS, FibroScan; FT, Fibrotest; HCV, hepatitis C virus. They were then centralized and re-analysed by the same trained pathologist (VP) blinded to the results of noninvasive tests. Liver fibrosis was staged on a F0-to-F4 scale according to the METAVIR scoring system [27], as follows: F0 = no fibrosis; F1= portal fibrosis without septa; F2 = portal fibrosis with rare septa; F3 = numerous septa without cirrhosis; F4 = cirrhosis. Specimens with pathology other than HCVassociated fibrosis were excluded from the analysis. Algorithms The SAFE algorithm Two distinct SAFE algorithms [20] for detection of significant fibrosis and cirrhosis, respectively, based on sequential use of APRI, FT and LB were applied to the data for the 116 patients and the results were compared with the histological diagnosis based on LB, which was taken as the gold standard. Both algorithms use APRI as the initial screening test, followed by FT as a second step, and limits the use of LB to those patients for whom noninvasive markers were inaccurate. The only differences between these algorithms are the thresholds set for each endpoint. Figures 1a and 2a describe the SAFE algorithms (respectively for significant fibrosis and cirrhosis), including cutoff values for APRI and FT and the related decisional tree. The Castera algorithm The proposed algorithm uses the combination of TE and FT as first-line assessment of fibrosis and is based on the agreement or disagreement of the results of these tests: when TE and FT agree, no LB is performed, whereas when TE and FT disagree, an LB is needed (Figs 1b,2b). Cut-offs used are those proposed in the original publication [18]. Statistical analysis Patients characteristics are given as median values [with interquartile range (IQR): Q1 Q3] or percentages. Cochran s Q test or McNemar s χ 2 test was used as appropriate for comparison of qualitative data. Tests were two-tailed and P-values < 0.05 were considered as significant. The diagnostic accuracy of the different tests was evaluated by constructing receiver-operating characteristic (ROC) curves and calculating the area under the ROC curve (AUROC) and its corresponding 95% confidence intervals (CIs). Sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios (LRs) and the percentage of correctly classified patients were calculated using previously described cutoffs for noninvasive tests [16 18]. AUROCs were adjusted according to the prevalence of fibrosis stages using DANA (difference between advanced and nonadvanced fibrosis = [(prevalence F2 2 + prevalence F3 3 + prevalence F4 4)/(prevalence F2 + prevalence F3 + prevalence F4)] [prevalence F1/(prevalence F0 + prevalence F1)]) as proposed by Poynard et al. [28]. The adjusted AUROCs were calculated as follows: adjauroc = obauroc + (0.1056) (2.5 DANA). Comparisons of AUROCs were performed using the method described by Delong et al. [29]. Performance of noninvasive tests was further assessed by calculating the percentage of correctly classified patients (true positive and true negative) taking LB as a reference in patients for whom LB was unnecessary. Overall performance of algorithms was assessed using both the percentage of saved LB (no need for LB according to the algorithm) as well as the percentage of correctly classified patients.

4 Algorithms combining noninvasive tests in HIV/HCV-infected patients 33 (a) HIV/HCV-coinfected patients (n = 116) (b) HIV/HCV-coinfected patients (n = 116) APRI (n = 116) FIBROSCAN + FIBROTEST (n = 116) <1 (n = 78) 1-2 (n = 22) >2 (n = 16) FIBROTEST (n = 38) <0.48 (n = 8) (n = 11) >0.75 (n = 19) LIVER BIOPSY NEEDED (n = 11) Disagree FS 12.5 kpa FS<12.5 kpa and FT <0.75 and FT 0.75 (n = 7) (n = 18) Agree FS<12.5 kpa FS 12.5 kpa and FT <0.75 and FT 0.75 (n = 74) (n = 17) Cirrhosis absent No need for liver biopsy (n = 86) Cirrhosis present No need for liver biopsy (n = 19) LIVER BIOPSY NEEDED (n = 25) Cirrhosis absent or present No need for liver biopsy (n = 91) Fig. 2 SAFE (a) or Castera (b) algorithms for cirrhosis (F4 by METAVIR score). Each figure reports the cut-offs used for each noninvasive test, and the number of patients for each branch when the algorithm was applied to the 116 HIV HCV coinfected patients. APRI, aspartate aminotransferaseto-platelet ratio index; FS, FibroScan; FT, Fibrotest; HCV, hepatitis C virus. Table 1 Characteristics of patients from the HEPAVIH cohort and FIBROSTIC study Results Patients Total (n = 116) Gender (male) [n (%)] 74 (64) Age (years) 44 (41 48) BMI (kg/m 2 ) 22 (20 25) AST (IU/L) 50 (38 70) ALT (IU/L) 54 (33 91) Platelets (10 9 /L) 202 ( ) CD4 count (10 9 /L)* 416 ( ) HIV RNA < 50 copies/ml (%)* 65 HCV RNA (log IU/mL)* 6.3 ( ) LB Length (mm) 19.5 Number of portal tracts 14 Delay between LB and NIT (months) 1 (0 5.7) Fibrosis stage according to METAVIR [n (%)] F0 3 (3) F1 65 (56) F2 25 (21) F3 10 (9) F4 13 (11) Results are expressed as median (interquartile range: Q1 Q3) unless otherwise stated. ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; HCV, hepatitis C virus; LB, liver biopsy; NIT, noninvasive test. *Available for 84 patients. The characteristics of the 116 patients are shown in Table 1. There were 74 men and 42 women, the median age being 44 (IQR: 41 48) years. The median LB length was 19.5 mm and the median number of portal tracts was 14. LB length was 15 mm in 87 patients (75%). Significant fibrosis (F 2) was present in 41% of patients and cirrhosis (F4) in 11%. The DANA was Comparison of TE, FT and APRI Diagnosis of significant fibrosis The performances of TE, FT and APRI for the diagnosis of significant fibrosis are described in Table 2. TE and FT had a better diagnostic performance than APRI (AUROC 0.87, 0.85 and 0.71, respectively; P < 0.004). The percentage of correctly classified patients according to LB was significantly higher for TE and FT than for APRI (80.2, 73.3 and 38.8%, respectively; P < ). Diagnosis of cirrhosis The performances of TE, FT and APRI for the diagnosis of cirrhosis are described in Table 2. TE had a better performance for cirrhosis than FT (AUROC 0.92 and 0.78, respectively; P = 0.005) or APRI (AUROC 0.92 and 0.73, respectively; P = 0.025). The percentage of correctly classified patients according to LB was significantly higher for TE than for FT or APRI (85.3, 72.4 and 68.1%, respectively; P < 0.003). Comparison of algorithms Diagnosis of significant fibrosis As shown in Figure 1a, with the SAFE algorithm, an LB was deemed unnecessary for diagnosing the presence of significant fibrosis in 57 patients (49.1%). Among these 57 patients, the presence of significant fibrosis was confirmed by LB in 37 patients (65%). Overall, the percentage of

5 34 L Castera et al. Table 2 Performance of transient elastography (TE), Fibrotest (FT) and the aspartate aminotransferase-to-platelet ratio index (APRI) for the diagnosis of significant fibrosis and cirrhosis (in patients from the HEPAVIH cohort and FIBROSTIC study; n = 116) Significant fibrosis Cirrhosis TE (FibroScan) Fibrotest APRI TE (FibroScan) Fibrotest APRI AUROC (95% CI) 0.87 ( ) 0.85 ( ) 0.71 ( ) 0.92 ( ) 0.78 ( ) 0.73 ( ) Cut-offs 7.1 kpa > > kpa > 2.0 Sensitivity (%) Specificity (%) Positive predictive value (%) Negative predictive value (%) Positive likelihood ratio Negative likelihood ratio Correctly classified (%) Significant fibrosis: global comparison between AUROCs: P = 0.004; TE vs. FT: P = 0.611; TE vs. APRI: P = 0.001; FT vs. APRI: P = 0.005; global comparison between correctly classified patients: P < ; TE vs. FT: P = 0.182; TE vs. APRI: P < ; FT vs. APRI: P < Cirrhosis: global comparison between AUROCs: P = 0.007; TE vs. FT: P = 0.005; TE vs. APRI: P = 0.025; FT vs. APRI: P = 0.593; global comparison between correctly classified patients: P = 0.001; TE vs. FT: P = 0.003; TE vs. APRI: P < 0.001; FT vs. APRI: P = AUROC, area under the receiver-operating characteristic curve; CI, confidence interval. Table 3 Performance of the two algorithms for the diagnosis of significant fibrosis and cirrhosis (for patients from the HEPAVIH cohort and FIBROSTIC study; n = 116) PPV (%) NPV (%) Saved biopsies (%) Correctly classified patients (%) Significant fibrosis SAFE algorithm 64.9 * Castera algorithm Cirrhosis SAFE algorithm Castera algorithm Comparison for correctly classified patients: F 2, P < ; F4, P = Comparison for saved biopsies: F 2, P = 0.003; F4, P = PPV, positive predictive value; NPV, negative predictive value. *Not applicable. correctly classified patients was 31.9% (37 of 116) (Table 3), which is much lower than that for TE (80.2%; P < ) or FT (73.3%; P < ) taken separately. Using Castera s algorithm, an LB was deemed unnecessary for diagnosing the presence or the absence of significant fibrosis in 80 patients (69.0%) (Fig. 1b). When there was agreement between TE and FT results for the absence (35 patients) or the presence (45 patients) of significant fibrosis, the result was also confirmed by LB for 33 (94%) and 38 (84%) patients, respectively. Overall, the percentage of correctly classified patients was 61.2% (71 of 116), which is significantly higher than that obtained for the SAFE algorithm (31.9%; 37 of 116; P < ) but lower than that obtained for TE (80.2%; P < ) or FT (73.3%; P = ) taken separately. However, the negative predictive value of the Castera algorithm for excluding significant fibrosis was very good (94.3%) and higher than those of TE, FT and APRI taken separately (88.1, 89.4 and 75.7%, respectively). Diagnosis of cirrhosis Using the SAFE biopsy algorithm, an LB was deemed unnecessary for diagnosing the absence or the presence of cirrhosis in 105 patients (90.5%) (Fig. 2a). Among the 86 patients in whom the SAFE algorithm suggested the absence of cirrhosis, this was confirmed by LB in 82 patients (95%), whereas in the 19 patients in whom SAFE suggested the presence of cirrhosis, this was confirmed by LB in seven patients (37%). Overall, the percentage of correctly classified patients was 76.7% (89 of 116), which was significantly lower than that for TE (85.3%; P = 0.003). Using the Castera algorithm, an LB was deemed unnecessary for diagnosing the absence or the presence of cirrhosis in 91 patients (78.4%) (Fig. 2b). Among the 74 patients in whom TE and FT agreed for the absence of cirrhosis, this was confirmed by LB in 72 patients (97%), whereas in the 17 patients in whom TE and FT agreed for the presence of cirrhosis, this was confirmed by LB in seven

6 Algorithms combining noninvasive tests in HIV/HCV-infected patients 35 patients (41%). Overall, the percentage of correctly classified patients was 68.1% (79 of 116), which is significantly lower than that for the SAFE algorithm (76.7%; 89 of 116; P = 0.05) as well as that for TE (85.3%; P < ) or FT (72.4%; P = 0.025) taken separately. Diagnostic performances of noninvasive tests and algorithms according to fibrosis stage distribution and liver biopsy length Given the uneven distribution of fibrosis stages in our population, we adjusted the AUROCs according to the DANA. After adjustment, TE and FT still had a better performance for diagnosing significant fibrosis than APRI (adjauroc 0.95, 0.93 and 0.78, respectively). Regarding cirrhosis, TE still had the best performance (adjauroc 0.99 vs and 0.80, respectively). Diagnostic performances of noninvasive tests were also analysed in the subgroup of 87 patients for whom LB was 15mm. For the diagnosis of significant fibrosis, TE and FT had a better diagnostic performance than APRI (AUROC 0.86, 0.87 and 0.71, respectively; P = 0.015); also, percentages of correctly classified patients for TE and FT were significantly higher than that for APRI (79.3 and 75.9 vs. 39.1%, respectively; P < ). TE had a better performance for cirrhosis than FT and APRI (AUROC 0.86, 0.69 and 0.70, respectively; P = 0.045) as well as a higher percentage of correctly classified patients (83.9, 71.3 and 66.7%, respectively; P = 0.004). Regarding the algorithms for the diagnosis of significant fibrosis, Castera s allowed the correct classification of a higher number of patients than the SAFE algorithm (62.1 vs. 33.3%, respectively; P < ). Conversely, for the diagnosis of cirrhosis, although there was a trend towards a higher percentage of patients correctly classified by SAFE (75.9%) compared with Castera s algorithm (66.7%), the difference did not reach statistical significance (P = 0.073). Discussion Noninvasive diagnosis of liver fibrosis is one of the fields that have evolved most rapidly in recent years. Noninvasive methods are now increasingly used in clinical practice in HCV-monoinfected patients, resulting in a significant decrease in the number of LBs performed in France [30]. In addition, the use of either TE or biomarkers for firstline assessment of liver fibrosis in HCV-infected patients without comorbidities has been approved, after an independent systematic review, by the French Health Authorities [31], and recently endorsed by the European Association for Study of Liver Clinical Practice guidelines [32]. As regards HIV/HCV-coinfected patients, noninvasive methods still remain to be validated and the performance of serum biomarkers may be not as good as in HCVmonoinfected patients [33,34]. In the present study, we assessed the performance of algorithms combining different noninvasive tests in HIV/ HCV-coinfected patients. Overall, our results suggest that the use of algorithms does not improve the diagnostic performance in this population. Among noninvasive methods available, we chose to evaluate two different and complementary approaches: (i) a physical approach based on the measurement of liver stiffness using TE, and (ii) a biological approach based on serum biomarkers including a patented algorithm (FT) and a free nonpatented index (APRI) [35]. These three noninvasive methods are by far the most widely used and validated in HCV monoinfection [36 38] and the study was conducted independently from the promoters of the tests. The population of HIV/HCV-coinfected patients we studied was selected from two large French cohorts of HCV-infected patients with or without HIV infection. Our methods, consisting of selecting high-quality LBs (median length = 19.5 mm) as a standard reference, performed within a median of 1 month (IQR 0 5.7) of noninvasive tests, and analysis of LBs by a single pathologist blinded to the results of noninvasive methods, are in accordance with recent recommendations for accurate fibrosis staging [39]. Also, detection of significant fibrosis and detection of cirrhosis are two clinically relevant endpoints. The presence of significant fibrosis in HIV/HCV-coinfected patients is considered a hallmark of a progressive liver disease and an indication for antiviral treatment. Although antiviral therapy has been shown to increase life expectancy, to improve quality of life and to be cost-effective, it can be associated with severe side effects. In addition, more than half of HIV/HCV-coinfected patients are still resistant to the current treatment regimen based on the combination of peginterferon and ribavirin, given for 48 weeks [1], and fibrosis stage is also an independent predictor of sustained HCV eradication [40]. Finally, early diagnosis of cirrhosis is important in HIV/HCV-coinfected patients not only because it triggers screening for complications such as oesophageal varices and hepatocellular carcinoma, but also because these patients have the most urgent need for antiviral therapy. In the first part of the study, we showed that, when taken separately, TE and FT had similar diagnostic accuracies for significant fibrosis, whereas TE had the best accuracy for cirrhosis. These results are consistent with previously reported findings in HCV-monoinfected patients [18,24,41]. Diagnostic performances of TE for significant fibrosis and cirrhosis (AUROC 0.87 and 0.92, respectively) were consistent with those reported in meta-analyses [36,42] as well as in HIV/HCV coinfection [43 45]. Similarly, diagnostic performances of FT

7 36 L Castera et al. in our population were consistent with those previously reported [46,47]. The fact that FT outperformed APRI in diagnosing significant fibrosis is consistent with the findings of Cacoub et al. [46]. Finally, the better diagnostic performance of TE for cirrhosis compared with APRI has already been reported by Sanchez-Conde et al. [48]. In the second part of the study, we evaluated algorithms combining TE and FT (Castera) or APRI and FT (SAFE) to assess whether their use might increase diagnostic accuracy. For the diagnosis of significant fibrosis, the number of saved LBs using the Castera algorithm was significantly higher than that for SAFE (69 vs. 49.1%, respectively; P = 0.003), a finding consistent with those reported in a recent comparative study in HCV monoinfection [22]. The percentages of LBs saved by both algorithms were similar to those reported in the original studies [18,21], as well as in a large-scale validation study [20]. However, although the percentage of correctly classified patients was significantly higher using the Castera algorithm than using SAFE (61.2 vs. 31.9%, respectively; P < ), it was lower than those obtained when using TE or FT separately (80.2 and 73.3%, respectively). Nevertheless, the important advantage of using the Castera algorithm is its high negative predictive value, allowing significant fibrosis to be ruled out more safely than when using single tests. For the diagnosis of cirrhosis, the percentages of correctly classified patients using Castera s algorithm or SAFE (68.1 and 76.7%, respectively) were also lower than that obtained using TE (85.3%). Thus, it seems that there is no advantage to using algorithms combining either TE and FT or APRI and FT in HIV/HCV-coinfected patients for the diagnosis of either significant fibrosis or significant cirrhosis. These results are in contradiction with those reported in HCV-monoinfected patients [18,20,21]. Although there is no clearcut explanation for this finding, differences in studied populations and the prevalence of fibrosis stages may account for this discrepancy. However, when the AUROCs were adjusted according to the DANA as proposed by Poynard et al. [28], although a global increase in the AUROCs of TE, FT and APRI was observed, our findings were unchanged, showing similar diagnostic accuracies of TE and FT for significant fibrosis and the best accuracy for TE for cirrhosis. Increasing LB length did not significantly change the results either, although the percentage of correctly classified patients for cirrhosis did not differ between the two algorithms. Furthermore, the performance of APRI and FT could be affected in HIV/HCV-coinfected patients by factors such as HIV-induced thrombocytopenia [49] and drug-related toxicity (e.g. bilirubin elevation caused by atazanavir or gamma-glutamyl transpeptidase abnormalities caused by nonnucleoside reverse transcriptase inhibitors) [1,50,51]. Finally, our results may not be applicable to all HIV/HCV-coinfected patients, given the retrospective nature of the present study and the potential selection bias of our population. In conclusion, in HIV/HCV-coinfected patients, TE and FT had similar diagnostic accuracies for significant fibrosis, whereas TE had the best accuracy for cirrhosis. The use of algorithms combining these tests does not seem to improve diagnostic performance. Acknowledgement Conflicts of interest: The authors have no conflicts of interest to declare. Appendix ANRS CO13 HEPAVIH cohort Scientific committee D. Salmon, F. Dabis, M. Winnock, M. A. Loko, P. Sogni, Y. Benhamou, P. Trimoulet, J. Izopet, V. Paradis, B. Spire, P. Carrieri, C. Katlama, G. Pialoux, I. Poizot-Martin, B. Marchou, E. Rosenthal, A. Bicart See, M. Bentata, A. Gervais, C. Lascoux-Combe, C. Goujard, K. Lacombe, C. Duvivier, D. Vittecoq, D. Neau, P. Morlat, F. BaniSadr, L. Meyer, F. Boufassa, S. Dominguez, B. Autran, A.M. Roque, C. Solas, H. Fontaine, L. Serfaty, G. Chêne, D. Costagliola and S. Couffin-Cadiergues. Clinical centres (ward/participating physicians) CHU Cochin (Médecine Interne et Maladies Infectieuses/D. Salmon; Hépato-gastro-entérologie/P. Sogni; Anatomopathologie/B. Terris, Z. Makhlouf, G. Dubost, F. Tessier, L. Gibault, F. Beuvon, E. Chambon and T. Lazure; Virologie/A. Krivine); CHU Pitié-Salpétrière (Maladies Infectieuses et Tropicales/C. Katlama, M. A. Valantin and S. Dominguez; Hépato-gastro-entérologie/Y. Benhamou; Anatomopathologie/F. Charlotte; Virologie/S. Fourati); CHU Sainte- Marguerite, Marseille (Hématologie et VIH/I. Poizot-Martin and O. Zaegel; Virologie/C. Tamalet); CHU Tenon (Maladies Infectieuses et Tropicales/G. Pialoux, P. Bonnard and F. Bani-Sadr; Anatomo-pathologie/P. Callard and F. Bendjaballah; Virologie/H. Assami); CHU Purpan Toulouse (Maladies Infectieuses et Tropicales/B. Marchou; Hépatogastro-entérologie/L. Alric, K. Barange and S. Metivier; Anatomo-pathologie/J. Selves; Virologie/F. Nicot); CHU Archet, Nice (Médecine Interne/E. Rosenthal; Infectiologie/C. Pradier; Anatomo-pathologie/J. Haudebourg and M. C. Saint-Paul); CHU Avicenne, Paris (Médecine Interne Unité VIH/A. Krivitzky and M. Bentata; Anatomo-pathologie/M. Ziol; Virologie/Y. Baazia); Hôpital Joseph-Ducuing, Toulouse (Médecine Interne/M. Uzan, A. Bicart-See and D. Garipuy; Anatomo-pathologie/J. Selves;

8 Algorithms combining noninvasive tests in HIV/HCV-infected patients 37 Virologie/F. Nicot); CHU Bichat Claude-Bernard, Paris (Maladies Infectieuses/P. Yéni and A. Gervais; Anatomopathologie/H. Adle-Biassette); CHU Saint-Louis (Médecine Interne/D. Séréni and C. Lascoux Combe; Anatomopathologie/P. Bertheau and J. Duclos; Virologie/P. Palmer); CHU Saint Antoine (Maladies Infectieuses et Tropicales/P. M. Girard, K. Lacombe and P. Campa; Anatomopathologie/D. Wendum, P. Cervera and J. Adam; Virologie/N. Harchi); CHU Bicêtre (Médecine Interne/J. F. Delfraissy, C. Goujard and Y. Quertainmont; Virologie/ C. Pallier); CHU Paul-Brousse (Maladies Infectieuses/ D. Vittecoq); CHU Necker (Maladies Infectieuses et Tropicales/O. Lortholary, C. Duvivier and S. Boucly), ANRS CO 3 Aquitaine cohort (D. Neau, P. Morlat, I. Raymond and I. Louis; Anatomo-pathologie/P. Bioulac-Sage; Virologie/P. Trimoulet and P. Pinson). Pathologists A. Martin, M. Ziol, A. Janin, J. Duclos, P. Bertheau, C. Guettier, T. Lazure, D. Henin, L. Choudat, H. Adle-Biassette, F. Capron, F. Charlotte, J. F. Flejou, D. Wendum, M. C. Vacher-Lavenu, F. Beuvon, L. Gibault, E. Chambon, B. Terris, R. Belkhir, P. Brousset, J. Selves, P. Callard, A. Martin, B. Guigui, A. Chollat-Namy, D. Basbous, F. Daussy Bonamour, J. C. Poluzzi, J. Haudebourg, M. C. Saint-Paul, P. Denis, P. Roux and M. Pizzi-Anselme. Data collection, management and statistical analyses D. Beniken, A. Ivanova, A. Fooladi, M. Azar, P Honoré, S. Breau, L. Serini, M. Mole, M. Malet, C. Bolliot, O. Eldbouni, A. Maignan, S. Mellul, G. Alexandre, A. Ganon, S. Thirrée, S. Gillet, J. Delaune, L. Dequae Merchadou, E. Pambrun, A. Frosch, J. Cohen, V. Villes, P. Kurkdji, M. A. Loko and M. Winnock. FIBROSTIC study List of coinvestigators APHP: Hôpital Beaujon: Professor P. Bedossa, Dr F. Degos, Professor P. Marcellin and Professor M. Vidaud; Hôpital Cochin-Necker: Professor S. Pol, Professor Ph. Sogni and Professor D. Salmon; Hôpital Jean Verdier: Dr A. Mahmoudi; Hôpital Paul Brousse: Dr B. Roche; Hôpital Saint Antoine: Professor R. Poupon and Dr L. Serfaty; Hôpital Pitié-Salpétrière: Professor C. Katlama, Dr M. Munteanu and Dr P. Lebray; CHU Amiens: Dr E. Nguyen Khac; CHU Angers: Professor P. Cales; CHU Besançon: Professor V. Di Martino; CHU Bordeaux Pessac: Professor V. de Ledinghen; CHU Brest: Professor J. B. Nousbaum; CHU Clermont Ferrand: Dr A. Abergel; CHU Grenoble: Professor J. P. Zarski; CHU Lille: Professor Ph. Mathurin; CHU Limoges: Dr D. Loustau-Ratti; CHU Hôtel Dieu Lyon: Professor C. Trépo; Hôpital Saint Joseph Marseille: Dr M. Bourlière; CHU Montpellier: Professor D. Larrey; CHU Nice: Professor A. Tran; CHU Nancy: Professor J. P. Bronowicki; CHU Nantes: Dr J. Gournay; Hôpital d Orleans: Dr X. Causse; CHU Rennes: Professor D. Guyader; CHU Tours: Professor Y. Bacq; CHU Strasbourg: Professor M. Doffoel. Data collection, management and statistical analyses J. Asselineau, I. Perrot and P. Perez. References 1 Soriano V, Vispo E, Labarga P, Medrano J, Barreiro P. Viral hepatitis and HIV co-infection. Antiviral Res 2010; 85: Benhamou Y, Bochet M, Di Martino V et al. Liver fibrosis progression in human immunodeficiency virus and hepatitis C virus coinfected patients. The Multivirc Group. Hepatology 1999; 30: Salmon-Ceron D, Lewden C, Morlat P et al. Liver disease as a major cause of death among HIV infected patients: role of hepatitis C and B viruses and alcohol. J Hepatol 2005; 42: Ghany MG, Strader DB, Thomas DL, Seeff LB. Diagnosis, management, and treatment of hepatitis C: an update. Hepatology 2009; 49: Bravo AA, Sheth SG, Chopra S. Liver biopsy. N Engl J Med 2001; 344: Castera L, Negre I, Samii K, Buffet C. Pain experienced during percutaneous liver biopsy. Hepatology 1999; 30: Cadranel JF, Rufat P, Degos F. Practices of liver biopsy in France: results of a prospective nationwide survey. For the Group of Epidemiology of the French Association for the Study of the Liver (AFEF). Hepatology 2000; 32: Castera L, Negre I, Samii K, Buffet C. Patient-administered nitrous oxide/oxygen inhalation provides safe and effective analgesia for percutaneous liver biopsy: a randomized placebo-controlled trial. Am J Gastroenterol 2001; 96: Piccinino F, Sagnelli E, Pasquale G, Giusti G. Complications following percutaneous liver biopsy. A multicentre retrospective study on 68,276 biopsies. J Hepatol 1986; 2: Maharaj B, Maharaj RJ, Leary WP et al. Sampling variability and its influence on the diagnostic yield of percutaneous needle biopsy of the liver. Lancet 1986; 1: Regev A, Berho M, Jeffers LJ et al. Sampling error and intraobserver variation in liver biopsy in patients with chronic HCV infection. Am J Gastroenterol 2002; 97:

9 38 L Castera et al. 12 Bedossa P, Dargère D, Paradis V. Sampling variability of liver fibrosis in chronic hepatitis C. Hepatology 2003; 38: Pinzani M, Vizzutti F, Arena U, Marra F. Technology Insight: noninvasive assessment of liver fibrosis by biochemical scores and elastography. Nat Clin Pract Gastroenterol Hepatol 2008; 5: Castera L, Pinzani M. Non-invasive assessment of liver fibrosis: are we ready? Lancet 2010; 375: Castera L. Noninvasive methods to assess liver disease in patients with hepatitis B or C. Gastroenterology 2012; 142: Wai CT, Greenson JK, Fontana RJ et al. A simple noninvasive index can predict both significant fibrosis and cirrhosis in patients with chronic hepatitis C. Hepatology 2003; 38: Imbert-Bismut F, Ratziu V, Pieroni L, Charlotte F, Benhamou Y, Poynard T. Biochemical markers of liver fibrosis in patients with hepatitis C virus infection: a prospective study. Lancet 2001; 357: Castera L, Vergniol J, Foucher J et al. Prospective comparison of transient elastography, Fibrotest, APRI, and liver biopsy for the assessment of fibrosis in chronic hepatitis C. Gastroenterology 2005; 128: Ziol M, Handra-Luca A, Kettaneh A et al. Noninvasive assessment of liver fibrosis by measurement of stiffness in patients with chronic hepatitis C. Hepatology 2005; 41: Sebastiani G, Halfon P, Castera L et al. SAFE biopsy: a validated method for large-scale staging of liver fibrosis in chronic hepatitis C. Hepatology 2009; 49: Sebastiani G, Vario A, Guido M et al. Stepwise combination algorithms of non-invasive markers to diagnose significant fibrosis in chronic hepatitis C. J Hepatol 2006; 44: Castera L, Sebastiani G, Le Bail B, de Ledinghen V, Couzigou P, Alberti A. Prospective comparison of two algorithms combining non-invasive methods for staging liver fibrosis in chronic hepatitis C. J Hepatol 2010; 52: Loko MA, Salmon D, Carrieri P et al. The French national prospective cohort of patients co-infected with HIV and HCV (ANRS CO13 HEPAVIH): early findings, BMC Infect Dis 2010; 10: Degos F, Perez P, Roche B et al. Diagnostic accuracy of FibroScan and comparison to liver fibrosis biomarkers in chronic viral hepatitis: a multicenter prospective study (the FIBROSTIC study). J Hepatol 2010; 53: Sandrin L, Fourquet B, Hasquenoph JM et al. Transient elastography: a new noninvasive method for assessment of hepatic fibrosis. Ultrasound Med Biol 2003; 29: Castera L, Forns X, Alberti A. Non-invasive evaluation of liver fibrosis using transient elastography. J Hepatol 2008; 48: Intraobserver and interobserver variations in liver biopsy interpretation in patients with chronic hepatitis C. The French METAVIR Cooperative Study Group. Hepatology 1994; 20 (1 Pt 1): Poynard T, Halfon P, Castera L et al. Standardization of ROC curve areas for diagnostic evaluation of liver fibrosis markers based on prevalences of fibrosis stages. Clin Chem 2007; 53: DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: Castera L, Denis J, Babany G, Roudot-Thoraval F. Evolving practices of non-invasive markers of liver fibrosis in patients with chronic hepatitis C in France: time for new guidelines? J Hepatol 2007; 46: HAS. Non invasive methods for the evaluation of hepatic fibrosis/cirrhosis (text in French). Available at -sante.fr/portail/upload/docs/application/pdf/ / document_avis_fibrose_cirrhose_dec_2008.pdf. 32 European Association for the Study of the Liver (EASL) Clinical Practice Guidelines: management of hepatitis C infection. J Hepatol 2011; 55: Loko MA, Castera L, Dabis F et al. Validation and comparison of simple noninvasive indexes for predicting liver fibrosis in HIV-HCV-coinfected patients: ANRS CO3 Aquitaine cohort. Am J Gastroenterol 2008; 103: Macias J, Giron-Gonzalez JA, Gonzalez-Serrano M et al. Prediction of liver fibrosis in human immunodeficiency virus/hepatitis C virus coinfected patients by simple non-invasive indexes. Gut 2006; 55: Castera L, Pinzani M. Biopsy and non-invasive methods for the diagnosis of liver fibrosis: does it take two to tango? Gut 2010; 59: Friedrich-Rust M, Ong MF, Martens S et al. Performance of transient elastography for the staging of liver fibrosis: a meta-analysis. Gastroenterology 2008; 134: Poynard T, Morra R, Halfon P et al. Meta-analyses of Fibrotest diagnostic value in chronic liver disease. BMC Gastroenterol 2007; 7: Shaheen AA, Myers RP. Diagnostic accuracy of the aspartate aminotransferase-to-platelet ratio index for the prediction of hepatitis C-related fibrosis: a systematic review. Hepatology 2007; 46: Guido M, Rugge M. Liver biopsy sampling in chronic viral hepatitis. Semin Liver Dis 2004; 24: Medrano J, Neukam K, Rallon N et al. Modeling the probability of sustained virological response to Therapy with

10 Algorithms combining noninvasive tests in HIV/HCV-infected patients 39 pegylated interferon plus ribavirin in patients coinfected with hepatitis C virus and HIV. Clin Infect Dis 2010; 51: Castera L, Le Bail B, Roudot-Thoraval F et al. Early detection in routine clinical practice of cirrhosis and oesophageal varices in chronic hepatitis C: comparison of transient elastography (FibroScan) with standard laboratory tests and non-invasive scores. J Hepatol 2009; 50: Talwalkar JA, Kurtz DM, Schoenleber SJ, West CP, Montori VM. Ultrasound-Based Transient Elastography for the Detection of Hepatic Fibrosis: systematic Review and Meta-analysis. Clin Gastroenterol Hepatol 2007; 5: de Ledinghen V, Douvin C, Kettaneh A et al. Diagnosis of hepatic fibrosis and cirrhosis by transient elastography in HIV/hepatitis C virus-coinfected patients. J Acquir Immune Defic Syndr 2006; 41: Kirk GD, Astemborski J, Mehta SH et al. Assessment of liver fibrosis by transient elastography in persons with hepatitis C virus infection or HIV-hepatitis C virus coinfection. Clin Infect Dis 2009; 48: Vergara S, Macias J, Rivero A et al. The use of transient elastometry for assessing liver fibrosis in patients with HIV and hepatitis C virus coinfection. Clin Infect Dis 2007; 45: Cacoub P, Carrat F, Bedossa P et al. Comparison of noninvasive liver fibrosis biomarkers in HIV/HCV co-infected patients: the fibrovic study ANRS HC02. J Hepatol 2008; 48: Cales P, Halfon P, Batisse D et al. Comparison of liver fibrosis blood tests developed for HCV with new specific tests in HIV/HCV co-infection. J Hepatol 2010; 53: Sanchez-Conde M, Montes-Ramirez ML, Miralles P et al. Comparison of transient elastography and liver biopsy for the assessment of liver fibrosis in HIV/hepatitis C viruscoinfected patients and correlation with noninvasive serum markers. J Viral Hepat 2010; 17: Cole JL, Marzec UM, Gunthel CJ et al. Ineffective platelet production in thrombocytopenic human immunodeficiency virus-infected patients. Blood 1998; 91: Bani-Sadr F, Miailhes P, Rosenthal E et al. Risk factors for grade 3 or 4 gamma-glutamyl transferase elevation in HIV/hepatitis C virus-coinfected patients. AIDS 2008; 22: Poynard T, Munteanu M, Imbert-Bismut F et al. Prospective analysis of discordant results between biochemical markers and biopsy in patients with chronic hepatitis C. Clin Chem 2004; 50: