Staging chronic hepatitis B into seven categories, defining inactive carriers and assessing treatment impact using a fibrosis biomarker (FibroTest®) and elastography (FibroScan®)

Accepted Manuscript Staging chronic hepatitis B into seven categories, defining inactive carriers and assessing treatment impact using a fibrosis biomarker (FibroTest) and elastography (FibroScan) Thierry Poynard, Julien Vergniol, Yen Ngo, Juliette Foucher, Vincent Thibault, Mona Munteanu, Wassil Merrouche, Pascal Lebray, Marika Rudler, Olivier Deckmyn, Hugo Perazzo, Dominique Thabut, Vlad Ratziu, Victor de Ledinghen PII: DOI: Reference:

S0168-8278(14)00457-7 http://dx.doi.org/10.1016/j.jhep.2014.06.027 JHEPAT 5229

To appear in:

Journal of Hepatology

Received Date: Revised Date: Accepted Date:

11 March 2014 16 June 2014 20 June 2014

Please cite this article as: Poynard, T., Vergniol, J., Ngo, Y., Foucher, J., Thibault, V., Munteanu, M., Merrouche, W., Lebray, P., Rudler, M., Deckmyn, O., Perazzo, H., Thabut, D., Ratziu, V., de Ledinghen, V., Staging chronic hepatitis B into seven categories, defining inactive carriers and assessing treatment impact using a fibrosis biomarker (FibroTest) and elastography (FibroScan), Journal of Hepatology (2014), doi: http://dx.doi.org/10.1016/j.jhep. 2014.06.027

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1 The correct author/affiliation combinations are placed below.:

Thierry Poynard1,2,3, Julien Vergniol4, Yen Ngo5, Juliette Foucher4, Vincent Thibault6, Mona Munteanu5, Wassil Merrouche4, Pascal Lebray1, Marika Rudler1, Olivier Deckmyn5, Hugo Perazzo 1, Dominique Thabut1, Vlad Ratziu 1, Victor de Ledinghen4,

1

Assistance Publique Hôpitaux de Paris (AP-HP), Groupe Hospitalier Pitié-Salpêtrière, Paris,

France; 2 University Pierre et Marie Curie (UPMC) Univ Paris 06, Paris, France; 3 INSERM, UMR_S 938, Paris, France 4

University of Bordeaux, Bordeaux, France

5

BioPredictive, Paris, France

6

Service de Biochimie Métabolique, Groupe Hospitalier Pitié-Salpêtrière (GHPS), AP-HP

Paris, France

2 Staging chronic hepatitis B into seven categories, defining inactive carriers and assessing treatment impact using a fibrosis biomarker (FibroTest) and elastography (FibroScan)

Thierry Poynard1,2,3, Julien Vergniol4, Yen Ngo5, Juliette Foucher4, Vincent Thibault6, Mona Munteanu5, Wassil Merrouche4, Pascal Lebray1, Marika Rudler1, Olivier Deckmyn5, Hugo Perazzo 1, Dominique Thabut1, Vlad Ratziu1, Victor de Ledinghen4, on behalf of FibroFrance Study Group, and the Bordeaux HBV Study Group.

1

Assistance Publique Hôpitaux de Paris (AP-HP), Groupe Hospitalier Pitié-Salpêtrière, Paris,

France; 2 University Pierre et Marie Curie (UPMC) Univ Paris 06, Paris, France; 3 INSERM, UMR_S 938, Paris, France 4

University of Bordeaux, Bordeaux, France

5

BioPredictive, Paris, France

6

Service de Biochimie Métabolique, Groupe Hospitalier Pitié-Salpêtrière (GHPS), AP-HP

Paris, France 5

Infectious Disease Unit GHPS, AP-HP, Paris, France

FibroFrance-GHPS group, and the Bordeaux HBV Study Group.

Corresponding author: Thierry Poynard, 47-83 Boulevard de l'Hôpital, 75651 Paris, France; [email protected]; Tel: +33 1 42 16 10 22.

Authors' involvement with the manuscript

3 TP: Study concept and design; acquisition of data; analysis and interpretation of data; drafting of the manuscript; critical revision of the manuscript for important intellectual content; statistical analysis; obtained funding, technical, or material support; study supervision. JV, YN, MM, WM, VT, MR, OD, PL, and DT: Acquisition of data; analysis and interpretation of data. VR: Acquisition of data; analysis and interpretation of data; critical revision of the manuscript for important intellectual content. VDL: Study concept and design; acquisition of data; analysis and interpretation of data; drafting of the manuscript; critical revision of the manuscript for important intellectual content. Possible conflict of interest TP is the inventor of patented FibroTest, with a capital interest in BioPredictive, the company marketing this test (FibroSure in USA). The patents belong to the French Public Organization “Assistance Publique Hopitaux de Paris”. YN, MM, OD are full employees of BioPredictive, the company marketing FibroTest. The other authors have nothing to declare. 4973 words, 4 Tables, 4 Figures

4 Abstract Background and Aims The first aim was to extend the validation of FibroTest (FT) and transient elastography (TE) as markers of occurrence of cirrhosis without complications (F4.1), esophageal varices (F4.2), and severe complications (F4.3) in patients with chronic hepatitis B (CHB). The second aim was to validate a previous definition of inactive carrier based on normal FT and ActiTest (normal-FTAT). The third aim was to assess the long-term dynamics of fibrosis in patients with sustained virological response. Methods The 10 years updated individual data of 1434 patients were pooled from two prospective cohorts. Results Of the 1312 patients without a history of complications, varices had occurred at 10 years in 14 patients [F4.2, incidence of 1.7%(0.6-2.8)], and severe complications in 25 [3.7%,95% CI (1.8-5.7)], including hepatocellular carcinoma (HCC) in 21 [3.7%(1.5-5.8)]. Using Coxmultivariate analysis adjusted for treatment, viral load, HBeAg status and ALT, FT and TE were predictive of liver-complications [n=37;AUROC=0.83(0.71-0.90);P<0.0001] and [n=8/844;AUROC=0.82(0.72-0.89); P<0.0001] respectively. Normal-FTAT better identified patients with lower fibrosis progression than the ALT-based standard: 3/163(1.8%) vs. 16/181(8.8%;P=0.004) in the Paris cohort, and 5/195(2.6%) vs. 15/228 (6.6%;P=0.05) in the Bordeaux cohort. Of the 582 responders, 23 had complications [incidence 6.2%(3.2-9.1)] including 19 HCC [5.8%(2.6-9.0)] and 10 with varices [2.6% (0.8-4.4)]. Of the 138 responders with advanced fibrosis, only 31%(15-47%) had fibrosis regression.

5 Conclusion. FibroTest and TE identified three categories of cirrhosis with increasing morbidity. Normal FibroTest and ActiTest were better able to identify inactive B carriers than the standard definition. Despite virological response, the overall incidence of cirrhosis increased, with a remaining 5.8% risk of HCC.

Keywords FibroTest; elastography; fibrosis stages; cirrhosis complications; prognostic factors; hepatocellular carcinoma; surrogate markers; regression fibrosis; progression of fibrosis in virological responder; chronic hepatitis B; FibroSure.

6 Introduction Fibrosis progression is an early estimate of chronic hepatitis B (CHB) severity.[1] Before the stage of cirrhosis, estimates of fibrosis progression have mainly been done using transition rates between successive stages, from normal liver (F0) to cirrhosis (F4), as defined by a histological score such as the METAVIR scoring system.[2] The last stage, F4, included patients with a wide range of severity. There was a need “to begin thinking of cirrhosis as a series of critical steps that culminate in hepatic decompensation".[3] Since 2003 FibroTest (FT)[4] and then transient elastography (TE)[5] have been extensively validated in CHB as markers of the METAVIR fibrosis stages using biopsy,[6,7,8,9,10] liver fibrosis dynamics,[10,11,12,13] morbidity, and mortality.[13,14,15] These studies showed the performances of FT and TE to be at least similar to those of biopsy. However, due to the very low incidence of each severe complication at 5 years,[14,15,16] more patients and a longer follow-up were needed to assess their specific performances for separately predicting the occurrence of hepatocellular carcinoma (HCC) and other liver related complications. The first aim of the study was to validate the performance of FT and TE in predicting three stages of cirrhosis as we did for hepatitis C.[17] The "seven-stages spectrum" includes the four non-cirrhotic stages and three in cirrhosis:[4] F4.1, defined as cirrhosis without varices or severe events; F4.2, defined as the presence of varices without severe events; and F4.3, defined by the occurrence of the first severe event. With regard to the other end of the spectrum, the availability of biomarkers simplified the definition of an inactive HBV carrier, independent of HBeAg status or viral load.[10] In 289 patients with ‘zero’ scores for FT-AT, the negative predictive value for liver-related complications or death at 4 years was 100%. In contrast, of the 275 patients with the classic

7 definition of inactive carrier, 62 (23%) had fibrosis as presumed by FT. The second aim was to prospectively validate this definition in a new population of inactive carriers. The third aim was to assess the long-term dynamics of fibrosis regression or progression in patients with virological response.[ 18]

8 Methods We analyzed the updated individual data of two prospective cohorts (Figure 1).[10,15] Details of methods are given in Supplementary file S1. The cohorts’ protocols (NCT01927133,NCT01241227) were approved as noninterventional studies by institutional review boards and regulatory agencies, and were conducted in accordance with principles of Good Clinical Practice and the declaration of Helsinki. Informed consent was obtained for all patients. All co-authors had access to the study data and had reviewed and approved the final manuscript. All consecutive patients over the age of 18 years, who were chronic carriers of the hepatitis B virus, were pre-included. The determination of chronic hepatitis B was made using the standard diagnostic criteria of positive HBs antigen for more than 6 months. Liver stiffness measurement by TE was introduced in June 2005. The exclusion criteria were the following: chronic hepatitis C virus infection and all other causes of chronic liver disease; liver transplantation done before the follow-up period; missing FT or viral load assessment. The baseline data were prospectively collected. For non-cirrhotic patients, follow-up consultations and non-invasive biomarkers were scheduled a minimum of every 2 years. Patients with cirrhosis were followed and treated according to guidelines from the European Association for the Study of Liver (EASL). Mortality and cause of death were assessed using the centralized French mortality office. FT measurements were done on fresh serum, performed blinded to the clinical data and according to the recommended pre-analytic and analytic methods. FT scores range from zero to 1.00. Patients without a high-risk profile of false positives/negatives were considered non-reliable.[19] TE was performed according to published recommendations using the M probe of FibroScan®. The results were expressed in kilopascals (kPa). Only procedures with at least 10

9 validated measurements, a greater than 60% success rate and an interquartile range <30% of the median, were considered reliable.[20] The cohorts were designed to prospectively assess the first severe clinical event (F4.3), defined as the occurrence of liver decompensation (including variceal bleeding; grade 2 or higher hepatic encephalopathy; jaundice (total bilirubin >50 µmol/L); ascites requiring therapeutic paracentesis and/or additional therapy) or the development of HCC. Death, whether liver-related or not, and liver transplantation were also recorded.(Figure 1) The cohorts were not specifically designed to prospectively assess the occurrence of varices (F4.2) with systematic endoscopy. The decision to perform endoscopy was left to each physician, and it was usually prescribed every 2 years for patients with presumed cirrhosis, using FT or TE, or progression of biomarkers to stage F3. Varices were recorded as being present or absent at endoscopy, without grading. The retrospective validation of biomarker' cutoffs, was performed in all patients (P1). For the prospective validation, the second group consisted only of patients without previous complications (P2). In order to assess the prognostic performance of repeated biomarkers and the impact of treatment on fibrosis progression, the third group consisted only of patients without complications at baseline and with paired FT (P3). We previously suggest a definition of inactive carrier based on FibroTest/ActiTest (F0/A0: FT≤0.27 and AT≤0.29).[13] This definition was compared to a "standard" ALTbased definition of inactive carrier, patients with repeated ALT≤40 IU/L (3 times), derived from guidelines.[21,22]SupplementaryFileS2

10 Statistical methods The incidence of events was calculated using time-to-event methods, the KaplanMeier estimate and the cumulative hazard function. Comparisons used the log-rank test and proportional hazard regression multivariate analysis standardized on treatment and the prognostic area under the ROC curves.[17,18] The predictive values of FT and TE were assessed semi-quantitatively and quantitatively. Predetermined cutoffs were used (SupplementaryFileS1).

For the new

cirrhotic stages, the previously suggested "5-year mortality cutoffs" (low F4.1, intermediate F4.2, high risk F4.3) were predetermined:[18,19] for FT, 0.74, 0.85, 0.95 and for TE, 12.5, 20 and 50kPa respectively. Survival rates or complications rates were calculated from the first FT or TE to the event. Fibrosis progression rates were calculated using hazard function (HR). To avoid overestimation of the incidence of complications, only events occurring at least 6 months after baseline inclusion were taken into account. Responders were treated patients with sustained viral load <100.000 for at least one year at the end of follow-up and non-responders (NR) the others; non-treated (NT) were patients never been treated during the entire follow-up. The multivariate analyses included the following factors: treatment response, baseline viral load, HBeAg status, BMI and ALT. The fibrosis regression rate (FRR) or fibrosis progression rate (FPR) were defined as the percentage of patients who had achieved a significant (0.20 for FT and 4kPa for TE) decrease or increase, respectively, of fibrosis during follow-up.[23] The progression to cirrhosis or the regression of cirrhosis as presumed by FT, were graphically detailed in order to better describe the changes in the number of cirrhotic patients during follow-up. NCSS8.0 software was used.[24]

11 Results Patient characteristics (Table1) A total of 1574 patients were pre-included, and 140 were excluded; 1434 were included in P1, 1312 in P2, and 898 in P3.(Figure 1) Compared with the Bordeaux cohort, the Paris cohort had half the number of inactive anti-HBe patients, more patients with complications, fewer Caucasians, more repeated FT, fewer TE, more treated patients and a longer follow-up.(SupplementaryFileS3) FT performance In P1, the prevalence of varices increased from 0.1% in patients with F0 (FT≤0.27) to 46.5% in patients with F4.2 (FT >0.85 and ≤0.95)(Table 2) (SupplementaryFileS4) The prevalence of severe complications increased (P<0.0001) from 0.3% in F0 patients to 91.3% in F4.3 patients (FT>0.95). The FT AUROCs for the diagnosis of varices/liver failure/HCC varied from 0.90 to 0.93. In P2, varices occurred in 14 patients (1.7% incidence)(Table 1 and Table 3). FT was predictive of the occurrence of varices [AUROC=0.83(0.63-0.93;P<0.0001)]. The probability of varices occurrence increased from 0.7%(0.0-1.9) in 736 F0 patients to 44.0%(0-91.0) in 19 patients with FT>0.85 (F4.2). In patients with FT>0.74, the probability of varices (n=11) was 1.2% (0.4-2.0). Severe complications occurred in 25 patients (3.7% incidence), mostly HCC (n=21). The total number of liver-related events was 37 (10.2%), including 6 transplantations (0.6%) and 8 liver-related deaths (0.7%). Twenty-three severe complications occurred in the 582 responders [6.2%(3.2-9.1)], including 19 cases of HCC [5.8%(2.6-9.0)] and 10 varices [2.6%(0.8-4.4)]. Approximately 50 % of patients with cirrhosis presumed during the followup using FT, TE or biopsy underwent the 2 year-interval check up by endoscopy and approximately 90% had at least one endoscopy.

12 In P2 FT was predictive of severe complications [AUROC=0.79(0.76-0.82);P<0.0001], including HCC [AUROC=0.84(0.80-0.87);P<0.0001]. According to the predetermined FT cutoffs, the probability of severe complications increased from 0.5%(0.0-1.1) in 737 F0 to 41.5%(6.8-76.3) in 18 patients with FT≤0.95 and >0.85. Only one patient NT, had a baseline FT>0.95. This 40-year-old male discontinued high alcohol consumption (140 g per day) and had a rapid improvement of fibrosis, with a decrease in FT from 0.97 to 0.52 and in TE from 17.2 to 5.4 kPa. When he was analyzed together with the 18 patients with FT>0.85 and ≤0.95, there were significant differences between the F4.2/F4.3 categories for all events versus F2 (P=0.01), F1 (P<0.0001) and F0 (P<0.0001).(Table 3) In P2, using multivariate analysis adjusted for cofactors, the baseline FT was independently associated with all events [RR=169(52-544),P<0.0001], liver-related events [RR=187(41-852),P<0.0001] and HCC [RR=183(24-1430),P<0.0001]. HBeAg, treatment response, and baseline viral load were not significantly associated with any events in any models. In P3, using multivariate analysis adjusted for treatment effect, it was possible to assess the additive predictive value of combining baseline FT and last FT as markers of fibrosis dynamics. Baseline FT [risk ratio RR=17(2-143),P=0.009] and last FT [(RR=35(3389) P=0.004] were independently associated with all events [n=37(14.0%; 7.7-20.3)]. The corresponding prognostic 0.90)P<0.0001

and

for

AUROCs were for baseline combining

baseline

and

FT; AUROC=0.84 (0.76last

FT

0.87

(0.80-0.92)

P<0.0001.(SupplementaryFileS5) The "FibroTest/ActiTest" definition of an inactive carrier was validated twice. The first validation was in the same Paris cohort but using a longer follow-up (year 10 vs. year 4). This cohort showed no complications and an incidence of 3/163 (1.8%) advanced fibrosis without cirrhosis. The second validation was in the Bordeaux cohort, which showed no

13 complications and an incidence of 5/195 (2.6%) advanced fibrosis without cirrhosis.(Figure 2). In comparison, when the "ALT-based" definition was used, more fibrosis progression were observed: 16/181 (8.8%) advanced fibrosis in the Paris and 15/228 (6.6%) in the Bordeaux cohort, (P=0.004 and P=0.05, respectively).(SupplementaryFileS2) None of the patients who developed cirrhosis was classified as an inactive carrier by the "FibroTest/ActiTest" definition, versus 14 out of 27 by "ALT-based" definition. Of the 43 patients with cirrhosis at baseline, 20 had ALT≤ 40 IU/L. TE performance (Table 2, SupplementaryFileS4) In P1 the prevalence of severe complications increased (P<0.0001) from 0.3% in patients with F0 (TE≤5kPa) to 88.9% in patients with F4.3 (>50kPa). The prevalence of varices increased from 0.9% in patients with F0 (TE≤5kPa) to 55.6% in patients with F4.2 (FT>20kPa). The AUROCs of TE for the diagnosis of complications varied from 0.88 to 0.95 when estimated in all patients. In P3 severe complications occurred in 4 patients, all with HCC [1.4%(0-3.3) incidence at 7 years]; death or transplantation occurred in 8 [10.1%(0-25.5)]; and varices in 5 patients [1.3%(0-2.7)]. TE was predictive of all events [n=14;AUROC=0.84(0.700.92);P<0.0001] and liver-related events [n=8;AUROC=0.82(0.72-0.89);P<0.0001], without difference with last TE or baseline and last FT. Multivariate analyses could not be performed, as the number of events was too small. Fibrosis progression and regression rates (Figure 3) Of the 138 responders with advanced fibrosis at baseline, only 31%(15-47%) had fibrosis regression. The only significant difference observed was between R and NT patients, but when the patient who discontinued alcohol consumption was excluded, there was no longer a difference in fibrosis regression rates (Log-rank=3.5; P=0.06).

14 Of the 855 patients without cirrhosis at baseline, 27 (26 males) had progressed to cirrhosis (19 R, 4 NR, 4 NT)(Figure 4A). Of the 43 patients with cirrhosis at baseline, 16 had regressed to a non-cirrhotic stage (Figure 4B). Therefore at the end of follow-up, there were 54 patients with cirrhosis, i.e. 12 more than the 42 at baseline. A total of 15 cases of HCC (14 R) occurred in the 898 patients with paired FT; 13 out of 15 had advanced fibrosis (minimum of F2) at baseline, and 14 out of 15 (93%) had fibrosis progression.(Figure 4C) The only patient without fibrosis progression was a 54-year-old subSaharan female without advanced fibrosis at baseline (Stage F1) but with morbid obesity (body mass index=37).

15 Discussion The present study, with larger sample and longer follow-up than previous publications, validates the performance of both FT and TE for predicting the occurrence of varices and HCC. FT permitted to assess the treatment impact on fibrosis changes, and permitted a simpler definition of an uncomplicated cirrhosis stage ("F4.1") as well as a validation of a simple inactive carrier stage ("F0A0") using the predetermined cutoffs. This prognostic study was possible as FT and TE had sufficient diagnostic performances to replace liver biopsy for estimating advanced fibrosis and cirrhosis.[6] Diagnosis of varices at baseline Retrospectively, FT and TE were highly associated with a history of varices or the presence of varices at baseline (Table 2). There was a clear cutoff for varices prevalence with F4.1. AUROCs varied according to the spectrum effect from 0.63 in patients with cirrhosis to 0.95 in all patients, similarly to those of external studies.[25,26] Prediction of varices occurrence Validation studies of biomarkers are difficult to carry out due to the low incidence (78% per year) of varices and the burden of repeated endoscopies.[4,27,28] No prospective study has assessed the value of TE for predicting varices occurrence in CHB. In chronic hepatitis C (CHC), FT had showed a significant level of performance despite a very low incidence of varices(4%).[17] The lack of a systematic diagnosis may underestimate the incidence of varices. We recorded the results of endoscopies performed by physicians in charge of patients, usually at baseline when severe fibrosis or cirrhosis were suspected, and then every 2 years in more than 50% of cases. We acknowledge that we did not prospectively record the size of varices, red signs, or variability between observers and centers, and no assessment of the hepatic venous gradient (HVPG) was performed. The presence of varices is subject to significant observer

16 variation, and the utility of a stage F4.2 is questionable.[3] FT and TE performance were not compared to HVPG, which is the best predictor of varices development and which also has predictive value for decompensation.[3] However, the main advantages of FT and TE are the lower cost and the absence of adverse events. Furthermore, FT had higher applicability, both for failure and reliability, and had lower interobserver variability compared with TE and HVPG.[6] The number of varices that occurred after TE was too small to validate prospectively the cutoffs. Prediction of severe complications We observed that both FT and TE accurately predicted the occurrence of complications mostly HCC. The higher proportion of HCC compared with retrospective estimates could be explained by a "tertiary center effect", with a high rate of patients admitted to the intensive care unit for bleeding and liver failure. Previous studies have already suggested these performance values for FT and TE but with shorter follow-up.[10,12,14,15,25,26,27,28] This is therefore the first validation of FT performance for predicting the occurrence of HCC at 10 years, as well as overall survival and liver-related death, independent of treatment effect, viral load, HBeAg status, and ALT activity. One limitation of the study is that we did not compare FT and TE to other fibrosis biomarkers or quantitative liver function tests. However, FT is the most validated biomarkers,[7,8] with higher prognostic performance than non-patented biomarkers such as APRI and FIB4.[14] The level of performance of FT was at least similar to other biomarkers combining viral load.[29] Another limitation is that due to the later availability of TE only 60% of patients had both FT and TE. In P1, the predetermined FT cutoffs were validated in cirrhotic patients for critical clinical events: 0.74 for F4.1; 0.85 for F4.2; and >0.95 for F4.3. In P2, one limitation was the

17 absence of significant difference in the incidence of events according to the predetermined cutoffs. We acknowledge a lack of power, partly explained by the exclusion of patients with complications in the first 6 months to prevent the tertiary center effect (patients hospitalized for previous ongoing complications). Pending further validation on larger cohorts, the results suggested however that FT could be a robust biomarker for an earlier selection of candidates for liver transplantation. For TE, the number of events was too small, and the predetermined cutoffs for TE cannot be validated. We acknowledge that many more patients and events are needed to demonstrate the benefits of repeating or combining such biomarkers as observed in CHC.[17] There was a small but insignificant increase in the performances for repeated FT or for the combination of baseline or repeated FT and TE versus baseline values alone. However 93% of patients who developed HCC had fibrosis progression as presumed by FT.(Figures4C) These observations strongly support the use of such fibrosis biomarkers for patient monitoring. Regression and progression of fibrosis and cirrhosis A disappointing result was the slow regression of advanced fibrosis among the patients with virological response, which was only 30% at 10 years, even lower than the 50% observed in CHC responder patients.[18] Our results using FT are also less optimistic than those observed in randomized trials using biopsy. In responders, despite longer follow-up, we observed 14/36(39%) of cirrhosis regression compared to 71/96(71%) at year 5 in patients treated with tenofovir.[30] Similarly, the progression to cirrhosis was also higher [19/420(5%) vs. 3/252(1%)]. According to the published patient characteristics, there were no obvious differences in fibrosis risk factors. The "extreme case" NT we observed, with cirrhosis regression after stopping high alcohol consumption, illustrated the influence of co-factors in patients with chronic hepatitis C. Because the number of patients with advanced fibrosis and NT was (logically) small,

18 [21(2.3%) cases out of 899] only one case had an important weight on fibrosis dynamic modeling. More patients will be needed to estimate the relative role of alcohol and metabolic factors. Differences in fibrosis progression/regression could be also due to false positives/negatives for both FT and biopsy. This estimated risk is less than 2% for FT,[19] but much higher for biopsy due to sampling error, as observed in paired longitudinal studies.[10,11,23] It is also reassuring that 11 years ago, using repeated biopsies in responders, cirrhosis regression was observed in 3/11 (27%) patients and progression to cirrhosis in 1/52 (2%) in non-cirrhotics at 3 years.[31] The present results using FT at 10 years were similar in responders: with 16/46 (35%) regression in cirrhotics and 19/436 (4%) progression to cirrhosis. Validation of FT-AT for defining inactive carriers The validation of FT-AT as a robust means of defining inactive carriers with better negative predictive value than ALT could simplify guidelines, and improve decisions regarding management.[6,21]. The 2012 EASL guidelines could be simplified. When such biomarkers of fibrosis and activity are used, the utility of repeating viral load or differentiating patients with or without HBeAg is questionable. Furthermore, as fibrosis stage is both an estimate of severity and a memory of previous flares impact, the interest of repeating ALT measurements is also questionable. In Conclusion, predetermined cutoffs of FT can be used to rank the severity of chronic hepatitis B from inactive carrier status to three stages in cirrhotic patients with increasing morbidity and mortality.

19 References [1]. Fattovich G, Bortolotti F, Donato F. Natural history of chronic hepatitis B: special emphasis on disease progression and prognostic factors. J Hepatol. 2008;48:335-52. [2]. Goodman ZD. Grading and staging systems for inflammation and fibrosis in chronic liver diseases. J Hepatol. 2007;47:598-607. [3]. Garcia-Tsao G, Friedman S, Iredale J, Pinzani M. Now there are many (stages) where before there was one: In search of a pathophysiological classification of cirrhosis. Hepatology. 2010;51:1445-9. [4]. Myers RP, Tainturier MH, Ratziu V, et al. Prediction of liver histological lesions with biochemical markers in patients with chronic hepatitis B. J Hepatol. 2003;39:222-30. [5].

Foucher J, Chanteloup E, Vergniol J, et al. Diagnosis of cirrhosis by transient

elastography (FibroScan): a prospective study. Gut. 2006;55:403-8. [6]. Castera L. Hepatitis B: are non-invasive markers of liver fibrosis reliable? Liver Int. 2014;34 Suppl 1:91-6. [7]. Salkic NN, Jovanovic P, Hauser G, Brcic M. Fibrosure for Significant Liver Fibrosis and Cirrhosis in Chronic Hepatitis B: A Meta-Analysis. Am J Gastroenterol. 2014 [8]. Poynard T, Ngo Y, Munteanu M, Thabut D, Ratziu V. Noninvasive Markers of Hepatic Fibrosis in Chronic Hepatitis B. Curr Hepat Rep. 2011;10:87-97. [9]. Chon YE, Choi EH, Song KJ, et al. Performance of transient elastography for the staging of liver fibrosis in patients with chronic hepatitis B: a meta-analysis. PLoS One. 2012;7:e44930. [10]. Ngo Y, Benhamou Y, Thibault V, et al. An accurate definition of the status of inactive hepatitis B virus carrier by a combination of biomarkers (FibroTest-ActiTest) and viral load. PLoS One. 2008;3:e2573.

20 [11]. Poynard T, Ngo Y, Munteanu M, et al. Biomarkers of liver injury for hepatitis clinical trials: a meta-analysis of longitudinal studies. Antivir Ther. 2010;15:617-31. [12]. Castéra L, Bernard PH, Le Bail B, et al. Transient elastography and biomarkers for liver fibrosis assessment and follow-up of inactive hepatitis B carriers. Aliment Pharmacol Ther. 2011;33:455-65. [13]. Poynard T, Munteanu M, Deckmyn O, et al. Validation of liver fibrosis biomarker (FibroTest) for assessing liver fibrosis progression: proof of concept and first application in a large population. J Hepatol. 2012;57:541-8. [14]. Poynard T, Ngo Y, Perazzo H, et al. Prognostic value of liver fibrosis biomarkers: a meta-analysis. Gastroenterol Hepatol (N Y). 2011;7:445-54. [15]. de Lédinghen V, Vergniol J, Barthe C, et al. Non-invasive tests for fibrosis and liver stiffness predict 5-year survival of patients chronically infected with hepatitis B virus. Aliment Pharmacol Ther. 2013;37:979-88. [16]. Lai CL, Yuen MF. Prevention of hepatitis B virus-related hepatocellular carcinoma with antiviral therapy. Hepatology 2013;57:399–408. [17]. Poynard T, Vergniol J, Ngo Y, et al. Staging chronic hepatitis C in seven categories using fibrosis biomarker (FibroTest™) and transient elastography (FibroScan®). J Hepatol 2013;60:706–714. [18]. Poynard T, Moussalli J, Munteanu M, et al. Slow regression of liver fibrosis presumed by repeated biomarkers after virological cure in patients with chronic hepatitis C. J Hepatol. 2013;59:675-83. [19]. Poynard T, Munteanu M, Deckmyn O, et al. Applicability and precautions of use of liver injury biomarker FibroTest. A reappraisal at 7 years of age. BMC Gastroenterol. 2011;11:39.

21 [20]. Castéra L, Foucher J, Bernard PH, et al. Pitfalls of liver stiffness measurement: a 5-year prospective study of 13,369 examinations. Hepatology. 2010;51:828-35. [21]. Yapali S, Talaat N, Lok AS. Management of hepatitis B: our practice and how it relates to the guidelines. Clin Gastroenterol Hepatol. 2014;12:16-26. [22]. Chao T, Lim JK , Ayoub W S, Nguyen LH, Nguyen MH. Systematic review with metaanalysis: the proportion of chronic hepatitis B patients with normal alanine transaminase guidet≤40 IU/L and significant hepatic fibrosis. Aliment Pharmacol Ther 2014; 39: 349–358. [23]. Poynard T, Lenaour G, Vaillant JC, et al. Liver biopsy analysis has a low level of performance for diagnosis of intermediate stages of fibrosis. Clin Gastroenterol Hepatol. 2012;10:657-63.e7. [24]. Hintze J. NCSS 8;NCSS, LLC, Kaysville, Utah, USA(2012) www.ncss.com [25]. Chon YE, Jung ES, Park JY, et al. The accuracy of noninvasive methods in predicting the development of hepatocellular carcinoma and hepatic decompensation in patients with chronic hepatitis B. J Clin Gastroenterol. 2012;46:518-25. [26]. Kim BK, Kim SU, Kim HS, et al. Prospective validation of FibroTest in comparison with liver stiffness for predicting liver fibrosis in Asian subjects with chronic hepatitis B. PLoS One. 2012;7(4):e35825. [27]. Shi KQ, Fan YC, Pan ZZ, et al. Transient elastography: a meta-analysis of diagnostic accuracy in evaluation of portal hypertension in chronic liver disease. Liver Int. 2013;33:6271. [28]. Wang JH, Chuah SK, Lu SN, et al. Transient elastography and simple blood markers in the diagnosis of esophageal varices for compensated patients with hepatitis B virus-related cirrhosis. J Gastroenterol Hepatol. 2012;27:1213-8. [29]. Wong GL, Chan HL, Chan HY, et al. Accuracy of risk scores for patients with chronic hepatitis B receiving entecavir treatment. Gastroenterology. 2013;144:933-44.

22 [30]. Marcellin P, Gane E, Buti M, et al. Regression of cirrhosis during treatment with tenofovir disoproxil fumarate for chronic hepatitis B: a 5-year open-label follow-up study. Lancet. 2013;381:468-75. [31]. Dienstag JL, Goldin RD, Heathcote EJ, Hann HW, Woessner M, Stephenson SL, Gardner S, Gray DF, Schiff ER. Histological outcome during long-term lamivudine therapy. Gastroenterology. 2003;124:105-17.

23 Table 1: Characteristics of patients included and incidence of complications

Preincluded Excluded Included Population P1 Characteristics at inclusion Gender male Age median yr (95%CI) Body Mass Index>27 HIV co-infection HBV characteristics Missing (HBeAg status unknown) HBeAg active HBeAg immunotolerant AntiHBe active AntiHBe inactive Caucasian origin Alcohol >30g/day Diabetes1 Biopsies baseline Reliable FibroTest baseline Stage presumed by FibroTest F0 (0-<=0.27) F1 (0.27-<=0.48) F2 (0.48-<=0.58) F3 (0.58-<=0.74) F4 All Cirrhosis F4.1 (0.74-<=0.85) F4.2 (0.85-<=0.95 F4.3 (>0.95) Elastography (TE) baseline Non reliable Reliable Stage presumed by TE reliable F0<=5 F1<=7.1 F2<=9.5 F3<=12.5 F4 All Cirrhosis F4.1(12.5-<=20) F4.2(20-<=50) F4.3(>50)

All 1574 140 1434 964 (67.2%) 39.8 (38.6-40.9) 252/1064 (23.7%) 37(2.6%) 1 402 (28.1%) 4 (0.3%) 625 (43.6%) 402 (28.1%) 515 (35.9%) 65/1254 (5.2%) 89/1264 (7.1%) 719/1434 (50.1%) 1434 739 (51.5%) 325 (22.6%) 107 (7.5%) 133 (7.5%) 131 (9.3%) 64 (4.5%) 43 (3.0%) 23 (1.7%) 911 67 (7.2%) 844 (92.8%) 347 (41.1%) 291 (34.5%) 97 (11.5%) 77 (9.1%) 32 (3.8%) 14 (1.7%) 9 (1.1%) 9 (1.1%)

24 All History of complications Varices only Severe complications Both Population for estimating complications' incidence No history of complication P2 Treatment during follow-up Response No response No treatment Incidence complications Varices (F4.2 Severe complications (F4.3) Ascites Primary liver cancer Bleeding Jaundice (Bili>50umol/L) or encephalopathy or Child C Death Liver transplantation Liver related death Liver related events All events: Patients with paired estimates P3 Repeated reliable FibroTest Duration between FT median Paired reliable FT TE Duration between paired TE

1

Data missing in 20 patients

122/1434 (8.5%) 44 44 34

1312 582 (44.3%) 132 (10.1%) 598 (45.6%) 14 (1.7%;0.6-2.8) 25 (3.7%;1.8-5.7) 7 (0.6%;0.1-1.0) 21 (3.7%;1.5-5.8) 2 (0.1%;0-0.2) 11 (0.9% ;0.4-1.4) 28 (2.6%;1.6-3.7) 6 (0.6 %;0.1-1.1) 8 (0.7 %;0.2-1.3) 37(10.2%;5.1-15.3) 57(13.7%;7.9-19.5) 898 5.6 (5.3-5.9) 415 4.0 (3.6-4.4)

25 Table 2: Retrospective validation of FibroTest and transient elastography cutoffs using history of cirrhosis complications n

Severe complications

Primary liver cancer

Liver failure

Varices

1434

78 (5.4%)

34 (2.4%)

55 (3.8%)

78 (5.4%)

FibroTest All stages 1 FT stages F0 <=0.27 F1 >0.27 <=0.48 F2 >0.48 <=0.58 F3 >0.58 <=0.74 F4.1 >0.74 <= 0.85 F4.2 >0.85 <=0.95 F4.3 >0.95 Test for increasing trend: Z value (P-value) AUROC all

739 325 107 133 64 43 23

2 (0.3%) 7 (2.2%) 4 (3.7%) 15 (11.2%) 11 (17.2%) 18 (41.9%) 21 (91.3%) -13.3 (<0.0001)

0 (0%) 1 (0.3%) 0 (0%) 12 (9.0%) 6 (9.4%) 8 (18.6%) 7 (29.2%) -9.3 (<0.0001)

2 (0.3%) 6 (1.9%) 4 (3.7%) 4 (3.0%) 7 (10.9%) 13 (30.2%) 19 (82.6%) -11.0 (<0.0001)

1 (0.1%) 10 (3.1%) 5 (4.7%) 17 (12.7%) 14 (21.9%) 20 (46.5%) 11 (47.8%) -12.9 (<0.0001)

1434

AUROC in cirrhosis

130

0.92 (0.88-0.94) <0.0001 0.81 (0.72-0.87) <0.0001

0.93 (0.91-0.95) <0.0001 0.67 (0.52-0.78) =0.02

0.91 (0.85-0.94) <0.0001 0.83 (0.73-0.89) <0.0001

0.90 (0.88-0.93) <0.0001 0.63 (0.52-0.72) =0.008

844

45 (4.7%) -8.4 (<0.0001)

16 (1.9%) -5.8 (<0.0001)

22 (2.6%) -6.2 (<0.0001)

17 (2.0%) -6.7 (<0.0001)

844

0.88 (0.82-0.92) (<0.0001) 0.79 (0.56-0.91) =0.0004

0.92 (0.85-0.96) (<0.0001) 0.66 (0.36-0.83) =0.19

0.88 (0.78-0.94) (<0.0001) 0.79 (0.55-0.91) =0.0007

0.95 (0.84-0.98) (<0.0001) 0.86 (0.66-0.95) <0.0001

Transient elastography All stages Test for increasing trend: Z value (P value) AUROC all AUROC in cirrhosis

1

32

FT= FibroTest; predetermined stages

26 Table 3: 10-year prospective validation of biomarkers' cutoffs n

Severe complication

Primary liver cancer

Death or transplantation

Varices

FibroTest reliable All stages

1312

25 (3.7%; 1.8-5.7)

21 (3.7%;1.5-5.8)

34 (3.2%;2.14.3)

14 (1.7%;0.62.8)

FibroTest predetermined stages F0 <=0.27 F1 >0.27 <=0.48

736 309

3 (0.5%;0-1.1) 1 (0.7%;0-2.1)

2 (0.4%;0-0.9) 1 (0.7%;0-2.1)

F2 >0.48 <=0.58

99

4 (6.0%;0.3-11.6)

3 (4.4%;0-9.4)

2 (0.7%;0-1.9%) 1 (0.4%;(01.3%) 0 (0%)

F3 >0.58 <=0.74

104

8 (12.2%;3.4-21.1)

F4.1 >0.74 <= 0.85

45

5 (13.6%;2.5-24.8)

1

19

4 (41.0%;6.2-76.0)

7 (13.6%;1.625.6) 4 (11.1%;0.821.5) 4 (44.7%;6.582.9)

4 (0.7%;0-1.5) 7 (2.9%;0.65.1) 4 (5.1%;0.110.1) 8 (9.7%;3.016.3) 7 (16.3%;5.227.4) 4 (22.3%;2.841.7)

21 (8.1%;017.4)

6 (1.2%;0.12.2)

F4.2 >0.85 <=1

Transient Elastography reliable All stages

1

5 (5.4%;0.810.1) 3 (6.9%;0-4.4) 3 (44.0%;091.1)

797 7893

9 (1.9%;0.3-3.4)

8 (1.7%;0.2-3.2)

Only one patient had a severe risk (FT>0.95, no event). Significant differences were observed between F4.2 category vs F2 (P=0.01), F1 (P<0.0001) and F0 (P<0.0001). 3 8 subjects with severe complications occurring before TE were excluded

27

Legends of Figures Figure 1: Chart flow of the study populations Figure 2: Inactive carriers defined by normal FibroTest-ActiTest Figure 3: Fibrosis regression (FR) or progression (FP) during follow-up. (A) Advanced fibrosis (AdF) at baseline. FR not different in R vs. NR (LogRank=1.2;P=0.28) but lower vs. NT (Log-Rank=6.7;P=0.01). (B) AdF at baseline. FP not different in R vs. NT (Log-Rank=0.5;P=0.48) or vs. NR (LogRank=0.8;P=0.36). (C) No-AdF at baseline. FR not different in R vs. NR (Log-Rank=0.1;P=0.73) and NT (LogRank=1.5;P=0.22). (D) No-AdF at baseline. FP not different in R vs. NR (Log-Rank=1.9;P=0.16) and NT (LogRank=0.9;P=0.33). (E) All patients. FR not different in R vs. NR (Log-Rank=0.9;P=0.76) and NT (LogRank=2.4;P=0.12). (F) All patients. FP not different in R vs. NR (Log-Rank=0.8;P=0.37) and lower vs. NT (Log-Rank=2.6;P=0.11) The number of dropouts increased between 6 and 8 years because Fibrotest was performed two years later in the Bordeaux cohorts than in the Paris cohort (Supplementary Files, Table1) Figure 4: Progression to cirrhosis, regression of cirrhosis and incidence of hepatocellular carcinoma according to FibroTest changes. The horizontal orange dotted lines correspond to the standard cutoffs for FibroTest F1(0.27), F2(0.48) and F4(0.74); horizontal purple lines correspond to the predetermined intermediate risk cutoff (0.85) and the high risk (0.95). (A): Progression to cirrhosis B): Regression of cirrhosis (C): Occurrence of hepatocellular carcinoma according to FibroTest progression

Figure

Paris Cohort n= 931

n=1574 Pre-included

Bordeaux Cohort n=643

135 excluded (28 HCV; 102 delta, 12 reactivation, 2 auto immune; 1 not reliable FibroTest)

5 excluded (4 not reliable FibroTest, 1 reactivation)

638 At least one FT

796 at least one FT

92 complications or varices before FibroTest (28 with both)

704 no complication at baseline

532 paired FT 70 paired TE

lundi 3 février 14

P1 n=1434 Retrospective cutoffs validation

P2 n=1312 Prospective cutoffs validation Inactive carrier definition P3 n=898 Paired FT Fibrosis dynamics Progression to cirrhosis n=30/856 Regression of cirrhosis n=17/42

30 complications or varices before FibroTest (6 with both)

608 no complication at baseline

366 paired FT 352 paired TE

Prospective population n=1312

Figure 2

954 Excluded (Treated n= 690, Untreated FT/AT >0.27/0.29 n=230, HIV pos n=1, Missing baseline ViralLoad n=68)

Paris Cohort n=163

Viral Load < 100,000 n=154

FibroTest <=0.27 and ActiTest<=0.29 n=358 untreated included

Viral Load >=100,000 n=9

Inactive Carrier 147 HBeAg-neg 7 HBeAg-pos

Immuno-tolerant 2 HBeAg-neg 7 HBeAg-pos

Baseline: 0 complication

FT: 154 F0/ 0 F1/ 0 F2/ 0 F3/ 0 F4 TE: 100 missing/ 27 F0/ 24 F1/ 2 F2/ 1 F3/ 0 F4

Baseline: 0 complication FT: 9 F0/ 0 F1/ 0 F2/ 0 F3/ 0 F4 TE: 8 missing/ 1 F0/ 0 F1/ 0 F2/ 0 F3/ 0 F4

FU: 0 Complications FT: 42 missing/ 27 F0/ 24 F1/ 2 F2/ 1 F3/ 0 F4 TE: 135 missing/ 6 F0/ 7 F1/ 6 F2/ 0 F3/ 0 F4

FU: 0 complication FT: 4 missing/ 4 F0/ 0 F1/ 1 F2/ 0 F3/ 0 F4 TE: 8 missing/ 0 F0/ 1 F1/ 0 F2/ 0 F3/ 0 F4

lundi 3 février 14

Bordeaux Cohort n=195

Viral Load < 100,000 n=182

Inactive Carrier 167 HBeAg-neg 15 HBeAg-pos Baseline: 0 complication

FT: 182 F0/ 0 F1/ 0 F2/ 0 F3/ 0 F4 TE: 3 missing/ 113 F0/ 50 F1/ 11 F2/ 5 F3/ 0 F4

FU: 0 Complications FT: 72 missing/ 80 F0/ 25 F1/ 4 F2/ 1 F3/ 0 F4 TE: 121 F0/ 49 F1/ 10 F2/ 0 F3/ 2 F4

Viral Load >=100,000 n=13

Immuno-tolerant 1 HBeAg-neg 12 HBeAg-pos Baseline: 0 complication FT: 13 F0/ 0 F1/ 0 F2/ 0 F3/0 F4 TE: 10 F0/ 3 F1/ 0 F2/ 0 F3/ 0 F4

FU: 0 Complication FT: 4 missing/ 7 F0/ 2 F1/ 0 F2/ 0 F3/ 0 F4 TE: 10 F0/ 3 F1/ 0 F2/ 0 F3/ 0 F4

Figure 3

A

Presumed advanced fibrosis at baseline, n = 184

B

Fibrosis regression presumed by FibroTest™

Fibrosis progression presumed by FibroTest™

NR NT R

Presumed non-advanced fibrosis at baseline, n =715

C

D

Fibrosis regression presumed by FibroTest™

Fibrosis progression presumed by FibroTest™

NR NT R

All patients, n = 899

E

F

Fibrosis regression presumed by FibroTest™

Fibrosis progression presumed by FibroTest™

NR NT R

All patients

Responders (R) Non-responders (NR) Non-treated (NT)

Number of patients at risk Years Baseline 2 4 6 8 455 411 325 240 129 94 78 65 43 21 350 273 195 124 38

10 26 5 5

Figure 4A

Figure 4B

Figure 4C