Rates and determinants of hepatitis B ‘e’ antigen and hepatitis B surface antigen seroclearance during longterm follow-up of patients coinfected with HIV and hepatitis B virus Anders Boyda, Joe¨l Gozlanb,c, Patrick Miailhesd,e, Caroline Lascoux-Combef, Manuela Se´bire-Le Camg, Hayette Rougierg, Fabien Zoulimd,h,i, Pierre-Marie Girardg,j and Karine Lacombeg,j Objective: This study determines rates and risk factors associated with hepatitis B surface antigen (HBsAg) and hepatitis B ‘e’ antigen (HBeAg) seroclearance, two important prognostic indicators during infection with hepatitis B virus (HBV), in a large contemporary cohort of patients coinfected with HIV–HBV. Design: Prospective cohort study of predominately antiretroviral therapy (ART) experienced, coinfected patients. Methods: Participants enrolled in the French HIV–HBV Cohort had complete HBV serological battery conducted at inclusion and every yearly visit. Piecewise-exponential survival models were used to determine risk factors associated with seroclearance. Results: A total of 290 patients, of whom 151 (52.1%) were HBeAg positive, had been followed for a median 7.4 years (interquartile range [IQR] ¼ 3.1–8.0). Tenofovir (TDF) containing ART became increasingly more frequent, as rates of undetectable HBV-DNA increased accordingly (at baseline ¼ 39.3%, end of follow-up ¼ 91.0%). In HBeAgpositive patients, 60 of 151 had HBeAg seroclearance (cumulative 46.4% at end of follow-up) after a median 3.0 years (IQR ¼ 2.0–4.9). Overall, 17 of 290 patients had HBsAg seroclearance (cumulative 7.4% at end of follow-up) after a median 4.6 years (IQR ¼ 2.1–7.2). Lower levels of time-averaged cumulative HBV-DNA were significantly associated with both HBeAg and HBsAg seroclearance (P < 0.001 and P ¼ 0.01, respectively). In post hoc analysis among patients initiating TDF, incidence rates of HBeAg seroclearance peaked at year 4 of TDF treatment (13.1/100 person-years), whereas a steep drop in HBsAg seroclearance incidence rates occurred after year 3 (at year 3 ¼ 1.2/100 person-years versus thereafter ¼ 0.6/100 person-years). Conclusion: HBsAg seroclearance and, to a lesser extent, HBeAg seroclearance remain difficult endpoints for patients coinfected with HIV–HBV to achieve. HBV-DNA

a INSERM, UMR_S1136, Institut Pierre Louis d’Epide´miologie et de Sante´ Publique, bLaboratoire de Virologie, Hoˆpital SaintAntoine, AP-HP, cUPMC UMRS CR7, INSERM U1135 CIMI, Paris, dCentre de Recherche sur le Cancer de Lyon, Equipes 15 et 16, INSERM, Unite´ 1052, CNRS, UMR 5286, eService des Maladies Infectieuses et Tropicales, Hoˆpital de la Croix-Rousse, Hospices Civils de Lyon, Lyon, fService des Maladies Infectieuses et Tropicales, Hoˆpital Saint-Louis, APHP, gDepartment of Infectious and Tropical Diseases, Saint-Antoine Hospital, AP-HP, Paris, hHepatology Unit, Hospices Civils de Lyon, iUniversite´ de Lyon, Lyon, and jSorbonne Universite´s, UPMC Univ Paris 06, UMR_S 1136, Institut Pierre Louis d’Epide´miologie et de Sante´ Publique, Paris, France. Correspondence to Dr Anders Boyd, Services des Maladies Infectieuses et Tropicales, Hoˆpital Saint-Antoine, 184 rue du Fbg St Antoine, 75571 Paris Cedex 12, France. Tel: +33 1 71 97 05 17; fax: +33 1 49 28 21 49; e-mail: [email protected] Received: 9 February 2015; revised: 19 June 2015; accepted: 24 June 2015.

DOI:10.1097/QAD.0000000000000795

ISSN 0269-9370 Copyright Q 2015 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

1963

1964

AIDS

2015, Vol 29 No 15 suppression, associated with effective treatment, is strongly linked to seroclearance, but this mostly occurs within the first years of ART-containing highly potent anti-HBV Copyright ß 2015 Wolters Kluwer Health, Inc. All rights reserved. activity.

AIDS 2015, 29:1963–1973 Keywords: exponential survival models, hepatitis B virus, immunosuppression, tenofovir, viral persistence

Introduction

Patients and methods

With the advent of potent antiretroviral therapy (ART), incidence of AIDS-related death has substantially declined among patients infected with HIV, bringing way to new causes of morbidity and mortality [1]. Chronic infection with hepatitis B virus (HBV) has been chief among these recent comorbidities, remaining a constant concern over the past decade [2]. Despite the highly effective anti-HBV treatments currently available, recent data continue to demonstrate a strong association with HIV–HBV coinfection and increased risk of liver cirrhosis, hepatocellular carcinoma (HCC), liver-related death, and overall mortality [3–5].

Study design and visits Patients from the French HIV–HBV Cohort were included in the present study, as described previously [20]. Briefly, a total of 308 patients were recruited from seven centers located in Paris and Lyon, France, during May 2002 to May 2003. Inclusion criteria were HIV-positive serology confirmed by western blot and HBsAg-positive serology for at least 6 months. Patients were then prospectively followed every 3 months until the month 36 visit (2005–2006). This period is referred to herein as the first study period.

One of the most important prognostic criteria in chronic HBV infection is hepatitis B surface antigen (HBsAg) seroclearance, often followed by decreased viral activity and improved liver-related prognosis [6]. In hepatitis B ‘e’ antigen (HBeAg) positive patients, HBeAg seroclearance is also considered a major clinical endpoint toward resolving HBV infection. HBV-monoinfected cohorts have alluded to a variety of markers associated with HBsAg and HBeAg seroclearance, such as increased transaminase levels, undetectable HBV-DNA, and possibly decreased HBsAg quantification [7–9]. Unfortunately, no strong risk factors have yet been identified with seroclearance in treated patients coinfected with HIV–HBV [10–12]. Many previous studies on the topic include only retrospective follow-up, small numbers of patients, and outcomes specifically after initiating either tenofovir (TDF) or lamivudine (LAM) containing ART [11,13–17]. Another impediment is that many of the large cohorts of HIV-infected persons cannot examine risk factors associated with seroclearance, as they do not have regularly collected HBV virological and serological data [18,19]. The aim of this study was to quantify rates of HBeAg seroclearance and HBsAg seroclearance in patients coinfected with HIV–HBV prospectively followed for up to 10 years. We then intended to examine the determinants associated with these important clinical endpoints, paying particular attention to virological, immunological, and therapeutic parameters.

During March 2007 to March 2008, 274 patients completing follow-up roughly 12–24 months after the last study visit were invited to participate in an extension of the French HIV–HBV Cohort, referred to herein as the second study period. Of them, 185 (67.5%) agreed to further follow-up (Supplemental Digital Content [SDC] Fig. 1, http://links.lww.com/QAD/A742). Clinical characteristics at the very first inclusion visit did not substantially differ between participants versus nonparticipants of the second study period (SDC Table 1, http:// links.lww.com/QAD/A742). At the beginning of the second study period, liver fibrosis levels were determined for all participants as described in the SDC Methods, http://links.lww.com/QAD/A742. Adapted from concurrent recommendations [21], subsequent study visits were guided by the degree of fibrosis. Patients with METAVIR F3-F4 fibrosis were asked to continue study visits every 6 months and patients with METAVIR F0-F2 every 12 months. If a patient changed liver fibrosis groups during follow-up, the frequency of study visits were modified accordingly. For all patients, follow-up continued until the month 36 visit (2010–2011). All patients provided written informed consent to participate in both study periods and the protocol was approved by the Pitie´-Salpeˆtrie`re Hospital Ethics Committee (Paris, France) in the first study period and by the Hoˆpital Saint-Antoine Ethics Committee (Paris, France) in the second study period, in accordance with the Helsinki Declaration.

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Seroclearance in HIV–hepatitis B virus coinfection Boyd et al.

Quantification of hepatitis B virus parameters Plasma HBV-DNA viral load was quantified at baseline and every 6 months in the first study period and every 6–12 months in the second study period using a commercial PCR-based assay (COBASAmpliPrep/ COBASTaqMan, detection limit: 12 IU/ml; or COBASAmplicor HBV Monitor, detection limit: 60 IU/ml; Roche Diagnostics, Meylan, France). Data on HBV genetic variability are described elsewhere [22]. Alanine aminotransferase (ALT) levels were quantified using standard methods for every study visit. ALT levels were regrouped in relation to the upper limit normal (ULN), defined at 35 IU/ml: less than 1X ULN, 1–2X ULN, and more than 2X ULN. Qualitative HBsAg, HBeAg, anti-hepatitis Bs antibody (HBsAb), and anti-hepatitis B ‘e’ antibody (HBeAb) were detected at baseline and every yearly visit during the first study period and at every 6–12 months in the second study period using a commercial enzyme immunoassay (DiaSorin, Antony, France; Monolisa HBsAg, Bio-Rad, Marnes-la-Coquette, France; or Architect, Abbott Diagnostics, Rungis, France). HBeAg seroclearance was defined as any patient with HBeAg loss during follow-up. HBsAg seroclearance was similarly defined as HBsAg loss during follow-up.

Quantification of HIV-related parameters HIV-1 viral loads were measured using either a branchedDNA (b-DNA Quantiplex 3.0, detection limit: 50 copies/ml, Bayer Diagnostics, Cergy Pontoise, France) or real-time PCR technique (COBAS AmpliPrep/ COBAS TaqMan HIV-1 Test, detection limit: 40 copies/ml, Roche Molecular Systems, Meylan, France) at each study visit. CD4þ T-cell counts were quantified using standard measurements at each study visit, while nadir CD4þ cell count was obtained from patient records prior to inclusion in the first study period. Clinical data All antiretroviral and anti-HBV treatment, along with concomitant treatment for other indications, were obtained from patient medical files and verified by the treating physician. Any liver-related and nonliver-related clinical event was reported and confirmed with the treating physician during follow-up. Statistical analysis For all analysis, continuous follow-up was divided into 1year intervals, while updating all time-varying data at the start of each interval. HBeAg and HBsAg seroclearance were treated as separate endpoints, while only considering the first seroclearance event and not taking into account transitioning back to antigen-positive status. Thus, follow-up was censored after event, loss to followup, study termination, or death. Incidence rates of seroclearance were modeled using a piecewise-

exponential survival model [23]. Hazard ratios (HRs) and 95% confidence interval (CI) were then obtained for each time-constant and time-varying variable in univariable analysis. With a predictive objective in mind, a multivariable model was constructed in which variables with P < 0.1 from univariable analysis were included in a forward-stepwise manner, while removing any variable below the P value threshold upon adjustment. CD4þ change from the previous visit (lag CD4þ) was modeled in absolute values and percentage and cumulative copyyears of HIV-RNA and HBV-DNA viral load were averaged over follow-up time (copy-yearsTAVG). More information on the model construction and how time-varying exposures in HIV and HBV parameters were calculated can be found in the SDC Methods, http://links.lww.com/QAD/A742. All statistical analyses were performed using STATA software (v12.1; College Station, Texas, USA) and significance was determined using a P value < 0.05.

Results In the present study, we initially selected all 308 patients participating in the cohort. Of them, 18 were excluded for the following reasons: only one study visit (n ¼ 16), death after the first study visit (n ¼ 1), and incomplete information on HBV serology (n ¼ 1, death occurring shortly after last study visit). In total, 290 patients were included in the analysis on HBsAg seroclearance and 151 patients with HBeAg-positive serology at study inclusion were included in the analysis on HBeAg seroclearance.

Description of the study population Demographic and HIV/HBV-related characteristics of the study population at inclusion are given in Table 1. Of the 265 ART-experienced patients, 242 (91.3%) were actively undergoing ART at study inclusion. Of those on active ART, 219 (90.5%) were undergoing an anti-HBVcontaining ART-regimen at inclusion, with previous exposure to anti-HBV treatment lasting 53.8 months (IQR ¼ 25.3–71.8) for LAM, 8.9 months (IQR ¼ 4.7– 11.7) for adefovir, and 4.9 months (IQR ¼ 2.4–9.0) for TDF. Evolution of HIV and hepatitis B virus parameters during follow-up Patients were followed for a median 7.4 years (IQR ¼ 3.1–8.0), amounting to 1727.5 person-years of follow-up. The proportion of patients actively undergoing ART substantially increased over time (83.5% at inclusion to 100% at the end of follow-up). Consequently, improvements in both CD4þ cell counts (P for trend < 0.0001) and proportion of patients with undetectable HIV-RNA (P for trend < 0.0001) were observed up until the end of follow-up (Fig. 1a).

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

1965

2015, Vol 29 No 15 (a) 700

CD4 HIV-RNA

Demographic characteristics Male : female ratio (% male) Age (years)a Originating from zone of high HBV endemicity HIV-related characteristics Estimated duration of HIV infection (years)a AIDS-defining illness CD4þ cell count (cells/ml)a CD4þ nadir (cells/ml)a ART naive Duration of ART (years)a Detectable HIV-RNA (>50 copies/ml) HIV-RNA viral load (log10 copies/ml)a,b HBV characteristics Estimated duration of HBV infection (years)a Detectable HBV-DNA (>60 IU/ml) HBV-DNA viral load (log10 IU/ml)a,c HBeAg positive ALT > ULN (35 IU/ml) Concomitant HBV treatment LAM FTC ADV TDF Standard-IFN Pegylated-IFN HBV genotype [n ¼ 158] A B D E G A/D or A/G mixed Precore mutations W28M [n ¼ 149] T13S [n ¼ 150] L-Nucleoside-associated pol gene mutations [n ¼ 158]

244/46 (84.1) 40 (35–45) 75 (25.9) 9.8 (3.6–14.0) 78 400 212 25 5.6 131

(26.9) (268–554) (102–324) (8.6) (2.8–7.4) (45.3)

Median CD4+ cell count (/mm3)

n ¼ 290

1.00

IQR 95%CL

0.80

600

0.60 500 0.40 400 0.20 300

Proportion with undetectable HIV-RNA

Table 1. Description of the study population at inclusion.

0.00 0

1

2

3

287

277

243

4

5

6

7

8

181

174

164

74

Year n 290

3.90 (2.54–4.44)

186

(b) 1.0

6.3 (2.3–11.0) 176 5.27 151 166

(60.7) (3.18–6.88) (52.1) (58.5)

201 2 15 55 6 8

(69.3) (0.7) (5.2) (19.0) (2.1) (2.8)

98 1 16 17 17 9

(62.0) (0.6) (10.1) (10.8) (10.8) (5.7)

39 (26.2) 77 (51.3) 90 (57.0)

ADV, adefovir; ALT, alanine aminotransferase; ART, antiretroviral therapy; FTC, emtricitabine; HBV, hepatitis B virus; IFN, interferon; IQR, interquartile range; LAM, lamivudine; TDF, tenofovir; ULN, upper limit normal. All statistics are given as n (%), except for those designated with (a) wherein the median (IQR) is given instead. b Among patients with detectable HIV-RNA. cAmong patients with detectable HBV-DNA.

TDF with LAM/FTC-containing ART became the prevailing treatment over time (Fig. 1b, P for trend < 0.0001). Of the 227 patients initiating TDFcontaining ARTregimen at some point during follow-up, 199 (87.7%) switched to TDF from a less-potent antiHBV regimen and 28 (12.3%) initiated TDF without any previous anti-HBV exposure. Pegylated-interferon was administered in 32 patients at some point during followup, 30 (93.8%) of whom were concomitantly undergoing ART with an anti-HBV nucleoside/nucleotide analogue (NA). Accordingly, the percentage of patients with

LAM/FTC ADV

1.0

TDF HBV-VL

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0

Proportion with undetectable HBV-VL

AIDS

Proportion with anti-HBV treatment

1966

0.0 0

1

2

3

287

277

243

4

5

6

7

8

181

174

164

74

Year n 290

186

Fig. 1. Evolution of HIV, hepatitis B virus, and antiviral treatment against hepatitis B virus over follow-up. The number of patients continuing follow-up, as divided in yearly intervals, are provided at the bottom of each figure. In (a), evolution of mean (95% CI) proportion of patients with undetectable HIV-DNA and median (IQR) CD4þ T-cell counts are displayed for each year. In (b), the proportion of patients with undetectable HBV-DNA viral loads, alongside the proportion undergoing antiviral therapy containing LAM or FTC), TDF, and/or ADV are given for each year. ADV, adefovir; CI, confidence interval; FTC, emtricitabine; HBV, hepatitis B virus; IQR, interquartile range; LAM, lamivudine; TDF, tenofovir.

undetectable HBV-DNA increased dramatically by the end of follow-up (Fig. 1b, P for trend < 0.0001).

Rates of hepatitis B ‘e’ antigen seroclearance and seroconversion The cumulative proportion of HBeAg seroclearance and its incidence rate are given at each yearly interval in Fig. 2a. Among the 151 HBeAg-positive patients at study inclusion, 60 (39.7%) lost HBeAg status after a median 3.0 years (IQR ¼ 2.0–4.9) of follow-up (incidence rate ¼ 8.4/100 person-years). Of them, 11 (18.3%) had

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Seroclearance in HIV–hepatitis B virus coinfection Boyd et al. (a)

(b) 100.0 Incidence rate of HBsAg-loss (/100 person years)

Incidence rate of HBeAg-loss (/100 person years)

100.0 80.0 60.0 40.0 20.0 0.0

80.0 60.0 40.0 20.0 0.0

0

1

2

3

4

5

6

7

0

8

1

2

0.80 0.60 0.40 0.20 0.0 1

2

3

4

5

6

7

8

Cumulative proportion of HBsAg-loss

Cumulative proportion of HBeAg-loss

1.00

0

N

4

5

6

7

8

1.00 0.80 0.60 0.40 0.20 0.00 0

1

2

Time (years) Events

3

Time (years)

Time (years)

3

4

5

6

7

8

Time (years)

0

5

9

17

11

4

2

4

3*

151

145

132

99

71

65

60

53

20

Events N

0

2

1

3

2

2

1

1

3**

290

285

274

239

180

173

165

154

71

Fig. 2. Incidence rate and cumulative proportion of patients with hepatitis B ‘e’ antigen (a) and hepatitis B surface antigen (b) seroclearance during follow-up. Incidence rate is depicted in the top panels and cumulative proportion in the bottom panels. Number of events is given for each yearly period, including the number of patients continuing follow-up below (i.e., those without antigen seroclearance and who were not lost to follow-up). A total of 60 HBeAg seroclearance events occurred in 151 HBeAgpositive patients (
of which five occurred >8 years and are not depicted in the figure) and 17 HBsAg seroclearance events in 290 patients (

of which two occurred >8 years and are not depicted in the figure). HBeAg, hepatitis B ‘e’ antigen; HBsAg, hepatitis B surface antigen.

unstable profiles of HBeAg seroclearance: six had reverted back to HBeAg-positive serology, one changed serostatus multiple times until ending HBeAg-positive, and four patients changed serostatus multiple times until ending HBeAg-negative. Of the 53 patients ending follow-up with HBeAgnegative serology, acquisition of anti-HBeAb was achieved in 24 (45.3%) patients either at the same visit as HBeAg seroclearance (n ¼ 17) or from 1.1–3.2 years after HBeAg seroclearance (n ¼ 7). Of the 24, seven (29.2%) had a temporary anti-HBeAb-positive test, ending follow-up with anti-HBeAb-negative status.

Determinants of hepatitis B ‘e’ antigen seroclearance In multivariable analysis (Table 2), higher cumulative HBV-DNA copy-yearsTAVG (P < 0.001) and longer cumulative duration of TDF-containing ART

(P < 0.001) were inversely and significantly associated with HBeAg seroclearance, whereas baseline F3–F4 fibrosis levels were positively associated with HBeAg seroclearance (P ¼ 0.001). Because of the significant and inverse association of HBeAg seroclearance with cumulative TDF-duration, we decided to describe HBeAg seroclearance rates within the subgroup of HBeAg-positive patients undergoing TDF-containing ART (n ¼ 123). In this group, 42 HBeAg seroclearance events occurred within a median 2.6 (IQR ¼ 1.9–4.8) years after TDF-initiation. Interval incidence rates of HBeAg seroclearance were constant during follow-up, with a peak occurring around 3–4 years of treatment: between TDF-initiation and year 1, incidence rate ¼ 4.3/100 person-years; years 1–2, incidence rate ¼ 9.8/100 person-years; years 2–3, incidence rate ¼ 8.4/100 person-years; years 3–4, incidence rate ¼ 13.1/100 person-years; and more than 4 years,

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

1967

1968

AIDS

2015, Vol 29 No 15

Table 2. Determinants of hepatitis B ‘e’ antigen seroclearance in HIV–hepatitis B virus coinfected patients. Multivariablea

Univariable nb

HR (95% CI)

P

HR (95% CI)

P

Sex (women versus men) Age (per year) BMI (per kg/m2) High HBV-endemic region AIDS-defining illness pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cells=ml CD4þ cell count

151 151 150 151 151 151

0.84 1.06 0.98 0.91 1.17 1.10

(0.26–2.71) (0.88–1.27) (0.91–1.05) (0.28–2.93) (0.55–2.47) (0.99–1.21)

0.8 0.6 0.5 0.9 0.7 0.08

CD4þ cell count >500 cells/ml CD4þ cell count >350 cells/ml CD4þ cell count >250 cells/ml Lag CD4þ (per 100 cells/ml) Lag CD4þ (% cells/ml)  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cells=ml CD4þ nadir

151 151 151 151 151 137

1.90 1.35 1.70 1.01 0.95 1.06

(0.89–4.05) (0.59–3.08) (0.54–5.35) (0.85–1.22) (0.56–1.61) (0.97–1.16)

0.09 0.5 0.4 0.9 0.8 0.2

CD4þ nadir >350 cells/ml CD4þ nadir >250 cells/ml Duration of ART (per year) Undetectable HIV-RNA Cumulative HIV-RNA (log10copy-yearsTAVG) Undetectable HBV-DNA Cumulative HBV-DNA (log10copy-yearsTAVG) Concomitant anti-HBV NA-treatment LAM TDF TDFþLAM/FTC Other Cumulative LAM duration (per year) Cumulative ADV duration (per year) Cumulative TDF duration (per year) ALT levels at baseline 2 ULN F3–F4 fibrosis at baseline HBV genotype A, A/D, A/G D E G L-Nucleoside-associated pol gene mutations Anti-HCV positive serology

137 137 151 151 151 149 149 151

0.89 1.16 0.92 4.38 0.67 3.29 0.69

(0.39–2.04) (0.52–2.60) (0.87–0.97) (1.19–16.10) (0.48–0.94) (1.56–6.94) (0.58–0.83)

0.8 0.7 0.003 0.03 0.02 0.002 350 cells/ml CD4þ cell count >250 cells/ml Lag CD4þ (per 100 cells/ml) Lag CD4þ(%500 cells/ml) pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cells=ml CD4þ nadir

290 290 290 290 290 263

0.50 0.74 0.59 1.12 2.19 0.97

(0.12–2.03) (0.18–3.13) (0.10–3.42) (0.72–1.72) (0.40–12.02) (0.87–1.08)

0.3 0.7 0.6 0.6 0.4 0.6

CD4þ nadir>350 cells/ml CD4þ nadir>250 cells/ml Duration of ART (per year) Cumulative HIV-RNA (log10copy-yearsTAVG) Cumulative HBV-DNA (log10copy-yearsTAVG) HBeAg-positive at baseline Concomitant anti-HBV NA-treatment LAM TDF TDFþLAM/FTC Other Cumulative LAM-duration (per year) Cumulative ADV-duration (per year) Cumulative TDF-duration (per year) ALT at baseline 2 ULN F3–F4 fibrosis at baseline L-Nucleoside-associated pol gene mutations Precore mutation (T13S or W28stop) Anti-HCV positive serology

263 263 290 290 285 290 290

0.38 0.58 0.87 1.04 0.28 0.15

(0.05–3.10) (0.10–3.20) (0.77–0.98) (0.40–2.68) (0.13–0.59) (0.03–0.71)

0.4 0.5 0.03 0.9 0.001 0.02

4.24 5.24 5.52 0.98 1.06 0.70

1.00 (0.14–129.04) (0.33–83.53) (0.59–51.58) (0.77–1.25) (0.43–2.62) (0.52–0.95)

0.4 0.2 0.14 0.9 0.9 0.02

2.30 0.11 1.04 0.37 0.35 0.69

1.00 (0.34–15.48) (0.01–1.86) (0.17–6.23) (0.01–10.65) (0.01–8.98) (0.04–11.58)

0.4 0.13 0.9 0.6 0.5 0.8

290 290 290 290

290 158 151 290

HR (95% CI)

P

0.32 (0.13–0.80)

0.01

0.80 (0.63–1.02)

0.07

ADV, adefovir; ALT, alanine aminotransferase; ART, antiretroviral therapy; CI, confidence interval; FTC, emtricitabine; HBV, hepatitis B virus; HCV, hepatitis C virus; HR, hazard ratio; IFN, interferon; IQR, interquartile range; LAM, lamivudine; TDF, tenofovir; ULN, upper limit normal. a Duration of ART was not included in the multivariable model due to collinearity with other cumulative anti-HBV duration variables. Baseline HBeAg-status was not included as it no longer had a P value below the prespecified threshold (P ¼ 0.60). b Total number of patients in analysis.

Determinants of hepatitis B surface antigen seroclearance As shown in Table 3, only increased cumulative HBVDNA copy-yearTAVG was inversely and significantly associated with HBsAg seroclearance (P ¼ 0.01), while increased cumulative TDF-exposure tended to be inversely associated with HBsAg seroclearance (P ¼ 0.07) in multivariable analysis. Again, because of the inverse relationship with cumulative TDF exposure and HBsAg seroclearance, we decided to conduct analysis within the subgroup of patients undergoing TDF-containing ART (n ¼ 227). In this group, 12 HBsAg seroclearance events occurred within a median 1.9 (IQR ¼ 0.7–3.6) years after TDF initiation. Accordingly, interval incidence rates of HBsAg seroclearance were highest during the first few years of treatment: between TDF initiation and year 1, incidence rate ¼ 1.9/100 person-years; years 1–2, incidence rate ¼ 1.6/100 person-years; years 2–3, incidence

rate ¼ 1.2/100 person-years; and more than 3 years, incidence rate ¼ 0.6/100 person-years. The association with cumulative HBV-DNA copy-yearTAVG and HBsAg seroclearance within this subgroup was as similarly pronounced as compared with the overall univariable analysis (HR ¼ 0.29, 95% CI ¼ 0.09–0.93). There was no significant difference between patients who initiated TDF without previous anti-HBV exposure versus those switching from a less-potent NA-therapy (HR ¼ 1.44, 95%CI ¼ 0.31–6.56).

Liver-related morbidity Overall, 19 (6.6%) events of liver-related morbidity were observed. Of them, four occurred prior to study inclusion: acute hepatic failure (n ¼ 2), HCC (n ¼ 1), and portal hypertension (n ¼ 1). During follow-up, 15 patients developed the following liver-related conditions after a median 4.6 years (range ¼ 1.0–5.9) of follow-up (incidence rate ¼ 0.9/100 person-years): biliary cirrhosis

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

1969

1970

AIDS

2015, Vol 29 No 15

(n ¼ 1), portal hypertension (n ¼ 6), hepatic failure (n ¼ 2), HCC (n ¼ 3), or some other liver condition (n ¼ 3). Among the 12 HBeAg-positive patients with liver-related morbidity, four had HBeAg seroclearance. Of the 19 patients overall, 18 remained HBsAg-positive during the entire follow-up, whereas one patient exhibited transient HBsAg loss before reconverting to HBsAg positive at the next study visit.

Overall mortality In total, there were 11 deaths during follow-up, occurring after a median 3.1 years (range ¼ 1.0–6.3) of follow-up (incidence rate ¼ 0.6/100 person-years). Almost all deaths were nonliver related (hematoma, n ¼ 1; pancreatic adenocarcinoma, n ¼ 1; colon adenocarcinoma, n ¼ 1; myocardial infarction, n ¼ 2; pneumonia, n ¼ 2; suicide, n ¼ 1; and unknown, n ¼ 1), with only two liver-related deaths (fulminant hepatitis, n ¼ 1; hepatic decompensation due to septic shock, n ¼ 1). Interestingly, of the nine patients with nonliver-related deaths, five (55.6%) had an underlying liver condition. Among the 10 HBeAg-positive patients who died during follow-up, five exhibited HBeAg-loss, four remained HBeAg-positive, and one reverted from negative to positive HBeAg-status. HBsAg-positive serology was observed throughout study follow-up for all 11 deaths.

Discussion

2.4% per year [27]. However, these rates are highly influenced by HBV genotype and can be unstable or transient during untreated infection. During long-term treatment with the less-potent nucleoside analogue LAM in patients monoinfected with HBV, specifically among those who had controlled HBV replication (i.e. did not develop resistance) and seroconverted HBeAg-negative during therapy, HBsAg seroclearance appears at 1.5% per year [28]. During long-term treatment with more potent NAs, such as TDF or entecavir, HBeAg and HBsAg seroclearance occur, respectively, at 9.1–12.4% per year and 1.7–2.9% in HBeAg-positive patients monoinfected with HBV, whereas HBsAg seroclearance is infrequent in HBeAg-negative patients at less than 0.1–2.1% per year [29,30]. Taken together, seroclearance rates during treated HBV monoinfection are not entirely different to those observed in our cohort. Our study enrolled patients during a period when most patients coinfected with HIV–HBV had years of experience with LAM-containing ART and either developed LAM-resistance or, less likely, controlled HBV replication as a result of effective therapy [31]. At the same time, the nucleotide analogue TDF, which had recently been approved for HIV infection, was shown to exhibit anti-HBV activity [32], thereby progressively making TDF-containing ART a more attractive choice for patients coinfected with HIV–HBV. The long-term use of these regimens is highly effective in suppressing HBV replication among coinfected patients without any evidence of antiviral resistance mutations [33], even for those with low-level replication during the course of therapy [34]. Hence, it is important to consider both the study population as a whole and among those switching to an antiretroviral regimen containing TDF.

In our study of HIV-infected patients with chronic hepatitis B, we demonstrate overall that HBeAg seroclearance among HBeAg-positive patients occurred at a modest rate of 8.4/100 person-years, and generally occurred early on during follow-up. Overall HBsAg seroclearance occurred at a meager rate of 1.0/100 person-years and appeared relatively constant over time. The seroclearance rates in our study fell near the lower limits of previous research in HIV–HBV coinfection with varying designs, wherein HBeAg seroclearance rates ranged from roughly 1.7–22.9% per year (in HBeAgpositive patients) and HBsAg seroclearance rates ranged from 0.6–6.0% per year [10,12,14–16,24–26]. These lower estimations are sensible, considering that our study has more ‘real world’ scenarios in which patients undergo antiviral treatment and includes a more representative profile of the coinfected population currently seen across Europe [18].

Contrary to what was expected, cumulative TDFduration appeared to negatively affect both HBeAg and HBsAg seroclearance in the overall population. However, when looking specifically at patients undergoing TDFcontaining ART, the incidence rates of seroclearance were highest during the early years of treatment and, for HBsAg seroclearance in particular, there was a clear drop in incidence rates after the third year of TDF-therapy. Some evidence as to why this occurs can be gathered from studies examining HBsAg quantification. TDF-treated coinfected patients with HBsAg seroclearance almost always have steep and early declines in HBsAg quantification, whereas those without HBsAg seroclearance have HBsAg levels generally stagnating above 3.0 log10 IU/ml [10,12,24]. The negative overall association observed in our study most likely stems from patients who continued follow-up and comprised this latter group.

During the natural course of chronic HBV monoinfection, HBeAg seroclearance occurs in roughly 2–15% of patients per year and as these patients enter the inactive phase of infection, HBsAg seroclearance follows at 0.7–

With lower time-averaged cumulative HBV-DNA being a significant determinant of seroclearance in multivariable analysis, suppressed HBV-DNA viremia would appear central to achieving these events. This finding mirrors

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Seroclearance in HIV–hepatitis B virus coinfection Boyd et al.

recent results among coinfected patients undergoing TDF-containing ART wherein only patients achieving and maintaining complete HBV-DNA suppression had achieved HBeAg seroclearance, HBeAg seroconversion, and HBsAg seroclearance [34]. Another factor associated with faster rates of specifically HBeAg seroclearance was higher levels of baseline fibrosis, which is surprising given the deleterious effect of cirrhosis on HBsAg seroclearance [29]. One possible explanation could be that the higher levels of cirrhosis in this group were associated with longer periods of HBeAg-positive infection, representing a subgroup with more advanced stages of disease [6]. As HBeAg-positive patients with high liver fibrosis tend to have lower levels of quantifiable HBeAg [35], these patients might have already been closer toward transitioning to an HBeAg-negative state [12]. Further work will require a more comprehensive understanding of how liver fibrosis changes in patients coinfected with HIV– HBV before and after seroclearance. Insofar as the virological and serological response rates are deemed similar between HBV-monoinfected and patients coinfected with HIV–HBV undergoing highly potent antiviral therapy [26,36], the effect of HIV-related immunosuppression poses a particularly interesting question on immune control of HBV. Intrahepatically, studies from liver biopsies have shown that coinfected patients with severe immunosuppression have decreased CD69þ T-cell and stellate cell activation when compared with monoinfected patients, while these levels do not change significantly after initiation of potent anti-HBVcontaining ART [37]. Clinical evidence suggests further that coinfected patients with high ALT levels and steep increases in CD4þ cell counts after treatment initiation are more likely to achieve HBsAg seroconversion, probably due to the effects of severe immunorestoration [17,24]. We modeled CD4þ cell counts at various levels, including the CD4þ changes from the previous visit, and found no significant relationship with either HBeAg or HBsAg seroclearance. As our study population had extensive experience undergoing ART at study inclusion with very gradual increases in CD4þ cell count over time, HBsAg loss as a result of this phenomenon would have been unlikely. Although we did gather information on the degree of previous immunodeficiency (via nadir CD4þ cell count), we cannot provide any clues as to the impact of ART-induced immune recovery on eventual seroclearance because changes in CD4þ cell count after ART initiation were not collected during our study. Treatment-naive study populations with larger ranges of CD4þ T cells would be needed to examine the profiles of CD4þ changes and hone its relationship with seroclearance. Large epidemiological studies have shown that patients coinfected with HIV–HBV not only have an increased risk of liver-related death compared with those with HIV monoinfection, but also death related to other or

unknown causes [4,38]. In our study, we observed almost no liver-related deaths but rather a wide range of other causes of mortality. The extent of HBV-replication suppression would likely explain the lack of liver-related deaths [39]. However, severe underlying liver conditions were noted in over half those who died of nonliverrelated causes. It could be hypothesized that coinfected patients with poor liver function are generally more prone to overall mortality; however, further data would be needed to confirm this. Certain limitations of our study need to be addressed. First, there were few HBsAg seroclearance events during follow-up, which could have reduced the power to detect certain risk factors. We conducted a number of simulations based on parameters from our study, demonstrating that any truly significant HR at least 4.0 of a time-invariant dichotomous covariate would have had adequate power (>0.80), whereas HRs of 2.0 for the same covariate would have yielded power at roughly 0.30 to detect a significant association. Unfortunately, this limitation is shared between all studies involving serological endpoints among coinfected patients, thereby warranting further collaboration. In addition, the definition for seroclearance was not ‘stable,’ whereby two consecutive visits with negative serology would be required. Sensitivity analysis using this definition did demonstrate a similar magnitude of effects (data not shown), yet with wider CIs. Second, there was a noticeable drop in participation between the first and second study periods. However, the differences in characteristics of those participating or not were almost negligible. It could be argued that the longer duration of LAM-containing ART and increased use of standard interferon among those not participating could mean that these patients had been more likely to have an indication for treatment due to higher degrees of liver disease, but this is difficult to distinguish from center-specific treatment preferences. Finally, the largely ART-experienced study population could have inadvertently excluded patients with seroclearance associated with rapid immune recovery, thus biasing our results toward lower seroconversion rates. However, most recent treatment recommendations suggest initiating ART at higher CD4þ count levels, implying that any present evaluation performed in HIV-infected patients would likely not observe this phenomenon. In conclusion, as roughly half of HBeAg-positive patients cleared HBeAg and 5% of all patients cleared HBsAg in our study, seroclearance remains a formidable challenge in HIV–HBV coinfection. Both of these endpoints are strongly associated with decreased HBV replication, pointing toward the benefit of highly potent antiviral treatment. However, seroclearance appears to more readily occur within the first years of treatment. These findings strongly underscore the persistent nature of HBV infection among patients coinfected with HIV–HBV, stressing the

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

1971

1972

AIDS

2015, Vol 29 No 15

need for other novel therapeutics in clearing HBsAg and initiatives toward finding other strategies that could elicit a cure for HBV. Meanwhile, it would be helpful if larger cohorts of coinfected patients could establish the additional benefit of liver-related and nonliver-related morbidity and mortality after seroclearance.

Acknowledgements The authors are grateful to the patients and to the clinical teams for their commitment to the French HIV–HBV Cohort Study. The authors acknowledge L. Roguet for managing the logistics of the French HIV–HBV Cohort; G. Pannetier and F. Carrat for their help in the data management; and Pr P. Cales for kindly providing Fibrometre scores. Finally, the authors would like to thank to Pr J.-F. Flejou and E. Roux of the Tumorothe`que H.U.E.P. at Saint-Antoine Hospital for storing samples. Authors’ contributions: A.B. managed data for the French HIV–HBV cohort, performed statistical analysis, and drafted the article; J.G. validated all virological and serological measures and gave critical input on the article; P.M. and C.L.-C. helped conceptualize and design of the French HIV–HBV cohort, participated in patient recruitment, and gave critical input on the article; M.S.-L.C. and H.R. coordinated and participated in data collection and helped draft parts of the article; F.Z. coordinated data collection on genetic variability and gave valuable comments on the article; P.-M.G. and K.L. conceptualized, designed, and sought funding for the French HIV–HBV cohort, coordinated data collection, and drafted parts of the article. All authors have approved the final version of the article. Source of funding: This work was supported in part by the Institut de Me´decine et d’Epide´miologie Applique´e and received additional grants from A.N.R.S. (Agence Nationale de Recherche sur le Sida et les He´patites). Gilead Sciences, Inc. provided an unrestricted grant for the French HIV–HBV cohort and was not involved in any part of the data collection, analysis, and article writing. Postdoctoral fellowships from the A.N.R.S. and SIDACTION were awarded to A.B.

Conflicts of interest There are no conflicts of interest.

References 1. Smith CJ, Ryom L, Weber R, Morlat P, Pradier C, Reiss P, et al. Trends in underlying causes of death in people with HIV from 1999 to 2011 (D:A:D): a multicohort collaboration. Lancet 2014; 384:241–248. 2. Lacombe K, Rockstroh J. HIV and viral hepatitis coinfections: advances and challenges. Gut 2012; 61 (Suppl 1):47–58.

3. Ioannou GN, Bryson CL, Weiss NS, Miller R, Scott JD, Boyko EJ. The prevalence of cirrhosis and hepatocellular carcinoma in patients with human immunodeficiency virus infection. Hepatology 2013; 57:249–257. 4. Falade-Nwulia O, Seaberg EC, Rinaldo CR, Badri S, Witt M, Thio CL. Comparative risk of liver-related mortality from chronic hepatitis B versus chronic hepatitis C virus infection. Clin Infect Dis 2012; 55:507–513. 5. Ly KN, Xing J, Klevens RM, Jiles RB, Ward JW, Holmberg SD. The increasing burden of mortality from viral hepatitis in the United States between 1999 and 2007. Ann Intern Med 2012; 156:271–278. 6. European Association for the Study of the Liver. EASL clinical practice guidelines: management of chronic hepatitis B virus infection. J Hepatol 2012; 57:167–185. 7. Liu J, Yang H-I, Lee M-H, Lu S-N, Jen C-L, Wang L-Y, et al. Incidence and determinants of spontaneous hepatitis B surface antigen seroclearance: a community-based follow-up study. Gastroenterology 2010; 139:474–482. 8. Yuen MF, Yuan HJ, Hui CK, Wong DK, Wong WM, Chan AO, et al. A large population study of spontaneous HBeAg seroconversion and acute exacerbation of chronic hepatitis B infection: implications for antiviral therapy. Gut 2003; 52:416–419. 9. Cornberg M, Ho¨ner Zu Siederdissen C. HBsAg seroclearance with NUCs: rare but important. Gut 2014; 63:1208–1209. 10. Zoutendijk R, Zaaijer HL, de Vries-Sluijs TEMS, Reijnders JGP, Mulder JW, Kroon FP, et al. Hepatitis B surface antigen declines and clearance during long-term tenofovir therapy in patients coinfected with HBV and HIV. J Infect Dis 2012; 206:974–980. 11. Kosi L, Reiberger T, Payer BA, Grabmeier-Pfistershammer K, Strassl R, Rieger A, et al. Five-year on-treatment efficacy of lamivudine-, tenofovir- and tenofovir R emtricitabine-based HAART in HBV-HIV-coinfected patients. J Viral Hepat 2012; 19:801–810. 12. Boyd A, Maylin S, Gozlan J, Delaugerre C, Simon F, Girard P-M, et al. Use of hepatitis B surface and ‘‘e’’ antigen quantification during extensive treatment with tenofovir in patients co-infected with HIV-HBV. Liver Int 2015; 35:795–804. 13. Martı´n-Carbonero L, Teixeira T, Poveda E, Plaza Z, Vispo E, Gonza´lez-Lahoz J, et al. Clinical and virological outcomes in HIV-infected patients with chronic hepatitis B on long-term nucleos(t)ide analogues. AIDS 2011; 25:73–79. 14. Strassl R, Reiberger T, Honsig C, Payer BA, Mandorfer M, Grabmeier-Pfistershammer K, et al. Viral determinants predicting hepatitis B surface antigen (HBsAg) seroclearance in HIV-/ HBV-coinfected patients. J Viral Hepat 2014; 21:508–516. 15. Psevdos G, Kim JH, Suh JS, Sharp VL. Predictors of loss of hepatitis B surface antigen in HIV-infected patients. World J Gastroenterol 2010; 16:1093–1096. 16. Plaza Z, Aguilera A, Mena A, Vispo E, Sierra-Enguita R, Tome´ S, et al. Influence of HIV infection on response to tenofovir in patients with chronic hepatitis B. AIDS 2013; 27:2219–2224. 17. Miailhes P, Trabaud M-A, Pradat P, Lebouche´ B, Chevallier M, Chevallier P, et al. Impact of highly active antiretroviral therapy (HAART) on the natural history of hepatitis B virus (HBV) and HIV coinfection: relationship between prolonged efficacy of HAART and HBV surface and early antigen seroconversion. Clin Infect Dis 2007; 45:624–632. 18. Soriano V, Mocroft A, Peters L, Rockstroh J, Antunes F, Kirkby N, et al. Predictors of hepatitis B virus genotype and viraemia in HIV-infected patients with chronic hepatitis B in Europe. J Antimicrob Chemother 2010; 65:548–555. 19. Price H, Bansi L, Sabin CA, Bhagani S, Burroughs A, Chadwick D, et al. Hepatitis B virus infection in HIV-positive individuals in the UK collaborative HIV cohort (UK CHIC) study. PLoS One 2012; 7:e49314. 20. Lacombe K, Massari V, Girard P-M, Serfaty L, Gozlan J, Pialoux G, et al. Major role of hepatitis B genotypes in liver fibrosis during coinfection with HIV. AIDS 2006; 20:419–427. 21. Rockstroh JK, Bhagani S, Benhamou Y, Bruno R, Mauss S, Peters L, et al. European AIDS Clinical Society (EACS) guidelines for the clinical management and treatment of chronic hepatitis B and C coinfection in HIV-infected adults. HIV Med 2008; 9:82–88. 22. Lacombe K, Boyd A, Lavocat F, Pichoud C, Gozlan J, Miailhes P, et al. High incidence of treatment-induced and vaccine-escape hepatitis B virus mutants among human immunodeficiency virus/ hepatitis B-infected patients. Hepatology 2013; 58:912–922.

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Seroclearance in HIV–hepatitis B virus coinfection Boyd et al. 23. Rabe-Hesketh S, Skrondal A. Multilevel and longitudinal modeling using Stata. 2nd ed. College Station, TX: StataCorp LP; 2008. 24. Matthews GV, Ali RJ, Avihingsanon A, Amin J, Hammond R, Bowden S, et al. Quantitative HBsAg and HBeAg predict hepatitis B seroconversion after initiation of HAART in HIVHBV coinfected individuals. PLoS One 2013; 8:e61297. 25. Schmutz G, Nelson M, Lutz T, Sheldon J, Bruno R, von Boemmel F, et al. Combination of tenofovir and lamivudine versus tenofovir after lamivudine failure for therapy of hepatitis B in HIV-coinfection. AIDS 2006; 20:1951–1954. 26. Bonacini M. Virologic and clinical outcomes in HIV/HBV coinfected patients on tenofovir-containing HAART. Gastroenterology 2010; 139:1827–1829. 27. Liaw Y-F, Brunetto MR, Hadziyannis S. The natural history of chronic HBV infection and geographical differences. Antivir Ther 2010; 15 (Suppl 3):25–33. 28. Fasano M, Lampertico P, Marzano A, Di Marco V, Niro GA, Brancaccio G, et al. HBV DNA suppression and HBsAg clearance in HBeAg negative chronic hepatitis B patients on lamivudine therapy for over 5 years. J Hepatol 2012; 56:1254– 1258. 29. Kim G-A, Lim Y-S, An J, Lee D, Shim JH, Kim KM, et al. HBsAg seroclearance after nucleoside analogue therapy in patients with chronic hepatitis B: clinical outcomes and durability. Gut 2014; 63:1325–1332. 30. Buti M, Fung S, Gane E, Afdhal NH, Flisiak R, Gurel S, et al. Long-term clinical outcomes in cirrhotic chronic hepatitis B patients treated with tenofovir disoproxil fumarate for up to 5 years. Hepatol Int 2015; 9:243–250. 31. Wolters LMM, Niesters HGM, Hansen BE, van der Ende ME, Kroon FP, Richter C, et al. Development of hepatitis B virus resistance for lamivudine in chronic hepatitis B patients coinfected with the human immunodeficiency virus in a Dutch cohort. J Clin Virol 2002; 24:173–181.

32. Dore GJ, Cooper DA, Pozniak AL, DeJesus E, Zhong L, Miller MD, et al. Efficacy of tenofovir disoproxil fumarate in antiretroviral therapy-naive and -experienced patients coinfected with HIV-1 and hepatitis B virus. J Infect Dis 2004; 189:1185–1192. 33. De Vries-Sluijs TEMS, Reijnders JGP, Hansen BE, Zaaijer HL, Prins JM, Pas SD, et al. Long-term therapy with tenofovir is effective for patients co-infected with human immunodeficiency virus and hepatitis B virus. Gastroenterology 2010; 139:1934–1941. 34. Boyd A, Gozlan J, Maylin S, Delaugerre C, Peytavin G, Girard PM, et al. Persistent viremia in human immunodeficiency virus/ hepatitis B coinfected patients undergoing long-term tenofovir: virological and clinical implications. Hepatology 2014; 60:497–507. 35. Wang H, Yan R, Zhou Y, Wang MS, Ruo GQ, Cheng MJ. Comparison of hepatitis B surface antigen and e antigen in predicting liver histology in hepatitis B e antigen-positive chronic hepatitis B patients. Hepatol Int 2014; 8:216–223. 36. Piroth L, Pol S, Lacombe K, Miailhes P, Rami A, Rey D, et al. Management and treatment of chronic hepatitis B virus infection in HIV positive and negative patients: the EPIB 2008 study. J Hepatol 2010; 53:1006–1012. 37. Iser DM, Avihingsanon A, Wisedopas N, Thompson AJ, Boyd A, Matthews GV, et al. Increased intrahepatic apoptosis but reduced immune activation in HIV-HBV co-infected patients with advanced immunosuppression. AIDS 2011; 25:197–205. 38. Data Collection on Adverse Events of Anti-HIV drugs (D:A:D) Study Group. Smith C, Sabin CA, Lundgren JD, Thiebaut R, Weber R, et al. Factors associated with specific causes of death amongst HIV-positive individuals in the D:A:D Study. AIDS 2010; 24:1537–1548. 39. Iloeje UH, Yang H-I, Jen C-L, Su J, Wang L-Y, You S-L, et al. Risk and predictors of mortality associated with chronic hepatitis B infection. Clin Gastroenterol Hepatol 2007; 5:921–931.

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

1973