Research Article Viral Hepatitis Viral and immune factors associated with successful treatment withdrawal in HBeAg-negative chronic hepatitis B patients Mireia García-López1,†, Sabela Lens1,†, Laura J. Pallett2, Barbara Testoni3, Sergio Rodríguez-Tajes1, Zoe Mariño1, Concepción Bartres1, Ester García-Pras1, Thais Leonel1, Elena Perpiñán1, Juan José Lozano4, Francisco Rodríguez-Frías5, George Koutsoudakis1, Fabien Zoulim3, Mala K. Maini2, Xavier Forns1,*, Sofía Pérez-del-Pulgar1,* 1 Liver Unit, Hospital Clínic, University of Barcelona, IDIBAPS, CIBERehd, Barcelona, Spain; 2Division of Infection and Immunity, Institute of Immunity and Transplantation, University College London, London, UK; 3INSERM U1052-Cancer Research Center of Lyon (CRCL), University of Lyon, UMR_S1052, CRCL, Lyon, France; 4Bioinformatics Platform, CIBERehd, Barcelona, Spain; 5Liver Pathology Unit, Department of Biochemistry and Microbiology, Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona, CIBERehd, Barcelona, Spain Background & Aims: Factors associated with a successful patients with HBeAg-negative CHB. The presence of functional outcome upon nucleos(t)ide analogue (NA) treatment with- HBV-specific T cells at baseline are associated with a successful drawal in HBeAg-negative chronic hepatitis B (CHB) patients outcome after treatment withdrawal. have yet to be clarified. The objective of this study was to analyse Lay summary: Nucleos(t)ide analogue therapy can be dis- the HBV-specific T cell response, in parallel with peripheral and continued in a high proportion of chronic hepatitis B patients intrahepatic viral parameters, in patients undergoing NA without cirrhosis. The strength of HBV-specific immune T cell discontinuation. responses may contribute to successful viral control after anti- Methods: Twenty-seven patients without cirrhosis with HBeAg- viral treatment interruption. Our comprehensive study provides negative CHB with complete viral suppression (>3 years) were in-depth data on virological and immunological factors than can studied prospectively. Intrahepatic HBV-DNA (iHBV-DNA), help guide individualised therapy in patients with chronic hep- intrahepatic HBV-RNA (iHBV-RNA), and covalently closed circu- atitis B. lar DNA (cccDNA) were quantified at baseline. Additionally, © 2020 European Association for the Study of the Liver. Published by serum markers (HBV-DNA, HBsAg, HBV core-related antigen Elsevier B.V. This is an open access article under the CC BY license [HBcrAg] and HBV-RNA) and HBV-specific T cell responses were (http://creativecommons.org/licenses/by/4.0/). analysed at baseline and longitudinally throughout follow-up. Results: After a median follow-up of 34 months, 22/27 patients Introduction (82%) remained off-therapy, of whom 8 patients (30% of the total Functional cure, defined as the loss of HBsAg, with or without the cohort) lost HBsAg. Baseline HBsAg significantly correlated with development of anti-HBs antibodies, is regarded as the optimal iHBV-DNA and iHBV-RNA, and these parameters were lower in treatment endpoint in patients with HBeAg-negative chronic patients who lost HBsAg. All patients had similar levels of hepatitis B (CHB). However, despite more than 10 years on detectable cccDNA regardless of their clinical outcome. Patients nucleos(t)ide analogue (NA) therapy, this endpoint is only ach- achieving functional cure had baseline HBsAg levels < −1,000 IU/ ieved in 1–5% of patients.1,2 The latest EASL guidelines propose ml. Similarly, an increased frequency of functional HBV-specific NA discontinuation in selected HBeAg-negative patients without CD8+ T cells at baseline was associated with sustained viral cirrhosis who have achieved long-term (>3 years) virological control off treatment. These HBV-specific T cell responses per- suppression on treatment, if close clinical monitoring can be sisted, but did not increase, after treatment withdrawal. A guaranteed after NA discontinuation.2 These recommendations similar, but not statistically significant trend, was observed for are based on data from studies evaluating treatment withdrawal HBV-specific CD4+ T cell responses. in Caucasian HBeAg-negative patients with complete viral sup- Conclusions: Decreased cccDNA transcription and low HBsAg pression on NA therapy3,4 where the rate of HBsAg loss after NA levels are associated with HBsAg loss upon NA discontinuation in discontinuation reached 20%. Importantly, a high proportion of these patients achieved sustained virological remission (low HBV-DNA levels and normal alanine aminotransferase [ALT] Keywords: Hepatitis B virus; Chronic hepatitis B; Nucleos(t)ide analogue; Antiviral levels). Despite this recommendation, some clinicians may be therapy discontinuation; HBsAg; HBcrAg; Covalently closed circular DNA; HBV-RNA; reluctant to offer NA withdrawal to their patients owing to the Immune response; HBV-Specific T cells. high probability of viral rebound after stopping antiviral therapy Received 7 May 2020; received in revised form 28 October 2020; accepted 16 November (serum HBV-DNA >2,000 IU/ml in about 55–70%).5 However, 2020; available online 2 December 2020 * Corresponding authors. Address: Liver Unit, Hospital Clínic, IDIBAPS, Barcelona, viral relapses are not always associated with ALT flares and, Spain. Tel.: +34-2275400, ext. 2093; Fax: +34-933129405. (X. Forns) (S. Pérez-del- when transitory, may lead to long-term virological remission. In Pulgar). addition, an accurate definition of relapse and criteria for NA re- E-mail addresses: xforns@clinic.cat (X. Forns), sofiapp@clinic.cat (S. Pérez-del-Pulgar). † These authors contributed equally to this work. introduction are needed to prevent premature re-treatment that https://doi.org/10.1016/j.jhep.2020.11.043 may impair the possibility of remission or functional cure.6 Journal of Hepatology 2021 vol. 74 j 1064–1074 Table 1. Characteristics of the entire cohort at baseline and during follow-up according to clinical outcome. Variables ALL patients (n = 27) HBsAg loss (n = 8) Virological control (n = 14) Treatment re-introduction (n = 5) p value Sex (male) 21 (78%) 7 (87%) 10 (71%) 4 (80%) 0.67 Age (years) 56 (45–61) 60 (57–63) 51 (43–56) 46 (44–65) 0.17 Duration of NA therapy (years) 8 (7–13) 13 (9–15) 8 (7–12) 8 (7–11) 0.18 NA therapy Tenofovir 20 (74%) 8 (100%) 9 (64%) 3 (60%) 0.13 Entecavir 7 (26%) 0 5 (36%) 2 (40%) Baseline elastography (kPa) 4.8 (4.0–5.5) 4.3 (3.3–5.1) 4.8 (3.9–5.3) 4.8 (3.9–5.3) 0.28 Last elastography (kPa; n = 19) 5.3 (4.3–6.1) 4.4 (3.7–5.4) 5.7 (4.2–6.7) 5.3 (4.9–5.7) 0.19 Fibrosis stage F0–1 25 (93%) 7 (88%) 13 (93%) 5 (100%) 0.70 F2 2 (7%) 1 (12%) 1 (7%) 0 HBV genotype A 3 (11%) 0 2 (14%) 1 (20%) 0.31 C 1 (4%) 0 1 (7%) 0 D 21 (78%) 6 (75%) 11 (78%) 4 (80%) F 2 (7%) 2 (25%) 0 0 Journal of Hepatology 2021 vol. 74 j 1064–1074 Baseline ALT (IU/L) 23 (17–26) 21 (17–26) 23 (19–28) 16 (15–27) 0.45 ALT peak > − 2 ULN 17 (63%) 3 (36%) 9 (64%) 5 (100%) 0.07 ALT peak − > 5 ULN 8 (29%) 3 (36%) 2 (14%) 3 (60%) 0.13 ALT peak > − 10 ULN 6 (22%) 3 (36%) 1 (7%) 2 (40%) 0.15 Last ALT (IU/L) 22 (16–30) 17 (16–20) 28 (23–33) 27 (22–36) 0.004 HBV-DNA peak 2,000 IU/ml 21 (78%) 3 (37%) 13 (93%) 5 (100%) 0.005 HBV-DNA peak > − 20,000 IU/ml 12 (44%) 2 (25%) 5 (36%) 5 (100%) 0.019 Serum virological markers Baseline qHBsAg (IU/ml) 1,310 (556–3,031) 70 (13–507) 2,020 (1,089–4,361) 2,122 (933–3,158) 0.002 < − 2,000 IU/ml 17 (63%) 8 (100%) 7 (50%) 2 (40%) 0.03 < 1,000 IU/ml − 11 (41%) 7 (87%) 3 (21%) 1 (20%) 0.006 < − 100 IU/ml 5 (18%) 5 (63%) 0 0 0.001 Last qHBsAg (IU/ml) 372 (0–1,689) 0 1,368 (175–2,749) 661 (350–1,953) 0.82 < 2,000 IU/ml − 21 (28%) n.a. 9 (64%) 4 (80%) 0.48 < − 1,000 IU/ml 17 (63%) n.a. 6 (43%) 3 (60%) 0.62 < − 100 IU/ml 10 (37%) n.a. 2 (14%) 0 0.53 Delta qHBsAg (Log IU/ml) -0.22 -(0.58–0.16) n.a. -0.22 -(0.66–0.14) -0.39 -(0.51–0.17) 0.92 Baseline HBcrAg (positive) 13 (48%) 2 (25%) 8 (57%) 3 (60%) 0.29 Baseline HBcrAg (log U/ml)* 3.2 (2.8–5.1) 3.9 (2.8–5.1) 3.0 (2.8–3.3) 3.0 (2.8–3.4) 0.97 Last HBcrAg (positive) 8 (29%) 1 (12%) 5 (36%) 2 (40%) 0.53 Baseline HBV-RNA (positive) 11 (41%) 1 (12%) 7 (50%) 3 (60%) 0.14 Baseline HBV-RNA (copies/ml)* 136 (96–842) n.a. 101 (96–477) 840 (603–922) 0.07 Last HBV-RNA (positive) 10 (37%) 0 7 (50%) 3 (60%) 0.53 Liver virological markers at baseline iHBV-DNA (copies/cell) 0.36 (0.14–1.02) 0.04 (0.02–0.13) 0.58 (0.25–1.15) 0.91 (0.53–1.33) 0.001 cccDNA (copies/cell) 0.09 (0.035–0.36) 0.04 (0.01–0.40) 0.11 (0.06–0.36) 0.05 (0.02–0.29) 0.32 iHBV-RNA (positive)† 12/18 (67%) 2/7 (14%) 7/8 (87%) 3/3 (100%) 0.009 iHBV-RNA (copies/25 ng total RNA)*,† 203 (37–4,417) n.a. 655 (28–4,417) 123 (29–184) 0.15 iHBV-RNA/cccDNA*,† 3,834 (76–34,780) n.a. 5,642 (76–34,780) 3,835 (3,174–28,996) 0.72 Categorical and quantitative data are expressed as n (%) and median (IQR), respectively. Groups were compared using the Chi-square test for categorical variables and the ANOVA test for continuous variables. ALT, alanine aminotransferase; cccDNA, covalently closed circular DNA; iHBV-DNA, intrahepatic HBV-DNA; iHBV-RNA, intrahepatic HBV-RNA; n.a., not applicable; ULN, upper limit of normal. *Analysis of median (IQR) values for HBcrAg, HBV-RNA, iHBV-RNA, and iHBV-RNA/cccDNA was performed only considering patients with positive results. † Data available for 18 patients. One patient in the HBsAg loss group had detectable iHBV-RNA 10 upper limit of normal the criteria for HBeAg-negative chronic infection with normal (ULN) regardless of the HBV-DNA level. ALT values and HBV-DNA <2,000 IU/ml. The remaining patients  ALT >5–10 ULN and HBV-DNA >2,000 IU/ml persisting for > −28 were in a ‘grey zone’; 7 had HBV-DNA levels >2,000 IU/ml with days (4 weeks). normal ALT values, with a further 2 patients presenting with  ALT >2–5 ULN and HBV-DNA >2,000 IU/ml persisting for > −6 elevated ALT levels that remained <2 ULN. Of the 27 CHB pa- months. tients, only 5 patients (18%) required NA re-introduction after  Need for immunosuppressive treatment. stopping NAs (Fig. 1A). The reasons for treatment re-introduction were: early flare in HBV-DNA and persistent ALT levels >10 ULN The primary efficacy endpoint was the proportion of patients (n = 1), high viraemia with persistent 2–5 ULN ALT levels for at achieving HBsAg loss (functional cure) or virological control least 4 weeks (n = 3), and the need for high dose corticosteroid (patients who remained off-therapy being HBsAg positive) after therapy for more than 7 days (n = 1, who also had an ALT >2 ULN treatment withdrawal. Secondary endpoints included the and HBV-DNA >20,000 IU/ml >8 weeks). We did not find any A 5 rho: 0.65 p = 0.0003 B 5 rho: 0.48 p = 0.04 C 5 rho: 0.45 p = 0.06 HBsAg (log IU/ml) HBsAg (log IU/ml) HBsAg (log IU/ml) 4 4 4 3 3 3 2 2 2 1 1 1 0 0 0 0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 1 2 3 4 5 iHBV-DNA (copies/cell) # iHBV-RNA (log copies/25 ng total RNA) iHBV-RNA/cccDNA# Fig. 2. Correlations between HBsAg and intrahepatic replication markers before NA discontinuation. Correlation between HBsAg and (A) iHBV-DNA, (B) iHBV-RNA, and (C) iHBV-RNA/cccDNA ratio as a measure of cccDNA transcriptional activity. Rho, Spearman’s correlation coefficient. #Data available for 18 pa- tients. cccDNA, covalently closed circular DNA; iHBV-DNA, intrahepatic HBV-DNA; iHBV-RNA, intrahepatic HBV-RNA; NA, nucleos(t)ide analogue. Journal of Hepatology 2021 vol. 74 j 1064–1074 1067 Research Article Viral Hepatitis A *** * *** ** iHBV-DNA (copies/cell) 6 2.0 1.5 cccDNA (copies/cell) HBsAg (log IU/ml) 1.5 4 1.0 1.0 2 0.5 0.5 0 0.0 0.0 HBsAg Virological Re-treated HBsAg Virological Re-treated HBsAg Virological Re-treated loss control loss control loss control B HBcrAg Serum HBV-RNA iHBV-RNA# ** Detectable C Undetectable 100 100 Probability of HBsAg loss 2 8 3 1 7 3 2 7 3 Log-Rank 9.9 p = 0.002 80 80 HBsAg ≤1,000 IU/ml Patients (%) 60 60 40 40 20 20 HBsAg >1,000 IU/ml 6 6 2 7 7 2 5 1 0 0 0 50 100 150 200 ca loss ca loss ca loss d d l l l d Re ntro Re ntro Re ntro te te te ea ea ea g g g Weeks after NA cessation o o o ro BsA ro BsA sA -tr -tr lc lc lc -tr HB H H gi gi gi Se Sp PPV NPV LR+ LR- lo lo lo ro 88% 79% 64% 94% 4.16 0.16 Vi Vi Vi Fig. 3. Association between baseline virological markers and outcome after NA discontinuation. (A) Baseline HBsAg, iHBV-DNA, and cccDNA levels according to patient outcome. The red lines indicate median values. (B) Proportion of patients with detectable HBcrAg, serum HBV-RNA, and iHBV-RNA according to outcome. Mann-Whitney (A) and Chi-squared tests (B) were used for comparison, *p <0.05, **p <0.01, ***p <0.001. (C) Kaplan-Meier curves for HBsAg loss after NA interruption according to HBsAg levels at baseline. #Data available for 18 patients. cccDNA, covalently closed circular DNA; iHBV-DNA, intrahepatic HBV-DNA; iHBV-RNA, intrahepatic HBV-RNA; LR, likelihood ratio; NA, nucleos(t)ide analogue; NPV, negative predictive value; PPV, positive predictive value; Se, sensibil- ity; Sp, specificity. difference according to outcome after NA discontinuation when 1,310 (556–3031) IU/ml, with 11 patients (41%) <−1.000 IU/ml and considering patient characteristics including sex, age, type, and 5 (18%) <− 100 IU/ml. Despite long-term NA therapy, all patients duration of antiviral therapy or HBV genotype (Table 1 and had detectable iHBV-DNA and cccDNA at baseline. In contrast, Supplementary Table S1). only 11 (41%) and 13 (48%) patients had detectable serum levels of HBV-RNA and HBcrAg, respectively. Intrahepatic HBV-RNA HBV-DNA, ALT, and HBsAg kinetics after NA withdrawal was detected in 67% of liver samples at baseline. Notably in all patients with undetectable iHBV-RNA (n = 6), we were also un- In this cohort, HBV-DNA viraemia peaked within 1–3 months of able to detect HBV-RNA in their matched serum samples. In 8 stopping NA therapy (Fig. 1B). Most patients had a virological relapse defined as either an increase in HBV-DNA above 2,000 patients we detected both iHBV-RNA and serum HBV-RNA, IU/mL (78%) or 20,000 IU/mL (44%). whereas in 4 patients only iHBV-RNA was detectable. In addi- tion, 5 out of 7 patients with virological control and 2 out of 3 ALT flares were observed in some patients during the ‘off- therapy’ period, although these were mostly mild (Table 1). Pa- patients needing re-treatment had detectable iHBV-RNA and tients with ALT flare >10 ULN presented with diverse clinical positive serum HBV-RNA levels. We analysed the correlations between baseline serum and outcomes with only 2 out of 6 patients needing treatment re- introduction, whereas 4 out of 6 remained off-therapy, intrahepatic viral markers (Fig. 2 and Supplementary Table S2). including 3 who achieved HBsAg loss (Table 1 and Fig. 1B). HBsAg correlated with both iHBV-DNA (rho = 0.65, p = 0.0003) and iHBV-RNA (rho = 0.48, p = 0.04), but not with cccDNA. Neither HBV-DNA nor ALT kinetics predicted the clinical evolu- Furthermore, serum HBV-RNA correlated with iHBV-DNA and tion during follow-up. The overall reduction in HBsAg levels was -0.22 -(0.58–0.16) log IU/ml during follow-up with no differences iHBV-RNA (rho = 0.50, p = 0.008 and rho = 0.56, p = 0.01, respectively). We did not observe any correlation between the noted between virological controllers and patients re-starting duration of prior NA therapy and the expression of any virolog- therapy (Fig. 1B). The evolution of the remaining markers (HBcrAg and serum HBV-RNA) is shown in Table 1 and ical parameter tested (data not shown). Supplementary Table S1. Low baseline HBsAg levels are associated with HBsAg loss Baseline HBsAg significantly correlates with iHBV-DNA and upon NA withdrawal iHBV-RNA, but not cccDNA Patients achieving HBsAg loss had lower HBsAg and iHBV-DNA Serum and liver parameters of HBV replication were analysed but similar cccDNA levels at baseline, compared with the in- before treatment interruption (Table 1 and Supplementary dividuals sustaining virological control and those requiring re- Table S1). The median baseline HBsAg titre in our cohort was treatment (Table 1 and Fig. 3A). Importantly, 5 out of 7 (71%) 1068 Journal of Hepatology 2021 vol. 74 j 1064–1074 A CD8+ T cells CD4+ T cells CD107a IFNγ TNFα CD107a IFNγ TNFα 100 * 3 3 1 1 6 5 1 Response against: 7 7 80 2 2 3 2 3 OLP pools Patients (%) 1 1 8 2 OLP pools 6 60 4 1 OLP pools 6 8 0 OLP pools 1 1 11 40 4 4 6 5 2 4 3 2 3 20 7 2 2 5 6 5 6 5 1 0 Re-treated Off-therapy Re-treated Off-therapy Re-treated Off-therapy Re-treated Off-therapy Re-treated Off-therapy Re-treated Off-therapy OLP pools: Core Env Pol B 10 * ** C 1.2 * ** 1.2 Off-therapy 8 Re-treated CD8+IFNγ+TNFα+ (%) CD4+IFNγ+TNFα+ (%) 0.9 0.9 CD8+CD107a+ (%) 6 4 0.6 0.6 2 0.3 0.3 2.0 0.3 0.3 1.5 0.2 0.2 1.0 0.5 0.1 0.1 0.0 0.0 0.0 BL W12 W48 BL W48 BL W48 Fig. 4. HBV-specific T cell responses at baseline and during follow-up. (A) Percentage of patients with baseline HBV-specific T cell responses (> −0.1% positive cells after subtracting the unstimulated control) and hierarchy of the T cell responses according to the different overlapping peptide (OLP) pools. (B) Degra- nulating CD8+ T cells after in vitro expansion with core OLP. (C) Polyfunctional T cells co-expressing IFNc and TNFa after core OLP stimulation at baseline and 48 weeks after treatment withdrawal. Bars represent the median and dotted lines the 0.1% cut-off for positive T cell responses minus the unstimulated control (B and C). Chi-square test (A) and Welch’s t test (B, C) were used to compare groups, *p <0.05, **p <0.01. patients who achieved HBsAg loss had undetectable iHBV-RNA at Supplementary Fig. S4. Interestingly, an increased proportion of baseline (p <0.01), whereas 91% of patients who remained patients that remained off-therapy had functional HBV-specific HBsAg-positive had detectable levels of iHBV-RNA (Table 1 and CD8+ T cell responses against epitopes from more than one Fig. 3B). In addition, baseline HBcrAg and serum HBV-RNA were HBV protein before NA withdrawal compared to those requiring more frequently undetectable in patients who achieved HBsAg NA re-introduction (68% vs. 20% p = 0.048 for IFNc production loss than in patients who did not (HBcrAg: 75% vs. 42%; p = 0.12 and 77% vs. 40% p = 0.099 for CD107a expression, respectively. and HBV-RNA: 88% vs. 47%; p = 0.053), although the differences Of note, we defined an HBV-specific T cell response as > −0.1% did not reach statistical significance. As shown in Fig. 3C, the after subtraction of the unstimulated control; Fig. 4A). Inter- probability of HBsAg loss was higher among patients with low estingly, for both CD8+ and CD4+ T cells, responding cells to all HBsAg < −1,000 IU/ml (log rank 9.9, p = 0.002). Combining HBsAg viral antigens were more likely to be detectable at baseline in levels < −1,000 IU/ml and undetectable serum HBV-RNA did not patients who remained off-therapy than in the re-treatment improve the predictive value for HBsAg loss (log rank 6.7, p = group. Indeed, CD4+ and CD8+ T cells producing IFNc and 0.009; Supplementary Fig. S2). Although virological parameters TNFa in response to any peptide pool were absent in all pa- predicted HBsAg loss, we did not see differences between the tients that had NA re-introduction. However, no statistically virological control and relapse groups. significant differences were observed within groups likely attributable to the number of patients falling in the re- HBV-specific T cell responses at baseline are associated with treatment group. virological control off-treatment The percentage of degranulating CD8+ T cells (CD107a) in To evaluate the impact of the HBV-specific T cell response on response to stimulation with HBV-core OLP was significantly the clinical outcome after NA discontinuation, we investigated higher at baseline and after 12 weeks of stopping NA treatment the effector capacity of HBV-specific CD4+ and CD8+ T cells after in patients remaining off-therapy than in those requiring NA re- in vitro stimulation with overlapping peptides (OLP) spanning introduction (p = 0.039 and 0.0093, respectively; Fig. 4B). How- the core, envelope, or polymerase proteins. A representative ever, these differences in CD107a responses between patient gating strategy and example flow cytometric plots are shown in groups were not maintained when focusing on OLP encom- Supplementary Fig. S3 and individual data showing CD8+ and passing the envelope and polymerase proteins (Supplementary CD4+ T cell responses at all time points are shown in Fig. S5). Journal of Hepatology 2021 vol. 74 j 1064–1074 1069 Research Article Viral Hepatitis A CD8+ T cells (all patients) CD4+ T cells (all patients) CD8+ T cells (off-therapy) CD4+ T cells (off-therapy) CD8+ T cells (re-treated) CD4+ T cells (re-treated) CD107a 3 3 3 3 3 3 IFNγ TNFα Expression (%) Expression (%) Expression (%) Expression (%) Expression (%) Expression (%) 2 2 2 2 2 2 1 1 1 1 1 1 0 0 0 0 0 0 BL W12 W48 BL W12 W48 BL W12 W48 BL W12 W48 BL W12 W48 BL W12 W48 B CD8+ T cells CD4+ T cells CD107a IFNγ TNFα CD107a IFNγ TNFα 1.0 BL Off-therapy 0.8 W12 W48 0.6 BL 0.4 Re-treated W12 0.2 W48 0.0 Core Env Pol Core Env Pol Core Env Pol Core Env Pol Core Env Pol Core Env Pol Fig. 5. HBV-specific T cell responses do not increase after NA withdrawal. (A) Longitudinal analysis of HBV-specific T cell responses after stopping therapy (entire cohort and broken down by clinical outcome: off-therapy and re-treated). Data are expressed as mean ± SEM. (B) Heat map showing normalised median frequencies of HBV-specific T cell responses against each OLP at baseline (BL), week 12 (W12) and week 48 (W48) after treatment discontinuation. Normalisation was performed by scaling maximum and minimum values per marker between 0 and 1. Absence of * indicates no differences compared to baseline in each group (Wilcoxon Signed-Rank test). Env, envelope; NA, nucleos(t)ide analogue; Pol, polymerase. When we investigated the presence of polyfunctional CD8+ significantly change during follow-up compared to baseline and CD4+ T cells (cells co-expressing IFNc and TNFa), we (Fig. 5A and Supplementary Fig. S4). Similarly, when classifying observed that the percentage of core-specific CD8+ T cells pro- patients according to whether they remained off-therapy or ducing IFNc/TNFa at baseline was increased among patients who required treatment re-introduction, no significant changes in any remained off-therapy compared with those patients needing effector functions to any of the peptide pools were detected at treatment re-introduction (p = 0.031; Fig. 4C). Importantly, this week 12 or 48 compared with baseline (Fig. 5B). increase persisted for more than a year of follow-up whilst off- As HBsAg loss with or without the production of anti-HBs therapy (p = 0.01). A similar trend was observed when probing represents the best predictor of success upon discontinuation effector CD4+ T cells co-producing IFNc/TNFa, however, the dif- of NA therapy, we analysed HBV-specific CD8+ and CD4+ T cell ferences between outcome groups did not reach statistical sig- responses within this small group of individuals. Notably, the nificance (Fig. 4C). Likewise, no differences were observed effector CD8+ T cell response to antigenic stimulation in these between groups of patients (off-therapy and treatment re- patients was heterogeneous, with a significantly greater magni- introduction) regarding the presence of polyfunctional T cell tude of response to HBV OLP stimulation in some patients, and responses against envelope and polymerase proteins with absent or very weak responses for all mediators (CD107a, (Supplementary Fig. S6). IFNc or TNFa) in others (Fig. 6A). The longitudinal analysis after We also evaluated if there was any association between T cell NA cessation did not show significant changes in the response responses and the development of clinically relevant ALT flares rates in this subgroup of patients (Fig. 6B). Similar results were during follow-up. As shown in Supplementary Fig. S7, we obtained for HBV-specific CD4+ T cell responses (data not observed that HBV-specific CD8+ and CD4+ T cell responses at shown). baseline were similar between patients regardless of whether they developed an ALT flare (using a threshold of >−2 ULN). The strength of HBV-specific T cell responses does not correlate with serum or intrahepatic virological markers HBV-specific responses do not increase following treatment We next explored whether differences in the HBV-specific T cell withdrawal response were associated with serum or intrahepatic virological To investigate the impact of NA discontinuation on functional parameters. As we noted, baseline HBsAg levels < −1,000 IU/ml were HBV-specific T cells, we performed a longitudinal analysis of the predictive of HBsAg clearance in our cohort, we classified patients T cell responses against HBV OLP at 12 and 48 weeks after NA accordingly and analysed their HBV-specific CD8+ and CD4+ T cell withdrawal. We found that the frequency of either HBV-specific responses at baseline. The proportion of patients with HBV- CD8+ or CD4+ T cells detectable by intracellular cytokine staining specific effector T cells as well as the frequency of polyfunctional for the antiviral mediators CD107a, IFNc, and/or TNFa did not T cells co-producing IFNc/TNFa were similar regardless of HBsAg 1070 Journal of Hepatology 2021 vol. 74 j 1064–1074 A Patient 8 (high responses) Patient 11 (low responses) BL W12 W48 BL W12 W48 3.28 3.00 0.39 2.02 0.48 0.46 0.39 CD107a CD107a 4.10 2.94 2.03 0.08 0.07 0.12 IFNγ IFNγ 1.12 1.46 3.71 0.28 0.11 0.11 CD8 CD8 TNFα TNFα B 6 6 6 4 4 4 2 2 2 CD8 CD107a (%) 2.0 CD8+TNFα+ (%) 2.0 2.0 CD8+IFNγ+ (%) + 1.5 1.5 1.5 1.0 1.0 1.0 + 0.5 0.5 0.5 0.0 0.0 0.0 BL W12 W48 BL W12 W48 BL W12 W48 Fig. 6. HBV core-specific CD8+ T cell responses in patients who achieved HBsAg loss after treatment withdrawal. (A) Representative dot plots of HBV-specific effector CD8+ T cells from 2 patients who lost HBsAg. (B) Longitudinal HBV-specific T cell responses against core OLP in patients that achieved HBsAg loss after stopping therapy. No differences were observed at weeks 12 and 48 when comparing to baseline (Wilcoxon Signed-Rank test). titres (Supplementary Fig. S8). We did not find any significant remains low.1,2 Several new direct-acting antivirals and immu- correlation between peripheral HBV-specific T cell responses and nomodulatory compounds are in clinical evaluation with the aim other serum (e.g. HBcrAg) or liver replication markers (iHBV-DNA, of achieving functional cure of CHB with finite treatment dura- cccDNA, iHBV-RNA, or iHBV-RNA/cccDNA ratio). tion.15 Alternatively, NA discontinuation has been proposed as a promising therapeutic strategy to promote HBsAg loss or at least Safety sustained off-treatment viral control in a considerable propor- Importantly, no patient developed hepatic decompensation, he- tion of patients.2–5,16 However, the reported rates of HBsAg loss patocellular carcinoma, or died during follow-up. Bilirubin levels in NA withdrawal studies have been enormously variable, in part greater than 2 mg/dl or prolongation of prothrombin time were owing to the heterogeneity of patient cohorts, and the absence of not detected at any time point, even amongst those individuals standardised treatment re-introduction criteria. To investigate requiring treatment re-introduction. No other adverse events the correlates of outcomes upon NA cessation, we analysed liver were detected during the follow-up period. All patients requiring and serum markers of HBV replication in parallel with assess- NA re-introduction had a subsequent good response with ente- ment of HBV-specific T cell immunity in a well-characterised cavir or tenofovir, returning to undetectable serum HBV-DNA cohort of patients with HBeAg-negative CHB undergoing NA and a normal ALT within 6 months. Furthermore, we observed discontinuation. no change in liver stiffness assessment at the 24-month follow- In our study, after a median follow-up of 2.8 years from up in 19 of our 27 patients compared with their matched base- treatment interruption, 82% of patients remain off-therapy and 8 line assessment or when comparing treatment outcome groups (30% of the total cohort) continued to remain HBsAg negative. (Table 1). Virological relapses were observed in the majority (78%) of pa- tients although these were not always associated with Discussion biochemical relapses. Indeed, most of the ALT flares were mild Despite lifelong treatment with NA being safe and effective in and temporary, leading to long-term virological and biochemical patients with HBeAg-negative CHB, the overall rate of HBsAg loss remission with no patients developing liver dysfunction. Our Journal of Hepatology 2021 vol. 74 j 1064–1074 1071 Research Article Viral Hepatitis results agree with those reported by Berg et al.,3 the only rand- control off-treatment. In both studies, ALT flares were not asso- omised trial comparing the strategy of NA-therapy maintenance ciated with significant increases in HBV-specific CD8+ T cell and therapy withdrawal, which included a comparable popula- frequency (in the Rivino et al.13 study they were lower in the tion in terms of genotypic and clinical characteristics. flaring group), raising the possibility of alternative immune Although treatment re-introduction criteria differ among the mechanisms driving liver damage. current published studies, we and others have shown that lower Finally, our study also has some limitations. First, although an HBsAg titres at the time of NA withdrawal are associated with an HBsAg threshold was not an inclusion criterion, our cohort is increased probability of HBsAg loss.3,4,6,17–19 In our study, a characterised by a proportion of patients with low HBsAg levels threshold value for HBsAg of < −1,000 IU/ml was a good predictor at baseline which might have influenced the improved success whether a patient went on to lose HBsAg. Likewise, patients who rates in this study. Nonetheless, when excluding the 2 patients achieved HBsAg loss had lower iHBV-DNA levels at baseline; this with lower qHBsAg (<1 IU/ml), the statistical analysis regarding marker includes cccDNA, relaxed circular DNA, and integrated baseline differences and predictive factors remained the same. HBV-DNA. Indeed, integration of HBV-DNA into the host genome Second, because of the limited amount of liver tissue, we were and its contribution to the formation of HBsAg particles20 may not able to assess HBV-DNA integration or to study innate im- explain the good correlation between iHBV-DNA and HBsAg in mune responses within the liver,26 or the contribution of liver- our cohort of HBeAg-negative patients chronically suppressed by resident T cells to viral control.27–29 Lastly, we cannot rule out NA therapy. future transition to a different HBV infection phase in patients We did not find an association between baseline cccDNA with virological control, thus these patients remain under clin- levels and virological outcome. However, most patients ical monitoring. achieving HBsAg loss had undetectable iHBV-RNA (71%) and, In summary, stopping NA therapy is feasible in a high pro- hence, reduced cccDNA transcriptional activity (iHBV-RNA/ portion of HBeAg-negative patients without cirrhosis. Although cccDNA ratio). This would explain, at least in part, a more lower cccDNA transcriptional activity after long-term antiviral favourable outcome in these patients. Indeed, 10 out of 11 pa- therapy is associated with successful outcome after NA with- tients not achieving HBsAg loss had detectable iHBV-RNA sug- drawal, intrahepatic viral parameters do not seem to improve the gesting transcriptionally active cccDNA and a higher risk of viral predictive capacity of readily available serum markers such as rebound. We also studied the prognostic value of other surrogate HBsAg. Indeed, low HBsAg levels (< −1,000 IU/ml) may offer pre- serum markers of cccDNA transcriptional activity such as HBcrAg dictive value in assessing those patients in whom no specific and serum HBV-RNA.7,21,22 The proportion of patients with un- intervention other than stopping antiviral therapy and close detectable serum HBV-RNA or HBcrAg at baseline in our cohort monitoring would be a valid therapeutic strategy for achieving was higher among patients achieving HBsAg loss (7/8 and 6/8 functional cure. For those patients with a less favourable viro- patients, respectively), however, these markers did not improve logical profile, the strength of HBV-specific T cell responses at the ability of HBsAg levels to predict functional cure. Impor- baseline may contribute to the outcome of treatment interrup- tantly, serum HBV-RNA and HBcrAg may have become unde- tion. The latter group may be more responsive to novel finite tectable in HBeAg-negative patients under long-term NA therapy immunotherapeutic options after NA discontinuation to increase thus limiting their predictive power.7,21 the rate of HBsAg loss. As adaptive immune responses are important for the control of HBV infection,12,23,24 we analysed the effector functionality of Abbreviations HBV-specific CD4+ and CD8+ T cells at the time of treatment ALT, alanine aminotransferase; BL, baseline; cccDNA, covalently withdrawal (baseline) and throughout NA discontinuation. At closed circular DNA; CHB, chronic hepatitis B; Env, envelope; baseline, the proportion of patients with HBV-specific CD8+ T HBcrAg, HBV core-related antigen; iHBV-DNA, intrahepatic HBV- cells expressing CD107a and IFNc after in vitro peptide stimula- DNA; iHBV-RNA, intrahepatic HBV-RNA; LR, likelihood ratio; tion was higher amongst patients who subsequently remained NPV, negative predictive value; NA, nucleos(t)ide analogue; OLP, off-therapy compared with those that needed treatment re- overlapping peptides; PBMC, peripheral blood mononuclear introduction. The presence of polyfunctional T cells has been cells; Pol, polymerase; PPV, positive predictive value; rho, associated with an improved control of viral replication.23,25 Spearman’s correlation coefficient; Se, sensibility; Sp, specificity; Accordingly, we found that the frequency of HBV-specific T cell ULN, upper limit of normal; W12, week 12; W48, week 48. co-producing IFNc/TNFa were higher both at baseline and 1 year after stopping therapy in patients who remained off-therapy. Financial support However, in contrast to recent findings,14 treatment with- This study was sponsored by the Instituto de Salud Carlos III drawal did not induce any significant change in the HBV-specific (ISCIII) through the Plan Estatal de Investigación Científica y CD8+ and CD4+ T cell response after 12 or 48 weeks of follow-up. Técnica y de Innovación 2013–2016, grants PI16/00111 (SPP), Interestingly, patients achieving HBsAg loss presented a het- PI15/00151 and PI18/00079 (XF), and PI18/01436 (FRF), co- erogeneous HBV-specific effector T cell response both at baseline funded by the European Regional Development Fund (ERDF), and during follow-up, with no clear pattern differentiating those by the Gilead Fellowship Programme, grant GLD15/00274 (XF) patients achieving HBsAg loss after an ALT flare or not. In line and by Plan Estratégico Nacional contra la hepatitis C (Spanish with the concept that HBV-specific CD8+ T cells can exert viral Health Ministry). XF has received support from Secretaria control without liver damage,24 a recent study showed higher d’Universitats i Recerca del Departament d’Economia i Coneixe- baseline frequencies of core and polymerase-specific CD8+ T cells ment (grant 2017_SGR_1753) and CERCA Programme/Generalitat were predictive of viral control in the absence of a flare after de Catalunya. SL has received grants from Asociación Española treatment interruption.13 Our findings similarly show that higher para el Estudio del Hígado (AEEH) and Societat Catalana de baseline HBV-specific CD8+ T cells were predictive of virological Digestologia (SCD). GK is supported by Fundación IBEROSTAR. 1072 Journal of Hepatology 2021 vol. 74 j 1064–1074 MGL is supported by the i-PFIS program (fellowship IFI18/00006) [3] Berg T, Simon KG, Mauss S, Schott E, Heyne R, Klass DM, et al. Long-term and SRT by the Rio Hortega program (fellowship CM17/00015) of response after stopping tenofovir disoproxil fumarate in non-cirrhotic HBeAg-negative patients – FINITE study. J Hepatol 2017;67:918–924. the ISCIII, both co-funded by the European Social Fund (ESF). SRT [4] Hadziyannis SJ, Sevastianos V, Rapti I, Vassilopoulos D, Hadziyannis E. is supported by an initiation research grant from SCD and the Sustained responses and loss of HBsAg in HBeAg-negative patients with Emili Letang end-of-residency prize from Hospital Clínic de chronic hepatitis B who stop long-term treatment with adefovir. Barcelona. MKM and LJP are supported by Wellcome Trust Gastroenterology 2012;143. 629–636.e1. [5] Papatheodoridis GV, Manolakopoulos S, Su TH, Siakavellas S, Liu CJ, Investigator award 214191/Z/18/Z and CRUK Immunology grant Kourikou A, et al. Significance of definitions of relapse after discontinu- 26603. This work was performed within the framework of the ation of oral antivirals in HBeAg-negative chronic hepatitis B. Hepatology RHU CirB-RNA (ANR-17-RHUS-0003) that is part of the second 2018;68:415–424. programme ‘Investissements d’Avenir’ operated by the French [6] Jeng WJ, Chen YC, Chien RN, Sheen IS, Liaw YF. Incidence and predictors of National Research Agency (ANR) (FZ and BT). hepatitis B surface antigen seroclearance after cessation of nucleos(t)ide analogue therapy in hepatitis B e antigen–negative chronic hepatitis B. Hepatology 2018;68:425–434. Conflicts of interest [7] Papatheodoridi M, Hadziyannis E, Berby F, Zachou K, Testoni B, SL has received lecture and advisory fees from Gilead, Abbvie, Rigopoulou E, et al. Predictors of hepatitis B surface antigen loss, relapse and retreatment after discontinuation of effective oral antiviral therapy in and MSD. XF has acted as advisor for Gilead and Abbvie. SRT has noncirrhotic HBeAg-negative chronic hepatitis B. J Viral Hepat received lecture fees from Gilead and Abbvie. ZM has received 2019;27:118–126. fees for lectures and consulting from Gilead and Abbvie, and a [8] Ha M, Zhang G, Diao S, Lin M, Sun L, She H, et al. A prospective clinical research grant from Gilead. The Maini lab has received research study in hepatitis B e antigen-negative chronic hepatitis B patients with funding from Gilead, Hoffmann La Roche and Immunocore. MKM stringent cessation criteria for adefovir. Arch Virol 2012;157:285–290. [9] Chi H, Hansen BE, Yim C, Arends P, Abu-Amara M, Van Der Eijk AA, et al. has sat on advisory boards/provided consultancy for Gilead, Reduced risk of relapse after long-term nucleos(t)ide analogue consoli- Hoffmann La Roche, Immunocore, VIR, Galapagos NV, GSK, dation therapy for chronic hepatitis B. Aliment Pharmacol Ther Abbvie, and Freeline. LJP has received fees from Gilead for con- 2015;41:867–876. sultancy work. FZ received research grants and consultancy [10] Jung KS, Park JY, Chon YE, Kim HS, Kang W, Kim BK, et al. Clinical out- comes and predictors for relapse after cessation of oral antiviral treatment honoraria from Hoffmann-La-Roche, Gilead Sciences, Janssen, in chronic hepatitis B patients. J Gastroenterol 2016;51:830–839. Assembly Biosciences, Arbutus, Contravir, Sanofi, and Transgene. [11] van Bömmel F, Bartens A, Mysickova A, Hofmann J, Krüger DH, Berg T, All other authors have declared no conflicts of interest. et al. Serum hepatitis B virus RNA levels as an early predictor of hepatitis B Please refer to the accompanying ICMJE disclosure forms for envelope antigen seroconversion during treatment with polymerase in- further details. hibitors. Hepatology 2015;61:66–76. [12] Maini MK, Pallett LJ. Defective T-cell immunity in hepatitis B virus infection: why therapeutic vaccination needs a helping hand. Lancet Authors’ contributions Gastroenterol Hepatol 2018;3:192–202. Contributed to the study concept and design: XF, SL, SPP. Patient [13] Rivino L, Bert N Le, Gill US, Kunasegaran K, Cheng Y, Tan DZMM, et al. selection/inclusion: SL, SRT, ZM, XF. Acquired data: MGL, SL, SRT, Hepatitis B virus–specific T cells associate with viral control upon nucleos(t)ide-analogue therapy discontinuation. J Clin Invest 2018;128:1– EGP, EP, TL, GK, CB, FRF, ZM, XF, SPP. Contributed to analysis and 14. interpretation of data: MGL, SL, XF, SPP, JJL, FZ, BT, LJP, MKM. [14] Rinker F, Zimmer CL, Siederdissen CHZ, Manns MP, Kraft ARM, Drafted the manuscript: SL, MGL, SPP. Contributed to critical Wedemeyer H, et al. Hepatitis B virus-specific T cell responses after revisions and approved the final manuscript: all authors. stopping nucleos(t)ide analogue therapy in HBeAg negative chronic hepatitis B. J Hepatol 2018;69:584–593. [15] Cornberg M, Lok ASF, Terrault NA, Zoulim F, Berg T, Brunetto MR, et al. Data availability Guidance for design and endpoints of clinical trials in chronic hepatitis B Data regarding patients included in the study are confidential. – Report from the 2019 EASL-AASLD HBV Treatment Endpoints Confer- Additional information relevant to readers is provided as ence. J Hepatol 2020;72:539–557. [16] Lampertico P, Berg T. Less can be more: a finite treatment approach for Supplementary data. HBeAg-negative chronic hepatitis B. Hepatology 2018;68:397–400. [17] Cao J, Chi H, Yu T, Li Z, Hansen BE, Zhang X, et al. Off-treatment hepatitis B virus (HBV) DNA levels and the prediction of relapse after discontinuation Acknowledgements of nucleos(t)ide analogue therapy in patients with chronic hepatitis B: a We are indebted to the Cytometry and Cell Sorting facility of the prospective stop study. J Infect Dis 2017;215:581–589. Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDI- [18] Papatheodoridis G, Rigopoulou E, Papatheodoridi M, Zachou K, BAPS) and the Peptide Synthesis Core Facility of the Pompeu Xourafas V, Gatselis N, et al. DARING-B: discontinuation of effective entecavir or tenofovir therapy in non-cirrhotic HBeAg-negative chronic Fabra University (UPF), Barcelona, Spain for the technical help. hepatitis B patients: a prospective Greek study. J Hepatol 2017;66:S26. [19] Chen CH, Lu SN, Hung CH, Wang J-HH, Hu T-HH, Changchien CS, et al. The Supplementary data role of hepatitis B surface antigen quantification in predicting HBsAg loss Supplementary data to this article can be found online at https:// and HBV relapse after discontinuation of lamivudine treatment. J Hepatol 2014;61:515–522. doi.org/10.1016/j.jhep.2020.11.043. [20] Wooddell CI, Yuen MF, Chan HLY, Gish RG, Locarnini SA, Chavez D, et al. Rnai-based treatment of chronically infected patients and chimpanzees References reveals that integrated hepatitis B virus DNA is a source of HBsAg. Sci Author names in bold designate shared co-first authorship Transl Med 2017;9:1–12. [21] Carey I, Gersch J, Wang B, Moigboi C, Kuhns M, Cloherty G, et al. Pre- genomic HBV RNA and HBcrAg predict outcomes in HBeAg negative [1] Marcellin P, Wong DK, Sievert W, Buggisch P, Petersen J, Flisiak R, et al. chronic hepatitis B patients suppressed on nucleos(t)ide analogue ther- Ten-year efficacy and safety of tenofovir disoproxil fumarate treatment apy. Hepatology 2019;0–1. for chronic hepatitis B virus infection. Liver Int 2019;39:1868–1875. [22] Wang J, Shen T, Huang X, Renuka Kumar G, Chen X, Zeng Z, et al. Serum [2] European Association for the Study of the Liver. EASL 2017 Clinical Prac- hepatitis B virus RNA is encapsidated pregenome RNA that may be tice Guidelines on the management of hepatitis B virus infection. J Hepatol 2017;67:370–398. Journal of Hepatology 2021 vol. 74 j 1064–1074 1073 Research Article Viral Hepatitis associated with persistence of viral infection and rebound. J Hepatol [26] Lebossé F, Testoni B, Fresquet J, Facchetti F, Galmozzi E, Fournier M, et al. 2016;65:700–710. Intrahepatic innate immune response pathways are downregulated in [23] Boni C, Laccabue D, Lampertico P, Giuberti T, Viganò M, Schivazappa S, untreated chronic hepatitis B. J Hepatol 2017;66:897–909. et al. Restored function of HBV-specific T cells after long-term effective [27] Pallett LJ, Davies J, Colbeck EJ, Robertson F, Hansi N, Easom NJW, et al. IL- therapy with nucleos(t)ide analogues. Gastroenterology 2012;143. 2(high) tissue-resident T cells in the human liver: sentinels for hepato- 963–73.e9. tropic infection. J Exp Med 2017;214:1567–1580. [24] Maini MK, Boni C, Lee CK, Larrubia JR, Reignat S, Ogg GS, et al. The role of [28] Gill US, Pallett LJ, Thomas N, Burton AR, Patel AA, Yona S, et al. Fine virus-specific CD8+ cells in liver damage and viral control during needle aspirates comprehensively sample intrahepatic immunity. Gut persistent hepatitis B virus infection. J Exp Med 2000;191:1269–1280. 2019;68:1493–1503. [25] Almeida JR, Price DA, Papagno L, Arkoub ZA, Sauce D, Bornstein E, et al. [29] Kim JH, Han JW, Choi YJ, Rha MS, Koh JY, Kim KH, et al. Functions of Superior control of HIV-1 replication by CD8+ T cells is reflected by their human liver CD69+ CD103- CD8+ T cells depend on HIF-2a activity in avidity, polyfunctionality, and clonal turnover. J Exp Med 2007;204:2473– healthy and pathologic livers. J Hepatol 2020;72:1170–1181. 2485. 1074 Journal of Hepatology 2021 vol. 74 j 1064–1074