Journal of Viral Hepatitis

ORIGINAL ARTICLE OPEN ACCESS

Predictive Factors for HBsAg Loss in Chronic

HBeAg-­Negative Hepatitis B Virus Infection: Insights From
a 5-­Year French Cohort
Xavier Causse1 | Pascal Potier1,2 | Antoine Valéry2 | Hélène Labadie3 | Gilles Macaigne4 | Jean-­François Cadranel5 |
Thierry Fontanges6 | Lina Mouna7 | Anne-­Marie Roque-­A fonso7 | for the PIBAC Study Group of Association Nationale des
Hépato-­Gastroentérologues des Hôpitaux Généraux (ANGH)

1Serviced'Hépato-­Gastroentérologie et d'Oncologie Digestive, Centre Hospitalier Universitaire d'Orléans, Orléans, France | 2Département d'Information
Médicale, Centre Hospitalier Universitaire d'Orléans, Orléans, France | 3Service d'Hépato-­Gastroentérologie, Centre Hospitalier de Saint Denis,
Saint Denis, France | 4Groupe Hospitalier Intercommunal Le Raincy-­Montfermeil, Service d'Hépato-­Gastroentérologie, Le Raincy Montfermeil,
France | 5Groupe Hospitalier Public du Sud de l'Oise, Service d'Hépato-­Gastroentérologie, Creil, France | 6Private Medical Practice, Maubec,
France | 7Inserm U1193, Assistance Publique-­Hôpitaux de Paris, Hôpital Paul Brousse, Service de Virologie, Université Paris-­Saclay, Villejuif, France

Correspondence: Anne-­Marie Roque-­A fonso (anne-marie.roque@aphp.fr)

Received: 10 September 2024 | Revised: 20 November 2024 | Accepted: 23 November 2024

Funding: The study was supported by the Association Nationale des Hépato-­Gastroentérologues des Hôpitaux Généraux (ANGH).

Keywords: HBcrAg | HBeAg-­negative chronic HBV infection | HBsAg | HBsAg loss prediction | HBV-­DNA | HBV genotype | HBV reactivation

ABSTRACT
Prognostic factors for the long-­term evolution of chronic hepatitis B e antigen (HBeAg)-­negative hepatitis B virus (HBV) infection
may vary depending on local epidemiology. We aimed to identify these factors in France, where the epidemiology is influenced by
diverse immigration. Hepatitis B surface antigen (HBsAg)-­positive, HBeAg-­negative adults with normal transaminase levels and
viral loads < 20,000 IU/mL for 1 year, without viral co-­infection or advanced liver disease, were enrolled for a 5-­year follow-­up.
A total of 564 patients were recruited from 23 centres (54.4% women, mean age 42.3 ± 12 years, 47.7% from sub-­Saharan Africa).
HBV DNA was detectable but < 2000 IU/mL for most (71.3%). Genotypes E (27.8%) and A (20.0%) were predominant. The mean
HBsAg titre was 3.8 ± 3.4 log IU/mL, > 1000 IU/mL in 60% of cases, and higher in genotype E (p < 0.0001). During follow-­up,
18 patients received antiviral treatment, 9 for viral reactivation (0.3% per year) and 9 preemptively. HBsAg loss occurred in 39
patients (1.4% per year). These patients were older (p < 0.0001), more frequently treated for dyslipidemia, hypertension or diabe-
tes (p < 0.05), and had lower baseline HBV DNA (p = 0.0112) and HBsAg (p < 0.0001), but similar levels of HBcrAg compared to
those who did not clear HBsAg. Baseline HBsAg was the only independent predictor of HBsAg loss (p = 0.009). In this cohort,
HBsAg < 153 IU/mL predicted clearance with 87% sensitivity and specificity. In conclusion, baseline HBsAg accurately predicted
seroclearance at 5 years in patients with chronic HBeAg-­negative infection, regardless of genotype, sex, or geographical origin,
indicating that this marker is widely applicable for reducing the frequency of patient monitoring.

Abbreviations: ALT, alanine aminotransferase; ANGH, Association Nationale des Hépato-­Gastroentérologues des Hôpitaux Généraux; BMI, body mass index; cccDNA, covalently closed and
circular DNA; EASL, European Association for the Study of the Liver; HBcrAg, hepatitis B core-­related antigen; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis
B virus; ROC curve, receiver operating characteristic curve.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original
work is properly cited, the use is non-commercial and no modifications or adaptations are made.

© 2024 The Author(s). Journal of Viral Hepatitis published by John Wiley & Sons Ltd.

Journal of Viral Hepatitis, 2025; 32:e14041 1 of 9

https://doi.org/10.1111/jvh.14041
Hepatitis B remains a public health problem in France, de- 1   |   Patients and Methods
spite availability of effective vaccines and treatments, which
currently only allow for a functional cure [1, 2]. The natural PIBAC is a multicentre cohort study involving 23 French and
history of chronic hepatitis B virus (HBV) infection has been 1 Belgian centres, in general and university hospitals, and in
divided into five phases, according to the presence of hepati- private practice. HBsAg positive and HBeAg negative adults
tis B e antigen (HBeAg), HBV DNA levels and alanine amino- with normal aminotransferases levels (ALT ≤ 40 IU/L) and
transferase (ALT) values. In most cases, multiple assessments HBV DNA < 20,000 IU/mL, repeated every 3 or 4 months for
of ALT levels and viral markers are required to define the at least 1 year were prospectively included. Other inclusion cri-
infection phase. HBeAg-­negative chronic HBV infection, for- teria were the absence of Hepatitis C, Hepatitis D or Human
merly ‘inactive carrier’ phase, is characterised by the presence Immunodeficiency Virus infection and the absence of signs of
of serum antibodies to HBeAg (anti-­H Be), undetectable or advanced liver disease, whether clinical, biological, ultrasono-
low (< 2000 IU/mL) HBV DNA levels and persistently normal graphic or elastographic. Transient elastography has been vali-
ALT in patients without viral coinfection and mild or absent dated to distinguish HBV patients with significant liver fibrosis
liver fibrosis. Some patients in this phase may have HBV DNA from those with minimal or absent fibrosis [19].
levels > 2000 IU/mL, and < 20,000 IU/mL [3]. Patients with
HBeAg-­negative infection have a low risk of progression to Data collected included sex, age, geographical origin, body
cirrhosis (< 0.1 per 100 persons/year) or hepatocellular car- mass index (BMI) and use of antihypertensive, antidiabetic
cinoma (0.05 per 100 persons/year), and little change in life or lipid-­lowering medications. Alcohol, tobacco and cannabis
expectancy (mortality 0.03 per 100 persons/year). Hepatitis consumption were collected at inclusion using an open-­ended
B surface antigen (HBsAg) loss and/or seroconversion may questionnaire. Patients were followed for 5 years, with an-
occur spontaneously in 0.7%–2.3% per year, but progression nual collection of clinical, biological, ultrasound and elasto-
to chronic hepatitis B may also occur in 0%–10% of cases, de- graphic data.
pending on the cohort [4–14]. Hence, long-­term surveillance
is recommended for this heterogeneous population. The 2017
guidelines of the European Association for the Study of the 1.1   |   Virological Analyses
Liver (EASL) advised to monitor ALT every 6–12 months in
patients with B viral load < 2000 IU/mL, with HBV DNA and HBsAg titre was determined using Elecsys* HBsAg II quant
liver fibrosis assessment every 2–3 years, depending on HBsAg II reagent on a Cobas 6000 system (Roche Diagnostics), with a
quantification. Those with HBV DNA > 2000 UI/mL need measurement range of 0.05 to > 52,000 IU/mL. The threshold
closer monitoring with ALT determination every 3 months for for HBsAg negativity was < 0.05 IU/mL. HBcrAg titre was de-
the first year and every 6 months thereafter, with yearly HBV termined using Lumipulse* G HBcrAg reagent (Fujirebio Inc.,
DNA and fibrosis assessment. If no treatment indication arises Japan, Research use only) on a Lumipulse G600II automated
during the first 3 years of surveillance, the follow-­up is then immunoassay system. The result was given in log10 arbitrary
the same as for patients with HBV DNA < 2000 IU/mL [3]. units (U)/mL: < 2 = negative, between 2 and 3 = detectable, non-­
quantifiable, from 3 to 7 = quantification range. HBV genotype
The correct and early identification of the stage of HBV disease was determined by sequencing and phylogenetic analysis of 600
has been the subject of much research. Indeed, the early identi- nucleotides encompassing the polymerase and surface genes.
fication of HBeAg-­negative infection would avoid the need for These analyses were carried out centrally at the virology labo-
close monitoring, particularly during the first year. In addition, ratory of the Paul Brousse Hospital in Villejuif. The viral load
identifying among patients with HBeAg-­negative infection those required for patient management was carried out on an ongoing
who will lose their HBsAg would further simplify follow-­up, basis at each centre.
given the very low risk of disease progression [15]. Typically,
such patients have > 15 years of chronic HBV infection [16], and
low HBsAg levels. Threshold values predicting HBsAg loss have 1.2   |   Statistical Analysis
been proposed, but may differ according to HBV genotype. In
Asian populations infected with genotypes B and C, HBsAg Three evolutionary profiles were defined: (1) Unfavourable:
level < 100 IU/mL was associated with HBsAg loss during fol- diagnosis of cirrhosis or liver cancer, HBV-­related death, or
low-­up, while in European populations infected with genotype HBV reactivation; (2) Favourable: HBsAg clearance; (3) Stable:
D, the threshold was < 1000 IU/mL [17]. The magnitude of absence of any of the events listed above. To investigate fac-
HBsAg titre reduction over time, and low or undetectable HBV tors associated with HBsAg loss, a univariate analysis was
RNA and Hepatitis B core-­related antigen (HBcrAg) levels were performed using chi-­2 or Fisher tests for categorical variables,
also predictive of HBsAg loss, but accurate cut-­offs have not yet and Student's t-­test for continuous variables. After interaction
been proposed [18]. analysis of the model variables and imputation of missing
data by the MICE package, a multivariate logistic regression
The aim of our study was to investigate the 5-­year outcome of analysis was conducted to identify independent predictors of
patients with HBeAg-­negative chronic HBV infection to identify HBsAg loss. Receiver operating characteristic (ROC) curves
factors predictive of HBsAg loss. The analysis of a French cohort were constructed to compare the respective diagnostic perfor-
seems valuable because French epidemiology differs from that mance of baseline HBV DNA and HBsAg titres in predicting
of Asia, Italy, or the United States, where similar studies have HBsAg loss. The area under the curve (AUC) and 95% con-
been conducted. fidence intervals were calculated. Statistical analysis was

2 of 9 Journal of Viral Hepatitis, 2025

performed using RStudio 2023.09 and Analyse-­it for Microsoft Among patients followed up to 5 years, 300 had no need for anti-
Excel 6.15.4. viral treatment, nor HBsAg clearance. Year-­5 HBsAg titre, avail-
able for 218 of these patients, was significantly lower than at
baseline, 3.8 ± 3.9 log IU/mL vs. 3.6 ± 3.8 log IU/mL (p < 0.0001),
2   |   Results which is a mean decrease of 0.2 log at 5 years.

From September 2014 to June 2016, 566 patients were pre-­

selected as ‘inactive carriers’. Of them, 2 were excluded due to 2.3   |   Factors Associated With HBsAg Clearance
elastography value > 10 kPa, indicating significant fibrosis, and
564, 1–100 per centre, were included. Table 1 compares the baseline characteristics of the 39 patients
with HBsAg clearance with those of the 300 patients who re-
mained HBsAg-­ positive without initiation of antiviral ther-
2.1   |   Baseline Characteristics apy. Patients with HBsAg loss were 10 years older (p < 0.0001),
their HBsAg level was > 10 times lower (p < 0.0001), and their
Fifty-­four percent were women, with a mean age of viral load was > 2 times lower (p = 0.0112). They were also
42.3 ± 12 years (Table 1). Sub-­S aharan Africa was the most fre- more often treated for dyslipidaemia (p = 0.0004), hypertension
quent area of origin (47.7%). The mean BMI was 25.7 ± 4.5 kg/ (p = 0.0056), or diabetes (p = 0. 0372). Viral genotype, HBcrAg
m 2 , 1.2% of men reported alcohol consumption of over 40 g/ level, geographic origin, sex, BMI or recruiting centre were not
day, 1% of women reported alcohol consumption of over 20 g/ associated with HBsAg loss.
day, 8.7% were current smokers and 1.2% reported canna-
bis use. In multivariate logistic regression, only baseline HBsAg titre
(p = 0.009) was associated with HBsAg clearance. Approximately
Most patients had detected but < 2000 IU/mL viral load 25% of patients with HBsAg < 1000 and half of patients with
(71.3%), HBsAg titre > 1000 IU/mL (59.8%) and low (< 3 log HBsAg < 100 have lost their HBsAg at 5 years, while only 13.8%
U/mL) or undetected HBcrAg (85.5%). Viral genotyping was of those with HBV DNA < 2000 and 41.2% of those with unde-
successful in 74.8% of subjects with a sufficient viral load. tectable HBV DNA had this outcome.
Genotype E was predominant (27.8%), followed by A (20.0%),
D (18.3%), B (4.4%), C (3.9%) and F (0.4%). Higher HBsAg titres The performance of baseline HBsAg and HBV DNA quanti-
were observed for patients infected by genotype E, compared fication to identify patients who will clear HBsAg during fol-
with all other genotypes except F, which was detected in only low-­up is shown in Figure 2, with areas under the ROC curve
2 patients (Figure 1a). Patients infected by genotypes B and C of 0.927 (CI 95% [0.889–0.965]), p < 0.0001, and 0.771 (CI 95%
presented higher HBcrAg titres, which were quantifiable in [0.689–0.853]), p < 0.0001, respectively. An HBsAg < 153 UI/mL
42 and 30% of cases respectively, compared with 7%–13% for predicted HBsAg loss at 5 years, with the highest sensitivity and
other genotypes (Figure 1b). HBV genotype had no impact on specificity, which is 87%. However, performance at this thresh-
viral load. old varied according to genotype, with sensitivity and specificity
of 100% and 97.6% for the E genotype, compared with 84% and
83% for the non-­E genotype.
2.2   |   Five-­Year Follow-­Up

The attrition rate at 5 years was 39% (Table S1). During fol- 3   |   Discussion
low-­up, two deaths occurred, unrelated to HBV (accidental fall,
breast cancer). No cases of acute liver failure, cirrhosis or hepa- Prospective follow-­up of a large cohort of patients with chronic
tocarcinoma were reported. HBeAg-­negative HBV infection in France confirms the favour-
able prognosis of this population, with rare viral reactivations,
Antiviral treatment was given to 18 patients, with a pre-­emptive occurring in 0.3% of cases per year. The other patients saw their
objective in nine cases: progressive cancer (n = 3), initiation of HBsAg levels decrease over time, with a very slow decline of 0.2
immunosuppressive therapy (n = 5) and occupational precau- log on average over the 5 years of follow-­up; functional cure, that
tion in a healthcare worker (n = 1). In the nine other patients, the is loss of HBsAg, was a rare event, occurring in 1.4% of cases
reason for antiviral treatment was HBV reactivation, defined as per year, similar to previously reported rates [3, 20–22], and was
viral load > 20,000 IU/mL and/or ALT > N. The reactivation rate accurately predicted over 5 years by a low baseline HBsAg titre.
was 0.3% per year and 1.6% at 5 years. Reactivation occurred in This simple tool can be used to tailor the care of patients with
7 women and 2 men, mean age 39 ± 9 years, mean HBsAg titre chronic HBeAg-­negative HBV infection.
at inclusion 4.06 + 4.15 log IU/mL, > 1000 IU/mL in all cases,
and mean viral load 3.6 ± 3.4 log IU/mL, > 2000 IU/mL in 6/9 During the natural history of chronic HBV infection, the for-
patients. Genotype E was identified in 6/9 patients. HBcrAg was merly ‘inactive carrier’ stage is reached after many years of evo-
tested in eight patients and undetected in 4/8 cases. lution and has a favourable long-­term outcome [3]. At this stage,
the decrease in HBsAg levels over time is expected [16, 23], with
Spontaneous HBsAg clearance was observed in 39 patients, infrequent loss of HBsAg, but when it does occur, seroclearance
that is 1.4%/year, 6.9% at 5 years. Among these 39 patients, 21 has been shown to be sustained in most cases [20, 21]. Among
had available anti-­HBs results: 15/21 (71.5%) seroconverted to the factors predictive of HBsAg loss in these patients, a low
anti-­HBs. baseline HBsAg titre and/or the dynamics of the decline in the

3 of 9
TABLE 1    |    Patient's characteristics at baseline, by outcome.

All patients, HBsAg clearance during HBsAg persistence at 5 years

N = 564 follow-­up, N = 39 w/o treatment, N = 300 p
Sex, n (%)
Female 307 (54.4) 19 (48.7) 166 (55.3) 0.4351
Male 257 (45.6) 20 (51.3) 134 (44.7)
Age mean (SD) 42.3 (11.9) 52 (11.6) 41.9 (11.2) < 0.0001
Origin, n (%)
Sub-­Saharian Africa 269 (47.7) 11 (28.2) 144 (48.0) 0.3330
France 87 (15.4) 8 (20.5) 45 (15.0)
North Africa 74 (13.1) 7 (17.9) 45 (15.0)
Asia/Oceania 68 (12.1) 7 (17.9) 34 (11.3)
Western Europe/North 39 (6.9) 4 (10.3) 18 (6.0)
America
Eastern Europe 19 (3.4) 2 (5.1) 10 (3.3)
Other 8 (1.4) 0 4 (1.3)
BMI mean (SD) 25.7 (4.5) 25.3 (3.8) 25.8 (4.5) 0.5424
Hypertension therapy, n (10.1) 9 (23.1%) 27 (9.0%) 0.0056
(%)
Diabetes therapy, n (%) (4.4) 4 (10.3%) 11 (3.7%) 0.0372
Hypolipidemic therapy, (5.0) 6 (15.4%) 9 (3.0%) 0.0004
n (%)
HBV genotype, n (%)
A 115 (20.0) 5 (12.8) 64 (21.3) 0.2187
B 26 (4.4) 3 (7.7) 15 (5.0)
C 22 (3.9) 2 (5.1) 9 (3.0)
D 105 (18.3) 5 (12.8) 57 (19.0)
E 158 (27.8) 8 (20.5) 85 (28.3)
F 2 (0.4) 0 1 (0.3)
ND 145 (25.2) 16 (41.1) 69 (23.0)
HBV DNA mean log IU/ 3.16 (3.44) 2.59 (3.02) 3.15 (3.4) 0.0112
mL (SD)
HBVDNA class
Not detected, n (%) (10.1) 14 (35.9) 20 (0.7) < 0.0001
Detected < 2000, n (%) (71.3) 23 (59) 220 (66.7)
2000–20,000, n (%) (18.6) 2 (5.1) 60 (32.6)
HBsAg mean log IU/mL 3.77 (3.91) 2.12 (2.69) 3.78 (3.90) < 0.0001
(SD)
HBsAg class
< 100, n (%) (16.5) 32 (82) 33 (11) < 0.0001
100–999, n (%) (23.8) 6 (15.4) 78 (26)
≥ 1000, n (%) (59.8) 1 (2.6) 189 (63)

(Continues)

4 of 9 Journal of Viral Hepatitis, 2025

TABLE 1    |    (Continued)

All patients, HBsAg clearance during HBsAg persistence at 5 years

N = 564 follow-­up, N = 39 w/o treatment, N = 300 p
HBcrAg class
Not detected, n (%) 234 (41.5) 21 (53.8) 125 (41.7) 0.0720
Detected, n (%) 237 (42.0) 16 (41.1) 120 (40.0)
Quantifiable, n (%) 68 (12.1) 1 (2.6) 41 (13.7)
NA, n (%) 25 (4.4) 1 (2.6) 14 (4.7)

≥3

FIGURE 1    |    Baseline HBsAg and HBcrAg levels, by genotype. (a) HBsAg titres (mean ± SD log IU/mL). Genotype A: 3.33 ± 0.8; B: 2.36 ± 0.9; C:
3.0 ± 0.2; D: 2.62 ± 1.1; E: 3.57 ± 0.9. HBsAg titres in genotype E were higher than titres in all other genotypes (p < 0.0001). (b) HBcrAg levels. The
frequency of quantifiable HBcrAg levels was higher in genotypes B and C compared to genotypes A, D and E (p < 0.05).

5 of 9
FIGURE 2    |    Performance of baseline HBsAg and HBV DNA quantification to identify patients who will lose HBsAg during 5-­years follow-­up.
Receiver operating characteristic curve (ROC) for prediction of HBsAg loss based on baseline serum levels of HBsAg and HBV DNA. Areas under
the curve were 0.927 for HBsAg levels, and 0.771 for HBV DNA levels.

titre are highly predictive of seroclearance in many studies, re- sub-­Saharan Africa. We show higher HBsAg titres in patients
flecting the decline in cccDNA levels. In Asian cohorts, an an- infected with genotypes F, E and A than in those infected with
nual decrease > 0.5 log IU/mL had a positive predictive value of genotypes B, C or D (Figure 1, Table S2). A Spanish study also
HBsAg clearance of 89% when the initial HBsAg was < 200 IU/ reported that HBsAg levels varied across genotypes in HBeAg-­
mL [23, 24]. In a large and diverse North American and Asian negative patients with higher titres for genotypes F, H, E and
Pacific population, the annual seroclearance rate reached 7% A than for genotype D [29]. To date, EASL guidelines for the
in patients aged over 55 years with HBsAg < 100 UI/mL [25]. identification of HBeAg-­negative chronic infections have been
HBsAg < 100 UI/mL was also the best predictor of spontaneous based on results from Asian or Caucasian cohorts where B and
seroclearance in an Italian cohort [22]. Combinations of factors C, or D genotypes predominate [3, 17, 30, 31]. Previously defined
have been proposed to improve the prediction of HBsAg loss thresholds may not be transposable to France due to the differ-
in patients with chronic HBeAg-­negative HBV infection: a low ence in genotype distribution. Indeed, most of our patients (60%)
HBsAg (< 100 IU/mL) with low anti-­HBc levels [26] or change had baseline HBsAg over 1000 UI/mL, the threshold proposed
in HBsAg levels and fibronectin level [27] for seroclearance at by Brunetto et al. [30] to define inactive infection among geno-
1 year; a low HBsAg with a low level of the IP-­10 cytokine [28] or type D-­infected patients, but the absence of liver events in most
a score combining sex, change in HBsAg, age and for the predic- of our patients suggests that they were true ‘inactive carriers’.
tion of HBsAg loss at 3 years [21].
In univariate analysis, older age and lower baseline viral load
In our study, where patients were enrolled with a strict defini- were associated with HBsAg seroclearance, a result consistent
tion of inactive HBeAg-­negative infection, baseline HBsAg titre with previous studies [18]. The association of HBsAg loss with
was the only independent predictor of functional cure: a value age and pathologies whose frequency increases with age (dys-
of 153 IU/mL was identified, with the best possible sensitivity lipidaemia, hypertension and diabetes) are consistent with a de-
and specificity (87%), patients who were HBsAg-­negative within crease in titre with the duration of infection [16]. However, most
5 years, performance was even better for genotype E-­infected of the migrants included in this study were diagnosed with B
patients. virus infection when they were screened on arrival in France,
so the age of the infection, although probably neonatal in most
One of the original features of our cohort is the great genotypic cases, is not known precisely. We therefore did not examine
diversity of our patients, with a majority of E and A, endemic in the reported age of infection in our analysis. Similarly, we did

6 of 9 Journal of Viral Hepatitis, 2025

not record which lipid-­lowering treatment was used (statin or The results of our cohort, which is characterised by great geno-
fibrate), which made it impossible to study whether there was a typic diversity and, by extension, great ethnic diversity, could
specific effect of statins. therefore be extrapolated to most European countries, and
probably also to countries of emigration: Once inactive carrier
In a large retrospective Taiwanese study, while HBsAg level status has been carefully characterised, a low HBsAg level reli-
was the best predictor of HBsAg clearance in treatment naïve ably identifies patients who will lose the HBsAg and who could
patients at 5 years of follow-­up, it was the HBcrAg level that was therefore safely benefit from reduced monitoring.
the best long-­term predictor, at 10, 15 or 20 years [32]. However,
inactive carriers have typically low or undetected HBcrAg lev-
els reflecting the low transcriptional activity of cccDNA at this 4   |   Conclusion
stage [18], and this was the case for most of our patients (83.5%).
Accordingly, HBcrAg level was not a relevant predictor of spon- This study of a large and diverse French cohort of patients with
taneous HBsAg seroclearance at this late stage of the natural strict diagnostic criteria for HBeAg-­negative chronic HBV in-
history. In addition, we observed a genotype-­dependent bias of fection confirms the rarity of clinical events and the low fre-
HBcrAg quantification that deserves further investigation. quency of functional cure after 5 years of follow-­up. In this
population, with predominantly E and A genotypes, HBsAg
Viral reactivation was infrequent and well controlled by the levels slowly declined over time and a low initial titre < 153 UI/
available nucleos(t)ide analogues (tenofovir and entecavir), mL was the best predictor of functional cure at 5 years. Patient
which were initiated as soon as reactivation was diagnosed, compliance with the monitoring protocol was limited, as con-
probably contributing to the absence of hepatic complications in firmed by the high attrition rate observed. This surveillance
this cohort. The small number of reactivations makes it impos- could be reduced, at low risk, for people whose HBsAg level
sible to define reliable risk thresholds. However, these patients is below this threshold, once it has been confirmed in a vali-
all had baseline HBsAg > 1000 IU/mL, and most were infected dation cohort. On the other hand, surveillance should not be
with genotype E, with viral load > 2000 IU/mL. It would be in- relaxed for those with a viral load > 2000 IU/mL and HBsAg
teresting to explore the factors associated with reactivation in > 1000 IU/mL. Only the availability of more potent antivirals,
a cohort with a larger number of patients who have HBV DNA capable of eradicating the virus, seems likely to change this
levels between 2000 and 20,000 IU/mL. For these patients, it is strategy.
important to emphasise the need for long-­term follow-­up, espe-
cially since treatment does not lead to HBV eradication.
Author Contributions
One might question the representativeness of our cohort, par-
ticularly given the high attrition rate. This high rate can be X.C., P.P. and A.-­M.R.-­A . analysed, interpreted the data, and wrote the
manuscript. A.V. contributed to data analysis A.-­M.R.-­A . and L.M. per-
explained by the absence of therapeutic indications. This is es-
formed HBcrAg and HBsAg quantification, and HBV genotyping. The
pecially true since many participants, being migrants, had other PIBAC study group (X.C., H.L., G.M., J.-­F.C., T.F.) recruited patients,
immediate concerns (housing with a change of address during performed clinical monitoring, and provided clinical expertise. All ap-
the study, access to employment, administrative situation, etc.). proved the manuscript.
In addition, the last monitoring period (2020–2021) was severely
disrupted by the COVID-­19 epidemic, which prevented or de-
Acknowledgements
layed many consultations. Nevertheless, despite high attrition
rate, our cohort is one of the largest prospective cohorts reported Roche Diagnostics kindly provided part of the reagents for HBsAg
quantification. We thank Barbara de Dieuleveult for database man-
to date [4–14]. This cohort is also representative of the popula-
agement, all the patients who participated in this study, and all the
tion treated in mainland France in several respects: (i) the ma- doctors in the PIBAC study group: H. Salloum, M. Picon-­ Coste, S
jority of patients were of African origin [1]; (ii) baseline BMI Hommel, J. Henrion, J.P. Arpurt, I. Rosa, I. Ollivier, S Bresson-­Hadni,
was comparable to that of the general population in the OBEPI-­ D. Zanditenas, M. Schnee, F. Heluwaert, S Cosconea, B. Lambare, A.
ROCHE 2020 study (Figure S1) [33]; (iii) Liver elastography data Garioud, P. Delasalle, C. Renou, B. Hanslik, N. Boyer, and C. Castelnau.
are consistent with previous reports [7, 12, 13, 34]. Alcohol, to- H. Salloum, M. Picon-­Coste, S Hommel, J. Henrion, J. P. Arpurt, I. Rosa,
bacco, and cannabis consumption were however lower than in I. Ollivier, S Bresson-­Hadni, D. Zanditenas, M. Schnee, F. Heluwaert, S
Cosconea, B. Lambare, A. Garioud, P. Delasalle, C. Renou, B. Hanslik,
the general population. Although self-­reporting may lead to an
N. Boyer, C. Castelnau.
underestimate, this reduced consumption could be linked to the
different lifestyles of the patients included, many of whom are
Ethics Statement
not of French origin. In our study, the main recruiting centres
were associated with large maternity wards. As screening for The ethics of the study was approved by the Comité de Protection des
hepatitis B is systematic during pregnancy, this probably con- Personnes (CPP) of Tours with number n° 2014-­s10.
tributed to the predominance of women in our cohort.
Conflicts of Interest
In Europe, the prevalence and the diversity of HBV are strongly The authors declare no conflicts of interest.
affected by immigration. France, and Europe in general, have
seen high rates of immigration from countries with intermediate
Data Availability Statement
or high HBV seroprevalence, including Asian and African coun-
tries [35]. HBV genotypes, which show distinct geographical dis- The data that support the findings of this study are available from the
tributions, or patient's genetics may impact disease progression. corresponding author upon reasonable request.

7 of 9
References 18. M. Riveiro-­Barciela, J. M. Pericas, and M. Buti, “How to Interpret
Viral Markers in the Management of Chronic Hepatitis B Infection,”
1. C. Tamandjou, S. Laporal, F. Lot, and C. Brouard, “Données épidé-
Clinical Microbiology and Infection 28, no. 3 (2022): 355–361.
miologiques récentes sur les hépatites C, B et Delta,” Bulletin Epidémi-
ologique Hebdomadaire 15–16 (2023): 311–317. 19. Y. Li, Y. S. Huang, Z. Z. Wang, et al., “Systematic Review With Meta-­
Analysis: The Diagnostic Accuracy of Transient Elastography for the
2. C. Hadey and P. Mélin, “«Vivre avec une hépatite B»: une enquête des
Staging of Liver Fibrosis in Patients With Chronic Hepatitis B,” Alimen-
Etats généraux de l'hépatite B en France métropolitaine et d'outre-­mer,”
tary Pharmacology & Therapeutics 43, no. 4 (2016): 458–469.
Bulletin Epidémiologique Hebdomadaire 3–4 (2022): 57–63.
20. D. Bruden, B. J. McMahon, M. Snowball, et al., “Rate and Durability
3. European Association for the Study of the Liver, “EASL 2017 Clinical
of the Clearance of HBsAg in Alaska Native Persons With Long-­Term
Practice Guidelines on the Management of Hepatitis B Virus Infection,”
HBV Infection: 1982–2019,” Hepatology 79, no. 6 (2024): 1412–1420.
Journal of Hepatology 67, no. 2 (2017): 370–398.
21. N. A. Terrault, A. S. Wahed, J. J. Feld, et al., “Incidence and Pre-
4. R. de Franchis, G. Meucci, M. Vecchi, et al., “The Natural History of
diction of HBsAg Seroclearance in a Prospective Multi-­Ethnic HBeAg-­
Asymptomatic Hepatitis B Surface Antigen Carriers,” Annals of Inter-
Negative Chronic Hepatitis B Cohort,” Hepatology 75, no. 3 (2022):
nal Medicine 118, no. 3 (1993): 191–194.
709–723.
5. J. P. Villeneuve, M. Desrochers, C. Infante-­R ivard, et al., “A Long-­
22. M. Barone, A. Iannone, M. Mezzapesa, et al., “Natural History and
Term Follow-­Up Study of Asymptomatic Hepatitis B Surface Antigen-­
Hepatitis B Virus Surface Antigen (HBsAg) Spontaneous Seroclearance
Positive Carriers in Montreal,” Gastroenterology 106, no. 4 (1994):
in Hepatitis B Virus e-­A ntigen (HBeAg)-­Negative Patients With Inac-
1000–1005.
tive Chronic Infection: A Multicenter Regional Study From South Italy,”
6. M. Martinot-­Peignoux, N. Boyer, M. Colombat, et al., “Serum Hepa- Pathogens 12, no. 10 (2023): 1198.
titis B Virus DNA Levels and Liver Histology in Inactive HBsAg Carri-
ers,” Journal of Hepatology 36, no. 4 (2002): 543–546. 23. W. K. Seto, D. K. Wong, J. Fung, et al., “A Large Case-­Control Study
on the Predictability of Hepatitis B Surface Antigen Levels Three Years
7. Y. S. Hsu, R. N. Chien, C. T. Yeh, et al., “Long-­Term Outcome After Before Hepatitis B Surface Antigen Seroclearance,” Hepatology 56, no.
Spontaneous HBeAg Seroconversion in Patients With Chronic Hepatitis 3 (2012): 812–819.
B,” Hepatology 35, no. 6 (2002): 1522–1527.
24. H. C. Lin, J. Liu, M. H. Pan, et al., “Rapid Decline Rather Than Ab-
8. M. Manno, C. Camma, F. Schepis, et al., “Natural History of Chronic solute Level of HBsAg Predicts Its Seroclearance in Untreated Chronic
HBV Carriers in Northern Italy: Morbidity and Mortality After Hepatitis B Patients From Taiwanese Communities,” Clinical and
30 Years,” Gastroenterology 127, no. 3 (2004): 756–763. Translational Gastroenterology 14, no. 8 (2023): e00586.
9. G. Fattovich, N. Olivari, M. Pasino, M. D'Onofrio, E. Martone, and 25. Y. H. Yeo, T. C. Tseng, T. Hosaka, et al., “Incidence, Factors, and
F. Donato, “Long-­Term Outcome of Chronic Hepatitis B in Caucasian Patient-­L evel Data for Spontaneous HBsAg Seroclearance: A Cohort
Patients: Mortality After 25 Years,” Gut 57, no. 1 (2008): 84–90. Study of 11,264 Patients,” Clinical and Translational Gastroenterology
10. F. Habersetzer, R. Moenne-­L occoz, N. Meyer, et al., “Loss of Hepa- 11, no. 9 (2020): e00196.
titis B Surface Antigen in a Real-­Life Clinical Cohort of Patients With 26. K. Kan, D. K. Wong, R. W. Hui, W. K. Seto, M. F. Yuen, and L. Y.
Chronic Hepatitis B Virus Infection,” Liver International 35, no. 1 Mak, “Anti-­H Bc: A Significant Host Predictor of Spontaneous HBsAg
(2015): 130–139. Seroclearance in Chronic Hepatitis B Patients—A Retrospective Longi-
11. W. P. Brouwer, H. L. Chan, M. R. Brunetto, et al., “Repeated Mea- tudinal Study,” BMC Gastroenterology 23, no. 1 (2023): 348.
surements of Hepatitis B Surface Antigen Identify Carriers of Inactive 27. F. Liu, W. K. Seto, D. K. Wong, et al., “Plasma Fibronectin Levels
HBV During Long-­ Term Follow-­ Up,” Clinical Gastroenterology and Identified via Quantitative Proteomics Profiling Predicts Hepatitis B
Hepatology 14, no. 10 (2016): 1481–1489 e1485. Surface Antigen Seroclearance in Chronic Hepatitis B,” Journal of Infec-
12. L. Castera, P. H. Bernard, B. Le Bail, et al., “Transient Elastography tious Diseases 220, no. 6 (2019): 940–950.
and Biomarkers for Liver Fibrosis Assessment and Follow-­Up of Inac- 28. K. Kan, D. K. Wong, R. W. Hui, W. K. Seto, M. F. Yuen, and L. Y.
tive Hepatitis B Carriers,” Alimentary Pharmacology & Therapeutics 33, Mak, “Plasma Interferon-­Gamma-­Inducible-­Protein 10 Level as a Pre-
no. 4 (2011): 455–465. dictive Factor of Spontaneous Hepatitis B Surface Antigen Seroclear-
13. G. L. Wong, H. L. Chan, Z. Yu, H. Y. Chan, C. H. Tse, and V. W. ance in Chronic Hepatitis B Patients,” Journal of Gastroenterology and
Wong, “Liver Fibrosis Progression Is Uncommon in Patients With In- Hepatology 39, no. 1 (2024): 202–209.
active Chronic Hepatitis B: A Prospective Cohort Study With Paired 29. M. Riveiro-­
Barciela, M. Bes, F. Rodriguez-­ Frias, et al., “Serum
Transient Elastography Examination,” Journal of Gastroenterology and Hepatitis B Core-­Related Antigen Is More Accurate Than Hepatitis B
Hepatology 28, no. 12 (2013): 1842–1848. Surface Antigen to Identify Inactive Carriers, Regardless of Hepatitis B
14. Y. Ngo, Y. Benhamou, V. Thibault, et al., “An Accurate Definition of Virus Genotype,” Clinical Microbiology and Infection 23, no. 11 (2017):
the Status of Inactive Hepatitis B Virus Carrier by a Combination of Bio- 860–867.
markers (FibroTest-­ActiTest) and Viral Load,” PLoS One 3, no. 7 (2008): 30. M. R. Brunetto, F. Oliveri, P. Colombatto, et al., “Hepatitis B Surface
e2573. Antigen Serum Levels Help to Distinguish Active From Inactive Hep-
15. K. Okada, Y. Nakayama, J. Xu, Y. Cheng, and J. Tanaka, “A Nation-­ atitis B Virus Genotype D Carriers,” Gastroenterology 139, no. 2 (2010):
Wide Medical Record Database Study: Value of Hepatitis B Surface 483–490.
Antigen Loss in Chronic Hepatitis B Patients in Japan,” Hepatology Re- 31. M. R. Brunetto, I. Carey, B. Maasoumy, et al., “Incremental Value of
search 54 (2024): 1004–1015. HBcrAg to Classify 1582 HBeAg-­Negative Individuals in Chronic In-
16. C. M. Chu and Y. F. Liaw, “Hepatitis B Surface Antigen Seroclear- fection Without Liver Disease or Hepatitis,” Alimentary Pharmacology
ance During Chronic HBV Infection,” Antiviral Therapy 15, no. 2 & Therapeutics 53, no. 6 (2021): 733–744.
(2010): 133–143. 32. T. C. Tseng, C. Chiang, C. J. Liu, et al., “Low Hepatitis B Core-­
17. H. L. Chan, A. Thompson, M. Martinot-­Peignoux, et al., “Hep- Related Antigen Levels Correlate Higher Spontaneous Seroclearance of
atitis B Surface Antigen Quantification: Why and How to Use It in Hepatitis B Surface Antigen in Chronic Hepatitis B Patients With High
2011—A Core Group Report,” Journal of Hepatology 55, no. 5 (2011): Hepatitis B Surface Antigen Levels,” Gastroenterology 164, no. 4 (2023):
1121–1131. 669–679 e666.

8 of 9 Journal of Viral Hepatitis, 2025

33. Ligue Contre l'Obésité, “Enquête épidémiologique nationale sur le
surpoids et l'obésité,” accessed June, 2024, https://​ligue​contr​elobe​site.​
org/​enque​t e-­​epide​m iolo​g ique​-­​n atio​n ale-­​sur-­​le-­​surpo​ids-­​et-­​lobes​ite/​
(2021).
34. M. Cornberg, V. W. Wong, S. Locarnini, M. Brunetto, H. L. A. Jans-
sen, and H. L. Chan, “The Role of Quantitative Hepatitis B Surface An-
tigen Revisited,” Journal of Hepatology 66, no. 2 (2017): 398–411.
35. S. Sharma, M. Carballo, J. J. Feld, and H. L. Janssen, “Immigration
and Viral Hepatitis,” Journal of Hepatology 63, no. 2 (2015): 515–522.

Supporting Information
Additional supporting information can be found online in the
Supporting Information section.

9 of 9