Correspondence: Latifa T. F. Yeung, MD, MSc, FRCPC, Rouge Valley Health System, Centenary Health Centre, Galaxy 12 Child & Teen Clinic, 2867 Ellesmere Road, 12thfl, Scarborough, ON M1E 4B9, Canada. E-mail: firstname.lastname@example.org
Whereas e-seroconversion represents the loss of hepatitis B e-antigen (HBeAg) followed by gain of antibody to HBeAg (anti-HBe), ‘inactive chronic infection’ extends this concept to include e-seroconversion with decreased serum viral load and biochemical remission. These events must be well-characterized before treatment outcomes can be evaluated. We examined the rates of e-seroconversion and achievement of inactive chronic infection among children with chronic HBV infection. Children who were HBsAg positive >6 months were identified retrospectively between 1983 and 2008 from the Hospital for Sick Children Liver Clinic. Inactive chronic infection was defined as loss of HBeAg, serum ALT ≤40 IU/mL, and HBV DNA <106 IU/mL. Both e-seroconversion and achievement of inactive chronic infection were characterized using survival analysis. The effect of transmission route, treatment, age at diagnosis, ethnicity, gender and baseline ALT on these rates was evaluated with univariate and multiple regression. Of 252 HBeAg-positive cases, 59.9% had HBV-infected mothers, 77% were Asian, and 33 received interferon-α. Untreated children were younger at last follow-up (mean 14.5 vs 17.6 years), had lower ALT (median 60 vs 116 IU/mL) and had shorter follow-up (6.6 vs 9.1 years, all P < 0.002) compared to treated children. Crude e-seroconversion rate was 41.7% over 0.5–19.1 years of follow-up, and this was not affected by transmission route (P = 0.93), gender (P = 0.62) nor treatment (P = 0.08). 49% achieved inactive chronic infection by age 19 years. Being non-Asian, age at diagnosis<3 years, and ALT ≥40 IU/mL were associated with a higher rate of e-seroconversion and achieving inactive chronic infection (P < 0.0001). Almost 50% of children achieved inactive chronic infection by early adulthood.
Despite an effective vaccine, chronic infection with hepatitis B virus (HBV) affects 350 million persons worldwide. Children infected with HBV are at risk of joining the 200 000–300 000 HBV-infected individuals who succumb to cirrhosis and hepatocellular carcinoma (HCC) each year [1, 2]. Whether it is best to offer treatment in childhood remains unclear .
E-seroconversion is defined as the loss of HBV e-antigen (HBeAg) followed by gain of anti-HBe antibody. Traditionally, e-seroconversion suggests a ‘chronic HBV carrier’ phase, now defined by negative HBeAg, normal alanine aminotransferase (ALT) and HBV DNA <103 IU/mL . For this study, we propose to classify children as having ‘inactive chronic infection’ (ICI), defined by loss of HBeAg, ALT ≤40 IU/mL and HBV DNA <106 IU/mL. This collective outcome recognizes the emerging importance of depressed HBV DNA levels in predicting favourable long-term outcomes [5, 6]. It also accommodates the changing minimum HBV DNA detection limits over 25 years and recognizes that DNA levels fluctuate in children . The rate of spontaneous e-seroconversion and achieving ICI among children must be well-characterized before treatment outcomes can be meaningfully evaluated.
We aimed to evaluate the natural history of chronic HBV infection in children using time-dependent analyses. We described the rates of (1) e-seroconversion and (2) ICI among children with chronic HBV infection. We examined the effect of route of transmission, age at diagnosis, ethnicity, gender and baseline ALT on these rates. As well, we evaluated the real-world effectiveness (rather than efficacy) of interferon-α in improving these rates.
Patients and Methods
Children aged <18 years who were positive for hepatitis B surface antigen (HBsAg) for at least 6 months were identified retrospectively between 1983 and 2008 from the Hospital for Sick Children Liver Clinic. Those followed <6 months or co-infected with HIV or hepatitis C virus were excluded.
Patients with chronic hepatitis B (CHB) were typically evaluated every 6–12 months. Liver biochemistry and α-fetoprotein levels were monitored using conventional assays. Viral markers (HBsAg, HBeAg, anti-HBe, anti-HBs) were tested by enzyme-linked immunoabsorbant assay. Serum samples were obtained and stored at minus 20 °C. Serum HBV DNA was quantitatively analysed using dot-blot hybridization between 1989 and 1993 (lower detection limit = 2.27 × 106 IU/mL), Digene Hybrid Capture Assay (I) between 1993 and 1999 (lower limit = 9.09 × 104 IU/mL), Digene II assay between 2000 and 2004 (lower limit = 940 IU/mL) and Roche PCR assay from 2005 onward (lower limit = 60 IU/mL).
E-seroconversion required the loss of HBeAg and gain of anti-HBe. Inactive chronic infection was defined as loss of HBeAg, ALT <40 IU/mL and HBV DNA <106 IU/mL. The effect of route of transmission, treatment, age at diagnosis, sex, ethnicity and baseline ALT on these outcomes was evaluated. Vertical transmission was defined as having a mother known to have chronic HBV infection. Horizontal transmission required a non-maternal household contact known to have HBV infection or history of blood transfusion. Ethnicity was categorized based on the 2006 Canadian Census Questionnaire . Patients were offered liver biopsy in response to a period of significant elevation of liver enzymes, prior to treatment, and to assess for alternate disease processes. If patients failed to e-seroconvert with ALT consistently > twice the upper limit of normal for >6 months and histology confirmed chronic hepatitis, other causes of liver disease were excluded and interferon-α was offered at 6 megaunits per m2 thrice weekly for 24 weeks. Response was evaluated 24 weeks following completion of therapy.
Additional outcomes of s-seroconversion (loss of HBsAg and gain of anti-HBs), abnormal growth, hepatic symptoms and hepatic decompensation were noted at each visit. Liver biopsy records were reviewed for cirrhosis or HCC.
The rate of achieving ICI status over time served as the primary outcome. The chi-squared test analysed categorical variables, Student's t-test compared continuous variables and Wilcoxon rank sum test evaluated skewed or ordinal data. Kaplan–Meier analysis was used to compute nonparametric estimates of the time to (1) e-seroconversion and (2) ICI achievement. Patients who had not e-seroconverted at last follow-up were censored. Rates of e-seroconversion and ICI achievement were evaluated with predictor variables for route of transmission (vertical, horizontal or unknown), history of interferon-α treatment, age at diagnosis, sex, ethnicity (Asian or non-Asian) and baseline ALT using univariate survival analysis by log-rank test. Variables found to be significant from univariate analysis were used in multivariate regression using the parametric Weibull variable hazard model. For sensitivity analysis, the ICI threshold for HBV DNA was lowered to <103 IU/mL, and descriptive and univariate tests were repeated only on patients who had access to serum HBV DNA testing with a detection limit <103 IU/mL. All statistical analyses utilized SAS version 9.2 (SAS Institute Inc., Cary, NC, USA). A P-value <0.05 was considered statistically significant.
This study received Hospital for Sick Children Research Ethics Board approval (REB #1000012452).
Of the 317 cases of chronic HBV infection identified, eight received lamivudine, one received entecavir and 56 had e-seroconverted by first assessment. Half of these 56 children were Asian¸ 40 were boys, with mean age at diagnosis of 8.8 years and median baseline ALT of 39 IU/L.
The remaining 252 children formed the basis of our analysis. Thirty-three patients received interferon-α. Treated patients were older at last follow-up, were followed longer and had higher baseline ALT compared to untreated patients (Table 1). Asians comprised 77% of the multi-ethnic cohort (Table 2). Two cases of vaccine failure were identified. Twenty children (six boys) were adopted.
Table 1. Epidemiological, clinical and biochemical features of HBeAg-positive patients
All values are mean (SD) unless otherwise specified.
Southeast or East Asian (Filipino, Korean, Burmese, Cambodian, Indonesian, Laotian, Malaysian, Vietnamese, Thai)
South Asian (East Indian, Pakistani, Sri Lankan)
European (Romanian, Polish, Ukraine, Portuguese, Italian, Bulgarian)
West Asian (Afghan, Iranian)
South American (Guyana, Uruguay)
African (Ghana, Angolian, Somalian, Black)
At initial presentation, one patient had fatigue, one had scleral icterus and one had jaundice, nausea and diarrhoea. During follow-up, 15 patients had signs or symptoms: one had jaundice, scleral icterus, anorexia and fatigue; one with beta-thalassemia had jaundice with dark urine; four demonstrated spider nevi. Nine patients had characteristics of non-alcoholic fatty liver disease (fatty liver on ultrasound, acanthosis nigricans and/or body mass index > 97th percentile). No child had HBV-attributable growth retardation or hepatic failure.
Among 53 children who underwent percutaneous liver biopsy, two demonstrated cirrhosis. One South Asian boy with unknown route of infection had a peak ALT of 210 IU/mL. Liver biopsy revealed mild cirrhosis. Six months later, he spontaneously e-seroconverted and ALT fell to 30 IU/mL. The second case involved a Chinese girl with vertically acquired HBV infection, peak ALT of 300 IU/mL and early cirrhosis on liver biopsy. At age 17, 3 months after biopsy, her ALT dropped to 226 IU/mL. She received interferon alfa-2b for 6 months. Nine months after starting treatment, she e-seroconverted and ALT normalized to 23 IU/mL. Neither case showed biomarkers or ultrasonographic evidence of portal hypertension.
Hepatocellular carcinoma was diagnosed in one Chinese boy with vertically acquired CHB. He presented at age 10 with α-fetoprotein of 38 400 mg/L. Abdominal ultrasound showed a 5 × 4.5 cm caudate lobe mass. Liver biopsy revealed primary HCC and low-grade chronic hepatitis B with multifocal nodular hyperplasia without evidence of fibrosis or cirrhosis. Six years earlier, a liver biopsy showed near normal histology without evidence of malignancy. He spontaneously s-seroconverted within 6 months of HCC diagnosis, and underwent liver transplantation 7 months after presentation. He never received anti-HBV therapy.
Thirty-three children (13%) at age 4.3–18.3 years (mean 13.5 years) were treated with interferon-α, seven of whom followed a clinical research protocol and were randomized to receive interferon alfa-2b . One patient with cirrhosis and thrombocytopenia received half the standard dose over fifteen months. He e-seroconverted at last follow-up (ALT 41 IU/mL).
Crude (time-independent) analyses
Ignoring the variable follow-up among patients, the crude rate of e-seroconversion was 42% (n = 105). Of the 219 untreated patients, 80 (37%) seroconverted to anti-HBe-positive status, compared with 25 (76%) of treated children.
S-seroconversion occurred in three (1.2%) children, all without treatment: a Caucasian girl who s-seroconverted by 5 years of age with HBV DNA <2.85 × 105 IU/mL and normalized ALT, a Vietnamese girl who s-seroconverted by 16 years of age with HBV DNA <9.09 × 104 IU/mL, and a Chinese girl who e-seroconverted with normalized ALT by 2.5 years of age and then s-seroconverted by 4 years-old with HBV DNA <60 IU/mL.
Achievement of e-seroconversion
Half of the 252 children achieved e-seroconversion before adulthood (Fig. 1). For the three patients who alternated between e-seroconversion and sero-reversion, their most recent serology was used for analysis. Following a brief period of being anti-HBe negative, all three cases ultimately e-seroconverted. On univariate analysis, those who were younger at diagnosis (<3 years), had higher ALT (≥40 IU/mL) and were non-Asian tended to e-seroconvert earlier (all P < 0.0001). Of note, after adjusting for variable follow-up, the median time to e-seroconversion for untreated (18.1 years) and treated (16.7 years) patients was neither statistically nor clinically significant. Similarly, route of transmission and gender were not significant. Multivariate regression confirmed the findings from univariate analysis (Tables S1 & S2).
Achievement of inactive chronic infection
Almost half (49%) of the cohort achieved ICI status before transitioning to adult care (Fig. 2). Similar to e-seroconversion, children were significantly more likely to achieve ICI if they were younger at diagnosis, had higher ALT or were non-Asian (all P < 0.003, Figs 3, 4, Table S1). Multivariate regression again confirmed the findings from univariate analysis (Table S2).
One hundred and forty-two patients (11 treated) had access to HBV DNA testing with detection limit <103 IU/mL. Analysing this subgroup (derived from the original 252 patients), previously significant differences in age at last follow-up, follow-up duration and proportion of Asians between the treated and untreated patients disappeared, leaving only a higher ALT for treated patients (P = 0.0003). For the outcome of e-seroconversion only, log-rank tests confirmed the favourable significance of younger age at diagnosis, higher baseline ALT, being non-Asian (all P ≤ 0.002) as well as interferon-α treatment (P = 0.02). However, for the outcome of e-seroconversion with normal ALT and a lower threshold of HBV DNA <103 IU/mL (Fig. S1), only younger age remained significant (P < 0.0001). Interferon-α did not significantly affect ICI attainment with a lower HBV DNA requirement (P = 0.82).
Our 25-year experience of childhood chronic HBV infection reflects a large multi-ethnic population. Crudely, 41% e-seroconverted; over time, 50% of e-seroconverted by early adulthood. These findings agree with the Montreal rate of 40.2%, although they identified fewer Asians (43% of 174 children were Asian, 25% had unknown ethnicity) . These estimates fall between measures from more homogenous groups: the crude e-seroconversion rate among Hong Kong children was 7% , whereas Europeans have reported rates as high as 98% .
The slope of both outcome curves was steepest during late adolescence, suggesting a period when e-seroconversion is especially likely. Other centres concur: 25% of 215 Asian Americans e-seroconverted by age 17 years and 50% by age 24 years . Fifty-eight Taiwanese children had mean age of e-seroconversion of 17.2 years . Earlier onset of puberty was associated with earlier e-seroconversion among Taiwanese males . These data highlight a particularly active period of host-virus interaction that warrants further study. Educating older teens on their increased likelihood of e-seroconversion might facilitate compliance with serological monitoring. Conversely, young adults who have not e-seroconverted may merit priority consideration for treatment.
Undergoing e-seroconversion and achieving ICI were more likely for children who were younger when diagnosed, were non-Asian and had higher baseline ALT. Similarly, Marx et al.  found that 13 years after diagnosis, 75% of Asian children, compared with 94% of non-Asians had seroconverted (P < 0.05). Polish children with CHB also showed that younger age and higher ALT favoured e-seroconversion . In our study, age at diagnosis served as a latest-possible date of infection and guides clinical care: once diagnosed, the younger child should be given a chance to spontaneously achieve ICI, whereas an older child may receive earlier consideration for treatment. Particularly when the complex, continued interaction between host immunity and viral evolution is initiated in early childhood, the passage of time favours the host, and more so than any other factor.
Route of transmission did not prove significant. While Zacharakis et al. found vertically transmitted children had lower rates of e-seroconversion, they compared vertical against horizontal or unknown transmission. By contrast, cases of unknown transmission were excluded from our analysis. Although Zacharakis et al.  found that ethnicity was insignificant, they studied only three (non-Asian) ethnic groups. Perhaps route of transmission and ethnicity are surrogates for viral genotype, which may determine different natural histories .
An efficacious treatment may yield low effectiveness when provided outside the careful confines of a clinical trial. Our findings emphasize the need for individualized consideration before initiating interferon-α. While the age at diagnosis was similar between treated and untreated groups, treated patients tended to have higher ALT and be non-Asian, factors shown to favour e-seroconversion. One would thus expect the treated to show a higher rate of e-seroconversion; yet this did not occur. Of note, treated children were considered at risk for progressive disease, thus are not directly comparable to untreated children. While no patient refused treatment, several spontaneously e-seroconverted while preparing to embark on treatment.
The similar outcomes between treated and untreated children may have several explanations. First, it may reflect an interferon-driven acceleration of e-seroconversion which loses impact as untreated patients naturally convert over time . Sensitivity analysis found that interferon-α may improve the e-seroconversion rate, but not the desired outcome of HBV DNA <103 IU/mL with normalized ALT. Second, interferon-α may help more severe cases of CHB achieve ICI at the same rate as the untreated milder cases. Third, interferon-α may benefit a yet-to-be-defined subset of patients. Finally, these rates may be irrelevant: while interferon-α may accelerate e-seroconversion in some [10, 19-21], it remains uncertain whether earlier e-seroconversion modifies the important long-term outcomes of cirrhosis and HCC. Besides interferon-α, newer anti-viral agents being tested in children may alter CHB natural history in the future.
A retrospective design was necessary to summarize our 25-year experience with childhood CHB. Consequently, histological data were not standardized to facilitate comparison. The estimated time of e-seroconversion depended on the frequency of blood work, generally performed yearly. Thus, the measured time of e-seroconversion may reflect a delay of almost a year from the true time. Treatment patients were more adherent. Overall these biases would over-estimate the time to event: the true time of e-seroconversion and ICI achievement may be earlier than measured, particularly among untreated patients.
Inactive chronic infection discretely characterizes a series of dynamic events. The individual events of losing HBeAg, gaining anti-HBe, normalizing ALT, and decreasing serum HBV DNA may each fluctuate over time. Consequently, the distinction between ICI and HBeAg-negative chronic hepatitis may be unclear. In a few, a substantial lag time exists between loss of HBeAg and gain of anti-HBe. Others oscillate between detectable and undetectable states of HBeAg or anti-HBe. Elevated ALT distinguishes ICI from HBeAg-negative reactivation hepatitis, but ALT may rise from non-alcoholic fatty liver disease, intercurrent illness or medication. HBV DNA levels fluctuate. More sensitive PCR techniques have revealed that low-level HBV replication may persist in children over 10 years after e-seroconversion and biochemical remission . Such low-level viremia may permit later reactivation hepatitis, subclinical disease and lead to HCC [16, 23]. Forty percentage of a Taiwanese cohort had fluctuating high viremia (>104 copies/mL) after spontaneous e-seroconversion, although follow-up revealed a benign course . These observations highlight the need for lifelong monitoring of all HBeAg-negative children, including those with ICI. Continuous clinical assessment, measurement of ALT, and most importantly, following the HBV DNA titre are necessary to ascertain that the patient merits a favourable prognosis. Long-term studies with follow-up into adulthood should examine whether consistently higher HBV DNA titres signify an increased risk of reactivation hepatitis in HBeAg-negative children.
Further insight into the host-viral interactions during e-seroconversion is warranted in future studies, including host cytokine gene polymorphisms and phenotype. While our definition of ICI pragmatically incorporated normal ALT and low HBV DNA, the insensitivity of older probes prevented adoption of HBV DNA <106 IU/mL. Although lowest-possible HBV DNA levels are desirable [22, 24], debate continues regarding the most suitable cut-off value. NIH defines ‘inactive carrier state’ as HBV DNA <2000 IU/mL, others propose HBV DNA <2000 IU/mL with HBsAg <100 IU/mL , while recent Taiwanese data with prolonged follow-up suggest a higher cut-off, HBV DNA <20 000 IU/mL, to identify adults with favourable prognosis . Our sensitivity analysis utilizing a lower threshold of HBV DNA <103 IU/mL supports our overall conclusions based on the higher HBV DNA cut-off. Newer techniques such as viral genotyping and sequence analysis were unavailable in our study. While several HBV mutants (G1896A, A1762T and G1764A) may play a significant role in HBeAg-negative adults, these mutations rarely occur in children .
Nearly half of children in our multi-ethnic cohort achieved ICI by early adulthood. These events were more likely for children who were diagnosed when younger, non-Asian, and had higher baseline ALT. These results were not influenced by interferon-α treatment. The maximal rate of e-seroconversion during late adolescence suggests a heightened window of host-immune and viral interaction that deserves more scrutiny. Those who fail to e-seroconvert by early adulthood may represent a higher risk group in whom treatment options should be considered. As cirrhosis and HCC occur later, it will be important to examine the predictive value of childhood e-seroconversion and low HBV DNA on these outcomes following transition to adult care.
We are grateful to Phyllis McGee MN and Constance O'Connor MN, NP-Paediatrics for their contribution to assembling patient data.
Statements of Interests
Dr. Ling has consulted for Gilead and receives research funding from Bristol-Myers-Squibb. The authors have no other conflicts of interest.