Chronic Hepatitis Is Common and Often Untreated Among Children With Hepatitis B Infection in the United States and Canada

Abstract

Objective:

To determine the outcomes of chronic hepatitis B virus (HBV) infection in a large, prospectively studied cohort of children in the USA and Canada.

Study design:

A prospective observational study of children with chronic HBV enrolled in 7 clinical centers and evaluated at baseline, weeks 24 and 48, and annually thereafter, with analysis of demographic, clinical, physical examination and blood test data.

Results:

Among 362 children followed for a median of 4.2 years, elevated alanine aminotransferase (ALT) levels (>1 ULN) were present in 72% at last evaluation, including in 60% of children with loss of HBeAg during follow-up and 70% of those who were HBeAg negative at baseline. Significant ALT flares (males ≥400 U/L, females ≥350 U/L) occurred in 13 children. Of 129 children who fulfilled AASLD treatment criteria during follow-up, anti-HBV treatment was initiated in only 25. One child died (unrelated to liver disease), 1 developed cirrhosis, but no episodes of cirrhotic decompensation or hepatocellular carcinoma were observed. Decline in platelet count was inversely associated with ALT elevations.

Conclusion:

In a cohort of children with chronic HBV infection in the US and Canada, many children remained at risk of progressive liver disease due to active hepatitis, but major clinical outcomes such as cirrhosis, cancer and death were rare. Many children who met criteria for treatment remained untreated.

Chronic infection with hepatitis B virus (HBV) is a major global health problem affecting approximately 257 million people, in view of which the World Health Organization (WHO) has endorsed a global strategy to eliminate hepatitis B infection and other viral hepatitis by 2030 (1,2). Although anti-HBV medications are available to reduce the incidence of adverse outcomes, including cirrhosis and hepatocellular carcinoma (3), 887,000 deaths due to HBV infection occurred in 2015 (4). Transmission of hepatitis B infection is largely preventable by immunization, but programs to provide vaccine to newborns often fail to achieve high levels of coverage for their at-risk populations, and some countries have yet to implement such programs (5). Globally, large numbers of children continue to acquire chronic infection with HBV each year.

Among children in the United States (US) and in Canada, chronic HBV infection is found mostly in those who were born (or whose parents were born) in countries of high endemicity (6). There is a significant paucity of published data on the risk of adverse clinical outcomes occurring among these North American children with HBV infection. Studies of the natural history of chronic hepatitis B infection in children are mostly from Asia or Europe, have tended to focus on changes over time in virological and serological markers that are thought to predict adverse clinical outcomes, and were mostly performed in the era before widespread availability of nucleot(s)ide analogue therapies (7–18).

The Hepatitis B Research Network (HBRN) therefore aimed to prospectively follow a large cohort of children with chronic hepatitis B infection in the USA and Canada to improve our understanding of disease progression during childhood in the modern era and among children linked to clinical care.

Methods

Children ≥6 months and <18 years of age with hepatitis B surface antigen (HBsAg) positivity were enrolled between 2010 and 2017 at 6 pediatric clinical sites in the United States and 1 in Canada. Subjects were excluded if they had a history of hepatic decompensation, hepatocellular carcinoma (HCC), or liver transplantation; human immunodeficiency virus, hepatitis C or hepatitis D co-infection; current antiviral therapy for HBV or chronic immunosuppressive therapy; inability or unwillingness to return for study visits; or evidence of severe illness or other medical or social condition that would make the subject, in the opinion of the investigator, unsuitable for the study. Informed consent was obtained from all participants or their parent/caregiver, and assent was obtained when appropriate from children without capacity to consent, according to the requirements of each site’s Institutional Review Board or Research Ethics Committee.

Study design

The HBRN and the enrollment criteria for its pediatric cohort study have been previously described in detail (6). Briefly, the HBRN was established and is funded by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) to conduct clinical, scientific, epidemiological and therapeutic research in acute and chronic HBV infection in children and adults.

In this prospective observational study, participants were evaluated during visits at baseline, week 24, week 48 and annually thereafter. For this analysis, week 24 visit blood test results were used to supplement any missing baseline data, and participants were required to have at least one follow-up visit occurring >24 weeks after baseline to be included. We also included data from adult follow-up visits for participants who transferred to the adult cohort study at 18 years of age.

Information collected at study visits included demographic, clinical and physical examination data including self-assessment of pubertal stage. Blood testing was performed at each visit. Participants experiencing significant alanine aminotransferase (ALT) flares (defined as ≥400 U/L in males and ≥350 U/L in females) underwent additional visits every 4 weeks until resolution of the enzyme elevations. Extra visits were undertaken as needed for participants with other unexpected clinical events. Ultrasound imaging of the liver and testing of blood alpha-fetoprotein (AFP) levels were not mandated by the study protocol, but results were collected if clinicians chose to undertake these tests.

During this HBRN cohort study, 3 participants were also enrolled in an industry-sponsored, placebo-controlled clinical trial of entecavir for immune active hepatitis B. For these participants, receipt of active treatment, timing of treatment start and laboratory results were identified in retrospect after completion of the treatment trial.

Also during this cohort study follow-up period, 52 children with normal or near-normal ALT levels (≤60 U/L in males and ≤40 U/L in females), positive HBeAg and high HBV DNA viral concentration > 10⁷ IU/mL were enrolled from this cohort study into the HBRN clinical trial of combination therapy for “immune tolerant” hepatitis B (19). Data for these children were censored at the time of enrollment into the clinical trial and included again in this analysis after completion of the post-treatment 48-week follow-up period.

Laboratory analyses

Blood was collected at each visit and the analyses presented here were based upon complete blood counts and biochemistry, including liver enzymes and bilirubin, which were tested in the site’s local laboratory. For these analysis, a standardized upper limit of the normal range for ALT was defined as ≤33 U/L for male and female infants <1 year old; ≤25 U/L for male and female children 1 year to <13 years old; ≤25 U/L for males and ≤22 U/L for female between 13 years and <18 years old; and ≤30 U/L for males and ≤20 U/L for females over 18 years of age (20–22). Laboratory-specific values were used for upper limit of normal for AST, due to variable methodology used in different laboratories.

Details of methodology for tests for HBsAg, HBeAg, anti-HBe, anti-HBs, HBV DNA, and HBV genotype are provided in the Appendix (available at www.jpeds.com).

Outcomes

Clinical outcomes

Death, liver transplantation, cirrhosis, decompensation of cirrhosis, development of HCC, and ALT flares were evaluated. These outcomes were adjudicated by an HBRN committee evaluating study data and all available clinical documentation submitted from the participating centers.

We also sought to evaluate progression of liver fibrosis. Serial liver biopsies were not considered part of routine clinical care and were not undertaken in this pediatric cohort solely for research purposes. Liver elastography measurements were not available in most centers for a major part of the study period. AST-to-platelet-ratio-index (APRI) varies markedly with acute changes in AST and therefore was expected to be an unreliable measure of fibrosis progression in this cohort. For these reasons, changes in platelet count were used to identify trends that might suggest advancing fibrosis with secondary thrombocytopenia. Age-specific, normal platelet count data were obtained from the Canadian Laboratory Initiative on Paediatric Reference Intervals (CALIPER) study investigators (personal communication, K Adeli) (23). Ultrasound scan reports were evaluated when available, to capture the appearance of the liver parenchyma, the size of the spleen, and imaging evidence of portal hypertension.

Serological and virological outcomes

We calculated the hepatitis B phenotype for participants at each follow-up time-point. Phenotypes were defined as follows: “immune tolerant” HBV infection, defined by normal serum ALT levels, positive HBeAg and high levels of HBV DNA in serum (≥10⁵ IU/mL); “immune active” chronic hepatitis B, defined by elevated ALT levels and high levels of HBV DNA (≥10⁵ IU/mL for HBeAg‐positive and ≥10⁴ IU/mL for HBeAg‐negative participants); “inactive carrier” state, defined by normal ALT levels, negative HBeAg and no detectable or only low levels of HBV DNA (≤10⁴ IU/mL). Participants who did not fulfil criteria for any one of these four categories were categorized as having an “indeterminant” phenotype (24).

We summarized the number of participants who at any time during follow-up became negative for hepatitis B e antigen (HBeAg), negative for hepatitis B surface antigen (HBsAg), whose measurement of HBV DNA became un-quantifiable (if quantifiable at baseline), and whose measurement of HBV DNA became quantifiable (if un-quantifiable at baseline). Additional analysis of these viral outcomes has been presented previously (25).

Commencement of anti-HBV treatment

We recorded the commencement of treatment for hepatitis B by the participant’s treating clinicians, including the choice of therapeutic agent. The status of participants who received treatment was evaluated at their last follow-up visit and when laboratory test results were missing at the last visit, results from the previous visit within 60 weeks were used. For comparison, we identified the number of participants who fulfilled criteria for treatment as defined in the clinical practice guideline of the American Association for the Study of Liver Diseases (AASLD) (26). The AASLD guideline suggests antiviral therapy in children (ages 2 to <18 years) who are positive for HBeAg with both elevated ALT and measurable HBV-DNA levels, noting that most studies consider ALT >1.3 times the upper limit of normal (ULN) for > 6 months and a DNA level >10⁴ IU/mL as criteria to initiate treatment (“AASLD criteria”). We also identified the number of participants fulfilling more extreme treatment criteria (ALT > 3 times the upper limit of normal for > 6 months and HBV DNA >10⁴ IU/mL) (“more extreme treatment criteria”).

Statistical analyses

Data are summarized using frequencies (percentages) for categorical variables and means (standard deviation) or medians (25^th and 75^th percentiles) for continuous variables. Outcomes are presented as the number of cases experiencing each outcome and as incident rates per 100 person-years (PY) with 95% confidence intervals (CI). Cumulative probability curves estimated by the Kaplan-Meier method were constructed for HBV treatment starts, significant ALT flares, and the virological and serological outcomes. For each outcome, we analyzed longitudinal data initially regardless of treatment status, and then censoring participants at the time of treatment initiation. Longitudinal ALT values and HBV phenotypes per individual were illustrated by heatmaps. Distributions of characteristics among participants meeting AASLD treatment criteria who received HBV treatment, and those who did not receive HBV treatment, were compared using exact Pearson chi-square or Kruskal-Wallis tests, as appropriate.

Platelet count was analyzed initially using cross-sectional enrollment data only, by age groups and compared graphically to the normal platelet count data obtained from the CALIPER study (no CALIPER data are available for children 6 month to <3 years old). We used mixed effect models to examine changes in platelet count per one-year increase in age and age-adjusted associations between platelet count and each clinical and laboratory variable, whilst accounting for correlations among repeated measures for the same participant. Changes in platelet count over time in individual participants were also examined using trajectory analysis to identify subgroups of participants with different rates of change of platelet count over time. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA) and R (version 3.5.3; R Foundation for Statistical Computing, Vienna, Austria).

Results

Of the 462 children enrolled into the HBRN pediatric cohort, 100 were excluded due to hepatitis C (n=2) or hepatitis D (n=7) co-infection, or inadequate follow-up visits (n=91), leaving 362 children for the current analysis. Baseline characteristics revealed a primarily Asian pediatric cohort (79%) with vertically acquired HBV infection (98%), mostly born outside of the USA and Canada (74%), among which HBV genotypes B (42%) and C (31%) predominated and ALT values were typically mildly elevated (66% with ALT >1–3 ULN) (Table I). The majority at enrollment were HBeAg positive (75%), with weight, height and body mass index appropriate for age.

Table 1.

Baseline characteristics (n=362)

Sex
Male	148 (41%)
Female	214 (59%)
Age at baseline
Mean(SD)	9.8 (4.4)
Median(25th:75th)	10.0 (6.1 : 13.5)
Min:Max	1.0 : 17.6
Age at baseline
6 months–<5 yrs	67 (19%)
5–<13 yrs	194 (54%)
13–<18 yrs	101 (28%)
Race	n=361
White	32 (9%)
Black	36 (10%)
Asian	286 (79%)
Other/Mixed	7 (2%)
Mode of Transmission of Hep B	n=292
Vertical transmission	286 (98%)
Horizontal/Household contact	6 (2%)
Place of Birth	n=361
US/Canada	93 (26%)
Other North America/South America	6 (2%)
Europe	21 (6%)
Asia/Australia	215 (60%)
Africa	26 (7%)
Patient Adopted
No	165 (46%)
Yes	197 (54%)
HBV Genotype	n=283
A	13 (5%)
B	118 (42%)
C	88 (31%)
D	52 (18%)
E	11 (4%)
Multiple	1 (0%)
Platelets (x103/mm3)	n=325
Median(25th:75th)	267 (229 : 311)
Min:Max	122 : 582
ALTxULN	n=355
Median(25th:75th)	1.6 (1.1 : 2.3)
Min:Max	0.3 : 82.7
ALTxULN	n=355
≤1 ULN	69 (19%)
>1–3 ULN	235 (66%)
>3–5 ULN	25 (7%)
>5 ULN	26 (7%)
ASTxULN	n=355
Median(25th:75th)	0.9 (0.8 : 1.2)
Min:Max	0.3 : 27.5
ASTxULN	n=355
≤1 ULN	209 (59%)
>1–3 ULN	134 (38%)
>3–5 ULN	7 (2%)
>5 ULN	5 (1%)
HBeAg	n=350
Negative	89 (25%)
Positive	261 (75%)
HBV DNA (log10 IU/mL)	n=358
Median(25th:75th)	8.1 (5.2 : 8.4)
Min:Max	BLD : 9.3
qHBeAg (log10 IU/mL)	n=252
Median(25th:75th)	3.1 (−0.6 : 3.3)
Min:Max	BLD : 3.8
qHBsAg (log10 IU/mL)	n=251
Median(25th:75th)	4.4 (4.0 : 4.8)
Min:Max	−0.2 : 5.6
Anti-HBe	n=317
Negative	227 (72%)
Positive	90 (28%)
Anti-HBs	n=305
Negative	289 (95%)
Positive	16 (5%)
BMI Z-score	n=349
Median(25th:75th)	0.1 (−0.7 : 0.8)
Min:Max	−2.9 : 2.2
Height Z-score	n=358
Median(25th:75th)	−0.2 (−0.9 : 0.5)
Min:Max	−3.6 : 2.6
Weight Z-score	n=361
Median(25th:75th)	−0.1 (−0.9 : 0.7)
Min:Max	−3.4 : 2.6
Tanner Stage: Physical Growth	n=339
I	186 (55%)
II	28 (8%)
III	49 (14%)
IV	58 (17%)
V	18 (5%)
Tanner Stage: Pubic Hair Growth	n=338
I	204 (60%)
II	23 (7%)
III	37 (11%)
IV	53 (16%)
V	21 (6%)

Outcomes observed during a median duration of follow-up of 4.2 years (range 0.8 to 7.9 years), their incidence rates and cumulative probabilities are presented in Table 2 and Figure 1. One participant died from a cause not related to liver disease (accidental choking). One female participant developed cirrhosis during follow-up at week 71 (at age 14 years) with liver histology showing fibrosis stage 4/4. This participant had a normal liver ultrasound scan at baseline and no associated co-morbidities. Baseline laboratory results revealed an ALT 152 U/L, AST 105 U/L and platelet count of 172 ×10³/mm³. Platelet count subsequently fell to 148 ×10³/mm³ at the time of liver biopsy.

Table 2.

Selected outcomes and incidence rates (n=362)

	Regardless of treatment status				Censored at HBV treatment initiation
Outcomes	na	Incident cases	person-years	Incidence rate per 100 person-years (95%CI)	na	Incident cases	person-years	Incidence rate per 100 person-years (95%CI)
Clinical outcomes
Death	362	1b	1489	0.07 (0.01, 0.48)	362	1b	1153	0.09 (0.01, 0.62)
Cirrhosis	362	1	1489	0.07 (0.01, 0.48)	362	1	1153	0.09 (0.01, 0.62)
Decompensation	362	0	-	-	362	0	-	-
HCC	362	0	-	-	362	0	-	-
Serological outcomes
ALT flare (male ≥400, female ≥350 U/L)	337	13	1214	1.07 (0.62, 1.84)	315	11	1055	1.04 (0.58, 1.88)
Ever become HBeAg-Negative	240	50	759	6.59 (4.99, 8.69)	220	29	636	4.56 (3.17, 6.56)
Ever become HBsAg-Negative	332	8	1188	0.67 (0.34, 1.35)	312	4	1045	0.38 (0.14, 1.02)
Virological outcomes
Ever become HBV DNA unquantifiable c	325	49	1063	4.61 (3.48, 6.10)	306	27	949	2.85 (1.95, 4.15)
Ever become HBV DNA quantifiable d	14	8	43	18.67 (9.34, 37.33)	14	8	43	18.67 (9.34, 37.33)

^aNumber of participants who were eligible at enrollment for each outcome and had at least one follow-up result.

^bCause of death: choking.

^cIf quantifiable at enrollment.

^dIf unquantifiable at enrollment.

Figure 1.

Cumulative probability of (A) onset of ALT flare; (B) becoming HBeAg negative for patients who were HBeAg positive at baseline; (C) becoming HBsAg negative; (D) developing unquantifiable HBV DNA for patients who had quantifiable HBV DNA at baseline; and, (E) commencing standard of care anti-HBV treatment.

No episodes of decompensation of cirrhosis, ascites, variceal hemorrhage or HCC were identified.

Serological and virological outcomes

Participants with HBeAg loss during follow-up

Of the 240 children who were HBeAg positive at baseline and had follow-up evaluation for HBeAg status, 50 became HBeAg negative (cumulative probability of 23% at the 4th year of follow up) (Table 2 and Figure 1, B). Among these 50, 40 were tested negative for anti-HBe before loss of HBeAg, of whom 34 developed anti-HBe.

The 50 participants who became HBeAg negative could be categorized into 4 groups according to the pattern of change in ALT, HBV DNA, and anti-viral treatment, after removal of 5 participants with inadequate follow-up after HBeAg loss. A significant decrease in DNA level was seen in 20 participants in whom ALT was initially either normal or elevated and became normal or near normal after HBeAg loss (Figure 2, A; available at www.jpeds.com). In 4 participants, HBV DNA remained high and ALT remained normal or elevated (Figure 2, B; available at www.jpeds.com). In 16 participants, treatment with entecavir (n=15) or lamivudine (n=1) was associated with a significant decline in HBV DNA values and an initially normal or high ALT level fell to normal or near normal after HBeAg loss (Figure 2, C; available at www.jpeds.com). Interferon treatment in 5 participants was associated with a significant reduction in HBV DNA and normalization or near normalization of initially elevated ALT levels (Figure 2, D; available at www.jpeds.com).

Figure 2.

Sequential lab plots in four example patients demonstrating four different patterns of HBeAg loss. (2A) A significant fall in DNA was seen in 20 participants in whom ALT was initially either normal or elevated and became normal or near normal after HBeAg loss. (2B) In 4 participants, HBV DNA remained high and ALT remained normal or elevated. (2C) In 16 participants, treatment with entecavir (n=15) or lamivudine (n=1) was associated with a significant fall in HBV DNA and an initially normal or high ALT fell to normal or near normal after HBeAg loss. (2D) Interferon treatment in 5 participants was associated with a significant reduction in HBV DNA and normalization or near normalization of initially elevated ALT.

At final follow-up, an elevated ALT level (>1 ULN) was noted in 27 of the 45 (60%) participants after HBeAg loss (21 with >1–2x ULN, 2 with >2–3x ULN, and 4 with >3x ULN). High HBV DNA (≥4 log ₁₀ IU/ml) was found at baseline in 44 of these 45 participants. At last follow-up, 12 (27%) had elevated HBV DNA ≥4 log ₁₀ IU/ml (Table 3; available at www.jpeds.com).

Table 3.

Baseline and last ALT and HBV DNA for (a) 88 participants who were HBeAg (−) at baseline and last follow-up, and (b) 45 participants who were HBeAg (+) and lost HBeAg during follow-up, stratified by whether receiving HBV treatment before losing HBeAg

	(a) HBeAg (−) at baseline & at last follow-up n=88	(b) Received HBV treatment before losing HBeAg(−)
		No n=241	Yes n=21
Baseline ALT x ULN	n=87	n=24	n=21
Median (25th: 75th)	1.5 (1.0 : 1.9)	2.3 (1.4 : 7.4)	5.6 (3.0 : 7.8)
Min:Max	0.3 : 50.1	0.8 : 82.7	1.3 : 54.6
Baseline ALTxULN	n=87	n=24	n=21
≤1 ULN	22 (25%)	2 (8%)	0 (0%)
>1–2 ULN	48 (55%)	9 (38%)	2 (10%)
>2–3 ULN	13 (15%)	4 (17%)	3 (14%)
>3 ULN	4 (5%)	9 (38%)	16 (76%)
Last ALT x ULN	n=88	n=24	n=21
Median (25th: 75th)	1.3 (1.0 : 1.7)	1.2 (1.0 : 1.5)	1.0 (0.8 : 1.3)
Min:Max	0.5 : 15.7	0.5 : 11.1	0.5 : 7.9
Last ALTxULN	n=88	n=24	n=21
≤1 ULN	26 (30%)	i 8 (33%)	10 (48%)
>1–2 ULN	52 (59%)	13 (54%)	8 (38%)
>2–3 ULN	9 (10%)	0 (0%)	2 (10%)
>3 ULN	1 (1%)	3 (13%)	1 (5%)
Baseline HBV DNA (log10 IU/mL)	n=88	n=24	n=21
Median (25th: 75th)	2.6 (2.0 : 3.8)	7.4 (6.1 : 8.4)	8.0 (7.9 : 8.3)
Min:Max	<LLOQ : 8.5	3.5 : 9.3	6.8 : 8.5
Baseline HBV DNA (log10 IU/mL)	n=88	n=24	n=21
<3	51 (58%)	0 (0%)	0 (0%)
3 – <4	20 (23%)	1 (4%)	0 (0%)
4 – <5	9 (10%)	2 (8%)	0 (0%)
5 – <7	6 (7%)	7 (29%)	2 (10%)
7 – <8	0 (0%)	5 (21%)	6 (29%)
≥8	2 (2%)	9 (38%)	13 (62%)
Last HBV DNA (log10 IU/mL)	n=88	n=24	n=20
Median (25th: 75th)	2.4 (1.3 : 3.3)	2.4 (1.7 : 4.4)	1.3 (<LLOQ : 4.2)
Min:Max	<LLOQ : 8.2	<LLOQ : 8.0	<LLOQ: 8.8
Last HBV DNA (log10 IU/mL)	n=88	n=24	n=20
<3	61 (69%)	14 (58%)	15 (75%)
3 – <4	20 (23%)	3 (13%)	0 (0%)
4 – <5	5 (6%)	5 (21%)	0 (0%)
5 – <7	1 (1%)	1 (4%)	3 (15%)
7 – <8	0 (0%)	0 (0%)	0 (0%)
≥8	1 (1%)	1 (4%)	2 (10%)

Participants with HBsAg loss during follow-up

Eight of 332 participants with follow-up evaluation of HBsAg became HBsAg negative (cumulative probability of 3% at the 4^th year of follow-up) (Table 2 and Figure 1, C). HBsAg loss occurred in 4 participants after initiation of anti-HBV treatment, each of whom had sustained HBsAg seroconversion (e.g. at least two consecutive results of HBsAg negative and anti-HBs positive). At the last follow-up, one of 8 participants who became negative became HBsAg positive again. Five of 7 children who had sustained HBsAg loss had normal ALT levels and two children had mildly elevated ALT levels (1.1 and 1.5 xULN) at the last follow-up. All 6 children who were still HBsAg negative had HBV DNA below level of quantification at the time of last follow up test and the remaining child did not have HBV DNA results from the last visit.

Phenotype

Phenotype was calculated in 312 participants at baseline, showing 27% (n=86) to be immune tolerant, 44% (n=136) with HBeAg positive hepatitis, 3% (n=10) with HBeAg negative hepatitis, 12% (n=37) inactive carriers and 10% (n=32) indeterminant phenotype and 4% (n=11) had started antiviral treatment after enrollment but before the initial phenotype could be determined. At last available follow-up, 66% (57/86) of those initially immune tolerant remained as immune tolerant, 49% (66/136) remained as HBeAg positive CHB, 30% (3/10) remained as HBeAg- CHB, 78% (29/37) remained as inactive carriers, and 56% (18/32) remained as indeterminant phenotype (Table 4 and Figure 3; available at www.jpeds.com).

Table 4.

Cross-listing of calculated phenotype at enrollment and at the last available follow-up (n=312)

		Calculated phenotype at enrollment
	All n=312	IT trial participants n=47 (15%)*		Immune tolerant	HBeAg+ CHB	HBeAg− CHB	Inactive carrier	Indeterminant	HBV drugs
		Immune tolerant	HBeAg+ CHB
		n=38 (12%)	n=9 (3%)	n=48 (15%)	n=127 (41%)	n=10 (3%)	n=37 (12%)	n=32 (10%)	n=11 (4%)
Last available phenotype
HBsAg negative	7 (2%)	1 (3%)	0 (0%)	0 (0%)	2 (2%)	1 (10%)	2 (5%)	0 (0%)	1 (9%)
Immune tolerant	80 (26%)	23 (61%)	3 (33%)	34 (71%)	18 (14%)	0 (0%)	0 (0%)	1 (3%)	1 (9%)
HBeAg+ CHB	80 (26%)	7 (18%)	2 (22%)	6 (13%)	64 (50%)	0 (0%)	0 (0%)	0 (0%)	1 (9%)
HBeAg− CHB	7 (2%)	0 (0%)	0 (0%)	2 (4%)	1 (1%)	3 (30%)	0 (0%)	1 (3%)	0 (0%)
Inactive carrier	67 (21%)	0 (0%)	0 (0%)	3 (6%)	14 (11%)	4 (40%)	29 (78%)	12 (38%)	5 (45%)
Indeterminant	36 (12%)	0 (0%)	0 (0%)	2 (4%)	7 (6%)	2 (20%)	6 (16%)	18 (56%)	1 (9%)
HBV drugs	26 (8%)	0 (0%)	2 (22%)	1 (2%)	21 (17%)	0 (0%)	0 (0%)	0 (0%)	2 (18%)
In the IT trial	9 (3%)	7 (18%)	2 (22%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Ever received HBV treatment during study follow-up		n=38	n=9	n=48	n=127	n=10	n=37	n=32	n=11
Yes		38 (100%)	9 (100%)	2 (4%)	33 (26%)	1 (10%)	0 (0%)	0 (0%)	3 (27%)
No		0 (0%)	0 (0%)	46 (96%)	94 (74%)	9 (90%)	37 (100%)	32 (100%)	8 (73%)

^*Among 47 participants who enrolled in the IT trial, 38 were immune tolerant and 9 were HBeAg+ CHB at the pediatric cohort enrollment. Pediatric IT clinical trial enrolled participants with normal or near-normal ALT (≤60 U/L in males and ≤40 U/L in females).

IT: Immune tolerant, CHB: chronic hepatitis B

Figure 3.

Heatmap of phenotype by baseline phenotype included 312 participants who had first available phenotype at week 0 and had at least one additional follow-up phenotype result.

Participants with negative HBeAg at baseline

Among the 89 children who were HBeAg negative at baseline, 2 children became HBeAg positive during follow-up, one with low levels of quantitative HBeAg levels (0.33 to 0.55 IU/mL) which then reverted to negative, and the other remained HBeAg positive at the last follow-up. At last follow-up, 62 of these 88 HBeAg-negative children (70%) had elevated ALT levels (52 with >1–2 ULN, 9 with >2–3 ULN, and 1 with 3> ULN) and 7 had HBV DNA levels ≥4 log ₁₀ IU/ml (Table 3).

HBV DNA

Of the 325 children with quantifiable HBV DNA at baseline, 82% had a high HBV DNA level ≥ 4 log ₁₀ IU/mL. 49 children had at least one unquantifiable HBV DNA level during follow-up (cumulative probability of 18% at the 4^th year of follow-up) (Table 2 and Figure 1, D). Twenty-two of these 49 children achieved this outcome after starting anti-HBV treatment and 27 without treatment initiation. Of the 27 children whose HBV DNA levels became unquantifiable spontaneously, 17 had HBV DNA < 4 log ₁₀ IU/mL and 20 were HBeAg negative at baseline. During follow-up 7 of the 27 children had undetectable HBV DNA levels and one was HBsAg negative at the time of the last visit. Fourteen children had detectable but unquantifiable HBV DNA levels at baseline, of whom 8 were found to develop quantifiable levels of HBV DNA with median 2.1 log ₁₀ IU/mL, ranging between 1.4 and 5.1 log ₁₀ IU/mL during follow-up.

Changes in ALT and ALT flares

ALT levels were above the ULN in 81% of the cohort at baseline. During follow-up, this proportion steadily decreased, such that 72% had elevated ALT levels at the last clinic visit. Measured ALT levels remained in the normal range throughout follow-up in 37 participants. In those with abnormal ALT levels recorded during follow-up, the peak value was below 2xULN in 138 participants (40%), 2–3xULN in 74 (21%), 3–5xULN in 41 (12%), and >5xULN in 56 (16%). ALT levels fluctuated significantly during follow-up in individual patients with, for example, 21 (38%) of 56 participants with peak ALT >5xULN also recording normal ALT values at other time points (Figure 4; available at www.jpeds.com).

Figure 4.

Heatmap of ALT level during follow up of 346 participants who had at least two visits with ALT values.

Onset of significant ALT flares were observed during follow-up in 13 children (Table 5; available at www.jpeds.com). Two children developed a second flare after resolution of the first, but this analysis includes only the first flare for each. The cumulative incidence of significant ALT flares at 4 years of follow-up was 4% (Table 2 and Figure 1, A). Only one flare occurred in a child within the usual age range for onset of puberty (8–13 years in girls, 9–14 years in boys). HBeAg status changed from positive before the flare to negative after the flare in 5 children, whose baseline characteristics, severity of flare and likelihood of receiving treatment (2 of these 5 were started on anti-HBV treatment) were similar to the children who did not become HBeAg negative. All five children undergoing HBeAg loss after a significant flare developed anti-HBe.

Table 5.

Characteristics of ALT flares in 13 participants.

Characteristics	n or median (min - max)
Sex	n=13
Female	6
Male	7
Age at flare onset (years)	14.2 (3.5 – 18.7)
Pubertal stage at flare onset	n=13
I (prepubertal)	6
II	0
III	1
IV	4
V	2
HBV genotype	n=13
A	1
B	8
C	2
D	1
Unknown	1
Biochemical changes	n=13
Peak ALT recorded (U/L)	560 (352 – 1710)
Elevation of bilirubin	n=5
Peak bilirubin recorded (mg/dL)	1.6 (1.2 – 2.5)
Treatment
On treatment at flare onset	n=0
Treatment started during flare	n=6
Entecavir	4
Interferon	1
Lamivudine	1
Time from flare onset to treatment start (weeks)	6.8 (1.3 – 20.4)
Flare Resolution Duration a	n=11
Time from flare onset to resolution (weeks)	8.6 (1.3 – 70.0)
HBeAg	n=11
HBeAg change from positive to negative after flare	5
HBeAg positive before and after flare	4
HBeAg negative before and after flare	2
Anti-HBe antibody	n=11
Anti-HBe change from negative to positive during flare	5
Anti-HBe positive before and after flare	2
Anti-HBe negative before and after flare	4
HBV DNA	n=11
HBV DNA at flare onset (log10 IU/mL)	8.2 (6.2 to 8.8)
HBV DNA nadir after flare (log10 IU/mL)	2.1 (<limit of detection - 8.2)
HBV DNA <4 log10 IU/mL after flare	7
Adjudicated cause of flare	n=13
HBV reactivation- rising HBV DNA during the flare	4 (including 1 after withdrawal of anti-HBV therapy)
Immune clearance- decrease in HBV DNA and loss of HBeAg	2 (including 1 who failed to sustain the loss of HBeAg)
ALT flare without change in HBV DNA	1
Unknown	6

^aResolution of flare defined as ALT falling below the flare diagnosis threshold of 400 U/L for males and 350 U/L for females. Eleven participants had available follow-up data after flare onset.

Commencement of anti-HBV treatment

Anti-HBV treatment was initiated at the treating physician’s discretion in 40 children during the study period with entecavir (n=25), interferon or peginterferon (n=11), lamivudine (n=3) or tenofovir disoproxil fumarate (n=1). The cumulative probability of treatment commencement in the study cohort was 13% by the 4^th year of follow-up (Figure 1, E). Incidence rates for treatment initiation were similar between most patient subgroups although higher among those with elevated ALT levels and lower among those negative for HBeAg (Table 6; available at www.jpeds.com). Five of these 40 children underwent a second course of treatment (2 with entecavir, 1 interferon, 2 tenofovir).

Table 6.

Incidence rates of standard-of-care HBV treatment initiation after enrollment among 362 participants

Baseline Characteristic		No. of Participants	No. w/Outcome	Total PYR	Incidence/100 PYR (95% CI) a
Age (Years)	6 months-<10	180	21	632	3.32 (2.17, 5.09)
	≥10	182	19	521	3.65 (2.33, 5.72)
Sex	Male	148	13	482	2.70 (1.57, 4.65)
	Female	214	27	671	4.02 (2.76, 5.86)
Asian	Asian	286	33	899	3.67 (2.61, 5.16)
	Non-Asian	75	7	247	2.83 (1.35, 5.94)
Patient Adopted	No	165	15	533	2.81 (1.70, 4.67)
	Yes	197	25	620	4.03 (2.73, 5.97)
Caregiver Highest Education	Below Bachelor’s degree	114	12	339	3.54 (2.01, 6.23)
	Bachelor’s degree and above	232	27	749	3.61 (2.47, 5.26)
Annual Household Income	<$50,000	60	4	181	2.21 (0.83, 5.88)
	$50,000– <$100,000	83	13	252	5.17 (3.00, 8.90)
	≥$100,000	119	12	406	2.95 (1.68, 5.20)
HBV Genotype	A	13	3	40	7.59 (2.45, 23.52)
	B	118	13	339	3.84 (2.23, 6.61)
	C	88	11	320	3.44 (1.91, 6.21)
	D	52	3	199	1.51 (0.49, 4.67)
	E	11	2	32	6.33 (1.58, 25.33)
	Multiple	1	0	2	–
HBeAg	Negative	89	2	331	0.60 (0.15, 2.41)
	Positive	261	38	777	4.89 (3.56, 6.72)
HBV DNA (log10 IU/mL)	<4	75	0	288	–
	4–<8	88	17	255	6.67 (4.15, 10.73)
	≥8	195	23	603	3.82 (2.54, 5.74)
qHBeAg (log10 IU/mL)	<3	113	14	367	3.81 (2.26, 6.43)
	3–<5	139	18	408	4.41 (2.78, 7.00)
qHBsAg (log10 IU/mL)	<3	17	0	69	–
	3–<5	206	28	607	4.62 (3.19, 6.69)
	≥5	28	4	91	4.42 (1.66, 11.77)
ALT x ULN	≤1 x ULN	69	0	232	–
	>1 – 2xULN	173	6	604	0.99 (0.45, 2.21)
	>2 x ULN	113	33	297	11.10 (7.89, 15.61)
Platelets (x103/mm3)	<200	41	6	106	5.69 (2.55, 12.66)
	≥200	284	34	929	3.66 (2.61, 5.12)

^aConfidence interval (CI) based on Wald

PYR = person-years

Before the 40 children received treatment, 37 were HBeAg positive, all were HBsAg positive and all had HBV DNA levels > 5.0 log ₁₀ IU/mL. At the last evaluation of the 39 patients who had available data after treatment initiation (median duration of treatment 96 weeks, range 20 – 361 weeks), 18 were still receiving treatment, 14 had normal ALT (≤ 1 ULN), 24 had elevated ALT (median 45 U/L, range 23 – 237 U/L), (ALT unknown in 1), 22 were HBeAg negative (13 positive and 4 unknown), 19 anti-HBe positive (13 negative and 7 unknown), 4 were HBsAg negative (28 positive, 7 unknown), and HBV DNA levels were less than 4.0 log ₁₀ IU/mL in 28, greater than this level in 10, and unknown in 1 patient.

Analysis of the cohort data revealed that 129 participants fulfilled the AASLD criteria for anti-HBV treatment (including ALT>1.3xULN for >6 months) during follow-up and 27 fulfilled more extreme treatment criteria (including ALT>3xULN for > 6 months), of whom only a subgroup received standard of care HBV treatment (25 [19%] and 14 [52%], respectively) (Table 7). Among children fulfilling the AASLD criteria, treatment rates were higher in the early years of the study (2011–2015, when 18 of 56 [32%] eligible children were treated) compared with the later years (2016–2018, when 7 of 73 [10%] eligible children were treated). Among treated children, the median pre-treatment peak ALT did not differ between the two eras.

Table 7.

Fulfilment of AASLD anti-HBV treatment criteria and more extreme criteria

	ALT elevated > 6 months	ALT elevated > 1 year
AASLD Guideline Criteria
ALT persistently elevated >1.3xULN a and other criteriab	129	110
Received standard of care HBV treatment after ALT elevation	25/129(19%)	21/110 (19%)
More extreme criteria
ALT persistently elevated >3xULN a and other criteriab	27	14
Received standard of care HBV treatment after ALT elevation	14/27 (52%)	7/14 (50%)

^aAASLD guideline recommends ULN for ALT of 35 U/L in males and 25 U/L in females.

^bOther criteria include HBeAg+, HBV DNA >10⁴, no HBeAg loss or anti-HBe positive at around 6 months.

Sociodemographic profiles were similar between treated versus non-treated groups except that children who were treated had significantly higher peak ALT before treatment initiation. At last follow-up, those who did not receive treatment had higher HBV DNA levels, and were more likely to be HBeAg positive and anti-HBe negative, than those who were treated (Table 8; available at www.jpeds.com). Platelet counts at last follow-up were similar between the two groups, and none of these untreated participants became HBsAg negative.

Table 8.

Sociodemographics and lab results among 129 children who fulfilled the AASLD criteria for anti-HBV treatment.

	Not treated with HBV therapy after meeting AASLD guideline n=104	Treated with HBV therapy after meeting AASLD guideline n=25
Age (years) at the beginning of flare
6 months-<10	64 (62%)	12 (48%)
≥10	40 (38%)	13 (52%)
Sex
Male	43 (41%)	8 (32%)
Female	61 (59%)	17 (68%)
Race
White	8 (8%)	0 (0%)
Black	10 (10%)	4 (16%)
Asian	83 (80%)	21 (84%)
Other/Mixed	3 (3%)	0 (0%)
Patient Adopted
No	56 (54%)	9 (36%)
Yes	48 (46%)	16 (64%)
Time Since Migration (years)
Born US/Canada	37 (36%)	6 (24%)
>10–20 years ago	7 (7%)	5 (20%)
≤10 years ago	57 (55%)	14 (56%)
Unknown Timing	3 (3%)	0 (0%)
Caregiver Highest Education
Below Bachelor’s degree	37 (36%)	7 (28%)
Bachelor’s degree and above	61 (59%)	17 (68%)
Missing	6 (6%)	1 (4%)
Annual Household Income
<$50,000	22 (21%)	4 (16%)
$50,000– <$100,000	19 (18%)	8 (32%)
≥$100,000	36 (35%)	7 (28%)
Missing	27 (26%)	6 (24%)
Type of Insurance
None or Self Pay	1 (1%)	0 (0%)
Medicaid	3 (3%)	1 (4%)
Tricare	1 (1%)	1 (4%)
Government	54 (52%)	8 (32%)
Private	43 (41%)	15 (60%)
Multiple	1 (1%)	0 (0%)
Prefer Not to Answer	1 (1%)	0 (0%)
Max of ALT during the flare or before treatment **
Median(25th:75th)	70 (53 : 93)	151 (96 : 231)
Min:Max	35 : 2067	62 : 1710
Total follow-up weeks
Median(25th:75th)	222.6 (131.3 : 282.5)	238.6 (187.1 : 336.6)
Min:Max	42.6 : 402.4	30.7 : 405.4
Age at the last visit
Median(25th:75th)	11.9 (8.4 : 17.2)	16.2 (10.3 : 17.5)
Min:Max	2.1 : 24.4	5.4 : 21.4
Lab results at the last follow-up visit
Platelets (x103/mm3)	n=95	n=22
Median(25th:75th)	243 (204 : 300)	248 (235 : 302)
Min:Max	123 : 390	133 : 424
ALT (U/L)	n=103	n=25
Median(25th:75th)	49 (36 : 68)	37 (23 : 66)
Min:Max	14 : 782	7 : 352
ALT x ULN-AASLD	n=103	n=25
Median(25th:75th)	1.8 (1.4 : 2.2)	1.4 (0.9 : 2.4)
Min:Max	0.6 : 31.3	0.3 : 14.1
HBV DNA (log10 IU/mL) **	n=102	n=24
Median(25th:75th)	8.1 (7.3 : 8.5)	2.0 (<LLOD : 5.9)
Min:Max	<LLOD : >ULOQ	<LLOD : 8.8
HBV DNA (log10 IU/mL) **
≤ 4	9 (9%)	15 (60%)
> 4	93 (89%)	9 (36%)
Missing	2 (2%)	1 (4%)
HBeAg **
Negative	13 (13%)	10 (40%)
Positive	89 (86%)	12 (48%)
Missing	2 (2%)	3 (12%)
HBsAg *
Negative	0 (0%)	1 (4%)
Positive	96 (92%)	20 (80%)
Missing	8 (8%)	4 (16%)
Anti-HBe **
Negative	80 (77%)	12 (48%)
Positive	13 (13%)	8 (32%)
Missing	11 (11%)	5 (20%)

LLOD: lower limit of detection; ULOQ: upper limit of quantification.

AASLD guideline recommends ULN for ALT of 35 U/L in males and 25 U/L in females.

^*p<0.05

^**p<0.01

Platelet count

Cross-sectional analysis of platelet count at baseline revealed an association with age (Figure 5; available at www.jpeds.com). Compared with a 1-year younger child, an older child’s platelet count was lower by 5.8 ×10³ per mm³ (95%CI: −7.3, −4.2) in the HBRN cohort. This result was similar to that observed in healthy children in the CALIPER study (−5.7 ×10³ per mm³, 95%CI: −7.3, −4.0) (23). A similar relationship between platelet count and age was also found by examining HBRN longitudinal pediatric cohort data (Table 9; available at www.jpeds.com). Age-adjusted analysis of longitudinal platelet count data using mixed effect models showed an association between lower platelet count and elevation of ALT levels (Table 9). Two patterns for this association were noted, including an acute fall in platelet count during a transient ALT flare, with subsequent return to baseline values (Figure 6, A; available at www.jpeds.com), and a steady sustained fall in platelet count associated with chronic ALT elevations (Figure 6, B; available at www.jpeds.com). Trajectory analysis did not reveal a subgroup of participants with more rapid decline in platelet count during study follow-up (results not shown).

Figure 5.

Platelet count (mean and 95% confidence intervals) by age in healthy children (CALIPER study) vs HBRN pediatric cohort at baseline (cross-sectional analysis). CALIPER data were obtained via personal communication from the CALIPER senior author (23)(Adeli 2015).

Table 9.

Age-adjusted differences in platelet counts (x10³/mm³)^a (longitudinal analysis: 354 children with 1414 observations with platelets).

Predictors	Observations	Age-adjusted association for each predictor
		Change or differences in platelets: beta (95% CI)	Type III p
Age, years	obs=1414	−4.9 (−5.7, −4.1)	<0.001
Sex	obs=1414		0.15
Female	848 (60%)	8.3 (−3.0, 19.6)
Male	566 (40%)	ref
ALTxULN	obs=1407		<0.001
≤ 1 ULN	366 (26%)	ref
> 1–3 ULN	898 (64%)	−7.1 (−13.6, −0.7)
> 3–5 ULN	78 (6%)	−18.2 (−29.5, −6.9)
> 5 ULN	65 (5%)	−26.7 (−37.7, −15.7)
ASTxULN	obs=1398		<0.001
≤ 1 ULN	1007 (72%)	ref
> 1–3 ULN	357 (26%)	−11.7 (−17.2, −6.2)
> 3–5 ULN	20 (1%)	−22.1 (−39.2, −5.0)
> 5 ULN	14 (1%)	−20.8 (−40.8, −0.7)
HBV DNA (log10 IU/mL)	obs=1364	0.4 (−0.9, 1.7)	0.56
qHBeAg (log10 IU/mL)	obs=1057	−0.4 (−2.8, 1.9)	0.72
qHBsAg (log10 IU/mL)	obs=1031	2.7 (−0.9, 6.4)	0.14
HBeAg status	obs=1336
Negative	474 (35%)	3.1 (−5.2, 11.4)	0.47
Positive	862 (65%)	ref
Currently treated with Peg-IFN/IFN (4 weeks washout period)	obs=1414		<0.001
Yes	11 (1%)	−73.9 (−96.6, −51.3)
No	1403 (99%)	ref

^aMixed effects models were used to account for correlations among repeated measures for the same child.

Figure 6.

Two case studies. Sequential platelet count and ALT measurements in two patients demonstrating different patterns of association between the two variables.

AFP monitoring

Clinicians measured AFP in 75% of participants at baseline, and in 63–79% at each subsequent annual follow-up visit. Recorded AFP values were normal (<10 ng/ml) in 99% of tests, and below 50ng/ml in all tests. Mild elevations of AFP in 15 children (median 21.4 ng/ml, range 10–44.8 ng/ml), returned to normal in 13 children, and reduced towards normal in 2 children at last follow-up (44.8 to 25.5 ng/ml in one child, 20 to 10 ng/ml in another child). In 5 of these children, mild AFP elevation coincided with significant elevation of ALT (median 654 U/l, range 143 to 2067 U/l). All but one of the 15 children had ultrasound scans performed that did not show focal lesions or nodular liver parenchyma.

Ultrasound imaging

Clinicians completed liver ultrasound imaging in 64% of participants at baseline and in 29–38% at each subsequent annual follow-up visit. Overall, 47% (range 17% to 77% in different centres) of participants underwent a liver ultrasound scan during follow-up. Clinical reports most commonly noted mild liver parenchymal abnormalities (13% of baseline scans), including appearances suggestive of steatosis. Splenomegaly was reported in a small number of participants (4% of baseline scans). One participant underwent ultrasound scan at their last study visit (week 96) that showed a 1.2cm left lobe lesion, AFP result was not available at this visit and the lesion was not demonstrated on a subsequent MRI scan.

Discussion

We prospectively followed a large cohort of children with HBV infection in North America, with data and biospecimen collection from a diverse group of 362 children occurring over 1,489 person-years of follow-up. Active hepatitis was common throughout follow-up, demonstrated by elevated ALT levels and an immune active phenotype, including among many participants who transitioned from HBeAg positive to negative during follow-up, and among those who were HBeAg negative throughout. Many children thus appeared to remain at risk for advancing severity of fibrotic liver disease and fulfilled recommended treatment criteria, although clinicians seldom provided anti-HBV treatment. A low incidence of cirrhosis and no other major clinical outcomes occurred during these childhood years, such as cirrhotic decompensation events, HCC, or death related to liver disease.

Our data confirm the rarity of end stage liver disease in children with chronic HBV infection. However, it is inaccurate to consider HBV infection in children as benign, in which most children are “immune tolerant” to the virus and will transition into an inactive carrier state over time. We show that active hepatitis phenotypes, elevated ALT levels and high HBV viral concentrations are prevalent in the cohort throughout follow-up, including participants with more significant degrees of ALT elevation (e.g. >3x ULN). Being HBeAg negative, or transitioning from HBeAg positive to negative during follow-up, did not ensure entry into an inactive carrier state although for some it did herald eventual loss of HBsAg. The variable progression of patients through different phenotypes, and occasional patients with unexpected combinations of laboratory results (eg. negative HBeAg, normal ALT, high HBV DNA) illustrates the need to better understand the complexity of host-virus interactions. Previous studies have shown variable findings, including similar proportions to our study with active hepatitis phenotypes during follow-up in Japan, but a lower proportion of children continuing with active hepatitis after HBeAg loss in studies from Italy, Spain, and Taiwan (7,8,10,13). These differences between studies may relate to race (e.g. Caucasian compared with Asian), route of infection (vertical vs. horizontal), length of follow-up, definition of normal range of ALT and AST, or other variables.

There is a lack of pediatric data describing risk factors for progression of liver fibrosis due to hepatitis B during childhood, although in adults with HBV infection the development of cirrhosis and HCC are associated with elevated ALT levels, HBeAg status and high HBV DNA titers (27). Extrapolation of these data to children would suggest that many of the participants in our cohort were at risk of progressive liver fibrosis. Unfortunately, serial liver biopsy and elastography measures were not available in our study; furthermore, a previous study of pediatric liver biopsies suggested that fibrosis stage may not correlate with age, duration of HBV infection, HBV genotype or HBV DNA level (28). We show that elevated ALT levels were associated with low platelet counts, which may be a biomarker for greater severity of liver fibrosis. However, further studies utilizing alternative markers of disease activity (other than ALT), and with serial liver biopsies or validated and accurate non-invasive tests of liver fibrosis in children, will be required to definitively address this issue.

Clinicians seldom started treatment for children in this cohort, although our data suggest that treatment should have been given more frequently according to published guidelines (26). We speculate that low treatment rates reflect clinicians’ perception of unsatisfactory outcomes of therapy including the rare achievement of clearance of HBsAg, concerns about the need for long-term therapy with nucleot(s)ide analogues, the possibility of antiviral resistance, the known, common adverse effect profile of interferon-based therapies, and awareness of the rarity of significant adverse outcomes of HBV infection during childhood (29). Clinicians seemed more likely to start treatment in children with higher ALT levels, suggesting that greater concern about poor clinical outcomes associated with more abnormal aminotransferase levels may be the major influence in decision making. Children with ALT flares were more likely to receive treatment probably because treating clinicians saw this high elevation of ALT as the indication for treatment. However, we did not collect data specifically to explore physician and patient/family approaches to treatment and why many of the children in our study remained untreated.

The low likelihood of children and adults with hepatitis B receiving treatment has been recognized as a problem globally (30). In particular, several barriers to anti-HBV care are common among immigrant populations, including lack of knowledge of HBV, language and cultural barriers, costs associated with healthcare and with potentially long-term drug treatment, conflicting family priorities related to new settlement, and adverse prior experience of health care (31,32). We showed no pattern of difference in incidence rate of treatment initiation related to family income, or the educational status of the caregiver. Some of the barriers identified in prior studies were probably less relevant to many of these families, including those with international adoptee children, as all of them were linked to healthcare in order to be included in the study. Higher incidence rates of treatment initiation were noted among international adoptee children compared with non-adopted children, although the confidence intervals for each of these estimates were wide and overlapping.

The small number of children treated precluded a detailed multivariable analysis of the impact of treatment on outcomes. Children who fulfilled AASLD treatment criteria but remained untreated had higher viral loads and were more likely to be HBeAg positive and anti-HBe negative at the end of follow-up, but the significance of these findings on clinical outcomes including progression of liver fibrosis remains unclear.

Major clinical outcomes during this childhood follow-up period were rare in our cohort. A low rate of diagnosis of cirrhosis and of HCC in children with hepatitis B has been previously reported (7–9,13–15). For example, 4 of 89 children in Italy had cirrhosis on liver biopsies at the beginning of a mean 14.5-year follow-up during which no further incident cases were identified, and 3 of 205 children in Japan were diagnosed with HCC during a median 8.2 years follow-up (8,13). Therefore, vigilance is required in the clinical follow-up of children with chronic HBV who are at risk of HCC even in the absence of cirrhosis (33). Clinicians caring for the children in our cohort chose to check AFP levels at most annual follow-up visits and undertook ultrasound scans in approximately one-third to one-half of the participants at each annual visit. In the absence of published data or guidelines for screening for HCC in children with HBV, we noted considerable variation in practice between centers.

Data describing ALT flares in children with chronic HBV infection are scarce, although these flares are well recognized in clinical practice. In Japan, flares of ALT >500 U/L were recognized in 18 of 149 untreated children and 12 of 43 treated children, during a median follow-up of 8.2 years, which represents a similar rate to that observed in the current study (13). The cumulative incidence of flares at 4 years in our pediatric cohort (4%) was a lower than the 5.7% cumulative incidence at 4 years that was seen in the HBRN adult cohort (34). In our cohort, flares occurred mostly in HBeAg positive children with high viral load who were not receiving anti-HBV treatment and we found no association between timing of flares with puberty. Flares were associated overall with a significant reduction in HBV load and transition from HBeAg positive to negative.

Our study had several limitations. Although involving a large cohort of infected children, the study had limited ability to accurately measure the incidence rates of severe liver disease complications in HBV-infected children, due to the long time-course for their development and their rare incidence during childhood. We were unable to include serial measurement of elastography to provide an additional measure of advancing fibrosis. Annual evaluations of blood tests may have missed events occurring between these evaluations.

In conclusion, we report outcomes observed in a large, prospectively studied cohort of children with chronic HBV infection in North America. Active hepatitis was common throughout follow-up, but treatment only seldom used, suggesting that many children remain at risk for advancing severity of fibrotic liver disease. To enable improved treatment guidelines to be developed, future studies should identify barriers to anti-HBV treatment in childhood and quantify the potential long-term benefits of treatment, such as prevention of progression of fibrosis during childhood, or the development of cirrhosis and HCC arising as major clinical outcomes during young adult life.

Abbreviations

AASLD

American Association for the Study of Liver Diseases

AFP

Alpha-fetoprotein

ALT

Alanine aminotransferase

APRI

Aspartate aminotransferase-to-platelet-ratio-index

AST

Aspartate aminotransferase

CALIPER

Canadian Laboratory Initiative on Paediatric Reference

DNA

Deoxyribonucleic acid

HBeAg

Hepatitis B e antigen

HBRN

Hepatitis B Research Network

HBsAg

Hepatitis B surface antigen

HBV

Hepatitis B Virus

HCC

Hepatocellular carcinoma

IR

Incident rates

IU

International unit

MRI

Magnetic resonance imaging

U/L

Units per liter

ULN

Upper limit of normal