Comparison of immunogenicity and persistence between inactivated hepatitis A vaccine Healive® and Havrix® among children: A 5-year follow-up study

ABSTRACT

Background: Inactivated vaccines for hepatitis A virus (HAV) infection are widely used in China. Mass vaccination programs drive the need for data on long-term persistence of vaccine-induced protection. Methods: A prospective, randomized, open-label clinical trial was conducted to compare geometric mean concentrations (GMCs) and seroconversion rates (SRs) of anti-HAV antibody elicited by the inactivated vaccines Healive and Havrix for 5 y post immunization, in which 400 healthy children were randomly assigned in a 3:1 ratio to receive 2 doses of Healive or Havrix at 0 and 6 month. Anti-HAV antibody concentration was detected by microparticle enzyme immunoassay (MEIA) during the study. Furthermore, an attempt was made to predict persistence of protective immunogenicity by using a suitable statistical model. Results: The GMCs were significantly higher after vaccination with Healive than after Havrix as comparator vaccine at 1, 6, 7, 18, 30, 42, 54 and 66 month (P < 0.01) with the peak point at 7 month (3427.2 mIU/ml for Healive and 1441.9 mIU/ml for Comparator). Similarly significant differences of SRs were found between the 2 groups at 1 and 6 month (P < 0.01). Afterwards, the SRs of both groups reached 100% at 7 month and did not decline until 66 month(99.1% for Healive and 97.5% for Comparator). A linear mixed model with a change point at 18 month(Model 3) was found to be suitable to predict persistence of protective immunogenicity induced by vaccines. It was estimated that the duration of protection for Healive was at least 20 y with a lower limit of GMC 95% confidence interval (CI) no less than 20 mIU/mL. Conclusions: Compared with Havrix, the new preservative-free inactivated hepatitis A vaccine (Healive) in 2 doses showed better persistence of antibody concentrations for 5 y after full-course immunization among children and the persistence of protective immunogenicity was estimated for at least 20 y.

KEYWORDS:

• immunogenicity; inactivated hepatitis A vaccine
• long term follow-up
• persistence of protective immunogenicity

View correction statement:

Erratum

Introduction

Hepatitis A which can cause asymptomatic-to-severe illness is a significant endemic and epidemic disease of global importance.^1-3 The prevalence of hepatitis A infection varies according to hygienic conditions, being very high in countries where the majority of infections are clinically undetected.^4-7 In May, 2013, an outbreak of symptomatic hepatitis A virus (HAV) infections occurred in the USA.⁸ In Japan, 342 cases of HAV infection had been reported in 2014, which was considered as a nationwide HAV outbreak.⁹ In China, hepatitis is also a huge public health problem and infection with HAV remains the leading cause of acute viral hepatitis, especially for children and young adults.¹⁰

China has long experience using inactivated HAV vaccines.^11,12 The review reported by Cui et al.¹² demonstrated that inactivated HAV vaccines manufactured in China were immunogenic, effective, and safe. Liu et al.² conducted a study in the healthy young adults at colleges in Nanchang City, China and found that 1 or 2 doses of inactivated hepatitis A vaccine Healive® (Sinovac Biotech, Beijing, China) induced high rates of seroconversion (anti-HAV IgG concentration ≥20 mIU/mL) persisting for at least 36 months. However, mass vaccination programs drive the need for data on long-term persistence of vaccine-induced protection.¹³

In 2006, a double-blind, randomized and controlled clinical trial was conducted in healthy volunteers aged 1-8 y from Changzhou city, Jiangsu province of China to compare the immunogenicity of anti-HAV antibodies among 3 consecutive production lots of Healive® and Havrix® (GlaxoSmithKline Biologicals) as comparator vaccine. Total 400 subjects were randomized into 4 groups with 100 subjects per group, receiving one of the 3 lots of Healive or the comparator vaccine. The vaccination was 2-dose regimen at 0-6 months. The three lots of Healive showed statistically indistinguishable clinical performance with 100% seroconversion rates (SRs) and 3237-3814 mIU/ml geometric mean concentrations (GMCs) 1 month after the second dose.¹⁴

The aim of the present analysis is to compare the immunogenicity between Healive and Comparator further for 5 y and predict persistence of protective immunogenicity induced by both vaccines with a suitable statistical model.

Results

Participants included in analysis

Of 400 children, 375 participants who completed their full-course immunization were invited to participate in the follow-up phase with 283 in Healive and 92 in Comparator (Fig. 1).

Figure 1. Follow-up of subjects in the 2 groups.

Demographic characteristics of both groups at baseline in the full analysis set (FAS) are presented in Table 1. There were no statistically significant differences in gender ratio, age, height and weight between 2 groups.

Table 1. Demographic characteristics of participants.

			Mean(95%CI)
Group	N	M/F	Age(y)	Height(cm)	Weight(kg)
Healive	283	142/141	3.8(3.6-4.0)	100.7(99.2-102.2)	17.2(16.8-17.7)
Comparator	92	48/44	3.7(3.4-4.0)	100.6(97.9-103.2)	17.4(16.6-18.2)

N, number of subjects; M/F, male/female ratio; 95% CI, 95% Confidence Interval on the mean value.

Observed immunogenicity of Anti-HAV antibodies over time

GMCs and SRs of anti-HAV antibodies in the 2 groups at each timepoint from 1 to 66 month are shown in Table 2. At 1 month after the first dose, GMCs of Healive and Comparator were 29.1 and 20.3 mIU/ml respectively and increased slowly in the following 5 months. At 7 month, i.e. 1 month after dose-2, GMCs of both groups reached the peak point (3427.2 mIU/ml for Healive and 1441.9 mIU/ml for Comparator). Later, there were a sharp decline with 572.0 and 386.0 mIU/ml at 18 month for Healive and Comparator respectively and a relatively slower rate of decline from 30 to 66 month (ranged from 468.7 to 257.1 mIU/ml for Healive and 300.1 to 168.1 mIU/ml for Comparator). The GMCs were significantly higher in Healive than in Comparator at each timepoint from 1 to 66 month (P < 0.01). Similarly significant differences of SRs were found between the 2 groups at 1 month (75.2% for Healive and 50.6% for Comparator) and 6 month (97.5% for Healive and 87.0% for Comparator) (P < 0.01) . Afterwards, the SRs of both groups reached 100% at 7 month and did not decline until 66 month(99.1% for Healive and 97.5% for Comparator).

Table 2. Geometric mean concentrations and seroconversion rates over time in healthy children.

Month	Healive	Comparator	P Value^a
1
Subjects, no	278	91
SR,%(95% CI)	75.2(69.7-80.1)	50.6(39.9-61.2)	<0.0001
GMC,mIU/ml(95% CI)	29.1(26.5-32.0)	20.3(17.2-24.1)	0.0002
6
Subjects, no	282	92
SR,%(95% CI)	97.5(95.0-99.0)	87.0(78.3-93.1)	<0.0001
GMC,mIU/ml(95% CI)	109.6(98.4-122.1)	46.2(38.1-56)	<0.0001
7
Subjects, no	279	92
SR,%(95% CI)	100(98.7-100.0)	100(96.1-100.0)	NS^b^a
GMC,mIU/ml(95% CI)	3427.2(3036.5-3868.1)	1441.9(1191.1-1745.5)	<0.0001
18
Subjects, no	257	87
SR,%(95% CI)	100(98.6-100.0)	100(95.8-100.0)	NS^b^a
GMC,mIU/ml(95% CI)	572.0(519.6-629.6)	386.0(323.7-460.3)	<0.0001
30
Subjects, no	245	81
SR,%(95% CI)	100(98.5-100.0)	100(95.5-100.0)	NS^b^a
GMC,mIU/ml(95% CI)	468.7(423.2-519.0)	300.1(250.2-360.0)	<0.0001
42
Subjects, no	242	83
SR,%(95% CI)	100(98.5-100.0)	100(95.7-100.0)	NS^b^a
GMC,mIU/ml(95% CI)	348.2(306.8-395.1)	234.3(187.2-293.2)	0.0021
54
Subjects, no	228	77
SR,%(95% CI)	100(98.4-100.0)	100(95.3-100.0)	NS^b^a
GMC,mIU/ml(95% CI)	322.3(284.3-365.2)	222.3(179.1-275.9)	0.0035
66
Subjects, no	230	79
SR,%(95% CI)	99.1(96.9-99.9)	97.5(91.2-99.7)	NS^b^a
GMC,mIU/ml(95% CI)	257.1(226.9-291.4)	168.1(135.6-208.4)	0.0008

Note.

^a Pearson or Fisher exact tests were used to compare seroconversion rates, and Student’s t-test was used to compare geometric mean concentrations; ^bnon-significant (P > 0.05).

Model results

278 participants in Healive and 92 participants in Comparator were included in the model research with 5 in Healive excluded for reasons of sample contamination and data missing (see model selection section for details).

The predicted GMCs and SRs from 7 to 306 month after the first dose using Model 3 are shown in Table 3. GMCs (95% CI) were predicted to 33.0 (24.3-45.0) mIU/mL at 246 month and 16.1 (11.0-23.6) mIU/mL at 306 month for Healive. During this period, 260 month was found to be the farthest timepoint with protective immunogenicity, where the GMC (95% CI) was estimated to 27.9 (20.2-38.7) mIU/mL with a lower limit no less than 20 mIU/mL. (Not listed in Table 3, the same below). In a similar way, the farthest timepoint with protective immunogenicity for Comparator was 209 month where the GMC (95% CI) was estimated to 33.2(20.0-55.2) mIU/mL. With regard to SRs, in order to guarantee a lower limit no less than 80%, it was estimated that the SR (95% CI) was 84.9% (80.1%-88.9%) at 157 month for Healive and 89.1% (80.9%-94.7%) at 116 month for Comparator.

Table 3. Observed and Predicted Geometric Mean Concentrations and Seroconversion Rates.

	Healive(N = 278)				Comparator (N = 92)
	GMC, mIU/mL(95%CI)		SR, %(95%CI)		GMC, mIU/mL(95%CI)		SR, %(95%CI)
Month	Observed	Predicted	Observed	Predicted	Observed	Predicted	Observed	Predicted
7	3441.9(3049.1-3885.3)	3441.9(3155.7-3754.1)	100.0(98.7-100.0)	100.0(98.7-100.0)	1441.9(1191.1-1745.5)	1441.9(1239.3-1677.6)	100.0(96.1-100.0)	100.0(96.1-100.0)
18	563.0(512.1-618.9)	563.0(515.4-614.9)	100.0(98.7-100.0)	100.0(98.7-100.0)	391.9(328.7-467.3)	391.9(335.8-457.5)	100.0(96.1-100.0)	100.0(96.1-100.0)
42	354.5(315.0-398.9)	376.6(341.4-415.4)	100.0(98.7-100.0)	100.0(98.7-100.0)	247.8(200.7-306.0)	260.7(218.8-310.6)	100.0(96.1-100.0)	100.0(96.1-100.0)
66	281.6(250.9-316.1)	282.8(252.2-317.2)	99.3(97.5,99.9)	99.3(97.5,99.9)	190.2(154.8-233.8)	193.8(157.1-239.1)	97.8(92.3-99.7)	97.8(92.3-99.7)
126		138.2(116.4-164.2)		91.0(87.0-94.1)		92.4(66.8-128.0)		87.0(78.1,93.1)
186		67.6(53.2-85.8)		73.4(67.8-78.5)		44.1(27.9-69.6)		67.4(56.8,76.8)
246		33.0(24.3-45.0)		57.9(51.9-63.8)		21.0(11.6-38.0)		44.6(34.2,55.3)
306		16.1(11.0-23.6)		45.3(39.3-51.4)		10.0(4.8-20.8)		34.8(25.2,45.4)

Note. The differences between observed and predicted GMCs and SRs could be attributed to the model error.

The predicted SRs using Model 3 for both groups are shown in Figure 2 and the SRs of Healive were higher than Comparator at all the timepoints until 360 month after the second dose.

Figure 2. Predicted Seroconversion Rates over time based on Model 3 for Healive and Comparator.

Discussion

This study was designed to compare GMCs and SRs of anti-HAV antibody elicited by the inactivated vaccines Healive and Havrix as comparator vaccine for 5 y post immunization, and then a suitable statistical model was used to predict the persistence of protective immunogenicity for the 2 vaccines.

The GMCs were significantly higher in Healive than in Comparator at each timepoint from 1 to 66 month (P < 0.01). The SRs of both groups reached 100% at 7 month and did not decline until 66 month. There were no significant differences of SRs between the 2 groups at each timepoint except 1 and 6 month. Since each company has different antigen dissociation technique and dissociation rate, which is a patent technology and not open to the public, the dose unit of U(Healive) and El.U(Havrix) is different and the antigen content could not be compared directly. However, it does not matter to make comparison of the immunogenicity above and protective immunogenicity persistence between the 2 licensed products which is what is most pertinent to the use of the products in the field and population.

Because the advantage of linear mixed models fitting to longitudinal data was recognized by several studies,^2,15-19 4 previously used linear mixed models were applied to fit our follow-up data and Model 3 was confirmed to be a suitable one. Similar to many clinical trials showed,^2,20,21 the GMCs declined rapidly in the first 12 months after 2 doses and slowly later, therefore, the timepoint of 18 month after the first dose was used as the chang point in Model 3. Apart from anti-HAV serostatus at 1 month, we also considered the inclusion of other covariates (age and gender), however, it had no statistically significant differences and did not improve goodness of fit.

A series of previous trials had explored the protective immunogenicity persistence of Havrix based on statistical models. Van Damme et al.²² assumed a constant basic exponential decline and predicted the persistence of protective immunogenicity for at least 20 y among healthy adults with a 0-1-6 month vaccination schedule. Hammitt et al.²³ used a log-linear model of decline in GMC of antibody, and predicted that children would retain protective immunogenicity levels of anti-HAV for more than 22 y with a 0-1-6 month vaccination schedule or for more than 27 y with a 0-1-12 month vaccination schedule. Based on 2 of the longest follow-up studies (participants were vaccinated according to a 0-12 mo and a 0-6 mo schedule, respectively) in adults,^24,25 Niel et al.¹⁶ reported a duration of protective immunogenicity at least 25 y with a linear mixed model, which was confirmed appropriate to describe the profile of HAV antibody decline phase and it was also applied in our persistence estimation.

In our study, the persistence of protective immunogenicity for Havrix was less than 20 y which was shorter compared with the studies mentioned above. ELISA method (Enzyme-Linked Immunosorbent Assay) to measure anti-HAV antibody concentration was adopted and anti-HAV antibody concentration no lower than 20 mIU/mL was considered the indicator of seroprotection for all the 4 studies including ours. Considering there are wide statistical bounds around the data in the published and current study, all results could lie within the range of variation of following such cohorts and modeling long-term changes based on a few years of data.

There were some limitations to our study as follows. The period of observations was limited to describe the whole profile of anti-HAV antibody elicited by the vaccines and a longer follow-up study need to be conducted to further adjust and verify the present results. In addition, results from single site might have resulted in selection bias.

In conclusion, compared with Havrix, the new preservative-free inactivated hepatitis A vaccine Healive with 0-6 month vaccination schedule showed better immunogenic for 5 y after full-course immunization among children and the persistence of protective immunogenicity was estimated for at least 20 y.

Materials and methods

Clinical trial methodology

The three consecutive production lots of Healive were combined as one group with the consistency of immunogenicity according to regulatory acceptance criteria¹⁴ taken into consideration. Therefore, the 400 healthy children were assigned in a 3:1 ratio to receive 2 doses of Healive (0.5 mL/dose) contained 250 U antigen and 0.25 mg alum without preservative or the comparator vaccine Havrix (0.5 mL/dose) contained 720 ELISA units (El.U) antigen and 0.25 mg alum with 2-phenoxyethanol as preservative.Volunteers who completed their full-course immunization were monitored annually for 5 y. Blood samples were collected at 0, 1, 6, 7, 18, 30, 42, 54 and 66 month with screening blood samples collected before immunization at 0 and 6 month. Anti-HAV antibody concentrations were assessed by microparticle enzyme immunoassay (MEIA) during the study. Both seroconversion and protective immunogenicity were defined as the ability of hepatitis A vaccines to elicit an immune response of at least 20 mIU/mL.^12,26

The study was conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practice guidelines. Study protocols were approved by the Ethics Review Committee of the Changzhou Center for Disease Control and Prevention. Written informed consents were obtained from all participants prior to the performance of the study. The trial was registered at Clinicaltrials.gov (NCT00534885).

Statistical methods

GMCs and SRs of anti-HAV antibody were calculated based on the actual observed data without missing data imputation. Logarithmic transformation (log10) had been made before the calculation of GMC and 95% CI. Student’s t-test or Mann–Whitney U test (for non-normally distributed data) was used to compare GMCs when relevant, and Chi-square test or Fisher’s exact test (when data were sparse) was used to compare SRs. The significance level was 0.05 (2-sided). All data were handled and analyzed using SAS version 9.3 (SAS Institute, Cary, NC).

Predicative models for persistence of protective Immunogenicity

In order to identify a suitable model for our data set, 4 linear mixed model structures used previously ^{15-17,20,27,28} to analyze antibody persistence following vaccination were investigated.

Model 1 is most widely used ^{15,17,20,27,28} to predict the persistence of protective immunogenicity induced by HAV. It is a linear mixed effects model involving linear antibody decay containing fixed and random effects for both slope and intercept parameters:$Y_{ij}=(a+a_i)+(b+b_i)t_j+{\epsilon }_{ij}$

Where Y_ij is log10 antibody concentration for subject i observed at time t_j, a and a_i are the population-level (fixed effect) and individual level (random effect) intercepts and b and b_i are the population-level and individual-level slope corresponding to the rate of linear antibody decay. ε_ij is the residual error between model prediction and the observed value. The model was fitted to the data from 7 month (timepoint of antibody concentration peak) to 66 month, that is, the model was only fitted to the antibody concentrations decline phase.

Model 2¹⁷ is a segmented linear mixed effects model which contains fixed and random effects for all slope and intercept parameters except for the indicator δ_ij (contains only fixed effects). This model was fitted to antibody concentrations from 1 to 66 month:${\displaystyle Y_{ij}=(a+a_i)+(b+b_i)t_j{\delta }_{ij}+c\times {\delta }_{ij}+(d+d_i)t_j(1-{\delta }_{ij})+{\epsilon }_{ij}}$

Where δ_ij = 1 for time up to the antibody concentrations peak and δ_ij = 0 for time post the peak.

Model 3^16,29 is a linear trend model with one change point during the antibody concentrations decline phase:${\displaystyle Y_{ij}=(a+a_i)+(b+b_i)t_j+c\times {\delta }_j+d\times t_j\times {\delta }_j+{\epsilon }_{ij}}$

Where δ_j = 1 for time post the change point and δ_j = 0 for time from the antibody concentrations peak up to the change point.

Model 4¹⁶ is a simple model including only different exponent parameters of time:${\displaystyle Y_{ij}=(a+a_i)+(b+b_i)t_j{}^{-0.5}+(f+f_i)t_j{}^{-1.5}+{\epsilon }_{ij}}$

Same as Model 1 and 3, Model 4 are only fitted to data during the antibody concentration decline phase.

Models were constructed and fitted using SAS 9.3 MIXED procedure with unconstrained variance–covariance matrices. AIC ³⁰ and BIC ³¹ were calculated to choose the suitable model to predict persistence of protective immunogenicity for Healive and Comparator, where smaller AIC and BIC values indicate better model fitting.

Model selection

Because of one sample contamination at 42 month, the observation (4C283) in Healive was excluded in the data set to build statistical models. In addition, 4 observations (4C397, 4C140, 4C115, 4C164) in Healive whose antibody concentrations were missing at 7 month were also excluded. Missing values were imputed with the last consecutive measurements for data from 1 to 7 month and the next consecutive measurements for data from 7 to 66 month, considering that the GMCs of both groups reached the highest level at 7 month.

Parameter estimates and fit statistics for each modeling approach are shown in Table 4. Serostatus at 1 month was retained in each model with statistically significant coefficient (P < 0.01). According to the result showed in Table 2, a much smaller average concentration decline rate after 18 month compared to the decline rate from 7 to 18 month were found. Therefore, the use of Model 1 that assumed the constant basic logarithmic decline rate would have resulted in a shorter duration of protection. Similar to Model 1, Model 2 with concentration increasing phase added also assumed the constant basic logarithmic decline rate from 7 month onward. Both Models above had considerably higher AIC or BIC than other 2 models. Considering the different antibody concentration decline rates, Model 3 with one change point (18 month) during the antibody concentrations decline performed relative smaller AIC and BIC. Compared to Model 3, Model 4 had an improvement in AIC and BIC, however, the positive coefficients of $t_j{}^{-1.5}$ implied an infinite duration of protection, as antibody concentrations were ultimately predicted to increase over time and it was unreasonable.

Table 4. Parameter estimates and fit statistics for each modeling approach.

		Healive				Comparator
Model	Parameters	Mean Parameters	95%CI	AIC	BIC	Mean Parameters	95%CI	AIC	BIC
1	Intercept(a+ai)	3.020	(2.943,3.097)	1379.56	1394.08	2.821	(2.729,2.914)	260.01	270.09
	Slope(b+bi)	−0.015	(−0.016,−0.014)			−0.012	(−0.014,−0.011)
	serostatus at 1 mo	0.344	(0.257,0.430)			0.335	(0.213,0.458)
2	Intercept(a+ai)	2.552	(2.471,2.632)	2784.88	2799.39	2.503	(2.386,2.62)	936.61	946.69
	Slope(b+bi)	−1.747	(−1.829,−1.666)			−1.714	(−1.857,−1.572)
	Slope(c)	0.269	(0.259,0.280)			0.231	(0.212,0.249)
	Slope(d+di)	−0.006	(−0.007,−0.005)			−0.006	(−0.008,−0.004)
	serostatus at 1 mo	0.380	(0.307,0.453)			0.355	(0.246,0.465)
3	Intercept(a+ai)	3.777	(3.694,3.860)	77.69	92.20	3.348	(3.244,3.452)	−54.24	−44.14
	Slope(b+bi)	−0.071	(−0.074,−0.069)			−0.051	(−0.056,−0.047)
	Slope(c)	−1.244	(−1.295,−1.192)			−0.878	(−0.957,−0.799)
	Slope(d)	0.066	(0.064,0.069)			0.046	(0.042,0.050)
	serostatus at 1 mo	0.344	(0.257,0.430)			0.335	(0.213,0.458)
4	Intercept(a+ai)	2.065	(1.949,2.181)	−208.38	−182.99	1.868	(1.697,2.039)	−191.08	−173.43
	Slope(b+bi)	1.198	(0.686,1.711)			2.059	(1.232,2.886)
	Slope(f+fi)	14.291	(11.938,16.645)			6.068	(2.417,9.719)
	serostatus at 1 mo	0.324	(0.239,0.408)			0.357	(0.237,0.478)

Note. all parameters P values are less than 0.01.

On the basis of modeling results, Model 3 was chosen as a suitable model. When observed and predicted values were compared using Model 3, R² ranged from 79.04% to 98.37% for Healive and 76.60% to 96.70% for Comparator at each timepoint from 7 to 66 month (data not shown). These values reflected a high level of agreement between the observed and predicted GMCs over time.

Disclosure of potential conflicts of interest

We declare that we have no conflict of interest exist in the submission of this manuscript, and manuscript is approved by all authors for publication. I would like to declare on behalf of my co-authors that the work described was original research that has not been published previously, and not under consideration for publication elsewhere, in whole or in part. All the authors listed have approved the manuscript that is enclosed.

Funding

This work was partially supported by Research Grants 81273176, 81302509 and 81473069 from the National Natural Science Foundation of China.