Long-term Durability of Immune Responses After Hepatitis A Vaccination Among HIV-Infected Adults

Abstract

Background.  Vaccination provides long-term immunity to hepatitis A virus (HAV) among the general population, but there are no such data regarding vaccine durability among human immunodeficiency virus (HIV)–infected adults.

Methods.  We retrospectively studied HIV-infected adults who had received 2 doses of HAV vaccine. We analyzed blood specimens taken at 1 year, 3 years, and, when available, 6–10 years postvaccination. HAV immunoglobulin G (IgG) values of ≥10 mIU/mL were considered seropositive.

Results.  We evaluated specimens from 130 HIV-infected adults with a median age of 35 years and a median CD4 cell count of 461 cells/mm³ at or before time of vaccination. Of these, 49% had an HIV RNA load <1000 copies/mL. Initial vaccine responses were achieved in 89% of HIV-infected adults (95% confidence interval [CI], 83%–94%), compared with 100% (95% CI, 99%–100%) of historical HIV-uninfected adults. Among initial HIV-infected responders with available specimens, 90% (104 of 116; 95% CI, 83%–95%) remained seropositive at 3 years and 85% (63 of 74; 95% CI, 75%–92%) at 6–10 years. Geometric mean concentrations (GMCs) among HIV-infected adults were 154, 111, and 64 mIU/mL at 1, 3, and 6–10 years, respectively, compared with 1734, 687, and 684 mIU/mL among HIV-uninfected persons. Higher GMCs over time among HIV-infected adults were associated with lower log₁₀ HIV RNA levels (β = −.12, P = .04).

Conclusions.  Most adults with well-controlled HIV infections had durable seropositive responses up to 6–10 years after HAV vaccination. Suppressed HIV RNA levels are associated with durable HAV responses.

Hepatitis A virus (HAV) infections remain a significant public health problem worldwide. Human immunodeficiency virus (HIV)–infected persons have an elevated risk of HAV coinfection; previous studies have shown that 20%–70% of HAV-positive patients have evidence of prior HIV infection [1–3]. Given the elevated prevalence of underlying liver disease among HIV-infected persons [4, 5], superimposed acute HAV infection may contribute to liver-related morbidity and mortality in this population [1, 6–8]. Furthermore, HIV-infected persons may experience prolonged HAV viremia, which has important public health implications for transmission within the community [9]. Previous investigators have recommended preventive strategies, including the administration of the HAV vaccine [10–12].

In the United States, 2 inactivated HAV vaccines have been licensed since 1995–1996 [13]. Although HAV vaccination generates robust initial antibody levels and provides long-term immunity in the general population [13–16], studies have shown that HIV-infected persons may generate less-robust antibody responses after a 2-dose series [17, 18]; hence, protection may wane more rapidly over time in this group. Despite numerous studies examining the initial immune responsiveness to HAV vaccination among HIV-infected persons [3, 17, 19–26], there are currently no data on the long-term durability of responses in this population.

METHODS

We retrospectively studied participants of the United States Military HIV Natural History Study (NHS) to evaluate the durability of antibody responses after 2 doses of the HAV vaccine and to determine the factors associated with seroresponse. The NHS is a prospective study of HIV-1–infected United States military beneficiaries. Participants are consenting military personnel including members of the Army, Navy/Marines, Air Force, and their dependents. Since 1985, new active duty members have undergone routine HIV testing, with repeat testing every 1–5 years. Those who consent to enroll in the NHS are evaluated every 6 months by an HIV specialist as part of the study, in addition to receiving routine clinical care. Data collected include demographics, markers of HIV disease progression, antiretroviral medication use, receipt of vaccinations, and clinical events [27]. Additionally, participants provide blood specimens during study visits; samples are frozen and maintained at a central repository under strict conditions including 24-hour continuous temperature monitoring. The study was approved by the governing institutional review board, and the participants provided written informed consent.

For this subset analysis, additional inclusion criteria included age ≥18 years; negative HAV immunoglobulin G (IgG) prior to vaccination; documented receipt of 2 HAV vaccinations administered 6–18 months apart after HIV diagnosis; and available blood specimens at 12 (± 6) months and ≥3 years after receipt of the second HAV vaccine to evaluate the initial and long-term antibody responses, respectively. We analyzed all participants in the NHS fulfilling these criteria to limit selection bias. Each participant had received his or her first hepatitis A vaccine between May 1996 and November 2003, and had relevant postvaccination data available through 2007. Study participants who received <2 or >2 HAV vaccine doses were excluded from this analysis. The study database did not contain information on the specific brand or manufacturer of the administered HAV vaccine; all participants received either VAQTA (Merck & Co.) or HAVRIX (GlaxoSmithKline) based on available supplies at each HIV NHS clinical site. Prior studies have demonstrated similar seroresponses between these 2 vaccines [28, 29], and the Advisory Committee on Immunization Practices (ACIP) considers these 2 brands of HAV vaccine interchangeable [11]. We did not evaluate recipients of TWINRIX hepatitis A and hepatitis B recombinant vaccine (GlaxoSmithKline) in this study due to its alternate dosing schedule.

At or during the visit proximal to the first dose of HAV vaccine, NHS investigators collected clinical data including demographics (age, gender, and self-reported ethnicity); body mass index (BMI); presence of concurrent medical conditions including diabetes mellitus, chronic hepatitis B (defined as a positive surface antigen), and chronic hepatitis C (defined as a positive antibody test); current and nadir CD4 cell counts; plasma HIV RNA level; history of AIDS-defining conditions (except an isolated CD4 count <200 cells/mm³) [30]; and receipt of antiretroviral therapy. Additional data included time between the administrations of the 2 doses of the HAV vaccine, as well as time-updated (every 6 mo) values for CD4 cell counts, plasma HIV RNA levels, and antiretroviral medication use from the time of the first dose of HAV vaccine until the date of the last available repository specimen for HAV antibody testing.

We analyzed stored serum samples (0.5 mL), collected as part of the NHS at 12 (± 6) months and 3 years (±9 mo) postvaccination, for HAV IgG level. Additional samples taken at the 12 (± 6) month time point were also analyzed, when available, to provide further data on the “initial” immune response, which was a priori defined to use all samples from 12 (± 6) months; analyses were adjusted for the use of multiple samples from the same individual. Finally, the HAV IgG level was determined at 6–10 years after the second HAV vaccine dose when this sample was available. Participants with and without 6-to-10-year data were similar, except that those without 6-to-10-year data were more likely to have been vaccinated more recently. For both the 3-year and 6-to-10-year specimens, the latest available time points during each of these intervals were used to evaluate seroresponse durability.

We compared historical data from HIV-uninfected adults (ages 18–41 y, median age 29 y) who had received 2 doses of the HAV vaccine (VAQTA) 6 months apart as part of a large clinical trial [31] with data from our HIV-infected cohort. Laboratory testing procedures were similar for both study groups; historical data used the modified HAVAB assay that has been shown to be comparable to the assay used in this study.

HAV IgG antibody levels were quantified by the Quantitative HAVK PLUS enzyme immunoassay at Merck Research Laboratories. The assay used a standard curve that was constructed with dilutions of a World Health Organization standard serum immunoglobulin. Levels were expressed as milli–international units per milliliter (mIU/mL). Samples with levels ≥10 mIU/mL were considered seropositive for HAV antibody similarly to prior studies [15, 19, 32]. Laboratory personnel were blinded to all clinical information, including the dates the samples were obtained.

Statistical analyses included descriptive statistics shown as numbers (percentages) for categorical variables, and medians (IQRs, interquartile ranges) for continuous variables. Data were analyzed in terms of both a “seropositive response” (ie, HAV IgG antibody levels ≥10 mIU/mL) and the geometric mean concentration (GMC) with 95% confidence intervals (CIs). Comparisons between our HIV-infected participants and HIV-uninfected historical controls [31] used the Satterthwaite approximation t test assuming unequal variances. Logistic regression models were used to evaluate factors associated with an initial seropositive response defined at 12 (± 6) months postvaccination. Univariate and multivariate analyses involved linear regression models used to evaluate factors associated with the initial GMC response. Multivariate analysis of this single outcome began with models including variables with a P value <.15 in the univariate models and a backward stepwise process was performed to derive the final model. Factors associated with the durability of GMC responses longitudinally were determined using mixed models for log-transformed antibody concentrations changing in a piecewise linear fashion over time postvaccination, using all available assay results. Multiple-risk-factor models initially included variables with a P value <.20 in the single-risk-factor models and a backward stepwise process was performed to derive the final model, maintaining piecewise linear terms for time after HAV vaccination throughout. All analyses were conducted using STATA version 10 and R version 2 (STATA Corporation; R Foundation for Statistical Computing, www.r-project.org).

RESULTS

Study Population Characteristics

The study cohort consisted of 130 HIV-infected adults with a median age of 35 years (IQR, 30–41); 96% were male; and ethnicity was Caucasian for 51%, African American for 42%, and other for 7% (Table 1). The median body mass index (BMI) was 25.3 kg/m², and 12% were classified as obese (>30 kg/m²). Chronic hepatitis B and C were present among 9% and 2% of participants, respectively. At first dose of HAV vaccination, the median CD4 count was 461 cells/mm³ (IQR, 322–617 cells/mm³). Of the total participants, 49% had a plasma HIV RNA level <1000 copies/mL. Of the 62% who were receiving highly active antiretroviral therapy (HAART), 63% had an HIV RNA level <1000 copies/mL. The median CD4 nadir before vaccination was 309 cells/mm³ (IQR 192–431 cells/mm³) and 14% had a prior AIDS-defining condition.

Table 1.

Study Population Characteristics at Time of Hepatitis A Virus Vaccination

Characteristica	All participantsbN = 130	Participants with initial seropositive responsecN = 116	Participants without initial seropositive response N = 14
Demographics
Age, y	35 (30−41)	34 (30–40)	37 (33–41)
Gender, male	125 (96%)	111 (96%)	14 (100%)
Ethnicity
Caucasian	66 (51%)	59 (51%)	7 (50%)
African American	54 (42%)	48 (41%)	6 (43%)
Other	10 (7%)	9 (8%)	1 (7%)
Clinical factors
Hepatitis B and C history
Hepatitis C Virus antibody positive	2 (2%)	1 (1%)	1 (7%)
Hepatitis B surface antigen positive	11 (9%)	8 (7%)	3 (21%)
Diabetes	8 (6%)	7 (6.%)	1 (7.1%)
Body mass index, kg/m2	25.3 (23.7–27.8) [n = 116]2	25.3 (23.7–28.2) [n = 102]2	25.3 (23.9–26.2) [n = 14]2
Obese (BMI >30 kg/m2)	14 (12%) [n = 116]2	12 (12%) [n = 102]2	2 (14%) [n = 14]2
HAV vaccine factors
Interval between 2 doses of vaccine, d	214 (183–306)	214 (183–306)	244 (183–297)
HIV-specific factors
Currentd CD4 cell count, cells/mm3	461 (322−617)	467 (344–630)	322 (225–491)
Current CD4 cell count < 200 cells/mm3	10 (8%)	6 (5%)	4 (29%)
Current CD4 cell count ≥ 200 to < 350 cells/mm3	30 (23%)	25 (22%)	5 (36%)
Current CD4 cell count ≥ 350 cells/mm3	90 (69%)	85 (73%)	5 (36%)
Plasma HIV RNA leveld < 1000 copies/mL	63 (49%) [n = 129]2	58 (50%) [n = 116]2	5 (38%) [n = 13]2
Currentd receipt of HAART, yes	81 (62%)	72 (62%)	9 (64%)
Nadir CD4 count, cells/mm3	309 (192−431)	312 (197–435)	227 (174–416)
CD4 nadir < 200 cells/mm3	33 (25%)	29 (25%)	4 (29%)
History of AIDS-defining condition, yes	18 (14%)	16 (14%)	2 (14%)

NOTE. BMI, body mass index; HAV, hepatitis A virus; HIV, human immunodeficiency virus; HAART, highly active antiretroviral therapy.

^aCategorical variables are shown as numbers (percentages) and continuous values as medians (interquartile ranges).

^bAll data represent 130 total participants, except where otherwise noted due to missing data: 14 participants were missing BMI data due to lack of recorded weights within the database, and 1 participant was missing an HIV RNA level at time of vaccination.

^cUsed all results within 12 ± 6 mo postvaccination.

^dCurrent was defined as the closest value at or prior to the first HAV vaccine administration.

Antibody Responses to HAV Vaccination

Initial positive vaccine seroresponses at 12 (± 6) months postvaccination were achieved in 89% (95% CI, 83%–94%) of HIV-infected persons. Among initial responders (n = 116), 90% (95% CI, 83%–95%) maintained an HAV IgG level ≥10 mIU/mL at 3 years postvaccination. Finally, among participants with an initial protective antibody response who also had available specimens (n = 74), 85% (95% CI, 75%–92%) continued to have a protective level at 6–10 years (median time 8.2 y; IQR, 7.1–8.8 y) postvaccination.

We examined the seropositive responses at each of the 3 time points (1, 3, and 6–10 y postvaccination) stratified by the CD4 count at the first dose of HAV vaccination (Table 2). Participants who had a CD4 count ≥350 cells/mm³ at the time of HAV vaccine receipt had a significantly higher 1-year postvaccination seropositivity rate than those with a CD4 count <350 cells/mm³ at HAV vaccination (94% vs 78%, P = .006). The seropositivity rate at 3 years postvaccination was also higher among those with more robust CD4 counts at initial vaccination (95% vs 87%), although this did not reach statistical significance (P = .09). There were no significant differences between the seropositivity rates at 6–10 years by CD4 strata, although the effective sample size was small for this time point. Likewise, the seropositive responses by plasma HIV RNA level strata (<1000 vs ≥1000 copies/mL) at the time of HAV vaccination had trends of higher rates of seropositivity among those with lower levels of viremia (Table 3; P values = .09, .11, and .12 for measures from initial, 3 y, and 6–10 y, respectively).

Table 2.

Seropositive Response Rates^a After HAV Vaccination Among HIV-Infected Persons, Stratified by CD4 Count at or Proximal to Vaccination

	CD4 count < 350 cells/mm3		CD4 count ≥ 350 cells/mm3
Time	Number seropositive/number evaluated	Seropositive rate (95% CI)b	Number seropositive/number evaluated	Seropositive rate (95% CI)b
Initial responsec	31/40	78% (62–89%)	85/90	94% (88–98%)
3 y	27/31	87% (70–96%)	75/79	95% (88–99%)
6–10 y	17/20	85% (62–97%)	46/54	85% (73–93%)

NOTE. HAV, hepatitis A virus; HIV, human immunodeficiency virus; 95% CI, 95% confidence interval.

^aData for 3 y and 6–10 y responses includes only those participants with a seropositive response at the initial time point.

^bSeropositive rate: Exact binomial 95% confidence interval calculated using Clopper–Pearson method.

^cUsed all results within 12 ± 6 mo postvaccination.

Table 3.

Seropositive Response Rates^a After HAV Vaccination Among HIV-Infected Persons, Stratified by Plasma HIV RNA Level at or Proximal to Vaccination^b

	HIV RNA <1000 copies/mL		HIV RNA ≥1000 copies/mL
Time	Number seropositive/number evaluated	Seropositive rate (95% CI)c	Number seropositive/number evaluated	Seropositive rate (95% CI)c
Initial responsed	75/80	94% (86–98%)	41/50	82% (69–91%)
3 y	68/71	96% (88–99%)	34/39	87% (73–96%)
6–10 y	41/46	89% (76–96%)	22/28	79% (59–92%)

NOTE. HAV, hepatitis A virus; HIV, human immunodeficiency virus; 95% CI, 95% confidence interval.

^aData for 3 y and 6–10 y responses includes only those participants with a seropositive response at the initial time point.

^bAnalysis restricted to those participants with an available plasma HIV RNA level.

^cSeropositive rate: Exact binomial 95% confidence interval calculated using Clopper–Pearson method.

^dUsed all results within 12 ± 6 mo postvaccination.

GMCs were evaluated at the 3 study time points, and were 154, 111, and 64 mIU/mL at 1, 3, and 6–10 years, respectively, among participants who had an initial seropositive response (Table 4). More robust GMCs were seen among those with CD4 counts ≥350 cells/mm³ and among those with low plasma HIV RNA levels (<1000 copies/mL) at the time of HAV vaccination during each of the 3 study time points (Table 4 and Figure 1).

Table 4.

Geometric Mean Concentrations^a of Antibody Responses After Hepatitis A Virus Vaccination in HIV-Infected Persons With Initial Seropositive Response,^b and Stratified by CD4 Cell Count and Plasma HIV RNA Level

		Geometric mean concentration,a mIU/mL (95% confidence interval)
		Stratified by prevaccination CD4 cell count (cells/mm3)		Stratified by prevaccination plasma HIV RNA level (copies/mL)
Time point	Total cohort	<350	≥ 350	<1000	≥1000
Initialc	154 (112–212) [n = 130]	87 (47−160) [n = 40]	199 (139−285) [n = 90]	196 (132−293) [n = 80]	104 (63−174) [n = 50]
∼3 y	111 (4−2999) [n = 116]	91 (3−3154) [n = 31]	121 (5−3131) [n = 79]	127 (5−3515) [n = 71]	88 (3−2481) [n = 39]
∼6–10 y	64 (2−2066) [n = 74]	50 (1−1755) [n = 20]	70 (2−2422) [n = 54]	79 (2−2541) [n = 46]	46 (1−1662) [n = 28]

NOTE. HIV, human immunodeficiency virus.

^aGeometric mean concentrations measured in mIU/mL.

^bInitial seropositive response: Data for 3 y and 6–10 y responses includes only those with a seropositive response at the initial time point. Some data for CD4 levels and HIV-RNA values were missing.

^cUsing all results within 12 ± 6 mo postvaccination. Initial response included 130 participants with a total of 230 values.

Figure 1.

Post–HAV vaccination geometric mean concentrations of antibody responses in HIV-infected persons stratified by CD4 cell count and plasma HIV RNA level at time of vaccination [42, 43].

Antibody Levels Among HIV-Infected Persons Compared With HIV-Uninfected Historical Controls

The initial proportion of seroresponders at 12 (± 6) months postvaccination was lower among HIV-infected persons in our study (89%) than among historical HIV-uninfected persons, in whom 100% (95% CI, 99%–100%) generated a seroprotective response [31]. In addition, the GMCs among HIV-uninfected persons were lower than those achieved at each of the 3 time points among HIV-uninfected persons (1734 mIU/mL [95% CI 1517–1981 mIU/mL], 687 mIU/mL [95% CI, 588–802 mIU/mL], and 684 mIU/mL [564–831 mIU/mL] for the 3 time points, respectively), although they were not statistically significantly different (P > .05), perhaps due to the wide range of GMC responses among our smaller-sized sample of HIV-infected participants.

Factors Associated With Antibody Responses After HAV Vaccination Among HIV-Infected Persons

We evaluated factors associated with achieving an initial seropositive (≥10 mIU/mL) response at 12 (± 6) months postvaccination but found no demographic or HIV-related factor to be significantly associated. Factors associated with trends for an initially higher log₁₀ GMC (at the 12 mo time point) in the univariate models included a younger age (β = −.04, P = .08), absence of chronic hepatitis B (β = −1.17, P = .05), higher CD4 cell count at vaccination (β = .11 per 100 cells/mm³, P = .07), and lower log₁₀ plasma HIV RNA level (β = −.25, P = .13) prior to HAV vaccination (Table 5). In the final multivariate model, younger age (β = −.05, P = .02) was associated with a higher initial GMC, with a trend for lower log₁₀ plasma HIV RNA levels (β = −.33, P = .09) with better response.

Table 5.

Factors Associated With Initial^a Geometric Mean Concentrations Generated After Hepatitis A Virus Vaccination Among HIV-Infected Persons

Univariate analysis: linear regression models	Regression coefficient (standard error)	P valueb
Age, y	−.04 (.02)	.08
Gender, male	1.1 (.86)	.19
Ethnicity: Caucasian, African American, Other [referent group]	.24 (.65) .25 (.66) 1.0 (-)	.75
Hepatitis C virus antibody	−1.22 (1.35)	.37
Hepatitis B surface antigen	−1.17 (.59)	.05
History of diabetes	−.96 (.69)	.17
Body mass index, kg/m2	−.03 (.06)	.60
Obese (BMI >30 kg/m2)	−.28 (.55)	.61
Interval between 2 HAV vaccine administrations, d	−.002 (.002)	.18
Currentc CD4 count, per 100 cells/mm3	.11 (.06)	.07
Currentc log10 plasma HIV RNA level, copies/mL	−.25 (.17)	.13
Currentc receipt of HAART	.13 (.34)	.71
Nadir prevaccination CD4 cell count, per 100 cells/mm3	.04 (.09)	.61
History of AIDS-defining condition	−.096 (.48)	.84
Finald multivariate analysis: linear regression model with final predictors	Regression coefficient (standard error)	P value
Age, y	−.05 (.02)	.02
Currentc log10 plasma HIV RNA level, copies/mL	−.33 (.19)	.09

NOTE.  HIV, human immunodeficiency virus; BMI, body mass index; HAV, hepatitis A virus; HAART, highly active antiretroviral therapy.

^aUsing all antibody concentration results within 12 ± 6 mo of vaccination.

^bP value: Wald-type test of null regression coefficient.

^cAt time of HAV vaccination was defined as the closest value prior to the first HAV vaccine administration.

^dThe final model was derived using a backward stepwise modeling approach.

We also explored time-updated factors associated with more robust GMC responses over time in a longitudinal analysis and found, in the multivariate model, lower log₁₀ plasma HIV RNA levels (β = −.12, P = .04) were longitudinally associated with maintaining higher GMCs (Table 6). In addition, there were trends for both higher CD4 counts as a time-varying covariate (β = .04 per 100 cells/mm³; P = .07) and absence of chronic hepatitis B at time of vaccination (β = −.98, P = .07), each being associated with higher GMCs over time. In addition, increasing number of years since HAV vaccination was associated with decreasing GMCs over time in the piecewise linear model assuming a changepoint at ∼16 months postvaccination (β = −.97 for 0–16 mo postvaccination and −.08 thereafter, P < .001). We also examined HIV RNA level as a categorical variable, and found that an undetectable viral load (<400 copies/mL) was associated with higher GMCs over time in the univariate (β = .23, P = .02) and multivariate (β = .20, P = .046) models.

Table 6.

Factors Associated With Longitudinal Changes in Geometric Mean Concentrations Over Time After Hepatitis A Virus Vaccination Among HIV-Infected Persons

Univariate longitudinal analysis via mixed models: Prevaccination factors, adjusted for piecewise linear postvaccination yearsa	Regression coefficient (standard error)	P valueb
Age, y	−.02 (.02)	.32
Gender, male	1.50 (.79)	.06
Ethnicity: Caucasian vs other, African American vs other	.08 (.59) .27 (.60)	.80
Hepatitis C virus antibody	−.88 (1.22)	.47
Hepatitis B surface antigen	−1.05 (.54)	.05
Body mass index, kg/m2	−.03 (.05)	.56
Obese (BMI >30 kg/m2)	−.50 (.50)	.32
Interval between 2 HAV vaccinations, d	−.003 (.002)	.11
CD4 cell count, per 100 cells/mm3	.08 (.06)	.19
Log10 plasma HIV RNA level, copies/mL	−.22 (.15)	.15
Receipt of HAART	−.17 (.31)	.59
Nadir prevaccination CD4 cell count per 100 cells/mm3	.03 (.08)	.75
History of AIDS-defining condition	.047 (.44)	.91
Finalc multivariate longitudinal analysis via a mixed model: time-updated and time-invariant predictors	Regression coefficient (standard error)	P value
Years post-HAV vaccination	−.97 (.12) 0–16 mo PV −.08 (.017) 17–120 mo PV	<.001
Log10 plasma HIV RNA level, copies/mL	−.12 (.06)	.04
CD4 cell count, per 100 cells/mm3	.04 (.02)	.07
Hepatitis B surface antigen positive prevaccination	−.98 (.53)	.07

NOTE. HIV, human immunodeficiency virus; BMI, body mass index; HAV, hepatitis A virus; HAART, highly active antiretroviral therapy.PV, postvaccination.

^aIncludes terms for y postvaccination (piecewise linear with change at ∼16 mo).

^bP value: Wald-type test of null regression coefficient.

^cThe final model was derived using a backward stepwise modeling approach.

DISCUSSION

Our study demonstrates that HIV-infected adults achieve high initial seroconversion rates (∼90%) after hepatitis A vaccination. Prior studies have demonstrated variable seroresponse rates from 40%–94%, depending on the study population's immune status and rates of antiretroviral coverage [3, 17–26]. Of note, our study was conducted in an early-diagnosed HIV cohort with access to antiretroviral therapy.

Our study is unique in that we evaluated the long-term durability of HAV vaccination among initial responders. Studies have previously demonstrated long-term (eg, ≥10 y) immunogenicity and efficacy, with projections for protection of >20 years, among the general population [13, 33–35]. Although administration of the HAV vaccine among HIV-infected persons is recommended [10–12], long-term immunogenicity data among HIV-infected persons have been absent, with most studies reporting only 28–56 week follow-up data [19]. Our study provides useful clinical data demonstrating that most HIV-infected adults who initially responded to HAV vaccination have good durability of responses for 6–10 years. However, some HIV patients failed to generate or maintain a response to the HAV vaccine, suggesting that improvements in vaccine immunogenicity or administration strategies would be valuable.

Though HIV-infected persons achieved high seropositive response rates, this group generated lower GMCs than historical HIV-uninfected persons. Of note, our study was not designed to directly compare HIV-infected and HIV-uninfected persons concurrently; rather, we used historical controls. Although the groups were similar in several regards, our HIV-infected cohort was slightly older than the historical cohort (median age, 35 y vs 29 y) which may have affected seroresponse rates. Nonetheless, given the large differences in GMCs between these 2 groups, it is likely that HIV-infected persons generate lower GMC responses postvaccination compared with HIV-uninfected persons, similar to the differences seen with other vaccine responses [36].

Given the lower initial antibody levels and the increasing life expectancy of HIV-infected persons, postvaccination booster doses may be necessary to maintain anti-HAV levels ≥10 mIU/mL over time (ie, >10 y). Some studies have investigated the use of a third dose of HAV vaccine to achieve higher initial postvaccination GMCs among HIV patients [37]; whether this strategy would result in more durable antibody responses is unknown. Our data support further investigations among HIV-infected persons evaluating antibody levels beyond 10 years and determining whether loss of detectable antibody levels results in loss of protection against natural infection. Of note, among immunocompetent hosts, memory responses to HAV may exist even in the absence of detectable antibody [38], but data among HIV patients are lacking. In addition, our data emphasize the importance of identifying predictors for durable seroresponses to identify HIV patients most at risk for reduced antibody levels over time.

In our study, factors associated with higher GMC over time included lower levels of HIV viremia. Other studies among HIV-infected persons have also found that plasma HIV RNA levels of <1000 copies/mL were associated with better initial seroresponses [3, 24]; our study adds to the current literature by demonstrating that time-updated plasma HIV RNA levels also are important in maintaining postvaccination antibody levels over time. Some studies have suggested that higher CD4 counts at the time of vaccination are associated with better antibody responses [17, 19, 20, 23, 25, 39]. We also noted trends for this association. The associations between seroresponses and both time-updated CD4 counts and plasma HIV RNA levels, and the lack of association between either the CD4 nadir or prior AIDS-defining conditions, suggest that the current and ongoing HIV status after vaccination may be the most important factors in generating and maintaining antibody responses [3, 25, 39, 40].

Our study has potential limitations. First, the study was a retrospective, nonrandomized study of samples collected as part of a natural history study. As part of an observational study, some participants did not have 6–10 year data due to HAV vaccination occurring <6 years ago, or due to loss to follow-up or death. Additionally, our study had a modest sample size that limited the narrowness of subgroup confidence intervals and the evaluation of factors associated with vaccine durability. A particular limitation exists in characterizing vaccinees who did not exhibit an “initial response”; however, our proportion of nonresponders is fairly consistent with estimates from other studies [19]. Second, patients who were vaccinated with alternative doses (other than 2 shots) were excluded from our analyses; hence, our study does not address seroresponses among these patients. Third, direct comparability between HIV-infected and HIV-uninfected persons was limited because our results are from a retrospective study whereas the HIV-uninfected data were from a clinical trial [31]. However, both studies used the similar antibody-testing procedures and were performed in the same laboratory. Fourth, we acknowledge that HAV IgG antibody levels may have been altered by exposures to HAV during the study period. Finally, although the clinical impact of waning HAV antibody concentrations was not evaluable, there is a concern that HIV patients with low antibody levels could develop clinical hepatitis, similar to that reported with hepatitis B [41]. The absolute lower limit of anti-HAV required to prevent HAV infection remains unclear; however, our study did use the cutoff reported as the minimal antibody concentration that provided protection in a prior study [32].

Our study had several strengths. This is the first study among HIV-infected persons to our knowledge to evaluate long-term (up to 10 y) responses after HAV vaccination. In addition, our study was conducted in a cohort with comprehensive clinical data as well as repeated measures that were correlated with the durability of immune responses. Finally, HAV antibody levels were evaluated using an approved and standardized, sensitive laboratory assay.

In summary, HIV-infected adults achieved high initial seroconversion rates after hepatitis A vaccination with most initial responders maintaining seropositive responses for up to 6–10 years. Because HIV patients achieved lower GMCs than HIV-uninfected persons, further studies evaluating the potential clinical implications of lower immune responses among this group are advocated. Maintaining suppressed plasma HIV RNA levels among HIV-infected persons may be an important strategy for sustaining durable antibody levels for vaccine preventable infections such as HAV.

Funding

 This work was supported by the Infectious Disease Clinical Research Program, a Department of Defense program executed through the Uniformed Services University of the Health Sciences (IDCRP-000-11); and the National Institute of Allergy and Infectious Diseases, National Institutes of Health (Inter-Agency Agreement Y1-AI-5072).