Introduction

HIV infected individuals are more susceptible to serious influenza, have longer illness duration, longer periods of viral shedding and suffer higher rates of influenza-related complications including deaths [14]. The recommended single-dose inactivated trivalent influenza vaccine in HIV-infected adults is safe but often elicits a suboptimal antibody response [47]. Although efficacy data are limited, at least one study of TIV in HIV-infected children suggests lower than expected efficacy in this population [8].

Vaccine adjuvants can improve immunogenicity in many vaccine hyporesponsive populations, such as dialysis patients [9], solid organ transplant recipients [10], and the elderly [11]. We previously assessed the AS03A adjuvant in combination with an inactivated split A/California/7/2009 (H1N1) influenza vaccine in a representative HIV study population during the 2009 influenza pandemic [12]. As measured by hemagglutination inhibition (HAI) titers, high-level responses were achieved with a single dose and antibody responses further improved with a booster.

Vaccine efficacy is best determined by the prevention of clinical disease, however, these studies are difficult to perform and require large numbers of subjects, and thus surrogate markers are needed. Although the HAI assay is widely used to assess influenza vaccine efficacy, it provides little information about the functional capabilities of the antibodies induced. Affinity/avidity maturation of B cells, which leads to the production of more strongly binding antibodies, is an important component of the normal immune response. As a result, the development of antibody avidity over time can be helpful in assessing both the kinetics and the quality of an immune response [13]. In the current work, we measured antibody avidity pre- and post-vaccination in a subgroup of HIV seropositive study participants who received the AS03A-adjuvanted vaccine.

Methods

Study Population and Vaccinations

As previously published, a randomized, multicenter, controlled, vaccine study was conducted under the auspices of the Public-Health Agency of Canada-Canadian Institutes of Health Research Influenza Research Network (PCIRN) [ClinicalTrials.gov as NCT01002040] [12]. HIV-infected volunteers were immunized with monovalent, AS03-adjuvanted H1N12009 pandemic (pdmH1N1) inactivated, split influenza vaccine (Arepanrix; GlaxoSmithKline, Laval, QC) containing 3.75 μg HA protein on day 0. Subjects in Group 1 received a single dose (day 0) while subjects in Group 2 received two doses 21 days apart. Sera were collected from all subjects at baseline (day 0), on days 21 and 42, and on day 180 or 6 months after the first dose.

HAI and Avidity Assays

We determined the avidity index (AI) of serum samples obtained at each sampling point (on days 0, 21, 42, and 180) from one of the four participating sites (The Ottawa Viral Hepatitis Program; n = 64). Avidity ELISAs were performed as previously described [14] with minor modifications in the analysis. Briefly, AI was calculated as the concentration of urea needed to displace 50 % of antigen-specific antibodies (based on changes in optical density (OD) following incubation with graded urea concentrations). Serum samples were processed at a dilution of 1:50 or lower to obtain an OD without urea below 2.0. Samples for each subject were run in duplicate with day 0, 21 and 42 samples from the same subject run at the same time and day 180 samples run separately. Control serum with known avidity index (AI) was included with all runs (data not shown). For serum samples with HAI titers below the limit of detection (HAI < 10), AI was considered unmeasurable; AI of these samples were omitted from graphs and identified in Table II as not detected (ND).

The HAI titers of all subjects were previously reported [12]. HAI titers below the limit of detection (<10) are identified in graphs as not detected (ND) and were assigned a value of 5 in the current work for statistical analysis.

Statistical Analyses

Comparison of HAI titer and AI between groups was determined by the Mann–Whitney U Test. Investigation of the involvement of HIV parameters was performed by Kruskal-Wallis non-parametric one-way analysis of variance (ANOVA) followed by Dunns post-test to compare all pairs of groups. All calculations for statistical significance were performed using GraphPad Prism 5.0 software. P value <0.05 was considered statistically significant.

Results

Study Population

Of the 64 subjects included in this sub-study, 28 (44 %) had received a single dose (Group 1) while the remaining 36 (56 %) received 2 doses (Group 2). The subset of subjects included in this sub-study had baseline characteristics similar to the entire original study population [12] and were well-balanced between the two groups (Table I). Most subjects (46/64, 72 %) had baseline HAI titers <10 (below the limit of detection) against pdmH1N1 and eight (12.5 %) had HAI titers ≥40 [geometric mean titer (GMT): 95]. Baseline HAI titers for the remaining ten subjects fell between 10 and 39. Long term (day 180) serum samples and HAI titer data were available for 79 % and 89 % of subjects in Groups 1 and 2, respectively.

Table I Baseline characteristics of vaccinated HIV study participants

Baseline HAI Titers ≥ 10

A minority of subjects were found to have detectable HAI titers (HAI ≥ 10) at the beginning of the study (18/64, 28 %) (Fig. 1a, c; Table II) and were analysed separately. The baseline HAI GMT of these subjects was 37. Among them, eight (8/18, 44 % or 12.5 % of the whole study population) had HAI titers ≥ 40 before immunization (Table II). After the first dose of vaccine, HAI titers increased in both groups (Fig. 1a, c) with 9/10 or 8/8 subjects seroprotected in Group 1 and 2, respectively (Table II). No significant increase in HAI titer was observed in subjects 21 days after the booster immunization (Fig. 1c) and the same proportion of subjects remained seroprotected (Table II). Similarly, no major difference was observed with one or two doses of vaccine in the maintenance of long-term HAI titers at 6 months (Fig. 1a, c; Table II).

Fig. 1
figure 1

Antibody responses in HIV-infected patients immunized with one or two doses of AS03-adjuvanted pdmH1N1 split vaccine. Subjects in Group 1 received the priming immunization only (day 0), while subjects in Group 2 received prime (day 0) and booster immunization (day 21). Serum was collected at baseline on day 0, and following immunizations on day 21, 42, and 180 (6 months) after priming immunization. The HAI titer data were previous determined and adapted with permission from [12]. The avidity index (AI) of each sample was determined by ELISA. HAI titer (a) and AI (b) of subjects with baseline HAI titer ≥ 10 in Group 1. HAI titer (c) and AI (d) of subjects with baseline HAI titer ≥ 10 in Group 2. HAI titer (e) and AI (f) of subjects with baseline HAI titer < 10 in Group 1. HAI titer (g) and AI (h) of subjects with baseline HAI titer < 10 in Group 2. Individual subjects are shown. ND not detected

Table II Antibody responses in HIV-infected patients immunized with one or two doses of AS03-adjuvanted pdmH1N1 split vaccine

The avidity maturation profiles of Group 1 and 2 subjects were similar with AI unchanged after the first immunization, and remained stable with or without booster immunization (Fig. 1b, d; Table II). Since AI was ~9 at all time-point with these subjects (Table II), this suggests that an AI value of 9 appears to be the highest AI achievable in this population detected by this assay.

Baseline HAI Titers <10

The remaining subjects (46/64; 72 %) had HAI titers below the limit of detection at baseline and can reasonably be considered as immunologically naïve for the pdmH1N1 virus (Fig. 1e, g; Table II). In both groups, AI and HAI titers increased after the first dose. At day 21, the AI of all the naïve subjects reached ~9 (Fig. 1f, g; Table II) which was comparable to the highest AI achieved in the pdmH1N1-experienced subjects at all time-points (Table II).

Virus-naïve subjects at baseline (HAI <10) who received one dose of vaccine failed to consistently produce substantial and sustained antibody responses. At day 42, 10/18 subjects in Group 1 were seropositive (Fig. 1e; Table II), in comparison, 24/28 subjects in Group 2 were seroprotected and these subjects also had a significantly higher HAI GMT (Fig. 1g; Table II). Furthermore, this trend of higher HAI titers after two doses of vaccine (Group 2) was maintained at 6 months after vaccination. The long-term HAI GMT of the subjects in Group 1 was calculated to be 8.6, but only a minority of subjects responded to the vaccine to produced detectable HAI titers (HAI ≥ 10) (Fig. 1e; Table II). At 6 months, the majority of naive subjects in Group 1 (9/16; 56 %) had undetectable HAI titers (HAI < 10). In comparison, long-term HAI titers of the subjects in Group 2 were considerably higher (Fig. 1g; Table II). Further, only one subject in Group 1 remained seroprotected at 6 months (1/16, 6 %) (Fig. 1e), while almost one third of the subjects in Group 2 maintained seroprotective HAI titers (8/25, 32 %) (Fig. 1g; Table II).

Interestingly, if detectable HAI titers (HAI ≥ 10) were produced after vaccination, optimal AI was rapidly achieved (Fig. 1f, h; Table II). Subjects in Groups 1 and 2 who produced detectable HAI titers (HAI ≥ 10) had AI of ~9 at days 21, 42 and 180 (Table II).

HIV Characteristics Sub-Analysis

Although we are limited by sample size, in general, HIV-related parameters did not correlate with AI profiles or HAI titers. Overall, very few subjects had detectable HIV viremia (4 and 7 subjects in Group 1 and Group 2, respectively). However, based on detectable or suppressed HIV RNA, no major differences in the HAI titer or AI were observed following either the first (at day 21) or second doses (at day 42 or 180), regardless of baseline HAI titer (Supplementary Fig. 1). Similarly, comparable AI and HAI titers were achieved in subjects irrespective of antiretroviral use after one or two immunizations, regardless of baseline HAI titer (Supplementary Fig. 2). All groups and subgroups had similar CD4 T-lymphocyte counts (Supplementary Fig. 3).

Discussion

AS03A and MF59 are examples of oil in water emulsions vaccine adjuvants [15]. The safety and immunogenicity of influenza subunit vaccines containing these adjuvants is now well-established in immune competent and vaccine hyporesponsive populations including those living with HIV [10, 1622]. HIV infection is associated with deficiencies in both humoral and cell-mediated immunity, which can both alter the course of common infections and influence vaccine immunogenicity [4, 5, 23]. Despite these deficiencies, our work and that of others has generally demonstrated high seroconversion and seroprotection rates with the use of AS03A-adjuvanted influenza vaccines in people living with HIV [12, 2426].

In our study, most subjects generated seroprotective HAI titers at early time-points (day 21 and 42); however, titers were reduced in comparison to those reported in healthy adults immunized with the same AS03-adjuvanted monovalent H1N1 vaccine (Arepanrix) [27, 28]. After one dose, HAI GMT ranging from 140 to 375 at day 21, ~250 at day 42 and ~125 at 6 months were reported in healthy adults [27, 28]. Those given two doses following the same immunization timeline as this study (day 0 and 21) achieved HAI GMT ranging from 300 to 500 at day 42 and 114–225 at 6 months [27, 28]. The inferior HAI responses in the HIV population of our study were especially evident at the 6 month time-point.

Pre-existing antibody titers specific for H1N12009 pandemic with high AI were identified at baseline in a portion of study participants and was consistent with other studies. Most H1N12009 trials in HIV populations report some degree of pre-existing H1N12009 immunity ranging from less than 10 % to nearly 50 % [29]. Natural infection with H1N12009 prior to study participation is the most likely explanation especially in studies conducted well into the H1N12009 pandemic [20]. The H1N1 pandemic presented in two waves in Ontario, Canada, the first wave ended in June 2009 and the second wave ended in November 2009 [30]. Baseline serum samples were collected during the first three weeks of November 2009, and although exclusion criteria included lab confirmed cases of pdmH1N1 [12], baseline HAI titers and AI strongly suggest that some of the subjects had had prior exposure to wild-type pdmH1N1 virus. Conversely, cross-immunity between pandemic and seasonal strains of influenza is another less likely potential explanation for baseline titers [31]. Only a small proportion of individuals (~5 %) below the age of 65 exhibited elevated cross-reactive antibodies to pdmH1N1 before the pandemic and the highest prevalence of cross-reactive antibodies was in subjects above the age of 65 (~15 %) [32]. The mean age of subjects in this study was 47 (Group 1) and 42 years (Group 2).

Avidity, or binding strength, is a key measure of antibody functionality. In the clinical setting, antibody avidity can be used to differentiate between recent and latent infections [3335], as well as to differentiate between primary and secondary vaccine failure [3638]. After the initiation of an immune response, antigen-specific antibody avidity usually increases until a maximal AI is achieved. This process requires both time and large enough amounts of antigen to drive antibody maturation to “completion”. Once a fully mature (avid) antibody response has been produced, AI values are typically sustained for prolonged periods of time, possibly for life [39]. Although there are currently no standard reagents for avidity analyses, and therefore no standardized correlates for “normal” or “maximal” avidity, the avidity profile following influenza vaccination would be expected to follow this same pattern; AI increasing to a maximal value, and then remaining stable over time.

Due to the nature of the assay and analysis methodology, we were unable to conclusively assess antibody avidity at very low titers. As HAI titer decreases, measurement of AI typically skews towards high values. Furthermore, as HAI titer decreases to undetectable levels, AI becomes unmeasurable. For these reasons, we omitted the calculated AI of samples with titers below the limit of detection of the HAI assay. Regardless of these limitations, the AI values of serum samples in this study ranged from 4 (minimum) to 27 (maximum). Furthermore, the AI and HAI datasets of this study are both negatively skewed and follow similar distribution, which supports the validity of our avidity assay.

In the current work, the expected avidity profile after immunization was observed in all subjects who mounted measurable HAI titers. The average AI achieved by our subjects was 9–10 which was within the same range as AI produced in healthy children immunized with the AS03-adjuvanted pdmH1N1 monovalent split vaccine [14]. Subjects with pre-existing antibodies generated very high HAI titers after one or two vaccinations and sustained high antibody avidity up to 6 months post-boost. These results suggest that optimal antibody avidity was achieved before this study as a result of prior exposure to virus, and furthermore suggests that these subjects would be well-protected against influenza infection, although we do not have supportive data on clinical protection.

Most of the subjects had undetectable HAI titers at study baseline but still generated high antibody avidity after vaccination. These subjects appeared to benefit the most from booster immunization and were able to achieve higher HAI titers in the long-term. More subjects retained seroprotective HAI titers at 6 months after two in comparison to one dose of vaccine, suggesting that the naive subjects who received two doses were better protected over the long term than subjects who received only one immunization. Of those who maintained long-term detectable HAI titers, high avidity antibodies were produced irrespective of the number of doses of vaccine. No major effect on avidity was observed in response to the booster immunization, AI was similar between Groups 1 and 2. Given that the purpose of influenza vaccines is to provide protection during a flu “season” lasting ~6 months, and the observation that the HIV-infected naïve subjects produced and maintained high avidity antibodies detectable in the serum up to 6 months indicates that they were likely well-protected against influenza infection.

In contrast to a previous report that demonstrated reduced B cell responses to influenza vaccination in HIV-infected individuals [5], we showed that once antibody titers are produced after vaccination, essentially normal B cell and avidity maturation are able to proceed. We show that affinity maturation proceeds fairly quickly and had reached peak AI by the first timepoint (three weeks after immunization). A limitation, however, is the lack of data from HIV-negative individuals in this study to directly compare avidity responses. We also lack a comparison group with individuals immunized with non-adjuvanted vaccine to investigate the involvement of AS03 adjuvant on the antibody avidity.

In the literature, data indicate the likelihood of achieving protective HAI antibody titers is lower in HIV-positive subjects with detectable HIV RNA levels and/or with low CD4 T-lymphocyte counts [1, 4042]. Protection from influenza disease provided by vaccination is also reported to be less than optimal in these subjects. In previous work, we demonstrated that influenza was responsible for at least 40 % of all febrile respiratory illnesses among an outpatient population of HIV-infected patients despite an influenza vaccination rate of 76 % [43]. In the current study, the presence of detectable HIV RNA, use of ARV therapy at the time of immunization and baseline CD4 T lymphocyte count did not appear to affect HAI titers or AI responses following vaccination. However, the small sample size limited our ability to identify relationships in the exploration of HIV characteristics and AI response. Further, other groups have identified transient decreases in CD4 T cell count [44] and increases in plasma viral RNA [45] after vaccination with the same AS03-adjuvanted pdmH1N1 split vaccine. Although transient, these changes could impact the ability to achieve and maintain long-term antibody titers with high avidity. Furthermore, these changes could theoretically have a greater influence on the vaccine immune response in subjects lacking pre-existing antibodies (naïve subjects) who receive a single dose of vaccine.

Conclusions

We show that immunization with two doses of AS03A-adjuvanted influenza vaccine is better at inducing long term (6 month) seroprotective HAI titers, especially in antigenically naïve HIV-positive subjects. High avidity antibodies are rapidly produced following immunization and once long term HAI titers are produced (following one or two immunizations) both antibody titer and avidity are well maintained up to 6 months. These data provide further support of a two-dose immunization strategy for HIV-positive subjects particularly in the setting of a novel (pandemic) influenza strain.