Responses to pandemic ASO3-adjuvanted A/California/07/09 H1N1 influenza vaccine in human immunodeficiency virus-infected individuals

General characteristics of 93 HIV-infected individuals, of whom 81 received and 12 declined the 2009 H1N1 vaccine, are shown in Table1. Of the vaccinated individuals, 63% (51/81) had plasma HIV viral loads below detectable levels (<1.6 log₁₀), as did 67% (8/12) of the unvaccinated group. Twelve vaccinated individuals were not receiving antiretroviral therapy at the time of vaccination and through the follow-up period. For the 81 vaccine recipients, the mean peripheral blood CD4⁺ T cell count ± SD pre-vaccination was 534 ± 277/μL and post-vaccination was 500 ± 266/μL. Excluding those vaccinated individuals whose viral load changed by > 1 log₁₀ over the observation period (12/81), the mean peripheral blood CD4⁺ T cell count ± SD pre-vaccination was 576 ± 268/μL pre-vaccination and 532 ± 261/μL post-vaccination. Neither change was statistically significant (p = 0.16 and p = 0.11, Mann–Whitney test for the entire group or with exclusion of those whose viral load changed by > 1 log₁₀ over the observation period respectively). Over a similar time period, the mean CD4⁺ T cell count of the 12 unvaccinated HIV-infected individuals (none of whom had a change in HIV plasma viral load of > 1 log₁₀ over the observation period) was 429 ± 235/μL initially and 418 ± 180/μL at the end of the observation period. There was no significant difference in the median change in CD4⁺ T cell counts between these groups over the autumn 2009 through early winter 2010 observation period whether the vaccinated individuals with a change in HIV plasma virus load of > 1 log₁₀ were included (−34, IQR −104-65) or excluded (−39, IQR −105-42) versus 10, IQR −21-28, (Figure1a). The small size of the group declining the vaccine (n = 12) precluded robust comparison of changes in CD4⁺ T cell numbers in inherently variable HIV-infected populations. Therefore, we also compared median change in circulating CD4⁺ T cell numbers within the vaccinated group over the period 1–5 months from receiving the vaccine to median change in CD4⁺ T cell numbers over an equivalent time period approximately 1 year later. While the seasonal flu vaccine in 2010 included A/California/07/09 H1N1, the ASO3 adjuvant was not used. Over the same time interval one year post A/California/07/09 H1N1 vaccination (2010/2011), the mean CD4⁺ T cell count of the H1N1-vaccinated group was 549 ± 283/μL initially and 573 ± 278/μL at the end of the observation period. There was no significant change in CD4⁺ T cell counts across this period (p = 0.3144, Mann–Whitney test). Excluding 4 cases where viral load rose or fell by > 1 log₁₀ over the observation period, the mean CD4⁺ T cell count of the H1N1-vaccinated group was stable at 584 ± 259/μL initially and 593 ± 259/μL at the end of the observation period. The median change in CD4⁺ T cell numbers over a 1–5 month interval spanning fall/winter 2010/2011 was significantly different from the median change in CD4⁺ T cell counts that occurred in the 1–5 month (2009/2010) immediate post A/California/07/09 H1N1 vaccination period. This was consistent whether the 12 cases in 2009/2010 and 4 cases in 2010/2011 where viral load rose or fell by > 1 log₁₀ were included (−34, IQR −104-65 versus 15, IQR −56-92, p = 0.0128, Mann–Whitney test, Figure1b) or excluded from analysis (−39, IQR −105-42 versus 7.0, IQR −57-92, p = 0.0098, Mann–Whitney test, Figure1c). These data suggest that primary A/California/07/09 H1N1 Arepanrix^TM vaccination had a slight and transient negative effect on circulating CD4⁺ T lymphocyte numbers in the group of HIV-infected individuals studied. However, the effect was relatively minor compared to the overall variation in CD4⁺ T cell counts observed over the follow-up period. Therefore, after measuring vaccine responses, we carried out a sub-analysis within the vaccinated group comparing the median change in CD4⁺ T cell numbers of responders to the median change in CD4⁺ T cell numbers of non-responders over the period 1–5 months from receiving the vaccine.

In cases where pre-existing antibodies reactive with A/California/07/09 H1N1 HA antigen were present [baseline mean fluorescence intensity (MFI) ≥ 125], an increase of ≥ 250 MFI units was considered a positive response. For those individuals without pre-existing antibody responses against A/California/07/09 H1N1 HA, an increase in MFI to ≥ 250 was considered a positive vaccine response. Consistent results were obtained with repeat testing of 80 pre- and post-vaccination sample pairs. Applying the criteria stated above, 28/80 individuals had pre-existing antibodies reactive with A/California/07/09 H1N1 HA antigen, 36/80 (45%, 95% CI 34%-56%) HIV-infected individuals responded to the vaccine and 44/80 were non-responders (Figure2a). Twenty-two of the 28 (79%, 95% CI 63%-94%) HIV-infected individuals with pre-existing antibodies against A/California/07/09 H1N1 HA antigen responded to the vaccine compared to only 14 of the 52 (27%, 95% CI 15%-39%) without pre-existing antibodies against A/California/07/09 H1N1 HA antigen. Thus, HIV-infected individuals with pre-existing antibodies against A/California/07/09 H1N1 HA antigen were considerably more likely to respond to the vaccine (p = 0.00001, Fisher’s exact test). If some of those individuals with pre-existing antibodies were actually infected with A/California/07/09 H1N1 HA antigen over the follow-up period for plasma collection, this could result in overestimation of the response rate in this group. A limitation in the retrospective design of our study was the variable timing in collection of post-vaccination plasma samples for analysis of anti- A/California/07/09 H1N1 HA antibody responses.

There was no significant difference in mean age between responder and non-responder groups. However, as shown in Figure2b, responders without pre-existing antibodies against A/California/07/09 H1N1 HA antigen had a lower mean age (41.4 ± 5.7 years) than either responders with pre-existing antibodies (46.1 ± 8.0 years, p = .03, Student’s t test) or non-responders (46.2 ± 7.0 years, p = .01, Student’s t test). Although there was a wide range of CD4⁺ T cell counts in both groups, the median CD4⁺ T cell count was significantly higher in the responder group than in the non-responder group (618, IQR 389–835 versus 446, IQR 314–582, p = 0.0157, Mann–Whitney test, Figure3a). There was no significant difference in median CD4⁺ T cell nadir between responders and non-responders (126, IQR 83–199 versus 143, IQR 61–202), nor between responders with or without pre-existing antibodies against A/California/07/09 H1N1 HA antigen (165, IQR 69–270 versus 210, IQR 63–314) respectively. Median plasma HIV viral load at the time of vaccination was not significantly different between groups and several responders had relatively high plasma levels of HIV (Figure3b). The effect of multiple variables on vaccine response was assessed by stepwise logistic regression. Age, gender, time since immunization, pre- and post-immunization CD4⁺ T cell counts, pre- and post- HIV viral load, and CD4⁺ T cell nadir did not independently affect response to influenza vaccine. Pre-existing influenza antibody was the only predictor (F = 119.9; p=0.002) of vaccine response in both univariate and multivariate analyses.

Although the median loss of CD4⁺ T cells over the period 1–5 months from receiving the vaccine was significantly greater than over the same time period 1 year later, the effect was small relative to overall variation in CD4⁺ T cell counts. To test for specificity of the effect in relation to the effect of vaccination, we compared CD4⁺ T cell losses between HIV-infected vaccine recipients who exhibited an increase in anti-A/California/07/09 H1N1 antibodies and those who did not. Our reasoning was that if CD4⁺ T lymphocyte counts were affected by the vaccine, the impact would be greatest in those making a measurable immune response against it. There was a significantly greater median loss of CD4⁺ T cells over the period from 1–5 months of receiving the A/California/07/09 H1N1 vaccine in the group of 36 individuals who responded to the vaccine compared to the 44 non-responders (−71, IQR −160-18 versus −31, IQR −80-78, p = 0.0214, Mann–Whitney test, Figure4a). Results were similar when 4 responders and 8 non-responders whose HIV virus load changed by > 1 log₁₀ over the observation period were excluded from analysis (−78, IQR −160-6.0 versus −36, IQR −80-70, p = 0.0204, Mann–Whitney test, Figure4b).

This was a single centre retrospective study carried out in the provincial HIV clinic located in St. John’s, NL, Canada between October, 2009 and March, 2011. Ethical approval was received from the Health Research Ethics Authority of Newfoundland and Labrador. Individuals attending the St. John’s HIV clinic in autumn 2009 received priority access to the A/California/07/09 H1N1 vaccine Arepanrix^TM (GlaxoSmithKline). The clinic nurse directly administered the vaccine to 81 HIV-infected individuals attending the clinic during the first week of October 2009 (vaccinated group), while 12 other HIV-infected individuals under the clinic’s care declined vaccination; these individuals comprise the unvaccinated group. Plasma samples were collected with informed consent as part of an ongoing research study on immune responses in HIV infection. Pre-and post-vaccination plasma samples were obtained from 80/81 vaccinated individuals. Lymphocyte subset counts and plasma virus load (Roche Amplicor®) measurements for time points immediately prior to (“pre-vaccine period”) and between 1 and 5 months post vaccination (“post-vaccine period”) were collected from clinical chart information. Measurements were taken from corresponding time periods for the unvaccinated group for comparison purposes. Changes in CD4⁺ T cell counts from the pre-vaccine to post-vaccine period plasma samples were compared between the vaccinated and unvaccinated groups. In 12 cases where the plasma HIV viral load either rose or fell by more than 1 log₁₀ across the pre and post vaccination span (4 responders and 8 non-responders), the changes in CD4⁺ T cell counts were excluded from comparison in sub-analyses, which were carried out separately. This was done because of the impact that large changes in HIV replication levels due to either de novo failure of, or response to antiretroviral therapy over the vaccine administration and follow-up period would likely have on CD4⁺ T cell numbers, irrespective of any vaccine effect. Comparison to changes in CD4⁺ T cell counts over a similar time period 1 year later was carried out within the vaccinated group using the same 1 log₁₀ change in plasma HIV viral load exclusion criteria for sub-analysis.

Plasma samples from the pre-vaccine period and post-vaccine period were collected and stored at −80°C until tested. Antibody levels against the A/California/07/09 H1N1 HA antigen and 11 additional influenza HA antigens before and after vaccination were measured as in[13]. Briefly, recombinant HA antigens were coupled to 5.5 μm microspheres, which were then incubated with plasma samples. Antibodies binding to the beads were subsequently detected with biotinylated anti-human IgG antibodies and fluorochrome-conjugated avidin. The intensity of the anti-human IgG bound fluorochrome is proportional to the amount of antibody bound to A/California/07/09 HA, which is calculated by flow cytometry with calibrated standards. Where antibodies reactive with A/California/07/09 H1N1 HA antigen were present at baseline (MFI ≥ 125), increases of ≥ 250 MFI units were considered to indicate positive humoral responses to the vaccine. Without pre-existing antibody responses against A/California/07/09 H1N1 HA, increases in MFI to ≥ 250 were considered to indicate positive humoral responses to the vaccine.