Antibody response durability following three-dose coronavirus disease 2019 vaccination in people with HIV receiving suppressive antiretroviral therapy

Background: Limited data exist regarding longer term antibody responses following three-dose coronavirus disease 2019 (COVID-19) vaccination, and the impact of a first SARS-CoV-2 infection during this time, in people with HIV (PWH) receiving suppressive antiretroviral therapy (ART). We quantified wild-type-specific, Omicron BA.1-specific and Omicron BA.5-specific responses up to 6 months post-third dose in 64 PWH and 117 controls who remained COVID-19-naive or experienced their first SARS-CoV-2 infection during this time. Design: Longitudinal observational cohort. Methods: We quantified wild-type-specific and Omicron-specific anti-Spike receptor-binding domain IgG concentrations, ACE2 displacement activities and live virus neutralization at 1, 3 and 6 months post-third vaccine dose. Results: Third doses boosted all antibody measures above two-dose levels, but BA.1-specific responses remained significantly lower than wild-type-specific ones, with BA.5-specific responses lower still. Serum IgG concentrations declined at similar rates in COVID-19-naive PWH and controls post-third dose (median wild-type-specific and BA.1-specific half-lives were between 66 and 74 days for both groups). Antibody function also declined significantly yet comparably between groups: 6 months post-third dose, BA.1-specific neutralization was undetectable in more than 80% of COVID-19 naive PWH and more than 90% of controls. Breakthrough SARS-CoV-2 infection boosted antibody concentrations and function significantly above vaccine-induced levels in both PWH and controls, though BA.5-specific neutralization remained significantly poorer than BA.1 even post-breakthrough. Conclusion: Following three-dose COVID-19 vaccination, antibody response durability in PWH receiving ART is comparable with controls. PWH also mounted strong responses to breakthrough infection. Due to temporal response declines, however, COVID-19-naive individuals, regardless of HIV status, would benefit from a fourth dose within 6 months of their third.

Longitudinal monitoring of immune responses post-third dose in PWH is critical to inform the timing of future immunizations in this group. Though some data are available on initial immunogenicity to third COVID-19 vaccine doses in PWH [14][15][16], no studies to our knowledge have assessed the longer term durability of post-third dose responses in this population. Furthermore, despite the high incidence of first-time SARS-CoV-2 infections after three vaccine doses, no studies to our knowledge have examined the impact of such infections on responses in PWH. Here, we extend prior observations from our cohort [14,17] by quantifying wildtype-specific, Omicron BA.1-specific and BA.5-specific responses up to 6 months post-third vaccine dose in 64 PWH and 117 controls who either remained COVID-19-naive or experienced their first (presumably Omicron [18]) SARS-CoV-2 infection, during this period.
the SARS-CoV-2 Spike Receptor Binding Domain (RBD) using the V-plex SARS-CoV-2 (IgG) ELISA kit (Panel 22; Meso Scale Diagnostics, Rockville, Maryland, USA), which features wild-type and Omicron-BA.1 RBD antigens, on a Meso QuickPlex SQ120 instrument. Serum was diluted 1 : 10000 and reported in WHO International Standard Binding Antibody Units (BAU)/ml using the manufacturer-supplied conversions. Surrogate virus neutralization activity [20] in serum was measured by competition ELISA using the same kit [Panel 22; V-plex SARS-CoV-2 (ACE2)] to measure blockade of the RBD-ACE2 receptor interaction. Sera were diluted 1 : 40 and results reported as % ACE2 displacement. Virus neutralizing activity in plasma was assessed using live wild-type (USA-WA1/2020; BEI Resources, Manassas, Virginia, USA), and two local isolates identified as Omicron BA.1 (GISAID Acces-sion# EPI_ISL_9805779) and Omicron BA.5 (GISAID Accession# EPI_ISL_15226696) on VeroE6-TMPRSS2 (JCRB-1819) target cells [19]. Virus stocks were diluted to 50 TCID 50 /200 ml in the presence of serial two-fold plasma dilutions (1/20 to 1/2560) and added to target cells in triplicate. Viral cytopathic effects (CPE) were recorded 3 days post-infection. Neutralization was reported as the highest reciprocal dilution able to prevent CPE in all three wells. Partial or no neutralization at 1/20 dilution was considered below the limit of quantification (BLOQ) and coded as a reciprocal dilution of 10.

Statistical analyses
Continuous variables were compared using the Mann-Whitney U test (unpaired data) or Wilcoxon test (paired data). Relationships between continuous variables were assessed using Spearman's correlation. In participants who remained COVID-19-naive, multiple linear regression was used to investigate the relationship between HIV infection and vaccine-induced immune measures using a confounder model that adjusted for variables that could influence vaccine responses or that differed in prevalence between groups. For Omicronspecific neutralization at 6 months post-third dose, multiple logistic regression was used because of the high proportion of results BLOQ. Included variables were: HIV infection (controls as reference group), age (per year), sex at birth (female as reference), ethnicity (nonwhite as reference), number of chronic conditions (per additional), dual ChAdOx1 as the initial regimen [mRNA or mixed (ChAdOx1/mRNA) regimen as the combined reference group] [14,17], third COVID-19 mRNA dose brand (BNT162b2 as reference) and the interval between second and third doses (per day). Plasma neutralization models also corrected for anticoagulant (ACD as reference). All tests were two-tailed, with P less than 0.05 considered statistically significant. Analyses were conducted using Prism v9.2.0 (GraphPad Software, San Diego, California, USA). Characteristics of the 64 PWH and 117 controls, all of  whom remained COVID-19-naive until at least 1 month  post-third dose, are shown in Table 1. All PWH had suppressed plasma HIV viremia on ART, median CD4 þ T-cell counts of 645 [interquartile range (IQR) 473-958] cells/ml, and median nadir CD4 þ T-cell counts of 225 (IQR 95-485) cells/ml, at enrolment. PWH were a median of 57 (IQR 42-65) years old and 90% men; controls were a median 47 (IQR 35-72) years old and 73% women. PWH had a higher proportion of white ethnicity (72%, compared with 55% in controls) and more chronic health conditions [median 1 (IQR 1-3) compared with 0 (IQR 0-1) in controls]. More PWH (10%) than controls (<1%) received two doses of the recombinant viral vector ChAdOx1 vaccine as their initial immunization series. All third doses were monovalent mRNA vaccines, either BNT162b2 (30 mg) or mRNA-1273 (50 or 100 mg). Most PWH (69%) and controls (62%) received mRNA-1273, where, per local guidelines, all adults aged at least 70 years were eligible for a 100 mg mRNA-1273 dose, as were PWH who met one or more of the following criteria: age at least 65 years, prior AIDS-defining illness, prior CD4 þ T-cell count less than 200 cells/ml, prior CD4 þ T-cell fraction 15% or less, any plasma HIV load greater than 50 copies/ ml in 2021, or perinatally acquired HIV [21]. Third vaccine doses were administered approximately 6.5 months after the second dose.

Participant characteristics
A total of 24 (38%) PWH and 45 (39%) controls experienced their first SARS-CoV-2 infection between 1 and 6 months post-third dose (one control experienced two infections during this period [22]). Based on the 56 (81%) participants for whom infection timing was available, infections occurred a median 105 (IQR 76-137) days post-third dose, or a median date of 10 April 2022 (IQR 28 February to 04 May), with no temporal differences between PWH and controls (P ¼ 0.4). Though SARS-CoV-2 variant information is unavailable for individual infections, most were likely Omicron BA.1 or BA.2 based on local epidemiology at the time [18].
Similarly, Omicron BA.1-specific IgG responses were comparable in COVID-19-naive PWH and controls at all post-third dose time points (all comparisons P ! 0.15; Fig. 1a). Nevertheless, BA.1-specific responses were significantly lower than wild-type-specific responses in all groups at all time points (all within-group comparisons of wild-type-and BA.1-specific responses were P < 0.0001; not shown). One month post-third dose, for example, BA.1-specific IgG concentrations were 2.96 (IQR 2.71-3.32) log 10 BAU/ml in PWH, which was 0.74 log 10 BAU/ml lower than wild-type-specific concentrations at this time. By 6 months, BA.1-specific IgG concentrations had declined to 2.47 (IQR 2.14-2.59) log 10 BAU/ml in PWH, which was 0.49 log 10 BAU/ml lower than wildtype-specific concentrations at this time.
We next performed multivariable analyses adjusting for sociodemographic, health and vaccine-related variables to identify variables associated with wild-type-specific and BA.1-specific IgG concentrations at 6 months post-third dose in the COVID-19-naive subgroup. These analyses revealed that a higher number of chronic health conditions -but not HIV infection -was associated with poorer IgG responses at this time (P ¼ 0.028 for wild-type-specific responses; 0.016 for BA.1-specific responses), as was male sex (P ¼ 0.012 for BA.1-specific responses) (Supplementary Table 1, http://links.lww. com/QAD/C779). In fact, adjusted IgG concentrations were slightly higher in PWH compared with controls, though this was not statistically significant. Of note, receipt of an mRNA-1273 third dose (rather than BNT162b2) was associated with stronger wild-typespecific IgG responses (P ¼ 0.029), though not BA.1specific responses (P ¼ 0.13), 6 months post-third dose. Among PWH, we also observed no significant relationship between most recent or nadir CD4 þ T-cell count and either wild-type-specific or BA.1-specific IgG concentrations at this time (P ! 0.3; Supplementary  Figure 1, http://links.lww.com/QAD/C786).
We estimated the half-lives of wild-type-specific IgG following three-dose vaccination to be a median 66 (IQR 47-89) days in COVID-19-naive PWH compared to 72 (IQR 54-96) days in controls, a difference that was not statistically significant (P ¼ 0.  Sociodemographic, health and vaccine data were collected by self-report and confirmed through medical records wherever available. b Chronic conditions were defined as hypertension, diabetes, asthma, obesity, chronic diseases of lung, liver, kidney, heart or blood, cancer, and immunosuppression because of chronic conditions or medication.  By contrast, the nearly 40% of PWH and controls who experienced their first SARS-CoV-2 infection between 1 and 6 months post-third dose exhibited markedly higher wild-type-specific and BA.1-specific IgG concentrations than their COVID-19-naive counterparts at all postinfection time points (all comparisons P < 0.0001 in Fig. 1a; also see Fig. 1b). In fact, at 6 months post-third dose, IgG responses in this 'hybrid immunity' group were significantly higher than those initially induced by vaccination alone, for example, BA.1-specific IgG concentrations were a median 3.26 log 10 BAU/ml (IQR 2.93-3.52), which was 0.27 log 10 BAU/ml higher than at 1 month post-third dose (P < 0.0001). Importantly, the magnitude of these 'hybrid' IgG responses was comparable between PWH and controls at all postinfection time points tested (all P ! 0.2; Fig. 1b, see P values above small brackets). Notably, while most participants experienced a marked boost in antibody levels following SARS-CoV-2 infection, IgG responses in a minority of PWH and controls remained constant or even declined post-infection (Fig. 1b).
BA.1-specific responses remained significantly lower than wild-type-specific responses at all time points (all withingroup comparisons for wild-type-specific and BA.1specific responses P < 0.0001; not shown), but these responses also declined similarly in COVID-19-naive PWH and controls (Fig. 2a). Between 1 and 6 months postthird dose, for example, BA.  counts were eligible for the higher (100 mg) third mRNA-1273 dose [14].
Similar to IgG concentrations, a SARS-CoV-2 breakthrough infection markedly boosted wild-type-specific and BA.1-specific ACE2 displacement activities at both 3 and 6 months post-third dose (all P < 0.0001), where activities at 6 months in this group were overall significantly greater than peak responses induced by three-dose vaccination (both P < 0.0001) (Fig. 2a). For instance, BA.1-specific ACE2 displacement activity was 95.5% (IQR 90.6-97.6) in the hybrid group 6 months post-third dose, which was 31% higher than that elicited by vaccination alone. Importantly, the magnitude of these 'hybrid' ACE2 displacement responses was comparable between PWH and controls at all post-infection time points tested (all P ! 0.2; Fig. 2b, see P values above small brackets). It is again notable, however, that a minority of PWH and controls did not show appreciable increases in this activity following infection (Fig. 2b).

Longitudinal viral neutralization activity following three-dose vaccination
One month post-third dose, wild-type-specific neutralization activity in PWH (median reciprocal plasma dilution 320; IQR 160-1280) was slightly higher than in controls (median 320, IQR 160-320; P ¼ 0.004) (Fig. 3a), though this did not remain significant after multivariable adjustment as previously reported [14]. COVID-19-naive PWH continued to maintain higher wild-type-specific neutralization throughout follow-up: by 3 months, neutralization declined to 160 (IQR 80-320) in PWH compared with 80 (IQR 40-160) in controls (P ¼ 0.02), while by 6 months, neutralization had declined to a median 80 (IQR 35-160) in PWH and 40 (IQR 20-80) in controls (P ¼ 0.006), values that were below the levels originally elicited by two-dose vaccination (both groups P 0.01). As reported previously [14], we attribute the higher (univariable) post-third dose neutralization in PWH to the fact that the majority of PWH met one or more of the eligibility criteria for receipt of a full (100 mg) third dose of mRNA-1273, rather than the standard 50 mg mRNA-1273 booster that was offered to the general population. Indeed, after adjustment for sociodemographic, health and vaccinerelated variables (including third-dose vaccine brand), HIV infection did not remain significantly associated with wild-type-specific neutralization activity at 6 months post-third dose (Supplementary Table 4, http://links. lww.com/QAD/C782). Rather, a higher number of health conditions was the only variable independently associated with poorer wild-type-specific neutralization at this time.
By contrast, individuals who experienced breakthrough infection showed significantly stronger wild-type-specific and BA.1-specific neutralization compared with their COVID-19-naive counterparts at 3 and 6 months (all P < 0.0001; Fig. 3a), responses that were significantly higher than those induced by vaccination alone (all comparisons P 0.0001) (Fig. 3a). At 6 months, for example, BA.1-specific responses in participants with hybrid immunity were a median 80 (IQR 80-280), compared with the cohort median of 40 (IQR 20-80) 1 month post-third dose. Importantly, the magnitude of these 'hybrid' neutralization responses was comparable between PWH and controls at all post-infection time points tested (all P ! 0.2; Fig. 3b, see P values above small brackets).
Responses to newer Omicron variants: BA.5 Since the emergence of Omicron BA.1, even more immune evasive variants have developed, including Omicron BA.5 that has dominated globally [23][24][25][26]. We, therefore, longitudinally assessed BA.5-specific neutralization activity in a subset of 18 PWH and 28 controls who experienced breakthrough infections, that were likely caused by BA.1 or BA.2 based on local epidemiology [18]. One month post-third dose, and prior to SARS-CoV-2 infection, the median reciprocal dilution required for BA.5 neutralization was 20 (IQR BLOQ-20) in this subset, which was two-fold lower than that for BA.1 (median 40, IQR 20-80; P ¼ 0.0005), and 16-fold lower than that for wild-type (median 320, IQR 160-640; P < 0.0001) (Fig. 4a). Although neutralization activity against all three virus strains rose significantly post-infection, (all P < 0.0001; Fig. 4b), BA.5-specific neutralization (median 160, IQR 80-200) remained lower than that against BA.1 (median 160, IQR 80-320; Fig. 3. Wild-type-specific and Omicron BA.1-specific live virus neutralization activity following three-dose coronavirus disease 2019 vaccination. Panel a: same as Fig. 1a, but for live virus neutralization activity, defined as the highest reciprocal plasma dilution at which neutralization was observed in all wells of a triplicate assay. Serial two-fold dilutions of 1/20 (lower limit of quantification; LLOQ) to 1/2560 (upper limit of quantification; ULOQ) were tested. Plasma samples showing neutralization in fewer than three wells at a 1/20 dilution are displayed as a reciprocal dilution of '10' and were reported as below limit of quantification (BLOQ) in the text. Omicron BA.1-specific neutralization was performed on a subset of samples only. Panel b: same data as panel a, but where neutralization activities are plotted longitudinally by participant (PWH in orange, CTRL in blue). The legend is the same as for Fig. 1b. Note that many data points are superimposed in both panels. P ¼ 0.007) and wild-type (median 640, IQR 320-2560; P < 0.0001) (Fig. 4a). Importantly, no significant differences were observed between PWH and controls in their ability to neutralize any virus strains after postvaccination SARS-CoV-2 infection (all P ! 0.4) (Fig. 4b, see P values above small brackets).

Discussion
Our results demonstrate that antibody response durability following three-dose COVID-19 vaccination in COVID-19-naive PWH receiving suppressive ART is comparable to controls without HIV. As we reported previously [14,17], initial peak antibody responses after three-dose vaccination were similar between PWH and controls, and our new data additionally show that wildtype-specific and Omicron-specific IgG concentrations, ACE2 displacement and virus neutralization activities declined at similar rates among PWH and controls who remained COVID-19-naive throughout follow-up. Multivariable analyses adjusting for sociodemographic, health and vaccine-related variables confirmed that there was no significant impact of HIV infection on any antibody outcome measure. Nevertheless, by 6 months post-third dose, all antibody responses in COVID-19naive participants, regardless of HIV status, had declined to similar or lower levels than peak responses observed after two vaccine doses. In fact, as early as 3 months postthird dose, BA.1-specific neutralization was already below the limit of quantification in $75% of COVID-19-naive participants, regardless of HIV status. Moreover, and consistent with recent reports [27][28][29], the ability to neutralize BA.5 was even poorer than BA.1.
By contrast, almost all participants who experienced their first SARS-CoV-2 infection (presumably BA.1 or BA.2 [18]) post-third dose displayed enhanced antibody responses to all viral variants tested. In fact, at 6 months post-third dose, all antibody measures in breakthrough infection cases were on average higher than those elicited by vaccination alone. Importantly, the magnitude of these 'hybrid' responses was equally strong in PWH and controls. Nevertheless, it was notable that responses to BA.5 remained significantly lower than those against wild-type and BA.1 even after infection, and that a minority of participants failed to show improved antibody responses post-infection, a phenomenon which requires further study.
Our study has several limitations. Our observations may not be generalizable to PWH with low CD4 þ T-cell counts and/or who are not receiving suppressive ART. Indeed, results from several studies indicate that PWH with CD4 þ T-cell counts below 500 cells/ml mount weaker responses to the first [30][31][32] and second [31][32][33][34][35][36][37][38] doses of COVID-19 vaccine. We did not assess cellular immunity in participants. We were unable to compare the duration or severity of breakthrough SARS-CoV-2 infections among PWH and controls as we did not systematically collect this information. Given recent reports that post-third dose breakthrough infections may be more frequent in PWH [39], future studies should address this. Finally, any recommendations based on our results are limited by the ongoing emergence of new SARS-CoV-2 variants, including strains that may be more immune-evasive than BA.5 (e.g. BA.2.75) [40][41][42][43][44].
In conclusion, our observations confirm the humoral immune benefits of third COVID-19 vaccine doses in PWH receiving suppressive ART, and further reveal that the durability of third-dose responses is comparable to that in persons without HIV. Nevertheless, regardless of HIV status, individuals who remain COVID-19-naive will benefit from a fourth dose within 3-6 months of their third dose, as antibody concentrations and neutralization activities declined markedly over time in this group, and the ability of vaccine-induced responses to neutralize the dominant Omicron BA.5 variant were even poorer than BA.1. By contrast, the majority of individuals who experienced their first SARS-CoV-2 infection post-third vaccine dose showed significantly higher antibody activities than those induced by vaccination alone (though anti-BA.5 responses remained weaker than anti-BA.1 responses, even after breakthrough infection). These observations suggest that a slightly delayed fourth dose (e.g. to 3-6 months following infection) would optimally benefit this group. Further studies of hybrid response durability are required, as are direct comparisons with immune responses elicited by a fourth dose in COVID-19-naive individuals, particularly in light of new bivalent formulations that include wildtype and Omicron Spike antigens.

Conflicts of interest
There are no conflicts of interest.