Safety, Effectiveness and Immunogenicity of heterologous mRNA-1273 Boost after Prime with Ad26.COV2.S among Healthcare Workers in South Africa: the single-arm, open-label, Phase 3 SHERPA Study

Study design and setting

SHERPA (Sisonke Heterologous mRNA-1273 boost after Prime with Ad26.COV2.S) was a single-arm, open-label, phase 3 study, nested in the Sisonke trial, in which healthcare workers in South Africa had received one or two doses of Ad26.COV2.S between February and December 2021. In SHERPA, Sisonke participants were offered a mRNA-1273 booster dose (ancestral strain) at 30 clinical research sites (CRS) across South Africa, between 23 May 2022 and 12 November 2022. Using routinely collected national data, we aimed to compare SARS-CoV-2 infections and COVID-19-related hospitalizations and deaths among SHERPA participants with Sisonke participants who did not receive the mRNA-1273 booster (see study schema in Supplementary Table 1). The data cut-off for outcome ascertainment was 31 December 2022. Seven CRS followed 200 SHERPA participants for six months in a safety and immunogenicity sub-study. SHERPA was registered in the Pan African Clinical Trials Registry (PACTR202310615330649) and the South African National Clinical Trial Registry (DOH-27-052022-5778). All participants provided written informed consent.

Participants

Sisonke participants were invited to join SHERPA via short messaging system (SMS) with a study website link, where they could register and choose a convenient CRS. SHERPA participants were 18 years or older with or without comorbidities, including HIV infection, were employed in the South African public or private healthcare sector, and had received either one or two doses of Ad26.COV2.S in Sisonke. Considering the high COVID-19 morbidity among pregnant women, pregnant and breastfeeding women were eligible. Sisonke participants who had received any COVID-19 vaccine other than one or two doses of Ad26.COV2.S (e.g., another mRNA booster) and those with evidence of SARS-CoV-2 infection within 14 days of enrolment were excluded (see Supplementary Table 2 for full eligibility criteria).

Study procedures

Once consented, the study team took a medical history, conducted a physical examination, and recorded any concomitant medications. Female participants of reproductive age had a pregnancy test, and all participants had a nasal swab taken for SARS-CoV-2 polymerase chain reaction (PCR) testing and a plasma sample to test for SARS-CoV-2 serology before vaccination. Participants in the safety and immunogenicity sub-study also had peripheral blood mononuclear cells (PBMC) and plasma samples taken before mRNA-1273 booster vaccination and 4 and 24 weeks after vaccination (see Supplementary Tables 3 and 4).

The 50mcg dose of the mRNA-1273 booster was prepared by research pharmacists. Clinical research staff then administered the vaccine in the deltoid muscle, and participants were monitored for 15 minutes for immediate reactogenicity events or allergic reactions. Adverse events (AE) experienced at the vaccination visit were managed and recorded by the CRS team. After the observation period, participants were issued with vaccination certificates, and CRS and Sisonke ‘Safety Desk’ contact details for AE reporting. Pregnant and breastfeeding women were followed up by the central safety team, through a pregnancy and breastfeeding registry that both the CRS and safety teams had access to.

Immunogenicity sub-study

Plasma and PBMC samples were taken before mRNA-1273 vaccination and 4 and 24 weeks after vaccination. We conducted SARS-CoV-2 spike enzyme linked immunosorbent assays (ELISA), SARS-CoV-2 nucleocapsid ELISA testing, a lentiviral pseudovirus neutralization assay, and an antibody-dependent cellular cytotoxicity (ADCC) assay at all timepoints. Neutralizing antibody responses in people living with HIV (PLWH) were tested using a vesicular stomatitis virus-based neutralization assay which was not subject to antiretroviral therapy interference⁷, and which produced equivalent titers to the lentiviral assay (Supplementary Fig.1). T-cell responses to SARS-CoV-2 spike were measured from cryopreserved PBMC using intracellular cytokine staining and flow cytometry. This was performed on the first 50 participants with available PBMC samples at the three sampling visits (see Supplementary materials for immunogenicity assays).

Data sources and data management

We obtained SARS-CoV-2 and severity endpoint data through the surveillance system that was built for Sisonke (Supplementary Fig.2). Data sources included the Notifiable Medical Conditions Sentinel Surveillance (NMCSS) database of SARS-CoV-2 cases, the DATCOV database of reported confirmed COVID-19 hospital admissions, and the national Electronic Vaccination Data System (EVDS) for COVID-19 vaccines. Automated near real-time deterministic and probabilistic linkage algorithms were applied to link these three routine COVID-19 datasets. Breakthrough SARS-CoV-2 infections, defined as a positive SARS-CoV-2 PCR or antigen test _≥14 days after booster vaccination, and COVID-19 hospitalisations or deaths were identified through linking the databases.

SHERPA study visits were recorded in a Research Electronic Data Capture (REDCap) database (Vanderbilt University, Nashville, US). After the screening/enrolment visit, solicited and unsolicited AEs were self-reported via an online data entry link shared by SMS with participants 1, 7 and 28 days after the booster. All safety events including Serious AEs (SAE) and AEs of Special Interest (AESI) were reviewed by safety clinicians and the Protocol Safety Review Team (PSRT). Data management systems were CFR Part 11 and POPIA compliant and processes followed Good Clinical Data Management Processes guidance.

Variables

For the relative vaccine effectiveness (rVE) analysis, we used demographic and clinical variables that are routinely collected in the EVDS, including age, sex, common comorbidities, and geographic location. The NMCSS database provided information about previous SARS-CoV-2 infections.

Outcomes

The co-primary outcomes were to assess the rVE of the mRNA-1273 booster by comparing the rates of SARS-CoV-2 infections and severe COVID-19 cases (defined as COVID-19 hospitalizations or deaths) among SHERPA participants compared to unboosted Sisonke participants. Secondary outcomes included an assessment of rVE of the mRNA-1273 booster against SARS-CoV-2 Infections, COVID-19 hospitalizations and deaths, in those who had received one or two doses of Ad26.COV2.S separately; an assessment of safety of the mRNA-1273 boost, including among pregnant women and PLWH; and comparisons of antibody and T-cell responses before and after vaccination.

Statistical analysis

Analyses for the primary and secondary outcomes were performed using SAS version 9.4 (Statistical Analysis Software, North Carolina, USA) and Stata version 17 (College Station, TX, USA). Demographic and clinical data of participants enrolled in SHERPA and Sisonke were summarized using descriptive statistics including frequencies, proportions, and medians with interquartile ranges (IQR). For the primary analysis, we used a statistical method proposed by Fintzi and Follmann⁸ referred to here as an adjusted cohort approach, which allowed inclusion of SHERPA participants within the larger Sisonke cohort (Supplementary Fig.3).

We used a Cox regression model with the mRNA-1273 booster status as a time-varying covariate, adjusting for age, sex, an indicator of whether a participant had previously received one or two doses of Ad26.COV2.S, geographic location, number of comorbidities and evidence of prior SARS-CoV-2 infection. This time-varying exposure was characterized by a continuous piecewise linear function of time elapsed since the first mRNA-1273 dose (23 May 2022) for the log hazard ratio. Participants who did not experience an event of interest were censored on 31 December 2022 (cut-off date). The rVE in reducing the risk of SARS-CoV-2 infection and COVID-19 hospitalizations and deaths was estimated as one minus the estimated hazard ratio with 95% CI. Further details regarding outcome analyses, including an alternative matched cohort analysis approach, definitions of variables, and immunogenicity analyses are detailed in the Statistical Analysis Plan.

Study approvals

The SHERPA and Sisonke trial protocols were approved by the South African Health Products Regulatory Authority (SAHPRA), and health research ethics committees of all participating CRSs. The study received full ethical approval by the following ethics committees in South Africa: Pharma-Ethics, Stellenbosch University Health Research Ethics Committee, University of KwaZulu-Natal Biomedical Research Ethics Committee, University of Cape Town Human Research Ethics Committee, Sefako Makgatho University Research Ethics Committee, The South African Medical Research Council Human Research Ethics Committee and the University of the Witwatersrand Human Research Ethics Committee.

Study Population

Of 501230 Sisonke participants who had received one or two Ad26.COV2.S doses between 17 February 2021 and 22 May 2022, 12342 enrolled into SHERPA and received the mRNA-1273 booster between 23 May 2022 and 12 November 2022 (Fig.1). Of these, 1094 (8.9%) were excluded from the rVE analysis, because they had either received a third Ad26.COV2.S or a BNT162b2 mRNA booster, or had received an Ad26.COV2.S vaccine after the earliest date of mRNA-1273 vaccination on 23 May 2022. Participants at one CRS had to be excluded from rVE and safety analyses due to concerns about data integrity. The mRNA-1273 booster rVE analysis included 11248 SHERPA participants, of whom 5102 (45.4%) had received one and 6146 (54.6%) two prior Ad26.COV2.S doses. These were compared to 413161 non-SHERPA Sisonke participants, of whom 185676 (44.9%) had received one and 227485 (55.1%) two doses of Ad26.COV2.S.

• Download figure
• Open in new tab

Figure 1: Flow Chart of the SHERPA Analysis Cohort nested in the Sisonke Implementation Trial

Baseline demographic and clinical characteristics of the SHERPA and non-SHERPA participants stratified by number of previous Ad26.COV2.S vaccinations are summarized in Table 1. The majority (79.3%) of SHERPA participants were female and median age was 41 years (IQR 35–48). One third (32.2%) of participants reported at least one comorbidity, most commonly HIV infection (18.7%), hypertension (12.9%) and diabetes (4.6%). Participants were recruited from all nine South African provinces. Non-SHERPA participants had similar characteristics, apart from fewer reporting HIV infection (8.3%).

Prior SARS-CoV-2 infections and time since last vaccination

Participants were checked against the NMCSS database for evidence of previous SARS-CoV-2 infection. Among SHERPA participants 1079 (9.2%) had evidence of a previous SARS-CoV-2 infection compared to 54770 (13.3%) among the non-SHERPA group. Overall, these estimates were likely an underestimate considering higher self-report rates (28%) and evidence of high prevalence of baseline seropositivity in the immunogenicity sub-study. The median days between the last Ad26.COV2.S dose and 23 May 2022 (SHERPA start) was 188 days (IQR 171-419) among SHERPA participants and 189 days (IQR 172-406) among the non-SHERPA group (Table 2).

SARS-CoV-2 infection and severe COVID-19 endpoints

As of 31 December 2022, after 3056 person-years of follow up, there were 13 recorded SARS-CoV-2 infections among SHERPA participants at median 125 days (IQR 90-154) after mRNA-1273 boosting. In contrast, 3352 infections were recorded among the non-SHERPA group after 251104 years of follow-up (Supplementary Table 5). There were 148 COVID-19 hospitalization or deaths among the non-SHERPA group compared to one among SHERPA participants. This immunocompromised PLWH, with a CD4 count of less than 200 cells/mm³ and high HIV viremia, died of COVID-19 pneumonia with respiratory distress syndrome, six months after receiving the mRNA-1273 booster.

Relative vaccine effectiveness of the mRNA-1273 booster

Comparing SHERPA participants to the non-SHERPA group, the unadjusted analysis showed a rVE of 59% (95%CI 29-76%) against SARS-CoV-2 infection, which did not change after adjusting for age, sex, prior SARS-CoV-2 infection, prior vaccination and geographic location (Table 3). The adjusted rVE of the booster among participants who had one prior Ad26.COV2.S dose was 77% (95%CI 9-94%) and 52% (95%CI 13-73%) for those who had two prior Ad26.COV2.S doses. In a separate model adjusting for HIV the rVE estimate only slightly changed to 57% (95%CI 25-75%). With only one severe COVID-19 case among SHERPA participants and few severe endpoints overall, the rVE estimate for COVID-19 hospitalizations and deaths lacked precision (Supplementary Table 6). In the alternative matched cohort analysis, the overall rVE against SARS-CoV-2 infection was slightly higher at 63% (95%CI 32-82%) (Supplementary Tables 7 and 8).

Safety profile of mRNA-1273 booster

Of 11798 SHERPA participants included in the safety analysis, 228 (1.9%) participants reported 575 reactogenicity events within 7 days after the mRNA-1273 booster. Of these, 230/575 (40%) were local reactogenicity events including 97/230 (42.2%) injection site pain, 69/230 (30%) swelling, 32/230 (13.9%) induration and 32/230 (13.9%) erythema events. A total of 345 systemic reactogenicity events were reported including 123 (35.7%) malaise and/or myalgia, 120 (34.8%) headache, 66 (19.1%) fever, 17 (4.9%) chills, 10 (2.9%) nausea, 5 (1.4%) vomiting and 4 (1.2%) arthralgia. All reported local and systemic reactogenicity were mild or moderate (grade 1 or 2) and all resolved without sequalae.

Unsolicited AEs collected for 28 days after vaccination and those meeting SAE/AESI criteria collected until the end of study participation were reported by 98 (0.8%) participants, reporting a total of 115 events. Of these events, 17 events met SAE criteria. Three of the 17 SAEs also met AESI criteria, the other 4 AESIs were grade 1 and 2 anosmia and ageusia, which resolved without sequalae. No related SAEs were reported during the study.

More reactogenicity events were reported by participants with prior SARS-CoV-2 infection (3.6% vs 1.8%), HIV-negative status (2.7% vs 1.2%) and those who received two compared to one prior Ad26.COV2.S dose (2.6% vs 2.0%). In a logistic regression model adjusting for age and sex, a prior SARS-CoV-2 infection (adjusted odds ratio [aOR] =2.03, 95%CI 1.59-2.59) and an HIV positive status (aOR=0.49, 95%CI 0.34-0.69) were associated with local/systemic reactions, but not having received two versus one Ad26.COV2.S dose before the mRNA-1273 booster (aOR=1.26, 95%CI 0.99-1.62), Supplementary Table 9.

Among 70 confirmed pregnancies, there were 39 (55.7%) full-term births, 4 (5.7%) preterm births (<37 weeks gestation), 11 (15.7%) early miscarriages (<12 weeks gestation), 7 (10.0%) stillbirths (>20 weeks gestation), and one (1.4%) ectopic pregnancy. Six (8.6%) women opted for an elective termination of pregnancy and two (2.9%) were lost to follow-up. All adverse outcomes, including miscarriages, preterm births and stillbirths were evaluated and deemed unrelated to vaccination by CRS investigators and the PSRT. Preterm births were reported in women with hypertension in pregnancy and multiple gestations, while the stillbirths (experienced by women aged 32 to 44 years) were evaluated as associated with abruptio placenta, postdate, cord around the fetal neck, and complicated home delivery (Supplementary Table 10).

Immune responses after mRNA-1273 booster

Baseline neutralizing antibody titers were 2-3 fold higher in participants who had received 2 doses versus 1 dose of Ad26.COV2.S (median: 1311 vs 3269 for D614G and 508 vs 1281 for BA.4) (Fig. 2A,B). The mRNA-1273 boost resulted in increased neutralizing antibodies titers for D614G after 4 weeks and then a decline by 24 weeks among those with one and two doses of Ad26.COV2.S (median baseline, 4-week and 24-week titers of 1311, 7033 and 2220 for the 1-dose group and 3269, 17653, 5526 for the 2-dose group), Fig. 2A-C. The antibody response after boosting was less pronounced towards the BA.4 pseudovirus despite a significant increase in rVE against this variant of concern. Baseline spike-binding antibody titers were comparable regardless of the number of previous Ad26.COV2.S doses. Binding titers significantly increased up to 6-fold 4 weeks post-boost (Fig. 2D,E), while ADCC had smaller but significant increases post-boost (Fig. 2F,G). The binding and ADCC titers were lower 6 months after the boost, with similar titers observed between both Ad26.COV2.S groups (Fig. 2D-G). At baseline, PLWH had up to 3-fold lower binding antibody titers and up to 2-fold lower neutralizing activity against D614G compared to HIV negative participants, irrespective of prior Ad26.COV2.S doses, while ADCC levels were similar (Supplementary Fig.4). Antibody functions in PLWH followed similar trends after the booster as HIV-negative participants.

• Download figure
• Open in new tab

Figure 2: Antibody Response elicited after mRNA-1273 heterologous Boost

Legend: Plasma from a subset of participants were assessed to determine the antibody response elicited upon heterologous boost (mJnJ) after 1 Ad26.COV2.S dose (JnJ x1; n=19) or 2 Ad26.COV2.S doses (JnJ x2; n=29). The neutralization activity was measured using a SARS-CoV-2 pseudovirus-based neutralization assay with (panel A) D614G, (panel B) Omicron BA.4 and (panel C) D614G in people living with HIV (PLWH). The plasma neutralization titer is measured as an ID₅₀. Black horizontal bars represent medians. The threshold of detection for the neutralization assay is an ID₅₀ of 20. Antibody binding was measured using an in-house SARS-CoV-2 assay using the D614G full spike protein (panel D and E). An EC50 was used to measure the binding titers of the samples. ADCC activity was measured by detecting the crosslinking ability of the antibodies present in the serum (panel F and G). Relative light units were measured which correlate with the levels of Fc_γRIIIa signalling. For all assays, statistical significance was measured with the Kruskal-Wallis test with Dunn’s multiple comparisons test. Significance is shown as: *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. Medians and fold changes are depicted under each graph. Samples were run in duplicate for all assays. Nucleocapsid ELISA was used to identify prior SARS-CoV-2 infections. Only individuals who were seropositive at baseline were included in the analysis as there was only 1 nucleocapsid-seronegative participants in the JnJ x1 group.

We examined T-cell responses in a subset of 50 participants (Fig. 3A). At baseline, spike-specific CD4+ T-cell responses were detected in 96% of participants (Fig. 3B). Four weeks after mRNA-1273 boosting, the frequency increased 2-fold (median: 0.074% vs 0.169%, p<0.0001), followed by contraction to a median of 0.113% by 24 weeks, which was not significantly different from baseline. Spike-specific CD8+ T-cell responses occurred in 52% of participants at baseline, and followed a similar trajectory as CD4 responses, increasing significantly at 4 weeks (median: 0.034% vs 0.057%, p=0.0029; Fig. 3C) and waning to 0.047% by 24 weeks. When examining the effect of prior vaccination with Ad26.COV2.S (Fig. 3D), there were significantly higher CD4+ responses at baseline in those who received two compared to one prior dose, which persisted at all timepoints after mRNA-1273 boosting. In contrast, the magnitude of CD8+ responses was similar between those who received one or two Ad26.COV2.S doses. HIV infection had no negative impact on vaccine booster responses at 4 or 24 weeks (Fig. 3E). When examining T-cell cross-reactivity to Omicron sub-lineages, we observed a high degree of preservation of the T-cell responses to Omicron BA.1, BA.4/5 and XBB.1 (Fig. 3F-G).

• Download figure
• Open in new tab

Figure 3: T-cell Response elicited after mRNA-1273 heterologous Boost

Legend: (A) Clinical characteristics of the T-cell sub-study cohort. Frequency of spike-specific CD4+ (B) and CD8+ (C) T-cells producing any of the measured cytokines (IFN-γ, IL-2 or TNF-α) in response to stimulation with a peptide pool spanning ancestral spike protein at baseline (BL), week 4 (W4) and week 24 (W24), in participants vaccinated with one or two prior doses of Ad26.COV2.S and an mRNA-1273 booster. The median response and proportion of participants with a spike-specific CD4+ or CD8+ T-cell responses are indicated below the graphs. (D) Spike-specific CD4+ and CD8+ T-cell cytokine responses stratified according to number of Ad26.COV2.S doses received. (E) Spike CD4+ and CD8+ T-cells in HIV-uninfected participants (pink) or PLWH (red). (F) Spike-specific CD4+ T-cell responses to spike from ancestral virus and Omicron sub-lineages BA.1, BA.4/5 and XBB.1. (G) Fold change of Omicron sub-lineage responses compared to ancestral spike. Bars represent the medians. Differences between responses at different time points post boost were assessed using a two-tailed Wilcoxon paired test. Differences between participants vaccinated with one or two doses of Ad26.COV2.S and responses to different SARS-CoV-2 variants were calculated using a Mann-Whitney U test. The number of participants included in each analysis is indicated on the graphs.