Participants were binned into the following groups: 1) COVID-19 vaccination without history of COVID-19 disease, 2) COVID-exposure: history of COVID-19 in the pregnancy or seropositive based on SARS-CoV-2 nucleocapsid IgG titers (10), and 3) control: no history of COVID vaccination, or COVID disease (confirmed both by history and seronegative for SARS-CoV-2 nucleocapsid).

Electronic medical records were reviewed for demographic, medical, and obstetric history, date of first positive SARS-CoV-2 PCR test, date of delivery, antenatal complications, and delivery outcomes. Standard of care at our institution includes a universal SARS-CoV-2 PCR test on admission and a neonatal SARS-CoV-2 PCR test if the mother had a positive diagnosis on admission for delivery.

For assessing antibody specificity, a multiplex testing platform (MesoScale Diagnostics, Rockville, MD) was employed: antigens were manufactured in a mammalian expression system (Expi 293 F) are printed onto 10-plex plates. The antigens used were: HA-trimer from Influenza A (Hong Kong H3), spike (soluble ectodomain with T4 trimerization domain) trimers from SARS-CoV-2, SARS-CoV-1, MERS-CoV, and the betacoronaviruses HKU-1 and OC43, as well as the spike N-terminal domain (NTD, Q14-L303 of the SARS-CoV-2 spike sequence), receptor binding domain (RBD, R319-F541 of the SARS-CoV-2 spike sequence), and nucleocapsid protein (N; full length) for SARS-CoV-2, and bovine serum albumin (BSA, as negative control). Assays were performed following the manufacturer’s instructions. In brief, plates were blocked using Blocker A Solution and incubated at room temperature (RT) for 1h on a plate shaker at 700 rpm. The plates were washed three times with 1x MSD Wash Buffer. Sera were diluted to 1:1000 with Diluent 100. Positive samples (pooled human serum from COVID-19 patients) and negative samples (pooled pre-pandemic human serum) were used as controls. Plates were sealed and incubated at RT for 2h on a plate shaker at 700 rpm, then washed three times with 1x MSD Wash Buffer. The detection antibody, SULFO-TAG anti-human IgG or anti-human IgM antibody was diluted to 2 µg/ml in Diluent 100 (MSD), added to the wells, and incubated at RT for 1h on a plate shaker 800rpm. After washing, MSD GOLD Read Buffer B was added to each well and the plates were read immediately on the MESO QuickPlex SQ 120 (MSD).

This assay has been described previously (11). The Spike expression plasmid sequences for SARS-CoV-2 were codon optimized and modified to remove the last 18 amino acids of the cytoplasmic tail, which improves S incorporation into pseudovirions (PSV). PSV were produced by cotransfection of HEK293T/17 cells with either a SARS-CoV-2 S plasmid based on the Wuhan-Hu-1 genome sequence (GenBank accession number MN908947.3) and an HIV-1 pNL4-3 luciferase reporter plasmid (pNL4-3.Luc.R-E-, NIH HIV Reagent Program, catalog number 3418). S expression plasmids for SARS-CoV-2 VOCs were similarly codon optimized and modified and included the following mutations: B.1.617.2/Delta (E156G, D614G, P681R, T19R, T478K, L452R, D950N, 157-158 del), B.1.1.529/Omicron (A67V, 69-70 del, T95I, G142D, 143-145 del, N211I, 212 del, G339D, S371L, S373P, S375F, S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, Y505H, T547K, D614G, H655Y, N679K, P681H, N764K, D796Y, N856K, N969K, Q954H, L981F). Infectivity and neutralizing titers were determined using ACE2-expressing HEK293 target cells (Integral Molecular). Test sera were serially diluted, mixed with an equal volume of diluted PSV, and plates were incubated for 1 h at 37°C. Target cells were added to each well (40,000 cells per well), and plates were incubated for an additional 48 h. Relative light units were measured with the Synergy Neo2 Hybrid Multi-Mode Microplate Reader (Agilent BioTek) using the Bright-Glo Luciferase Assay System (Promega). Neutralizationdose–response curves were fitted by nonlinear regression using GraphPad Prism, and titers are reported as the reciprocal of the serum dilution necessary to achieve 50% inhibition of SARS-CoV-2 infectivity (ID50).

The primary outcome was neutralizing capacity, measured by the reciprocal of the serum dilution necessary to achieve 50% inhibition of SARS-CoV-2 infectivity (ID50) against wild-type, Delta, and Omicron-B1 between cohorts. Secondary outcomes include a comparison of cord-blood ID50 as well as efficiency of vertical transfer, measured by cord-blood:maternal blood ID50 for each variant, correlation between titers and variant-specific immunity, and factors associated with maternal ID50.

Statistical analysis was conducted using SPSS v. 26.0 (Chicago, SPSS, Inc) and R. Regarding baseline characteristics- categorical comparisons were done with chi-square analysis or Fisher’s exact test as appropriate and continuous variables were compared with one-way ANOVA. Serological titers and ID-50 values were log-transformed, and compared with ANCOVA, unadjusted, and adjusted taking into consideration potential covariates that were identified to be different between groups at p<0.05. Correlation between latency, titers, and ID50 was assessed with bivariate correlation and reported with Pearson correlation coefficient (r). Comparisons of paired maternal and cord-blood serology were carried out using linear regression to determine the slope, correlation coefficient, and R (2). P<0.05 is considered significant for all analyses. Pair-wise comparison of the difference in variant-specific immunity was done with the Wilcox Rank Sum test. Figures were generated using the SPSS, and stats, ggplot2, and corrplot packages in R.

There were 192 participants who had samples available for additional analyses. They delivered between February 2021-August 2021. This time period includes the Delta wave, which began in India in late 2020, emerged in the United States in April 2021, and dominated in the summer of 2021. This time period does not include the Omicron variant wave, which was first identified in South Africa in November 2021, and quickly spread around the world. Analyzing this sample set provides insights into how future variants from lineages divergent from the vaccine-included sequences will perform and make the results of this study relevant. There were 88 in the control group, 57 in the COVID-exposed group, and 47 in the COVID-vaccine group. Of 57 in the COVID-exposed group, there were 28 participants who did not have a documented COVID-19 infection but were seropositive. Maternal-cord-blood dyads at delivery were available for 114 participants, N=54 control, N=32 COVID-exposed, and N=28 COVID- vaccine. There were some significant baseline differences between groups, including race, maternal age and maternal BMI, and time from exposure ( Table 1 ). All of those in the vaccine group had received both doses of an mRNA-based COVID vaccine and were >7 days from the first vaccine dose. Of those in the COVID-exposed group, only 19 had a specific date recorded for their COVID-19 infection and the majority of these were >7 days from delivery ( Table 1 ).

COVID-19 vaccination was associated with higher mean full-length spike IgG titers compared to those with COVID-19 disease and controls. Those with COVID-19 exposure had lower titers compared to the vaccine group. Participants with COVID exposure had, as expected, higher nucleocapsid titers than controls and COVID vaccine recipients ( Figure 1 ).

Those with COVID-19 mRNA vaccination had higher ID50 against wildtype and Delta virus, and to a lesser degree, Omicron, compared to controls in unadjusted and adjusted analyses. In contrast, those with COVID-exposure did not have any improved immunity against Omicron compared to controls ( Figure 2 ). In pairwise comparisons for those who had received the COVID vaccine, there was progressively reduced ID-50 from wild-type (5.74 ± 1.45) to delta (4.30 ± 0.92) to Omicron (3.83 ± 0.60) (p<0.001 for all comparisons). For those with COVID exposure, there was progressively reduced ID-50 with Delta (3.91 ± 0.75) and Omicron (3.76 ± 0.60) compared to wild-type (4.7 ± 1.19) (p<0.001, p<001), and with Omicron compared to Delta (p=0.02).

Among those with COVID-19 exposure or vaccination, time from exposure (days) was negatively correlated with neutralizing capacity for both wild-type (r=-0.35, p=0.005) and Delta (-0.31, p=0.01). Maternal full-length spike IgG titers were positively correlated with neutralizing capacity against wild-type (r=0.74, p<0.001), Delta (r=0.56, p<0.001) and Omicron-B1 (r=0.30, p=0.003) variants ( Figure 3 ). The correlation matrices demonstrated the distinct relationships between antibody titers and neutralizing activities in the COVID vaccine vs. COVID disease cohorts: the COVID vaccine group showed a strong correlation between maternal spike titers and wild-type and Delta ID50 in maternal and cord-blood. Moreover, a weak correlation between maternal Omicron ID50 and maternal spike titers and wild-type and Delta neutralization capacity was observed ( Figure 3A ). In contrast, strong correlations in the COVID exposure cohort were only seen between maternal spike titers vs. wild-type and Delta neutralizing capacity. This correlation for antibody titer and neutralizing capacity in cord blood was strong for wildtype, but already diminished for Delta. Due to a lack of correlation with Omicron ID50, this analysis was removed from the correlation matrix in Figure 3B . In summary, the COVID vaccine cohort reveals more cross-reactivity against variant strains, including even some activity against Omicron. The latter was not observed in the COVID exposure cohort.

In a multivariable logistic regression analysis that takes into consideration race, age, BMI, latency, along with cohort sub-group, COVID-vaccination was positively associated with wild-type (B=1.3 (0.4 – 2.1), p=0.005), Delta (B=0.5 (0.1-1.1), p=0.047), and Omicron (B=0.4 (0.1-0.7), p=0.01) ID50 compared to COVID-exposed. These results are limited to those with time from exposure data available ( Table 1 ).

There was a positive association between maternal and cord-blood titers for both COVID vaccine and COVID-19-exposed cohorts ( Figure 3 ). Although overall neutralizing capacity decreased with increased time from exposure, the efficiency of transfer of neutralizing capacity increased with time from exposure ( Figure 4 ). In a linear regression model, both time from exposure and COVID vaccination vs disease were positively associated with increased vertical transfer efficiency for the wild-type and the Delta variant. There was a 0.20 (0.07-0.33, p=0.004) and 0.12 (0.0-0.24, p=0.05) increase in cord-blood:maternal neutralizing capacity with COVID vaccination compared to COVID-19 exposure for wild-type and Delta neutralizing capacity respectively. Finally, in comparing the efficiency of vertical transfer of immunity to variants, the Wilcox rank sum comparison found that there was reduced efficiency from wild-type to Delta to Omicron (p<0.001); and among the COVID-exposure cohort reduced efficiency from wild-type to Delta variants (Omicron not evaluated due to lack of significant immunity against Omicron in the COVID-exposed cohort) ( Figure 5 ).

The maternal and neonatal benefits of COVID vaccination are well documented in the form of both reduced maternal morbidity and mortality as well as reduced infant COVID-related morbidity and mortality in the setting of initial vaccination series and earlier variants (14, 15). A recent study on patients boosted in the third trimester receiving the bivalent booster demonstrated increased Omicron spike titers compared to primary vaccination (16) but did not compare Omicron neutralizing capacity with prior COVID exposure. An earlier study evaluated the efficacy of COVID vaccination against wild-type and Alpha and Beta variants and found reduced responses against the Alpha and Beta variants, but did not include a comparison to control or disease exposure (17), and provided no data for the Delta or Omicron variants.

Our work adds to the current body of knowledge by evaluating how well an initial vaccination series, or early COVID exposure, protects against current variants of concern. Due to the size and heterogeneity of our cohort, we were able to study individuals with a range of time from infection and adjust for differences in baseline characteristics. Consistent with earlier studies comparing the COVID vaccine and disease exposure and wild-type immunity (18), we found vaccination afforded improved immunity, although attenuated, against new variants as well. Interestingly, we found reduced neonatal immunity against progressive variants, reflecting the limitations of passively acquired cross-variant immunity. This highlights the need to promote the use of variant-specific booster immunizations in pregnancy if not just for maternal benefit, but also for neonatal protection.

Our results have several important implications for research. First, given the continued, global circulation of SARS-CoV-2 exposure, comparing vaccine-induced immunity to disease-induced immunity and the impact of vaccination and/or exposure on the immunological profile hold tremendous value. Second, our results highlight the limited utility of serological titers alone when analyzing immunity against newer variants of concern. Future research on SARS CoV-2 immunity should focus on neutralizing the capacity of immune serum and not just serological titers. For example, COVID-19 infections resulted in broad antibody response to SARS-CoV-2 spike protein and nucleocapsid, but, unlike vaccination, resulted in negligible Omicron neutralizing activity in both maternal and cord-blood sera. Third, we noted significantly higher neutralizing capacity against wild-type in cord-blood compared to maternal blood. The efficiency of vertical transfer of immunity is an important area for future research to optimize maternal immunization with the potential for neonatal benefit. This transfer, however, did not translate into equivalent protection against the Omicron variant, sub highlighting the distinct nature of the passive immunity acquired by neonates and the need for further research into whether, for example, the lack of IgM, or preference of transfer of specific IgG isotypes results in the difference in neutralizing capacity between maternal and cord-blood. This has important implications for the degree of functional neonatal immunity afforded by maternal vaccination, and optimal time periods of maternal vaccination for infant immunity. Fourth, the observed differences in the persistence of neutralizing antibody titers induced by vaccination vs SARS-CoV-2 exposure in maternal sera and the corresponding neutralizing activity in cord-blood strongly support the recommendation to vaccinate pregnant patients. Finally, the observation that 28% of control subjects were seropositive already by mid-2021 confirms the increasing endemicity of SARS-CoV-2 infections and the importance of taking existing immunity into consideration in future studies.

Our results have a number of important implications for clinical care. First and foremost, we highlight the importance of bivalent boosters during pregnancy; an initial vaccine series resulted in reduced maternal and no neonatal functional immunity against the recent variant of concern, i.e., Omicron. Poor bivalent booster uptake may be related to a false sense of security from the initial vaccine course or prior infection, and this data provides evidence to counter those assumptions. Second, our findings regarding the benefit of vaccination and latency in the vertical transfer of immunity, and the limited acquired cross-variant immunity, have important implications for the timing of maternal booster doses in pregnancy for the purpose of inducing infant immunity.