Immune imprinting and vaccine interval determine antibody responses to monovalent XBB.1.5 COVID-19 vaccination

Categories: Disease & Virus

May 18, 2025

Cohort and study design

Forty-nine participants were initially recruited; however, serum samples were analyzed for anti-nucleocapsid seropositivity by enzyme-linked immunosorbent assay (ELISA) to screen for recent prior infection. Nine seropositive participants were excluded from the final cohort. Within the final cohort of 40 participants (Table 1), 30 received three or four doses of monovalent Wuhan-Hu-1 followed by one dose of bivalent Wuhan-Hu-1/BA.4-5 vaccine (bivalent recipients), and 10 received three or four doses of monovalent Wuhan-Hu-1 vaccine (bivalent non-recipients) (Fig. 1a). At time of recruitment, all participants received the XBB.1.5 monovalent mRNA vaccine manufactured by Moderna Therapeutics. Suspected and rapid antigen-confirmed infection history was collected from participants. Serum samples were collected from each participant prior to XBB.1.5 vaccination and again at an average of 22 days post-vaccination. These were tested for 50% effective antibody IgG, IgA, IgM, and total IgG/A/M isotype concentrations (EC50) by ELISA against the ancestral spike RBD as well as 50% live SARS-CoV-2 neutralizing titers by focus reduction neutralization tests (FRNT50) against the ancestral WA1, vaccine-matched XBB.1.5, and emergent EG.5.1 and JN.1 strains.

Table 1 Final cohort demographics

**Fig. 1: XBB.1.5 monovalent vaccination boosts antibody responses.**

XBB.1.5 monovalent vaccination boosts antibody titers

In both bivalent recipient and non-recipient groups, the XBB.1.5 monovalent vaccine successfully boosts neutralizing titers against XBB.1.5 (recipient p < 0.0001, non-recipient p = 0.0008), EG.5.1 (recipient p < 0.0001, non-recipient p < 0.0001), and JN.1 (recipient p < 0.0001, non-recipient p = 0.0013) and back-boosts titers against the ancestral WA1 strain (recipient p < 0.0001, non-recipient p = 0.0490). Of note, the JN.1 variant, from which currently dominant strains emerged, demonstrates significant escape from vaccinated sera with neutralizing titers that are 4-fold lower than against the vaccine-matched XBB.1.5, regardless of vaccine history (recipient p = 0.0002, non-recipient p = 0.0490) (Fig. 1b, c). In both groups, the XBB.1.5 vaccine similarly back-boosts IgG (recipient p = 0.0003, non-recipient p = 0.0199), IgA (recipient p = 0.0056, non-recipient p = 0.0286), and total IgG/A/M (recipient p = 0.0127, non-recipient p = 0.0123) isotype antibodies targeting the ancestral spike RBD, but does not significantly boost RBD-specific IgM (recipient p = 0.1457, non-recipient p = 0.0731) (Figs. 1d, e).

Bivalent non-recipients exhibit diminished neutralizing breadth and potency prior to XBB.1.5 vaccination

To visually assess breadth of neutralizing antibody responses, neutralizing titers against the XBB.1.5, EG.5.1, and JN.1 variants were plotted against ancestral WA1 titers. 95% confidence ellipses were drawn to delineate the distribution of each participant group. Antibodies in both groups demonstrate a strong preference toward neutralizing WA1 over variants at pre-vaccination, but these cross-neutralizing responses are improved by XBB.1.5 vaccination as denoted by the upward shift toward increased variant-specific titers (Fig. 2a–f).

**Fig. 2: Bivalent non-recipients exhibit diminished variant cross-neutralization and neutralizing potency at pre-vaccination.**

In comparing groups at the pre-vaccination timepoint, bivalent non-recipients have a slight skew toward WA1 neutralization relative to recipients. To quantitate these visual observations, relative neutralization ratios were calculated by dividing the neutralizing titer against a given variant by WA1 titers. At pre-vaccination, bivalent non-recipients have significantly lower relative neutralization ratios for XBB.1.5 (p = 0.0305) and EG.5.1 (p = 0.0111) compared to recipients, possibly due to absence of prior vaccine exposure to the BA.4/5 spike variant which would enhance Omicron cross-neutralizing responses. The trend for JN.1 is similar but not statistically significant (p = 0.1935) (Figs. 2a–c). However, after XBB.1.5 vaccination, the bivalent non-recipient group clusters to the upper, right distribution of the total cohort, suggesting a trend toward improved cross-neutralization in addition to overall potency of neutralization. However, this enhancement in cross-neutralization is not significant upon comparing relative neutralization ratios (Figs. 2d–f).

Antigenic cartography was performed to further evaluate variant cross-neutralization. Here, neutralizing titer values against each SARS-CoV-2 strain for each participant serum sample were used to generate antigenic maps. One map was generated for each group (bivalent recipient and non-recipient) and at each time point (pre- and post-vaccination). One square-length map distance represents a ~2-fold change in FRNT50 neutralizing titer and is a measure of antigenic difference. Across all maps, XBB.1.5 and EG.5.1 are positioned closest together and appear most antigenically similar from the perspective of serum titers; this is consistent with the fact that EG.5.1 is a closely related descendent of the XBB.1.5 lineage while WA1 and JN.1 are phylogenetically more distant²⁴. In both groups, map distances between WA1 and variants are shorter at post-vaccination compared to pre-vaccination, demonstrating successful boosting of cross-neutralizing responses by the XBB.1.5 monovalent vaccine (Fig. 2g).

Unsurprisingly, bivalent non-recipients exhibit longer map distances than recipients at pre-vaccination between WA1 and XBB.1.5 (6.4 versus 5.2), EG.5.1 (7.2 versus 5.6), and JN.1 (6.8 versus 5.8), suggesting diminished cross-neutralization from lack of prior vaccination with the BA.4/5 spike variant. At post-vaccination, however, bivalent non-recipients have slightly shorter but overall similar map distances between WA1 and XBB.1.5 (3.6 versus 3.7), EG.5.1 (3.6 versus 3.9), and JN.1 (4.9 versus 5.0) (Fig. 2g). These analyses suggest that despite reduced pre-vaccination cross-neutralization due to lack of the bivalent vaccine, non-recipients recover comparable cross-neutralization to bivalent recipients upon XBB.1.5 vaccination.

Finally, neutralizing potency of antibody responses at pre-vaccination was measured by assessing the neutralizing potency index (NPI), which is calculated by dividing the serum neutralizing titer for a given variant by the total IgG/A/M titer. Biologically, each variant-specific NPI quantifies the subset of spike RBD-binding antibodies that lead to neutralization of infection by the assessed variant. The neutralizing potency of serum antibodies are significantly lower against XBB.1.5 (p = 0.0432), EG.5.1 (p = 0.0195), and JN.1 (p = 0.0371) but not WA1 (p = 0.4560) in bivalent non-recipients prior to XBB.1.5 vaccination. As with significantly diminished variant cross-neutralization, reduced neutralizing potency is likely the result of lacking exposure to the BA.4/5 spike variant and highlights the importance of bivalent vaccination in enhancing Omicron-targeting responses (Fig. 2h).

XBB.1.5 vaccine elicits greater boosting of variant-neutralizing antibodies in bivalent non-recipients

Pre-vaccination neutralizing titers against the contemporary variants trend lower in bivalent non-recipients against XBB.1.5 (1.6-fold; p = 0.3107), EG.5.1 (2-fold; p = 0.1040), and JN.1 (1.4-fold; p = 0.3792), consistent with greater waning of antibody responses in this group due to the missing bivalent vaccine and longer time interval since prior vaccination. Strikingly, however, after final vaccination, trends toward higher absolute titers are observed in the bivalent non-recipient group against WA1 (1.7-fold; p = 0.1158), XBB.1.5 (1.9-fold; p = 0.3188), EG.5.1 (2.2-fold; p = 0.1773), JN.1 (1.8-fold; p = 0.1500), suggesting an anchoring effect due to immune imprinting in bivalent recipients (Figs. 1b, 1c). Although these differences in post-vaccination absolute titers between bivalent recipients and non-recipients do not reach statistical significance, boosting of variant-specific neutralizing antibodies (i.e. fold-change after vaccination) is significantly greater in non-recipients against XBB.1.5 (18.4X versus 6.2X, p = 0.0134), EG.5.1 (30.9X versus 7.0X, p = 0.0027), and JN.1 (9.2X versus 3.7X, p = 0.0389). However, no significant difference is observed against WA1 (2.7X versus 2.2X, p = 0.5910) (Fig. 3a–d).

**Fig. 3: Bivalent non-recipients exhibit stronger boosting of variant-neutralizing and spike RBD-binding isotype titers.**

The bivalent recipient group was further stratified based upon having received either three or four doses of ancestral monovalent vaccine prior to the bivalent vaccine. Using an ANOVA follow-up test called “test for linear trend” (a.k.a. “test for linear contrast,” “post-test for trend”), which statistically assesses systematic decreases in left-to-right column order, these recipient subgroups were then compared to the bivalent non-recipient group in order to assess the effect of ancestral spike vaccinations on XBB.1.5 vaccine immunogenicity. This reveals a generally inverse relationship between number of ancestral spike vaccinations and absolute post-titer that approaches significance for XBB.1.5 (p = 0.0803) and EG.5.1 (p = 0.0504) while boosting has similar and statistically significant relationships for XBB.1.5 (p = 0.0124) and EG.5.1 (p = 0.0030). Within the bivalent recipient group, however, having 3 or 4 doses of ancestral spike results in similar responses against JN.1 and thus lack of linear trend in terms of absolute post-titer (p = 0.1875) and boosting (p = 0.1219), suggesting that vaccine immunogenicity against JN.1 and strains emerging from this distinct lineage may suffer less from ancestral spike imprinting. While the number of ancestral vaccinations has an impact on absolute post-titers against WA1 (p = 0.0323), no difference is observed in boosting (p = 0.6491), likely due to relatively high baseline neutralizing antibody titers from repetitive vaccine- and infection-mediated back-boosting across all participant groups (Supplementary Fig. 1).

XBB.1.5 vaccine elicits greater boosting and absolute ancestral spike-binding isotype titers in bivalent non-recipients

Pre-vaccination titers trend higher for bivalent non-recipients despite longer interval since prior vaccination for IgG (1.7-fold; p = 0.0926), IgA (1.7-fold; p = 0.1283), and total IgG/A/M (2.2-fold; p = 0.0236) (Fig. 1d, e). Our assay was optimized to detect antibody isotypes targeting the ancestral spike RBD and, though unexpected, these higher pre-vaccination titers in the non-recipient group are validated by similarly elevated neutralizing titers against the ancestral WA1 strain (Fig. 1b, c, Supplementary Fig. 1a). These increased pre-vaccination responses against ancestral antigen in bivalent non-recipients may be explained by natural infection histories skewed toward older strains. In other words, non-recipients opted out of the bivalent vaccine and were thus more susceptible to infection by early or pre-Omicron lineages that circulated in 2022. Infection by these lineages would more likely maintain responses against older spike variants, whereas an infection history more dominated by post-XBB Omicron would more drastically draw responses away from ancestral antigen. However, it is important to note that both bivalent recipient and non-recipient groups have similar overall numbers of prior infection.

At post-vaccination, absolute titers are significantly higher in the bivalent non-recipient group for IgG (2.7-fold; p = 0.0098), IgA (3.6-fold; p = .0089), and total IgG/A/M (5.1-fold; p = 0.0004) (Fig. 1d, e). Furthermore, more profound boosting of isotypes is observed in bivalent non-recipients than in recipients for IgA (3.1X versus 1.5X, p = 0.0131) and total IgG/A/M (3.5X versus 1.6X, p = 0.0130), though weaker and not statistically significant for IgG (3.2X versus 2.0X, p = 0.1422) and indifferent for IgM (1.5X versus 1.4X, p = 0.8288) (Fig. 3e–h).

When stratifying study participants by number of ancestral Wuhan-Hu-1 doses received prior to bivalent vaccination and performing ANOVA follow-up column “tests for linear trend,” significant inverse relationships are found with post-vaccination titers for IgG (p = 0.0222), IgA (p = 0.0003), and total IgG/A/M (p = 0.0014) as well as for boosting of IgA (p = 0.0154). This relationship approaches statistical significance for boosting of total IgG/A/M (p = 0.0799). No relationship is observed for absolute IgM post-titer (p = 0.4371) nor boosting of IgG (p = 0.6057) and IgM (p = 0.9400) (Supplementary Fig. 2).

Bivalent receipt and vaccination interval are tightly linked variables

It has been previously demonstrated that longer intervals between exposures, whether through natural infection or vaccination, may result in enhanced potency and breadth of neutralizing antibody responses^18,19,20. Compared to bivalent recipients, the non-recipient group have a greater interval between the XBB.1.5 monovalent vaccine and prior vaccination (mean of 361.2 ± 48.3 days versus mean of 706.9 ± 58.9 days). It is therefore of interest to evaluate the contribution of these longer intervals to the increased boosting observed in bivalent non-recipients. However, despite being statistically significant, representative multiple predictors linear regression modeling on EG.5.1 titer fold-change reveals that bivalent receipt and vaccine interval are highly collinear variables within our cohort as evident in variance inflation factors >10 and R² > 0.9 between variables (Supplementary Fig. 3). In other words, those who opted to not receive the bivalent vaccine reliably had longer rest intervals between vaccines. This is an expected problem due to the inherently intertwined nature of these variables in real-world vaccinated communities and indicates that regression modeling within this cohort cannot be used to fully isolate the independent effect of vaccine intervals on antibody boosting. However, it is important to emphasize that ancestral titers are independently back-boosted within each participant group, providing evidence that immune imprinting persists in the context of XBB.1.5 monovalent vaccination regardless of the differences, including vaccine intervals, between the two groups and is likely remnant of original ancestral monovalent vaccines (Fig. 1b–e).

Source link