A longitudinal cohort of vaccinees with previous COVID-19 displayed improved SARS-CoV-2 neutralization compared with those with vaccination alone. Between December 2020 and March 2021, we recruited 10 individuals who experienced PCR-confirmed COVID-19 prior to vaccination and collected blood samples before and after a standard 2-dose BNT162b2 vaccine regimen (Table 1) and 20 age- and sex-matched individuals with no self-reported history of prior COVID-19 infection, verified by negative nucleocapsid ELISA, and collected blood samples before and after vaccination. We then measured and compared serum neutralizing titers in these two groups using a live-virus focus reduction neutralization test (FRNT) (Figure 1, A and B). Serum neutralizing titers increased in both groups before and after vaccination and were significantly higher among those with prior infection compared with those with vaccination alone for all strains tested, including ancestral strain of SARS-CoV-2 (WA1) and the early VOCs Alpha, Beta, and Gamma (Figure 1C). These results suggested that hybrid immunity from the combination of vaccination and natural infection may result in meaningfully improved neutralizing serum antibody titers.

Longitudinal cohort of previously infected vaccinees shows improved variant neutralization compared with vaccination alone. (A) Representative focus reduction neutralization test (FRNT) results showing wells infected with live SARS-CoV-2 with the addition of serially diluted serum, which were stained and counted. (B) Representative focus reduction neutralization curve showing the average neutralization of duplicates as a percentage of no-serum controls, fit to a dose-response curve to find the 50% neutralizing titer (FRNT50). (C) Live-virus FRNT50 measurements against original SARS-CoV-2 (WA1) and the Alpha, Beta, and Gamma variants before and after vaccination. (D) Time line depicting the prevalence of current and former variants of concern at the study location, i.e., Oregon, USA. (41) Vaccine-only participants are represented by red circles, and hybrid immune participants are represented by cyan squares. Error bars represent the geometric mean, with 95% CIs. P values in C show the result of Mann-Whitney U tests. All P values are 2 tailed, and P = 0.05 was considered significant. For C, n = 20 for the vaccine only group and n = 10 for the prior infection group.

A cross-sectional cohort of hybrid immune individuals, including both those with prior infection and those with vaccine breakthrough. To more comprehensively study our initial findings that suggested infection followed by vaccination elicited higher levels of SARS-CoV-2–specific antibodies compared with those with vaccination alone, we next expanded on our cohort by recruiting additional vaccinated persons with or without hybrid immunity due to previous COVID-19 (Table 1). This larger hybrid immune group included 23 individuals with PCR-confirmed infections prior to vaccination and 23 individuals with vaccine breakthrough infections, as both vaccination/infection histories have been shown to provide similar levels of serological immunity (11). To assure a more uniform comparison, sera were collected less than 60 days following vaccination or PCR-confirmed breakthrough infection. The participants with infection prior to vaccination had all contracted COVID-19 during the pre-VOC era and are thus believed to have been infected with ancestral SARS-CoV-2 variants, while breakthrough cohort participants were recruited after the emergence of the VOCs but prior to the Omicron era (Figure 1D). Using a subset of individuals for whom appropriate samples were available, viral sequences were obtained from 17 of 23 breakthrough participants, showing that the majority of infections were caused by the Alpha and Delta VOCs (Table 2).

Elevated antibody levels and neutralizing titers with hybrid immunity. We next measured spike-specific antibody levels in our larger cohort with a series of ELISA experiments. Against purified receptor binding domain (RBD) protein, total antigen-specific antibody levels were 3.6-fold greater in the group of individuals with hybrid immunity compared with those with vaccine only (Figure 2A). Class-specific ELISAs showed that this was primarily driven by increases in IgG levels, which increased 3.7-fold (Figure 2B), while the less abundant IgA improved by 3.2-fold (Figure 2C), and IgM levels showed no significant difference between groups (Figure 2D). Total antibody levels against the full-length spike protein, which includes the entire S1 and S2 domains, were also improved with hybrid immunity by 3.1-fold (Figure 2E).

Cross-sectional cohort of individuals with hybrid immunity, showing improved antibody levels and variant neutralization. Levels of SARS-CoV-2 spike receptor binding domain–specific (RBD-specific) (A) total (IgG/IgA/IgM) antibody, (B) IgG, (C) IgA, and (D) IgM. (E) Levels of full-length spike-specific total antibody. (F) Live-virus FRNT50 measurements against original SARS-CoV-2 (WA1) and the Alpha, Beta, Delta, and Omicron (BA.1 and BA.2) variants. Vaccine-only participants are represented by red circles, and hybrid immune participants are represented by blue squares. Error bars represent the geometric mean, with 95% CIs. P values show the result of Mann-Whitney U tests. All P values are 2 tailed, and P = 0.05 was considered significant. For all panels, n = 20 for the vaccine-only group and n = 46 for the hybrid immunity group.

Similarly, neutralizing antibody titers against SARS-CoV-2 and every SARS-CoV-2 variant tested rose significantly in the hybrid immune group compared with the vaccination-alone group (Figure 2F). Neutralizing titers increased by 8.4-fold against WA1, 12.5-fold against Alpha, 22.7-fold against Beta, 9.6-fold against Delta, 19.0-fold against Omicron BA.1, and 13.3-fold against Omicron BA.2. The largest fold increases were seen against the most vaccine-resistant variants, Beta and Omicron (BA.1 and BA.2). Furthermore, it appears that these increases were not restricted to variants with which the cohort was experienced, as all samples were collected prior to the emergence of Omicron.

Improved antibody quality among hybrid immune individuals. To assess the breadth of the neutralizing antibody response, we then looked at the relative ability to neutralize variants. This was measured by dividing the neutralizing titer for each variant by the neutralizing titer for WA1. The hybrid immunity cohort showed considerably greater cross-reactivity against Alpha and Beta variants compared with the vaccine-only cohort, where an appreciable deficit in cross-neutralization against Alpha and Beta were seen (Figure 3, A and B). In contrast, cross-reactivity against Delta was comparable in the two cohorts, where neutralization against Delta and WA1 were similar (Figure 3C). Cross-neutralization against Omicron BA.1 and BA.2 was substantially reduced in both cohorts, but it was less so in the hybrid immunity group, where high titers were associated with better cross-reactivity (Figure 3, D and E, and summarized in Figure 3F and Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.165265DS1).

Antibody quality and variant cross-neutralization are improved with hybrid immunity. Individual neutralizing FRNT50 values against WA1 versus (A) Alpha, (B) Beta, (C) Delta, (D) Omicron (BA.1), and (E) Omicron (BA.2). Diagonal broken lines indicate equal neutralization of WA1 and variant in A–D. (F) Relative neutralization, calculated as the neutralizing titer against each of the variants divided by the neutralizing titer against WA1. (G) Neutralizing potency index, indicating the neutralizing FRNT50 against the indicated variant divided by full-length spike protein EC50 antibody levels. Vaccine-only participants are represented by red circles, and hybrid immune participants are represented by blue squares. Error bars represent the geometric mean, with 95% CIs. P values in F and G show the result of Mann-Whitney U tests with the Holm-Šídák multiple comparison correction. All P values are 2 tailed, and P = 0.05 was considered significant. For all panels, n = 20 for the vaccine-only group and n = 46 for the hybrid immunity group.

To assess the potency of the neutralizing antibody responses, we calculated the neutralizing potency index (NPI) for the individuals in each cohort against each variant. The NPI is the neutralizing titer divided by the quantity of full-length spike-specific total antibody levels, as measured by ELISA. NPI scores indicate the efficiency with which antigen-specific antibodies neutralize virus on a per total antibody basis in which higher scores indicate that fewer antibodies are necessary to achieve a given neutralization titer. We found that the NPI of hybrid immune individuals increased significantly for all variants tested, with indexes of 2.7-fold (WA1), 4.0-fold (Alpha), 7.2-fold (Beta), 3.0-fold (Delta), 6.1-fold (Omicron BA.1), and 4.2-fold (Omicron BA.2), indicating a significant improvement in the neutralizing efficiency of the antibodies produced by hybrid immunity compared with those produced by vaccination alone (Figure 3G).

The interval between vaccination and natural infection dictates neutralizing titer levels. The hybrid immune cohort included individuals who developed COVID-19 between 40 and 404 days after vaccination as well as individuals who were vaccinated between 35 and 283 days after testing positive for COVID-19. This range of hybrid exposure intervals allowed us to determine the effect of time intervals on the resulting neutralizing antibody response. We also characterized the correlation among antibody levels and neutralizing titers with our demographic data on age, exposure interval, sex, and the time form last exposure to sample collection. Only neutralizing antibody titers and antibody levels were significantly correlated with exposure interval. The strongest correlations were seen for full-length spike-specific antibody level as well as neutralization of WA1, Alpha, Beta, Delta, Omicron BA.1, and Omicron BA.2 (Figure 4, A–G).

Exposure interval determines strength of hybrid immunity. Comparison of exposure interval, the time between first and last antigen exposure, with (A) full-length spike EC50 antibody levels and neutralization of (B) WA1, (C) Alpha, (D) Beta, (E) Delta, (F) Omicron (BA.1), and (G) Omicron (BA.2). (H) Neutralization of variants binned by exposure interval in days. (I) Heatmap of correlation significance between explanatory and response variables. Individual values in A–G are shown as filled circles, and the shaded area indicates the linear fit with 95% CI. R2 is indicated for each curve fit. P values in A–G show the result of an F test using a 0-slope null hypothesis. P values in H show the result of Mann-Whitney U tests with the Holm-Šídák multiple comparison correction. Colors in I represent the P values of Pearson’s r correlation coefficients according to the scale bar. All P values are 2 tailed, and P = 0.05 was considered significant. For A–G and I, n = 46. For H, n = 7 for the 35–100 days group, n = 10 for the 101–200 days group, n = 18 for the 201–300 days group, and n = 11 for the 301–404 days group.

The magnitude of increase seen over time was also different for each of the variants. Using linear regression, we found the neutralizing titer against WA1 increased 5.3-fold by day 400 (Figure 4). This increase was 4.8-fold for Alpha, 11.5-fold for Beta, 11.2-fold for Delta, 17.6-fold for Omicron BA.1, and 14.3-fold for Omicron BA.2. The largest increases were seen against the more contemporary variants, which also tend to be more vaccine resistant (Figure 2F). To validate that these trends were not an artifact of linear regression, we also subdivided the cohort into 100-day exposure interval bins, which recapitulated the previous findings (Figure 4H). Steady increases were seen each 100 days, resulting in a final increase of 4.2-fold against WA1, 4.1-fold against Alpha, 9.6-fold against Beta, 7.1-fold against Delta, 12.5-fold against Omicron BA.1, and 10.7-fold against Omicron BA.2 between the 35- to 100-day and 300- to 404-day exposure interval groups. Both methods of analysis found that a large and significant improvement in neutralizing antibody titers occurred over an increased duration between antigen exposures provided by vaccination and natural infection. Furthermore, these correlations were maintained when measured separately for individuals with infection prior to vaccination and individuals with vaccine breakthrough infections (Supplemental Figures 2 and 3). Observed separately, neutralizing titers from individuals from the breakthrough group appeared to increase faster than those in the prior infection group, but no statistically significant difference could be measured. RBD-specific total antibody and IgG levels correlated less strongly, while RBD-specific IgA and IgM did not correlate significantly with exposure interval (Supplemental Figure 4).

We then assessed for interactions between exposure interval and other variables that could confound our analyses, including age, sex, or the time between final antigen exposure (either vaccination or COVID-19 infection) and serum sample collection, all of which have been previously shown to affect antibody levels (4, 23, 24). As expected, titers weakly correlated with age and sex but did not approached the relative contribution of exposure interval (Figure 4I). Collection interval was not significantly correlated with any variable, likely due to our strict 60-day limit on collection interval for inclusion in the study.

Variant cross-neutralization improves with greater exposure intervals. After observing the increases in variant cross-neutralization between the hybrid immunity and vaccine-only groups, we sought to determine whether there was an equivalent dependence on the exposure interval duration. Alpha is the least vaccine-resistant variant and did not improve relative to WA1 because it started at a ratio of 1 from the beginning (Figure 5A). For the more vaccine-resistant variants, which started well below 1, all saw increased variant cross-neutralization with increasing exposure interval (Figure 5, B–E). This indicates that the neutralizing antibody response is becoming more broadly neutralizing over time, between exposures. No significant trends were seen with NPI over time (Supplemental Figure 5). This indicates that while the variant cross-reactivity is increasing with longer exposure intervals, the proportion of antibodies that are capable of neutralization is maintained.

Exposure interval increases variant cross-neutralization by hybrid immune sera. Comparison of exposure interval, the time between first and last antigen exposure, with relative neutralization of (A) Alpha, (B) Beta, (C) Delta, (D) Omicron (BA.1), and (E) Omicron (BA.2) over WT (WA1). Individual values are shown as filled circles, and the shaded area indicates the linear fit with 95% CI. R2 is indicated for each curve fit. P values show the result of an F test using a 0-slope null hypothesis. All P values are 2 tailed, and P = 0.05 was considered significant. For all panels, n = 46.