Longitudinal effects of SARS-CoV-2 breakthrough infection on imprinting of neutralizing antibody responses

Abstract

Background: The impact of the infecting SARS-CoV-2 variant of concern (VOC) and the vaccination status was determined on the magnitude, breadth, and durability of the neutralizing antibody (nAb) profile in a longitudinal multicentre cohort study.

Methods: 173 vaccinated and 56 non-vaccinated individuals were enrolled after SARS-CoV-2 Alpha, Delta, or Omicron infection and visited four times within 6 months and nAbs were measured for D614G, Alpha, Delta, BA.1, BA.2, BA.5, BQ.1.1, XBB.1.5 and JN.1.

Findings: Magnitude-breadth-analysis showed enhanced neutralization capacity in vaccinated individuals against multiple VOCs. Longitudinal analysis revealed sustained neutralization magnitude-breadth after antigenically distant Delta or Omicron breakthrough infection (BTI), with triple-vaccinated individuals showing significantly elevated titres and improved breadth. Antigenic mapping and antibody landscaping revealed initial boosting of vaccine-induced WT-specific responses after BTI, a shift in neutralization towards infecting VOCs at peak responses and an immune imprinted bias towards dominating WT immunity in the long-term. Despite that bias, machine-learning models confirmed a sustained shift of the immune-profiles following BTI.

Interpretation: In summary, our longitudinal analysis revealed delayed and short lived nAb shifts towards the infecting VOC, but an immune imprinted bias towards long-term vaccine induced immunity after BTI.

Funding: This work was funded by the Bavarian State Ministry of Science and the Arts for the CoVaKo study and the ForCovid project. The funders had no influence on the study design, data analysis or data interpretation.

Keywords: Antigenic map; COVID-19 breakthrough infection; Immune imprinting; Machine learning; Magnitude-breadth; Neutralization; SARS-CoV-2.

Fig. 1

Neutralization assays against different SARS-CoV-2 variants for vaccinated and unvaccinated Alpha, Delta and Omicron infected study groups. Neutralizing antibody responses of vaccinated (V, n = 171) and unvaccinated (U, n = 56) individuals at visit 1, 2, 4 and 5 after infection with Alpha (A, n = 32), Delta (D, n = 132) or Omicron (O, n = 63) (as indicated on the x-axis). Shown are the dilution factors for 50% Inhibitory concentration (IC50) for WT/D614G, Alpha, Delta, Omicron BA.1, Omicron BA.2 and Omicron BA.5 spike, as indicated by the individual graph titles. The assay cut-off (IC50 = 20) as well as saturation (IC50 = 2560) are indicated by dashed lines. IC50s were measured in three replicates. Median and interquartile ranges are indicated by the red horizontal line and whiskers.

Fig. 2

Magnitude-breadth analysis for all study groups at the different visits. Magnitude-breadth curves of the vaccinated & Alpha infected (VA, n = 24), unvaccinated Alpha infected (UA, n = 8), vaccinated & Delta infected (VD, n = 92), unvaccinated Delta infected (UD, n = 40), vaccinated & Omicron infected (VO, n = 55) and unvaccinated Omicron infected (UO, n = 8) groups at visit 1 (V1), visit 2 (V2), visit 4 (V4) and visit 5 (V5) (a–d). e Magnitude-breadth curves at V5 with the variant BQ.1.1 included in the analysis. f Calculated Areas under the curve (AUC) of the magnitude-breadth curves in a-d for each infection variant-vaccination status combination sorted by visits. Statistically significant differences are indicated by p values (∗p < 0.05, ∗∗p < 0.005, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001) [Kruskal–Wallis test and post-hoc tests adjusted by Dunn's correction for multiple tests].

Fig. 3

Antigenic maps for SARS-CoV-2 variants. Antigenic maps for four different study time points (V1, V2, V4 and V5) based on sera of vaccinated (n = 171) and unvaccinated (n = 56) participants with Alpha, Delta or Omicron (n = 32, 132, 63, respectively) infections. The SARS-CoV-2 variants are shown as big, coloured circles on the map Wuhan/D614G (red), Alpha (purple), Delta (blue), Omicron BA.1 (orange), BA.2 (yellow), BA.5 (green) and BQ.1.1 (grey) and respective individual sera are indicated as small squares (vaccinated) or small circles (unvaccinated) in purple, blue and orange for Alpha, Delta or Omicron infections, respectively. The y- and x-axes of the map both represent antigenic distance. Each grid square (1 antigenic unit) represents a 2-fold change in neutralization titre. Row one are maps including all sera, while rows two and three show only the sera of vaccinated or unvaccinated individuals, as indicated.

Fig. 4

Antibody landscapes for each study group. Coloured surfaces show the fitted geometric mean titre (GMT) antibody landscapes for the different study groups vaccinated & Alpha infected (dark purple, n = 24), unvaccinated Alpha infected (light purple, n = 8), vaccinated & Delta infected (dark blue, n = 92), unvaccinated Delta infected (light blue, n = 40), vaccinated & Omicron infected (dark yellow, n = 55) and unvaccinated Omicron infected (light yellow, n = 8). While the base x-y plane corresponds to the antigenic map shown in Fig. 3, grey impulses show the height of the GMT for specific SARS-CoV-2 variants. The vertical z-axis in each plot corresponds to the GMT on the log2 scale, each two-fold increment is marked, starting from a titre of 20 at the map surface. The antibody landscapes for visit 1 can be seen in the upper left panel, visit 2 in the upper right panel, visit 4 in the lower left panel and visit 5 in the lower right panel.

Fig. 5

Machine learning model accuracy and error rates. a Out of bag error rate of random forest models for the vaccinated, breakthrough infected and unvaccinated, infected groups at the visits 1,2,4 and 5. b Random forest model accuracies for the vaccinated and breakthrough infected or unvaccinated and infected groups at the visits 1,2,4 and 5 on an independent test dataset. c Variant-specific accuracy in vaccinated and breakthrough infected individuals for visit 1,2,4 and 5 d Variant-specific accuracy in unvaccinated and infected individuals for visit 1,2,4 and 5. e Exemplary neuron configuration of the neural networks used. With 6 input neurons, one hidden layer of 5 neurons and three output neurons giving the probability for A (Alpha), D (Delta) and O (Omicron) infection. f Neural network model accuracies for the vaccinated and breakthrough infected or unvaccinated and infected groups at the visits 1,2,4 and 5 on independent test datasets (n (vaccinated) = 171, thereof 24 Alpha, 92 Delta and 55 Omicron; n (unvaccinated) = 56, thereof 8 Alpha, 40 Delta, 8 Omicron, each n simulated to 333 as described in the methods, model replicates (Random Forest) = 10,000, model replicates (neural networks) = 100).

Supplementary Figure S1

Adjustment for follow up time and study centre effects. Tobit models for every visit were utilized to adjust the neutralization data for follow up time (scaled to mean) and study centre effects with centre 5 as reference. a Visualization of assumed linear follow-up time effects by vaccination status (0 = no vaccination n = 56, 1 = vaccinated n = 171) and visit. b Visualization of the adjustment for each visit and within the different groups analysed later (FA = Vaccinated Alpha BTI n = 24, FD = Vaccinated Delta BTI n = 92, FO = Vaccinated Omicron BTI n = 55, UA = Unvaccinated Alpha Infection n = 8, UD = Unvaccinated Delta Infection n = 40, UO = Unvaccinated Omicron infection n = 8). Shown are adjusted values on the y axis and unadjusted values on the X axis with loess regression to visualize trends. c Visualization of study centre stratified nAb titres and study centre adjustment, shown are median titres for each visit by study centre. The left panel shows unadjusted values, the right panel the adjusted values.

Supplementary Figure S2

Neutralization assays against the SARS-CoV-2 variant BQ.1.1 for vaccinated and unvaccinated Alpha, Delta and Omicron infected study groups at visit 5. Antibody responses of vaccinated (V, n = 171) and unvaccinated (U, n = 56) individuals at visit 5 after infection with Alpha, Delta or Omicron (as indicated on the x-axis). Serum dilution factors needed to reach 50% inhibitory concentration (IC50) for Omicron BQ.1.1. The assay cut-off (IC50 = 20) as well as saturation (IC50 = 2560) are indicated by dashed lines. IC50 were determined by three replicates. Median and interquartile ranges are indicated in blue. Differences between vaccinated and unvaccinated were analysed by Mann-Whitney test, statistically significant differences are indicated by p values (∗∗ p < 0.005). [Mann-Whitney-U] Differences between infection variant groups were analysed by Kruskal-Wallis tests with post-hoc Dunn’s corrected multiple comparisons tests, statistically significant differences are indicated by p values (∗ p < 0.05,∗∗ p < 0.005, ∗∗∗ p< 0.001, ∗∗∗∗ p < 0.0001)[Kruskal-Wallis with post-hoc Dunn’s].

Supplementary Figure S3

Estimates of fold-change compared to the homologous variant. Comparison of median fold differences for all measured SARS-CoV-2 variants relative to the titres against the variant of infection: Alpha panel a (vaccinated, n = 24) and d (unvaccinated, n = 8), Delta panel b (vaccinated, n = 92) and e (unvaccinated, n = 40), Omicron panel c (vaccinated, n = 55) and f (unvaccinated, n = 8). Visits 1, 2, 4 and 5 are shown simultaneous in each panel. Points show the median estimate for the fold-change difference while lines show the interquartile range for the estimate.

Supplementary Figure S4

Comparison of neutralization assays against different SARS-CoV-2 variants for double- and triple-vaccinated Omicron groups. Antibody responses of double vaccinated (2V, n = 47) and triple vaccinated (3V, n = 8) individuals at visit 1, 2, 4 and 5 after infection with Alpha, Delta or Omicron (as indicated on the x-axis). Shown are the dilution factors for 50 % Inhibitory concentration (IC50) for WT/D614G, Alpha, Delta, Omicron BA.1, Omicron BA.2 and Omicron BA.5 spike, as indicated by the individual graph titles. The assay cut-off (IC50 = 20) as well as saturation (IC50 = 2560) are indicated by dashed lines. IC50s were determined in triplicates. Median and interquartile ranges are indicated in blue. Statistically significant differences are indicated by p values (∗ p < 0.05, ∗∗ p < 0.005).[Mann-Whitney-U].

Supplementary Figure S5

Comparison of neutralization in adenoviral vector- or mRNA vaccinated individuals. a Shown are pooled nAb titres against all tested variants (D614G, Alpha, Delta, BA1, BA2, BA5) in either individuals who received an adenoviral vaccination at any point (n = 18) or in individuals who only received mRNA vaccines (n = 153), for the different visits. Statistically significant differences are indicated by p values (∗ p < 0.05, ∗∗ p < 0.005)[Mann-Whitney-U] b Shown are variant specific nAb titres at four different visits in either individuals who received an adenoviral vaccination at any point (n = 18) or in individuals who only received mRNA vaccines. Statistically significant differences are indicated by p values (∗ p < 0.05, ∗∗ p < 0.005). [Mann-Whitney-U].

Supplementary Figure S6

Non-linear association of time intervals between vaccination and BTI on the nAb response (IC50) towards the infecting variant. Generalized additive model smooths based on thin plate regression splines, with automatic selection of the effective degrees of freedom, for alpha (red, n = 24), delta (green, n = 92) and omicron (blue, n = 55) BTI. Shown are separate models for Visit 1, Visit 2, Visit 4 and Visit 5. Shaded regions indicate approximate 95%-prediction intervals (Prediction ± 2×SE).

Supplementary Figure S7

Magnitude breadth curves with confidence 95% confidence intervals and log-rank tests as alternative comparison method. Shown are the four panels (one for each visit) with magnitude breadth curves colour coded for the different groups (Purple = Alpha BTI n = 24, Dark blue = Delta BTI n = 92, brown = Omicron BTI n = 55 and pink = Alpha infection n = 8, light blue = Delta infection n = 40, orange = Omicron infection n = 8). Below are the results of the log rank tests for comparison of all groups (FA = Alpha BTI, FD = Delta BTI, FO = Omicron BTI and UA = Alpha infection, UD = Delta infection, UO = Omicron infection), plotted as a heatmap (grey = not significant, white to red = significant, 0 denotes a p value < 0.0001) for each visit. [log-rank-test].

Supplementary Figure S8

Comparison of magnitude-breadth analysis between double- and triple-vaccinated Omicron groups. To the magnitude breadth curves of Figure 2 the curves for double- (O2xV n = 48) and triple-(O3xV, n = 7) vaccinated & Omicron infected subgroups are added at visit 1 (V1), visit 2 (V2), visit 4 (V4) and visit 5 (V5). Area under the curve (AUC) for the double- and triple-vaccinated & Omicron infected groups are calculated and compared at V1, V2, V4 and V5. Statistically significant differences are indicated by p values (∗ p < 0.05, ∗∗ p < 0.005, ∗∗∗ p <0.001)[Mann-Whitney U].

Supplementary Figure S9

Assessing the effect of the uncertainty in variant reactivity for the antigenic maps using bootstrapping. Antigenic maps for four different study time points (V1, V2, V4 and V5, as indicated) based on sera of vaccinated and unvaccinated participants with Alpha, Delta or Omicron infections. The SARS-CoV-2 variants are shown as big coloured circles on the map WT/D614G (red), Alpha (purple), Delta (blue), Omicron BA.1 (orange), BA.2 (yellow), BA.5 (green) and BQ.1.1. (grey) Individual sera are not shown. The x- and y-axes of the map both represent antigenic distance. Each grid square (1 antigenic unit) represents a 2-fold change in neutralization titre. 1000 resampling bootstrap repeats were performed with 100 optimizations per repeat. at all visits. Each coloured region indicates the area in which 68% (one standard deviation) of the positional variation of variant.

Supplementary Figure S10

Antibody landscapes for Omicron double-vaccinated versus triple-vaccinated cases. Coloured surfaces show the fitted geometric mean titre (GMT) antibody landscapes for the different study groups vaccinated & Alpha infected (grey), unvaccinated Alpha infected (grey), vaccinated & Delta infected (grey), unvaccinated Delta infected (grey), triple vaccinated & Omicron infected (dark yellow, n = 48), double vaccinated & Omicron infected (red, n = 7) and unvaccinated Omicron infected (grey). While the base x-y plane corresponds to the antigenic map shown in Figure 3, grey impulses show the height of the GMT for specific SARS-CoV-2 variants. The vertical z-axis in each plot corresponds to the GMT on the log2 scale, each two-fold increment is marked, starting from a titre of 20at the map surface. The antibody landscapes for visit 1 can be seen in the upper left panel, visit 2 in the upper right panel, visit 4 in the lower left panel and visit 5 in the lower right panel.

Validation-cellline