Severe Acute Respiratory Syndrome Coronavirus 2 Infection History and Antibody Response to 3 Coronavirus Disease 2019 Messenger RNA Vaccine Doses

Abstract

Background

Data on antibody kinetics are limited among individuals previously infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). From a cohort of healthcare personnel and other frontline workers in 6 US states, we assessed antibody waning after messenger RNA (mRNA) dose 2 and response to dose 3 according to SARS-CoV-2 infection history.

Methods

Participants submitted sera every 3 months, after SARS-CoV-2 infection, and after each mRNA vaccine dose. Sera were tested for antibodies and reported as area under the serial dilution curve (AUC). Changes in AUC values over time were compared using a linear mixed model.

Results

Analysis included 388 participants who received dose 3 by November 2021. There were 3 comparison groups: vaccine only with no known prior SARS-CoV-2 infection (n = 224); infection prior to dose 1 (n = 123); and infection after dose 2 and before dose 3 (n = 41). The interval from dose 2 and dose 3 was approximately 8 months. After dose 3, antibody levels rose 2.5-fold (95% confidence interval [CI] = 2.2–3.0) in group 2 and 2.9-fold (95% CI = 2.6–3.3) in group 1. Those infected within 90 days before dose 3 (and median 233 days [interquartile range, 213–246] after dose 2) did not increase significantly after dose 3.

Conclusions

A third dose of mRNA vaccine typically elicited a robust humoral immune response among those with primary vaccination regardless of SARS-CoV-2 infection >3 months prior to boosting. Those with infection <3 months prior to boosting did not have a significant increase in antibody concentrations in response to a booster.

COVID-19, mRNA vaccine, antibody response, antibody kinetics, booster

Since February 2020, the United States has experienced unparalleled levels of community transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), resulting in a high seroprevalence of SARS-CoV-2 infection-induced antibodies in the population [1, 2]. Introduction of coronavirus disease 2019 (COVID-19) messenger RNA (mRNA) vaccines substantially mitigated the impact of the pandemic by preventing infection or attenuating disease severity [3–5]. Although vaccine effectiveness studies show relatively prolonged protection against severe disease and hospitalizations, waning of neutralizing antibodies after vaccination and a corresponding reduction in the protective effect against infection have led to questions about the need and timing of additional doses to increase duration of protection [4, 6–8]. Infection and vaccination elicit discrete effects on adaptive immunity [9–12], with a third vaccine dose demonstrating relatively sustained effectiveness against severe outcomes of infection [3, 4, 13]. Less is understood, however, about the combined effects of vaccine and infection-induced (hybrid) immunity against SARS-CoV-2, the added benefit of a third vaccine dose following natural infection, the timing of booster doses, and the impact on antibody waning following a combination of vaccine and natural infection. The extent to which hybrid immunity, including order and timing of immune-modifying events as well as length of time between infection and vaccine doses, influences immune response is not well understood [5]. Improved understanding of these factors can help optimize the frequency and timing of additional doses [14].

In a prospective cohort of essential and frontline workers, we evaluated SARS-CoV-2–specific antibody decline against the S2 domain of the spike protein and the receptor-binding domain (RBD) of the Wuhan/Hu-1/2019 sequence virus for up to 9 months after vaccine dose 2 and compared antibody levels before and after dose 3 among 3 participant groups with a history of: vaccination only; infection prior to vaccination; and infection after dose 2 and before dose 3. We were specifically interested in the effects of a third dose among those with infection that occurred <90 days or >90 days prior to receipt of the third dose.

METHODS

Study Design and Setting

The HEROES–RECOVER essential and frontline worker study consists of 2 large prospective cohorts—HEROES (the Arizona Healthcare, Emergency Response, and Other Essential Workers Surveillance Study) and RECOVER (Research on the Epidemiology of SARS-CoV-2 in Essential Response Personnel)—that operate under similar previously described protocols and initiated active weekly surveillance for SARS-CoV-2 infection in August 2020 [15, 16]. Participants were enrolled in 6 US states: Arizona (Phoenix, Tucson, and other areas), Florida (Miami), Minnesota (Duluth), Oregon (Portland), Texas (Temple), and Utah (Salt Lake City). The Centers for Disease Control and Prevention relied upon institutional review board approval that was determined at each participating institution, and all participants gave written consent upon enrollment [15, 16].

Sociodemographic information and medical history were reported upon enrollment. Past infection status with SARS-CoV-2 prior to enrollment could be determined by either self-report of laboratory-confirmed infection (via reverse-transcription polymerase chain reaction [RT-RCR] or antigen test) or serology test result at enrollment. If either self-report or serology results indicated past infection, participants were categorized as having infection prior to enrollment. COVID-19 vaccination was self-reported and verified by vaccine card at all sites. In addition, sites in Minnesota, Oregon, Texas, and Utah reviewed occupational health records, electronic medical records, or state immunization registries.

Specimen and Symptom Data Collection

Participants self-collected a mid-turbinate nasal swab weekly and at the onset of COVID-19–like illness (CLI) symptoms, defined as fever, chills, cough, shortness of breath, sore throat, diarrhea, muscle aches, and/or a change in smell or taste. When CLI was identified, participants completed electronic surveys at the beginning and end of symptoms to describe the duration of illness, the presence or absence of fever, and any visit with a medical professional, not including self-treatment. Self-collected mid-turbinate swab specimens were tested for SARS-CoV-2 using RT-PCR assay at the Marshfield Clinic Laboratory (Marshfield, Wisconsin).

Serum was collected upon enrollment and at 11- to 13-week intervals during the study period. Additionally, participants were asked to submit serum within 21–60 days after qualitative RT-PCR–confirmed SARS-CoV-2 infection and 14–60 days after each COVID-19 vaccine dose.

Laboratory Testing

Sera were sent to the University of Arizona Genetics Core Laboratory for testing using a semiquantitative enzyme-linked immunosorbent assay (ELISA) for RBD and S2 antibodies, as previously described [17]. Anti-RBD and anti-S2 antibody levels were measured as area under the serial dilution curve (AUC) of optical density values from 5 serial 1:3 dilutions beginning at a 1:60 dilution of serum and ending at 1:4860. The AUC measurement is a well-established method for summarizing semiquantitative ELISA results and has improved coverage probabilities of titration curves compared with other values (eg, end point titer) [18, 19].

Statistical Analyses

Participants were included in the analysis if they received 3 mRNA COVID-19 vaccine doses by November 2021 and submitted scheduled routine, post-vaccination, and post-infection serum specimens, as relevant. Three participant groups were identified for analysis: those without a history of SARS-CoV-2 infection (“vaccination only”), those with infection prior to receipt of dose 1 (“previous infection”), and those with infection after receipt of dose 2 and before dose 3 (“breakthrough infection”). Those who did not receive 3 doses of vaccine or participated in a clinical vaccine trial and those with an infection between dose 1 and dose 2 were excluded from analysis (Supplementary Figure 1). Participants infected prior to enrollment were excluded from descriptive statistics of infection characteristics, as these details were unknown. Breakthrough infections that occurred within 14 days after dose 2 were also excluded.

Geometric mean AUC values were calculated at several time points including: post-infection (median 42 days, interquartile range [IQR] = 27–118), between infection and blood draw); post-14 days after dose 1; at 5 intervals of 50 days after dose 2, and after dose 3 (median 21 days, IQR = 15–28). Fifty-day intervals were arbitrarily chosen in the absence of established thresholds for measuring antibody decline. The period of follow-up after dose 3 corresponds with the study protocol that designated “post-vaccine blood specimens” as those received 14–60 days after vaccine receipt. Not all participants provided serum during each analytic time period; therefore, antibody levels calculated for each period included a different number of participants. To evaluate changes in AUC values over time within participant groups, geometric mean ratios (GMRs) of the AUC values and 95% confidence intervals (CIs) were estimated using linear mixed effects models adjusting for age, vaccine manufacturer, race, chronic conditions, and repeated measures. Adjustment for vaccine manufacturer accounted for whether participants received homologous doses of either BNT162b2 or mRNA-1273. Sample size did not permit adjustment for heterologous vaccine dosing. We evaluated associations between participant characteristics and the geometric mean AUC following dose 3 using the Pearson χ² test, Fisher exact test, or 1-way analysis of variance. To compare antibody levels across participant groups following dose 3, the GMR of the AUC values from the most recent collection prior to dose 3 to AUC values after dose 3 were estimated using linear models to adjust for age, vaccine manufacturer, race, and chronic conditions. Among vaccinated-only and previously infected participants, we assessed the rate of antibody decline after dose 2 using mixed effects models with a natural spline for time since dose 2, adjusting for vaccine manufacturer and chronic conditions and repeated measures. Statistical significance was defined as P < .05 or whether 95% CIs did not overlap.

AUC values were calculated using GraphPad Prism (v9). All analyses were conducted with SAS software, version 9.4 (SAS Institute) and R software, version 4.2.0 (R Foundation for Statistical Computing).

RESULTS

Participant Characteristics

From July 2020 to November 2021, 388 study participants received 3 mRNA COVID-19 vaccine doses and met inclusion criteria, including 224 with COVID-19 vaccine-only exposure, 123 previously SARS-CoV-2 infected, and 41 with breakthrough infection after at least 2 vaccine doses (Table 1, Supplementary Figure 1). Most participants resided in Arizona (50%) and Duluth, Minnesota (40%). More than half of participants were female (69%), aged 18–39 years (58%), and healthcare personnel (76%). Participants commonly reported at least 1 chronic condition (36%) and taking more than 1 daily medication (41%; Table 1). Most (88%) participants received the BNT162b2 vaccine product for all 3 doses; 10% received mRNA-1273, and 2% received a combination of vaccine products over the 3 doses (Table 1).

Table 1.

Coronavirus Disease 2019 Messenger RNA Vaccine Dose 3 Recipient Characteristics According to Severe Acute Respiratory Syndrome Coronavirus 2 Infection Status and Timing

Characteristic	Total Participants, No. (%)	Vaccination-Only (ie, Never Infected), No. (%)	Previous SARS-CoV-2 Infection, No. (%)	SARS-CoV-2 Breakthrough Infection, No. (%)	P Value
All participants (row %)	388	224 (0.6)	123 (0.3)	41 (0.1)
Cohort location
Tucson, Arizona	113 (29.1)	74 (33.0)	22 (17.9)	17 (41.5)	<.001*
Phoenix and other areas in Arizona	82 (21.1)	51 (22.8)	27 (22.0)	4 (9.8)
Miami, Florida	1 (0.3)	0 (0.0)	1 (0.8)	0 (0.0)
Duluth, Minnesota	154 (39.7)	97 (43.3)	47 (38.2)	10 (24.4)
Portland, Oregon	12 (3.1)	0 (0.0)	7 (5.7)	5 (12.2)
Temple, Texas	4 (1.0)	0 (0.0)	3 (2.4)	1 (2.4)
Salt Lake City, Utah	22 (5.7)	2 (0.9)	16 (13.0)	4 (9.8)
Sex
Female	267 (68.8)	147 (65.6)	92 (74.8)	28 (68.3)	0.210
Male	121 (31.2)	77 (34.4)	31 (25.2)	13 (31.7)
Age, y
18–39	224 (57.7)	48 (21.4)	47 (38.2)	12 (29.3)	.002*
40–64	123 (31.7)	164 (73.2)	70 (56.9)	23 (56.1)
≥65	41 (10.6)	12 (5.4)	6 (4.9)	6 (14.6)
Race and ethnicity
Non-Hispanic, White	336 (86.6)	193 (86.2)	105 (85.4)	38 (92.7)
Non-Hispanic, Black	1 (0.3)	0 (0.0)	1 (0.8)	0 (0.0)	.826
Non-Hispanic, Asian	5 (1.3)	3 (1.3)	2 (1.6)	0 (0.0)
Hispanic	39 (10.1)	23 (10.3)	13 (10.6)	3 (7.3)
Other	7 (1.8)	5 (2.2)	2 (1.6)	0 (0.0)
Occupationa
Primary HCP	123 (31.7)	79 (35.3)	34 (27.6)	10 (24.4)	.252
Nurses and other allied HCP	170 (43.8)	95 (42.4)	59 (48.0)	16 (39.0)
First responders	18 (4.6)	7 (3.1)	8 (6.5)	3 (7.3)
Essential and other frontline workers	77 (19.8)	43 (19.2)	22 (17.9)	12 (29.3)
Body mass indexb
Underweight/Normal weightc	142 (36.6)	82 (36.6)	48 (39.0)	12 (29.3)	.522
Overweight (25–29.9 kg/m2)	129 (33.2)	80 (35.7)	32 (26.0)	17 (41.5)
Obese	111 (28.6)	61 (27.2)	40 (32.5)	10 (24.4)
Chronic conditiond
None	247 (63.7)	150 (67.0)	72 (58.5)	25 (61.0)	.512
1 or more	138 (35.6)	74 (33.0)	49 (39.8)	15 (36.6)
Daily medications
0	142 (36.6)	82 (36.6)	46 (37.4)	14 (34.1)	.957
1	84 (21.6)	51 (22.8)	25 (20.3)	8 (19.5)
2 or more	158 (40.7)	90 (40.2)	50 (40.7)	18 (43.9)
Smoking
Not current smoker	319 (82.2)	184 (82.1)	96 (78.0)	39 (95.1)	.702
Smoke tobacco products	66 (17.0)	40 (17.9)	25 (20.3)	1 (2.4)
Influenza vaccination history in past 5 y
No vaccination history	20 (5.2)	11 (4.9)	8 (6.5)	1 (2.4)	.294
1–3 y of vaccination	19 (4.9)	7 (3.1)	9 (7.3)	3 (7.3)
≥4 y of vaccination	346 (89.2)	206 (92.0)	104 (84.6)	36 (87.8)
SARS-CoV-2 clinical presentatione
CLI symptoms			31 (81.6)	19 (90.5)	.369
Non-CLI associated symptomsf			2 (5.3)	0 (0.0)
Asymptomatic			5 (13.2)	2 (9.5)
SARS-CoV-2 detection duration, wk
1			8 (21.1)	10 (47.6)	.034*
≥2			30 (78.9)	11 (52.4)
Febrile COVID-19
No			15 (39.5)	10 (47.6)	.442
Yes			23 (60.5)	10 (47.6)
Medically attended COVID-19
No medical visit			25 (65.8)	17 (81.0)	.120
1 or more medical visits			13 (34.2)	3 (14.3)
COVID-19 vaccine products (all 3 doses)
BNT162b2	340 (87.6)	202 (90.2)	106 (86.2)	32 (78.0)	.222
mRNA-1273	40 (10.3)	19 (8.5)	14 (11.4)	7 (17.1)
Combination	8 (2.1)	3 (1.3)	3 (2.4)	2 (4.9)

Abbreviations: CLI, COVID-19-like illness; COVID-19, coronavirus disease 2019; HCP, healthcare provider; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

*Statistically significant.

Primary HCPs include medical doctor, doctors of osteopathic medicine, nurse practitioners, and physician assistants.

Body mass index, kg/m²: underweight, ≤18.4; normal, 18.5–24.9; overweight, 25–29.9; obese, BMI ≥30.

Only 1 participant was underweight by 0.6 and therefore combined with normal weight.

Chronic conditions include asthma, autoimmune disease, chronic lung disease, cancer, diabetes, heart disease, hypertension, immunosuppression, kidney disease, liver disease, and neurologic or neuromuscular disease.

Includes only 59 infections that occurred during the active surveillance study period. Those with a self-reported infection prior to enrollment are excluded from this section as infection characteristics are unknown.

Non-CLI symptoms included headache, nausea, and fatigue.

Among 59 participants with RT-PCR–confirmed infection that occurred after serologic testing at enrollment and during the active surveillance study period (July 2020 to November 2021) when complete symptom data were collected during follow-up, 38 had infections prior to vaccination (group 2) and 21 had breakthrough infections (group 3). Respectively, 31 of 38 (82%) and 19 of 21 (90%) reported CLI symptoms, and 30 of 38 (79%) and 11 of 21 (52%) had SARS-CoV-2 detectable for at least 2 weeks after the first RT-PCR–positive specimen (Table 1).

After each vaccine dose, sera were collected after a median interval of 17 days (IQR = 14–20). Between dose 2 and dose 3, routine serum collections occurred every 3 months with a median interval of 181 days (IQR = 87–265) from dose 2 to serum collection.

There were no large differences in geometric mean AUC values after dose 3 among various sociodemographic factors including sex, age, and race and ethnicity. Significant differences were observed between categories of occupation and location; however, the difference in occupation was not consistent across groups (Supplementary Tables 1 and 2).

Response to COVID-19 Vaccine Among Vaccinated-Only Participants

Among vaccinated-only participants, the GMR comparing the AUC values for RBD and S2 antibody levels were 2.6 (95% CI = 2.4–2.9) and 1.8 (95% CI = 1.7–2.0) after dose 2 (<50 days) compared with after dose 1, respectively (Tables 2 and 3). Following the dose 2 response (<50 days), we observed a significant decline in antibody levels measured at each time point until dose 3. Compared with sera collected after dose 2 (<50 days), RBD and S2 titers from sera collected 200 days or more after dose 2 had decreased 32% and 55%, respectively (Figure 1). Following dose 3, RBD antibody levels increased with a GMR of 2.9 (95% CI = 2.6–3.3), and S2 antibody levels increased with a GMR of 2.6 (95% CI =2.3–2.9; Tables 2 and 3).

Figure 1.

Quantity of RBD and S2 binding antibodies followed longitudinally after each of 3 doses of mRNA coronavirus disease 2019 vaccine, comparing vaccinated-only (ie, never infected) vs previously SARS-CoV-2 infected. Antibody levels were measured via enzyme-linked immunosorbent assays as AUC. AUC values were categorized within 50-day periods following mRNA dose 1 (D1), dose 2 (D2), and dose 3 (D3). Geometric mean AUCs were calculated via mixed effects linear regression on log-transformed AUC values. RBD antibody levels (A) and S2 domain antibody levels (B). Abbreviations: AUC, area under the serial dilution curve; mRNA, messenger RNA; RBD, receptor-binding domain; S2, spike-protein; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

Table 2.

Quantity of Receptor-Binding Domain Antibodies to Severe Acute Respiratory Syndrome Coronavirus 2 Following 3 Doses of Vaccination With Messenger RNA Coronavirus Disease 2019 Vaccines by Infection History

Time Period of Blood Collection by Infection History	Blood Samples	Days from Event to Blood Collection,	Quantitative Receptor-Binding Domain Enzyme-Linked Immunosorbent Assaya
	N	Median (Interquartile Range)	Geometric Mean Area Under the Serial Dilution Curve	(95% CI)	Geometric Mean Ratio With Previous Time Pointa	(95% CI)
Vaccination-only (ie, never SARS-CoV-2 infected)
After dose 1	189	16 (15–18)	0.0042	(.0039–.0044)
After dose 2 (0–50 d)	247	17 (15–21)	0.0110	(.0103–.0116)	2.63	(2.4–2.9)*
After dose 2 (51–100 d)	92	92 (90–97)	0.0076	(.0070–.0084)	0.70	(.6–.8)*
After dose 2 (101–150 d)	83	111 (105–134)	0.0063	(.0057–.0070)	0.83	(.7–1.0)
After dose 2 (151–200 d)	144	181 (173–190)	0.0041	(.0038–.0045)	0.65	(.6–.7)*
After dose 2 (201 d-dose 3)	81	249 (225–257)	0.0036	(.0033–.0040)	0.87	(.8–1.0)
After dose 3	302	20 (15–28)	0.0106	(.0101–.0111)	2.93	(2.6–3.3)*
Previous SARS-CoV-2 infection
After infection	87	64 (28–127)	0.0033	(.0030–.0037)		(.0–.0)
After dose 1	65	15 (14–17)	0.0108	(.0096–.0122)	3.24	(2.8–3.8)*
After dose 2 (0–50 d)	83	17 (13–21)	0.0123	(.0111–.0137)	1.14	(1.0–1.3)
After dose 2 (51–100 d)	35	92 (89–96)	0.0094	(.0080–.0111)	0.76	(.6–.9)*
After dose 2 (101–150 d)	35	113 (106–139)	0.0077	(.0065–.0091)	0.82	(.7–1.0)
After dose 2 (151–200 d)	68	178 (169–189)	0.0068	(.0060–.0076)	0.88	(.7–1.1)
After dose 2 (201 d-dose 3)	46	244 (222–249)	0.0043	(.0038–.0050)	0.64	(.5–.8)*
After dose 3	128	21 (15–28)	0.0111	(.0102–.0121)	2.55	(2.2–3.0)*
SARS-CoV-2 breakthrough infection
After dose 1	13	17 (15–21)	0.0036	(.0028–.0047)		(.0–.0)
After dose 2 (0–50 d)	20	17 (14–22)	0.0101	(.0082–.0124)	2.81	(2.0–3.9)*
After infection and dose 2 (51–100 d)	0					(.0–.0)
After infection and dose 2 (101–150 d)	8	135 (116–142)	0.0075	(.0053–.0108)	0.75	(.5–1.1)
After infection and dose 2 (151–200 d)	14	176 (166–190)	0.0072	(.0056–.0093)	0.96	(.6–1.5)
After infection and dose 2 (201 d-dose 3)	22	248 (222–265)	0.0077	(.0063–.0093)	1.06	(.8–1.5)
After dose 3	48	27 (19–35)	0.0133	(.0116–.0152)	1.74	(1.4–2.2)*

Abbreviations: CI, confidence interval; S2, spike-protein; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

*Statistically significant.

Adjusted for age, race, chronic conditions, and vaccine manufacturer.

Table 3.

Quantity of S2 Domain Antibodies Following 3 Doses of Vaccination With Messenger RNA Coronavirus Disease 2019 Vaccines by Infection History

Time Period of Blood Collection by Infection History	Blood Samples,	Days from Event to Blood Collection,	Quantitative S2 Enzyme-Linked Immunosorbent Assaya
	N	Median (Interquartile Range)	Geometric Mean Area Under the Serial Dilution Curve	(95% CI)	Geometric Mean Ratio With Previous Time Pointa	(95% CI)
Vaccination-only (ie, never SARS-CoV-2 infected)
After dose 1	189	16 (15–18)	0.0035	(.0033–.0038)
After dose 2 (0–50 d)	247	17 (15–21)	0.0065	(.0062–.0069)	1.84	(1.7–2.0)*
After dose 2 (51–100 d)	92	92 (90–97)	0.0064	(.0059–.0070)	0.99	(.9–1.1)
After dose 2 (101–150 d)	83	111 (105–134)	0.0053	(.0048–.0059)	0.83	(.7–0.9)*
After dose 2 (151–200 d)	144	181 (173–190)	0.0045	(.0042–.0048)	0.84	(.7–1.0)
After dose 2 (201 d-dose 3)	81	249 (225–257)	0.0036	(.0033–.0040)	0.81	(.7–0.9)*
After dose 3	302	20 (15–28)	0.0092	(.0088–.0097)	2.55	(2.3–2.9)*
Previous SARS-CoV-2 infection
After infection	87	64 (28–127)	0.0071	(.0064–.0079)		(.0–.0)
After dose 1	65	15 (14–17)	0.0112	(.0099–.0125)	1.57	(1.3–1.8)*
After dose 2 (0–50 d)	83	17 (13–21)	0.0116	(.0105–.0129)	1.04	(.9–1.2)
After dose 2 (51–100 d)	35	92 (89–96)	0.0086	(.0074–.0101)	0.74	(.6–.9)*
After dose 2 (101–150 d)	35	113 (106–139)	0.0088	(.0075–.0103)	1.02	(.8–1.3)
After dose 2 (151–200 d)	68	178 (169–189)	0.0084	(.0075–.0094)	0.96	(.8–1.2)
After dose 2 (201 d-dose 3)	46	244 (222–249)	0.0052	(.0045–.0059)	0.62	(.5–.7)*
After dose 3	128	21 (15–28)	0.0107	(.0099–.0116)	2.07	(1.8–2.4)*
SARS-CoV-2 breakthrough infection
After dose 1	13	17 (15–21)	0.0040	(.0032–.0049)		(.0–.0)
After dose 2 (0–50 d)	20	17 (14–22)	0.0070	(.0059–.0084)	1.76	(1.3–2.3)*
After infection and dose 2 (51–100 d)	0			(.0000–.0000)		(.0–.0)
After infection and dose 2 (101–150 d)	8	135 (116–142)	0.0062	(.0047–.0083)	0.94	(.7–1.3)
After infection and dose 2 (151–200 d)	14	176 (166–190)	0.0071	(.0058–.0087)	1.10	(.8–1.6)
After infection and dose 2 (201 d-dose 3)	22	248 (222–265)	0.0073	(.0062–.0087)	0.97	(.7–1.3)
After dose 3	48	27 (19–35)	0.0104	(.0093–.0116)	1.45	(1.2–1.8)*

Abbreviations: CI, confidence interval; S2, spike-protein; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

*Statistically significant.

Adjusted for age, race, chronic conditions, and vaccine manufacturer.

Response to COVID-19 Vaccine Among Participants With Previous Infection

Among previously infected participants, the GMR after dose 1 compared with post-infection levels was 3.2 (95% CI = 2.8–3.8) for RBD antibody levels and 1.6 (95% CI = 1.3–1.8) for S2 antibody levels (Tables 2 and 3). After dose 2 (<50 days), the GMR compared with post-dose 1 was modestly higher for RBD and S2 antibody levels (GMR, 1.14; 95% CI = 1.0–1.3 and GMR, 1.04; 95% CI = .9–1.2, respectively). The GMR comparing post-dose 1 to post-dose 2 antibody levels according to days from infection to dose 1 is shown in Supplementary Tables 3 and 4. Compared with sera collected after dose 2 (<50 days), antibody levels for both RBD and S2 ≥ 200 days after dose 2 had decreased 37% and 46%, respectively (Figure 1). Receipt of dose 3 elicited a 2.5-fold (95% CI = 2.2–3.0) increase in RBD antibody levels and 2.1-fold (95% CI = 1.8–2.4) increase in S2 antibody levels (Tables 2 and 3).

Rate of Antibody Decline Among Vaccine-Only and Previously Infected Participants

Following dose 2, previously infected participants had consistently higher RBD and S2 antibody levels over time compared with those with vaccination only until ≥200 days post-dose 2 (Tables 2 and 3, Figure 1). The pattern of antibody decline among vaccinated-only compared with previously infected participants differed significantly (P < .001; Figure 2). By the final time point >200 days after dose 2, the S2 antibody levels among previously infected participants were low but remained significantly higher than those among vaccinated-only participants (95% CI of AUC values did not overlap; Tables 2 and 3).

Figure 2.

Restricted cubic spline models estimating waning of RBD and S2 binding antibodies after receipt of 2 doses of mRNA coronavirus disease 2019 vaccine and before receipt of a third mRNA dose, comparing vaccinated-only (ie, never infected) vs previously SARS-CoV-2 infected. Antibody levels were measured via enzyme-linked immunosorbent assays as AUC. RBD antibody levels (A) and S2 antibody levels (B). Abbreviations: AUC, area under the serial dilution curve; mRNA, messenger RNA; RBD, receptor-binding domain; S2, spike-protein; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

Comparison of Antibody Response to COVID-19 Vaccine Dose 3 by Infection History

Antibody levels after dose 3 were significantly higher for all 3 groups compared with the final measurement after dose 2 (>200 days; Tables 2 and 3, Supplementary Figure 2). The geometric mean AUC of RBD following dose 3 was not significantly different among groups; however, participants who were previously infected and those with breakthrough infections had significantly higher S2 antibody levels after dose 3 compared with the vaccinated-only group (Supplementary Figure 2).

Participants with recent breakthrough infection that occurred within 90 days before dose 3 had a 28% higher mean RBD antibody level following dose 3 than those with breakthrough infection >90 days before dose 3, though not a statistically significant difference (Figure 3, Supplementary Table 5). Notably, among those with a recent infection, the antibody level final measurement before dose 3 was similar to other post-vaccination levels (AUC values >0.01), and no significant change in RBD or S2 antibody levels from before to after dose 3 were detected (Figure 3). In contrast, among breakthrough infections that occurred more than 90 days before dose 3, RBD and S2 antibody levels were 2.3-fold (95% CI = 1.7–3.0) and 1.7-fold (95% CI = 1.4–2.1) higher after dose 3 compared with the final measurement before dose 3 (Supplementary Tables 5 and 6).

Figure 3.

Magnitude of RBD binding-antibody levels after messenger RNA dose 3, comparing participants with severe acute respiratory syndrome coronavirus 2 breakthrough infection (ie, after dose 2) <90 days before dose 3 vs participants with breakthrough infection that occurred >90 days before dose 3. RBD antibody levels were measured via enzyme-linked immunosorbent assay as AUC. For both anti-RBD and anti-spike protein, geometric mean AUCs were estimated from mixed effects linear regression on log-transformed AUC values. Abbreviations: AUC, area under the serial dilution curve; RBD, receptor-binding domain.

End point titers were also calculated for all time points and analytic groups, and overall results were similar to those calculated from AUC values (Supplementary Tables 7 and 8).

DISCUSSION

We followed a cohort of essential and frontline workers and evaluated the effects of hybrid immunity on humoral immune response following COVID-19 mRNA vaccination. These findings indicate that serum antibody levels decreased following 2 vaccine doses regardless of infection history. During the initial 6 months after receipt of a second dose, participants with previous infection sustained higher antibody levels for a longer period compared with vaccinated-only participants. Antibody levels increased substantially after receipt of a third dose, and these increases were significant among participants with vaccine-only–induced immunity, as well as those with infection prior to receipt of the first dose and those with breakthrough infection that occurred more than 3 months before the third dose.

Importantly, we did not observe a significant rise in antibody response to a third dose among those with a relatively recent infection, that is, within 3 months prior to receipt of dose 3. This suggests that individuals with recent infection may have already reached a maximum antibody level and thus did not produce a measurable response to dose 3. Findings from previous research have also suggested that longer intervals between infection and vaccination may improve the immune response to vaccination [14].

Our study is one of few, outside of randomized, controlled trials, to follow the same individuals over a significant amount of time to assess binding-antibody levels after each of 3 doses of vaccine. Over a 9-month period, we observed that participants with previous infection had elevated binding to S2 and RBD across all analytic time points relative to vaccinated-only participants. These data are consistent with previous studies that indicated higher vaccine-elicited titers and neutralizing antibodies among individuals infected with SARS-CoV-2 months before receipt of an mRNA vaccine [20–25]. This also reflects immunological memory of past infection and corresponding anamnestic response beginning with the first vaccination that may stimulate superior antibody binding compared with vaccinated-only individuals [24].

The pattern of antibody waning after dose 2 of mRNA vaccine also appears to be influenced by the presence of prior infection. Among participants with previous infection, higher levels of vaccine-elicited antibodies were maintained for a longer period relative to vaccinated-only participants. Among vaccinated-only participants, a rapid decline was observed after the first 3 months post-vaccination. A similar rapid decline was not observed among previously infected participants until approximately 5 months post-vaccination, suggesting that hybrid immunity may confer longer stability of RBD and S2 binding antibodies from the time of dose 2 receipt.

This study has several limitations. First, although we were unable to adjust for unmeasured confounding of the association between infection history and humoral immune response, we did control for several important factors that may influence immune parameters and risk of infection, or are expected to be associated with vaccination status. Second, the majority (86.6%) of participants were non-Hispanic White. Assuming the possibility of heterogeneity of antibody kinetics according to race or ethnicity, results may not be generalizable to all populations. Third, a proportion of AUC values were calculated from titration curves that did not cross the limit of detection prior to the final series dilution, which caused censoring among participants with high antibody levels [19]. However, SARS-CoV-2 antibody measurement used a validated assay against 2 separate regions of the spike protein that each contain neutralizing epitopes (RBD and S2), which guarded against detection of antibodies that may be cross reactive from previous seasonal coronavirus infections while also providing a better picture of the full breadth of humoral antibody response [26]. Fourth, both the AUC and the end titer for SARS-CoV-2 are not standardized measures that allow for comparison to titer values from other assays outside of the University of Arizona Laboratory. Fifth, correlates of protection analyses have not been determined to allow for true clinical interpretations or implications of the titer values. Finally, this study did not look at antibody levels after receipt of bivalent COVID-19 booster doses, which are currently recommended across the United States to protect against newer variants. Although titers did not increase significantly among those with infection <3 months prior to a third dose, the potential benefit of a third dose should not be negated, especially in the context of the bivalent vaccine that confers protection against the Omicron variant. The assay used in this study precedes the emergence of Omicron and therefore measures antibody binding to antigen targets that are specific to the ancestral SARS-CoV-2 virus, rather than the Omicron variant. An investigation of antibody response to Omicron-specific antigens was beyond the scope of this analysis. Furthermore, viral sequencing results were not available, precluding an examination of antibody response according to infection with specific variants.

In conclusion, our study among a cohort of frontline workers demonstrated increased vaccine-elicited antibodies after the primary series of mRNA COVID-19 vaccination, followed by gradual waning over time irrespective of previous SARS-CoV-2 infection. The combination of infection and vaccination led to slightly slower waning of antibody levels over time, indicating that decisions around timing of additional booster doses might consider timing and type of previous immune-modifying events. However, we demonstrated a clear immunological benefit from a third dose of vaccine regardless of previous infection, including among many of those with a breakthrough infection after the primary vaccine series. Only those with infection close to the time of dose 3 receipt demonstrated a minimal response. These data support data from previous studies [14, 27, 28] that suggest waiting at least 3 months post-infection to maximize the boost in antibody titers. While immune response and kinetics that are specific to the bivalent vaccine or infection with Omicron variant were not explored in our study, these results highlight the importance of tracking COVID-19 immune-modifying events and lay the groundwork for decisions about the timing of vaccine doses.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors thank the following:

Centers for Disease Control and Prevention (CDC). Eduardo Azziz-Baumgartner, Melissa L. Arvay, William Brannen, Stephanie Bialek, Allison Ciesla, Alicia M. Fry, Aron Hall, Adam MacNeil, Clifford McDonald, Sue Reynolds, Robert Slaughter, Matthew J. Stuckey, Rose Wang, Ryan Wiegand. Abt Associates: Steve Pickett, Rekha Balachandran, Tyler Morrill, Kim Groover, Deanna Fleary, Robin Bloodworth, Laura Edwards, Peenaz Mistry. Baylor Scott and White Health: Kayan Dunnigan, Nicole Calhoun, Leah Odame-Bamfo, Clare Mathenge, Michael E. Smith, Kempapura Murthy, Tnelda Zunie, Eric Hoffman, Martha Zayed, Ashley Graves, Joel Blais, Jason Ettlinger, Sharla Russell, Natalie Settele, Tiya Searcy, Rupande Patel, Elisa Priest, Jennifer Thomas, Muralidhar Jatla, Madhava Beeram, Javed Butler, Alejandro Arroliga. Kaiser Permanente Northwest Center for Health Research: Holly Groom, Yolanda Prado, Daniel Sapp, Mi Lee, Chris Eddy, Matt Hornbrook, Donna Eubanks, Danielle Millay, Dorothy Kurdyla, Kristin Bialobok, Ambrosia Bass, Kristi Bays, Kimberly Berame, Cathleen Bourdoin, Rashyra Brent, Carlea Buslach, Lantoria Davis, Stephen Fortmann, Jennifer Gluth, Kenni Graham, Tarika Holness, Kelley Jewell, Enedina Luis, Abreeanah Magdaleno, DeShaun Martin, Joyce Smith-McGee, Martha Perley, Sam Peterson, Aaron Peipert, Krystil Phillips, Joanna Price, Ana Reyes, Sperry Robinson, Katrina Schell, Emily Schield, Natosha Shirley, Anna Shivinsky, Valencia Smith, Britta Torgrimson-Ojerio, Brooke Wainwright, Shawn Westaway. Marshfield Clinic Research Laboratory: Saydee Benz, Adam Bissonnette, Krystal Boese, Emily Botten, Jarod Boyer, Michaela Braun, Julianne Carlson, Caleb Cravillion, Amber Donnerbauer, Tim Dziedzic, Joe Eddy, Heather Edgren, Alex Ermeling, Kelsey Ewert, Connie Fehrenbach, Rachel Fernandez, Wayne Frome, Sherri Guzinski, Mitch Hertel, Garrett Heuer, Erin Higdon, Cressa Huotari, Lynn Ivacic, Lee Jepsen, Steve Kaiser, Bailey Keffer, Tammy Koepel, Sarah Kohn, Alaura Lemieux, Carrie Marcis, Megan Maronde, Isaac McCready, Nidhi Mehta, Dan Miesbauer, Collin Nikolai, Brooke Olson, Jeremy Olstadt, Lisa Ott, Cory Pike, Nicole Price, Chris Reardon, Alex Slenczka, Elisha Stefanski, Lydia Sterzinger, Kendra Stoltz, Melissa Strupp, Lyndsay Watkins, Roxann Weigel, Ben Zimmerman. University of Miami: Damena Gallimore-Wilson, Roger Noriega, Cynthia Beaver, Alexandra Cruz, Annabel Reyes, Brigitte Madan, Addison Testoff, John Jones. Whiteside Institute for Clinical Research: Jessica Lundgren, Karley Respet, Jennifer Viergutz, Daniel Stafki, Mary Robinson, Jill Dolezilek, Leiah Hoffman, Tyna O’Connor, Catherine Diluzio, Samantha Kendrick. St. Luke’s: Marilyn J. Odean. University of Arizona: Ariyah Armstrong, Nora Baccam, Zoe Baccam, Tatum Butcher, Shelby Capell, Andrea Carmona, Karysa Carson, Alissa Coleman, Hannah Cowling, Carly Deal, Kiara Earley, Sophie Evans, Julia Fisher, Ashlyn Flangos, Joe K. Gerald, Lynn Gerald, Anna Giudici, Erika Goebert, Taylor Graham, Sofia Grijalva, Hanna Hanson, Olivia Healy, Chloe Hendrix, Katherine Herder, Adrianna Hernandez, Raven Hilyard, James Hollister, Rezwana Islam, Krystal S. Jovel, Caroline Klinck, Karl Krupp, Karla Ledezma, Sally Littau, Amelia Lobos, Ashley Lowe, Jeremy Makar, Natalya Mayhew, Kristisha Mevises, Flavia Nakayima Miiro, Cierra Morris, Sarah Murray, Janko Nikolich-Žugich, Assumpta Nsengiyunva, Kennedy Obrien, Mya Pena, Riley Perlman, Celia Pikowski, Cynthia Porter, Ferris A. Ramadan, Patrick Rivers, Jen Scott, Priyanka Sharma, Alison Slocum, Saskia Smidt, Lili Steffen, Jayla Sowell, Danielle Stea, Xiaoxiao Sun, Nicholas Tang, Gianna Taylor, Ta’Nya Tomas, Heena Timsina, Italia Trejo, April Yingst. University of Utah: Kurt T. Hegmann, Matthew S. Thiese, Rachel T. Brown, Camie Schaefer, Arlyne Arteaga, Matthew Bruner, Daniel Dawson, Emilee Eden, Jenna Praggastis, Joseph Stanford, Jeanmarie Mayer, Marcus Stucki, Riley Campbell, Kathy Tran, Madeleine Smith, Braydon Black, Christina Pick, Madison Tallman, Chapman Cox, Derrick Wong, Michael Langston, Adriele Fugal, Fiona Tsang, Maya Wheeler, Gretchen Maughan, Megan Wilson, Pasha Stinson, Jesse Williams, Seon Reed, Jinyi Mao, Nikki Gallacher, Kendal Chatard, Jenna Vo, Katie Luong, Ryder Jordin, Grace Stewart, Brock Bourdelle, Timina Powaukee, Max Minoughan.

Disclaimer. The findings and conclusions presented here8 are those of the authors and do not necessarily represent the official position of the CDC. Use of trade names and commercial sources is for identification only and does not imply endorsement by the US Department of Health and Human Services.

Financial support. This work was supported by the CDC, National Center for Immunization and Respiratory Diseases (contracts 75D30120R68013 to Marshfield Clinic Research Institute, 75D30120C08379 to the University of Arizona, and 75D30120C08150 to Abt Associates). M. G. reports an institutional subcontract for the Research on the Epidemiology of SARS-CoV-2 in Essential Response Personnel–Pediatric Research Observing Trends and Exposures in COVID-19 Timelines cohort studies from the CDC and Abt Associates. M. G. W. reports prior employment at Abt Associates, contracted by the CDC to execute this study. J. L. B. reports support from the CDC paid to the University of Arizona.

Author notes

Potential conflicts of interest. A. L. N. reports research funding from Pfizer and Vir Biotechnology for unrelated studies. A. J. C.-M. reports Federal Emergency Management Agency grants (EMW-2019-FP-00526, EMW-2019-FP-00517, and EMW-2017-FP-00860) and State of Florida Appropriation 2369A and 2369A. M. G. reports grants or contracts unrelated to this work from the CDC–Abt Associates, CDC, CDC–VUMC, CDC–Westat and a role as co-chair of the Infectious Diseases and Immunization Committee, Texas Pediatric Society, Texas Chapter of the American Academy of Pediatrics. A. A. L. reports grants or contracts from Mercy C.A.R.E.S., National Heart Lung and Blood Institute, and the US Environmental Protection Agency and a leadership or fiduciary role with the Arizona Asthma Coalition. R. S. and J. L. U. report patents planned, issued, or pending (serological assays for SARS-CoV-2). All other authors: No reported conflicts of interest.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.