Antibody response in elderly vaccinated four times with an mRNA anti-COVID-19 vaccine

The humoral response after the fourth dose of a mRNA vaccine against COVID-19 has not been adequately described in elderly recipients, particularly those not exposed previously to SARS-CoV-2. Serum anti-RBD IgG levels (Abbott SARS-CoV-2 IgG II Quant assay) and neutralizing capacities (spike SARS-CoV-2 pseudovirus Wuhan and Omicron BA.1 variant) were measured after the third and fourth doses of a COVID-19 mRNA vaccine among 46 elderly residents (median age 85 years [IQR 81; 89]) of an assisted living facility. Among participants never infected by SARS-CoV-2, the mean serum IgG levels against RBD (2025 BAU/ml), 99 days after the fourth vaccine, was as high as 76 days after the third vaccine (1987 BAU/ml), and significantly higher (p = 0.030) when the latter were corrected for elapsed time. Neutralizing antibody levels against the historical Wuhan strain were significantly higher (Mean 1046 vs 1573; p = 0.002) and broader (against Omicron) (Mean 170 vs 375; p = 0.018), following the fourth vaccine. The six individuals with an Omicron breakthrough infection mounted strong immune responses for anti-RBD and neutralizing antibodies against the Omicron variant indicating that the fourth vaccine dose did not prevent a specific adaptation of the immune response. These findings point out the value of continued vaccine boosting in the elderly population


Serum anti-RBD IgG levels.
The first blood sample (Supplementary Table S1) obtained the week before the first boost (third vaccine dose) and around seven months after the first vaccine course (two doses, three weeks apart) showed that anti-RBD IgG levels were low (median 50 BAU/mL [Q1-Q3: 24; 126]) but 76 days (median) after the third vaccine dose, the median level increased by a factor of 21 (Table 2).At a median time of 99 days after the second booster (fourth dose of vaccine), the median anti-RBD IgG level increased by 1.5-fold compared to the titers achieved after the third vaccine dose.The geometric mean values were not statistically different in the two time points (Table 2, Fig. 1).
Based on the data published by Canetti et al. 3 , Gilboa et al. 17 and Grassi et al. 18 showing at least a 0.90% reduction by day of the antibody titers after the third vaccine dose among non-infected vaccinated individuals, here we applied a very conservative ratio of 1.2 for correction and reduced by 20% the values of the IgG titers of the second blood samples obtained 76 days after the first boost (third dose of vaccine) to compare it with the third Table 1.Characteristics of study participants, by COVID-19 infection status.Median [IQR = Q1;Q3], Q1 is first quarter and Q3 is third quarter; DM is diabetes mellitus; HTN is high blood pressure; Frailty score were calculated according to the literature [13][14][15] .Anti-RBD IgG response in uninfected residents.Anti-RBD IgG were measured in samples taken 76 and 99 days (median) after third (post third vaccine) and fourth (post fourth vaccine) vaccines, respectively, by Chemiluminescent Microparticle ImmunoAssay (CMIA) SARS-CoV-2 IgG II Quant (Abbott, IL, USA), expressed as binding antibody units (BAU) per ml.Each triangle represents one individual.There was initially no significant difference in mean antibody levels between blood sample taken after the third and the fourth vaccines.After time correction of values post third vaccine (20% lower), the increase between third and fourth vaccines became significant.Mean values are indicated below the x-axis.Bars represent 95%CI around the mean.ns: p = 0.8587; *p = 0.0301.

Participants never infected
Frailty and anti-RBD antibodies levels.The population studied had frailty scores from 0 to 5 with no severe frailty at all (Table 1 and Supplementary Table S1).There was no difference in anti-RBD IgG titers between frailty categories for all three samples (Kruskal-Wallis p = 0.648, 0.158, 0.802, respectively, for the entire cohort; p = 0.611, 0.189, 0.376, respectively, for the 40 patients who were not infected by COVID-19) (Fig. 2 and Supplementary Fig. S2).
Neutralizing antibodies (nAb).The nAb levels followed a more striking evolution.For the 40 residents not previously infected by SARS-CoV-2, if the nAb levels against the Wuhan based pseudovirus (psVWT) were low or under the LOD for the majority of the residents a week before the third dose of vaccine (data not shown), the NT 50 titers 10-11 weeks (76 days median) after the third vaccine dose reached a median of 669 [Q1-Q3: 333; 1629] and emerged for the Omicron BA.1 based pseudovirus (psVBA1) with a median of 50 [Q1-Q3: 2; 234] (Table 2).The increase in nAb was even stronger and more statistically significant 14 weeks (99 days, median) after the fourth dose of the vaccine with a 1.7-fold increase in NT 50 titers (median) for the psVWT versus three doses (Wilcoxon signed ranks test p = 0.001) and a fourfold increase in NT 50 titers (median) for the psVBA.1 (Wilcoxon signed ranks test p = 0.001) (Table 2, Figs.3).
While all the residents' sera reached NT 50 titers over the LOD against the psVWT after the third and fourth dose of vaccine, still 14 out of 40 (versus 18 out of 40 after three vaccine doses) residents' sera were below the LOD (NT 50 < 60) against the psVBA1, even after the fourth vaccine (Supplementary Table S1).

Residents who contracted COVID-19.
In January first 2022, one resident was symptomatic with COVID-19 (RT-PCR test positive) but nevertheless was vaccinated with the fourth dose eight days later.Between February and April 2022, five more residents, all fourth-dose-vaccinated in January 2022, developed a laboratory confirmed COVID-19.All the vaccine escape cases of COVID-19 presented a mild to moderate disease and none required hospitalization.During that period of time, BA.1 and then BA.2 were the dominant variants.Forty residents were never found or suspected of having been exposed to SARS-CoV-2.
Fisher's exact test found no significant difference in the frailty category between patients who did not contract COVID-19 and those who did (p = 0.747).Anti-RBD IgG titers distribution according to Frailty of the whole cohort.There was no difference in antibody titers between frailty categories for all three samples (Kruskal-Wallis p = 0.648, 0.158, 0.802, respectively, for the entire cohort; p = 0.611, 0.189, 0.376, respectively, for patients who were not infected by SARS-CoV-2, Supplementary Fig. S2).Each circle represents one individual.Bars (horizontal) represent the median values, dashes represent first (lower) and third (upper) quartiles.In white the second blood sample data, in grey-the third blood sample data.ns: Kruskal-Wallis test p > 0.05.Frailty score were calculated according to the literature [13][14][15] .
Vol.:(0123456789) www.nature.com/scientificreports/For the six breakthrough cases after the third vaccine dose and before the fourth dose, the mean and median anti-RBD IgG titers were slightly lower, but not significantly, than those of the 40 non-infected residents (Table 2).The median value of anti-RBD IgG levels in the six breakthrough infections, including the individual infected a week before his fourth dose of vaccine, was 2.3-fold higher and 3.3-fold higher ((Mann-Whitney U test, p = 0.002) than among the 40 residents who were never infected but were vaccinated four times (Table 2).
For the six residents who had a breakthrough infection, the NT 50 values for the third blood sample against psVWT were significantly higher than for the second sample (Wilcoxon signed ranks test p = 0.028) (Table 2).The response in nAb was even more important for the psVBA.1 with NT 50 levels reaching close to half of the titer levels of the psVWT titers although five out of six breakthrough individuals had anti Omicron NT 50 below the LOD prior to the breakthrough infection (Table 2, Fig. 5).
In summary, the NT 50 titers were increased both against WT and BA.1 even eight days after the infection and persisted at least 151 days after the breakthrough infection (Table 2 and Supplementary Table S1).
As a control of potential earlier exposure to SARS-CoV-2, we measured the presence of IgG antibodies against the nucleocapsid antigen, which is not present in the vaccine formulation.The levels were below the limit of detection in all samples except for one from an individual with a breakthrough infection (data not shown).
Altogether, if looking at paired data for anti-RBD IgG, WT nAb and BA.1 nAb levels for each resident, and if we take into account the results for two pairs out of three, six individuals were down, 11 equal and 23 up.Therefore, we can consider that 58% benefited, 27% regained previous status, and 15% did not improve after the fourth dose of vaccine (Supplementary Table S1).Altogether, 85% of the residents recovered or improved their immune response after the fourth dose of vaccine, as compared to their immune status after the third dose.

Discussion
The data presented, based on paired blood samples collected after the third and the fourth doses of the same mRNA vaccine, describe the humoral response of a well-protected elderly population (85 years, median) of whom 87% (n = 40) were never infected by SARS-CoV-2.Subsequently, as the residents were tested on a weekly  19 .
Comparing this study to other published data is complex.The first challenge is the timing between vaccination and blood sampling that are different among the different studies 1,3,4,10,[20][21][22] .Additionally, serological data expressed in BAU/ml are not always comparable when using different technical platforms 23 .Finally, although data obtained by nAb determination assays based on pseudovirus VSV constructs or based on live viruses are well correlated with each other, the pseudovirus lentivirus-based assays are less correlated 24 .
In the elderly population studied, the mean anti-RBD IgG levels two and a half months after the third dose and three and a half months after the fourth dose were 1987 BAU/ml (95% CI 1462; 2512) and 2025 BAU/ml (95% CI 1522; 2528) respectively and were grossly half of the values found by Eliakim-Raz et al. 20 , in a younger population (median age of 70 y [Q1-Q3: 66-74 years]) two weeks after the third and the fourth dose of the BNT162b2 vaccine, 3926 BAU/ml and 4161 BAU/ml respectively 20 .However, the increase between the third dose and the fourth dose of the mean anti-RBD IgG titers were comparable to our data, 6% and 2% respectively.These differences can be explained by the fact that the blood sampling in the study of Eliakim-Raz et al. 20 was done two weeks after the mRNA vaccination which corresponds to the peak of the antibody response and not 10-14 weeks after vaccination as we did, which corresponds to a timing when anti-RBD IgG and nAb have decreased from their peak but reached a stable plateau over a long period of time 3,25 .Indeed, Tut et al. 26 showed in a study of paired samples following the third dose of an mRNA vaccine that in a median time between samples of 77 days, [73;83 ], anti-RBD titers fell by 58% from 1619 BAU/ml to 684 BAU/ml in an infection-naive group 26 .

Figure 4.
Neutralizing antibodies and frailty scores of the whole cohort.There was no difference in NT 50 against WT between frailty categories for blood samples two (in white) or three (in grey) (Kruskal-Wallis p = 0.125, 0.975, respectively, for the entire cohort; p = 0.195, 0.863, respectively, for patients who were not infected by SARS-CoV-2, Supplementary Fig. S3).Likewise there was no difference in NT 50 against Omicron between frailty categories for blood samples two or three (Kruskal-Wallis p = 0.279, 0.964, respectively, for the entire cohort; p = 0.245, 0.725, respectively, for patients who were not infected by SARS-CoV-2, Supplementary Fig. S3).White rectangle: second blood sampling at 76 days (median) after the third vaccine.Grey rectangle: third blood sampling after 99 days (median) after the fourth vaccine.Each circle represents one individual.Horizontal bars represent the median values, lower and upper borders of the boxes represent the first and third quartile, respectively.Whiskers represent the minimal (lower) and maximal (upper) values.ns: Kruskal-Wallis test p > 0.05.Frailty score were calculated according to the literature [13][14][15]  www.nature.com/scientificreports/Nevertheless, in another study with blood samples taken 56 days median after the third dose of an mRNA vaccine, Bruel et al. 27 reported anti-Spike IgG levels in those without breakthrough infection of 2528.3BAU/mL (range: 695.4-8832.0)which are similar to those we found 76 days (median) after the third dose of vaccine e.g.1987 BAU/ml (95% CI 1462; 2512), considering the time lapse in the sampling in the two studies −20 days, and the consistently lower amount of anti-RBD IgG found when compared to anti-Spike IgG levels 27 .In addition, in the same study of Bruel et al. 27 , the data were also similar to what we found for breakthrough cases who had lower BAU values after the third vaccine dose than individuals without breakthrough infection, e.g., 1429.9BAU/ ml (670.9-3818.3)vs 2528.3BAU/ml (range 695.4-8832.0) 28alike in our study with 1064 BAU/ml (95% CI 571; 1557) and 1987 (95% CI 1462; 2512) BAU/ml respectively (Table 2).
Even more, similar BAU levels to those reported in this study were found among a much younger population of health care workers (60.8 median age) and using the same vaccine (BNT162b2 vaccine) 1 with levels of anti-RBD antibodies of 2102 BAU/ml four weeks after the third dose versus 1987 BAU/ml (95% CI 1462, 2512), 11 weeks after the first boost in the present study and 2684 BAU/ml (2372-3038) three weeks after the fourth dose versus 2025 BAU/ml (95% CI 1522; 2528), 14 weeks after the fourth dose in the present study.
Finally, antibody decay after the first and the second boost of mRNA vaccines have been documented in the literature 3,17,18,22 .So accordingly, the anti-RBD IgG levels obtained after the second boost were significantly higher than the anti-RBD IgG levels after the first boost, when the latter were corrected for elapsed time (Table 2, Fig. 1).
Altogether infection-naïve elderly people, even at an advance age (85 median) but with no profound Frailty (Table 1), developed anti-RBD antibodies after the third dose of an mRNA vaccine at levels comparable to those found in younger populations 1,10,29 and three and a half months after the fourth dose of vaccine regained or slightly improved the level of anti-RBD IgG that they had two and a half months after the third dose of vaccine (Table 2, Fig. 1).
Neutralizing antibodies have been correlated to protection against severe COVID-19 disease more than anti RBD IgG levels as shown by Gilbert et al. 30 and by Nugent et al. 28 but were not always assessed, especially in elderly populations, in the studies evaluating booster vaccine efficacy [5][6][7][8][9] .
In the previously mentioned population of health workers with a median age of 60.8 years Regev-Yochay et al. 1 , using a live virus-based assays, nAb titers of 1542 against WT and of 136.3 against BA.1 two weeks after four doses of vaccine are comparable to the titers developed by the elderly residents in this study, 1573 against WT and 375 against BA.1 (based on psVSV based assays) even taking into account the differences of methodologies 24 .Bruel et al. 27 found that two months after the 3rd dose of vaccine in un-infected elderly population, EC50 of 1283 and 188 titers respectively for Delta and Omicron which are similar to the titers of 1046 for WT and of 169 for Omicron (Table 2) found after the third dose of vaccine in the present study.Similarly to what Bruel et al. 27 described, we also found that the breakthrough infection cases had lower nAb after the third dose (696 and 9 for WT and Omicron respectively) (Table 2) compared to non-infected individuals and that it may be correlated to frailty scores.www.nature.com/scientificreports/Definitively, the fourth dose notably increased the level of neutralizing antibodies against not only the SARS-CoV-2 wild type but also against the variant BA.1 (Fig. 3) in COVID-19 naïve elders, and vaccine escape infections resulted in enhanced immune response (Table 2, Fig. 5), supporting the fact that the immune system of these elders did not show signs of exhaustion.
As the Elderly are the most at-risk population, their ability to respond to four vaccine doses delivered within a year speaks to a robustness of immune response, rather than exhaustion.As such, part of a multifaceted response to the COVID-19 pandemic should continue to include booster schedules.There is a definitive need for new and better vaccine compositions that would include new antigens beside the spike antigen and antigens shared by many variants providing a wider protection for emerging and future variants.

Methods
Study design and participants.This study was performed within a cohort of elderly residents living in an assisted living facility where there were no confirmed COVID cases among residents at time of recruitment (May-August 2021).Participants were interviewed to complete a structured questionnaire, which included sociodemographic characteristics, current health condition, a measurement of frailty and known exposures to COVID-19 and subsequent tests and infections [13][14][15] .Three blood samples were collected between August 2021 and May 2022.The first blood sample was taken at a median of 194 days after the second vaccine dose (and a week before the third vaccine dose), the second blood sample was taken at a median of 76 days after the third vaccine dose, and the third blood sample was taken at a median of 99 days after the fourth vaccine dose (Supplementary Fig. S1).Within the framework of the MoH "Defense for Our Parents" Guidelines 16 , all participants were regularly screened for COVID-19, using either Real-Time Polymerase Chain Reaction (RT-PCR) or antigen tests.People who were immunocompromised were excluded from the cohort study.
The assisted living facility was without known COVID-19 cases at time of recruitment but cases were detected starting January 2022, during the Omicron wave.Among cohort study participants who contributed all three blood samples, six participants were diagnosed with COVID-19 between January and April 2022, one participant a week before the fourth dose and five after the fourth dose (Supplementary Table S1).All the vaccine escape cases of COVID-19 presented a mild to moderate disease and none required hospitalization.The remaining 40 participants were not diagnosed with COVID-19, confirmed by weekly antigen tests.It is important to note that this private institution has permanent medical staff (physician and nurses) dedicated to the elderly and is composed of private units (studio apartments), which facilitated residents' ability to isolate throughout the pandemic response.
Our research protocol was approved by the National Committee for Human Medical Research (The Israeli Minister of Health-MOH) in the office of the Chief Scientist (CoR-MOH-058-2020) and performed in accordance with the Declaration of Helsinki.Informed consent was obtained from all participants and/or their legal guardians.The study was performed in compliance with the provisions of the Declaration of Helsinki from the World Medical Association and good clinical practice (GCP) guidelines.All methods were performed in accordance with the relevant guidelines and regulations.
Serology.Blood samples were collected in VACUETTE 8 ml CAT Serum Separator Clot Activator tubes and after coagulation at room temperature, the tubes were centrifuged (21 °C, 3500g, 10 min).Sera were aliquoted and immediately frozen at − 74 °C.After one cycle of freeze-thawing, sera were tested for immunoglobulin G (IgG) antibodies against the SARS-CoV-2 spike RBD using the commercial Chemiluminescent Microparticle ImmunoAssay (CMIA) SARS-CoV-2 IgG II Quant (Abbott, IL, USA), according to the manufacturer's instructions at the certified laboratory of the Hebrew University medical school affiliated Shaare Zedek Medical Center (Jerusalem, Israel).The antibody titers were obtained in arbitrary antibody units (AU/ml) and were converted by multiplying them by a factor of 0.142 (according to the manufacturer's instructions) to Binding Antibody Unit (BAU/ml), adapted to the WHO standard for SARS-CoV-2 immunoglobulin 33 .Anti-Nucleocapsid (N) IgG antibodies were evaluated in a made-in-house Enzyme-Linked Immunosorbent Assay (ELISA) test, as described in detail by Stolovich-Rain et al. 40 .
Neutralizing antibodies.Neutralization assay was performed using propagation-incompetent, singleround infectious particles, i.e.-vesicular stomatitis virus spike SARS-CoV-2 pseudovirus (psV) with GFPreporter 40 , pseudotyped with the historical Wuhan (psVWT) or the BA.1 sequence (psVBA.1)that were constructed as described by Stolovich-Rain et al. 40 .Neutralization assays based on psSARS-CoV-2 have been shown to be highly correlative to authentic SARS-CoV-2 virus micro-neutralization assay 40 .Following titration, 100-300 focus forming units (FFU) of psSARS-2 were incubated with two to threefold serial dilutions of heat inactivated (56 °C for 30 min) tested sera.After incubation for 60 min at 37 °C, virus/serum mixture was transferred to Vero E6 cells grown to 75-80% confluence in 96-well plates.The plates were incubated for 18-22 h, and 50% focus reduction titer (NT 50 ) was calculated by counting green fluorescent foci using an automated fluorescence microscope (Cytation5, Agilent, BioTek) 24,41 .NT 50 was calculated using the method established by Reed and Muench 42 .Sera not capable of reducing viral replication by 50% at a final dilution of 1:60 were considered non-neutralizing; limit of detection (LOD) = 60.This value is two to three times-fold higher than the value corresponding to a level of protection against infection or serious disease as described by Khoury et al. 25 .For a clear graphical presentation, samples below the LOD were marked as a titer of two 42 .

Statistical analysis.
Descriptive statistics were presented as frequencies and percentages for categorical variables and as median values with interquartile range or mean values with standard deviations for continuous variables, as appropriate.95% Confidence Intervals (CIs) were presented when performing a graphical

FitFigure 2 .
Figure 2.Anti-RBD IgG titers distribution according to Frailty of the whole cohort.There was no difference in antibody titers between frailty categories for all three samples (Kruskal-Wallis p = 0.648, 0.158, 0.802, respectively, for the entire cohort; p = 0.611, 0.189, 0.376, respectively, for patients who were not infected by SARS-CoV-2, Supplementary Fig.S2).Each circle represents one individual.Bars (horizontal) represent the median values, dashes represent first (lower) and third (upper) quartiles.In white the second blood sample data, in grey-the third blood sample data.ns: Kruskal-Wallis test p > 0.05.Frailty score were calculated according to the literature[13][14][15] .

Figure 3 .
Figure 3. Wild type and Omicron neutralizing antibodies after the third and fourth vaccines in non-infected residents.WT: wild type SARS-CoV-2.Omicron: BA.1 variant.Neutralization capacity [NT 50 ] is expressed as a function of reciprocal values of sera dilutions on a log 10 scale.Starting at a final 1:60 dilution of the sera, the participants were assessed for both WT and Omicron neutralization antibodies by SARS-CoV-2 spikepseudotyped VSV-GFP-ΔG reporter assay on Vero-E6 cells.Sera with NT 50 that could not be calculated at the 1:60 dilutions were graded 2 (LOD) for graphical representations.Post 3rd vaccine: blood sampling was done at a median time of 76 days after the third vaccine.Post 4th blood: sampling was done at a median time of 99 days after the fourth vaccine.Each point corresponds to on individual.Bars represent 95%CI around the mean.**WT paired p-value = 0.0023; *Omicron paired p-value = 0.0179.

Figure 5 .
Figure5.WT and Omicron neutralizing antibodies in residents before and after breakthrough infections.Serum neutralization in breakthrough infections was assessed for both wild type SARS-CoV-2 (WT) and BA.1 variant (Omicron) neutralization capacity by SARS-CoV-2 spike-pseudotyped VSV-GFP-ΔG reporter assay on Vero-E6 cells.NT5 50 is expressed as a function of reciprocal values of sera dilutions on a log2 scale.Each circle or square represents one individual.Post 3rd vaccine: blood sampling was done at a median time of 76 days after the third vaccine.Post 4th blood: sampling was done at a median time of 99 days after the fourth vaccine.Mean values are indicated below the x-axis.*WT paired p-value = 0.0222; ns: Omicron (BA.1) paired p-value = 0.0531 but the median difference was highly significant (Wilcoxon signed ranks test p = 0.028) (Table2).