Accelerated waning of the humoral response to COVID-19 vaccines in obesity

EAVE II study

The EAVE II surveillance platform drew on near-real-time nationwide healthcare data for 5.4 million individuals (~99%) in Scotland26,27,28,29. It includes information on clinical and demographic characteristics of each individual, their vaccination status and type of vaccine used and information on positive SARS-COV-2 infection and subsequent hospitalization or death. Ethical approval was granted by the National Research Ethics Service Committee, Southeast Scotland 02 (reference 12/SS/0201) for the study using the EAVE II platform. Approval for data linkage was granted by the Public Benefit and Privacy Panel for Health and Social Care (reference 1920-0279). Individual written patient consent was not required for this analysis.

Using data from the EAVE II platform, we examined the impact of obesity (using BMI measurements) and clinical and demographic characteristics, including time since receiving the second and third vaccine doses, previous history of testing positive for COVID-19, gap between vaccine doses and dominant variant in the background, of fully vaccinated adults in Scotland who experienced severe COVID-19 outcomes. The cohort analyzed for this study consisted of individuals aged 18 years and older who were administered with at least two doses of BNT162b2 mRNA, ChAdOx1 nCoV-19 or mRNA-1273 vaccines between 8 December 2020 and 19 March 2022. Follow-up began 14 d after receiving the second dose until COVID-19-related hospitalization, COVID-19-related death or the end of study period (that is, 19 March 2022). All the COVID-19-related hospital admissions or deaths were selected between 14 September 2021 and 19 March 2022. We excluded events that occurred within the first 14 d after completing the primary vaccination schedule to allow for time for a full immune response to be mounted. Patients without immunosuppression had their primary vaccination schedule with two doses, and so the third dose is a booster. For people with immunosuppression, the primary vaccination schedule was for three vaccine doses. BMI was available for individuals based on last recorded measurement within their primary care record. Where the BMI was missing, it was imputed using ordinary least squares regression with all QCovid risk groups39 together with age, sex and deprivation included as predictors. The coding systems used in Scotland are Read for GP data and ICD-10 for hospitalization data. This information is provided in more detail in the EAVE II protocol and cohort profile39 and data dictionary (https://www.ed.ac.uk/usher/eave-ii) and at https://github.com/EAVE-II/EAVE-II-data-dictionary.

Definition of outcomes

The primary outcome of interest was severe COVID-19, which was defined as COVID-19-related hospital admission or death 14 d or more after receiving the second vaccine or booster dose40. COVID-19-related hospital admission was defined as hospital admission within 14 d of a positive RT–PCR test or COVID-19 as reason for admission or a positive SARS-CoV-2 RT–PCR test result during an admission where COVID-19 was not the reason for admission. COVID-19-related mortality was defined as either death for any reason within 28 d of a positive RT–PCR test or where COVID-19 was recorded as the primary reason for death on the death certificate.

Population characteristics and confounders

Characteristics of interest were defined at baseline on 8 December 2020 and included age, sex, socioeconomic status based on quintiles of the SIMD, urban or rural place of residence (which is a measure of rurality based on residential settlement), BMI, previous natural infection from SARS-CoV-2 before second dose of the vaccine (classified as 0–3 months, 3–6 months, 6–9 months and ≥9 months before second vaccine dose), number of pre-existing comorbidities known to be linked with severe COVID-19 outcome and being in a high-risk occupational group, defined as someone with undergoing regular RT–PCR testing27. Time since vaccination was distributed in the periods of 3–10 weeks, 11–15 weeks, 16–20 weeks and ≥21 weeks from completion of second dose of the primary vaccination and 3–5 weeks, 6–8 weeks and ≥9 weeks for the booster doses separately. To allow for variation in background levels of community infection, we split the data by calendar week. BMI was grouped as <18.5 kg/m2 (underweight), 18.5–24.9 kg/m2 (normal weight), 25–29.9 kg/m2 (overweight), 30–39.9 kg/m2 (obese) and ≥40 kg/m2 (severely obese) according to World Health Organization (WHO) criteria.

EAVE II statistical analysis

We calculated the frequency and rate per 1,000 person-years of severe COVID-19 outcomes for all demographic and clinical factors. Generalized linear models (GLMs) assuming a Poisson distribution with person-time as an offset representing the time at risk were used to derive rate ratios (RRs) with 95% CIs for the association between demographic and clinical factors and COVID-19-related hospitalization or death. aRRs were estimated adjusting for all confounders, including age, sex, SIMD, time since receiving the second dose of vaccine, pre-existing comorbidities, the gap between vaccine doses, previous history of SARS-CoV-2 infection and calendar time. SIMD looks at the extent to which an area is deprived across seven domains: income, employment, education, health, access to services, crime and housing. SIMD was allocated based on an individual’s home postcode, with quintiles of population ranging from 1 for the most deprived 20% to 5 for the least deprived 20% of the population.

We conducted a post hoc sensitivity analysis in clinically confirmed cases. Because BMI was not available for all the individuals, missing BMI was imputed using ordinary least squares regression with all other independent variables included as predictors. We conducted a post hoc sensitivity analysis by imputing the missing BMI using an average of 10 least squares regressions (multiple imputation). R (version 3.6.1) was used to carry out all statistical analyses.

Ethical approval and study populations

SCORPIO clinical study

Clinical studies in individuals with severe obesity and normal BMI controls were approved by the National Research Ethics Committee and the Health Research Authority (East of England–Cambridge Research Ethics Committee (SCORPIO study amendment of ‘NIHR BioResource’ 17/EE/0025)). Human tonsil samples were collected under ethical approval from the UK Health Research Authority (reference 16/LO/0453). Each participant provided written informed consent. All studies were conducted in accordance with the Declaration of Helsinki and the guidelines for Good Clinical Practice.

Individuals with severe obesity (class II/III WHO criteria of BMI ≥ 40 kg/m2 or BMI ≥ 35 kg/m2 with obesity-associated medical conditions, such as type 2 diabetes and hypertension) who attended the obesity clinic at Cambridge University Hospitals NHS Trust and had received two doses of COVID-19 vaccination (first and second doses of ChAdOX1 nCoV-19 or BNT162b2 mRNA) between December 2021 and May 2022 were invited to take part. Individuals with acquired (HIV, immunosuppressant drugs) or congenital immune deficiencies and cancer were excluded. Third-dose vaccinations (BNT162b2, Pfizer BioNTech or half-dose mRNA1273 (Moderna)) were administered as part of the NHS vaccination program. UK Health Security Agency policy recommends the use of longer needles (38 mm versus 25 mm) in individuals with severe obesity.

Additional normal BMI controls were recruited in Oxford, UK, as part of the PITCH study under the GI Biobank Study 16/YH/0247, approved by the Yorkshire & Humber Sheffield Research Ethics Committee, which was amended for this purpose on 8 June 2020. Samples obtained 6 months after the primary course were included.

Clinical and immunological measurements were taken before the booster vaccination and 8 d (−3), 28 d (±7) and 105 d (±7) after vaccination. Third-dose vaccinations were administered as part of the NHS vaccination program and were mRNA vaccines (BNT162b2 or mRNA1273 (Moderna)). Clinical data regarding comorbidities associated with obesity were obtained from the medical records. Supplementary Data Table 8 details the demographic characteristics of this cohort. Healthy healthcare workers were enrolled into the longitudinal OPTIC cohort in Oxford, UK, between May 2020 and May 2021 as part of the PITCH consortium. PITCH participants were sampled between July and November 2021, a median of 185 d (range, 155–223) after receiving a second vaccination with ChAdOX1 nCoV-19 or BNT162b2 mRNA vaccine. All PITCH participants were classified as infection-naive, as defined by never having received a positive lateral flow or PCR test for SARS-CoV-2 and negative anti-nucleocapsid antibodies at the time of their first vaccination. Therefore, a total of 28 individuals with severe obesity and 41 normal BMI controls were evaluated 6 months after the primary course of vaccination, whereas, for the response to third-dose vaccination, 16 normal BMI controls were studied.

Of the 28 recruited individuals with severe obesity, two had positive anti-nucleocapsid antibodies and reported a positive RT–PCR test before their third-dose vaccination. They were excluded from further analysis. In addition, between day 28 and day 105, two individuals with severe obesity reported positive SARS-CoV-2-tests (lateral flow test or RT–PCR tests as per UK guidelines at the time). They were excluded from the day 28 to day 105 analysis. An additional three people with severe obesity were recruited after they had had their booster for day 28 and day 105 analysis only. In addition, one of the normal-weight individuals had positive anti-nucleocapsid antibodies who had not had a PCR test, before their third-dose vaccination. This individual was excluded from the before and after third-dose analysis. In addition, between day 28 and day 105, two normal-weight individuals reported positive SARS-CoV-2-tests (lateral flow test or PCR tests as per UK guidelines at the time, one of those individuals on two separate occasions). They were excluded from the day 28 to day 105 analysis. Missing data in addition to this were due to (1) occasional difficult venepuncture in individuals with obesity; (2) insufficient peripheral blood mononuclear cells (PBMCs) isolated for both T and B cell analysis; or (3) insufficient sample to run both wild-type and Omicron neutralization assays. Therefore, we specified how many individuals were included per analysis, per figure.

Peripheral blood samples were acquired in either lithium heparin (PBMCs) or serum-separating tubes. Serum tubes were centrifuged at 1,600g for 10 min at room temperature to separate serum from the cell pellet before being aliquoted and stored at −80 °C until use. PBMCs were isolated by layering over Lymphoprep density gradient medium (STEMCELL Technologies), followed by density gradient centrifugation at 800g for 20 min at room temperature. PBMCs were isolated and washed twice using wash buffer (1× PBS, 1% FCS, 2 mM EDTA) at 400g for 10 min at 4 °C. Isolated PBMCs were resuspended in freezing media, aliquoted and stored at −80 °C for up to 1 week before being transferred to liquid nitrogen until use.

SARS-CoV-2 serology by multiplex particle-based flow cytometry

Recombinant SARS-CoV-2 nucleocapsid, spike and RBD were covalently coupled to distinct bead sets (Luminex) to form a three-plex and analyzed as previously described41. Specific binding was reported as mean fluorescence intensity (MFI).

Neutralizing antibodies to SARS-CoV-2

Luminescent HEK293T-ACE2-30F-PLP2 reporter cells (clone B7) expressing ACE2 and SARS-CoV-2 papain-like protease-activatable circularly permuted firefly luciferase (FFluc) are available from the National Institute for Biological Standards and Control (NIBSC, https://www.nibsc.org/, cat. no. 101062)30. They were cultured in IMDM supplemented with 4 mM GlutaMAX (Gibco), 10% FCS, 100 U ml−1 penicillin and 0.1 mg ml−1 streptomycin at 37 °C in 5% CO 2 , regularly screened and confirmed to be mycoplasma negative (Lonza MycoAlert).

The SARS-CoV-2 viruses used in this study were a wild-type (lineage B) isolate (SARS-CoV-2/human/Liverpool/REMRQ0001/2020), a kind gift from Ian Goodfellow (University of Cambridge), isolated by Lance Turtle (University of Liverpool), David Matthews and Andrew Davidson (University of Bristol)42,43, and an Omicron (lineage B.1.1.529) variant, a kind gift from Ravindra Gupta44. Unless otherwise indicated, all data shown refer to neutralization of wild-type virus.

Sera were heat inactivated at 56 °C for 30 min before use, and NT 50 values were measured as previously described30,45. In brief, luminescent HEK293T-ACE2-30F-PLP2 reporter cells (clone B7) expressing SARS-CoV-2 papain-like protease-activatable circularly permuted FFluc were seeded in flat-bottomed 96-well plates. The next day, SARS-CoV-2 viral stock (multiplicity of infection (MOI) = 0.01) was pre-incubated with a three-fold dilution series of each serum for 2 h at 37 °C and then added to the cells. Sixteen hours after infection, cells were lysed in Bright-Glo Luciferase Buffer (Promega) diluted 1:1 with PBS and 1% NP-40, and FFluc activity was measured by luminometry.

Experiments were conducted in duplicate. To obtain NT 50 values, titration curves were plotted as FFluc versus log (serum dilution) and then analyzed by nonlinear regression using the Sigmoidal, 4PL, X is log(concentration) function in GraphPad Prism. NT 50 values were reported when (1) at least 50% inhibition was observed at the lowest serum dilution tested (1:10) and (2) a sigmoidal curve with a good fit was generated. For purposes of visualization and ranking, samples with no neutralizing activity were assigned an arbitrary NT 50 value of 2. Samples for which visual inspection of the titration curve indicated inhibition at low dilutions, but that did not meet criteria (1) and (2) above, were assigned an arbitrary NT 50 value of 4.

To confirm the linearity of the assay, a high-titer positive control serum sample was spiked into FCS (serum dilution series), and then each dilution was treated as a separate sample. Expected and obtained NT 50 values against wild-type SARS-CoV-2 were compared by linear regression, generating a coefficient of determination (R2) of 1.00 for dilutions above the limit of quantification (Extended Data Fig. 8a,b).

As a measure of intermediate precision46, inter-assay variability was quantified for a medium-titer control serum sample tested against wild-type SARS-CoV-2 in 18 independent experiments conducted over a period 18 months by two different laboratory scientists, revealing a coefficient of variation (CV) of 27% (Extended Data Fig. 8c).

For external validation, a panel of 28 serum samples from NHS Blood and Transplant convalescent plasma donors participating in the C-VELVET study (approved by the West Midlands Solihull Research Ethics Committee, reference 21/WM/0082, IRAS project ID 296926) was tested blinded against both wild-type and Omicron variant SARS-CoV-2. NT 50 values were compared with previously reported focus reduction neutralization test (FRNT) results obtained at the University of Oxford47, revealing a Spearman’s rank correlation coefficient (rho) of 0.9696 (Extended Data Fig. 8d).

Finally, to enable comparison with other studies, the neutralizing capacity of WHO International Standard 20/136 against wild-type SARS-CoV-2 was measured in five independent experiments, yielding a geometric mean NT 50 of 1,967 (Extended Data Fig. 8e,f). This standard comprises pooled convalescent plasma obtained from 11 individuals which, when reconstituted, is assigned an arbitrary neutralizing capacity of 1,000 IU ml−1 against early 2020 SARS-CoV-2 isolates48. NT 50 values against wild-type SARS-CoV-2 from this study may, therefore, be converted to IU ml−1 using a calibration factor of 1,000/1,967 (0.51), with a limit of quantification of 5.1 IU ml−1 (corresponding to an NT 50 value of 10).

T cell cytokine production

Antigen-specific T cell responses were assessed using an ELISpot assay as previously described49. Results are expressed as spot forming units (SFU) per 106 PBMCs. Analysis was completed using GraphPad Prism software version 9.3.1. The comparison of means between groups was performed using two-way, mixed-model ANOVA.

Quantitation of lymphocyte types and subsets by spectral flow cytometry

Generation of RBD-specific B cell probes and measurement of RBD-specific B cells were measured by high-dimension flow cytometry as described previously50,51. In brief, for flow cytometry stains, a single-cell suspension was prepared from cryopreserved PBMC samples as follows. First, 1 ml of PBMC samples was de-frosted in a 37 °C water bath and then immediately diluted into 9 ml of pre-warmed RPMI + 10% FBS. Cells were washed twice with 10 ml of FACS buffer (PBS containing 2% FBS and 1 mM EDTA). Cells were then resuspended in 500 µl of FACS buffer, and cell numbers and viability were determined using a Countess automated cell counter (Invitrogen). Next, 5 × 106 viable cells were transferred to 96-well plates for antibody staining (dilutions used are provided in Supplementary Table 9). Cells were then washed once with FACS buffer and stained with 100 µl of surface antibody mix (including B cell probes) for 2 h at 4 °C. Cells were then washed twice with FACS buffer and fixed with the eBioscience Foxp3/Transcription Factor Staining Buffer (Thermo Fisher Scientific, 00-5323-00) for 30 min at 4 °C. Cells were then washed with 1× eBioscience Foxp3/Transcription Factor Permeabilization Buffer (Thermo Fisher Scientific, 00-8333-56) twice and stained with intracellular antibody mix in permeabilization buffer at 4 °C overnight. After overnight staining, samples were washed twice with 1× permeabilization buffer and once with FACS buffer and acquired on a Cytek Aurora. Cells for single-color controls were prepared in the same manner as the fully stained samples. Manual gating of flow cytometry data was performed using FlowJo version 10.8 software (Tree Star).

SCORPIO study statistical analysis

Analysis was completed using GraphPad Prism software version 9.3.1. The comparison of means or medians between groups was performed using two-sided parametric t-tests, non-parametric Mann–Whitney U-tests or mixed-model tests when appropriate. tSNE, FlowSOM and heat map analyses were performed using R (version 4.1.2), using code that was previously described52.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.