Since July 2020, a repeatitive serological survey has been being conducted to monitor the spread of SARS-CoV-2 infection among the staff members of the National Center for Global Health and Medicine (NCGM), Japan [16,17,18]. Participants were asked to donate a blood sample for the measurement of anti-SARS-CoV-2 antibodies. We also collected the information on medical history, health-related lifestyle, and COVID-19 (e.g., COVID-19 infection and vaccination) via an online questionnaire. The participation in the survey was completely voluntary; and a written informed consent was obtained from each participant.

In the present study, we used data of the third round of survey conducted in June 2021, 2 months after the completion of an in-house vaccination program (Pfizer–BioNTech). We additionally obtained annual health check-up information which was collected in the same year as the survey (June 2021). Eligible participants were NCGM’s staff of all occupations (including doctors, nurses, administrative staff, and allied healthcare professionals) who had completed two doses of the vaccine. We excluded those who disagreed to provide their health check-up data, received antibody test within 14 days of the second vaccination, or lacked information on MetS components or covariates.

We quantitatively measured IgG (AU/mL) against the SARS-CoV-2 spike protein, using AdviseDx SARS-CoV-2 IgG II assay, Abbott ARCHITECT®. In a subgroup of the vaccine recipients (n = 68), the Spearman’s rank correlation coefficient (95% CI) between the above SARS-CoV-2 spike IgG titer and neutralizing antibody titer was 0.497 (0.286–0.661), 0.250 (0.005–0.467), and 0.683 (0.526–0.795) against Wuhan, Alpha, and Delta strains, respectively. We also qualitatively measured antibodies against SARS-CoV-2 nucleocapsid protein using the SARS-CoV-2 IgG assay (Abbott) and used these data to identify those with possible infection. The sensitivity and specificity for the identification of past infection with SARS-CoV-2 viruses using this assay were 100% and 99.9%, respectively [19].

The information on MetS components, i.e., waist circumference (WC), blood pressure (BP), fasting plasma glucose (FPG), triglycerides (TG), and high-density lipoprotein cholesterol (HDL-C), was collected during the health check-up. WC was measured at the umbilical level in a standing position using a measuring tape (maximum:150 cm); Systolic and diastolic BP were measured with an automated sphygmomanometer (HEM-907, Omron Health Care Co. Ltd., Kyoto, Japan); FPG was measured using an enzymatic (Hexokinase UV) method (Cica Liquid GLU, Kanto Chemical Co., Tokyo, Japan); TG level was measured by an enzymatic method using the Pureauto S TG-N (Sekisui Medical Co., Ltd., Tokyo, Japan); and HDL-C concentration was measured by a direct enzymatic method using the Cholestest-N HDL (Sekisui Medical Co., Ltd., Tokyo, Japan).

MetS was defined, according to the Joint Interim Statement [20], as a clustering of any three or more of the following components: high FPG (≥ 100 mg/dL or using anti-diabetic medication), central obesity (WC ≥ 90 cm for men, or ≥ 80 cm for women), high TG (≥ 150 mg/dL or using lipid-lowering medication), high BP (systolic BP ≥ 130 mmHg, diastolic BP ≥ 85 mmHg or using antihypertensive medication) and reduced HDL-C (< 40 mg/dL for men or < 50 mg/dL for women). The cut-off values for WC were based on the recommendation of the World Health Organization for Asian populations [21].

We selected covariates according to epidemiological evidence for their association with the immune response to SARS-CoV-2 vaccines: age, sex [4, 5], smoking [5], alcohol drinking [22], physical activity [23], underlying comorbidities (i.e., cancer, heart, or lung diseases) [5, 24], history of SARS-CoV-2 infection [22, 25], and the time interval (in day) between the second dose of SARS-CoV-2 vaccine and the day of blood draw (vaccination-to-IgG time) [22]. The history of infection with SARS-CoV-2 was defined as the positive result of either polymerase chain reaction test or antibodies against SARS-CoV-2 nucleocapsid protein.

The background characteristics of the study population, according to MetS status, were described as arithmetic mean and standard deviation (SD), or median and range/interquantile range for continuous variables, and percentages for categorical variables.

Linear regression modeling was used to estimate the means (95% confidence interval [CI]), and the beta-coefficients (95% CI) of log10-transformed SARS-CoV-2 spike IgG titers, relative to MetS. Two models were fitted: Model 1 was adjusted for age and sex; and Model 2 was further adjusted for smoking (non-smoker, or smoker), alcohol drinking (non-drinker, drinker consuming < 23 or ≥ 23 g ethanol/day), leisure-time physical activity (non-engagement, < 150, or ≥ 150 min/week), comorbid cancer (all types), heart or lung diseases, history of SARS-CoV-2 infection, and vaccination-to-IgG time. The marginal means (95% CI) predicted from Model 2 were then back-transformed to obtain the adjusted geometric mean titer (GMT) (95% CI) of SARS-CoV-2 spike IgG. The beta-coefficients (95% CI) from Model 2 were back-transformed to obtain the geometric mean ratio (GMR) (95% CI) for SARS-CoV-2 spike IgG titer.

We also examined the association between the number of MetS components and SARS-CoV-2 spike IgG titers, using Model 1 and Model 2 in which those with five components were regrouped together with those having four components. The trend in this association was assessed by assigning an ordinal number (1–5) to each group, which was treated as a continuous variable when fitted in regression models.

To eliminate the potential impact of comorbidities and history of SARS-CoV-2 infection on the association between MetS and the immunogenicity of Pfizer–BioNTech vaccine, we conducted a sensitivity analysis using Model 1 and Model 2 in which we excluded participants with either condition. Statistical significance was set at p < 0.05 for trend and p < 0.1 for interaction tests. All statistical analyses were conducted in RStudio (version 3.2.4) using the package “emmeans” (version 1.6.3) [26].