Age-dependent association of thyroid peroxidase antibodies with incident cardiovascular disease events: Insights from a longitudinal cohort study

Abstract

Background

The relationship between thyroid autoimmunity and cardiovascular disease (CVD) is controversial, and age-specific effects are not well characterized. We examined associations between thyroid peroxidase antibodies (TPOAb) status and incident CVD.

Methods

In this prospective population-based cohort, 4825 adults aged ≥20 years without baseline CVD were identified. Survival analyses were conducted using a 1.5-year landmark approach. Mean follow-up after the landmark was 14.8 ± 3.3 years. TPOAb was analyzed as a log-transformed continuous variable, dichotomous status, and categorical levels. Cox proportional hazards models were used to estimate hazard ratios (HRs) with progressive adjustment for demographic, lifestyle, cardiometabolic, and thyroid-related factors. Effect modification by age was assessed using interaction and spline analyses.

Results

A total of 4825 participants (56.5% female and mean age of 39.7 ± 14.0 years) were included. During follow-up, 576 incident CVD events occurred. In the overall population, TPOAb was not associated with CVD risk (adjusted HR per 1-SD increase: 1.04, 95%CI 0.95-1.13). The primary continuous-age spline analysis demonstrated significant effect modification by age (global interaction P = 0.024). In exploratory age-stratified analyses, higher TPOAb levels were associated with increased CVD risk among participants aged <40 years (adjusted HR: 1.35, 95%CI 1.10-1.66), but not among those aged ≥40 years (HR, 1.00; 95% CI, 0.92–1.10). The association among younger participants remained significant in the euthyroid subgroup and in sensitivity analyses using Firth penalized regression and hard cardiovascular outcomes. Dose-response analyses further suggested increased risk at higher TPOAb levels among younger participants.

Conclusions

Thyroid autoimmunity was not associated with incident CVD in the overall cohort but demonstrated a significant age-dependent association. Exploratory analyses suggested increased risk among participants aged <40 years. These hypothesis-generating findings identify age as a potentially important modifier of the thyroid autoimmunity–CVD association and warrant validation in independent cohorts.

1. Introduction

Thyroid peroxidase antibody (TPOAb) is a hallmark marker of autoimmune thyroid disease (AITD) and reflects humoral immune reactivity against thyroid peroxidase, a key enzyme in thyroid hormone synthesis [1,2]. In population-based studies, the prevalence of TPOAb positivity generally ranges from 10 to 12% [3] in the general adult population, to 17-23% among women and older adults. Even among euthyroid individuals, TPOAb positivity is common, observed in approximately 11% [[3], [4], [5]], and is regarded as a predictor of future thyroid dysfunction and hypothyroidism [3,6]. Beyond its established role in identifying individuals at risk of thyroid disease, accumulating evidence suggests that TPOAb positivity may be associated with broader systemic effects, including low-grade inflammation, altered lipid metabolism, and endothelial dysfunction [[7], [8], [9], [10]].

Mechanistically, TPOAb-related autoimmunity could contribute to cardiovascular risk through immune-mediated inflammatory cascades that promote vascular injury [7,11] and atherogenesis [12]. Associations between TPOAb and cardiometabolic and vascular risk factors, such as arterial wall thickening, dyslipidemia, and atherosclerosis, have been reported even in euthyroid populations [[13], [14], [15]]. Nevertheless, findings based on subclinical vascular measures cannot be directly extrapolated to the occurrence of clinical cardiovascular events. These pathophysiologic links provide a plausible biological rationale for examining the relationship between thyroid autoimmunity and cardiovascular outcomes, even in the absence of manifest thyroid disease.

While the association of TPOAb status and CVD-related outcomes (primarily mortality rather than incident events) has been investigated in previous studies, there remains controversy regarding its clinical and statistical significance, as a recent population-based study on the ELSA dataset [16] showed no significant association between TPOAb status and CVD mortality, whereas an investigation on middle-aged individuals from the Rotterdam cohort [17] observed a statistically significant association between TPOAb detectability and cardiovascular mortality, with a more pronounced dose–response pattern in men. In addition, most of these conducted investigations [16,17] focus more prominently on middle-aged or elderly populations, and only a limited number of population-based studies have examined whether age modifies the relationship between thyroid autoimmunity and cardiovascular outcomes, and most available cohorts did not formally evaluate age-specific associations. Although a prior pooled analysis [18] explored age-stratified associations between thyroid autoimmunity and cardiovascular outcomes, age was evaluated only as a secondary stratification variable using broad <65/≥65-year categories rather than being modeled continuously. This approach may have reduced sensitivity to associations concentrated in younger adulthood. Furthermore, heterogeneity across pooled cohorts (e.g., differences in assay methods, TPOAb cut-offs, follow-up duration, and outcome ascertainment) could have obscured potential effect modification.

In the present study, we aimed to evaluate the association between thyroid peroxidase antibody (TPOAb) levels and the risk of incident cardiovascular disease (CVD) events in a prospective population-based cohort and to determine whether this association varied across age. TPOAb was modeled using continuous, categorical, and binary definitions to comprehensively assess potential dose–response relationships. Given the limited and inconsistent evidence from prior studies, most of which have focused on cardiovascular mortality rather than incident events and have predominantly included middle-aged and older populations, we further examined effect modification by age using interaction and stratified analyses. This approach was designed to determine whether thyroid autoimmunity confers differential cardiovascular risk across age groups and to better characterize its role in the development of incident CVD events, complementing prior studies focused primarily on mortality.

2. Methods

2.1. Study design and population

Our study population was derived from phases 1 (1999-2001) and 2 (2001-2005) of the Tehran Thyroid Study (TTS), a prospective cohort designed to investigate the prevalence, natural course, and associated clinical outcomes of thyroid disorders among the adult population of Tehran, Iran, an iodine-sufficient region. The TTS is conducted as a substudy within the Tehran Lipid and Glucose Study (TLGS). Detailed descriptions of the TTS and TLGS design, sampling framework, and data collection methods have been published previously [[19], [20], [21]]. According to the TTS protocol, the enrolled participants were followed up at a 3-year interval. The study comprised six examination phases spanning more than 18 years. Of the 5470 cases aged ≥20 years, 645 were excluded from the study due to CVD history (N = 187), use of systemic glucocorticoids (N = 32), missing covariates (N = 41), pregnancy (N = 13), thyroid medication use (N = 64), TSH<0.3 mIU/L (N = 296), and cancer (N = 12). The eligible baseline cohort comprised 4825 participants (Fig. 1). For participants enrolled in phase 1, phase 1 served as the index examination; for those first enrolled in phase 2, phase 2 served as the index examination. Follow-up for the survival analyses began 1.5 years after the individual index examination and continued until the first incident CVD event, non-CVD death, loss to follow-up, or administrative censoring at the end of 2018, whichever occurred first. Individual examination and event or censoring dates were used to calculate elapsed follow-up, thereby accommodating staggered calendar-time entry; time since the landmark was used as the analysis time scale.

Fig. 1.

Flowchart of participant selection for the eligible baseline cohort.

2.2. Clinical, outcome, and laboratory measurements

Baseline data, including demographic characteristics, medication use, CVD history, and smoking status, were obtained by trained interviewers using a standardized questionnaire. Anthropometric measurements were performed with participants wearing light clothing and no shoes. Body weight was measured using a calibrated digital scale (Seca 707, Seca Corp., Hanover, MD, USA; range 0.1–150 kg) and recorded to the nearest 0.1 kg, while height was measured in the standing position without shoes using a tape measure with the shoulders in normal alignment. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m²). Blood pressure was measured twice on the right arm in the seated position after at least 15 min of rest, using a standardized mercury sphygmomanometer calibrated by the Iranian Institute of Standards and Industrial Research. The mean of the two measurements was considered the participant's blood pressure [21].

Participants in the TLGS were contacted annually to document any hospitalizations or other medical events occurring during the preceding year [21,22]. A trained nurse conducted the initial inquiry, and if an event was reported, a physician subsequently collected relevant medical records through follow-up visits at home or in the hospital. Mortality data were obtained from hospital records via authorized local physicians or from official death certificates. All reported events were independently adjudicated by the outcome committee [22] and classified according to pre-specified clinical criteria, as well as the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, and American Heart Association guidelines for cardiovascular events [21,23].

Venous blood samples were collected from each participant between 7:00 and 9:00 a.m. following a 12- to 14-h overnight fast, using anticoagulant-free tubes. Samples were centrifuged within 30-45 min of collection, and serum aliquots were stored at −70 °C and subsequently transferred to the hormone laboratory of the Research Institute for Endocrine Sciences for hormonal analyses.

Thyroid peroxidase antibody (TPOAb) levels were measured using a microplate enzyme immunoassay (sequential type 1 ELISA; AccuBind™, Monobind, Costa Mesa, CA, USA) with a Sunrise ELISA reader (Tecan Co., Salzburg, Austria). The intra- and inter-assay coefficients of variation (CVs) for the TPOAb assay were 3.9% and 4.7%, respectively. All immunoassays were performed on serum samples stored at −70 °C using the same analytical method, analyzer, and commercial assay. The assay's lower limit of detection was 1.5 IU/mL; values below this limit were recorded as 1.5 IU/mL before log transformation. No post hoc batch correction was applied.

Serum free thyroxine (FT4) and thyroid-stimulating hormone (TSH) concentrations were determined by electrochemiluminescence immunoassay using Roche Diagnostics kits and a Roche/Hitachi Cobas e−411 analyzer (GmbH, Mannheim, Germany). The analytical sensitivities for FT4 and TSH were 0.30 pmol/L and 0.005 μIU/mL, respectively. Assay accuracy was monitored using lyophilized quality control materials (Lyphochek Immunoassay Plus Control, Bio-Rad Laboratories). The intra- and inter-assay CVs were 1.3% and 3.3% for FT4, and 1.4% and 4.6% for TSH, respectively. Details of additional laboratory measurements, including fasting plasma glucose (FPG), high-density lipoprotein cholesterol (HDL-C), and total cholesterol (TC), which were performed on the day of sampling, have been reported elsewhere [21].

2.3. Definition of variables

TPO antibody (TPOAb) positivity was defined as serum TPOAb concentrations exceeding the upper limits of the sex-specific reference ranges previously established for the Iranian population, namely >32.80 IU/mL for men and >35.04 IU/mL for women [24]. Furthermore, TPOAb levels were also classified and investigated as tertiles and clinical categories (cut-off: 35 and 300 IU/mL); the cut-offs for the clinical categories were arbitrary and intended for exploratory dose-response analyses.

Thyroid functional status was classified based on serum FT4 and TSH concentrations. Euthyroidism was defined as an FT4 level between 0.91 and 1.55 ng/dL with a TSH concentration within the reference range of 0.32–5.06 mIU/L. Subclinical hypothyroidism was defined by elevated TSH levels (>5.06 mIU/L) in the presence of normal FT4 concentrations (0.91–1.55 ng/dL), whereas overt hypothyroidism was characterized by TSH levels >5.06 mIU/L accompanied by reduced FT4 levels (<0.91 ng/dL). Subclinical hyperthyroidism was defined as suppressed TSH levels (<0.32 mIU/L) with normal FT4 concentrations (0.91–1.55 ng/dL), while overt hyperthyroidism was diagnosed when TSH levels were <0.32 mIU/L in conjunction with elevated FT4 levels (>1.55 ng/dL) [25,26].

Education level was classified as primary (<6 years), secondary (6–12 years), or higher education (>12 years). Physical activity was assessed using metabolic equivalent of task (MET) minutes, with <600 MET-minutes per week classified as low physical activity, consistent with established recommendations and previous TLGS studies [27]. Baseline dyslipidemia was defined as the use of lipid-lowering medication or the presence of at least one of the following: TC ≥ 240 mg/dL, TG ≥ 200 mg/dL, HDL-C <40 mg/dL in men or <50 mg/dL in women, or LDL-C ≥160 mg/dL [28,29]. Diabetes mellitus was defined as fasting plasma glucose ≥126 mg/dL, and/or 2-h postprandial plasma glucose ≥200 mg/dL, or current use of antidiabetic medications. Current smoking status was defined as cigarette smoking on a daily or occasional basis. Hypertension was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg, or the use of antihypertensive medications.

In this study, clinical cardiovascular disease (CVD) was considered as a composite outcome including coronary heart disease (CHD), definite or probable myocardial infarction based on electrocardiographic findings together with cardiac biomarkers and clinical signs, angiographically documented subclinical coronary atherosclerosis, history of cardiac surgery, or unstable angina pectoris; definite or probable stroke, defined as an acute neurological deficit lasting ≥24 h; and non-CHD conditions such as heart failure and atrial fibrillation. Cardiovascular mortality was defined as death attributable to CVD-related events. Stroke events were defined as subarachnoid hemorrhage, intracerebral hemorrhage, other intracranial hemorrhage, occlusion or stenosis of precerebral or cerebral arteries, or death due to stroke, as adjudicated by the endpoint committee. Mortality ascertainment was confirmed through review of death certificates, forensic medical reports, and, when available, verbal autopsy records.

In this 18-year longitudinal cohort study, we evaluated the association between TPOAb status (positivity clinical categories, tertiles, and continuous serum levels) and time-to-event incidence of cardiovascular events.

2.4. Statistical analysis

Continuous variables are presented as mean ± standard deviation (SD), and categorical variables as counts (percentages). Between-group comparisons were performed using independent-samples t-tests for approximately normally distributed continuous variables, Mann–Whitney U tests for non-normally distributed continuous variables, and χ² tests for categorical variables.

Time-to-event analyses for incident cardiovascular disease (CVD) were conducted using Cox proportional hazards regression models, with results reported as hazard ratios (HRs) and 95% confidence intervals (CIs). Only the first incident CVD event was considered in the time-to-event analyses. TPOAb was analyzed as a log-transformed continuous variable (per 1-SD increase), dichotomous variable (positivity), and across exploratory clinical categories and tertile categories to assess dose–response relationships. Standardization of log-transformed TPOAb used the cohort-wide SD, which was applied unchanged in all stratified analyses. Given the limited number of events among participants aged <40 years, Firth-penalized Cox regression was performed to assess the robustness of the estimates to small-sample bias. Additionally, a sensitivity analysis restricted to hard cardiovascular outcomes (a composite outcome of MI, stroke, and cardiovascular mortality) was performed to assess the robustness of the findings. To minimize potential reverse causation, a 1.5-year landmark analysis was performed. Only participants who were alive, under observation, and free of CVD at the landmark entered the risk set, and analysis time was measured from the landmark onward.

Four progressively adjusted models were constructed: Model 1 included age and sex; Model 2 additionally included familial CVD history, smoking status, physical activity, BMI, and education; Model 3 further adjusted for baseline diabetes, baseline hypertension, and baseline dyslipidemia; Model 4 additionally included TSH. Model 3 was considered the primary confounder-adjusted model, whereas Model 4 was treated as a secondary TSH-adjusted model because TSH may mediate part of the association.

Effect modification by age was assessed primarily by retaining age as a continuous variable and modeling its interaction with standardized log-transformed TPOAb using restricted cubic splines. The spline interaction terms were evaluated jointly using a global likelihood-ratio test. Age-specific hazard ratios for CVD per 1-SD increase in log-transformed TPOAb were estimated with 95% confidence intervals and plotted across the age range together with the distribution of CVD events. A linear interaction between standardized log-transformed TPOAb and continuous age was also examined. Based on exploratory examination of the spline findings, an additional categorical interaction using age groups of <40 and ≥ 40 years was evaluated, followed by exploratory age-stratified analyses.

Subgroup analyses were conducted using Model 3 as the primary confounder-adjusted model, excluding the stratification variable from the adjustment set when applicable. Model 4, which additionally included TSH, was considered secondary. Non-linear associations between log-transformed TPOAb and CVD were assessed using restricted cubic spline models, including separate exploratory analyses within the two age groups. Kaplan–Meier-derived cumulative hazard curves and log-rank tests were used to compare TPOAb groups. The proportional hazards assumption was evaluated using Schoenfeld residuals. Analyses were conducted using complete cases; participants with missing required covariate data (n = 41) were excluded, and no imputation was performed.

All statistical analyses were performed using R (version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria). A two-sided P value < 0.05 was considered statistically significant.

3. Results

3.1. Baseline characteristics

Baseline characteristics of the overall cohort and comparisons between TPOAb groups are summarized in Table 1. A total of 4825 participants (56.5% female), with a mean age of 39.7 ± 14.0 years, constituted the eligible baseline cohort, of whom 653 (13.5%) were TPOAb-positive, and 4172 (86.5%) were TPOAb-negative. 88.6% of the eligible baseline cohort were euthyroid. The participants were followed up for 14.8 ± 3.3 years after the landmark. During follow-up, 576 incident CVD events were recorded, including 68 among participants aged <40 years and 508 among those aged ≥40 years. The cohort comprised 2749 (56.9%) individuals aged <40 years and 2076 (43.0%) aged ≥40 years.

Table 1.

Baseline characteristics of the eligible baseline cohort according to TPOAb status.

Characteristics	Overall	Negative TPOAb	Positive TPOAb	P value
Number of participants	4825	4172	653	-
Age (years)	39.72 + 14.05	39.43 + 14.04	41.57 + 13.93	<0.001
Male	2097 (43.46)	1906 (45.69)	191 (29.25)	<0.001
Body mass index (kg/m2)	26.59 + 4.56	26.49 + 4.54	27.26 + 4.65	<0.001
Waist circumference (cm)	86.37 + 12.30	86.22 + 12.28	87.39 + 12.40	0.052
Education				0.003
Illiterate/primary school (<6 yrs)	1722 (35.69)	1450 (34.76)	272 (41.65)
High school (6-12 years)	2437 (50.51)	2141 (51.32)	296 (45.33)
Higher education (>12 years)	666 (13.80)	581 (13.93)	85 (13.02)
Smokers	564 (11.68)	493 (11.81)	71 (10.87)	0.21
Low physical activity	1516 (31.41)	1315 (31.51)	201 (30.78)	0.27
SBP (mmHg)	117.57 + 17.83	117.47 + 17.65	118.18 + 18.99	0.41
DBP (mmHg)	77.02 + 10.33	77.05 + 10.36	76.84 + 10.16	0.67
FPG (mg/dL)	95.26 + 28.22	95.19 + 28.39	95.69 + 27.18	0.67
Triglyceride (mg/dL)	160.21 + 107.31	159.66 + 103.80	163.69 + 127.46	0.37
TC	204.38 + 45.02	203.79 + 44.65	208.29 + 47.25	0.04
HDL-C (mg/dL)	41.42 + 10.77	41.20 + 10.68	42.80 + 11.22	<0.001
Anti-hypertensive drug use	251 (5.20)	213 (5.10)	38 (5.81)	0.58
Lipid-lowering drug use	100 (2.08)	87 (2.09)	13 (1.99)	0.87
Antidiabetic medication use	138 (2.87)	119 (2.86)	19 (2.91)	0.72
TSH	2.88 + 10.83	2.11 + 4.34	7.80 + 26.83	<0.001
FT4	1.19 + 0.19	1.21 + 0.19	1.10 + 0.23	<0.001
Euthyroidism	4276 (88.62)	3857 (92.44)	419 (64.16)	<0.001

SBP systolic blood pressure, DBP diastolic blood pressure, FPG fasting plasma glucose, HDL-C high-density lipoprotein cholesterol, TC total cholesterol.

*The categorical and continuous variables were reported as counts (percentages) and means ± SD, respectively.

Participants with positive TPOAb were notably older than those with negative TPOAb (41.57 ± 13.93 vs. 39.43 ± 14.04 years, P < 0.001), with a considerably higher proportion of female gender (70.75% vs. 54.31%, P < 0.001). They also had a higher mean body mass index (27.26 ± 4.65 vs. 26.49 ± 4.54 kg/m², P < 0.001).

Educational level differed significantly between groups (P = 0.003), whereas smoking status and physical activity were comparable. No significant differences were observed in systolic or diastolic blood pressure, FPG, triglycerides, or rate of antihypertensive, lipid-lowering, and antidiabetic medication use.

Regarding lipid profile, total cholesterol was modestly higher among TPOAb-positive individuals (208.29 ± 47.25 vs. 203.79 ± 44.65 mg/dL, P = 0.04), as was HDL-C (42.80 ± 11.22 vs. 41.20 ± 10.68 mg/dL, P < 0.001). Thyroid function markers differed notably between groups, with TPOAb-positive participants having substantially higher TSH levels (7.80 ± 26.83 vs. 2.11 ± 4.34, P < 0.001), lower FT4 (1.10 ± 0.23 vs. 1.21 ± 0.19, P < 0.001), and a lower prevalence of euthyroidism (64.1% vs. 92.4%, P < 0.001).

3.2. Primary association between TPOAb and incident CVD

In the overall cohort, TPOAb levels were not associated with incident CVD risk. Across Models 1-4, HRs per 1-SD increase in log-transformed TPOAb ranged from 1.02 to 1.05, none of which reached statistical significance. In the primary confounder-adjusted model (Model 3), the HR was 1.04 (95% CI 0.95–1.13; P = 0.38). Additional adjustment for TSH in Model 4 did not materially alter the result (HR = 1.03, 95% CI 0.94–1. 12; P = 0.47). TPOAb positivity was likewise not associated with CVD in Model 3 (HR = 1.00, 95% CI 0.79–1.27; P = 0.99), with a similar estimate in Model 4 (HR = 0.97, 95% CI 0.76–1.25; P = 0.84) (Table 2). Evaluation using Schoenfeld residuals showed no evidence of violation of the proportional hazards assumption for the TPOAb exposures or the fitted models.

Table 2.

Association of thyroid peroxidase antibody status with CVD event risk.

	Total		< 40 yearsa		≥40 yearsa
	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
Log TPOAb, Per 1-SD
No.event	576		68		508
Model 1	1.05 (0.97-1.14)	0.19	1.37 (1.11-1.68)	0.003	1.02 (0.94-1.11)	0.61
Model 2	1.02 (0.94-1.11)	0.53	1.37 (1.11-1.69)	0.003	0.99 (0.91-1.08)	0.94
Model 3	1.04 (0.95-1.13)	0.38	1.35 (1.10-1.66)	0.004	1.00 (0.92-1.10)	0.83
Model 4	1.03 (0.94-1.12)	0.47	1.32 (1.07-1.63)	0.009	1.02 (0.93-1.12)	0.66
TPOAb positive vs. negative.
No.event (P/N)	85/491		14/54		71/437
Model 1	1.09 (0.86-1.37)	0.48	2.25 (1.24-4.09)	0.008	0.97 (0.75-1.25)	0.80
Model 2	0.98 (0.77-1.25)	0.87	2.34 (1.28-4.28)	0.006	0.88 (0.68-1.15)	0.35
Model 3	1.00 (0.79-1.27)	0.99	2.25 (1.23-4.13)	0.009	0.91 (0.70-1.18)	0.45
Model 4	0.97 (0.76-1.25)	0.84	2.09 (1.11-3.93)	0.022	0.92 (0.70-1.20)	0.53

Model 1 adjusted for: age, sex.

Model 2 adjusted for: Age + sex + smoking + physical activity + Education + BMI + family history of CVD.

Model 3 adjusted for: Model 2 + baseline diabetes + baseline hypertension + baseline dyslipidemia.

Model 4 adjusted for: Model 3 + TSH.

^aPrimary continuous-age spline interaction: global P = 0.024; exploratory categorical interaction (<40 vs ≥ 40 years): P = 0.016.

In the primary effect-modification analysis, age was retained continuously and modeled using restricted cubic splines. The global interaction between continuous age and standardized log-transformed TPOAb was significant (likelihood-ratio χ² = 9.43, df = 3; P = 0.024), indicating that the association varied across age (Fig. 2). A complementary linear interaction analysis also demonstrated significant effect modification by continuous age (interaction HR = 0.79, 95% CI 0.63–0.98; P = 0.034; Table S1). For secondary exploratory analyses, age was categorized as <40 versus ≥40 years using a pragmatic rounded boundary approximately corresponding to the age range over which the spline-estimated association began to attenuate. The categorical interaction was significant (interaction HR = 0.45, 95% CI 0.24–0.86; P = 0.016), indicating a weaker association among participants aged ≥40 years (Table S2).

Fig. 2.

Age-specific hazard ratios for incident CVD per 1-SD increase in log-transformed TPOAb, estimated using a continuous-age restricted cubic spline interaction model. Shading indicates 95% CIs, the horizontal dashed line indicates HR = 1, the vertical dotted line marks the exploratory 40-year cutoff, and bars show the distribution of CVD events.

3.3. Age-stratified analyses

Among participants aged <40 years, higher TPOAb levels were consistently associated with increased CVD risk across all models. In the primary Model 3, the HR per 1-SD increase in log-transformed TPOAb was 1.35 (95% CI 1.10–1.66; P = 0.004), while the HR for TPOAb positivity was 2.25 (95% CI 1.23–4.13; P = 0.009). Additional adjustment for TSH produced comparable estimates of 1.32 (95% CI 1.07–1.63; P = 0.009) and 2.09 (95% CI 1.11–3.93; P = 0.022), respectively. Among participants aged ≥40 years, neither continuous TPOAb (Model 3: HR = 1.00, 95% CI 0.92–1.10; P = 0.83) nor TPOAb positivity (HR = 0.91, 95% CI 0.70–1.18; P = 0.45) was associated with CVD risk (Table 2).

Furthermore, Kaplan-Meier curves (Fig. 3) showed a significantly higher cumulative CVD hazard among TPOAb-positive than TPOAb-negative participants aged <40 years (log-rank P = 0.024). (Fig. 3A). In contrast, no significant difference was observed between the two TPOAb subgroups among participants ≥40 years (log-rank P = 0.28) (Fig. 3B). In the sensitivity analysis restricted to hard cardiovascular outcomes, defined as myocardial infarction, stroke, or cardiovascular mortality (439 events), the findings remained consistent with the primary analysis. Per 1-SD increase in log-transformed TPOAb, the adjusted HR was 1.04 (95% CI 0.95–1.14) in the overall cohort, 1.38 (95% CI 1.09–1.74) among participants aged <40 years, and 1.01 (95% CI 0.91–1.12) among those aged ≥40 years. Similarly, TPOAb positivity was associated with hard CVD among participants aged <40 years (HR = 2.45, 95% CI 1.23–4.87), but not in the overall cohort or among participants aged ≥40 years (Table S3).

Fig. 3.

Kaplan–Meier-derived cumulative hazard curves for incident CVD according to TPOAb status, stratified by age group. (Numbers at risk represent participants contributing follow-up after the 1.5-year landmark.)

Given the relatively limited number of events among participants aged <40 years, Firth penalized Cox regression was used to reduce potential small-sample bias. The findings remained consistent with the primary analysis, demonstrating significant associations between TPOAb and incident CVD. In Model 3, the HR was 1.36 (95% CI 1.10–1.68) per 1-SD increase in log-transformed TPOAb, while the HR for TPOAb positivity was 2.25 (95% CI 1.20–4.25). Additional adjustment for TSH in Model 4 produced comparable estimates of 1.34 (95% CI 1.07–1.66) and 2.17 (95% CI 1.13–4.19), respectively, with both associations remaining statistically significant (Table 3).

Table 3.

Association between TPOAb and incident CVD according to four hierarchical adjustment models among participants aged <40 years: Firth-penalized Cox regression.

TPOAb exposure	Event	Model 1 HR (95% CI)	Model 2 HR (95% CI)	Model 3 HR (95% CI)	Model 4 HR (95% CI)	Ridge HRa
Continuous TPOAb
per 1-SD increase	68	1.40 (1.14–1.73)	1.41 (1.14–1.75)	1.36 (1.10–1.68)	1.34 (1.07–1.66)	1.26
TPOAb
negative	54	1.0 (Reference)	1.0 (Reference)	1.0 (Reference)	1.0 (Reference)
positive	14	2.36 (1.28–4.36)	2.49 (1.33–4.65)	2.25 (1.20–4.25)	2.17 (1.13–4.19)	1.80
TPOAb Tertile
T1	20	1.0 (Reference)	1.0 (Reference)	1.0 (Reference)	1.0 (Reference)
T2	19	1.30 (0.67–2.49)	1.22 (0.62–2.38)	1.23 (0.63–2.39)	1.23 (0.63–2.41)	1.08
T3	29	2.13 (1.16–3.94)	2.17 (1.15–4.08)	2.17 (1.15–4.10)	2.13 (1.12–4.04)	1.64
TPOAb categories
<35	54	1.0 (Reference)	1.0 (Reference)	1.0 (Reference)	1.0 (Reference)
36 – 300	8	2.14 (1.02–4.49)	2.24 (1.05–4.75)	2.06 (0.97–4.39)	1.99 (0.92–4.34)	1.58
>300 vs < 35	6	3.14 (1.27–7.72)	3.40 (1.36–8.50)	2.97 (1.18–7.50)	2.82 (1.10–7.21)	2.30

Abbreviations: CVD, cardiovascular disease; HR, hazard ratio; CI, confidence interval; TPOAb, thyroid peroxidase antibody.

Note: All analyses were restricted to participants aged <40 years. Models were fitted using Firth-penalized Cox proportional hazards regression. The four models represent sequential hierarchical adjustment sets.

Model 1 adjusted for: age, sex.

Model 2 adjusted for: Age + sex + smoking + physical activity + Education + BMI + family history of CVD.

Model 3 adjusted for: Model 2 + baseline diabetes + baseline hypertension + baseline dyslipidemia.

Model 4 adjusted for: Model 3 + TSH.

^aRidge-penalized Cox regression based on Model 3 was additionally performed as a coefficient-shrinkage sensitivity analysis.

3.4. Dose-response relationship between TPOAb and incident CVD

Dose-response analyses suggested increasing CVD risk at higher TPOAb levels among participants aged <40 years. In the Firth penalized Model 3, compared with TPOAb levels <35 IU/mL, the HR was 2.06 (95% CI 0.97–4.39) for levels of 36–300 IU/mL and 2.97 (95% CI 1.18–7.50) for levels >300 IU/mL. Additional adjustment for TSH produced comparable estimates of 1.99 (95% CI 0.92–4.34) and 2.82 (95% CI 1.10–7.21), respectively. Similarly, the highest TPOAb tertile was associated with increased CVD risk compared with the lowest tertile in both Model 3 (HR = 2.17, 95% CI 1.15–4.10) and Model 4 (HR = 2.13, 95% CI 1.12–4.04) (Table 3). Conventional Cox analyses yielded comparable findings, with increased risk among younger participants with TPOAb >300 IU/mL (HR = 3.11, 95% CI 1.31–7.41) and those in the highest tertile (HR = 2.12, 95% CI 1.17–3.86).

In contrast, no evidence of a dose-response association was observed in the overall cohort or among participants aged ≥40 years (Table S4).

Restricted cubic spline analyses suggested different exposure–response patterns across the two age groups (Fig. 4). Among participants aged <40 years, the adjusted HR did not increase consistently across lower-to-moderate TPOAb levels but rose substantially at higher log-transformed TPOAb concentrations, with increasing uncertainty at the upper exposure range (Fig. 4). Among participants aged ≥40 years, estimates varied non-linearly above and below the null without a consistent directional dose–response pattern (Fig. 4).

Fig. 4.

Restricted cubic spline associations between log-transformed TPOAb and incident CVD in participants aged <40 and ≥ 40 years. Lines represent Model 3-adjusted HRs, shaded areas indicate 95% CIs, and dotted lines indicate HR = 1.

3.5. Subgroup analyses

Among participants aged <40 years, hazard ratios per 1-SD increase in log-TPOAb were 1.28 (95% CI 0.95–1.72) in males and 1.44 (95% CI 1.06–1.96) in females. Corresponding estimates were 1.53 (95% CI 1.18–1.97) among euthyroid individuals and 1.18 (95% CI 0.79–1.77) among those with thyroid dysfunction. Across TSH strata, HRs were 1.31 (95% CI 0.85–2.03), 1.37 (95% CI 0.96–1.95), and 1.21 (95% CI 0.70–2.10) for TSH levels 0-2, 2-5, and >5 mIU/L, respectively. Significant associations were also observed among nonsmokers, participants without hypertension, and those who were overweight. Among participants aged ≥40 years, no significant association was observed within any subgroup (Table 4).

Table 4.

Subgroup analyses of the association between continuous log-transformed TPOAb levels and incident CVD (per 1-SD increase).†

	Total	< 40 years	≥ 40 years
Sex
Males	1.01 (0.89-1.15)	1.28 (0.95-1.72)	1.00 (0.87-1.14)
Females	1.06 (0.95-1.19)	1.44 (1.06-1.96)	1.03 (0.92-1.16)
Thyroid statuses
euthyroidism	1.09 (0.98-1.22)	1.53 (1.18-1.97)	1.06 (0.95-1.20)
Thyroid dysfunction	0.96 (0.84-1.11)	1.18 (0.79-1.77)	0.92 (0.79-1.06)
TSH levels (mIU/L)
0 - 2	1.03 (0.90-1.18)	1.31 (0.85-2.03)	1.03 (0.90-1.18)
2 - 5	1.08 (0.93-1.27)	1.37 (0.96-1.95)	1.03 (0.86-1.23)
>5	1.31 (1.00-1.71)	1.21 (0.70-2.10)	1.33 (0.97-1.82)
Smoking
No	1.08 (0.98-1.20)	1.47 (1.14-1.88)	1.04 (0.94-1.16)
Yes	0.90 (0.71-1.13)	1.22 (0.79-1.89)	0.85 (0.65-1.12)
Hypertension
No	1.13 (1.01-1.26)	1.36 (1.07-1.74)	1.11 (0.98-1.25)
Yes	1.05 (0.91-1.20)	1.30 (0.83-2.05)	1.04 (0.90-1.20)
BMI statuses
Normal-weight	0.91 (0.75-1.12)	1.22 (0.79-1.87)	0.90 (0.72-1.11)
Over-weight	1.14 (1.01-1.29)	1.53 (1.12-2.09)	1.09 (0.96-1.24)
Obesity	0.99 (0.85-1.14)	1.14 (0.74-1.73)	0.99 (0.85-1.15)

Adjusted Age + sex + smoking + physical activity + Education + BMI + family history of CVD + baseline diabetes + baseline hypertension + baseline dyslipidemia.

† Subgroup analyses were conducted using Model 3-adjusted models. When the stratification variable was included among adjustment covariates, it was excluded from that specific model to avoid overadjustment.

4. Discussion

In this large prospective population-based cohort, we found that thyroid peroxidase antibody (TPOAb) levels were not associated with incident cardiovascular disease (CVD) risk in the overall population. However, the primary continuous-age analysis demonstrated significant effect modification by age. In exploratory age-stratified analyses, higher TPOAb levels were associated with increased risk of incident CVD events among individuals aged <40 years, but not among those aged ≥40 years. This pattern was generally consistent across complementary TPOAb exposure definitions. Sensitivity analyses using Firth-penalized regression and a conventional hard cardiovascular endpoint yielded generally consistent findings. Furthermore, our spline analyses did not demonstrate a strictly monotonic association among participants aged <40 years; instead, risk estimates increased primarily at higher TPOAb concentrations, with greater uncertainty at the extremes of the exposure distribution. Among participants aged ≥40 years, estimates varied around the null without a consistent directional pattern. These findings extend the existing literature by suggesting that the association between thyroid autoimmunity and incident CVD may vary with age, although the exploratory age-stratified findings require external validation.

The age-dependent association between TPOAb and incident cardiovascular disease (CVD) in our cohort persisted after comprehensive adjustment for demographic, lifestyle, and baseline cardiometabolic factors, including BMI, smoking, physical activity, education, familial CVD history, diabetes, hypertension, and dyslipidemia. Additional adjustment for TSH did not materially alter the findings. This suggests that thyroid autoimmunity may confer cardiovascular risk through non-traditional, immune-mediated pathways, particularly in younger adults. Subgroup analyses were broadly consistent with this pattern, as the association was evident among euthyroid participants and those without hypertension or smoking exposure, whereas it was less apparent in higher-risk subgroups, consistent with a mechanism that operates independently of, or is less detectable in the presence of, classical cardiometabolic and overt thyroid-related factors. However, these exploratory findings should be interpreted cautiously.

Taken together, these findings raise the possibility that thyroid autoimmunity may represent a form of non-classical cardiovascular risk, operating through pathways not fully captured by conventional cardiometabolic factors. The age-dependent pattern observed in our study suggests that these mechanisms may be more biologically relevant in younger individuals, in whom vascular changes may be at earlier and potentially more modifiable stages. In this context, a set of interrelated immune and vascular processes, encompassing immune-mediated vascular injury (e.g., endothelial activation and low-grade inflammation), functional cardiovascular alterations (e.g., autonomic imbalance and microvascular dysfunction), and early-life susceptibility mechanisms (e.g., genetic and immune factors), may plausibly link TPOAb positivity to incident cardiovascular disease, rather than a single dominant pathway.

TPOAb positivity may reflect chronic low-grade autoimmune activity and has been associated with a pro-inflammatory state and markers of subclinical atherosclerosis, including increased carotid intima–media thickness and incident coronary artery calcification, independent of thyroid function and conventional cardiovascular risk factors [[30], [31], [32]]. However, associations with subclinical atherosclerosis do not establish prediction of clinical cardiovascular events. In addition, emerging evidence suggests that thyroid autoimmunity may be linked to endothelial dysfunction, as reflected by elevated adhesion molecules and impaired vasodilatory signaling, providing a plausible biological pathway for early vascular injury [33]. In a broader context, accelerated and subclinical atherosclerosis has also been described in autoimmune rheumatic diseases and antiphospholipid antibody–positive populations [34,35]. In younger adults, intact vasculature and low baseline risk may render these immune-mediated effects proportionally more impactful, potentially making these effects more detectable in relative terms. In contrast, in older adults, pre-existing vascular remodeling, fibrosis, and cumulative subclinical damage may attenuate the relative contribution of such immune mechanisms, consistent with prior evidence showing that the impact of individual risk factors diminishes with advancing age [36,37].

In addition, thyroid autoimmunity may perturb autonomic balance, characterized by reduced parasympathetic tone and relative sympathetic predominance, as reflected by alterations in heart rate variability parameters and their correlation with antibody titers [38,39]. Such autonomic dysregulation has been associated with adverse cardiovascular profiles and may contribute to arrhythmic susceptibility and cardiovascular risk, as reduced heart rate variability has been shown to independently predict incident cardiac events beyond traditional risk factors [40]. In younger individuals, cardiac function may be more dynamically responsive to autonomic inputs, potentially amplifying the impact of these subtle shifts, whereas in older adults, age-related reductions in autonomic responsiveness and baroreflex sensitivity may attenuate this pathway [41,42].

Beyond these functional pathways, TPOAb may serve as a marker of underlying genetic or immune susceptibility, clustering with other autoimmune conditions such as type 1 diabetes and reflecting shared immune regulatory pathways that contribute to early atherogenesis [43]. Evidence from longitudinal cohorts indicates that thyroid autoimmunity can emerge early in life, particularly in individuals carrying high-risk human leukocyte antigen (HLA) genotypes, supporting the role of inherited immune predisposition and early-life environmental triggers in its development [43]. Consistent with this early-life perspective, studies in pediatric populations have characterized Hashimoto thyroiditis in both euthyroid and hypothyroid phases and documented its coexistence with childhood-onset systemic lupus erythematosus [44,45]. In this context, TPOAb positivity in younger individuals may reflect an early-expressing pro-inflammatory immune milieu associated with greater susceptibility to incident cardiovascular events [37]. In older adults, selective survival and the increasing dominance of classical cardiometabolic risk factors may attenuate the observable contribution of this pathway, consistent with the age-dependent pattern observed in our study [36].

Furthermore, even after adjustment for classical risk factors, thyroid autoimmunity may reflect residual immunometabolic dysregulation not captured by standard models. TPOAb positivity has been linked to insulin resistance and low-grade inflammation, including higher homeostatic model assessment of insulin resistance (HOMA-IR) and high-sensitivity C-reactive protein (hsCRP) levels, as well as a greater burden of metabolic syndrome traits in euthyroid individuals [33,46]. Population-level data further show that TPOAb tracks with progressive increases in adiposity, glycemic indices, and lipid parameters even within the normal thyroid range [47]. In younger adults, these modest but systemic disturbances may meaningfully influence the timing of first cardiovascular events in a low-risk setting, whereas in older individuals, their impact may be attenuated by accumulated cardiometabolic burden and treatment effects.

Finally, from a life-course perspective, the absolute hazard of cardiovascular disease rises steeply with age; consequently, the relative contribution of early immune-mediated risk may diminish over the life course as classical cardiometabolic factors increasingly dominate risk variation [36]. In this context, TPOAb-related effects may be most apparent during an early-adulthood window of susceptibility, when vascular integrity is relatively preserved and competing cardiometabolic risk factors are less prominent. As individuals age, cumulative exposure to established risk factors and their treatment may attenuate the observable impact of autoimmune-related processes. In addition, a depletion of susceptible individuals, whereby those most vulnerable to immune-mediated risk may experience events earlier in life, may further contribute to the attenuation of associations in older age groups. These observations are compatible with a possible life-course-dependent association of thyroid autoimmunity, rather than a uniform or persistent influence across age strata.

Together, these findings suggest that TPOAb may serve as a marker of non-traditional, immune-driven cardiovascular risk, with an association that may be more apparent in younger adults. Further studies are warranted to clarify the underlying mechanisms and to determine whether TPOAb has potential value in early cardiovascular risk assessment.

These findings have several potential clinical and research implications. Younger individuals are often classified as having low short-term cardiovascular risk by conventional prediction models; however, current prevention frameworks emphasize that risk assessment in this population is inherently challenging, as clinical events may occur over a prolonged time horizon and short-term estimates may not fully capture lifetime risk trajectories [48]. Consistent with this, young adults with low short-term but elevated long-term risk have been shown to experience significantly higher rates of incident atherosclerotic cardiovascular disease events, underscoring the challenges of accurately characterizing risk in younger populations [49]. In this context, our findings raise the possibility that thyroid autoimmunity may be associated with higher relative cardiovascular risk in some younger adults, although these exploratory analyses do not establish incremental predictive value beyond conventional risk factors. TPOAb should not be interpreted as a causal factor or an established cardiovascular risk marker, but rather, our findings highlight the potential role of immune-mediated pathways in early disease development. Current cardiovascular prevention guidelines do not incorporate TPOAb into routine cardiovascular risk assessment; therefore, our findings should be considered hypothesis-generating [50]. Nevertheless, they support further investigation into whether immune-related biomarkers such as TPOAb could contribute to refined risk assessment or early prevention strategies in selected populations.

Previous studies evaluating the association between TPOAb and cardiovascular outcomes have yielded inconsistent findings. The ELSA study [16] reported no significant association between TPOAb status and cardiovascular mortality, whereas the Rotterdam Study [17] observed a significant association when TPOAb was analyzed using a lower detectability threshold, but not when using standard positivity cut-offs. This discrepancy suggests that exposure definition may influence observed associations. Although these overall null findings are broadly consistent with our results in the total population, neither study formally evaluated age as an effect modifier.

In line with a recent individual participant data meta-analysis [18], which found no significant overall association of TPOAb positivity with CHD or stroke, we observed no association in the overall study population. Because TTS contributed participants to both studies, the source populations and follow-up periods substantially overlapped. However, the analytic populations were not identical: Hysaj et al. included 5143 TTS participants with a median follow-up of 17.9 years and retained participants with baseline CVD and thyroid-medication use in their primary analysis, whereas the present study included 4825 participants with a mean follow-up of 14.8 ± 3.3 years after the 1.5-year landmark and applied more restrictive eligibility criteria to define a baseline CVD-free primary-prevention population. Endpoint overlap was substantial for CHD, although TTS did not contribute stroke-event data to the Hysaj analysis, whereas the present primary outcome additionally included stroke and other adjudicated cardiovascular events. Furthermore, Hysaj et al. modeled TPOAb only as a binary exposure and assessed age modification across the pooled 14-cohort population using <65/≥65-year strata, without reporting a dedicated TTS-specific age interaction. This broad age categorization and between-cohort heterogeneity may have obscured associations concentrated in younger adults. Thus, the present study extends rather than independently replicates the previous TTS contribution by characterizing the TPOAb-by-age association across the continuous adult age spectrum using complementary TPOAb exposure definitions.

Our findings extend the existing literature by providing evidence of age-dependent heterogeneity, with exploratory analyses suggesting a positive association among individuals aged <40 years but not among those aged ≥40 years. These observations suggest that differences in exposure modeling, outcome definition, and age assessment may contribute to inconsistencies across previous studies.

This study has several strengths. Its prospective population-based design with long-term follow-up and assessment of incident cardiovascular events, rather than mortality alone, supports the temporal ordering of baseline TPOAb measurement and incident CVD. The use of both continuous and categorical TPOAb definitions, along with spline modeling and penalized sensitivity analyses, allowed the association to be evaluated using complementary analytical approaches. In addition, comprehensive adjustment for a wide range of cardiometabolic and thyroid-related factors reduces the potential for confounding.

Several limitations should be acknowledged. As an observational study, causal inferences cannot be established, and residual confounding by unmeasured factors remains possible. TPOAb levels, thyroid function, and thyroid-related medication use were assessed at baseline, and longitudinal changes in these measures were not incorporated into the analyses. However, evidence from the TTS, which followed participants over 18 years with repeated TPOAb measurements, indicates that the vast majority (94.7%) maintained stable trajectories (either persistently low or persistently high), with a low incidence of new positivity (5.6 per 1000 person-years). These findings support baseline TPOAb status as a reasonably stable indicator of long-term thyroid autoimmunity in community-dwelling adults [51]. Nevertheless, TPOAb represents a serological marker of thyroid autoimmunity, and its assessment at baseline may not fully capture changes in autoimmune status during follow-up. Furthermore, data on inflammatory biomarkers and detailed immune profiles were not available, limiting direct mechanistic inference. The primary composite outcome included clinically heterogeneous cardiovascular events; however, the findings were generally similar in a sensitivity analysis using a conventional hard cardiovascular endpoint comprising cardiovascular death, myocardial infarction, and stroke. The number of events among participants aged <40 years remained limited, particularly within individual TPOAb categories, reducing statistical precision. Although the principal Firth-penalized estimates for continuous TPOAb and TPOAb positivity remained statistically significant and the ridge estimates were directionally consistent with the conventional Cox estimates, the age-stratified, categorical, and subgroup analyses were exploratory and remain susceptible to sparse-data and multiple-testing concerns. Furthermore, the specific 40-year cutoff was not prespecified and should not be interpreted as an established biological or clinical threshold. Finally, the generalizability of these findings to other populations with different demographic or clinical characteristics may be limited.

5. Conclusion

In conclusion, thyroid autoimmunity, as reflected by TPOAb levels, was not associated with incident cardiovascular disease in the overall population, but the continuous-age analysis indicated significant effect modification by age. Exploratory age-stratified analyses suggested that higher TPOAb levels were associated with increased CVD risk among individuals aged <40 years but not among those aged ≥40 years. These findings, while hypothesis-generating, identify age as a potentially important modifier of the cardiovascular relevance of thyroid autoimmunity and provide a rationale for independent validation in younger populations.

6. Disclosure summary

The authors declare that they have no competing interests. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CRediT authorship contribution statement

Farhad Shaker: Writing – original draft, Project administration, Methodology, Investigation, Data curation, Conceptualization. Safdar Masoumi: Validation, Software, Methodology, Formal analysis, Data curation, Conceptualization. Ladan Mehran: Writing – review & editing, Methodology, Investigation. Mohamadamin Tarighat-Payma: Writing – original draft, Investigation. Atieh Amouzegar: Writing – review & editing, Supervision, Project administration, Methodology, Conceptualization. Fereidoun Azizi: Writing – review & editing, Validation, Supervision, Project administration. Salman Razvi: Writing – review & editing, Validation, Supervision.

Ethics declaration

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was approved by the Human Research Review Committee of the Endocrine Research Center of Shahid Beheshti University of Medical Sciences, Tehran, Iran. (Approval No. IR.SBMU.ENDOCRINE.REC.1404.189).

Ethics approval and consent to participate

This study conformed to the ethical guidelines of the Helsinki Declaration and was approved by the Human Research Review Committee of the Endocrine Research Center of Shahid Beheshti University of Medical Sciences, Tehran, Iran (IR.SBMU.ENDOCRINE.REC.1404.189). All participants provided informed written consents.

Consent for publication

Not applicable.

Availability of data and materials

Datasets generated and analyzed during the current Studies are not publicly available due to institutional policies, but are available from the corresponding author on reasonable request.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Abbreviations:

AITD

autoimmune thyroid disease

BMI

body mass index

CHD

coronary heart disease

CI

confidence interval

CVD

cardiovascular disease

DBP

diastolic blood pressure

ELISA

enzyme-linked immunosorbent assay

FPG

fasting plasma glucose

FT4

free thyroxine

HDL-C:

high-density lipoprotein cholesterol

HR

hazard ratio

LDL-C:

low-density lipoprotein cholesterol

MET

metabolic equivalent of task

MI

myocardial infarction

SBP

systolic blood pressure

SD

standard deviation

TC

total cholesterol

TG

triglycerides

TLGS

Tehran Lipid and Glucose Study

TPOAb

thyroid peroxidase antibody

TSH

thyroid-stimulating hormone

TTS

Tehran Thyroid Study

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data will be made available on request.