This nationwide population-based cohort study used data from the Korean National Health and Nutrition Examination Survey (KNHANES). The KNHANES includes a health interview, a health examination survey, and a nutrition survey. The Korean Centers for Disease Control and Prevention (KCDC) has conducted these nationally representative cross-sectional surveys annually since 2007, when it transitioned from a triennial schedule to promptly assess the health and nutrition status of Koreans [16]. The surveys sample non-institutionalized civilians aged 1 year and older using a stratified multistage clustered probability sampling design. For the nutrition survey, biochemical screenings are conducted on participants aged 10 years and older. The KCDC obtained approval from the KCDC Institutional Review Board and collected all samples in accordance with the ethical principles of the Helsinki Declaration for medical research involving human subjects.

From the KNHANES database between 2013 and 2015, we identified 2,425 participants aged 10–18 years. Among them, 1,021 adolescents had undergone thyroid function tests. Because our participants of interest were adolescents in the euthyroid TSH range, 963 participants (94.3%) were included as the study cohort. After excluding participants with incomplete data, the final population consisted of 940 adolescents, 498 of whom were male (Fig. 1, Supplementary Table 1).

The primary outcome was MetS and its components. MetS in adolescents was defined according to the International Diabetes Federation (IDF) guidelines [17]. The new IDF definition of metabolic disease varies by age group, and the following criteria were applied for adolescents. For participants aged 10–15 years, the criteria are abdominal obesity (90th percentile of waist circumference) [18] plus at least 2 other clinical features from the following list: triglycerides ≥150 mg/dL, high-density lipoprotein cholesterol (HDL-C) <40 mg/dL, systolic blood pressure (SBP) ≥130 mmHg or diastolic blood pressure ≥85 mmHg, and fasting plasma glucose (FPG) ≥100 mg/dL or diagnosed type 2 diabetes mellitus. For those aged 16 years and older, the IDF adult criteria were used [19]. The cutoff for abdominal obesity was ≥90 cm for males and ≥80 cm for females.

Serum TSH and free thyroxine (fT4) levels were measured using an electrochemiluminescence immunoassay (Roche Diagnostics, Germany). The reference interval of serum TSH was defined as the 2.5th to 97.5th percentile of the serum TSH levels of the reference population. The definition of euthyroid in the Korean reference population is a serum TSH concentration of 0.62–6.86 mIU/L and an fT4 of 0.96–1.60 ng/dL [20]. The reference population was defined as subjects with no history of thyroid disease and no history of taking medication that could influence thyroid function. However, because a health interview about thyroid disease history was not conducted for participants aged 10–18 years, we assumed that these young participants within the euthyroid range had no prior history of thyroid disease.

To investigate the relationship between TSH levels within the normal range and MetS and its components, we classified participants within the euthyroid reference range into 4 quartiles. In our study population, nonhigh TSH levels ranged from 0.62 to 3.34 mIU/L, and high-normal TSH levels ranged from 3.35 to 6.86 mIU/L.

Anthropometry, blood pressure measurement, and blood analysis protocols are well-described in the KNHANES-VI guidelines [21]. Waist circumference was measured at the narrowest point between the lower borders of the rib cage and the iliac crest at the end of normal expiration. Blood pressure was monitored using an oscillometric device 3 times at 30-second intervals after 5 minutes of rest; the average of the second and third measurements was then recorded. Blood samples were obtained in the morning after an overnight fast of at least 8 hours. FPG, total cholesterol, low-density lipoprotein cholesterol, HDL-C, and triglyceride levels were measured with a Hitachi Automatic Analyzer 7600 (Hitachi Ltd., Japan). The reported results for TSH and fT4 met the specifications for accuracy, general chemistry, special immunology, and ligands of the quality control and quality assurance program of the College of American Pathologists [22].

Using weighted data, we compared participant clinical and biochemical characteristics by MetS status to ensure internal validity, identify key differences, and detect potential confounders for subsequent analyses. We conducted 2-sample t-tests for continuous variables to examine whether the characteristics of participants with and without MetS differed significantly. To handle the complexities of survey data with clustering and stratification design, we applied the Rao-Scott chi-square test and assessed differences in categorical variables between the 2 groups by MetS status as well. For that comparison, we weighted the data to ensure the representativeness of the population of interest, correcting for any potential biases in the sample and balancing the distribution of variables in the 2 groups. Because clinical variables such as triglycerides, aspartate aminotransferase (AST), and alanine aminotransferase (ALT) exhibited right-skewed distributions, geometric means were calculated to minimize the influence of outliers. Additionally, log-transformations were applied to these variables to reduce skewness and ensure compatibility with statistical tests assuming normally distributed residuals. Using weighted data, we also conducted analysis of covariance tests to examine the association between MetS and the 4 quartiles of TSH levels, adjusting for relevant covariates. We performed analyses in 3 models: unadjusted, adjusted for age and sex, and further adjusted for log-transformed ALT. We performed t-tests to compare MetS and its components in the high-normal TSH versus nonhigh TSH groups. We further used logistic regression analyses to examine their associations. Covariate adjustment was used with 3 models: model 1 was an unadjusted crude model; model 2 was adjusted for age and sex; and model 3 was further adjusted for log-transformed ALT. We applied a 95% confidence level using a 2-sided test to determine the significance of the P-values. All analyses were performed using the SAS 9.3 (SAS Institute Inc., USA).

This study was approved by the Institutional Review Board of Myongji Hospital (MJH 2018-07-013) in accordance with the Declaration of Helsinki. The requirement for informed consent was waived by the board because this study is retrospective and KNHANES participants provided consent to participate in the surveys.

Using the weighted data, we compared the clinical and biochemical characteristics of study subjects by MetS status, as shown in Table 1. Compared with the non-MetS group, the MetS group showed significantly higher values in body mass index (BMI) (21.2±0.1 vs. 29.6±0.7 kg/m²), waist circumference (71±0.4 cm vs. 93.3±2.3 cm), SBP (108.5±0.4 mmHg vs. 120.7±2.8 mmHg), triglyceride levels (73.1 mg/dL vs. 179 mg/dL), AST (17.8 IU/L vs. 22.2 IU/L), and ALT (12.7 IU/L vs. 26.8 IU/L), and HDL-C levels were significantly lower in the MetS group (51.9±0.4 mg/dL vs. 38±1.6 mg/dL). Although the differences were not statistically significant, adolescents with MetS also tended to have higher TSH levels (2.71±0.05 mg/dL vs. 2.93±0.33 mg/dL), and a greater percentage of them had high-normal TSH (24.1% vs. 37.4%).

We examined the association between MetS and TSH levels in euthyroid adolescents by stratifying the TSH levels into 4 quartiles. As shown in Fig. 2, after adjusting for covariates, a significant association with MetS was observed across TSH quartiles. Although the association between MetS and TSH quartiles was not strictly linear, it tended to increase with higher TSH quartiles, reaching its peak in the second quartile (Q2) (Supplementary Table 2). This trend remained significant after adjusting for age and sex (model 2: P for trend=0.012) and further adjusting for log-transformed ALT (model 3: P for trend=0.006), compared with the crude model (model 1: P for trend=0.010).

To further investigate how higher TSH levels were associated with an increased risk of MetS and its components, we differentiated between the high-normal TSH group and the nonhigh TSH group (Table 2). The prevalence of MetS was numerically higher in the high-normal TSH group (3.5%) than the nonhigh TSH group (1.9%), but the difference was not statistically significant (P=0.1929). Similarly, the prevalence of some MetS components—elevated blood pressure, hyperglycemia, hypertriglyceridemia, and low HDL-C—varied between the 2 TSH groups; however, none of those differences reached statistical significance. Among the MetS components, only abdominal obesity differed significantly between the high-normal TSH and nonhigh TSH groups, with a higher prevalence in the high-normal TSH group (16.2% vs. 9.6%, P=0.0153).

Table 3 presents the results from logistic regression conducted to determine whether and how the high-normal TSH and nonhigh TSH groups were associated with MetS and its components. The risk of abdominal obesity was significantly higher in the high-normal TSH group than the nonhigh TSH group (model 1: odds ratio [OR], 1.828; 95% confidence interval [CI], 1.115–2.996; P=0.0168; model 2: adjusted OR, 1.896; 95% CI, 1.155–3.122; P=0.011; model 3: adjusted OR, 1.975; 95% CI, 1.104–3.531, P=0.0218). Consistent with the findings in Table 2, however, no significant association was observed between the high-normal TSH group and hypertriglyceridemia, low HDL-C, elevated blood pressure, hyperglycemia, or MetS.