Cardiometabolic-related dietary patterns and thyroid function: a population-based cross-sectional study

The principal hormones released by the thyroid gland, thyroxine (T4) and triiodothyronine (T3) play a crucial role in various physiological processes within the body [1]. These hormones are essential for the normal functioning of the cardiovascular, metabolic, nervous, musculoskeletal, and reproductive systems. Therefore, both subclinical and clinical thyroid dysfunction have detrimental effects on numerous aspects of health [2‐6].

Genetic factors, sex, age, stress, lifestyle factors, pollutants, and diet have been demonstrated to affect thyroid function [7, 8]. It is well-known that adequate dietary iodine and selenium intakes are necessary for normal thyroid function. Moreover, there are numerous indications that other food parameters, such as vitamin D, zinc, iron, copper, soy, brassica vegetables, seaweed, and ultra-processed foods, affect thyroid function; however, insufficient research has been conducted on these issues to date [8]. A few studies are also available on the associations of a certain dietary pattern with thyroid-stimulating hormone (TSH) and thyroid hormones [9‐11]. In a cross-sectional study involving Italian participants, a higher score for the Mediterranean diet was associated with reduced levels of free T3 (FT3) and free T4 (FT4) but not TSH [9]. A cross-sectional study using the National Health and Nutrition Examination Survey (NHANES) also found a positive association between the dietary inflammatory index score and total T4 in males [10]. In addition, a cross-sectional study of healthy Croatian adults identified multiple dietary factors with distinct characteristics that were associated with FT3, FT4, and TSH. These factors were identified using principal component analysis (PCA) [11].

PCA identifies patterns of commonly consumed foods by grouping food items or food groups based on their correlation with each other. PCA discovers linear combinations (known as principal components) of food items or food groups that capture the most significant variations in individuals’ dietary choices. Unlike PCA, the reduced rank regression (RRR) does not prioritize explaining the variance between foods. Instead, RRR aims to identify linear functions of predictors (such as food groups) that can explain the maximum amount of variation in a set of intermediate response variables (such as biomarkers) [12‐14]. Therefore, the dietary pattern that RRR has maintained may better reflect the significance of the dietary pattern in relation to a condition or disease.

Except for the study by Brdar et al. [11] that investigated the association of dietary patterns derived from PCA with TSH and thyroid hormones, no other study has examined the association of dietary patterns with thyroid function. Therefore, the aim of the present study was to identify dietary patterns related to the six cardio-metabolic variables (body mass index (BMI), fasting serum concentrations of glucose, triglycerides, HDL-C, systolic blood pressure (SBP), and diastolic blood pressure (DBP) in our study population and then to investigate the association of the dietary patterns with TSH and thyroid hormones in euthyroid participants and the odds of thyroid dysfunction. We also examined whether a cardiometabolic-related dietary pattern suggested by the Framingham Offspring Study (FOS) [15] was associated with cardio-metabolic variables and thyroid function in our study population.

The mean ± SD age of 3520 participants included in this study was 42.9 ± 12.3 years, and 57.2% were women. Of the participants, 133 had clinical hypothyroidism, 341 had subclinical hypothyroidism, 25 had clinical hyperthyroidism, and 30 had subclinical hyperthyroidism. The characteristics of the participants according to thyroid function state are presented in Table 1. Participants with clinical hypothyroidism and hyperthyroidism were significantly older than those with normal thyroid function. Those with subclinical hypothyroidism had significantly lower fasting serum glucose, whereas those with subclinical hyperthyroidism had significantly higher fasting serum glucose than those with normal function. The prevalence of smokers and full-time employment was highest in those with subclinical hyperthyroidism. TPOAb positivity was most prevalent in clinical hypothyroidism, followed by clinical hyperthyroidism.

Both exploratory dietary patterns accounted for 4% of the food group variance. The native-based pattern explained 2.1% of the variance in the total response variables, while the EPIC-based pattern explained 1.9%. The native-based pattern was highly (factor loading ≥ 0.2) positively loaded with fast foods, pickled vegetables, breads, soft drinks, legumes, and cooked vegetables but negatively loaded with dried fruits, confectionary, potatoes, butter, and jam and honey. The EPIC-based pattern was highly positively associated with pizza, soft drinks, legumes, and processed meats and negatively associated with fried potatoes, confectionery, cooked potatoes, jam and honey, and cake and cookies. The factor loadings of each food group for the native-based pattern are described in Additional file 1: Table S1, while those for the EPIC-based pattern are described in Additional file 1: Table S2. Both patterns were positively correlated with all response variables, with the highest and lowest correlations with fasting serum glucose and HDL-C, respectively. The FOS-based pattern only showed a weak positive association with BMI, SBP, and DBP (Table 2).

Table 3 shows associations between dietary patterns, TSH, FT4, and TPOAb in individuals with normal thyroid function. In the unadjusted model, there was a 0.08 mU/L (95% CI 0.04, 0.12) and 0.06 mU/L (95% CI 0.03, 0.11) increase in serum TSH concentration per one SD increase in the scores of the native-based pattern and the EPIC-based pattern, respectively. The associations remained significant after taking into account the covariates. These two patterns showed no significant association with FT4. A marginally significant inverse association was observed between the FOS-based pattern and TSH, which became non-significant in the multivariable adjusted model (p value = 0.069).

Table 4 presents the associations between dietary pattern scores and the odds of clinical and subclinical hypothyroidism. There was an increase in the odds of clinical hypothyroidism across the tertiles of the native-based pattern (p-trend = 0.033) in the multivariable-adjusted model. The odds of clinical hypothyroidism was significantly higher in tertile 3 compared to tertile 1 (HR = 1.65; 95% CI 1.04, 2.62) in the adjusted model 2. The other dietary patterns were not related to clinical hypothyroidism. The risk of subclinical hypothyroidism was significantly greater in the second and third tertiles of the native-based pattern compared to the first tertile, with ORs (95% CIs) of 1.49 (1.11, 1.98) and 1.54 (1.41, 2.09) in the adjusted model 2 (p-trend = 0.005). As a continuous variable, one SD increase in the dietary pattern score 1 was associated with a 14% higher odds of subclinical hypothyroidism after adjusting for covariates (OR = 1.14, 95% CI 1.01, 1.28). In addition, the odds of subclinical hypothyroidism increased across the tertiles of the EPIC-based pattern, but the odds was significantly higher only in tertile 3 compared to tertile 1, with an OR (95% CI) of 1.44 (1.07, 1.94) in the adjusted model 2. One SD increase in the score of the EPIC-based pattern was also related to a 16% higher odds of subclinical hypothyroidism (OR = 1.16, 95% CI 1.03, 1.31) in the adjusted model 2.

Figure 2 illustrates the odds of clinical hyperthyroidism and subclinical hyperthyroidism per SD of each dietary pattern. The dietary pattern was not related to the odds of hyperthyroidism. In addition, no significant association was observed between dietary patterns and the odds of TPOAb positivity (Table 5).

We identified two dietary patterns that were positively correlated with all six cardiometabolic markers, including BMI, serum glucose, HDL-C, triglycerides, SBP, and DBP. The native-based pattern was characterized by high intakes of fast foods, pickled vegetables, breads, soft drinks, legumes, and cooked vegetables and low intakes of dried fruits, confectionary, potatoes, butter, and jam–honey. The EPIC-based pattern was characterized by high intakes of pizza, soft drinks, legumes, and processed meats and low intakes of fried potatoes, confectionery, jam and honey, cooked potatoes, butter, cake, and cookies. These two cardiometabolic-related dietary patterns were associated with higher TSH levels in participants with normal thyroid function and an increased odds of subclinical hypothyroidism.

Insufficient studies have examined the relationship between nutrition and thyroid function within the context of dietary patterns [9‐11]. Two cross-sectional studies that computed an a priori dietary index indicated a negative association between the Mediterranean diet and FT3 and FT4 [9] and a positive association between the dietary inflammatory index and total T4 [10]. In addition, a cross-sectional study employing the method of PCA identified 18 dietary components with varying associations with thyroid function. High consumption of fruit juices, Cedevita vitamin drinks, and non-alcoholic drinks was associated with lower TSH and higher FT4, whereas high consumption of bacon and sausage was associated with higher FT3 and FT4. A factor with a high positive loading for white bread and a high negative loading for whole grains was also significantly associated with FT4 [11].

Using the RRR method and the EPIC food groups, the FOS also kept a dietary pattern that was linked to the 5 cardio-metabolic variables (BMI, serum glucose, triglycerides, HDL-C, SBP, and DBP). The FOS-suggested dietary pattern was positively associated with BMI, fasting glucose, triglycerides, SBP, and DBP and negatively associated with HDL-C [15]. However, the contribution of food groups to the dietary pattern was different from ours. We also made a simplified dietary pattern based on the FOS factor loadings. However, the FOS-based pattern only showed very weak correlations with SBP and DBP among the cardiometabolic variables, suggesting this pattern was unrelated to cardiometabolic parameters in our population. Using total cholesterol, triglycerides, and HDL-C as intermediate variables and the RRR method, the Whitehall I study suggests two dietary patterns [25]. Moreover, the Multi-Ethnic Study of Atherosclerosis (MESA) identified a dietary pattern that was positively correlated with waist circumference, serum glucose, triglycerides, SBP, and DBP and negatively correlated with HDL-C [26]. The characteristics of diets related to cardio-metabolic variables vary between studies. However, food groups that were consistent in these patterns and ours revealed that a pattern with high intakes of processed meats, pizza, and soft drinks is associated with higher glucose, BMI (or waist circumference), triglycerides, SBP, and DBP. Some prospective studies have also found an association between processed meat and soft drink consumption and metabolic abnormalities such as obesity, elevated blood pressure, and abnormal blood glucose [27, 28].

Cardiometabolic variables and thyroid function are closely and reciprocally associated. Some studies confirmed the causal associations between thyroid function and cardiometabolic health [29‐31]. On the other hand, BMI and HDL-C have been suggested to have a causal effect on thyroid function [29, 30, 32]. Moreover, diabetes is also associated with a higher prevalence of thyroid dysfunction [33]. A prospective study also demonstrated a higher incidence of subclinical hypothyroidism among those with metabolic syndrome during an average follow-up period of 4.2 years. Among the components of the metabolic syndrome, high blood pressure and high triglyceridemia showed significant positive associations with subclinical hypothyroidism [34]. The association between these dietary patterns and subclinical hypothyroidism was, therefore, partially explained by cardio-metabolic variables, particularly serum glucose and BMI. Legumes and vegetables were positively loaded into the exploratory dietary patterns identified in the current study. Some plant constituents, such as cyanogenic glucosides, polyphenols, phenolic acids, and alkaloids, have been proposed to have anti-thyroid effects. However, the interference of the constituents with thyroid function depends on their amount of ingestion and iodine status, and the amount consumed through diet is insufficient to affect thyroid function [35].

This study investigates for the first time the association between RRR-retained dietary patterns and thyroid function. The RRR method allows for the generation of hypotheses regarding potential mediators between patterns and thyroid function [14]. Other strengths of the study include the use of population-based data, a relatively large sample size, and a validated dietary assessment instrument. The findings of the current study are clinically important because subclinical hypothyroidism is a common endocrinopathy that can increase the risk of metabolic syndrome, cardiovascular diseases, cognitive decline, and mortality [36‐38]. According to this study, dietary patterns may be an important modifiable risk factor for subclinical hypothyroidism.

However, there are also some limitations that should be considered. The cross-sectional design of the study means that causality cannot be established and reverse causation cannot be ruled out. In addition, the study was conducted on the population of Tehran, so the results may not be generalizable to other populations. In addition, a small number of participants had hyperthyroidism. Therefore, the sample size used to examine the association between diet and hyperthyroidism was insufficient. Furthermore, concern for residual or unmeasured confounders remained because of the observational design of the study.

In conclusion, a dietary pattern positively related to cardio-metabolic variables, characterized by high intakes of fast foods (pizza and processed meat), soft drinks, and legumes and low intakes of confectionery, potatoes, butter, and jam and honey, was associated with higher levels of TSH in participants with euthyroidism and an increased odds of subclinical hypothyroidism.