Abstract
Objective
Thyroid-stimulating hormone (TSH) is influenced by genetic and environmental factors such as socioeconomic position (SEP). However, interactions between TSH-related genetic factors and indicators of SEP have not been investigated to date. The aim of the study was to determine whether education and income as SEP indicators may interact with TSH-related genetic effect allele sum scores (GESTSH_2013 and GESTSH_2020) based on two different GWAS meta-analyses that affect TSH values in a population-based study.
Methods
In 4085 participants of the Heinz Nixdorf Recall Study associations between SEP indicators, GESTSH and TSH were quantified using sex- and age-adjusted linear regression models. Interactions between SEP indicators and GESTSH were assessed by GESTSH × SEP interaction terms, single reference joint effects and calculating genetic effects stratified by SEP group.
Results
Participants within the highest education group showed the strongest genetic effect with on average 1.109-fold (95% CI: 1.067–1.155) higher TSH values per GESTSH_2013 SD, while in the lowest education group, the genetic effect was less strong (1.061-fold (95% CI: 1.022–1.103)). In linear regression models including interaction terms, some weak indication for a positive GESTSH_2013 by education interaction was observed showing an interaction effect size estimate of 1.005 (95% CI: 1.000–1.010) per year of education and GESTSH_2013 SD. No indication for interaction was observed for using income as SEP indicator. Using the GESTSH_2020, similar results were observed.
Conclusion
Our results gave some indication that education may affect the expression of TSH-related genetic effects. Stronger genetic effects in high-education groups may be explained by environmental factors that have an impact on gene expression and are more prevalent in high SEP groups.
Introduction
Thyroid hormones play an important role in almost all biological processes in humans such as energy metabolism, cell growth, development, cardiovascular system, CNS, immunity and bone metabolism (1, 2). Importantly, several studies have shown associations between overt or subclinical thyroid dysfunction and clinical endpoints such as cardiovascular disease, atrial fibrillation (3), dyslipidaemia (3), hypertension (4), atherosclerosis (5), type 2 diabetes (3) and bone fractures (6). Even psychological symptoms such as depression (7) or reduced quality of life (8) are associated with subclinical thyroid dysfunction. Because thyroid-stimulating hormone (TSH) reacts strong to minor changes in thyroid hormone values, abnormal TSH values are more sensitive markers for early thyroid dysfunction including hypothyroidism as well as hyperthyroidism than the measurement of thyroid hormones values (9). Further, it has been reported that thyroid function is influenced by genetic variation as well as social factors such as socioeconomic position (SEP), both contributing to substantial inter-individual differences observed in healthy populations (10, 11, 12). For instance, Völzke et al. (13) have detected higher TSH values in unemployed participants compared to employed participants in a population-based cohort. In addition, it has been indicated that there is a greater chance of developing thyroid dysfunction with lower income and higher deprivation levels (10, 12), while higher education may reduce the risk for low TSH values (14).
While every human has an individual hypothalamic–pituitary–thyroid axis setpoint within a range that is predominately determined by genes mainly responsible for intra-individual differences (15), twin studies have also suggested a heritability of 57–71% for inter-individual variance in TSH serum values (16, 17). Genome-wide association studies (GWAS) have already identified genetic loci that are associated with thyroid function. Porcu et al. (2013) (18) performed a GWAS on TSH values within the normal range in over 25,000 participants and found 20 robustly associated genetic loci, which explained 5.6% of the total variation in serum TSH values of the study population. In a more recent GWAS, Zhou et al. (2020) (19) identified 74 genome-wide significant loci, which had on average weaker individual effects on TSH compared to previous GWAS results, but explained 13.3% of the total variation in serum TSH values. Results have also suggested that the total genetic effect on TSH is determined by many common genetic variants with a small individual effect (15). While it has been hypothesised that interactions between genetic and environmental factors (G×E) may partly account for the missing heritability of common traits and the low penetrance of single common genetic variants (20), interactions between TSH-related genetic factors and indicators of SEP have not been studied to date.
The aim of the present study was to examine whether SEP indicators’ education and income interact with TSH-related genetic effect allele sum scores to affect TSH serum values in a population-based study.
Material and methods
Study population
All analyses were based on the population-based Heinz Nixdorf Recall Study. From December 2000 to August 2003, a total of 4814 participants aged 45–75 years were randomly selected and recruited from mandatory citizen registries of the three cities Bochum, Essen and Mülheim/Ruhr. These cities are located in the Western part of Germany. The baseline response proportion was 55.8% (21). More information about the study design has been provided in detail elsewhere (22). The study was approved by the local ethics committee of the University of Duisburg-Essen. The study involves extended quality management procedures with a certification to DIN ISO 9001:2000. All participants gave written informed consent.
Indicators of socioeconomic position
Education and income were used as indicators of SEP. Information was collected at study baseline by using standardised computer-assisted face-to-face interviews. The International Standard Classification of Education (ISCED) was used to define education as total years of formal education by combining school and vocational training (23). It was then either used as continuous variable or categorised into four groups for stratified analyses with the lowest group of ≤10 years of education, which is equivalent to a basic school degree with no vocational training, and the highest group of ≥18 years of education, which is equivalent to a university degree. Income was measured as the monthly household equivalent income calculated by dividing the participants’ household net income by a weighting factor for each household member. Income was used as a continuous variable or was categorised into sex-specific quartiles. Education and income were analysed separately to consider their different mechanism in causing social inequalities in health (24, 25).
TSH
Serum TSH values were assessed from frozen blood samples with the Roche Modular Analytics E170 electro-chemiluminescence immunoassay (Roche Diagnostic). The working range for this method was 0.005–100 mIU/L. The functional sensitivity amounts to 0.014 mIU/L. TSH values were used as a continuous variable except for some of the analysis, for which, the main analysis population was categorised into euthyroid, hypothyroid and hyperthyroid participants. The reference range of 0.27–4.20 mIU/L by the manufacturer was adopted to define participants as euthyroid. Hyperthyroid status was defined as TSH values <0.27 mIU/L, while hypothyroid status was defined as TSH values >4.20 mIU/L.
Genetic data
Lymphocyte DNA was isolated from EDTA anti-coagulated venous blood using the Chemagic Magnetic Separation Module I (Chemagen, Baesweiler, Germany). The meta-analyses of genome-wide association studies by Porcu et al. (2013) (18) and Zhou et al. (2020) (19) were used to identify lead single nucleotide polymorphisms (SNPs) representing independent genetic loci robustly associated with TSH on the global screening array 24v1.0 available for all study participants included in the analysis (Supplementary Tables 1 and 2, see section on supplementary materials given at the end of this article). Genotype imputation was carried out using IMPUTE version 2 (v.2.3.1) (26, 27) with reference data from 1000 Genomes Phase 1 (v3). No deviation from Hardy–Weinberg equilibrium (HWE) (P ≤ 0.001) was detected. To compare the genetic effects from two different GWAS meta-analyses that reflect differences in the average individual effect size of the reported genome-wide significant SNPs, two weighted genetic effect allele sum scores (i.e. GESTSH_2013 including 19 SNPs based on the Porcu et al. meta-analysis and GESTSH_2020 including 126 SNPs based on the Zhou et al. meta-analysis) were calculated by aggregating the total number of TSH increasing alleles for each individual across the selected SNPs using the software PRSice-2 (28). Both scores were weighted by the corresponding effect-size estimate of each SNP reported in the literature. For SNP clumping, a linkage disequilibrium (LD) cut-off of r2 = 0.1 was used. To reflect the number of TSH-increasing alleles, both GES were rescaled by multiplying the weighted sum score by the number of SNPs included and then dividing it by the sum of effect-size estimates of the SNPs reported in the respective meta-analysis. For comparing results between both GES, effect size estimates were reported per s.d. of the respective GES. Participants of the Heinz Nixdorf Recall Study are of European ancestry and genetically homogeneous.
Statistical analysis
Out of the baseline study population, 4085 participants were included in the main analysis population, as participants with missing genetic information or missing TSH values were excluded from analysis. In addition, two participants with TSH values > 80 mIU/L were excluded to avoid extreme outliers. For sensitivity analysis, additional 1293 participants reporting a history of thyroid dysfunction (including thyroid cancer or thyroid removal) or intake of thyroid medication were excluded (Fig. 1).
Flowchart of participants out of the Heinz Nixdorf Recall study included in the analysis.
Citation: Endocrine Connections 12, 2; 10.1530/EC-22-0127
For statistical analysis, TSH values were log-transformed using the natural logarithm to normalise the distribution. To prevent numerical errors, the TSH values were corrected by adding 1 before log-transformation (ln(TSH+1)). Results of linear regression analysis were back-transformed and reported as Exp(β) with 95% CI. Exp(β) can be interpreted as the factor change of the geometric mean (TSH+1) value per unit change of the independent variable, while it has to be stated that the addition of the relatively high value of 1 to TSH prior to log-transformation limits the interpretation of the obtained effect size estimates as relative change. To assess the associations of SEP indicators and the GESTSH_2013/GESTSH_2020 with TSH values, linear regression analysis with continuous ln(TSH+1) as dependent variable was performed. In addition, associations of SEP indicators with thyroid status (euthyroid/hypothyroid/hyperthyroid) as dependent variable were assessed using logistic regression analysis in order to account for a possible U-shaped relationship between SEP and TSH. In the logistic regression analyses, the thyroid status was set as the dependent variable with euthyroid participants as the reference group and either hyperthyroid or hypothyroid participants as cases. Further, sex-stratified analyses were performed.
The main effect of the GESTSH_2013/GESTSH_2020 on TSH was calculated using linear regression models with continuous ln(TSH+1) as dependent variable. The effect of GESTSH_2013/GESTSH_2020 on TSH was also stratified by education groups and income quartiles to assess heterogeneity of the genetic effect across SEP strata. Additionally, sex-stratified analysis of the effect of GESTSH_2013/GESTSH_2020 on TSH was performed. The GESTSH_2013 was then divided into tertiles and all possible combinations of GESTSH_2013 tertiles and SEP indicators were entered as dummy variables into a linear regression model with the group of lowest GESTSH_2013 and lowest education or income as reference to analyse single reference joint effects between the genetic effect and SEP indicators on TSH. In addition, linear regression models were fitted including GESTSH_2013/GESTSH_2020 by SEP indicator interaction terms. All regression models were adjusted for sex and age to control for confounding. Statistical analyses were performed using the SPSS software (SPSS Inc., version 27.0).
Results
Half of the study participants were women (Table 1). The mean (±s.d.) age of the main analysis population was 59.7 ± 7.8 years. The median TSH value in this main analysis population was 1.26 mIU/L (interquartile range: 0.83–1.84). Men and women did not differ strongly in median TSH values in the study population. However, there was a substantial difference between men and women in SEP. As compared to women, men were on average higher educated and had a higher household income in the study population.
Characteristics of main analysis population stratified by sex.
| All (n = 4085) | Men (n = 2039) | Women (n = 2046) | |
|---|---|---|---|
| Age (years)a | 59.7 ± 7.8 | 59.7 ± 7.8 | 59.7 ± 7.8 |
| Income (€/month) c nmiss=258 | 1448 (1107–1874) | 1520 (1107–2072) | 1278 (937–1874) |
| Lowest income quartile | 1206 (31.5%) | 518 (26.5%) | 688 (36.7%) |
| Second income quartile | 885 (23.1%) | 422 (21.6%) | 463 (24.7%) |
| Third income quartile | 533 (13.9%) | 294 (15.0%) | 239 (12.8%) |
| Highest income quartile | 1203 (31.4%) | 720 (36.8%) | 483 (25.8%) |
| Education (years of training) b nmiss=12 | |||
| ≤10 years | 474 (11.6%) | 97 (4.8%) | 377 (18.4%) |
| 11–13 years | 2249 (55.2%) | 971 (47.9%) | 1278 (62.5%) |
| 14–17 years | 922 (22.6%) | 687 (33.9%) | 235 (11.5%) |
| ≥18 years | 428 (10.5%) | 274 (13.5%) | 154 (7.5%) |
| TSH values (mIU/L)c | 1.26 (0.83–1.84) | 1.24 (0.87–1.74) | 1.28 (0.77–1.94) |
| Hyperthyroid (TSH value <0.27 mIU/L)b | 183 (4.5%) | 50 (2.5%) | 133 (6.5%) |
| Hypothyroid (TSH value >4.2 mIU/L)b | 120 (2.9%) | 35 (1.7%) | 85 (4.2%) |
| Thyroid medication | 525 (12.9%) | 89 (4.4%) | 436 (21.3%) |
| TSH-associated genetic effect allele sum score (GESTSH_2013)a | 24.7 ± 2.7 | 24.7 ± 2.7 | 24.6 ± 2.7 |
| TSH-associated genetic effect allele sum score (GESTSH_2020)a | 32.3 ± 4.6 | 32.3 ± 4.5 | 32.3 ± 4.6 |
aMean ± s.d.;
b n (%);
cMedian (first quartile to third quartile).
nmiss, number of participants with missing values.
The calculation of the association between GESTSH_2013 and TSH showed an on average 1.088-fold (95% CI: 1.073–1.100) higher TSH value per standard deviation of the GESTSH_2013 (Table 2), while for the GESTSH_2020, similar results were observed (Supplementary Table 3). An insignificantly stronger GESTSH_2013 effect was observed for men compared to women in the study population (Supplementary Table 4). An association between SEP indicators and TSH could not be observed (Table 2). Using categorical TSH as dependent variable in logistic regression analysis, there was some indication for an association of education and income, having TSH values outside of the reference range in crude regression models (Table 3 and 4). However, results after adjustment for age and sex indicated that this association was mainly explained by confounding while female sex was associated with having TSH values outside of the reference range in all regression models.
Sex- and age-adjusted effects (Exp(β)) and corresponding 95% CIs on thyroid-stimulating hormone (TSH) values in separate linear regression models including main effects of education (per year) and income (per 1000 €/month) as indicators of socioeconomic position (SEP) and a TSH-associated genetic effect allele sum score (GESTSH_2013; per s.d.).
| Exp(β) (95% CI) | P | |
|---|---|---|
| Education | ||
| Sex (female) | 1.015 (0.989–1.041) | 0.25 |
| Age | 1.000 (0.998–1.001) | 0.84 |
| Education (per year) | 0.997 (0.991–1.002) | 0.22 |
| Income | ||
| Sex (female) | 1.021 (0.996–1.048) | 0.10 |
| Age | 1.000 (0.998–1.002) | 0.92 |
| Income (per 1000 €/month) | 1.000 (0.982–1.018) | 0.99 |
| GESTSH_2013 | ||
| Sex (female) | 1.022 (0.998–1.047) | 0.07 |
| Age | 1.000 (0.999–1.002) | 0.91 |
| GESTSH_2013 (per s.d.) | 1.088 (1.073–1.100) | 2.1 × 1−42 |
Odds ratios (OR) and 95% CIs for the association of income (per 1000 €/month) and education (per year) with hyperthyroid vs euthyroid status.
| OR (95% CI) | P | |
|---|---|---|
| Education | ||
| Hyperthyroid ~ education |
||
| Education (per year) | 0.918 (0.862–0.979) | 8.7 × 10−3 |
| Hyperthyroid ~ education + age + sex | ||
| Education (per year) | 0.999 (0.932–1.070) | 0.97 |
| Age | 1.024 (1.004–1.044) | 0.02 |
| Sex (female) | 2.830 (1.996–4.014) | 5.3 × 10−9 |
| Income | ||
| Hyperthyroid ~income | ||
| Income (per 1000 €/month) | 0.736 (0.581–0.932) | 0.01 |
| Hyperthyroid ~ income + age + sex | ||
| Income (per 1000 €/month) | 0.840 (0.662–1.065) | 0.15 |
| Age | 1.020 (1.000–1.040) | 0.05 |
| Sex (female) | 2.845 (2.014–4.019) | 3.0 × 10−9 |
Odds ratios (OR) and 95% CIs for the association of income (per 1000 €/month) and education (per year) with hypothyroid vs euthyroid status.
| OR (95% CI) | P | |
|---|---|---|
| Education | ||
| Hypothyroid ~ education | ||
| Education (per year) | 0.875 (0.808–0.947) | 9.9 × 10−4 |
| Hypothyroid ~ education + age + sex
|
||
| Education (per year) | 0.921 (0.846–1.004) | 0.06 |
| Age | 0.997 (0.974–1.021) | 0.82 |
| Sex (female) | 2.286 (1.505–3.472) | 1.0 × 10−4 |
| Income | ||
| Hypothyroid ~ income | ||
| Income (per 1000 €/month) | 0.771 (0.582–1.023) | 0.07 |
| Hypothyroid ~ income+ age + sex
|
||
| Income (per 1000 €/month) | 0.842 (0.634–1.119) | 0.24 |
| Age | 0.999 (0.975–1.023) | 0.91 |
| Sex (female) | 2.636 (1.743–3.988) | 4.0 × 10−6 |
In stratified analyses, the effect of GESTSH_2013 on TSH increased with increasing education levels (Fig. 2). Participants with the lowest education (≤10 years) had the weakest genetic effect with on average 1.061-fold (95% CI: 1.022–1.103) higher TSH values per s.d. of the GESTSH_2013, while participants with the highest education (≥18 years) had the strongest genetic effect with on average 1.109-fold (95% CI: 1.067–1.155) higher TSH values per SD. Although a trend of increasing genetic effects on TSH with increasing income quartile was indicated, the differences between income quartiles were less marked. Overall, results showed a positive trend in the genetic main effect on TSH across education groups.
Sex-and age-adjusted effects and corresponding 95% CI of the genetic effect per standard deviation (GESTSH_2013) on thyroid-stimulating hormone (TSH) values stratified by education groups (in years) and income quartiles in linear regression models.
Citation: Endocrine Connections 12, 2; 10.1530/EC-22-0127
The analysis of joint effects between SEP and GESTSH_2013 categories revealed a clear trend within but not between the different SEP/GESTSH_2013 groups (Table 5). Regarding the analysis of education, the joint effect was the strongest in the group with high GESTSH_2013 in the highest education group, with on average 1.564-fold (95% CI: 1.217–2.011) higher TSH values compared to the reference category of low education and low GESTSH_2013.
Sex- and age-adjusted effects (Exp(β)) and corresponding 95% CIs on thyroid-stimulating hormone (TSH values) in linear regression models of the joint effects of tertiles of a TSH-associated effect allele sum score (GESTSH_2013) and socioeconomic position indicators, calculated separately for income quartiles and education categories with the group of having a low genetic effect allele sum score and the lowest socioeconomic position as reference.
| n | Exp(β) (95% CI) | P | |
|---|---|---|---|
| Education/GESTSH_2013 groups | |||
| ≤10 years | |||
| Low GESTSH_2013 | 159 | Reference category | Reference category |
| Middle GESTSH_2013 | 162 | 1.223 (0.970–1.543) | 0.09 |
| High GESTSH_2013 | 153 | 1.518 (1.200–1.925) | 5.1 × 10−4 |
| 11–13 years | |||
| Low GESTSH_2013 | 737 | 0.946 (0.790–1.136) | 0.56 |
| Middle GESTSH_2013 | 749 | 1.184 (0.986–1.418) | 0.07 |
| High GESTSH_2013 | 763 | 1.526 (1.275–1.833) | 5.0 × 10−6 |
| 14-17 years | |||
| Low GESTSH_2013 | 317 | 0.839 (0.686–1.027) | 0.09 |
| Middle GESTSH_2013 | 291 | 1.146 (0.934–1.406) | 0.19 |
| High GESTSH_2013 | 314 | 1.506 (1.230–1.848) | 7.2 × 10−5 |
| ≥18 years | |||
| Low GESTSH_2013 | 147 | 0.858 (0.674–1.088) | 0.20 |
| Middle GESTSH_2013 | 159 | 1.177 (0.931–1.485) | 0.17 |
| High GESTSH_2013 | 122 | 1.564 (1.217–2.011) | 4.6 × 10−4 |
| Income/GESTSH_2013 groups | |||
| Lower quartile | |||
| Low GESTSH_2013 | 400 | Reference category | Reference category |
| Middle GESTSH_2013 | 399 | 1.187 (1.027–1.376) | 0.02 |
| High GESTSH_2013 | 407 | 1.684 (1.457–1.951) | 2.7 × 10−12 |
| Second quartile | |||
| Low GESTSH_2013 | 279 | 0.921 (0.784–1.082) | 0.32 |
| Middle GESTSH_2013 | 304 | 1.306 (1.115–1.530) | 8.9 × 10−4 |
| High GESTSH_2013 | 302 | 1.497 (1.278–1.754) | 6.0 × 10−7 |
| Third quartile | |||
| Low GESTSH_2013 | 179 | 0.957 (0.795–1.152) | 0.64 |
| Middle GESTSH_2013 | 170 | 1.036 (0.858–1.254) | 0.71 |
| High GESTSH_2013 | 184 | 1.652 (1.376–1.989) | 9.7 × 10−8 |
| Highest quartile | |||
| Low GESTSH_2013 | 416 | 0.931 (0.806–1.076) | 0.34 |
| Middle GESTSH_2013 | 414 | 1.313 (1.136–1.518) | 2.3 × 10−4 |
| High GESTSH_2013 | 373 | 1.608 (1.387–1.868) | 4.7 × 10−10 |
Linear regression models including interaction terms between SEP indicators and the GESTSH_2013 gave some weak indication for GESTSH_2013 by education interaction showing an effect size estimate for the interaction term of 1.005 (95% CI: 1.000–1.010) per year of education and s.d. of the GESTSH_2013. No indication for a GESTSH_2013 by income interaction was observed, although the direction of the effect size estimate (1.008 (95% CI: 0.991–1.027) per 1000 €/month and s.d. of the GESTSH) was consistent with the observed GESTSH_2013 by education interaction (Table 6). The effect size estimates for the observed GESTSH_2013 by education interaction were slightly stronger in men compared to women (Supplementary Table 5) and slightly stronger compared to the GESTSH_2020 by education interaction effect size estimate (Supplementary Table 6).
Sex- and age-adjusted effects (Exp(β)) and corresponding 95% CIs on thyroid-stimulating hormone (TSH values) in linear regression models including interaction terms of a TSH-associated genetic effect allele sum score (GESTSH_2013; per s.d.), education (per years of education) and income (per 1000 €/month) as indicator of socioeconomic position.
| Exp(β) (95% CI) | P | |
|---|---|---|
| Education | ||
| Sex (female) | 1.019 (0.994–1.044) | 0.14 |
| Age | 1.000 (0.998–1.001) | 0.93 |
| Education (per year) | 0.953 (0.910–0.997) | 0.04 |
| GESTSH_2013 | 1.013 (0.944–1.085) | 0.72 |
| GESTSH_2013 × education | 1.005 (1.000–1.010) | 0.045 |
| Income | ||
| Sex (female) | 1.024 (0.999–1.050) | 0.06 |
| Age | 1.000 (0.998–1.002) | 0.96 |
| Income (per 1000 €/month) | 0.920 (0.785–1.078) | 0.31 |
| GESTSH_2013 | 1.073 (1.041–1.106) | 2.0 × 10−6 |
| GESTSH_2013 × income | 1.008 (0.991–1.027) | 0.29 |
In the sensitivity analysis, participants were excluded if they had a positive history of thyroid dysfunction or reported taking medications that affect the thyroid gland. Thus, three times more women than men were excluded (Supplementary Table 7). Overall, effect size estimates of the main results did not differ compared to the main analysis population (Supplementary Tables 8, 9 and Supplementary Fig. 1). However, due to the smaller sample size in the sensitivity analysis, the precision of the effect size estimation was decreased, leading to wider 95% CIs.
Discussion
The results of the present study gave some indication that SEP indicator educational attainment may affect serum TSH values by interacting with a TSH-related weighted genetic effect allele sum score. With higher education, a stronger effect of the GESTSH_2013 on TSH values was observed. This is supported by education-stratified analyses of the genetic effect on TSH values. Also, joint effects of all possible combinations of GESTSH_2013 tertiles and education groups showed the strongest effect on TSH values for participants with the highest GESTSH_2013 within the highest education group. Effect size estimates for the GESTSH_2013 by education interaction term gave some support for interaction. Results for the GESTSH_2020 showed slightly less strong effect size estimates. This may reflect the on-average weaker effects of individual SNP on TSH in the GESTSH_2020.
The direction of the GESTSH_2013 main effect on TSH was consistent with previous studies conducted by Porcu et al. and Chaker et al. (18, 29). In these studies, the same set of SNPs has been used in a polygenic risk score. Importantly, the genetic effect in our study differs only slightly between sexes. A stronger genetic effect in men has been explained by sex-specific strength of effects of genetic loci included in the GESTSH_2013. Porcu et al. (18) detected that TSH-elevating alleles in gene MAF/LOC440389 and gene PDE10A were related to a stronger TSH increase in men, while this sex difference did not replicate in the extended GWAS of Teumer et al. (2018) (30) Interestingly, recent studies have reported that women showed a higher heritability for TSH values compared to men (31, 32). An explanation for this result might be that the previously identified loci just explain a fraction of the total variation in serum TSH values indicating substantial missing heritability. Thus, the exact nature of sex differences in the genetic effects on TSH might not be fully reflected by the GESTSH used in the present study.
Sex and age are potential confounders for the association between SEP and thyroid function. In the present study, an association was observed between female sex having TSH values outside the reference range, which is in line with previous studies (10, 33, 34, 35). Associations between age and TSH values have been discussed controversially in the literature. Similar to the present study, some studies have shown no indication of an association between age and TSH (36, 37). In contrast, an increase in TSH values with increasing age was observed in other studies (38, 39). These discrepancies may be due to the age structure of each cohort and regional differences in iodine supply.
Having a higher income or more years of education has been reported to decrease the chance of being hypothyroid or hyperthyroid depending on serum TSH values in previous studies, even though this association was not strongly indicated and confounded by age and sex in the present study. Santos Palacios et al. (10) suggested that the risk of developing hyper- or hypothyroidism may be reduced with increasing income. Other studies have shown that higher education is associated with lower incidence of hyperthyroid serum TSH values (13, 14). Wilson et al. (12) have used a deprivation score as SEP indicator to evaluate the association between TSH and SEP. Similarly, they showed that subclinical hyperthyroidism is positively associated with higher deprivation score values. However, subclinical hypothyroidism was more common in participants with the lowest deprivation score. Probably the relative effect of income and education decreased when combining the SEP indicators with other domains, for example, employment or overall health (12).
It has been hypothesised that individuals with low SEP in general show stronger genetic effects on health than individuals with high SEP. Johnson et al. (40) reported that the variance in physical health explained by genetic factors decreases with increasing income in a twin study. Individuals with low SEP are exposed to more health-related risk factors, for example, greater exposure to stress, stronger exposure and vulnerability to behavioural risk factors, and have a higher prevalence of different diseases (41, 42, 43, 44, 45). There are indications that adverse environmental factors may interact with genetic factors and influence gene expression through epigenetic modifications (46, 47). A possible explanation could be that genes for disease susceptibility can magnify their effect in a disadvantaged environment with higher rates of risk factors (48). This effect was also detected in studies assessing the association between genetic factors, SEP and BMI, coronary artery calcification or incidence of coronary events (49, 50). However, the interaction between genetic factors and SEP on TSH values in the present study pointed in the opposite direction, as stronger genetic effects were observed in groups of higher SEP. There is an alternative explanation for gene by SEP interactions that hypothesised that individuals with high SEP are exposed to less non-genetic risk factors, which leads to an increased genetic effect in groups of high SEP due to a decrease in the effect of non-genetic factors (51). A Danish study investigated the gene–environment interaction by reference to childcare, SEP and problem behaviour in children. They detected a lower heritability of problem behaviour in children with lower SEP. Likewise, decrease in heritability was explained by an increase in the influence of the environment (52). Further, Ge et al. (53) showed that heritability of education increased with increasing SEP. Notably, these studies investigated the effect on heritability, but their results revealed the same direction of interactions as in the present findings. Interestingly, previous studies indicated that genes for mathematical skills are more strongly expressed in high SEP groups, while environmental factors are less important (54, 55). In line with these results, the reported direction of interaction may be explained by a stronger effect of genetic factors in high-education groups, which may present with lower rates of TSH-related non-genetic risk factors.
However, as the overall impact of environmental factors on thyroid dysfunction is largely unknown, there may also be unknown environmental risk factors for thyroid dysfunction which are more prevalent in high education groups. These unknown environmental factors could influence gene expression leading to a stronger genetic effect on TSH values. For instance, several recent studies detected an association between moderate alcohol consumption and lower risk of developing hypo- or hyperthyroidism (14, 56). It is well known that people in high SEP groups are more often moderate drinkers (57). Alcohol and its metabolites play an important epigenetic role in gene expression in neurons through histone acetylation, suggesting a possible influence on gene expression of TSH as well (58, 59). Further, regular physical activity may reduce the risk of developing hypothyroidism, while obesity may be a risk factor for hypothyroidism (14). There are indications that people in high SEP groups are more physically active in their leisure time, while obesity is more frequent in people with low SEP position (60, 61). Further research is required to investigate potential environmental risk factors and their influence on gene expression of TSH-associated genetic loci.
Even further, sensitivity analyses considering only participants without thyroid dysfunction and without thyroid medication did not change the strength and direction of the observed interaction between TSH-related genetic factors and SEP, suggesting that the observed interaction is not mainly triggered by clinical relevant thyroid dysfunction.
Strengths and limitations
The strengths of the present study are its population-based study design and the inclusion of two different SEP indicators. All serum TSH values were measured standardised by the same laboratory with the same immunoassay. The interaction between genetic and socioeconomic factors was not only explored by interaction terms. Stratified analyses and analyses of single reference joint effects supported the main result.
A limitation is the sample size and the limited statistical power for single SNP analyses. Further, time at blood sampling was not standardised, but the secretion of TSH is related to a circadian rhythm showing higher values in the morning and a decrease at noon (62).
Conclusion
To our knowledge, the present study is the first study to explore the interaction between SEP and a genetic effect allele score on TSH. The results gave some weak indication for a modification of the TSH-related genetic effect by education in a population-based study showing stronger genetic effects in groups of high education. Higher SEP groups might be less exposed to non-genetic risk factors so the genetic effect has a stronger impact. Further research is needed to understand the complex relationship between genetic environmental factors and TSH.
Supplementary materials
This is linked to the online version of the paper at https://doi.org/10.1530/EC-22-0127.
Declaration of interest
None of the authors has a relevant conflict of interest.
Funding
Parts of the study were also supported by the German Research Council (DFG) (DFG project: EI 969/2-3, ER 155/6-1;6-2, HO 3314/2-1;2-2;2-3;4-3, INST 58219/32-1, JO 170/8-1, KN 885/3-1, PE 2309/2-1, SI 236/8-1;9-1;10-1,), the German Ministry of Education and Science (BMBF project: 01EG0401, 01GI0856, 01GI0860, 01GS0820_WB2-C, 01ER1001D, 01GI0205).
Data availability statement
Due to data security reasons (i.e. data contain potentially participant identifying information), the Heinz Nixdorf Recall Study does not allow sharing data as a public use file. However, others can access the data used upon request, which is the same way the authors of the present paper obtained the data. Data requests can be addressed to: recall@uk-essen.de.
Acknowledgements
The authors thank the Heinz Nixdorf Foundation (Chairman: Martin Nixdorf; Past Chairman: Dr jur. Gerhard Schmidt (†)), for their generous support of this study. The authors are indebted to all study participants and to both the dedicated personnel of the study centre of the Heinz Nixdorf Recall study and to the investigative group, in particular to U. Slomiany, U. Roggenbuck, E. M. Beck, A. Öffner, S. Münkel, R. Peter, and H. Hirche. Advisory Board: Meinertz T., Hamburg, Germany (Chair); Bode C., Freiburg, Germany; deFeyter P. J., Rotterdam, Netherlands; Güntert B, Halli, Austria; Gutzwiller F., Bern, Switzerland; Heinen H., Bonn, Germany; Hess O., Bern, Switzerland; Klein B., Essen, Germany; Löwel H., Neuherberg, Germany; Reiser M., Munich, Germany; Schmidt G., Essen, Germany; Schwaiger M., Munich, Germany; Steinmüller C., Bonn, Germany; Theorell T., Stockholm, Sweden; Willich S. N., Berlin, Germany.
References
- 1↑
Ortiga-Carvalho TM, Chiamolera MI, Pazos-Moura CC, Wondisford FE. Hypothalamus-pituitary-thyroid axis. Comprehensive Physiology 2016 6 1387–1428. (https://doi.org/10.1002/cphy.c150027)
- 2↑
Gnocchi D, Steffensen KR, Bruscalupi G, Parini P. Emerging role of thyroid hormone metabolites. Acta Physiologica 2016 217 184–216. (https://doi.org/10.1111/apha.12648)
- 3↑
Manolis AA, Manolis TA, Melita H, Manolis AS. Subclinical thyroid dysfunction and cardiovascular consequences: an alarming wake-up call? Trends in Cardiovascular Medicine 2020 30 57–69. (https://doi.org/10.1016/j.tcm.2019.02.011)
- 4↑
Åsvold BO, Bjøro T, Nilsen TIL, Vatten LJ. Association between blood pressure and serum thyroid-stimulating hormone concentration within the reference range: a population-based study. Journal of Clinical Endocrinology and Metabolism 2007 92 841–845. (https://doi.org/10.1210/jc.2006-2208)
- 5↑
Syamsunder AN, Pal P, Pal GK, Kamalanathan CS, Parija SC, Nanda N, Sirisha A. Decreased baroreflex sensitivity is linked to the atherogenic index, retrograde inflammation, and oxidative stress in subclinical hypothyroidism. Endocrine Research 2017 42 49–58. (https://doi.org/10.1080/07435800.2016.1181648)
- 6↑
Blum MR, Bauer DC, Collet TH, Fink HA, Cappola AR, da Costa BR, Wirth CD, Peeters RP, Åsvold BO & den Elzen WP et al.Subclinical thyroid dysfunction and fracture risk: a meta-analysis. JAMA 2015 313 2055–2065. (https://doi.org/10.1001/jama.2015.5161)
- 7↑
Ittermann T, Völzke H, Baumeister SE, Appel K, Grabe HJ. Diagnosed thyroid disorders are associated with depression and anxiety. Social Psychiatry and Psychiatric Epidemiology 2015 50 1417–1425. (https://doi.org/10.1007/s00127-015-1043-0)
- 8↑
Gulseren S, Gulseren L, Hekimsoy Z, Cetinay P, Ozen C, Tokatlioglu B. Depression, anxiety, health-related quality of life, and disability in patients with overt and subclinical thyroid dysfunction. Archives of Medical Research 2006 37 133–139. (https://doi.org/10.1016/j.arcmed.2005.05.008)
- 9↑
Soh SB, Aw TC. Laboratory testing in thyroid conditions–pitfalls and clinical utility. Annals of Laboratory Medicine 2019 39 3–14. (https://doi.org/10.3343/alm.2019.39.1.3)
- 10↑
Santos Palacios S, Llavero Valero M, Brugos-Larumbe A, Díez JJ, Guillén-Grima F, Galofré JC. Prevalence of thyroid dysfunction in a large southern European population. Analysis of modulatory factors. The APNA study. Clinical Endocrinology 2018 89 367–375. (https://doi.org/10.1111/cen.13764)
- 11↑
Kus A, Chaker L, Teumer A, Peeters RP, Medici M. The genetic basis of thyroid function: novel findings and new approaches. Journal of Clinical Endocrinology and Metabolism 2020 105 1707–1721.(https://doi.org/10.1210/clinem/dgz225)
- 12↑
Wilson S, Parle JV, Roberts LM, Roalfe AK, Hobbs FD, Clark P, Sheppard MC, Gammage MD, Pattison HM & Franklyn JA et al.Prevalence of subclinical thyroid dysfunction and its relation to socioeconomic deprivation in the elderly: a community-based cross-sectional survey. Journal of Clinical Endocrinology and Metabolism 2006 91 4809–4816. (https://doi.org/10.1210/jc.2006-1557)
- 13↑
Völzke H, Craesmeyer C, Nauck M, Below H, Kramer A, John U, Baumeister S, Ittermann T. Association of socioeconomic status with iodine supply and thyroid disorders in northeast Germany. Thyroid 2013 23 346–353. (https://doi.org/10.1089/thy.2012.0416)
- 14↑
Huang Y, Cai L, Zheng Y, Pan J, Li L, Zong L, Lin W, Liang J, Huang H & Wen J et al.Association between lifestyle and thyroid dysfunction: a cross-sectional epidemiologic study in the She ethnic minority group of Fujian Province in China. BMC Endocrine Disorders 2019 19 83. (https://doi.org/10.1186/s12902-019-0414-z)
- 15↑
Medici M, Visser WE, Visser TJ, Peeters RP. Genetic determination of the hypothalamic-pituitary-thyroid axis: where do we stand? Endocrine Reviews 2015 36 214–244. (https://doi.org/10.1210/er.2014-1081)
- 16↑
Hansen PS, Brix TH, Sørensen TI, Kyvik KO, Hegedüs L. Major genetic influence on the regulation of the pituitary-thyroid axis: a study of healthy Danish twins. Journal of Clinical Endocrinology and Metabolism 2004 89 1181–1187. (https://doi.org/10.1210/jc.2003-031641)
- 17↑
Panicker V, Wilson SG, Spector TD, Brown SJ, Falchi M, Richards JB, Surdulescu GL, Lim EM, Fletcher SJ, Walsh JP. Heritability of serum TSH, free T4 and free T3 concentrations: a study of a large UK twin cohort. Clinical Endocrinology 2008 68 652–659. (https://doi.org/10.1111/j.1365-2265.2007.03079.x)
- 18↑
Porcu E, Medici M, Pistis G, Volpato CB, Wilson SG, Cappola AR, Bos SD, Deelen J, den Heijer M & Freathy RM et al.A meta-analysis of thyroid-related traits reveals novel loci and gender-specific differences in the regulation of thyroid function. PLoS Genetics 2013 9 e1003266. (https://doi.org/10.1371/journal.pgen.1003266)
- 19↑
Zhou W, Brumpton B, Kabil O, Gudmundsson J, Thorleifsson G, Weinstock J, Zawistowski M, Nielsen JB, Chaker L & Medici M et al.GWAS of thyroid stimulating hormone highlights pleiotropic effects and inverse association with thyroid cancer. Nature Communications 2020 11 3981. (https://doi.org/10.1038/s41467-020-17718-z)
- 20↑
Hofker MH, Fu J, Wijmenga C. The genome revolution and its role in understanding complex diseases. Biochimica et Biophysica Acta 2014 1842 1889–1895. (https://doi.org/10.1016/j.bbadis.2014.05.002)
- 21↑
Stang A, Moebus S, Dragano N, Beck EM, Möhlenkamp S, Schmermund A, Siegrist J, Erbel R, Jöckel KH & Heinz Nixdorf Recall Study Investigation Group. Baseline recruitment and analyses of nonresponse of the Heinz Nixdorf Recall Study: identifiability of phone numbers as the major determinant of response. European Journal of Epidemiology 2005 20 489–496. (https://doi.org/10.1007/s10654-005-5529-z)
- 22↑
Schmermund A, Möhlenkamp S, Stang A, Grönemeyer D, Seibel R, Hirche H, Mann K, Siffert W, Lauterbach K & Siegrist J et al.Assessment of clinically silent atherosclerotic disease and established and novel risk factors for predicting myocardial infarction and cardiac death in healthy middle-aged subjects: rationale and design of the Heinz Nixdorf RECALL Study. Risk factors, evaluation of coronary calcium and lifestyle. American Heart Journal 2002 144 212–218. (https://doi.org/10.1067/mhj.2002.123579)
- 23↑
UNESCO United Nations Educational, Scientific and Cultural Organization International Standard Classification of Education ISCED. International Standard Classification of Education ISCED. Montreal, Québec, Canada: UNESCO-UIS, 1997.
- 24↑
Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (Part 1). Journal of Epidemiology and Community Health 2006 60 7–12. (https://doi.org/10.1136/jech.2004.023531)
- 25↑
Geyer S, Hemström O, Peter R, Vågerö D. Education, income, and occupational class cannot be used interchangeably in social epidemiology. Empirical evidence against a common practice. Journal of Epidemiology and Community Health 2006 60 804–810. (https://doi.org/10.1136/jech.2005.041319)
- 26↑
Howie BN, Donnelly P, Marchini J. A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS Genetics 2009 5 e1000529. (https://doi.org/10.1371/journal.pgen.1000529)
- 27↑
Howie B, Marchini J, Stephens M. Genotype imputation with thousands of genomes. G3 2011 1 457–470. (https://doi.org/10.1534/g3.111.001198)
- 28↑
Choi SW, O'Reilly PF. PRSice-2: polygenic Risk Score software for biobank-scale data. GigaScience 2019 8 giz082 . (https://doi.org/10.1093/gigascience/giz082)
- 29↑
Chaker L, Korevaar TI, Medici M, Uitterlinden AG, Hofman A, Dehghan A, Franco OH, Peeters RP. Thyroid function characteristics and determinants: the Rotterdam study. Thyroid 2016 26 1195–1204. (https://doi.org/10.1089/thy.2016.0133)
- 30↑
Teumer A, Chaker L, Groeneweg S, Li Y, Di Munno C, Barbieri C, Schultheiss UT, Traglia M, Ahluwalia TS & Akiyama M et al.Genome-wide analyses identify a role for SLC17A4 and AADAT in thyroid hormone regulation. Nature Communications 2018 9 4455. (https://doi.org/10.1038/s41467-018-06356-1)
- 31↑
Lee YK, Shin DY, Shin H, Lee EJ. Sex-specific genetic influence on thyroid-stimulating hormone and free thyroxine levels, and interactions between measurements: KNHANES 2013–2015. PLoS One 2018 13 e0207446. (https://doi.org/10.1371/journal.pone.0207446)
- 32↑
Samollow PB, Perez G, Kammerer CM, Finegold D, Zwartjes PW, Havill LM, Comuzzie AG, Mahaney MC, Göring HH & Blangero J et al.Genetic and environmental influences on thyroid hormone variation in Mexican Americans. Journal of Clinical Endocrinology and Metabolism 2004 89 3276–3284. (https://doi.org/10.1210/jc.2003-031706)
- 33↑
Führer D, Brix K, Biebermann H. Understanding the healthy thyroid state in 2015. European Thyroid Journal 2015 4 1–8. (https://doi.org/10.1159/000431318)
- 34↑
Estaquio C, Valeix P, Leenhardt L, Modigliani E, Boutron-Ruault MC, Chérié-Challine L, Legrand M, Hercberg S, Castetbon K. Serum thyrotropin and free thyroxine reference ranges as defined in a disease-free sample of French middle-aged adults. Clinical Chemistry and Laboratory Medicine 2009 47 1497–1505. (https://doi.org/10.1515/CCLM.2009.334)
- 35↑
Qiu L, Wang DC, Xu T, Cheng XQ, Sun Q, Hu YY, Liu HC, Lu SY, Yang GH, Wang ZJ. Influence of gender, age and season on thyroid hormone reference interval. Zhonghua Yi Xue Za Zhi 2018 98 1582–1587. (https://doi.org/10.3760/cma.j.issn.0376-2491.2018.20.011)
- 36↑
Burkhardt K, Ittermann T, Heier M, Kirchberger I, Völzke H, Wallaschofski H, Below H, Nauck M, Meisinger C. TSH-reference range of adults: results from the population-based study KORA F4. Deutsche Medizinische Wochenschrift 2014 139 317–322. (https://doi.org/10.1055/s-0033-1360046)
- 37↑
Schalin-Jäntti C, Tanner P, Välimäki MJ, Hämäläinen E. Serum TSH reference interval in healthy Finnish adults using the Abbott Architect 2000i Analyzer. Scandinavian Journal of Clinical and Laboratory Investigation 2011 71 344–349. (https://doi.org/10.3109/00365513.2011.568630)
- 38↑
Surks MI, Hollowell JG. Age-specific distribution of serum thyrotropin and antithyroid antibodies in the US population: implications for the prevalence of subclinical hypothyroidism. Journal of Clinical Endocrinology and Metabolism 2007 92 4575–4582. (https://doi.org/10.1210/jc.2007-1499)
- 39↑
Vadiveloo T, Donnan PT, Murphy MJ, Leese GP. Age- and gender-specific TSH reference intervals in people with no obvious thyroid disease in Tayside, Scotland: the thyroid Epidemiology, Audit, and Research Study (TEARS). Journal of Clinical Endocrinology and Metabolism 2013 98 1147–1153. (https://doi.org/10.1210/jc.2012-3191)
- 40↑
Johnson W, Krueger RF. Genetic effects on physical health: lower at higher income levels. Behavior Genetics 2005 35 579–590. (https://doi.org/10.1007/s10519-005-3598-0)
- 41↑
Cockerham WC, Hamby BW, Oates GR. The social determinants of chronic disease. American Journal of Preventive Medicine 2017 52 S5–S12. (https://doi.org/10.1016/j.amepre.2016.09.010)
- 43↑
Dalstra JA, Kunst AE, Borrell C, Breeze E, Cambois E, Costa G, Geurts JJ, Lahelma E, Van Oyen H & Rasmussen NK et al.Socioeconomic differences in the prevalence of common chronic diseases: an overview of eight European countries. International Journal of Epidemiology 2005 34 316–326. (https://doi.org/10.1093/ije/dyh386)
- 44↑
Moore GF, Littlecott HJ. School- and family-level socioeconomic status and health behaviors: multilevel analysis of a national survey in wales, United Kingdom. Journal of School Health 2015 85 267–275. (https://doi.org/10.1111/josh.12242)
- 45↑
Stringhini S, Sabia S, Shipley M, Brunner E, Nabi H, Kivimaki M, Singh-Manoux A. Association of socioeconomic position with health behaviors and mortality. JAMA 2010 303 1159–1166. (https://doi.org/10.1001/jama.2010.297)
- 46↑
Alegría-Torres JA, Baccarelli A, Bollati V. Epigenetics and lifestyle. Epigenomics 2011 3 267–277. (https://doi.org/10.2217/epi.11.22)
- 47↑
McGuinness D, McGlynn LM, Johnson PC, MacIntyre A, Batty GD, Burns H, Cavanagh J, Deans KA, Ford I & McConnachie A et al.Socio-economic status is associated with epigenetic differences in the pSoBid cohort. International Journal of Epidemiology 2012 41 151–160. (https://doi.org/10.1093/ije/dyr215)
- 48↑
Tiret L Gene-environment interaction: a central concept in multifactorial diseases. Proceedings of the Nutrition Society 2002 61 457–463. (https://doi.org/10.1079/pns2002178)
- 49↑
Dinescu D, Horn EE, Duncan G, Turkheimer E. Socioeconomic modifiers of genetic and environmental influences on body mass index in adult twins. Health Psychology 2016 35 157–166. (https://doi.org/10.1037/hea0000255)
- 50↑
Frank M, Dragano N, Arendt M, Forstner AJ, Nöthen MM, Moebus S, Erbel R, Jöckel KH, Schmidt B. A genetic sum score of risk alleles associated with body mass index interacts with socioeconomic position in the Heinz Nixdorf Recall Study. PLoS One 2019 14 e0221252. (https://doi.org/10.1371/journal.pone.0221252)
- 51↑
Bronfenbrenner U, Ceci SJ. Nature-nurture reconceptualized in developmental perspective: a bioecological model. Psychological Review 1994 101 568–586. (https://doi.org/10.1037/0033-295x.101.4.568)
- 52↑
Middeldorp CM, Lamb DJ, Vink JM, Bartels M, van Beijsterveldt CE, Boomsma DI. Child care, socio-economic status and problem behavior: a study of gene-environment interaction in young Dutch twins. Behavior Genetics 2014 44 314–325. (https://doi.org/10.1007/s10519-014-9660-z)
- 53↑
Ge T, Chen CY, Neale BM, Sabuncu MR, Smoller JW. Phenome-wide heritability analysis of the UK Biobank. PLoS Genetics 2017 13 e1006711. (https://doi.org/10.1371/journal.pgen.1006711)
- 54↑
Schwabe I, Boomsma DI, van den Berg SM. Mathematical ability and socio-economic background: IRT modeling to estimate genotype by environment interaction. Twin Research and Human Genetics 2017 20 511–520. (https://doi.org/10.1017/thg.2017.59)
- 55↑
Rhemtulla M, Tucker-Drob EM. Gene-by-socioeconomic status interaction on school readiness. Behavior Genetics 2012 42 549–558. (https://doi.org/10.1007/s10519-012-9527-0)
- 56↑
Carlé A, Pedersen IB, Knudsen N, Perrild H, Ovesen L, Rasmussen LB, Jørgensen T, Laurberg P. Moderate alcohol consumption may protect against overt autoimmune hypothyroidism: a population-based case-control study. European Journal of Endocrinology 2012 167 483–490. (https://doi.org/10.1530/EJE-12-0356)
- 57↑
Grittner U, Kuntsche S, Gmel G, Bloomfield K. Alcohol consumption and social inequality at the individual and country levels-results from an international study. European Journal of Public Health 2013 23 332–339. (https://doi.org/10.1093/eurpub/cks044)
- 58↑
Pandey SC, Bohnsack JP. Alcohol makes its epigenetic marks. Cell Metabolism 2020 31 213–214. (https://doi.org/10.1016/j.cmet.2020.01.008)
- 59↑
Mews P, Egervari G, Nativio R, Sidoli S, Donahue G, Lombroso SI, Alexander DC, Riesche SL, Heller EA & Nestler EJ et al.Alcohol metabolism contributes to brain histone acetylation. Nature 2019 574 717–721. (https://doi.org/10.1038/s41586-019-1700-7)
- 60↑
Gidlow C, Johnston LH, Crone D, Ellis N, James D. A systematic review of the relationship between socio-economic position and physical activity. Health Education Journal 2006 65 338–367. (https://doi.org/10.1177/0017896906069378)
- 61↑
McLaren L Socioeconomic status and obesity. Epidemiologic Reviews 2007 29 29–48. (https://doi.org/10.1093/epirev/mxm001)
- 62↑
Brabant G, Prank K, Hoang-Vu C, Hesch RD, von zur Mühlen A. Hypothalamic regulation of pulsatile thyrotopin secretion. Journal of Clinical Endocrinology and Metabolism 1991 72 145–150. (https://doi.org/10.1210/jcem-72-1-145)
