Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
OPEN International Journal of Obesity (2013), 1–7 & 2013 Macmillan Publishers Limited All rights reserved 0307-0565/13 www.nature.com/ijo ORIGINAL ARTICLE Thyroid hormone receptor alpha gene variants increase the risk of developing obesity and show gene–diet interactions JM Fernández-Real1,2, D Corella2,3, L Goumidi4, JM Mercader5, S Valdés6,7,8, G Rojo Martı́nez6,7, F Ortega1,2, M-T Martinez-Larrad9, JM Gómez-Zumaquero6,7, J Salas-Salvadó2,10, MA Martinez González11, MI Covas2,12, P Botas6,7, E Delgado6,7, D Cottel4, J Ferrieres13, P Amouyel4, W Ricart1,2, E Ros2,14, A Meirhaeghe4, M Serrano-Rios9, F Soriguer4 and R Estruch2,15 OBJECTIVE: Thyroid hormone receptor-beta resistance has been associated with metabolic traits. THRA gene sequencing of an obese woman (index case) who presented as empirical thyroid hormone receptor-a (THRA) resistance, disclosed a polymorphism (rs12939700) in a critical region involved in TRa alternative processing. DESIGN AND SUBJECTS: THRA gene variants were evaluated in three independent europid populations (i) in two population cohorts at baseline (n ¼ 3417 and n ¼ 2265), 6 years later (n ¼ 2139) and (ii) in 4734 high cardiovascular risk subjects (HCVR, PREDIMED trial). RESULTS: The minor allele of the index case polymorphism (rs12939700), despite having a very low frequency (4%), was significantly associated with higher body mass index (BMI) (P ¼ 0.042) in HCVR subjects. A more frequent THRA polymorphism (rs1568400) was associated with higher BMI in subjects from the population (P ¼ 0.00008 and P ¼ 0.05) after adjusting for several confounders. Rs1568400 was also strongly associated with fasting triglycerides (P dominant ¼ 3.99 10 5). In the same sample, 6 years later, age and sex-adjusted risk of developing obesity was significantly increased in GG homozygotes (odds ratio 2.93 (95% confidence interval, 1.05–6.95)). In contrast, no association between rs1568400 and BMI was observed in HCVR subjects, in whom obesity was highly prevalent. This might be explained by the presence of an interaction (P o0.001) among the rs1568400 variant, BMI and saturated fat intake. Only when saturated fat intake was high (424.5 g d 1), GG carriers showed a significantly higher BMI than A carriers after controlling for energy intake and physical activity. CONCLUSIONS: THRA gene polymorphisms are associated with obesity development. This is a novel observation linking the THRA locus to metabolic phenotypes. International Journal of Obesity advance online publication, 12 February 2013; doi:10.1038/ijo.2013.11 Keywords: Thyroid hormone receptors; genetics; polymorphisms. INTRODUCTION The epidemic of obesity and associated comorbidities, such as diabetes, is one of the most important global public health problems.1,2 Obesity results from a chronic imbalance between energy acquisition (food intake), energy expenditure (basal metabolic rate, thermogenesis and physical activity), and the tendency to deposit any excess energy as either fat or lean mass (nutrient partitioning). Nearly 70% of interindividual differences in adiposity may be attributable to genetic factors.3–5 At least 20 single gene disorders that affect the central sensing and control of energy balance result in an autosomal form of human obesity.3,4,6 Several endocrine abnormalities predispose to obesity,7 including thyroid dysfunction.8 Triiodothyronine (T3) and tetraiodothyronine (T4) regulate energy metabolism and thermogenesis, and have critical roles in glucose and lipid metabolism, food intake and fatty acid oxidation.8,9 Even slight variations in thyroid function contribute to variations in body mass index (BMI) and weight gain with aging.10–12 Thyroid hormones are linked to lipogenesis in lipogenic tissues through activated thyroid hormone receptor-induced gene expression.13–15 The two major thyroid hormone receptor isoforms are differentially expressed during development and distributed in adult tissues.16 TRa has specific roles in the heart and in cardiac pacemaking.17,18 Mice lacking TRa have an average heart rate 20% lower than that of control animals.18. TRa also mediates adaptive thermogenesis in brown adipose tissue.19,20 TRb regulates thyroid stimulating hormone (TSH) and the metabolic actions of T3 in the liver.21,22 The two major thyroid hormone receptor isoforms, TRa1 and TRa2, are functionally antagonistic.19,22 TRa1 is an authentic thyroid hormone receptor 1 Department of Diabetes, Endocrinology and Nutrition. Institut d’Investigació Biomédica de Girona, Girona, Spain; 2CIBER Fisiopatologia de la Obesidad y la Nutrición, CIBEROBN (CB06/03), Instituto de Salud Carlos III, Girona, Spain; 3Genetic and Molecular Epidemiology Unit, Department of Preventive Medicine. University of Valencia, Valencia, Spain; 4 INSERM, U744; Institut Pasteur de Lille; Univ. Lille Nord de France; UDSL, Lille, France; 5Joint IRB-BSC program on Computational Biology. Barcelona Supercomputing Center, Barcelona, Catalonia, Spain; 6Servicio de Endocrinologı́a y Nutrición, Hospital Universitario Carlos Haya, Malaga, Spain; 7CIBER of Diabetes and Metabolism (CIBERDEM) (CB07/08), Instituto de Salud Carlos III, Malaga, Spain; 8Hospital Central de Asturias, Oviedo, Spain; 9Spanish Biomedical Research Centre in Diabetes and Associated Metabolic Disorders (CIBERDEM) and Instituto de Investigación Sanitaria del Hospital Clı́nico San Carlos (IdISSC), Madrid, Spain; 10Human Nutrition Department, School of Medicine, University Rovira i Virgili, Reus, Spain; 11Department of Preventive Medicine and Public Health, University of Navarra, Spain; 12Cardiovascular Risk and Nutrition Research Group, IMIM-Institut de Recerca del Hospital del Mar, Barcelona, Spain; 13INSERM, U558, Faculté de Médecine, Université Paul Sabatier, Toulouse, France; 14Lipid Unit, Service of Endocrinology, Hospital Clinic, Barcelona, Spain and 15Department of Internal Medicine, Hospital Clinic, IDIB APS, University of Barcelona, Barcelona, Spain. Correspondence: Dr JM Fernández-Real, Section of Diabetes, Endocrinology and Nutrition, Hospital of Girona ‘Dr Josep Trueta’, Carretera de Franc¸a s/n, Girona 17007, Spain. E-mail: [email protected] Received 28 September 2012; revised 29 November 2012; accepted 19 December 2012 TRa gene polymorphism and obesity JM Fernández-Real et al 2 that binds T3 and mediates its activity. TRa2 is a variant receptor that lacks a functional hormone-binding site and antagonizes the ability of TRa1 to activate gene transcription in the presence of T3.23 Thus, the regulation of TRa alternative processing is important for determining the cellular levels of TRa1 and TRa2 mRNAs, which, in turn, are critical for modulating the response to T3. TSH levels are increased in some obese subjects, suggesting thyroid hormone resistance. This is supported by the observation of decreased thyroid hormone receptors in circulating mononuclear cells of obese individuals24 and decreased negative feedback between TSH and peripheral T3 levels.25,26 The current study stems from a clinical observation of an obese woman who showed empirical thyroid hormone receptor-a (THRA) resistance. THRA gene sequencing showed a polymorphism in a critical region involved in TRa alternative processing. We then hypothesized that THRA polymorphisms could be associated with obesity. MATERIALS AND METHODS General population samples Cohort 1. Data and samples from 3417 apparently healthy subjects were obtained from population-based prospective studies performed in four regions of Spain between 1996 and 1999 (700 subjects from the North27– Asturias, 760 from North-East28–Girona, 906 from the Center29–Segovia and 1051 from the South30 of Spain–Malaga). Eligible participants (all Caucasians) were selected at random from the census and after a screening visit were invited to participate. The participation rate was higher than 70%. Baseline studies included a standardized questionnaire, physical examination and laboratory tests. Fasting serum and plasma were withdrawn and stored at 80 1C until analysis. Two thousand one hundred thirty-nine participants from the centers in the North, North-East and South regions of Spain were re-evaluated between 2003 and 2005 (63% of the original cohort). Cohort 2. Participants were recruited as part of the World Health Organization-MONICA population survey conducted from 1995 to 1997 in the Lille Urban Community in northern France (n ¼ 1155) and the HauteGaronne county in southern France (n ¼ 1170).31 The protocol was approved by the appropriate independent ethics committee in each center. High cardiovascular risk (PREDIMED) cohort From October 2003 to October 2007, subjects at high cardiovascular risk were selected by physicians in Primary Care Centers affiliated with 10 teaching hospitals in Spain to participate in the PREDIMED study.32 Eligible subjects were community-dwelling people who fulfilled at least one of two criteria: type 2 diabetes; or 3 or more cardiovascular risk factors (current smoking, hypertension, dyslipidemia, BMI X25 kg m 2 and family history of premature cardiovascular disease). Hypertension was defined as blood pressure X140/90 mm Hg or treatment with antihypertensive medication, and dylipidemia as high plasma low-density lipoprotein-cholesterol (X160 mg dl 1 or lipid-lowering therapy) or low plasma high-density lipoprotein-cholesterol (p40 mg dl 1 in men and p50 mg dl 1 in women). Exclusion criteria included history of cardiovascular disease, any severe chronic illness (terminal cancer, severe hepatic or renal diseases and/or dementia), and drug or alcohol addiction. In this substudy, we analyzed the first 4734 subjects who entered the trial and provided informed consent for genetic analyses. The study protocol was approved by the local ethics committees of each participating institution and all participants provided informed consent. Laboratory determinations Analytes determined for each participant were: serum glucose levels (glucose–oxidase method); cholesterol and triglyceride levels by enzymatic procedures; and high-density lipoprotein cholesterol level after precipitation with phosphotungstic acid and magnesium chloride, as previously reported.27,32 Serum TSH, free T4 and free T3 were measured in all centers by electrochemiluminescence (Roche Diagnostics, GmbH, Mannheim, Germany) with intra- and inter-assay coefficients of variation less than 5%. International Journal of Obesity (2013) 1 – 7 In the MONICA studies, all measurements were performed in a central laboratory (Purpan Hospital, Toulouse, France). Fasting triglyceride levels were measured using enzyme assays (Boehringer Mannheim, Mannheim, Germany). Genotyping of THRA gene polymorphisms THRA rs12939700 and rs1568400 polymorphisms were determined using a 7900HT Sequence Detection system (Applied Biosystems by Life Technologies, Norwalk, CT, USA) and predesigned fluorescent allelic discrimination TaqMan assay by standard procedures (Assay ID C_9511338_10). For quality control purposes, 10% of randomly selected samples were genotyped a second time, with no discrepancies. The genotyping success rates were between 96.3 and 98.5%. The populations were in Hardy–Weinberg equilibrium for the studied polymorphims in all populations studied. Statistical analyses t-test and ANOVA were used to compare crude differences of means between genotypes when indicated. The SNP association was computed using a general linear model implemented in SNPassoc software (Barcelona, Spain).33,34 Logistic regression analyses were carried out to estimated the odds ratio and 95% confidence intervals for obesity depending on THRA genotypes. The reference class was set as the homozygosity for the major allele among controls in all cases. P-values were derived from likelihood ratio tests. The best model was chosen using the Akaike Information Criteria.34 Multivariate adjustment of means was undertaken by analysis of covariance. The interaction between the THRA polymorphism and fat intake in determining BMI or other anthropometric variables was carried out by linear regression analysis. The P-value for the interaction term between fat intake (as dichotomic (cut-off the median value) or as a 3-category variables (tertiles)) and the THRA polymorphism was obtained in the hierarchical multivariate interaction model containing dummy variables for fat intake, the THRA polymorphism, their multiplicative term,and the additional control for the indicated covariates. Stratified analyses by genotype were also carried out, and the lineal trend between categories of fat intake was estimated by analysis of covariance. In the MONICA cohorts general linear models were used to compare mean values of variables by genotype groups. The statistical analyses were performed using SPSS software (version 15.0; Chicago, IL, USA). RESULTS Index case The index case was an obese (BMI 31.2 kg m 2) 48-year-old woman referred to the hospital outpatients’ clinics for evaluation of thyroid dysfunction. Hypothyroidism had been diagnosed 6 months earlier on the basis of serum thyroxine values of 0.9 ng dl 1 (normal 0.8–1.8) and TSH of 12.8 mU l 1 (normal 0.2–4.5), and treatment with l-thyroxine gradually increased up to 175 mg d 1. During this period she sustained an 18 kg weight loss while remaining entirely asymptomatic. The general examination, including the thyroid gland, was normal; her heart rate was 56 beats per min and an electrocardiogram disclosed no abnormalities. T4 was 2.2 ng dl 1 (normal 0.8–1.8) and TSHo0.001 mU l 1. An overdose of l-thyroxine was suspected and the dose was reduced to 150 mg d 1. Two months later, TSH was 1.2 mU l 1 and T4 1.5 ng dl 1 and the heart rate remained unchanged. Given the relatively low heart rate in the presence of hyperthyroidism, some form of THRA resistance was suspected. Sequencing of the entire THRA locus showed that the patient was heterozygous for a rs12939700 (C/A) polymorphism located in intron 9, considered as critical in the regulation of splicing.35 We also decided to study a polymorphism in the THRA gene with the highest heterozygosity (rs1568400, 635 A/G) in linkage disequilibrium with rs12939700. Polymorphism of the index case in the high-risk population (PREDIMED) This THRA polymorphism rs12939700 (A/C) polymorphism was analyzed in 4734 high-risk individuals, in whom a very low & 2013 Macmillan Publishers Limited TRa gene polymorphism and obesity JM Fernández-Real et al 3 frequency of the variant allele was shown. The prevalence of genotypes in high-risk subjects was 93% CC (n ¼ 4401 subjects), 6.8% CA (n ¼ 323) and 0.2% AA (n ¼ 10). Carriers of the A allele had a higher weight and BMI than CC homozygotes (P ¼ 0.042; Table 1). Moreover, carriers of the A allele had a higher prevalence of obesity (BMI X30 kg m 2) than CC subjects (53.3% vs 45.2%; P ¼ 0.030). Cross-sectional studies General population study. rs1568400 ( 635A/G) and obesity-related phenotypes. The frequencies of the different genotypes were 50.2% AA, 42.1% AG and 7.7% GG, and were similar in the four regions (that is, frequency of GG in Girona 8.0%, in Asturias 7.9%, in Segovia 7.3% and in Málaga 8.7%). The merged sample of general population was composed of 3417 individuals (54.3% AA, 51.4% GA and 47.6% GG subjects were women). Among them, G allele carriers (similar in age to AA homozygotes) were significantly more obese and showed an increased average waist circumference, total cholesterol and fasting triglyceride concentrations (Table 2 and Figure 1). The association with BMI remained significant after adjusting for age, gender, triglycerides and region (P ¼ 0.00008 in a dominant model, P ¼ 0.00034 in a codominant model, and P ¼ 0.00011 in a log-additive model). There was also a strong association between the rs1568400 SNP and fasting triglyceride levels (p dominant ¼ 3.99 10 5), such as G allele carriers showed the highest triglyceride levels. This association remained significant after controlling for age, BMI, sex, fasting glucose and region (Table 3A). This polymorphism was not significantly associated with fasting glucose, low-density lipoprotein cholesterol or high-density lipoprotein cholesterol. The associations with obesity and fasting triglycerides were independently replicated in the MONICA cohorts (Table 3B). Longitudinal study Relationship between rs1568400 SNP and development of obesity. Of the initial general population sample, 2139 participants were re-evaluated after a mean follow-up of 6.2±0.8 years. In this Table 1. Anthropometric and biochemical variables according to thyroid hormone receptor alpha gene polymorphism (rs12939700) in 4734 participants in the PREDIMED trial rs12939700 N Age (years) Weight (kg) BMI (kg m 2) Waist (cm) Total cholesterol (mg dl 1) LDL cholesterol (mg dl 1) HDL cholesterol (mg dl 1) Triglycerides (mg dl 1) Fasting glucose (mg dl 1) Energy intake (kcal d 1) Carbohydrates (g d 1) Proteins (g d 1) Total fat (g d 1) Saturated fat (g d 1) MUFA (g d 1) PUFA (g d 1) CC CA þ AA Mean (s.d.) Mean (s.d.) 4401 67.1 (6.3) 76.3 (11.7) 29.8 (4.1) 100.0 (10.6) 211.6 (38.5) 131.2 (35.6) 53.8 (13.2) 133.7 (74.2) 127.3 (46.2) 2325 (654) 242 (87) 94 (24) 102 (32) 26 (10) 51 (17) 16.3 (7.4) 333 66.6 (6.1) 77.6 (12.9) 30.3 (4.3) 100.6 (11.6) 214.3 (40.5) 133.8 (35.7) 53.1 (13.2) 136.2 (72.8) 131.6 (47.5) 2324 (716) 240 (89) 96 (29) 102 (35) 26 (11) 50 (18) 16.2 (7.1) P — 0.131 0.042 0.042 0.304 0.224 0.218 0.404 0.564 0.128 0.982 0.729 0.196 0.989 0.342 0.708 0.794 Abbreviations: HDL, high-density lipoprotein; LDL, low-density lipoprotein; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids. & 2013 Macmillan Publishers Limited analysis, subjects with abnormal TSH values (o0.2 and X4.0 mU l 1) were excluded from the calculations.(Table 4A) shows the main characteristics of this population. Age, gender and years of follow-up were similar in carriers and non-carriers of the G allele of rs1568400. Compared with the other genotypes, GG homozygotes had a significantly higher BMI (P ¼ 0.001) both at baseline and at follow-up evaluations. Similarly, average weight and waist circumference increases were also significantly higher among GG homozygotes (P ¼ 0.020 and 0.048, respectively). After adjusting for age, gender and thyroid function, the obesity incidence was significantly increased in GG homozygotes (Odds ratio 2.93 (95% confidence interval 1.05 to 6.95); P ¼ 0.01) compared with their nonhomozygous counterparts (Table 4B). Interaction of rs1568400 SNP with fat intake in the PREDIMED cohort. This polymorphism was determined in baseline samples of 4695 high-risk subjects from the PREDIMED trial. The prevalence of the genotypes was 50.9% AA, 40.2% AG and 8.9% GG. We found no significant associations between the rs1568400 SNP and obesity-related variables in this population (Table 5). Considering the previous association of rs1568400 SNP with BMI in the general population sample, the lack of association in these high-risk subjects prompted us to look for a potential gene–diet interaction for this polymorphism. When we examined the main macronutrients of the diet (that is, carbohydrates, proteins and lipids) estimated by food frequency questionnaires, we observed a statistically significant interaction with total fat intake in determining BMI (Po0.001). More in depth evaluation of this interaction showed saturated fat to have the highest effect on BMI. On testing the interaction between rs1568400 SNP and saturated fat intake expressed as a dichotomous variable (above and below the median intake in the whole population: 24.5 g d 1) in determining BMI, we found a highly statistically significant interaction term (Po0.001). Thus, when saturated fat was below the median (o24.5 g d 1), the rs1568400 SNP was not associated with BMI (29.74 þ / 0.09 Kg m 2 for Table 2. Anthropometric and biochemical variables of the subjects from the general population according to thyroid hormone receptor alpha -635A/G gene polymorphism (rs1568400) n (3417) Sex (men/women) Age (years) Body mass index (Kg m 2) Waist circumference (cm) Men Women Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Total cholesterol (mg dl 1) LDL cholesterol (mg dl 1) HDL cholesterol (mg dl 1) Fasting triglyceridesa Fasting glucose (mg dl 1) AA G carriers P 1719 791/928 50.6±12.4 28.28±5.2 1698 847/851 50.42±12.65 28.78±5.56 — 0.60 0.007 93.73±13.09 95.1±13.43 0.009 91.6±13.7 95.9±11.9 129.84±21.51 92.9±14.4 97.09±12.1 130.12±20.72 0.08 0.09 0.69 76.74±10.47 77.86±11.28 0.24 201.11±33.33 210.28±38.35 0.034 126.5±31.46 133.66±37.08 0.24 56.88±15.94 56.31±16.28 0.30 93 (66–134) 101.9±34.8 100 (71–147) 103.24±38.6 o0.0001 0.31 Abbreviations: HDL, high-density lipoprotein; LDL, low-density lipoprotein Values are mean±s.d. aexpressed as medians and interquartile range. International Journal of Obesity (2013) 1 – 7 TRa gene polymorphism and obesity JM Fernández-Real et al 4 2.03 95% CI Log fasting triglycerides (mg/dl) 95% CI Body mass index (kgr/m2) 29.0 28.8 p=0.007 28.6 28.4 28.2 2.02 p<0.0001 2.01 2.00 1.99 1.98 1.97 28.0 AA G allele carriers thyroid hormone receptor alpha gene polymorphism (rs1568400) AA G allele carriers thyroid hormone receptor alpha gene polymorphism (rs1568400) Figure 1. Effect of the rs1568400 polymorphism of the THRA gene on mean BMI and mean log fasting triglycerides in a general population sample. Bars are 95% confidence intervals. Table 3A. Association of the rs1568400 SNP with log plasma fasting triglyceride levels in the general population sample, controlling for body mass index, age, sex, glucose and region of origin Number Of subjects Log 10 trigl (mean) Se dif 95% CI P Codominant AA AG GG 1735 1442 266 1.98 2.02 2.03 0.006 0.007 0.014 0 0.027 0.034 0 0.01–0.04 0.01–0.06 0.00071 Dominant AA AG-GG 1735 1708 1.98 2.02 0.006 0.006 0 0.028 0 0.01–0.04 0.00016 rs1568400 AA þ AG vs 29.37 þ / 0.33 kg m 2 for GG; P ¼ 0.306). However, when saturated fat intake was above the median (424.5 g d 1), this polymorphism was associated with significantly higher BMI in GG subjects (29.38 þ / 0.09 Kg m 2 for AA þ AG vs 30.49 þ / 0.29 kg m 2 for GG; P ¼ 0.015), replicating the association found in the general population samples. This interaction remained statistically significant even after further controlling for age, sex, total energy intake and physical activity. Moreover, to examine a possible dose-response relationship, population tertiles of saturated fat intake were considered (o21.1 g d 1; 21.1 to 28.3 g d 1 and 428.3 g d 1). In this case, we again observed significant effects of the rs1568400 SNP for this interaction. In the regression model, a highly statistical significant interaction term (Po0.001) was obtained both in the crude model and after adjustment for age, gender, diabetes, total energy intake and physical activity. In the stratified analysis depending on the genotype, we observed a statistically significant P-value for lineal trend across tertiles of fat intake in subjects homozygous for the G allele. Figure 2 shows the interaction effect between saturated fat intake and the rs1568400 variant in determining BMI when saturated fat was considered as tertiles of intake. When saturated fat intake was low, the variant allele was not associated with a higher BMI. However, when saturated fat intake was high, the GG genotype was significantly associated with a higher BMI. Similar interaction results were obtained when weight or waist circumference were used as outcomes instead of BMI (data not shown). The results of this interaction were homogeneous for both men and women (data not shown). International Journal of Obesity (2013) 1 – 7 DISCUSSION Genome-wide association studies have allowed the identification of many gene variants related to obesity. Another way of identifying ‘candidate’ genes is through genetic analysis of peculiar clinical cases. The index case presented here was an obese woman overtreated with l-thyroxine. Being biochemically hyperthyroid, this woman lost 18 kg in 6 months, with no changes in heart rate characteristic of elevated thyroid hormones. Sequencing of the THRA locus disclosed a polymorphism in a critical region involved in the regulation of splicing. As this index case was obese, we analyzed two polymorphisms of the THRA locus that belonged to different blocks in association with BMI in three samples, two obtained from population-based studies and the other from subjects at high cardiovascular risk participating in the PREDIMED trial. In the latter, the polymorphism present in the index case (rs12939700) was associated with obesity. Another polymorphism (rs1568400) that has a relatively high frequency in the population was also associated with BMI according to both cross-sectional (in two independent general populations of Spain and France) and follow-up analyses in the Spanish populationbased cohort. This association with obesity was also confirmed in the publicly available genome-wide association study of the Genetic Investigation of ANthropometric Traits (GIANT) consortium (P ¼ 0.01, n ¼ 119 524 subjects).6 Interestingly, a SNP 45.6 Kb from rs1568400, rs1045929 (D’ ¼ 0.43, r2 ¼ 0.056) in the thyroid hormone receptor-associated protein (TRAP100/MED24) gene was also associated with obesity (P ¼ 0.004).6 The lack of association of rs12939700 and the weak association of rs1568400 in the GIANT consortium genome-wide association study data could be due to the different models of inheritance used to test the association in the present study, compared with those used by the GIANT consortium. While only the P-values for the association under the additive model is reported by GIANT consortium, our association was only significant when using the dominant model for both rs1568400 and rs12939700. This emphasizes the need of reporting the genome-wide association study summary data in other models of inheritance besides the commonly used additive model. In addition, large meta-analyses, such as those performed by the GIANT consortium rely on genotype imputation, and inaccuracy in this process, which particularly affects low frequency variants, such as rs12939700, could result in loss of power to detect a significant association. The THRA rs1568400 variant was not associated with obesity in the large cohort of subjects at increased cardiovascular risk from the PREDIMED cohort. This may be explained by the fact that almost all of these subjects had a BMI X25 kg m 2 as one of the & 2013 Macmillan Publishers Limited TRa gene polymorphism and obesity JM Fernández-Real et al 5 Table 3B. Z-scores of clinical variables according to THRA genotypes in combined MONICA Lille and Toulouse rs1568400 Zscore_BMI Zscore_WAIST Zscore_HIP Zscore_WHR Zscore_TG* AA (1230) AG (866) GG (169) P (dominant) P (recessive) 0.03±0.98 0.02±0.99 0.04±0.98 0.0002±1.00 0.004±1.05 0.04±1.03 0.04±1.01 0.05±1.02 0.01±0.98 0.03±1.00 0.06±1.00 0.03±1.01 0.09±1.01 0.04±1.04 0.12±0.66 0.05 0.15 0.02 0.68 0.57 0.57 0.77 0.29 0.41 0.02 P-values are adjusted for age, sex, alcohol consumption, smoking habit, physical activity level and center ( þ BMI for TG). *P values were calculated on logtransformed data. Significant values are depicted in bold. Table 4A. Subjects’ characteristics according to thyroid hormone receptor alpha -635A/G gene polymorphism (rs1568400) North–Asturias n (%) Gender (% women) Baseline age, year Baseline BMI, kg m 2 Age at follow-up, year Years of follow-up BMI at follow-up, kg m 2 Weight increase, kg South—Pizarra AA AG GG AA AG GG 358 (52.4) 53.9 52.1±12.7 27±3.9 58.3±12.7 6.26±0.2 27.63±4.2 1.67±5 271(39.7) 53.1 52.4±11.9 27.44±4.4 58.6±11.9 6.24±0.2 28.12±4.6 1.85±5.4 54(7.9) 59.38a 54.9±11.1a 29.45±6.1c 61.2±13.2a 6.24±0.2a 30.13±6.5c 1.88±4.9d 443 (47.7) 64.4 39.8±13.6 27.62±5.2 46.7±14.1 6.46±2.3 28.87±5.5 2.65±5.9 404(43.5) 61.2 40.6±13.9 27.22±4.9 47.2±14.1 6.44±1.6 28.33±4.9 2.18±6.2 81(8.7) 51.9a 36.42±14.41b 26.95±5d 43.9±13.7b 6.03±1.7a 29.70±5.2d 4.44±6.8e a P ¼ NS within each study. bP ¼ 0.01 for within study comparisons. cP ¼ 0.001 adjusted for age and sex. dP ¼ NS adjusted for age and sex and education level. P ¼ 0.02 adjusted for age and sex and education level. e Table 4B. Risk of developing obesity according to thyroid hormone receptor alpha -635A/G gene polymorphism (rs1568400) OR 95 CI % P Asturias AA AG GG reference 1.43 2.95a 0.76–2.68 1.05–7.73 0.26 0.03 Pizarra AA AG GG reference 1.14 2.93a 0.65–1.98 1.23–6.95 0.63 0.01 Pizarra AA AG GG reference 0.97 3.19b 0.53–1.75 1.031–7.75 0.92 0.01 Pizarra AA AG GG reference 1.25 2.99c 0.71–2.21 1.34–7.16 0.42 0.01 a Adjusted for age and sex and education level. bAdjusted for age and sex and education level and TSH at baseline. cAdjusted for age and sex and education level and TSH at follow-up. inclusion criteria and, therefore, there were few non-obese patients in the cohort. Nevertheless, in depth analysis of this selected population showed a gene–diet interaction between the THRA rs1568400 polymorphism and fat intake. Only individuals with the GG genotype and a high intake of saturated fat showed a significant association with increased BMI. The findings in high-risk & 2013 Macmillan Publishers Limited subjects were different to those observed in the general population, probably because some of the former were counseled by their family physicians to reduce total energy and saturated fat intake. The association of this polymorphism and obesity traits may only be observed in an obesogenic environment. Importantly, the THRA rs1568400 gene polymorphism was not only crosssectionally associated with obesity. Longitudinally, subjects with this gene variant had a threefold increased risk of developing obesity. Thyroid hormone stimulates the basal metabolic rate and adaptive thermogenesis. In fact, the relationship among thyroid function, energy metabolism and body fat is firmly established.19 Up to now, evidence concerning the possible role of THRA in obesity and metabolism has been exclusively obtained from animal models. The introduction of a dominant negative mutation derived from a resistance to thyroid hormone-associated THRB mutation into the THRA gene (P398H) resulted in increased adiposity and sensitivity to diet-induced obesity.36 However, introduction of other types of mutations led to a lean phenotype.37,38 In recent years it has been increasingly recognized that the hypothalamus-pituitary-thyroid axis regulates satiety signals through MC4R (melanocortin receptor-4) and NPY/AgRP neurons and UCP2.39–41 Unlike T3, the administration of the TRb-specific agonist GC-1 had no effect on food intake in rats, suggesting that the orexigenic effects of thyroid hormone may also be mediated by THRA.42 It has been speculated that hyperphagia in classical resistance to thyroid hormones (TRb-driven) is mediated by elevated thyroid hormones acting centrally via THRA.43 The THRA rs1568400 polymorphism was also associated with fasting triglycerides (Tables 3A and 3B). This association remained significant after controlling for the main confounders. In addition, there was an interaction between this polymorphism and fat intake. In fact, THRA is a gene expressed in the rat jejunum and International Journal of Obesity (2013) 1 – 7 TRa gene polymorphism and obesity JM Fernández-Real et al 6 Table 5. Anthropometric and biochemical variables according to thyroid hormone receptor alpha-635A/G gene polymorphism (rs1568400) in 4695 participants in the PREDIMED trial N Age (years) Weight (kg) BMI (kg m 2) Waist (cm) Total cholesterol (mg dl 1) LDL-C (mg dl 1) HDL-C (mg dl 1) Triglycerides (mg dl 1) Fasting glucose (mg dl 1) AA AG GG Mean (s.d.) Mean (s.d.) Mean (s.d.) 2387 67.2 (6.1) 76.5 (11.2) 29.8 (4.1) 100.3 (10.8) 212.4 (38.5) 131.7 (34.0) 54.0 (13.4) 133.9 (74.0) 126.4 (46.1) 1889 66.8 (6.3) 76.0 (11.6) 29.9 (4.4) 99.7 (10.5) 210.6 (38.5) 130.5 (34.0) 53.4 (12.8) 132.0 (71.1) 128.8 (46.3) 419 67.2 (6.7) 76.5 (12.1) 30.0 (4.1) 100.4 (10.8) 210.8 (40.1) 132.2 (50.2) 53.4 (13.2) 138.1 (95.5) 128.2 (44.0) P 0.149 0.502 0.826 0.210 0.355 0.494 0.350 0.333 0.296 Abbreviations: HDL, high-density lipoprotein; LDL, low-density lipoprotein P-value for comparison of means between genotypes (ANOVA test). 32 BMI (Kg/m2) P interaction THRA x saturated fat <0.001 P<0.05 30 28 AA AG+GG tertile 1 AA AG+GG tertile 2 AA AG+GG tertile 3 Saturated fat intake (tertiles) Figure 2. Interaction between the rs1568400 polymorphism of the THRA gene with saturated fat intake (tertiles) in determining BMI in the PREDIMED cohort. involved not only in carbohydrate and lipid absorption but also in the development of the intestine.44,45 Interestingly, patients with classical resistance to thyroid hormones (usually associated with mutations in thyroid hormone receptor b) also showed slightly elevated fasting triglycerides43 and the targeting of thyroid hormone receptor with receptor-beta agonists has previously shown to reduce both plasma cholesterol and triglyceride levels.46 In fact, dyslipidemia is a well-known characteristic of subjects with hypothyroidism,47 suggesting that G carriers show some degree of tissue hypothyroidism. In summary, both the study of crude associations between the selected SNPs at the THRA locus and the study of gene– environment (fat intake) interactions have yielded interesting results regarding their association with obesity and metabolic phenotypes. The mechanisms for these associations should be further studied given that TRa1 could contribute to adipose tissue expandability.48 CONFLICT OF INTEREST The authors declare no conflict of interest. ACKNOWLEDGEMENTS We greatly appreciate the technical assistance of Gerard Pardo and Oscar Rovira (Unit of Diabetes, Endocrinology and Nutrition. Institut d’Investigació Biomèdica de Girona, Hospital Universitari de Girona Dr Josep Trueta). JMM was supported by Sara Borrell International Journal of Obesity (2013) 1 – 7 fellowship from the Instituto de Salud Carlos III. This work was supported by research grants from the Ministerio de Educación y Ciencia (MEC) (SAF2008-02073 and PI070954), the Instituto de Salud Carlos III (CIBERobn) and Consejerı́a de Salud Junta de Andalucı́a (PI-0327-2010). JMM was supported by Sara Borrell Fellowship from the Instituto Carlos III. CIBERobn de Fisiopatologı́a de la Obesidad y Nutrición and CIBER de Diabetes y Enfermedades Metabólicas Asociadas are ISCIII projects. Funding was also obtained from the Commission’s Sixth Framework Program (CRESCENDO consortium, integrated project LSHM-CT-2005-018652), ANR GENOPAT (2008P006850) and INSERM. The French arm of the World Health Organization-MONICA population study was funded by grants from the Conseil Régional du Nord-Pas de Calais, the Caisse Primaire d’Assurance Maladie de Sélestat, the Association Régionale de Cardiologie d’Alsace, ONIVINS, Parke-Davis, the Mutuelle Générale de l’Education Nationale (MGEN), the Réseau National de Santé Publique, the Direction Générale de la Santé, the INSERM, the Institut Pasteur de Lille and the Unité d’Evaluation du Center Hospitalier et Universitaire de Lille, and to the THRA study group for their help in recruitment of subjects (Enrique Gómez-Gracia (University of Málaga, Málaga, Spain), Miguel Fiol (University Institute for Health Sciences Investigation, Palma de Mallorca, Spain), Fernando Arós (Hospital Txagorritxu, Vitoria, Spain), José Lapetra (San Pablo Health Center, Sevilla, Spain), Luis Serra-Majem (Las Palmas University, Las Palmas, Spain, Xavier Pintó (Hospital de Bellvitge, Hospitales de Llobregat, Spain), Carolina Ortega-Azorı́n and Marı́a Arregui (Genetic and Molecular Epidemiology Unit. Department of Preventive Medicine. University of Valencia, Valencia, Spain)). AUTHOR CONTRIBUTIONS RE researched data and contributed to discussion; DC, LG, JMM, SV, GRM, FO, JMG-Z, JS-S, MIC, ER, MS-R, FS researched data; M-TM-L, MAMG, PB, ED, DC, JF, PA, WR, AM: contributed to discussion and JMF-R: researched data and wrote the manuscript. REFERENCES 1 Hu FB. Obesity epidemiology. Oxford Univ Press: New York, 2008. 2 Berrington de Gonzalez A, Hartge P, Cerhan JR, Flint AJ, Hannan L, MacInnis RJ et al. Body-mass index and mortality among 1.46 million white adults. N Engl J Med 2010; 363: 2211–2219. 3 McCarthy M. Genomics, type 2 diabetes, and obesity. N Engl J Med 2010; 363: 2339–2350. 4 O’Rahilly S. Human genetics illuminates the paths to metabolic disease. Nature 2009; 462: 307–314. 5 Maes HH, Neale MC, Eaves LJ. Genetic and environmental factors in relative body weight and human adiposity. Behav Genet 1997; 27: 325–351. 6 Speliotes EK, Willer CJ, Berndt SI, Speliotes EK, Willer CJ, Berndt SI et al. Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat Genet 2010; 42: 937–948. 7 Kokkoris P, Pi-Sunyer FX. Obesity and endocrine disease. Endocrinol Metab Clin North Am 2003; 32: 895–914. 8 Reinehr T. Obesity and thyroid function. Mol Cell Endocrinol 2010; 316: 165–171. 9 Biondi B. Thyroid and obesity: an intriguing relationship. J Clin Endocrinol Metab 2010; 95: 3614–3617. & 2013 Macmillan Publishers Limited TRa gene polymorphism and obesity JM Fernández-Real et al 7 10 Knudsen N, Laurberg P, Rasmussen LB, Bülow I, Perrild H, Ovesen L et al. Small differences in thyroid function may be important for body mass index and the occurrence of obesity in the population. J Clin Endocrinol Metab 2005; 90: 4019–4024. 11 Fox CS, Pencina MJ, D’Agostino RB, Murabito JM, Seely EW, Pearce EN et al. Relations of thyroid function to body weight: cross-sectional and longitudinal observations in a community- based sample. Arch Intern Med 2008; 168: 587–592. 12 Manji N, Boelaert K, Sheppard MC, Holder RL, Gough SC, Franklyn JA. Lack of association between serum TSH or free T4 and body mass index in euthyroid subjects. Clin Endocrinol 2006; 64: 125–128. 13 Jiang W, Miyamoto T, Kakizawa T, Sakuma T, Nishio S, Takeda T et al. Expression of thyroid hormone receptor alpha in 3T3-L1 adipocytes; triiodothyronine increases the expression of lipogenic enzyme and triglyceride accumulation. J Endocrinol 2004; 182: 295–302. 14 Ying H, Araki O, Furuya F, Kato Y, Cheng SY. Impaired adipogenesis caused by a mutated thyroid hormone alpha1 receptor. Mol Cell Biol 2007; 27: 2359–2371. 15 Alevizaki M, Saltiki K, Voidonikola P, Mantzou E, Papamichael C, Stamatelopoulos K. Free thyroxine is an independent predictor of subcutaneous fat in euthyroid individuals. Eur J Endocrinol 2009; 161: 459–465. 16 Forrest D, Sjoberg M, Vennstrom B. Contrasting developmental and tissue-specific expression of alpha and beta thyroid hormone receptor genes. EMBO J 1990; 9: 1519–1528. 17 Kahaly GJ, Dillmann WH. Thyroid hormone action in the heart. Endocr Rev 2005; 26: 704–728. 18 Wikström L, Johansson C, Saltó C, Barlow C, Campos Barros A, Baas F et al. Abnormal heart rate and body temperature in mice lacking thyroid hormone receptor alpha 1. EMBO J 1998; 17: 455–461. 19 Cheng SY, Leonard JL, Davis PJ. Molecular aspects of thyroid hormone actions. Endocr Rev 2010; 31: 139–170. 20 Reyne Y, Nougues J, Cambon B, Viguerie-Bascands N, Casteilla L. Expression of c-erbA alpha, c-erbA beta and Rev-erbA alpha mRNA during the conversion of brown adipose tissue into white adipose tissue. Mol Cell Endocrinol 1996; 116: 59–65. 21 Weiss RE, Murata Y, Cua K, Hayashi Y, Seo H, Refetoff S. Thyroid hormone action on liver, heart, and energy expenditure in thyroid hormone receptor beta-deficient mice. Endocrinology 1998; 139: 4945–4952. 22 Lazar MA. Thyroid hormone receptors: multiple forms, multiple possibilities. Endocr Rev 1993; 14: 184–193. 23 Lazar MA, Chin WW. Nuclear thyroid hormone receptors. J Clin Invest 1990; 86: 1777–1782. 24 Burman KD, Latham KR, Djuh YY, Smallridge RC, Tseng YC, Lukes YG et al. Solubilized nuclear thyroid hormone receptors in circulating human mononuclear cells. J Clin Endocrinol Metab 1980; 51: 106–116. 25 Reinehr T, Andler W. Thyroid hormones before and after weight loss in obesity. Arch Dis Child 2002; 87: 320–323. 26 Reinehr T, de Sousa G, Andler W. Hyperthyrotropinemia in obese children is reversible afterweight loss and is not related to lipids. J Clin Endocrinol Metab 2006; 91: 3088–3091. 27 Valdés S, Botas P, Delgado E, Alvarez F, Cadórniga FD. Population-based incidence of type 2 diabetes in northern Spain: the Asturias Study. Diabetes Care 2007; 30: 2258–2263. 28 Moreno-Navarrete JM, Ortega FJ, Bassols J, Castro A, Ricart W, Fernández-Real JM. Association of circulating lactoferrin concentration and 2 nonsynonymous LTF gene polymorphisms with dyslipidemia in men depends on glucose-tolerance status. Clin Chem 2008; 54: 301–309. 29 Martinez-Larrad MT, Fernandez-Perez C, Gonzalez-Sanchez JL, Gonzalez-Sanchez JL, López A, Fernández-Alvarez J et al. Prevalence of the metabolic syndrome (ATP-III criteria). Population-based study of rural and urban areas in the Spanish province of Segovia. Med Clin (Barc) 2005; 125: 481–486. & 2013 Macmillan Publishers Limited 30 Soriguer F, Rojo-Martinez G, Almaraz MC, Esteva I, Ruiz de Adana MS, Morcillo S et al. Incidence of type 2 diabetes in southern Spain (Pizarra study). Eur J Clin Invest 2008; 38: 126–133. 31 The World Health Organization Monica Project. Ecological analysis of the association between mortality and major risk factors of cardiovascular disease. Int J Epidemiol 1994; 23: 505–516. 32 Estruch R, Martı́nez-González MA, Corella D, Salas-Salvadó J, Ruiz-Gutiérrez V, Covas MI et al. Effects of a Mediterranean-style diet on cardiovascular risk factors: a randomized trial. Ann Intern Med 2006; 145: 1–11. 33 Gonzalez JR, Armengol L, Sole X, Guinó E, Mercader JM, Estivill X et al. SNPassoc: an R package to perform whole genome association studies. Bioinformatics 2007; 23: 644–645. 34 Balding DJ. A tutorial on statistical methods for population association studies. Nat Rev Genet 2006; 7: 781–791. 35 Hastings ML, Wilson CM, Munroe SH. A purine-rich intronic element enhances alternative splicing of thyroid hormone receptor mRNA. RNA 2001; 7: 859–874. 36 Liu YY, Schultz JJ, Brent GA. A thyroid hormone receptor gene mutation (P398H) is associated with visceral adiposity and impaired catecholamine-stimulated lipolysis in mice. J Biol Chem 2003; 278: 38913–38920. 37 Sjogren M, Alkemade A, Mittag J, Nordström K, Katz A, Rozell B et al. Hypermetabolism in mice caused by the central action of an unliganded thyroid hormone receptor a1. EMBO J 2007; 26: 4535–4545. 38 Pelletier P, Gauthier K, Sideleva O, Samarut J, Silva JE. Mice lacking the thyroid hormone receptor-alpha gene spend more energy in thermogenesis, burn more fat, and are less sensitive to high-fat diet-induced obesity. Endocrinology 2008; 149: 6471–6486. 39 Dhillon H, Zigman JM, Ye C, Lee CE, McGovern RA, Tang V et al. Leptin directly activatesSF-1 neurons in the VMH and this action by leptin is required for normal body weight homeostasis. Neuron 2006; 49: 191–203. 40 Decherf S, Seugnet I, Kouidhi S, Lopez-Juarez A, Clerget-Froidevaux MS, Demeneix BA. Thyroid hormone exerts negative feedback on hypothalamic type 4 melanocortin receptor expression. Proc Natl Acad Sci USA 2010; 107: 4471–4476. 41 Coppola A, Liu ZW, Andrews ZB, Paradis E, Roy MC, Friedman JM et al. A central thermogenic-like mechanism in feeding regulation: an interplay between arcuate nucleus T3 and UCP2. Cell Metab 2007; 5: 21–33. 42 Villicev CM, Freitas FR, Aoki MS, Taffarel C, Scanlan TS, Moriscot AS et al. Thyroid hormone receptor beta-specific agonist GC-1 increases energy expenditure and prevents fat mass accumulation in rats. J Endocrinol 2007; 193: 21–29. 43 Mitchell CS, Savage DB, Dufour S, Schoenmakers N, Murgatroyd P, Befroy D et al. Resistance to thyroid hormone is associated with raised energy expenditure, muscle mitochondrial uncoupling, and hyperphagia. J Clin Invest 2010; 120: 1345–1354. 44 Suzuki T, Mochizuki K, Goda T. Localized expression of genes related to carbohydrate and lipid absorption along the crypt-villus axis of rat jejunum. Biochim Biophys Acta 2009; 1790: 1624–1635. 45 Mochizuki K, Yagi E, Sakaguchi N, Mochizuki H, Takabe S, Kuranuki S et al. The critical period for thyroid hormone responsiveness through thyroid hormone receptor isoform alpha in the postnatal small intestine. Biochim Biophys Acta 2007; 1770: 609–616. 46 Ladenson PW, Kristensen JD, Ridgway EC, Olsson AG, Carlsson B, Klein I et al. Use of the thyroid hormone analogue eprotirome in statin-treated dyslipidemia. N Engl J Med 2010; 362: 906–916. 47 Pearce EN. Hypothyroidism and dyslipidemia: modern concepts and approaches. Curr Cardiol Rep 2004; 6: 451–456. 48 Ortega FJ, Moreno-Navarrete JM, Ribas V, Esteve E, Rodriguez-Hermosa JI, Ruiz B et al. Subcutaneous fat shows higher thyroid hormone receptor-alpha1 gene expression than omental fat. Obesity (Silver Spring) 2009; 17: 2134–2141. This work is licensed under a Creative Commons AttributionNonCommercial-NoDerivs 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/3.0/ International Journal of Obesity (2013) 1 – 7