Published online before print June 23, 2003, doi: 10.1161/01.HYP.0000081965.51761.AA
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Hypertension. 2003;42:e3.)
© 2003 American Heart Association, Inc.
Letters to the Editor |
Glucocorticoid Receptor Gene and Coronary Artery Disease: Right Idea, Wrong Gene Variant?
Department of Clinical Chemistry, Sahlgrenska University Hospital, Göteborg, Sweden
To the Editor:
Recent data reported by Lin et al1 suggest that coronary artery disease (CAD) is associated with an Asn363Ser variant in exon 2 of the glucocorticoid receptor gene (GRL). The study is based on the notion that a dysfunctional glucocorticoid receptor may add to the adverse health effects of excessive cortisol concentrations.
In a previous study, comprising Anglo-Celtic descent with a strong hereditability for hypertension and formerly designed to test the relationship between GRL and essential hypertension, the authors found an association between the Ser363 variant and elevated body mass index (BMI).2 However, no association or linkage with hypertension was detected.3 This lack of effect on blood pressure has been reported previously in a Dutch cohort.4
The authors fail to provide valuable pieces of information regarding previously published observations. The fact is that the only studies indicating an association between the Ser363 variant and obesity, measured as BMI, is that of Lin et al2 and Huizenga et al.4 In the referenced study of the Newcastle Heart Project,5 a significant association of the Ser allele was detected with increased waist-to-hip ratio (WHR) and not BMI in white men. To see an association with WHR in this report the authors needed to correct for a number of other factors that associate with central obesity, hence the multivariate approach (Christopher P.F. Redfern, personal communication, 2001). More importantly, however, is that a majority of studies indicates that the ATT to GTT point variation in exon 2 of the GRL is neither associated with obesity nor with obesity-related metabolic and circulatory perturbations.6
The Asn363Ser variant is located in the transactivation (
1) domain, and phosphorylation of serine residues is important in DNA binding by glucocorticoid receptors (GRs). However, a series of studies clearly indicates that this variant lacks any effect on the function of the receptor in vitro.4,7–9 Still, Lin et al1 suggest that the underlying mechanism for the observed association with CAD might be that "adipocytes having the S363 variant of GR could be more sensitive to glucocorticoids...."
During the past decade, mutations affecting liability to obesity have been discovered at a phenomenal rate.10 The current literature linking obesity to genetic variants is teeming with reports of associations that cannot be replicated. Explanations for this lack of reproducibility are well rehearsed and typically include poor study design, incorrect assumptions about the underlying genetic architecture, and simple overinterpretation of data.
In this regard, the authors of this paper apply a case-control strategy of association study for characterizing the genetic contribution to CAD and obesity. This approach, however, is inherently more susceptible to identifying gene variants that prove to be spuriously associated with diseases.11 Furthermore, findings beyond the study design and study objective are most likely to be misleading and spurious in nature.2 Although publication is a crucial portion of the scientific process, investigators and editors should be encouraged to avoid publishing non–hypothesis-driven results, especially in excellent journals,2 to prevent their unfortunate persistence into the present time.
References
1. Lin RCY, Wang XL, Morris BJ. Association of coronary artery disease with glucocorticoid receptor n363s variant. Hypertension. 2003; 41: 404–407.
2. Lin RC, Wang WY, Morris BJ. High penetrance, overweight, and glucocorticoid receptor variant: case-control study. BMJ. 1999; 319: 1337–1338.
3. Lin RC, Wang WY, Morris BJ. Association and linkage analyses of glucocorticoid receptor gene markers in essential hypertension. Hypertension. 1999; 34: 1186–1192.
4. Huizenga NA, Koper JW, De Lange P, Pols HA, Stolk RP, Burger H, Grobbee DE, Brinkmann AO, De Jong FH, Lamberts SW. A polymorphism in the glucocorticoid receptor gene may be associated with an increased sensitivity to glucocorticoids in vivo. J Clin Endocrinol Metab. 1998; 83: 144–151.
5. Dobson MG, Redfern CP, Unwin N, Weaver JU. The n363s polymorphism of the glucocorticoid receptor: potential contribution to central obesity in men and lack of association with other risk factors for coronary heart disease and diabetes mellitus. J Clin Endocrinol Metab. 2001; 86: 2270–2274.
6. Rosmond R. The glucocorticoid receptor gene and its association to metabolic syndrome. Obes Res. 2002; 10: 1078–1086.[Medline] [Order article via Infotrieve]
7. Karl M, Lamberts SW, Detera-Wadleigh SD, Encio IJ, Stratakis CA, Hurley DM, Accili D, Chrousos GP. Familial glucocorticoid resistance caused by a splice site deletion in the human glucocorticoid receptor gene. J Clin Endocrinol Metab. 1993; 76: 683–689.[Abstract]
8. Koper JW, Stolk RP, de Lange P, Huizenga NA, Molijn GJ, Pols HA, Grobbee DE, Karl M, de Jong FH, Brinkmann AO, Lamberts SW. Lack of association between five polymorphisms in the human glucocorticoid receptor gene and glucocorticoid resistance. Hum Genet. 1997; 99: 663–668.[CrossRef][Medline] [Order article via Infotrieve]
9. de Lange P, Koper JW, Huizenga NA, Brinkmann AO, de Jong FH, Karl M, Chrousos GP, Lamberts SW. Differential hormone-dependent transcriptional activation and -repression by naturally occurring human glucocorticoid receptor variants. Mol Endocrinol. 1997; 11: 1156–1164.
10. Chagnon YC, Rankinen T, Snyder EE, Weisnagel SJ, Perusse L, Bouchard C. The human obesity gene map: the 2002 update. Obes Res. 2003; 11: 313–367.[Medline] [Order article via Infotrieve]
11. Daly AK, Day CP. Candidate gene case-control association studies: advantages and potential pitfalls. Br J Clin Pharmacol. 2001; 52: 489–499.[CrossRef][Medline] [Order article via Infotrieve]
Response
Basic & Clinical Genomics Laboratory, School of Medical Sciences, and Institute for Biomedical Research, The University of Sydney, Sydney, Australia
Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, Texas, Cardiovascular Genetics Laboratory, Prince of Wales Hospital, Centre for Thrombosis and Vascular Research, University of New South Wales, Sydney, Australia
We are perplexed by Dr Rosmond’s letter, which simply restates what we already said in our paper,1 and then dismisses most genetic association studies.
Our Introduction cites all positive and negative studies of the glucocorticoid receptor (GR) N363S variant with obesity, body mass index (BMI), and/or waist-to-hip ratio (WHR). Moreover, a morbidly obese group in Sydney further supports N363S in obesity.2 Neither WHR nor BMI accurately reflects the metabolic disturbance associated with obesity, but they do classify obesity status. Regardless, a person with BMI >40 kg/m2 (as in our recent study2) is obese, no matter what the marker. Although WHR indicates central obesity, it is affected by gender and age. Indeed, both BMI and WHR will grossly associate with, but neither will finely mirror, the metabolic changes.
There are also data on a BclI restriction fragment length polymorphism (RFLP) in obesity. Interstudy differences are commonplace, so what Rosmond says is hardly news. In the case of GR and hypertension, most data are negative. Despite finding occasional weak, gender-specific associations for some polymorphisms, our overall conclusion favored no association.3 For N363S, this applied to our original study,3 as well as to hypertension in coronary artery disease (CAD),1 obesity, and diabetes.2 Although others have studied indices of CAD, ours is the first to examine CAD itself. This involved 556 patients, with a highly significant finding for N363S (case versus control=0.15 versus 0.04; P=0.00005), rising to 0.45 for carriers with unstable angina.1
We nowhere claim that N363S is functional. Nor do we say that position 363 is a proven phosphorylation site. We cite some of the in vitro studies and Rosmond adds others. However, they are limited by choice of cell and cannot mimic in vivo physiology. We stated which amino acids have shown phosphorylation and that N363S is merely a potential phosphorylation site.1 Our statements apply to any variant in linkage disequilibrium (LD) with N363S. Thus "adipocytes having the S363 variant ..." is correct, because it does not say that S363 itself is the reason that these cells might be "more sensitive to glucocorticoids".
Rosmond’s inference that our study is misleading and spurious is grossly biased and could be leveled at almost any association study. There is much debate about optimum design and no consensus has yet been reached.4,5 In the meantime, journals have the right to publish papers judged as scientifically sound in the context of prevailing views. Association studies have the potential to be spurious for a variety of reasons, such as population stratification.6 That is why we also examined N363S in morbid obesity2 and obesity in CAD,1 in diabetes,2 and in hypertensive2,7 and healthy2,7 populations. Moreover, in the largest, most fully characterized group (CAD), S363 was associated with lipid elevation, thus providing a basis for us to speculate on mechanism.1
Rosmond’s reasons for lack of study reproducibility are incomplete. He fails to acknowledge other possibilities. For example, a positive association could be real, but specific to a certain population,4 because of different environmental influences on genotype. The negative genomic finding in a Melbourne study involved linkage, well-known to be less powerful.
References
1. Lin RC, Wang XL, Morris BJ. Association of coronary artery disease with glucocorticoid receptor N363S variant. Hypertension. 2003; 41: 404–407.
2. Lin RCY, Wang XL, Dalziel B, Caterson ID, Morris BJ. Association of obesity, not diabetes or hypertension, with glucocorticoid receptor N363S variant. Obesity Res. 2003; 11: 802–808.[Medline] [Order article via Infotrieve]
3. Lin RCY, Wang WY, Morris BJ. Association and linkage analyses of glucocorticoid receptor gene markers in essential hypertension. Hypertension. 1999; 34: 1186–1192.
4. Lomueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN. Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease. Nat Genet. 2003; 33: 177–182.[CrossRef][Medline] [Order article via Infotrieve]
5. Glazier AM, Nadeau JH, Aitman TJ. Finding the genes that underlie complex traits. Science. 2002; 298: 2345–2349.
6. Cardon LR, Palmer LJ. Population stratification and spurious allelic association. Lancet. 2003; 361: 598–604.[CrossRef][Medline] [Order article via Infotrieve]
7. Lin RCY, Wang WYS, Morris BJ. High penetrance, overweight, and glucocorticoid receptor variant: case-control study. BMJ. 1999; 319: 1337–1338.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||