Sex steroids influence the risk profile for cardiovascular diseases. This is exemplified by different cardiovascular risk profiles in men and women with changes occurring after menopause 1 and hormone replacement or suppression therapy 2–5. The principal circulating androgen testosterone can potently affect cardiovascular physiology. Testosterone acts through both nongenomic and genomic mechanisms, of which the latter is mediated by the androgen receptor (AR). Besides the fact that circulating testosterone levels are higher in men than in women, the effects of androgens differ between the sexes.
In the myocardium, testosterone might exert complex sexual dimorphic effects on conductivity and remodeling after ischemia or increased load 6. Direct testosterone administration significantly decreased QTc time in men, but not in women 7. Sexual dimorphism of testosterone effects is even more pronounced in the endothelium of the coronary arteries and thus influences the risk of myocardial infarction (MI). Whereas testosterone seems to have an antiatherogenic effect in men 8–10, in women, it has a stimulating effect on atherosclerotic plaque formation 5,8,11. Sex-specific differences in steroid hormone metabolism or receptor activation might mediate sex-specific responses, but the precise mechanisms are still not completely understood.
In nongonadal tissues, testosterone can be formed from circulating androstenedione through local intracrine conversion by the type 5 17β-hydroxysteroid dehydrogenase (HSD), encoded by AKR1C3 (=HSD17B5), and possibly also by 17β-HSD type 3 (HSD17B3), the principal testosterone-forming enzyme in the testis 12. 5α-Dihydrotestosterone (DHT) is a more active androgen and is produced by conversion of testosterone by the enzyme 5α-reductase. DHT binds to the AR with a two to three times higher affinity than testosterone and has a slower dissociation rate 13. There are two types of 5α-reductase: type I, encoded by the SRD5A1 gene, and type II, encoded by SRD5A2. Expression of the type II enzyme is mainly confined to the male reproductive tract, whereas SRD5A1 expression is widely distributed throughout the body 14,15. Both genes are involved in testosterone metabolism and single nucleotide polymorphisms (SNPs) in both genes have been associated with polycystic ovary syndrome (PCOS) and hirsutism-related traits 16. These findings originate from a candidate gene study of modest size, lacking replication.
We hypothesized that sex-dependent differences in the effects of testosterone may be caused by sex-dependent tissue expression of or SNPs in genes encoding androgen-metabolizing enzymes. Because of the sexual dimorphic effect of testosterone in both the risk of MI as well as cardiac remodeling, we postulate that the presence of these enzymes in human endothelial as well as myocardial tissues can modulate the local conversion of androstenedione and testosterone into DHT. Variation in either steroidogenic enzyme expression or activity can result in differences in cellular response leading to sex-specific modulation of the risk of MI or cardiac remodeling.
In this study, we first investigated the potential of intracrine conversion of sex steroids in physiological and pathological states by measuring the presence of steroidogenic enzymes and AR mRNA in human hearts. Second, we studied whether variations in genes encoding the androgen-metabolizing enzymes are associated with a change in the risk of MI. Together, these studies aimed at providing evidence that sex-specific expression of, or presence of, SNPs in androgen-metabolizing enzymes might contribute toward clinical sexual dimorphism in the risk of MI and heart failure.
Materials and methods
Measurement of steroidogenic enzymes
Human heart samples
To study mRNA expression of androgen-metabolizing enzymes and the AR, human heart tissues were obtained as described previously 17. Ventricular samples were taken from seven heart-beating organ donors (median age 37, range 7–47 years, four women) from eight patients with hypertrophic cardiomyopathy (median age 43, range 22–59 years, five women) and from five patients with end-stage dilated heart failure (median age 38, range 30–52, one woman). Left and right atrial specimens were taken from 11 different heart-beating organ donors (median age 48, range 5–63, six women). Control samples from organ donors were provided by the Rotterdam Heart Valve bank after the removal of heart valves; all donors died because of cerebral hypoxia or hemorrhage and had not taken any cardiovascular medication. Within 24 h of death or operation, myocardial pieces were dissected and frozen and stored at −70°C. These studies were approved by the Medical Ethics Committee of the Erasmus Medical Center.
Quantitative RT-PCR of androgen-metabolizing enzymes and AR
Heart samples were homogenized and RNA was isolated and reverse transcribed as described previously 17. Primer and probe sequences have been specified previously 18. The HSD17B3 assay (Hs00970002_m1) was purchased from Applied Biosystems (Nieuwerkerk aan den IJssel, the Netherlands). Quantitative RT-PCR for expression of the genes AKR1C3, SRD5A1, SRD5A2, AR, and hypoxanthine ribosyl transferase 1 (HPRT1) was performed as reported previously 17. All assays showed more than 90% efficiency and did not detect genomic DNA. The quantity of mRNA of steroidogenic enzymes and the AR was expressed relative to that of the reference gene HPRT1 using the ΔCt method 19.
Data for genetic analysis were obtained from the participants of the Rotterdam Study (RS), an ongoing prospective population-based cohort study of chronic diseases in elderly Caucasians. All inhabitants of Ommoord, a suburb of the city of Rotterdam in the Netherlands, aged 55 years or older were invited in 1990 to participate in the study. The medical ethics committee approved the study and informed consent was obtained from all participants. The rationale and design of the study have been described elsewhere 20. The first cohort included 7983 individuals who were all interviewed and examined at baseline in the period 1990–1993 (RSI). In 2000, all inhabitants of Ommoord aged 55 years and older at that time and not yet participating in RSI were invited to participate in the extended cohort (RSII). This cohort included 3011 individuals, who entered the study with consent. Follow-up examinations are carried out periodically, while all participants are monitored continuously for major morbidity and mortality through linkage with the general practitioner and municipality records.
All participants with successful genotyping who did not have a history of MI at baseline were included. Prevalent MI was based on medical records from general practitioners and cardiologists, ECG data, laboratory measurements, and data on hospital admissions. The cohort was followed until the first occurrence of a MI, death, or the end of the study period (1 January 2007), whichever came first.
The study outcome MI was defined on the basis of a hospital discharge diagnosis, or in case a patient was not hospitalized, when signs and symptoms, electrocardiography recordings, and cardiac enzyme data were diagnostic of MI 21. Two research physicians independently validated all potential cardiac events according to the International Classification of Diseases, 10th version, as reported previously 22. In cases of disagreement, a cardiologist reviewed the events and decided on the definite validation.
Genomic DNA was extracted from whole-blood samples using the salting-out method and microarray genotyping was performed using the Infinium II Human Hap 550K Genotyping BeadChip version 3 (Illumina Inc., San Diego, California, USA) as described previously 23.
SNPs were selected from the gene-coding regions of SRD5A1, SRD5A2, and AKR1C3. The first two genes are located on chromosomes 5p15 and 2p23, respectively, position 6685–6722 kbp and 31 603–31 659 kbp, respectively. The AKR1C3 gene is located on chromosome 10p15, position 5127–5140 kbp. We selected all SNPs within the gene-coding regions, plus 10 kbp at each end of the gene. A minimum of 90% coverage of alleles within these areas was aimed for, with a threshold of an r2 of 0.75. Using this selection, 95 and 92% allele coverage was reached for SRD5A1 and SRD5A2, respectively. For AKR1C3, three extra SNPs were selected between 10 and 50 kbp at both sides of the gene to reach 92% coverage. The following cut-off points were used: Hardy–Weinberg P-value cut-off 0.05 divided by the number of SNPs drawn from the database to correct for multiple testing, minimal genotype percentage 95%, and minimal minor allele frequency 0.05. SNPs that did not fulfill the above-mentioned criteria were excluded from the analysis.
Information on several potential confounders and effect modifiers, such as age, sex, BMI (kg/m2), serum cholesterol, blood pressure, diabetes mellitus, and smoking, (classified as never/former/current) was gathered at baseline and nonfasting blood samples were drawn 20. Diabetes mellitus was defined as the use of blood-glucose lowering medication and/or a nonfasting serum glucose level of 11.1 mmol/l or higher.
Serum levels of testosterone and sex hormone-binding globulin (SHBG) were estimated in a random sample of 1035 participants using coated-tube or a double-antibody radioimmunoassay, respectively (Diagnostic Systems Laboratories, Webster, Texas, USA) 24. As a measure of bioavailable testosterone, non-SHBG-bound testosterone was calculated on the basis of hormone, SHBG, and albumin levels, and respective affinity constants according to the method described previously 25,26.
mRNA expression levels in the human heart samples were logarithmically converted and analyzed using an unpaired t-test or one-way analysis of variance, followed by post-hoc evaluation according to Tukey. Correlations between gene expressions were studied using Pearson’s correlation coefficient, followed by Bonferroni correction for multiple testing.
The association between SRD5A1, SRD5A2, and AKR1C3 genotypes and MI was evaluated using survival analysis techniques, stratified by sex. Subsequently, we tested whether significant effect modification was present between sex and genotype using interaction terms. Per SNP, the major allele was used as the reference. We studied the genotype effect and the allele effect of the SNPs on incident MI. Cox proportional hazard models were used, adjusted for the potential confounding effect of age, total cholesterol, HDL-cholesterol, BMI, diabetes mellitus, systolic blood pressure, diastolic blood pressure, and smoking habits, and, additionally, for bioavailable testosterone. Covariates that were univariately associated with MI (at a P<0.1) were included in the regression analyses to adjust for their potential confounding effect if they changed the point estimate of the association between genotype and MI by more than 10%. Bonferroni correction was applied to adjust for multiple testing. Haplotype analyses were carried out to assess the influence of haplotype on the association between genotype and MI using Haploview (Broad Institute, Cambridge, MA, USA). Haploblocks were composed manually, on the basis of linkage disequilibrium, and χ2 tests were used to measure associations. All statistical analyses were carried out using SPSS software (version 15.0; SPSS Inc., Chicago, Illinois, USA), with the exception of the replication analyses, which were carried out using R software (version 2.7.1; The R Foundation for Statistical Computing, Statistics Department, University of Auckland, New Zealand).
Significant results from the genetic analysis were replicated within two cohorts: first, within the Atherosclerosis Risk in Communities (ARIC) Study, a prospective study, set up to investigate the etiology of atherosclerosis and variation in cardiovascular disease in four US communities, in adults aged 45–64 years 27. Second, replication was performed within the second cohort of the RSII 20 according to the methods described above.
Measurement of mRNA expression of steroidogenic enzymes and AR in hearts
The mRNA expression levels of HPRT1, HSD17B3, AKR1C3, SRD5A1, SRD5A2, and AR were studied in a series of normal left and right atria and in different types of left ventricular tissues, that is, from normal hearts, and hearts from patients with hypertrophic cardiomyopathy and end-stage dilated heart failure. SRD5A2 expression was negative in all except one sample, derived from a right atrium (threshold cycle of 35.7). All other genes tested positive for all samples. Individual gene expression levels were not significantly different between groups of normal left ventricles, and left and right atria (data not shown). The ventricular samples of patients with dilated heart failure showed a trend toward lower AKR1C3 expression (P=0.07) compared with the controls (Fig. 1). Overall, the expression of AKR1C3 was ∼100-fold higher than that of HSD17B3. The expression of SRD5A1, HSD17B3, and AR did not differ between the groups of left ventricular tissues (Fig. 1). In the left ventricle group, however, tissues from female patients showed higher SRD5A1 expression than those from male patients (P<0.05). The mRNA levels of AKR1C3 and SRD5A1 (r=0.47, P=0.04) and of HSD17B3 and AR (r=0.60, P=0.005) were positively correlated.
Of the 5974 participants for whom results of genotyping were available, 701 were excluded because of a history of MI. One-hundred and eight participants were excluded because of missing data on MI. After exclusion, 5165 participants remained for analysis.
The population included 1974 men (38%) and 3191 women (62%). The mean age at baseline was 67.7 years of age (range 55.0–97.8 years) in men and 69.8 years of age (range 55.0–99.2) in women. Table 1 shows the baseline characteristics of the study population. During follow-up, 358 participants developed an MI: 209 men (10.6%) and 149 women (4.7%).
Analysis of polymorphisms within SRD5A1, SRD5A2, and AKR1C3
Eleven SNPs were selected in the area coding for the SRD5A1 gene, of which three SNPs were located outside SRD5A1. In the SRD5A2 gene coding area, six SNPs were selected, of which two SNPs were located around the SRD5A2 gene. In the AKR1C3 gene coding area, 10 SNPs were selected, of which four were within the borders of the gene.
SRD5A1, SRD5A2, and AKR1C3 and MI
All polymorphisms were analyzed for their association with MI (Supplementary Tables 1–3, Supplemental digital content 1, http://links.lww.com/CAEN/A11). One tagging SNP was found to be significantly associated with MI in women: SNP rs248805G>A in SRD5A1 (Fig. 2) 28. The effect of this SNP was significantly different between men and women (P-value of interaction <0.001; data not shown).
For women carrying two variant alleles of rs248805, the hazard ratio (HR) for MI was 2.12 [95% confidence interval (CI) 1.36, 3.32] (Table 2). Adjustment for bioavailable testosterone increased the HR to 5.00 (95% CI 1.42, 17.61). In men, the association was not significant, but the effect of the polymorphism tended to be protective: homozygous variant genotype showed an HR of 0.84 (95% CI 0.56, 1.27) and after adjustment for bioavailable testosterone an HR of 0.84 (95% CI 0.33, 2.13). We tested the cardiovascular risk factors for potential confounding effects. However, none of them resulted in a 10% change in point estimate. Adjustment for all included risk factors only resulted in subtle changes in HRs (Table 2). SNP rs248805 was not significantly associated with bioavailable testosterone levels. Testosterone and SHBG levels in participants with and without MI are summarized in Table 3. Analyses using haplotypes did not show an increased association with MI (data not shown).
For the polymorphisms within SRD5A2 and AKR1C3, no significant associations with MI were found, with or without adjustment for the enzyme substrate.
Within the ARIC study population, SNP rs248805 was also associated with MI only in women, although in the opposite direction, and only for allele effect (HR 0.84; 95% CI 0.71, 0.99) (Table 4). Adjustment for the included cardiovascular risk factors did not alter point estimates and yielded similar results (data not shown).
Within RSII, carriage of one or more variant alleles of SNP rs248805 showed an effect on MI risk in the same direction as in the first cohort of the RS, although not statistically significant (Table 4). The number of MI cases was low because of the limited follow-up time. The risk of MI was increased in women (HR 1.44; 95% CI 0.65, 3.22); no effect was observed in men. Adjustment for covariates did not significantly alter the point estimate (data not shown).
In this study, we have shown that SRD5A1 has significantly higher cardiac expression levels in women than men. Furthermore, we found that genetic variation within SRD5A1 is associated with an increased risk of MI in western European women. This genetic effect was in contrast to that in an American population, where the same SNP appeared to be cardioprotective.
We have shown that SRD5A1 mRNA was locally expressed in normal and pathological human myocardial tissues, whereas SRD5A2 was not. A previous semiquantitative study found increased SRD5A1 mRNA in hypertrophic cardiomyopathy compared with normal samples 29. We could not replicate this finding in a larger series, but did detect a difference in ventricular SRD5A1 expression between men and women. A potential explanation for this difference might be that the upregulated SRD5A1 expression in women results from lower local androgen concentration, as we have described earlier for the expression of SRD5A1 in the prostate 18. In this tissue, AKR1C3 was also upregulated in the absence of androgens, supporting our present observation of a positive correlation between the levels of expression of SRD5A1 and AKR1C3. In agreement with the results of animal studies 30, no sex-dependent difference was shown in cardiac AR expression in the present study. As androgen concentrations, including DHT levels, are lower in women than in men, the effect of an increased SRD5A1 enzyme activity, and thus a higher conversion of testosterone to DHT, might be relatively more pronounced in women.
The results from our cohort data analysis indicate that there is an increased risk of MI in western European women carrying at least one variant allele of rs248805G>A within the SRD5A1 gene, with an almost 2.5 times higher risk in carriers of two variant alleles compared with the wild-type genotype. These results suggest that there is an increase in enzyme expression or activity in those carrying the variant allele. This will result in the increased production of DHT and therefore increased AR activation and steroid-related effects. In women, this effect is associated positively with an increased risk of atherosclerosis 5,8,31. A nested case–control study carried out by Rexrode et al.32 showed higher androgen levels in women with a cardiovascular event. In our study, a nonsignificant protective effect was observed in men. This remarkable dimorphic effect corresponds with the known effects of testosterone on the cardiovascular system in both sexes 3,8,10. In our study, adjustment for bioavailable testosterone in a subgroup with steroid level measurements increased the risk of MI. As discussed above, lower substrate androgen levels might stimulate SRD5A1 expression, which could account for this. In the ARIC cohort, an effect of the SNP was observed in women, although in the opposite direction. Analyses in a second cohort of the RSII, however, showed the same effect as in the original cohort of the RS, although statistically not significant. One explanation could be that there are differences in enzyme expression levels between Dutch white and American white populations with consequences for steroid effects. Unfortunately, we only obtained myocardial samples from western Europeans, which showed increased expression of SRD5A1 in women. Possibly, this increase could be absent in female hearts from other populations, leading to different effects of genetic variation in the SRD5A1 gene. These differences in expression levels or enzyme activity might be mediated by differences in dietary patterns. Nutrients have been shown to influence genetic effects through epigenetic mechanisms 33 and may contribute directly or indirectly toward the onset and severity of diseases 34. As an example, soy intake influenced the activity of 5α-reductase in mice. Mice with a low soy-containing diet had higher activity of 5α-reductase 35. In addition, an interaction with selenium was observed, showing a decrease in activity. Differences in dietary patterns could influence circulating steroid levels and their receptors and affect the final contribution of genetic variations, depending on the attributable risk of the separate factors. So far, no definitive conclusions have been drawn on this point.
Genetic polymorphisms in SRD5A1 have been described in earlier studies and variants in 5α-reductase have been associated with androgen-related diseases 16. The polymorphism of interest in our study, rs248805, is located in an intronic region of the SRD5A1 gene. This SNP is tightly linked to rs248793 (r2=0.81) and rs566202 (r2=0.94) 33, which are associated with peripheral arterial disease. Variant alleles of both rs248793 and rs566202 were associated with a protective effect on peripheral arterial disease 36. However, this study mainly included men (111 men vs. 34 women) and focused on another vascular system, in which expression patterns of SRD5A1 might differ. SNP rs248793G>C, previously detected as a restriction fragment length polymorphism (RFLP) at position 6633 kbp, is also referred to as the HinfI RFLP of SRD5A136. A study of 62 men showed a significant association between this RFLP and the DHT/T ratio 37. Carriage of two variant alleles was associated with a higher DHT/T ratio. This strengthens the suspicion of increased steroid effect, which we expected with the variant alleles of SNP rs248805.
The strength of our study is that it consists of two parts, indicating sex-specific effects on the myocardial expression of SRD5A1, combined with sex-specific effects of an SNP within this candidate gene on the risk of MI. Although the tissues examined are comprised of distinct cell types, both studies point toward sex-specific modulation of androgen action through SRD5A1, because of either transcriptional or genetic variances. However, as the genetic analysis could not be replicated in an independent cohort, we must be careful in drawing our conclusions.
Adjustment was performed for known cardiovascular confounders, but this showed only a minor difference with the crude analysis. It is unlikely that selection bias occurred by exclusion of study participants of whom no data on genotype were available as it is highly improbable that genotyping error or nonparticipation is dependent on genotype. The outcome, MI, is a fairly specific diagnosis. Therefore, misclassification is unlikely. Also, information bias is improbable as data on genotype and MI were prospectively recorded without knowledge of the research hypotheses.
This study is limited by the availability of human cardiac tissue, particularly that of coronary arteries. Although SRD5A1 levels were similar among the three myocardial tissue groups, the current cohort of myocardial samples precludes any definitive conclusions on sex-specific stimulation of enzyme levels during disease. Myocardial mRNA levels also do not necessarily reflect endothelial expression, but the increased SRD5A1 expression in women suggests a sex-specific difference in mRNA transcription or degradation. Combined with changes in serum testosterone levels, this could explain the association between MI and genetic variation in SRD5A1, only significant in women of western European descent. Besides, we studied mRNA expression, which may not completely correspond to protein levels. Studying coronary artery samples and local protein and steroid concentrations in a larger cohort could further substantiate conclusions on the local effects of SRD5A1 expression on the MI risk profile.
Our study suggests sex-specific cardiovascular effects of local testosterone conversion by SRD5A1. High testosterone levels were already known to be atheroprotective in men. This study is the first to suggest that a local effect of androgens in the heart is associated with an increased risk of MI in western European women, mediated by the local presence of androgen-metabolizing enzymes. Further research should be focused on the combination of genetics, expression levels, and dietary patterns, in a homogeneous population, ideally from conception onwards. The effect of nutritional genomics has been studied, although this has not yet been elucidated. This will be challenging, however, because of different environmental agents involved and the potential influence of health of the participants and lifelong exposure to dietary components.
Conflicts of interest
There are no conflicts of interest.
1. Kannel WB, Hjortland MC, McNamara PM, Gordon T. Menopause and risk of cardiovascular disease: the Framingham study. Ann Intern Med. 1976; 85:447–452.
2. Hodis HN, Mack WJ, Azen SP, Lobo RA, Shoupe D, Mahrer PR, et al.. Hormone therapy and the progression of coronary-artery atherosclerosis in postmenopausal women. N Engl J Med. 2003; 349:535–545.
3. Vitale C, Mendelsohn ME, Rosano GM. Gender differences in the cardiovascular effect of sex hormones. Nat Rev Cardiol. 2009; 6:532–542.
4. McGrath KC, McRobb LS, Heather AK. Androgen therapy and atherosclerotic cardiovascular disease. Vasc Health Risk Manag. 2008; 4:11–21.
5. Hak AE, Westendorp IC, Pols HA, Hofman A, Witteman JC. High-dose testosterone is associated with atherosclerosis in postmenopausal women. Maturitas. 2007; 56:153–160.
6. Bell JR, Bernasochi GB, Varma U, Raaijmakers AJ, Delbridge LM. Sex and sex hormones in cardiac stress – mechanistic insights. J Steroid Biochem Mol Biol. 2013; 137:124–135.
7. Schwartz JB, Volterrani M, Caminiti G, Marazzi G, Fini M, Rosano GM, Iellamo F. Effects of testosterone on the Q-T interval in older men and older women with chronic heart failure. Int J Androl. 2011; 34Pt 2e415–e421.
8. Bruck B, Brehme U, Gugel N, Hanke S, Finking G, Lutz C, et al.. Gender-specific differences in the effects of testosterone and estrogen on the development of atherosclerosis in rabbits. Arterioscler Thromb Vasc Biol. 1997; 17:2192–2199.
9. Malkin CJ, Pugh PJ, Jones RD, Jones TH, Channer KS. Testosterone as a protective factor against atherosclerosis – immunomodulation and influence upon plaque development and stability. J Endocrinol. 2003; 178:373–380.
10. Hak AE, Witteman JC, de Jong FH, Geerlings MI, Hofman A, Pols HA. Low levels of endogenous androgens increase the risk of atherosclerosis in elderly men: the Rotterdam study. J Clin Endocrinol Metab. 2002; 87:3632–3639.
11. Rajkhowa M, Glass MR, Rutherford AJ, Michelmore K, Balen AH. Polycystic ovary syndrome: a risk factor for cardiovascular disease? BJOG. 2000; 107:11–18.
12. Labrie F, Luu-The V, Lin SX, Simard J, Labrie C. Role of 17 beta-hydroxysteroid dehydrogenases in sex steroid formation in peripheral intracrine tissues. Trends Endocrinol Metab. 2000; 11:421–427.
13. Grino PB, Griffin JE, Wilson JD. Testosterone at high concentrations interacts with the human androgen receptor similarly to dihydrotestosterone. Endocrinology. 1990; 126:1165–1172.
14. Normington K, Russell DW. Tissue distribution and kinetic characteristics of rat steroid 5 alpha-reductase isozymes. Evidence for distinct physiological functions. J Biol Chem. 1992; 267:19548–19554.
15. Martel C, Melner MH, Gagne D, Simard J, Labrie F. Widespread tissue distribution of steroid sulfatase, 3 beta-hydroxysteroid dehydrogenase/delta 5-delta 4 isomerase (3 beta-HSD), 17 beta-HSD 5 alpha-reductase and aromatase activities in the rhesus monkey. Mol Cell Endocrinol. 1994; 104:103–111.
16. Goodarzi MO, Shah NA, Antoine HJ, Pall M, Guo X, Azziz R. Variants in the 5alpha-reductase type 1 and type 2 genes are associated with polycystic ovary syndrome and the severity of hirsutism in affected women. J Clin Endocrinol Metab. 2006; 91:4085–4091.
17. Chai W, Hofland J, Jansen PM, Garrelds IM, de Vries R, van den Bogaerdt AJ, et al.. Steroidogenesis vs. steroid uptake in the heart: do corticosteroids mediate effects via cardiac mineralocorticoid receptors? J Hypertens. 2010; 28:1044–1053.
18. Hofland J, van Weerden WM, Dits NF, Steenbergen J, van Leenders GJ, Jenster G, et al.. Evidence of limited contributions for intratumoral steroidogenesis in prostate cancer. Cancer Res. 2010; 70:1256–1264.
19. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods. 2001; 25:402–408.
20. Hofman A, van Duijn CM, Franco OH, Ikram MA, Janssen HL, Klaver CC, et al.. The Rotterdam Study: 2012 objectives and design update. Eur J Epidemiol. 2011; 26:657–686.
21. de Bruyne MC, Mosterd A, Hoes AW, Kors JA, Kruijssen DA, van Bemmel JH, et al.. Prevalence, determinants, and misclassification of myocardial infarction
in the elderly. Epidemiology. 1997; 8:495–500.
22. World Health Organization. International statistical classification of diseases and health related problems, 10th revision. 1992.Geneva:World Health Organization.
23. Rodenburg EM, Visser LE, Danser AH, Hofman A, van Noord C, Witteman JC, et al.. Genetic variance in CYP2C8 and increased risk of myocardial infarction
. Pharmacogenet Genomics. 2010; 20:426–434.
24. de Ronde W, Hofman A, Pols HA, de Jong FH. A direct approach to the estimation of the origin of oestrogens and androgens in elderly men by comparison with hormone levels in postmenopausal women. Eur J Endocrinol. 2005; 152:261–268.
25. Södergård R, Bäckström T, Shanbhag V, Carstensen H. Calculation of free and bound fractions of testosterone and estradiol-17 beta to human plasma proteins at body temperature. J Steroid Biochem. 1982; 16:801–810.
26. Van den Beld AW, de Jong FH, Grobbee DE, Pols HA, Lamberts SW. Measures of bioavailable serum testosterone and estradiol and their relationships with muscle strength, bone density, and body composition in elderly men. J Clin Endocrinol Metab. 2000; 85:3276–3282.
27. The ARIC investigators. The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. Am J Epidemiol. 1989; 129:687–702.
28. Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005; 21:263–265.
29. Thum T, Borlak J. Testosterone, cytochrome P450, and cardiac hypertrophy. FASEB J. 2002; 16:1537–1549.
30. Lizotte E, Grandy SA, Tremblay A, Allen BG, Fiset C. Expression, distribution and regulation of sex steroid hormone receptors in mouse heart. Cell Physiol Biochem. 2009; 23:75–86.
31. Liu PY, Death AK, Handelsman DJ. Androgens and cardiovascular disease. Endocr Rev. 2003; 24:313–340.
32. Rexrode KM, Manson JE, Lee IM, Ridker PM, Sluss PM, Cook NR, Buring JE. Sex hormone levels and risk of cardiovascular events in postmenopausal women. Circulation. 2003; 108:1688–1693.
33. Clarke SD. Nutrient regulation of gene and protein expression. Curr Opin Clin Nutr Metab Care. 1999; 2:287–289.
34. Kaput J, Rodriguez RL. Nutritional genomics: the next frontier in the postgenomic era. Physiol Genomics. 2004; 16:166–177.
35. Christensen MJ, Quiner TE, Nakken HL, Lephart ED, Eggett DL, Urie PM. Combination effects of dietary soy and methylselenocysteine in a mouse model of prostate cancer. Prostate. 2013; 73:986–995.
36. Signorelli SS, Barresi V, Musso N, Anzaldi M, Croce E, Fiore V, Condorelli DF. Polymorphisms of steroid 5-alpha-reductase type I (SRD5A1) gene are associated to peripheral arterial disease. J Endocrinol Invest. 2008; 31:1092–1097.
37. Ellis JA, Panagiotopoulos S, Akdeniz A, Jerums G, Harrap SB. Androgenic correlates of genetic variation in the gene encoding 5alpha-reductase type 1. J Hum Genet. 2005; 50:534–537.