Haplotype analyses of CYP17A1 genetic polymorphisms and coronary artery disease in a Uygur population

Abstract

Background:

The relationship between CYP17A1 genetic polymorphisms and coronary artery disease (CAD) remains unclear. The aim of the present study was to assess the association between CYP17A1 gene polymorphism and CAD in a Chinese Uygur population.

Methods:

A total of 493 people including 266 patients and 227 controls were selected for the present study. All CAD patients and controls were genotyped for the same five single nucleotide polymorphisms (SNPs) (rs4919686, rs1004467, rs4919687, rs10786712, and rs2486758) by a real-time PCR method.

Results:

The rs4919686, rs1004467, and rs4919687 polymorphisms were found to be associated with CAD in genotypes, dominant model, recessive model, and allele frequency (rs4919686: all p<0.05, rs1004467: all p≤0.001, rs4919687: all p<0.001); the significant difference was retained (all p<0.05) after adjustment for the major confounding factors. The overall distribution of haplotypes established by SNP1–SNP4 (in total subjects and men) and SNP1–SNP4–SNP5 (in total subjects) were significantly different between the CAD patients and the control subjects (p=0.006, men: p=0.026, and p=0.030, respectively).

Conclusion:

Polymorphisms rs4919686, rs4919687 and rs1004467 were found to be associated with CAD in this Uygur population.

Keywords

CYP17A1 single nucleotide polymorphism coronary artery disease case-control study

Introduction

Coronary artery disease (CAD) and its most severe complications, such as myocardial infarction, are leading causes of mortality worldwide.^1,2 It is a complex multifactorial disorder thought to result from several susceptibility genes and multiple environmental determinants.^3,4 Recently, the genetic basis of CAD has gained considerable interest; heritable factors were found to account for 40–60% of occurrence and development of CAD.⁵ With the development of genome-wide association studies concerning CAD, a growing number of susceptibility genes related to CAD have been revealed, and some large-scale association analyses have identified many common, uncommon and functional variants for CAD.^6,7

The cytochrome P450 proteins catalyze many reactions involved in drug metabolism and synthesis of cholesterol, steroids and other lipids. They are responsible for not only the metabolism of xenobiotics but also a host of endogenous substances whose metabolites have critical roles in the maintenance of cardiovascular health.^8,9 Mounting evidence has demonstrated that CYP enzymes are involved in the pathogenesis of CAD. For example, the CYP2C19 genetic variants involved in the conversion of clopidogrel to its active metabolite are associated with an increased rate of recurrent cardiovascular events, including stent thrombosis (ST);^10,11 recent reports suggest that CYP8A1, predominantly in vascular endothelial and smooth muscle cells, is a member of the CYP family and acts mainly as an enzyme that converts prostaglandin H2 (PGH2) into prostacyclin (PGI2); and some studies suggest that gene polymorphisms are associated with cardiovascular risk.¹² In addition, CYP1A1, CYP1A2 (which metabolize tobacco polycyclic aromatic hydrocarbons and aromatic amines during smoking),^13,14 CYP2C8, CYP2J2 (EET synthesis),^15,16 CYP3A4 (the main metabolic enzyme of fentanyl),¹⁷ and CYP17 and CYP19 (synthesis of hormones)¹⁸ have been demonstrated to have a relationship with CAD.

In humans, the CYP17A1 gene is located on chromosome 10q24.3, and is mainly expressed in the adrenal glands and gonads. This gene encodes a member of the cytochrome P450 superfamily of enzymes. The CYP17A1 gene produces P450c17 proteins; it has both 17alpha-hydroxylase and 17,20-lyase activities and is a key enzyme in the steroidogenic pathway that produces progestins, mineralocorticoids, glucocorticoids, androgens, and estrogens. CYP17A1 gene mutations affect the synthesis of steroids, which are the precursors of sex hormones. Some evidence has indicated that the levels of sex hormones can affect the development of cardiovascular and cerebrovascular diseases.¹⁹ Sex hormones such as estrogens protect against oxidative stress and are known to be vasoprotective.^20-22

In recent years, several studies have investigated the association between CYP17A1 heritable variation and CAD.^7,23 However, the relationship between CYP 17A1 genetic polymorphisms and CAD in the Uygur population remains unclear. In the present study, we aimed to assess the association between the polymorphism of CYP17A1 and CAD in the Uygur population in Xinjiang, China.

Methods

Ethical approval of the study protocol

This study was approved by the Ethics Committee of the First Affiliated Hospital of Xinjiang Medical University (Xinjiang, China). Written informed consent was obtained from each participant, and explicitly provided permission for DNA analyses as well as collection of relevant clinical data. This study was conducted according to the standards of the Declaration of Helsinki.

Study population

The subjects were from a Uygur population who lived in the Xinjiang Uygur Autonomous Region of China, and comprise 493 people (326 men, 167 women), with 266 patients and 227 ethnically and geographically matched controls. All subjects attended as inpatients in the First Affiliated Hospital of Xinjiang Medical University from 2007 to 2013. CAD group patients were defined as having the presence of at least one significant coronary artery stenosis of >50% luminal diameter on coronary angiography. Patients with congenital heart disease, valvular disease, multiple organ failure syndrome, malignancy or chronic inflammatory disease were excluded from this study. Each control subject also underwent a coronary angiogram and did not show coronary artery stenosis, and the electrocardiogram did not indicate CAD.

Biochemical analyses

Before cardiac catheterization, a 5 ml sample of fasting venous blood drawn by venipuncture in the Cardiac Catheterization Laboratory was taken from all participants. The blood samples were collected into tubes containing ethylene diamine tetraacetic acid (EDTA), and to separate the plasma content were centrifuged at 4000 ×g for 5 min. Genomic DNA was extracted using a standard phenol-chloroform method. Serum concentrations of glucose (Glu), total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), blood urea nitrogen (BUN), creatinine (Cr) and uric acid (UA) were measured using standard methods in the Central Laboratory of First Affiliated Hospital of Xinjiang Medical University as described previously.^24-26

Genotyping of the CYP17A1 gene

Using Haploview 4.2 software and International HapMap Project website phase II database (http://www.hapmap.org), we obtained five tag single nucleotide polymorphisms (SNPs) (SNP1: rs4919686, SNP2: rs1004467, SNP3: rs4919687, SNP4: rs10786712, SNP5: rs2486758) by using minor allele frequency (MAF) ≥0.05 and linkage disequilibrium (LD) patterns with r²≥ 0.8 as a cut off. Genotyping was confirmed by the TaqMan SNP Genotyping Assay (Applied Biosystems, Foster City, CA) according to the previous report.

Statistical analysis

Differences in categorical variables (e.g. age, body mass index (BMI), TC, TG, HDL-C, LDL-C, Glu, BUN, Cr, UA) were analyzed using means ± standard deviation (SD). The difference between the CAD and control groups was analyzed using an independent-sample t-test. Differences in frequencies of smoking, drinking, hypertension, diabetes mellitus, and CYP17A1 genotypes were analyzed using χ2 test or Fisher’s exact test where appropriate. Hardy–Weinberg equilibrium was assessed by χ2 analysis. Logistic regression analyses with effect ratios (odds ratio (OR) and 95% confidence intervals (CI)) were used to assess the contribution of major risk factors. Data analysis was carried out using the computer software Statistical Package for Social Sciences SPSS 17.0 for Windows (SPSS Institute, Chicago, USA). Statistical significance was established at a p-value <0.05. In addition, based on the genotype data of the genetic variations, we performed LD analysis and haplotype-based case-control analysis, using the expectation maximization algorithm and the software SHEsis (http://analysis.bio-x.cn/SHEsisMain.htm). The pairwise LD analysis was performed using five SNP pairs. We used|D’|values of >0.25 to assign SNP locations to one haplotype block. Single nucleotide polymorphisms with an r² value of <0.5 were selected as tagged. In the haplotype-based case-control analysis, haplotypes with a frequency of <0.03 were excluded. The frequency distribution of the haplotypes was calculated by performing a permutation test using the bootstrap method.

Results

Characteristics of the study participants

As shown in Table 1, there was no significant difference in age between patients with CAD and control subjects, which indicated that the study was an age-matched case-control study. For total and female subjects, the systolic blood pressure (SBP), serum concentration of glucose, TG, UA, Cr, and the incidence of diabetes were significantly higher in subjects with CAD than in the controls, and the serum concentration of HDL-C was significantly lower for patients with CAD than for control participants. For all, the diastolic blood pressure (DBP) was significantly higher in CAD subjects than in the controls. For women, the serum concentration of BUN was significantly higher for the CAD patients than in the control subjects. There was no significant difference in the following variables between the CAD patients and the control participants in total and female subjects: the incidence of hypertension, smoking, drinking, BMI, plasma concentration of TC, and LDL-C. For men, the incidence of smoking, SBP, and glucose were significantly higher for the CAD patients as compared with the control subjects, and HDL-C was significantly higher for control participants than for patients with CAD; for other variables there was no significant difference in men.

Table 1.

Demographic and clinical characteristics of study participants.

	Total			Men			Women
	CAD	Control	p-value	CAD	Control	p-value	CAD	Control	p-value
Number (n)	266	227		181	145		85	82
Age, mean (SD)	57.32 (9.25)	56.09 (9.57)	0.146	56.62 (9.63)	55.83 (10.00)	0.460	58.85 (9.18)	56.54 (8.80)	0.104
EH, n (%)	99 (37.2)	82 (36.1)	0.802	58 (32.0)	45 (31.0)	0.845	41 (48.2)	37 (45.1)	0.687
Diabetes, n (%)	37 (13.9)	15 (6.6)	0.009	21 (11.6)	9 (6.2)	0.094	16 (17.6)	6 (7.3)	0.044
Smoking, n (%)	29 (10.9)	16 (7.0)	0.139	36 (19.9)	9 (6.2)	<0.001	0	0	1
Drinking, n (%)	19 (7.1)	14 (6.2)	0.666	23 (12.7)	9 (6.2)	0.050	1 (1.2)	0	1
BMI, mean (SD)	26.95 (3.46)	26.51(4.45)	0.230	27.27 (3.45)	26.46 (4.14)	0.060	26.28 (3.41)	26.62 (4.97)	0.610
SBP, mean (SD)	142.32 (28.67)	133.0 (23.95)	<0.001	140.63 (28.95)	132.55 (22.81)	0.007	145.79 (27.96)	133.77 (25.94)	0.005
DBP, mean (SD)	87.47 (20.17)	83.77 (17.53)	0.033	87.10 (20.14)	83.45 (17.32)	0.089	88.24 (20.34)	84.31 (17.99)	0.193
Glu, mean (SD)	6.13 (2.50)	5.19 (1.92)	<0.001	6.04 (2.40)	5.10 (1.80)	<0.001	6.32 (2.70)	5.35 (2.11)	0.012
TG, mean (SD)	1.96 (1.14)	1.65 (1.03)	0.002	1.86 (1.01)	1.67 (1.05)	0.103	2.19 (1.36)	1.61 (1.01)	0.003
TC, mean (SD)	4.26 (1.38)	4.31 (0.89)	0.633	4.19 (1.35)	4.37 (0.92)	0.198	4.43 (1.43)	4.23 (0.84)	0.303
HDL, mean (SD)	0.92 (0.29)	1.13 (0.37)	<0.001	0.91 (0.27)	1.12 (0.39)	<0.001	0.96 (0.32)	1.15 (0.34)	0.001
LDL, mean (SD)	2.64 (1.09)	2.75 (0.94)	0.242	2.63 (1.16)	2.76 (0.88)	0.285	2.67 (0.92)	2.75 (1.04)	0.628
UA, mean (SD)	307.97 (98.76)	279.91 (87.20)	0.001	323.71 (94.29)	305.29 (83.38)	0.072	273.90 (100.21)	235.50 (75.55)	0.007
Cr, mean (SD)	76.25 (38.13)	69.18 (18.60)	0.010	80.44 (36.52)	76.30 (17.95)	0.221	67.18 (40.16)	56.72 (12.03)	0.029
BUN, mean (SD)	5.41 (1.90)	5.35 (1.45)	0.693	5.39 (1.81)	5.62 (1.47)	0.239	5.45 (2.10)	4.88 (1.28)	0.042

Continuous variables are expressed as mean ± SD. Categorical variables are expressed as percentages.

BMI: body mass index; BUN: blood urea nitrogen; Cr: creatinine; DBP: diastolic blood pressure; DM: diabetes mellitus; Glu: glucose; TG: triglyceride; TC: total cholesterol; HDL: high-density lipoprotein; LDL: low-density lipoprotein; EH: essential hypertension; SBP: systolic blood pressure; UA: uric acid.

The p-value of the continuous variables was calculated by independent t-test. The p-value of the categorical variables was calculated by Fisher’s exact test.

Distributions of CYP17A1 genotypes

As shown in Table 2, the genotype distributions for each SNP were in good agreement with the predicted Hardy–Weinberg equilibrium values (data not shown). For total, male and female subjects, the distribution of SNP4 (rs10786712) and SNP5 (rs2486758) genotypes did not show a significant difference between CAD and control subjects (p>0.05 respectively). For total subjects, distribution of rs4919686 genotypes, dominant model (AC + CC vs. AA), recessive model (AA+AC vs. CC) and allele frequency showed significant difference between CAD and control subjects (p=0.034, p=0.030, p=0.047 and p=0.011, respectively); the dominant model, recessive model, and allele frequency of rs4919686 were significantly higher in subjects with CAD than in controls (37.7% vs. 28.2%, 5.3% vs. 1.8%, and 21.5% vs. 15.0%). Distribution of rs1004467 genotypes; dominant model (CC + CT vs. TT), recessive model (CT+TT vs. CC) and allele frequency showed significant difference between CAD and control subjects (p<0.001, p=0.001, p<0.001 and p<0.001), and the dominant model, recessive model, and allele frequency of rs1004467 were significantly higher in controls than in CAD subjects (83.8% vs. 70.9%, 32.0% vs. 18.1%, and 57.9% vs. 44.5%). Distribution of rs4919687 genotypes; dominant model (AA + AG vs. GG), recessive model (AG+GG vs. AA) and allele frequency showed significant difference between CAD and control subjects (all p<0.001), and the dominant model, recessive model, and allele frequency of rs4919687 were significantly higher in CAD patients than in controls subjects (58.4% vs. 41.7%, 19.6% vs. 7.6%, and 39.0% vs. 24.7%).

Table 2.

Genotype and allele distributions in patients with CAD and control participants.

Variants		Total			Man			Woman
Variants		CAD n (%)	Control n (%)	p-value	CAD n (%)	Control n (%)	p-value	CAD n (%)	Control n (%)	p-value
Rs4919686 (SNP1)
Genotyping	AA	154 (62.3)	158 (71.8)	0.034	107 (62.6)	100 (71.9)	0.166	47 (61.8)	58 (71.6)	0.136
	AC	80 (32.4)	58 (26.4)		56 (32.7)	36 (25.9)		24 (31.6)	22 (27.2)
	CC	13 (5.3)	4 (1.8)		8 (4.7)	3 (2.2)		5 (6.6)	1 (1.2)
Recessive model	CC	13 (5.3)	4 (1.8)	0.047	8 (4.7)	3 (2.2)	0.233	5 (6.6)	1 (1.2)	0.184
Recessive model	AA+AC	234 (94.7)	216 (98.2)	0.047	163 (95.3)	136 (97.8)	0.233	71 (93.4)	80 (98.8)	0.184
Dominant model	AA	154 (62.3)	158 (71.8)	0.030	107 (62.6)	100 (71.9)	0.082	47 (61.8)	58 (71.6)	0.194
Dominant model	AC+CC	93 (37.7)	63 (28.2)	0.030	64 (37.4)	39 (28.1)	0.082	29 (38.2)	23 (28.4)	0.194
Additive model	AC	80 (32.4)	58 (26.4)	0.154	56 (32.7)	36 (25.9)	0.189	24 (31.6)	22 (27.2)	0.543
Additive model	AA+CC	167 (67.6)	162 (73.6)	0.154	115 (67.3)	103 (74.1)	0.189	52 (68.4)	59 (72.8)	0.543
Allele	A	388 (78.5)	37.4 (85.0)	0.011	270 (78.9)	236 (84.9)	0.057	118 (77.6)	138 (85.2)	0.085
Allele	C	106 (21.5)	66 (15.0)	0.011	72 (21.1)	42 (15.1)	0.057	34 (22.4)	24 (14.8)	0.085
Rs1004467 (SNP2)
Genotyping	CC	46 (18.1)	71 (32.0)	<0.001	26 (14.8)	29 (20.6)	0.359	20 (25.6)	42 (51.9)	<0.001
	CT	134 (52.8)	115 (51.8)		99 (56.3)	77 (54.6)		35 (44.9)	38 (46.9)
	TT	74 (19.1)	36 (16.2)		51 (29.0)	35 (24.8)		23 (29.5)	1 (1.2)
Recessive model	CC	46 (18.1)	71 (32.0)	<0.001	26 (14.8)	29 (20.6)	0.176	20 (25.6)	42 (51.9)	0.001
Recessive model	CT+TT	208 (81.9)	151 (68.0)	<0.001	150 (85.2)	112 (79.4)	0.176	58 (74.4)	39 (48.1)	0.001
Dominant model	TT	74 (19.1)	36 (16.2)	0.001	51 (29.0)	35 (24.8)	0.408	23 (29.5)	1 (1.2)	<0.001
Dominant model	CC+CT	180 (70.9)	186 (83.8)	0.001	125 (71.0)	106 (75.2)	0.408	55 (70.5)	80 (98.8)	<0.001
Additive model	CT	134 (52.8)	115 (51.8)	0.835	99 (56.3)	77 (54.6)	0.770	35 (44.9)	38 (46.9)	0.796
Additive model	CC+TT	120 (47.2)	107 (48.2)	0.835	77 (43.8)	64 (45.4)	0.770	43 (55.1)	43 (53.1)	0.796
Allele	C	226 (44.5)	257 (57.9)	<0.001	151 (42.9)	135 (74.9)	0.211	75 (48.1)	122 (75.3)	<0.001
Allele	T	282 (55.5)	187 (42.1)	<0.001	201 (57.1)	147 (52.1)	0.211	81 (51.9)	40 (24.7)	<0.001
Rs4919687 (SNP3)
Genotyping	AA	50 (19.6)	17 (7.6)	<0.001	44 (24.9)	12 (8.5)	<0.001	6 (7.7)	5 (6.1)	0.380
	AG	99 (38.8)	76 (34.1)		68 (38.4)	51 (36.2)		31 (39.7)	25 (30.5)
	GG	106 (41.6)	130 (58.3)		65 (36.7)	78 (55.3)		41 (52.6)	52 (63.4)
Recessive model	AA	50 (19.6)	17 (7.6)	<0.001	44 (24.9)	12 (8.5)	<0.001	6 (7.7)	5 (6.1)	0.690
Recessive model	AG+GG	205 (80.4)	206 (92.4)	<0.001	133 (75.1)	129 (91.5)	<0.001	72 (92.3)	77 (93.9)	0.690
Dominant model	GG	106 (41.6)	130 (58.3)	<0.001	65 (36.7)	78 (55.3)	0.001	41 (52.6)	52 (63.4)	0.164
Dominant model	AA+AG	149 (58.4)	93 (41.7)	<0.001	112 (63.3)	63 (44.7)	0.001	37 (47.4)	30 (36.6)	0.164
Additive model	AG	99 (38.8)	76 (34.1)	0.283	68 (38.4)	51 (36.2)	0.681	31 (39.7)	25 (30.5)	0.220
Additive model	AA+GG	156 (61.2)	147 (65.9)	0.283	109 (61.6)	90 (63.8)	0.681	47 (60.3)	57 (69.5)	0.220
Allele	A	199 (39.0)	110 (24.7)	<0.001	43 (27.6)	35 (21.3)	0.195	43 (27.6)	35 (21.3)	0.195
Allele	G	311 (61.0)	336 (75.3)	<0.001	113 (72.4)	129 (78.7)	0.195	113 (72.4)	129 (78.7)	0.195
Rs10786712 (SNP4)
Genotyping	CC	83 (32.9)	66 (29.9)	0.481	58 (33.5)	40 (28.6)	0.476	25 (31.6)	26 (32.1)	0.879
	CT	116 (46.0)	114 (51.6)		76 (43.9)	71 (50.7)		40 (50.6)	43 (53.1)
	TT	53 (21.0)	41 (18.6)		39 (22.5)	29 (20.7)		14 (17.7)	12 (14.8)
Recessive model	TT	53 (21.0)	41 (18.6)	0.500	39 (22.5)	29 (20.7)	0.696	14 (17.7)	12 (14.8)	0.618
Recessive model	CC+CT	199 (79.0)	180 (81.4)	0.500	134 (77.5)	111 (79.3)	0.696	65 (82.3)	69 (85.2)	0.618
Dominant model	CC	83 (32.9)	66 (29.9)	0.473	58 (33.5)	40 (28.6)	0.347	25 (31.6)	26 (32.1)	0.951
Dominant model	CT+TT	169 (67.1)	155 (70.1)	0.473	115 (66.5)	100 (71.4)	0.347	54 (68.4)	55 (67.9)	0.951
Additive model	CT	116 (46.0)	114 (51.6)	0.228	76 (43.9)	71 (50.7)	0.232	40 (50.6)	43 (53.1)	0.756
Additive model	CC+TT	136 (54.0)	107 (48.4)	0.228	97 (56.1)	69 (49.3)	0.232	39 (49.4)	38 (46.9)	0.756
Allele	C	282 (56.0)	246 (55.7)	0.927	192 (55.5)	151 (53.9)	0.696	90 (57.0)	95 (58.6)	0.761
Allele	T	222 (44.0)	196 (44.3)	0.927	154 (44.5)	129 (46.1)	0.696	68 (43.0)	67 (41.4)	0.761
Rs2486758 (SNP5)
Genotyping	CC	12 (5.2)	10 (4.4)	0.434	7 (4.1)	7 (4.9)	0.560	5 (8.5)	3 (3.7)	0.421
	CT	79 (34.2)	90 (40.0)		58 (33.7)	56 (38.9)		21 (35.6)	34 (42.0)
	TT	140 (60.6)	125 (55.6)		107 (62.2)	81 (56.3)		33 (55.9)	44 (54.3)
Recessive model	CC	12 (5.2)	10 (4.4)	0.709	7 (4.1)	7 (4.9)	0.733	5 (8.5)	3 (3.7)	0.230
Recessive model	CT+TT	219 (94.8)	215 (95.6)	0.709	165 (95.9)	137 (95.1)	0.733	54 (91.5)	78 (96.3)	0.230
Dominant model	TT	140 (60.6)	125 (55.6)	0.274	107 (62.2)	81 (56.3)	0.282	33 (55.9)	44 (54.3)	0.850
Dominant model	CC+CT	91 (39.4)	100 (44.4)	0.274	65 (37.8)	63 (43.8)	0.282	26 (44.1)	37 (45.7)	0.850
Additive model	CT	79 (34.2)	90 (40.0)	0.200	58 (33.7)	56 (38.9)	0.341	21 (35.6)	34 (42.0)	0.445
Additive model	CC+TT	152 (65.8)	135 (60.0)	0.200	114 (66.3)	88 (61.1)	0.341	38 (64.4)	47 (58.0)	0.445
Allele	C	103 (22.3)	110 (24.4)	0.443	72 (20.9)	70 (24.3)	0.311	31 (26.3)	40 (24.7)	0.764
Allele	T	359 (77.7)	340 (75.6)	0.443	272 (79.1)	218 (75.7)	0.311	87 (73.7)	122 (75.3)	0.764

CAD: coronary artery disease; n: number of participants; SNP: single nucleotide polymorphism.

For men, the distribution of rs4919687 genotypes, dominant model (AA + AG vs GG), and recessive model (AG+GG vs. AA) showed significant difference between CAD and control subjects (p<0.001, p=0.001, and p<0.001), and the dominant model and recessive model of rs4919687 were significantly higher in CAD patients than in control subjects (63.3% vs. 44.7%, 24.9% vs. 8.5%, respectively). For women, the distribution of rs1004467 genotypes, dominant model (CC + CT vs. TT), recessive model (CT+TT vs. CC) and allele frequency showed significant difference between CAD and control subjects (p<0.001, p<0.001, p=0.001 and p<0.001), and the dominant model, recessive model, and allele frequency of rs1004467 were significantly higher in controls than in CAD subjects (98.8% vs. 70.5%, 51.9% vs. 25.6%, and 75.3% vs. 48.1%).

Logistic regression analyses

Multivariable logistic regression analysis combed genotypes with the following variables: the incidence of hypertension, diabetes, smoking, and drinking, BMI, SBP, DBP, Glu, TG, TC, HDL-C, LDL-C, UA, Cr, and BUN. For total subjects and women (Table 3), after multivariate adjustment, rs1004467 remain significantly associated with CAD in the recessive model (total: OR=2.259, 95%CI: 1.345–3.795, p=0.002; women: OR=3.782, 95%CI: 1.565–9.140, p=0.003), and in the dominant model (total: OR=0.550, 95%CI: 0.324–0.934, p=0.027; women: OR=0.04, 95%CI: 0.004–0.365, p=0.004). For total subjects (Table 4), after multivariate adjustment, rs4919686 remain significantly associated with CAD in the dominant model (OR=1.749, 95%CI: 1.093–2.798, p=0.020) and in the recessive model (OR=0.153, 95%CI: 0.031–0.762, p=0.022). For total subjects and men (Table 5), after multivariate adjustment, rs4919687 remain significantly associated with CAD in the dominant model (total: OR=1.801, 95%CI: 1.161–2.795, p=0.009; men: OR=2.448, 95%CI: 1.415–4.234, p=0.001) and in the recessive model (total: OR=0.261, 95%CI: 0.126–0.541, p<0.001; men: OR=0.170, 95%CI: 0.071–0.408, p<0.001)

Table 3.

Multiple logistic regression analysis for CAD patients and control subjects (rs1004467).

	Total			Men			Woman
	OR	95% CI	p	OR	95% CI	p	OR	95% CI	p
Recessive model	2.259	1.345–3.795	0.002	1.694	0.834–3.441	0.145	3.782	1.565–9.140	0.003
Hypertension	0.772	0.473–1.261	0.301	0.895	0.489–1.639	0.710	0.404	0.150–1.089	0.073
Diabetes	1.384	0.619–3.094	0.428	0.913	0.322–2.588	0.864	3.281	0.786–13.696	0.103
Smoking	0.635	0.206–1.958	0.430	0.812	0.248–2.660	0.731	0	0	1
Drinking	1.093	0.309–3.864	0.891	0.789	0.206–3.017	0.729	0	0	1
Glucose	1.155	1.013–1.316	0.031	1.221	1.031–1.447	0.021	1.053	0.818–1.354	0.690
Triglyceride	1.092	0.874–1.363	0.438	0.865	0.641–1.167	0.342	1.659	1.091–2.522	0.018
HDL	0.197	0.094–0.417	<0.001	0.163	0.063–0.423	<0.001	0.183	0.045–0.739	0.017
Creatinine	1.008	0.997–1.019	0.138	1.005	0.993–1.016	0.421	1.044	1.003–1.087	0.037
Uric acid	1.002	0.999–1.004	0.254	1.002	0.998–1.005	0.328	1.002	0.996–1.008	0.430
SBP	1.022	1.007–1.038	0.004	1.018	0.999–1.037	0.057	1.040	1.007–1.074	0.017
DBP	0.985	0.965–1.006	0.152	0.993	0.968–1.019	0.618	0.964	0.926–1.004	0.075

CAD: coronary artery disease; OR: odds ratios; 95%CI: 95% confidence intervals; SBP: systolic blood pressure; DBP: diastolic blood pressure; HDL: high-density lipoprotein.

Table 4.

Multiple logistic regression analysis for CAD patients and control subjects (rs4919686).

	Total			Men			Woman
	OR	95% CI	p	OR	95% CI	p	OR	95% CI	p
Dominant model	1.749	1.093–2.798	0.020	2.377	1.313–4.301	0.004	0.975	0.391–2.433	0.957
Hypertension	0.746	0.450–1.236	0.255	0.853	0.455–1.600	0.620	0.411	0.148–1.146	0.089
Diabetes	1.422	0.626–3.232	0.400	0.807	0.261–2.495	0.709	2.876	7.38–11.218	0.128
Smoking	0.816	0.269–2.482	0.721	0.818	0.254–2.637	0.737	0	0	1
Drinking	0.660	0.183–2.378	0.525	0.651	0.172–2.468	0.528	0	0	1
Glucose	1.126	0.984–1.289	0.085	1.226	1.020–1.473	0.030	1.039	0.819–1.318	0.754
Triglyceride	1.187	0.944–1.493	0.142	0.876	0.637–1.205	0.416	1.808	1.197–2.732	0.005
HDL	0.156	0.070–0.345	<0.001	0.110	0.039–0.311	<0.001	0.144	0.033–0.638	0.011
Creatinine	1.011	0.999–1.023	0.079	1.007	0.994–1.020	0.294	1.039	1.000–1.079	0.053
Uric acid	1.002	0.999–1.004	0.252	1.002	0.998–1.005	0.293	1.002	0.996–1.008	0.475
SBP	1.023	1.007–1.039	0.005	1.019	1.000–1.039	0.053	1.039	1.007–1.071	0.016
DBP	0.987	0.967–1.008	0.214	0.995	0.970–1.022	0.729	0.966	0.930–1.003	0.073

CAD: coronary artery disease; OR: odds ratios; 95%CI: 95% confidence intervals; SBP: systolic blood pressure; DBP: diastolic blood pressure; HDL: high-density lipoprotein.

Table 5.

Multiple logistic regression analysis for CAD patients and control subjects (rs4919687).

	Total			Men			Woman
	OR	95% CI	p	OR	95% CI	p	OR	95% CI	p
Dominant model (GG+GA vs GG)	1.801	1.161–2.795	0.009	2.448	1.415–4.234	0.001	0.909	0.375–2.200	0.832
Hypertension	0.832	0.508–1.360	0.462	1.028	0.556–1.902	0.930	0.394	0.143–1.083	0.071
Diabetes	1.544	0.686–3.479	0.294	1.124	0.380–3.328	0.833	2.709	0.688–10.666	0.154
Smoking	0.854	0.277–2.630	0.783	1.123	0.336–3.751	0.851	0	0	1
Drinking	0.818	0.233–2.873	0.754	0.564	0.146–2.180	0.406	0	0	1
Glucose	1.155	1.011–1.318	0.034	1.215	1.023–1.444	0.027	1.064	0.837–1.354	0.611
Triglyceride	1.089	0.861–1.378	0.478	0.858	0.631–1.167	0.330	1.704	1.130–2.569	0.011
HDL	0.203	0.096–0.427	<0.001	0.145	0.055–0.382	<0.001	0.173	0.041–0.735	0.017
Creatinine	1.009	0.998–1.021	0.120	1.004	0.993–1.016	0.495	1.041	1.002–1.081	0.040
Uric acid	1.001	0.999–1.004	0.282	1.001	0.998–1.004	0.443	1.003	0.997–1.009	0.404
SBP	1.020	1.004–1.035	0.012	1.016	0.997–1.034	0.098	1.042	1.010–1.074	0.010
DBP	0.987	0.967–1.007	0.205	0.996	0.971–1.022	0.757	0.961	0.925–0.999	0.045

CAD: coronary artery disease; OR: odds ratios; 95%CI: 95% confidence intervals; SBP: systolic blood pressure; DBP: diastolic blood pressure; HDL: High-density lipoprotein.

LD analysis

Table 6 shows patterns of LD analysis in the CYP17A1 gene. For all five SNPs the r² values were below 0.5; for one haplotype block |D’| values were beyond 0.25. Because the |D’| for SNP2–SNP3 was <0.25, this meant that SNP2 and SNP3 could not be used to construct haplotypes. We constructed haplotypes using SNP1, SNP4, and SNP5.

Table 6.

Pairwise linkage disequilibrium for five SNPs.

\|D’\| values
r² values		SNP1	SNP2	SNP3	SNP4	SNP5
	SNP1		0.290	0.790	1.000	0.867
	SNP2	0.019		0.230	0.269	0.324
	SNP3	0.296	0.026		0.466	0.346
	SNP4	0.284	0.057	0.130		0.865
	SNP5	0.051	0.031	0.018	0.186

|D’| above the diagonal and r² below the diagonal.

Haplotype analyses

As shown in Table 7, in the haplotype-based case-control analysis, haplotypes were established through the use of different combinations of the three SNPs. For total subjects, the overall distribution of the haplotypes established by SNP1–SNP4 and SNP1–SNP4–SNP5 were significantly different between the CAD patients and the control subjects (p=0.006, and p=0.030, respectively). In total, frequencies of the C–T haplotype established by SNP1–SNP4, the C–T haplotype established by SNP1–SNP5, and the C–T–T haplotype established by SNP1–SNP4–SNP5 were significantly higher for the patients with CAD as compared with the control subjects (p=0.007, p=0.024, and p=0.013, respectively). The frequency of the A–T haplotype established by SNP1–SNP4, and the A–T–T haplotype established by SNP1–SNP4–SNP5 was significantly higher for the control subjects as compared with the CAD patients (p=0.017, and p=0.046). For men, the overall distribution of the haplotypes established by SNP1–SNP4 was significantly different between the subjects with CAD and the controls (p=0.026). In men, the frequency of the A–T haplotype established by SNP1–SNP4, and the A–T–T haplotype established by SNP1–SNP4–SNP5 was significantly lower for the control subjects as compared with the CAD patients (p=0.022 and p=0.039). The frequency of the C–T haplotype established by SNP1–SNP4 was significantly higher for CAD patients than for controls (p=0.043). For women, the overall distribution of haplotypes was not significantly different between the CAD patients and the control subjects.

Table 7.

Haplotype analysis in patients with CAD and in control subjects.

Haplotype	No. 1	No. 2	No. 3	Overall p-value			Frequency in total			Frequency in man			Frequency in woman
Haplotype	No. 1	No. 2	No. 3	Total	Man	Woman	CAD	Control	p-value	CAD	Control	p-value	CAD	Control	p-value
	SNP1	SNP4		0.006	0.026	0.155
H1	A	C					0.557	0.557	0.997	0.558	0.540	0.638	0.553	0.586	0.546
H2	A	T					0.227	0.295	0.017	0.231	0.313	0.022	0.217	0.265	0.318
H3	C	T					0.217	0.148	0.007	0.211	0.147	0.043	0.230	0.148	0.063
	SNP1		SNP5	0.057	0.140	0.251
H1	A		C				0.206	0.246	0.182	0.195	0.244	0.157	0.242	0.250	0.951
H2	A		T				0.577	0.604	0.533	0.596	0.606	0.895	0.517	0.600	0.229
H3	C		T				0.206	0.150	0.024	0.201	0.150	0.090	0.224	0.150	0.106
		SNP4	SNP5	0.701	0.378	0.455
H1		C	C				0.209	0.230	0.424	0.225	0.197	0.362	0.246	0.239	0.883
H2		C	T				0.341	0.321	0.584	0.308	0.363	0.187	0.272	0.342	0.228
H3		T	T				0.440	0.431	0.882	0.446	0.431	0.610	0.469	0.408	0.315
	SNP1	SNP4	SNP5	0.030	0.052	0.237
H1	A	C	C				0.206	0.232	0.344	0.193	0.226	0.293	0.247	0.238	0.871
H2	A	C	T				0.344	0.321	0.500	0.367	0.310	0.160	0.272	0.343	0.218
H3	A	T	T				0.229	0.287	0.046	0.226	0.297	0.039	0.241	0.269	0.610
H4	C	T	T				0.210	0.145	0.013	0.204	0.149	0.085	0.228	0.138	0.057

CAD: coronary artery disease; SNP: single nucleotide polymorphism.

Haplotypes with frequencies >0.03 were estimated using SHEsis software; the p-value was calculated by permutation test using the bootstrap method.

Discussion

An association with CYP genetic polymorphism has been established in numerous diseases such as hypertension, CAD, heart failure, stroke, cardiomyopathy and arrhythmias.⁹ In our study, we found the variation in CYP17A1 gene is associated with CAD in a Chinese Uygur population, and after multivariate adjustment the associations between CYP17A1 gene polymorphisms and CAD are still maintained. The etiology and pathogenesis of CAD include disorders of lipoprotein metabolism,²⁷ hypertension,²⁸ insulin resistance or diabetes,²⁹ the impairment and inflammation of vascular endothelium,^30,31 and the disturbance of blood coagulation and the fibrinolytic system.³²

As previously mentioned, CYP17A1 has two kinds of enzyme activity: 17alpha-hydroxylase and 17,20-lyase activities convert 17-alpha OH pregnenolone to dehydroepiandrosterone, which is a precursor to sex hormones. P450c17 protein is an important enzyme that catalyzes the formation of all endogenous androgens. Therefore, CYP17A1 genetic mutations potentially reduce androgen biosynthesis. In the clinical setting, countering androgen biosynthesis by inhibition of CYP17A1 is a rational approach for therapeutic intervention in castration-resistant prostate cancer.³³

In recent years, many clinical studies have shown that testosterone levels play an important role in the progress of CAD among elderly men,³⁴ whereas lower testosterone levels promote CAD.³⁵ In addition, some evidence has indicated that testosterone is also related to CAD risk factors, for example, disorders of lipoprotein metabolism is also one of the primary mechanisms of CAD,²⁷ and research shows a positive correlation between the levels of testosterone and HDL-C and a negative correlation between the levels of testosterone and TG, TC, LDL-C and VLDL. Malkin et al. confirmed that the low levels of testosterone can lead to lipid disorders, and supplementing testosterone can correct dyslipidemia.³⁶ Insulin resistance and diabetes are significant independent risk factors for CAD.²⁹ Selvin et al. indicated that there is a relationship between diabetes and low free or low bioactive testosterone levels.³⁷ The dysfunction of vascular endothelium is a key mechanism in the occurrence and development of CAD,^30,31 the levels of physiological testosterone can promote endothelial cells to release nitric oxide through improved vascular endothelial function, and low levels of testosterone can decrease vascular endothelial function and promote the occurrence of CAD.³⁸ Blood coagulation and the fibrinolytic system is an important mechanism in CAD;³² physiological levels of testosterone can improve the function of endothelial cells³⁹ promote expression of tissue-plasminogen activator and inhibit the secretion of plasminogen activator inhibitor; and low levels of testosterone can increase the levels of proteins related to clotting factor VIII, cause endothelial dysfunction and vascular inflammatory reaction, and increase the incidence of CAD. Normal estrogen signaling is also dependent on CYP17A1. Estrogen plays a very important role in many physiological and pathological process, such as mediation of vasoconstriction, vascular endothelium repair, lipid metabolism, glucose metabolism and insulin-related signal transduction pathways, etc., which directly or indirectly affect the function of cardiovascular system. In women, estrogen seems to be protective for the cardiovascular system.^20-22 Wellons et al. thought early menopause is associated with an increased risk of CAD,⁴⁰ and high levels of endogenous estrogen may explain the low prevalence of CAD in premenopausal women.⁴¹

In our study, we found that polymorphisms of the CYP17A1 gene were associated with CAD in a Uygur population. For rs4919686, in total, the frequency of the C allele is higher in CAD patients than in control subjects; there were significant differences in genotypes, dominant model, and recessive model. After multivariate adjustment of confounding factors such as plasma concentration of TG, UA, HDL, Glu, Cr, incidence of hypertension, diabetes, drinking, and smoking, the significant difference was retained. This indicated that the C allele is a risk factor for CAD. For rs1004467, compared with men, in total subjects and women, the frequency of C allele was higher in control subjects than in CAD patients, and there were significant differences in genotypes, dominant model, and recessive model. The significant difference was retained after multivariate adjustment of confounding factors for CAD, and there was no difference for the C allele between CAD patients and control subjects in men. This indicated that the C allele might be protecting against for CAD in total subjects and in women. For rs4919687, compared with women, in total subjects and men, the genotypes, dominant model, and recessive model were significantly higher in CAD patients than in control subjects. The significant difference was retained after multivariate adjustment of confounding factors for CAD, in total; the frequency of the A allele is higher in CAD patients than in control subjects, there is no difference for the A allele between CAD patients and control subjects in women. This indicated that the A allele is a risk factor for CAD in total subjects, and the AA genotype of rs4919687 could be a risk genetic marker of CAD in total subjects and in men. In addition, based on these findings, we hypothesized that haplotype analysis would be useful for the assessment of associations between haplotypes and CAD. For total subjects and men, we succeeded in identifying two susceptible haplotypes, the C-T of SNP1-SNP4 as well as the C-T-T of SNP1-SNP4-SNP5, and two protective haplotypes, the A-T of SNP1-SNP4 as well as the A-T-T of SNP1-SNP4-SNP5. These haplotype analysis results are consistent with the genotype analysis results of SNP1 (rs4919686), that the C allele confers risk and the A allele is protective. For the total subjects we discovered a susceptible haplotype, the C-T of SNP1-SNP5; however, the overall distribution of this haplotype was not significantly different between the CAD patients and the control subjects (p=0.057).

In conclusion, this is first study to investigate the differences between the human CYP17A1 gene and CAD in the Chinese Uygur population, and is the first haplotype-based case-control study to correlate its association with CAD. The C allele of rs4919686 in the CYP17A1 gene could be a genetic marker for susceptibility. The C allele of rs1004467 could be a protective genetic marker, and the T allele could be a genetic marker for risk of CAD in total subjects and women. The A allele of rs4919687 may be a genetic risk marker and the G allele may be a genetic protective marker of CAD in total subjects. The AA genotype of rs4919687 could be a genetic risk marker of CAD in total subjects and men. Additional studies will need to be undertaken in order to isolate the functional mutations which associate polymorphisms of the CYP17A1 gene with CAD.

Limitation of the study

The present study was limited by the relatively small sample size. This may have led to weak statistical significance and wide CIs when estimating OR.

Footnotes

Conflict of interest

None declared.

Funding

This study was supported by National Natural Science Foundation of China (81160017 and 81470014) and Xinjiang Science and Technology Projects (201491181).

References

Roger

Lloyd-Jones

. Heart disease and stroke statistics—2011 update: A report from the American Heart Association. Circulation 2011; 123: e18–e209.

. Major causes of death among men and women in China. N Engl J Med 2005; 353: 1124–1134.

Manace

Godiwala

Babyatsky

MW.

Genomics of cardiovascular disease. Mt Sinai J Med 2009; 76: 613–623.

Zee

Cheng

Erlich

. Genetic variants of arachidonate 5-lipoxygenase-activating protein, and risk of incident myocardial infarction and ischemic stroke: A nested case-control approach. Stroke 2006; 37: 2007–2011.

Roberts

Stewart

AF.

Genes and coronary artery disease: Where are we?

J Am Coll Cardiol 2012; 60: 1715–1721.

O’Donnell

Kavousi

Smith

. Genome-wide association study for coronary artery calcification with follow-up in myocardial infarction. J Circulation 2012; 124: 2855–2864.

Schunkert

König

Kathiresan

. Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease. J Nat Genet 2011; 43: 333–338.

Zordoky

El-Kadi

AO.

Effect of cytochrome P450 polymorphism on arachidonic acid metabolism and their impact on cardiovascular diseases. Pharmacol Ther 2010; 125:446–463.

Elbekai

El-Kadi

AO.

Cytochrome P450 enzymes: Central players in cardiovascular health and disease. Pharmacol Ther 2006; 112: 564–587.

10.

Sibbing

Stegherr

Latz

. Cytochrome P450 2C19 loss-of-function polymorphism and stent thrombosis following percutaneous coronary intervention. Eur Heart J 2009; 30: 916–922.

11.

Harmsze

Van Werkum

Ten Berg

. CYP2C19*2 and CYP2C9*3 alleles are associated with stent thrombosis: A case–control study. Eur Heart J 2010; 31: 3046–3053.

12.

Nakayama

Soma

Rehemudula

. Association of 5’ upstream promoter region of prostacyclin synthase gene variant with cerebral infarction. Am J Hypertens 2000; 13: 1263–1267.

13.

Wang

Greco

Sim

. Effect of CYP1A1 MspI polymorphism on cigarette smoking related coronary artery disease and diabetes. Atherosclerosis 2002; 162: 391–397.

14.

Cornelis

El-Sohemy

Campos

Genetic polymorphism of CYP1A2 increases the risk of myocardial infarction. J Med Genet 2004; 41: 758–762.

15.

Yasar

Bennet

Eliasson

. Allelic variants of cytochromes P450 2C modify the risk for acute myocardial infarction. Pharmacogenetics 2003; 13: 715–720.

16.

Spiecker

Darius

Hankeln

. Risk of coronary artery disease associated with polymorphism of the cytochrome P450 epoxygenase CYP2J2. Circulation 2004; 110: 2132–2136.

17.

Shi

Qiu

. A functional polymorphism in the CYP3A4 gene is associated with increased risk of coronary heart disease in the Chinese Han population. Basic Clin Pharmacol Toxicol 2011; 108: 208–213.

18.

Letonja

Peterlin

Bregar

. Are the T/C polymorphism of the CYP17 gene and the tetranucleotide repeat (TTTA) polymorphism of the CYP19 gene genetic markers for premature coronary artery disease in Caucasians? Folia Biol (Praha) 2005; 51: 76–81.

19.

Fan

Raitakari

Kiiah6nen

. Hepatic lipase promoter C-480T polymorphism is associated with serum lipids levels, but not subclinical atherosclerosis: the Cardiovascular Risk in Young Finns Study. Clin Genet 2009; 76: 46–53.

20.

Haschke-Becher

Kirchheiner

Trummer

. Impact of CYP2C8 and 2C9 polymorphisms on coronary artery disease and myocardial infarction in the LURIC cohort. Pharmacogenomics 2010; 11: 1359–1365.

21.

Kondo

Hirose

Kageyama

Roles of oxidative stress and redox regulation in atherosclerosis. J Atheroscler Thromb 2009; 16: 532–538.

22.

Kaur-Knudsen

Bojesen

Nordestgaard

BG.

Common polymorphisms in CYP2C9, sub-clinical atherosclerosis and risk of ischemic vascular disease in 52000 individuals. Pharmacogenomics J 2009; 9: 327–332.

23.

Mehta

NN.

Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease. Circ Cardiovasc Genet 2011; 4: 327–329.

24.

Xie

. Haplotype analysis of the CYP8A1 gene associated with myocardial infarction. Clin Appl Thromb-Hem 2009; 15: 574–580.

25.

Xie

. Association of polymorphisms of PTGS2 and CYP8A1 with myocardial infarction. Clin Chem Lab Med 2009; 47: 347–352.

26.

Yang

Wang

. Association of interaction between smoking and CYP 2C19*3 polymorphism with coronary artery disease in a Uighur population. Clin Appl Thromb Hemost 2010; 16: 579–583.

27.

Arsenault

Lemieux

Despre’s

. The hypertriglyceridemic-waist phenotype and the risk of coronary artery disease: Results from the EPIC-Norfolk prospective population study. CMAJ 2010; 182: 1427–1432.

28.

Nakamura

Saitoh

Takagi

. Impact of abnormal glucose tolerance, hypertension and other risk factors on coronary artery disease. Circ J 2007; 71: 20–25.

29.

Raz

Relationship between blood glucose control and improved cardiovascular outcome after stent implantation in diabetic patients. Cardiology 2010; 116: 48–50.

30.

Kitta

Obata

Nakamura

. Persistent impairment of endothelial vasomotor function has a negative impact on outcome in patients with coronary artery disease. J Am Coll Cardiol 2009; 53: 323–330.

31.

Farmer

Torre-Amione

Atherosclerosis and inflammation. Curr Atheroscler Rep 2002; 4: 92–98.

32.

Folsom

Aleksic

Park

. Prospective study of fibrinolytic factors and incident coronary heart disease: The Atherosclerosis Risk in Communities (ARIC) Study. Arterioscler Thromb Vasc Biol 2001; 21: 611–617.

33.

Yap

Carden

Attard

. Targeting CYP17: Established and novel approaches in prostate cancer. Curr Opin Pharmacol 2008; 8: 449–457.

34.

Naessen

Sjogren

Bergquist

. Endogenous steroids measured by high-specificity liquid chromatography-tandem mass spectrometry and prevalent cardiovascular disease in 70-year-old men and women. J Clin Endocrinol Metab 2010; 95: 1889–1897.

35.

Malkin

Pugh

Morris

. Low serum testosterone and increased mortality in men with coronary heart disease. Heart 2010; 96: 1821–1825.

36.

Malkin

Pugh

Jones

. The effect of testosterone replacement on endogenous inflammatory eytokines and lipid profiles in hypogonadal men. J Clin Endocrinol Metab 2004; 89: 3313–3318.

37.

Selvin

Feinleib

Zhang

. Androgens and diabetes in men results from the Third National Health and Nutrition Examination Survey (NHANES III). Diabetes Care 2007; 30: 234–238.

38.

Maturana

Breda

Lhullier

. Relationship between endogenous testosterone and cardiovascular risk in early postmenopausal women. Metabolism 2008; 57: 961–965.

39.

Jin

Lin

. Physiological testosterone stimulates tissue plasminogen activator and tissue factor pathway inhibitor and inhibits plasminogen activator inhibitor type l release in endothelial cells. Biochem Cell Bid 2007; 85: 246–251.

40.

Wellons

Ouyang

Schreiner

. Early menopause predicts future coronary heart disease and stroke: The Multi-Ethnic Study of Atherosclerosis. Menopause 2012; 19: 1081–1087.

41.

Mendelsohn

Karas

RH.

The protective effects of estrogen on the cardiovascular system. N EngI J Med 1999; 340: 1801–1811.