Significant Association between OPG/TNFRSF11B Variant and Common Complex Ischemic Stroke

Abstract

Background:

The serum level of osteoprotegerin (encoded by OPG or TNFRSF11B) was previously shown to be increased in patients with ischemic stroke. A single nucleotide polymorphism rs3134069 in the TNFRSF11B gene was previously associated with ischemic stroke in a population of diabetic patients in Italy. It remains to be determined whether rs3134069 is associated with ischemic stroke in the general population or populations without diabetes.

Materials and Methods:

We genotyped rs3134069 and performed a case-control association study to test whether rs3134069 is associated with ischemic stroke in 2 independent Chinese Han populations, including a China-Central population with 1629 cases and 1504 controls and a China-Northern population with 1206 cases and 720 controls.

Results:

rs3134069 showed significant association with ischemic stroke in the China-Central population (P = 9.24 × 10⁻³, odds ratio [OR] = 1.50). The association was replicated in the independent China-Northern population (P = 2.45 × 10⁻⁴, OR = 1.53). The association became more significant in the combined population (P = 7.09 × 10⁻⁶, OR = 1.41). The associations remained significant in the male population, female population, and population without type 2 diabetes. Our expression quantitative trait loci analysis found that the minor allele C of rs3134069 was significantly associated with a decreasedexpression level of TNFRSF11B (P = .002).

Conclusions:

This study demonstrates that rs3134069 in TNFRSF11B increases risk of ischemic stroke by decreasing TNFRSF11B expression.

Introduction

Stroke is a common cause of death worldwide,¹ and accounts for over 20% of total deaths in China.² Ischemia is responsible for about 87% of stroke cases.³ Atherosclerosis in the brain is the major cause for formation of thrombi, and is estimated to cause 70%−80% of ischemic strokes.⁴ Ischemic stroke is known to be affected by genetic risk factors, environmental factors, and their interactions. Genome-wide association studies have identified some loci conferring risk to ischemic stroke, including 12p13 (NINJ2), 12q24 (ALDH2), 7p21 (HDAC9), 9p21, 4q25 (PITX2), 16q22 (ZFHX3), 9q34 (ABO), and 1p13.2 (TSPAN2).^5–10 Candidate gene approaches also identified ALOX5AP, NOS3, PCSK9, PDE4D, SGK1, VKORC1, and other genes confer risk to ischemic stroke.^11–16 However, all these genomic variants in aggregate explain only a small proportion of heritability of ischemic stroke, a phenomenon referring to as “missing heritability”.¹⁷ Therefore, many more genomic variants associated with ischemic stroke need to be identified to fully elucidate the genetic architecture of this important disease.

Osteoprotegerin (OPG), which belongs to the tumor necrosis factor receptor superfamily, is encoded by the TNFRSF11B gene on chromosome 8q23.¹⁸ Several studies have demonstrated that OPG could promote endothelial cell survival.^19–21 Moreover, several studies have shown a correlation between increased endogenous serum OPG levels and the presence and severity of clinical coronary artery disease (CAD),^22,23 stroke,^24–26 cardiovascular morbidity and mortalit²⁷ as well as the progression of atherosclerosis in humans.^28,29 Further evidence suggests that circulating OPG levels also are associated with the extent of vascular calcification.^30–32 Moreover, increased OPG serum levels were found in Ldlr^−/− knockout (KO) mice on a Western diet.³³ Recently, a variant in the TNFRSF11B gene was found to be associated with coronary atherosclerosis in patients with rheumatoid arthritis.³⁴ Moreover, Biscetti et al reported that in a small population of 364 diabetic patients with a history of ischemic stroke and 492 diabetic subjects without history of ischemic stroke, single-nucleotide polymorphism (SNP) rs3134069 showed a significant genotypic association with ischemic stroke under the background of diabetes.³⁴ However, it is unknown whether SNP rs3134069 is associated with ischemic stroke in subjects without type 2 diabetes (T2D). Khouloud et al found that PPARγ variant C161T was associated with risk to ischemic stroke in populations with T2D, but not in populations without T2D.³⁴

Moreover, only 15% of patients with ischemic stroke have a history of T2D and remaining 85% do not have a history of T2D.³⁵ Therefore, it is necessary to determine whether TNFRSF11B variants are associated with ischemic stroke in a more complex population, especially in populations without T2D.

In this study, we analyzed the association between TNFRSF11B SNP rs3134069 and ischemic stroke in 2 independent case-control ischemic stroke populations with a total of 5059 Chinese Han subjects. Moreover, we studied whether rs3134069 was a functional variant associated with the expression level of TNFRSF11B mRNA by eQTL analysis.

Materials and Methods

Study Subjects

Samples enrolled in this study were selected from the GeneID database, which is an ongoing study in the Chinese Han population and has collected more than 80,000 DNA samples and also all available clinical data. The goal of the GeneID database is to identify susceptibility genes or other risk factors of cardiovascular and cerebrovascular diseases in the Chinese Han population.^36–43

In this study, we enrolled a case-control population with a total of 5059 samples, including 2835 patients with ischemic stroke and 2224 comparable controls. The study contained 2 independent populations matched by geographical areas they were enrolled. The China-Central population consisted of 1629 patients with ischemic stroke and 1504 controls, and the cases enrolled were from the patients who were under treatment of ischemic stroke in the hospitals of Wuhan city, whereas the controls of the China-Central population were enrolled from people who subjected physical examinations from the same hospitals of Wuhan city. The China-Northern population included 1206 patients with ischemic stroke enrolled from the patients who were under treatment of ischemic stroke in the hospitals of Beijing, and 720 controls enrolled from other patients who showed no sign of ischemic stroke by related medical examinations in the hospitals of Beijing. All subjects were self-reported to be of Chinese Han origin.

The diagnosis of ischemic stroke was made based on the standard World Health Organization criteria.⁴⁴ The clinical diagnosis was made carefully by at least 2 independent neurologists based on a medical history of stroke, stroke signs by neurological examinations, and cerebral ischemia by computed tomography (CT) or magnetic resonance imaging (MRI) images. In this study, we selected patients with ischemic stroke only, and excluded those patients with subarachnoid hemorrhages, embolic brain infarction, brain tumors, and those with a relevant brain stem or subcritical hemispheric lesions with a diameter of greater than 1.5 cm based on the data from CT or MRI examinations. We selected controls as those which were diagnosed to be stroke-free, had no medical history of stroke, and showed normal brain CT or MRI examinations in the China-Northern population, or common health individuals in the China-Central population.

Other basic demographical and clinical characteristics such as the age, sex, smoking history, hypertension, and T2D were also obtained from medical records or the data from physical examinations. The diagnostic criteria for hypertension are a systolic blood pressure of 140 mm Hg or higher or a diastolic blood pressure of 90 mm or higher. The diagnostic criteria for T2D are a fasting plasma glucose concentration of 126 mg/dL or higher after at least 8 hours of fasting or a 2-hour plasma glucose level of 200 mg/dL or higher during an oral glucose tolerance test.

This study was approved by the Ethics Committees on human subject research of Huazhong University of Science and Technology and local institutions. Written informed consent was obtained from all subjects. This study followed the guidelines by the Declaration of Helsinki.

SNP Genotyping

The Syto 9 fluorescent dye-based high resolution melt (HRM) method was used to genotype SNP rs3134069 on a Rotor-gene 6200 System (Corbett Life Science, Mortlake, Australia) as described by us.^{13,38,39,45–47} In brief, a total volume of 25 μl of polymerase chain reaction (PCR) reaction mixture contained 2.5 μl of 10 × PCR buffer (containing 15 mM MgCl₂), .5 μl of dNTPs (10 mM), .5 μl of each primer (10 μM), 1 μl of template DNA (25 ng/μl), 5 μmol/L of SYTO 9 fluorescent dye, .15 U of Taq DNA polymerase (TIANGEN, Beijing, China), and ddH2O. The forward PCR primer is 5’- CCACCATCATCAAAGGGCTAT −3’ and the reverse primer is 5’-CAGGGAATTAATGGGGGAGAC −3’. The PCR was performed on an ABI 9700 PCR System (Life Technologies, Gaithersburg, MD) with a profile of 95°C for 3 minutes, 40 cycles of 95°C for 10 seconds, 59°C for 10 seconds and 72°C for 15 seconds, and a final step of 72°C for 10 minutes. PCR products were analyzed and the genotype of each individual was analyzed using the HRM program. During each round of HRM analysis, 3 DNA samples with known genotypes of CC, AC, and AA were included as positive controls, and 1 negative control of ddH₂O without genomic DNA was also included. The total success rate of HRM genotyping was 96.3%.

Direct Sanger sequencing was used to validate the accuracy of HRM genotyping data, and the details were described in the Supplementary Material.

Real-Time Quantitative Reverse Transcription PCR Analysis

The ^ΔΔCq method was used to determine the mRNA expression level of the TNFRSF11B gene among study subjects with different genotypes of SNP rs3134069. Quantitative real-time RT-PCR analysis was carried out according to the standard Minimum Information for Publication of Quantitative Real-Time PCR Experiments guidelines,⁴⁸ and as described previously by us.^49,50 The details were described in the Supplementary Material.

Statistical Analysis

The Hardy-Weinberg disequilibrium test was performed in each control population using the PLINK program (1.07 version).⁵¹

The case-control association analysis was performed as described previously.^13,40,52 Chi-square (χ²) tests were performed using Pearson’s 2 × 2 and 2 × 3 contingency tables to evaluate the association between allelic frequencies or genotypic frequencies and ischemic stroke using SPSS v17.0 (IBM Corporation, Armonk, NY). Multiple logistic regression analysis was used to adjust covariates, including sex, age, hypertension, smoking history, and T2D using SPSS v17.0 (IBM Corporation). Statistical power analysis of study populations was conducted using program PS (Power and Sample size Calculations, version 3.0.43).⁵³

For statistical analysis of real-time RT-PCR data of the TNFRSF11B expression level, a Student’s t-test was used to compare the differences for the mean RQ values between different genotypes of SNP rs3134069 using SPSS v17.0.

Results

Significant Allelic Association between TNFRSF11B SNP rs3134069 and Ischemic Stroke

We performed a case-control association study to test whether SNP rs3134069 in the TNFRSF11B gene was associated with ischemic stroke. First, we genotyped SNP rs3134069 in a Central China population (China-Central) and analyzed the potential association between rs3134069 with ischemic stroke. The significant association was then replicated in a Northern China population (China-Northern). Power analysis indicates that the 2 independent populations in this study can provide enough statistics power to test the association. Under the assumptions of the type I error of .05, a minor allele frequency of .13 for SNP rs3134069 in the Chinese Han population (HapMap data), and with an odds ratio (OR) of 1.3, the power of China-Central population was 95% and 80% for the China-Northern population.

The China-Central population consisted of 1629 patients with ischemic stroke and 1504 controls (Table 1). SNP rs3134069 was genotyped in the China-Central population. The genotyping data of the controls in the China-Central population did not deviate from a Hardy-Weinberg equilibrium tests (P = .76). In the China-Central population, the frequency of the minor allele C in cases is significant higher in controls (.11 in cases versus .09 in controls). The minor allele C of rs3134069 was significantly associated with an increased risk of ischemic stroke (observed P_obs=4.26 × 10⁻³ with an OR = 1.27) (Table 2). The data indicate that SNP rs3134069 is a significant genetic risk variant for ischemic stroke. The association remained significant with an OR of 1.50 (P_adj=9.24 × 10⁻³) after adjusting for covariates, including sex, age, hypertension, smoking history, and T2D (Table 2).

Table 1.

Clinical and demographical characteristic of study subjects

Item	China-Central population		China-Northern population		Combined population
	Ischemic stroke	Control	Ischemic stroke	Control	Ischemic stroke	Control
	(n=1629)	(n=1504)	(n=1206)	(n = 720)	(n = 2835)	(n = 2224)
Male, n (%)	929 (57.03)	994 (66.09)	671 (55.64)	469 (65.14)	1600 (56.44)	1463 (65.78)
Age (y)*	65.88 ± 14.04	45.30 ± 10.09	60.50 ±10.76	55.73 + 8.93	63.67 ± 13.07	48.15 ±10.83
Hypertension, n (%)	395 (24.25)	14 (.93)	366 (30.35)	152 (21.11)	761 (26.84)	166 (7.46)
Smoking, n (%)	327 (20.07)	23 (1.53)	203 (16.83)	64 (8.89)	530 (18.69)	87 (3.91)
T2D, n %	1181 (72.50)	3 (.20)	106 (8.79)	40 (5.56)	1287 (45.40)	43 (1.93)

Abbreviation: T2D, type 2 diabetes.

^*Age at the first diagnosis of the disease for CAD cases and age at enrollment for CAD controls.

Table 2.

Allelic association of SNP rs3134069 with stroke in the Chinese Han population^*

Study population	Sample size	Phwe	Without adjustment				Adjustment
	Case/Control		RA	Frequency	Pobs	OR (95%CI)	Padj	OR (95%CI)
China-Central	1629/1504	.76	C	.11/. 09	4.26 × 10−3	1.27 (1.08–1.50)	9.24 × 10−3	1.50 (1.11–2.05)
China-Northern	1206/720	1.00	C	.13/.10	5.60 × 10−3	1.35 (1.09–1.67)	2.45 × 10−4	1.53 (1.22–1.92)
Combined	2835/2224	.70	C	.12/.09	3.51 × 10−5	1.31 (1.15–1.50)	7.09 × 10−6	1.41 (1.21–1.64)
Female	1170/750	.68	C	.11/.08	.002	1.44 (1.15–1.81)	.001	1.43 (1.20–. 1.65)
Male	1665/1474	.75	C	.12/.10	.002	1.28 (1.09–1.50)	.003	1.30 (1.10–1.52)

Abbreviations: CI, confidential interval; OR, odds ratio; P_adj, P value for association after adjusting for covariates of sex, age, and hypertension by multiple logistic regression analysis using SPSS v17.0; P_hwe, P value for Hardy–Weinberg equilibrium (HWE) tests using PLINK version 1.07; P_obs, P value for association before adjusting for covariates by 2 × 2 contingence tables using PLINK version 1.07.

^*Frequency, frequency of the risk allele (RA) in the case group and control group.

We enrolled another population from northern China (China-Northern) to verify the significant association between SNP rs3134069 and ischemic stroke. The China-Northern population included 1206 patients with ischemic stroke and 720 controls (Table 1). The genotyping data in the China-Northern population did not deviate from HWE test (P = 1.00). The frequency of the minor allele C in the patient group was also significantly higher than in the control group (.13 in cases versus .10 in controls). Significant association was identified between SNP rs3134069 and ischemic stroke in the China-Northern population with a P_obs value of 5.60 × 10⁻³ and an OR of 1.35 (Table 2). After adjusting for covariates of sex, age, hypertension, smoking history, and T2D, the association remained significant (P_adj=2.45 × 10⁻⁴, OR = 1.53) (Table 2).

We combined the 2 populations and generated a larger population for further association analysis. In the combined population, the association between SNP rs3134069 and ischemic stroke became more significant (P_obs=3.51 × 10⁻⁵, OR = 1.31; P_adj=7.09 × 10⁻⁶, OR = 1.41). Together, the data from 2 independent populations strongly suggest that the minor allele C of SNP rs3134069 confers a significant risk of ischemic stroke.

We divided the study subjects by sex, and found that the allelic association between SNP rs3134069 and ischemic stroke was significant in both the female population (P_obs=.002, OR = 1.44; P_adj=.001, OR = 1.43) and in the male population (P_obs=.002, OR = 1.28; P_adj=.003, OR = 1.30) (Table 2). The ORs were higher in the female population than in the male population, however, the differences were not significant by a Breslow-Day test (P > .05).

As described above, the association between rs3134069 and ischemic stroke remained significant after adjusting T2D and other covariates in the China-Central population (Table 2). However, the rate of T2D in cases was markedly higher than that in controls (Table 1), and may still confound the association analysis. Therefore, to further assess whether T2D affects the association between rs3134069 and ischemic stroke, we excluded the subjects with T2D from both cases and controls and reanalyzed the association. After excluding subjects with T2D, the frequency of the minor allele C in cases was significantly higher than that in controls (.12 in 448 cases versus .09 in 1501 controls; P_obs=.03, OR = 1.31; P_adj=.04, OR = 1.32) (Table 3). The results were validated in the China-Northern population (1100 cases versus 680 controls, P_obs=.03, OR = 1.28; P_adj=.04, OR = 1.28) and in the combined population (1548 cases and 2181 controls, P_obs=6.76 × 10⁻⁵, OR = 1.31; P_adj,=6.76 × 10⁻⁴, OR = 1.30) (Table 3). These data suggest that SNP rs3134069 is a risk factor for ischemic stroke independent of T2D.

Table 3.

Allelic association of SNP rs3134069 with stroke in the population without type 2 diabetes

Study population	Sample size		Without adjustment				Adjustment
	Case/Control	Phwe	RA	Frequency*	Pobs	OR (95%CI)	Padj	OR (95%CI)
China-Central	448/1501	.88	C	.12/.09	.03	1.31 (1.03–1.66)	.04	1.32 (1.03–1.67)
China-Northern	1100/680	.83	C	.13/.10	.03	1.28 (1.03–1.58)	.04	1.28 (1.03–1.56)
Combined	1548/2181	.86	C	.12/.09	6.76 × 10−5	1.31 (1.16–1.56)	1.17 × 10−4	1.30(1.16–1.55)

Abbreviations: CI, confidential interval; OR, odds ratio; P_hwe, P value for Hardy-Weinberg equilibrium (HWE) tests using PLINK version 1.07; P_adj, P value for association after adjusting for covariates of sex, age, and hypertension by multiple logistic regression analysis using SPSS v17.0; P_obs, P value for association before adjusting for covariates by 2 × 2 contingence tables using PLINK version 1.07.

^*Frequency, frequency of the risk allele (RA) in the case group and control group.

Significant Genotypic Association between SNP rs3134069 and Ischemic Stroke

We also analyzed the genotypic association under 3 common genetic models: additive, dominant, or recessive model (Table 4). The data showed that significant genotypic association was identified between SNP rs3134069 and ischemic stroke under an additive model or a dominant model, but not under a recessive model in the China-Central population before and after adjusting for covariates of sex, age, hypertension, smoking history, and T2D (P_obs=1.14 × 10⁻², P_adj=1.00 × 10⁻³ under an additive model; P_obs=1.13 × 10⁻², P_adj=1.20 × 10⁻² under a dominant model) (Table 4). The finding was confirmed in the China-Northern population (P_obs=1.84 × 10⁻², P_adj=1.54 × 10⁻⁴ under an additive model; P_obs=4.92 × 10⁻⁴, P_adj=2.04 × 10⁻⁴ under a dominant model) (Table 4). The association became more significant in the combined population (P_obs=1.76 × 10⁻⁴, P_adj=5.91 × 10⁻⁶ under an additive model; P_obs=8.50 × 10⁻⁶, P_adj=2.29 × 10⁻⁵ under a dominant model) (Table 4). When the combined population was divided into the male population and female population, the genotypic association between SNP rs3134069 and ischemic stroke was significant in both the female population (P_obs=.008, P_adj=.002 under an additive model; P_obs=.004, P_adj=.002 under a dominant model) or in the male population (P_obs=.008, P_adj=.003 under an additive model; P_obs=.003, P_adj=.004 under a dominant model) (Table 5).

Table 4.

Genotypic association of SNP rs3134069 with ischemic stroke under 3 different genetic models

Study panels	Model*	Genotypes (CC/CA/AA)		Without adjustment		Adjustment
		Cases	Controls	Pobs	OR (95%CI)	Padj	OR (95%CI)
China-Central	Additive	25/317/1287	11/251/1242	1.14 × 10−2	−	1.00 × 10−2	1.50 (1.10–2.04)
	Dominant	342/1287	262/1242	1.13 × 10−2	1.26 (1.05–1.51)	1.20 × 10−2	1.54 (1.10–2.16)
	Recessive	25/1604	11/1493	3.51 × 10−2	2.12 (1.04–4.31)	2.79 × 10−1	1.97 (.58–6.71)
China-North	Additive	15/272/919	6/126/588	1.84 × 10−2	−	1.54 × 10−4	1.57 (1.24–1.99)
	Dominant	287/919	132/588	4.92 × 10−3	1.39 (1.11–1.75)	2.04 × 10−4	1.61 (1.25–2.07)
	Recessive	15/1191	6/714	4.01 × 10−1	1.50 (.58–3.88)	1.47 × 10−1	2.11 (.77–5.80)
Combined	Additive	40/589/2206	17/377/1830	1.76 × 10−4	−	5.91 × 10‒6	1.42 (1.22–1.66)
	Dominant	629/2206	394/1830	8.50 × 10−5	1.32 (1.15–1.52)	2.29 × 10‒5	1.43 (1.21–1.68)
	Recessive	40/2795	17/2207	3.10 × 10−2	1.86 (1.05–3.29)	1.00 × 10‒2	2.38 (1.23–4.61)

Abbreviations: OR, odds ratio; CI, confidential interval; P_adj, P value for association after adjusting for covariates of sex, age, hypertension, and T2D by multiple logistic regression analysis using SPSS v17.0; P_obs, P value for association before adjusting for covariates by 2 × 2 contingence tables using PLINK version 1.07.

^*Additive model = CC/CA/AA; dominant model = CC+CA/AA; recessive model = CC/CA+AA.

Table 5.

Genotypic association of SNP rs3134069 with ischemic stroke under 3 different genetic models in both females and males of the combined population

Study panels	Model*	Genotypes (CC/CA/AA)		Without adjustment		Adjustment
		Cases	Controls	Pobs	OR (95%CI)	Padj	OR (95%CI)
Female	Additive	18/220/932	5/108/637	.008	−	.002	1.40 (1.20–1.65)
	Dominant	238/932	113/637	.004	1.44 (1.23–1.84)	.002	1.45 (1.25–1.77)
	Recessive	18/1152	5/745	.13	2.33 (.86–6.29)	.22	2.40 (.80–6.33)
Male	Additive	22/369/1274	12/269/1193	.008	−	.003	1.44 (1.23–1.74)
	Dominant	391/1274	281/1193	.003	1.30 (1.10–1.55)	.004	1.34 (1.10–1.65)
	Recessive	22/1463	12/1462	.22	1.63 (.85–3.31)	.18	1.88 (.88–3.51)

Abbreviations: CI, confidential interval; OR, odds ratio; P_adj, P value for association after adjusting for covariates of sex, age, hypertension and T2D by multiple logistic regression analysis using SPSS v17.0; P_obs, P value for association before adjusting for covariates by 2 × 2 contingence tables using PLINK version 1.07.

^*Additive model = CC/CA/AA; dominant model = CC+CA/AA; recessive model = CC/CA+AA.

C allele of SNP rs3134069 is Associated with the Decreased Expression Level of TNFRSF11B

SNP rs3134069 is located at a position of 245 bp upstream of the TNFRSF11B transcriptional start site. Based on its position, we hypothesized that SNP rs3134069 was a functional variant which may be associated with the expression level of TNFRSF11B. We measured the expression levels of TNFRSF11B from 89 people randomly selected from the general population undergoing physical examinations by real-time RT-PCR analysis and genotyping of these individuals for SNP rs3134069. The analysis showed that the mean relative expression level of TNFRSF11B in the 13 AC genotype carriers was significantly lower than that in the 76 AA genotype carriers (P = .002) (Fig 1). These results suggest that the minor allele C of SNP rs3134069 is associated with a decreased expression level of TNFRSF11B.

Figure 1.

Assessment of the relationship between SNP rs3134069 and the expression level of TNFSF11B mRNA by real-time RT-PCR analysis. Total RNA samples were isolated from 89 blood human samples (lymphocytes), converted into cDNA, and used for real time RT-PCR analysis. Genomic DNA samples were isolated from the 89 study subjects and genotyped for SNP rs3134069 by the HRM analysis. A Student’s t-test was used to compare the differences for the mean RQ values between different genotypes (AA and AC) of SNP rs3134069. Due to the low minor allele frequency, no homozygous CC genotype was found.

Discussion

In the present study, we identified a novel, highly significant association between SNP rs3134069 in the TNFRSF11B gene on chromosome 8q23 and ischemic stroke in the Chinese Han population. First, significant allelic association between SNP rs3134069 and ischemic stroke was found in a Central Chinese population from GeneID with 1629 ischemic stroke cases and 1504 controls before and after adjusting for covariates (P_adj=9.24 × 10⁻³) (Table 2). The minor allele C of SNP rs3134069 confers a risk of ischemic stroke (OR = 1.50). We also found significant genotypic association under an additive or dominant inheritance model (Table 4). Second, both the novel allelic association and genotypic association between SNP rs3134069 and ischemic stroke were verified in the China-Northern population (1206 patients with ischemic stroke and 720 controls) (Tables 2 and 4). Third, after combining the 2 populations, the P values for the association became more significant (P_adj=7.09 ×10⁻⁶, OR = 1.41 for allelic association; P_adj=5.91 ×10⁻⁶ under an additive model and P_adj=2.29 ×10⁻⁵ under a dominant model for genotypic association) (Tables 2 and 4). Finally, we analyzed the association between rs3134069 and ischemic stroke in population without T2D and the association remained significant. These data demonstrated significant association between TNFRSF11B SNP rs3134069 and risk of ischemic stroke, especially in population without T2D.

Previously, Biscetti et al found that in a small population of 364 patients with diabetes and a history of ischemic stroke and 492 diabetic subjects without history of ischemic, SNP rs3134069 showed a significant genotypic association with ischemic stroke under the background of diabetes (P < 10⁻⁴).⁵⁴ It was unknown whether SNP rs3134069 showed a significant allelic association with diabetes-associated ischemic stroke as no data were presented. Moreover, it is unknown whether the TNFRSF11B variant is associated with ischemic stroke in population without T2D. Here we show significant association between TNFRSF11B SNP rs3134069 and ischemic stroke in the general population and in the population without T2D. Therefore, our study significantly expands the spectrum of TNFRSF11B association with stroke from a diabetic population to the complex population.

Importantly, we demonstrated significant association between risk allele C of SNP rs3134069 and a decreased expression level of TNFRSF11B (Fig 1). The data suggest that rs3134069 is a functional variant for TNFRSF11B and that a decreased mRNA expression level of TNFRSF11B is a risk factor for the pathogenesis of ischemic stroke. This finding was surprising because several previous studies reported a positive association between increased serum OPG levels and ischemic stroke. In 2001, Browner et al reported that although the serum OPG level was not associated with the risk of incident thrombotic strokes, it was associated with the risk of fatal strokes.²⁶ Guldiken et al later showed that serum OPG levels were significantly higher in 51 patients with ischemic stroke than in 28 control subjects.²⁵ Ustundag et al showed that serum OPG levels were significantly higher in patients with cardioembolic and atherothrombotic strokes than in healthy controls or in patients with the transient ischemic attack.²⁴ Moreover, Jensen et al followed up 244 patients with acute ischemic stroke for 47 months and found that patients with lower serum OPG levels measured at admission had significantly improved survival.⁵⁵ Moreover, Morony et al demonstrated that a treatment with recombinant osteoprotegerin in Ldlr^−/−KO mice did not affect atherosclerotic lesion sizes or number.²⁴ More importantly, a study with TNFRSF11B^−/− KO mice clearly demonstrated that the TNFRSF11B deficiency accelerated advanced atherosclerotic lesion progression under the ApoE^−/− KO background.⁵⁵ Our data are consistent with the data from KO mice that decreased TNFRSF11B expression increases risk of atherosclerosis-related ischemic stroke, however, this needs to be confirmed in the future. Thus, it is likely that the increased OPG levels in serum samples from patients with stroke may represent an adaptive response of an organism to inflammation associated with atherosclerosis and stroke.

A limitation of the present study is that it enrolled patients with the large vessel subtype of stroke only. It will be interesting to study whether SNP rs3134069 is associated with other subtypes of stroke such as cardioembolic stroke, large vessel disease, small vessel disease, and cryptogenic stroke in the future.

In conclusion, the data in the present study demonstrate significant association between TNFRSF11B SNP rs3134069 and ischemic stroke in the complex population, especially in the population without T2D. We also showed that the risk allele C of SNP rs3134069 was associated with decreased TNFRSF11B expression. Our results suggest that TNFRSF11B is a susceptibility gene for common complex ischemic stroke and that the minor allele C of SNP rs3134069 increases risk of ischemic stroke by decreasing the expression level of TNFRSF11B.