Effect of Genetic Variants Associated With Plasma Homocysteine Levels on Stroke Risk
Background and Purpose—Elevated total plasma homocysteine (tHcy) levels are known to be associated with increased risk of ischemic stroke (IS). Given that both tHcy and IS are heritable traits, we investigated a potential genetic relationship between homocysteine levels and stroke risk by assessing 18 polymorphisms previously associated with tHcy levels for their association with IS and its subtypes.
Methods—Previous meta-analysis results from an international stroke collaborative network, METASTROKE, were used to assess association of the 18 tHcy-associated single-nucleotide polymorphisms (SNPs) in 12 389 IS cases and 62 004 controls. We also investigated the associations in regions located within 50 kb from the 18 tHcy-related SNPs and the association of a genetic risk score, including the 18 SNPs.
Results—One SNP located in the RASIP1 gene and a cluster of 3 SNPs located at and near SLC17A3 were significantly associated with IS (P<0.0003) after correcting for multiple testing. For stroke subtypes, the sentinel SNP located upstream of MUT was significantly associated with small-vessel disease (P=0.0022), whereas 1 SNP located in MTHFR was significantly associated with large-vessel disease (P=0.00019). A genetic risk score, including the 18 SNPs, did not show significant association with IS or its subtypes.
Conclusions—This study found several potential associations with IS and its subtypes: an association of an MUT variant with small-vessel disease, an MTHFR variant with large-vessel disease, and associations of RASIP1 and SLC17A3 variants with overall IS.
The relationship between total plasma homocysteine (tHcy) levels and stroke risk has been investigated by numerous observational studies, which together provide compelling evidence that elevated homocysteine levels are associated with an increased risk of ischemic stroke (IS).1 However, residual confounding and reverse causation impair causal inference from the results of observational studies.2 Mendelian randomization studies investigating a potential causal relationship between tHcy and IS risk have yielded inconclusive results.3
Recent studies showed that lowering homocysteine levels through vitamin B (folic acid and vitamin B12) intervention reduced the risk of stroke in patients with normal renal function and with normal vitamin B12 metabolism.4–6 These findings may explain the negative results obtained by earlier studies,7,8 where renal function and vitamin B12 metabolism status were not taken into account. In addition, the folic acid fortification of grains in the United States and some European countries may have also reduced the benefit of folic acid intervention trials.9 Thus, the relationship between stroke and homocysteine is complex, and careful consideration is required in designing future clinical trials.10
Recently, a genome-wide association study (GWAS) meta-analysis of plasma homocysteine levels in 44 147 individuals of European ancestry identified 13 associated genetic loci at genome-wide significance (P<5×10–8).11 Within the 13 associated loci, this meta-analysis identified 18 independent single-nucleotide polymorphisms (SNPs) explaining 6% of variation in tHcy levels.
Since both tHcy and IS have a genetic component, the reported association between tHcy and stroke risk may result from shared genetic risk factors. Thus, we investigated the 18 SNPs previously associated with homocysteine levels for their association with IS and its subtypes: large-vessel disease (LVD), small-vessel disease (SVD), and cardioembolic stroke. We also evaluated association of a genotype risk score (GRS), including the18 tHcy-associated SNPs for association with IS and its subtypes.
Materials and Methods
The study population included 12 389 IS cases and 62 004 controls of European ancestry from 15 cohorts contributing to the METASTROKE collaboration.12 Details of the study designs for each participating study are included in the online-only Data Supplement.
Plasma homocysteine-associated SNPs reaching genome-wide significance (P<5×10–8) in a published meta-analysis of GWAS11 were selected for inclusion. In total, 18 independent SNPs (r2<0.2) were found to be significantly associated with tHcy at a P<5×10–8 and selected for assessment in ischemic stroke and its subtypes (Table I in the online-only Data Supplement). As a secondary analysis, to account for potential population differences in linkage disequilibrium (LD) between functional and tag-SNPs, we included additional SNPs located ±50 kb around the 18 tHcy-associated SNPs, and a total of 3160 SNPs were selected of which 166 variants were independent (r2<0.2).
Summary statistics for association of the 18 tHcy-associated SNPs with IS and its subtypes were provided by METASTROKE Consortium. Details on genotyping, imputation, and quality control methods are provided in the Methods in the online-only Data Supplement. Logistic regression was performed to test association of individual SNPs with IS and its subtypes assuming an additive model and adjusting for study-specific covariates: age, sex, and ancestry principal components (Methods in the online-only Data Supplement).
Considering that stroke subtypes are independent of each other,12–16 we prespecified a Bonferroni-adjusted significance threshold of α=0.0027 (where α=0.05/18 SNPs) to adjust for primary analyses testing 18 independent SNPs for association with IS and its subtypes.
For the secondary analysis of SNPs located within ±50 kb of the 18 tHcy-related SNPs, we specified a Bonferroni-adjusted significance threshold of 0.0003 (α=0.05/166=0.0003, for 166 independent SNPs tested).
We then tested the association between an additive GRS of the 18 homocysteine-associated SNPs and increased risk of IS and its subtypes. The GRS was calculated using a previously described method17 (Methods in the online-only Data Supplement).
Using the CaTS (Power Calculator for Two Stage Association Studies) genetic power calculator,18 our study had 80% power to detect odds ratio (OR) of 1.06 to 1.11 for IS, and subtype-specific OR of 1.13 to 1.23 for variants with allele frequency 10% to 50% at a Bonferroni-corrected threshold of 0.0027 (α=0.05/18).
Study Population Characteristics
The discovery meta-analysis of IS studies included 12 389 cases and 62 004 controls of European descent. Stroke subtypes include cardioembolic, LVD, and SVD accounted for 19% (n=2365), 17.4% (n=2167), and 15.2% (n=1894), respectively, of all IS cases. Detailed characteristics of the participating studies have been summarized previously.12
Association of tHcy-Associated SNPs With Overall IS
First, we assessed the association of each tHcy-associated SNP with IS risk and subsequently with its subtypes. Then, we tested the combined effect of the 18 sentinel tHcy-associated SNPs on risk of IS and its subtypes.
For overall IS, the OR for the 18 tested SNPs ranged from 0.96 to 1.04 (Table II in the online-only Data Supplement). Of the 18 SNPs tested, 2 SNPs located at/near FUT1 (rs838133: OR, 1.04; 95% confidence interval [CI], 1.00–1.07; P=0.013) and CPS1 (rs7422339: OR, 0.96; 95% CI, 0.92–0.99; P=0.045) were associated with IS at a nominal P value (0.05), but did not pass Bonferroni-corrected P=0.0027 (P=0.05/18).
The combined GRS, including the 18 independent tHcy SNPs, did not show a significant association with IS (OR, 1.02; 95% CI, 0.91–1.15; P=0.63; Table III in the online-only Data Supplement).
Association of tHcy-Associated SNPs With IS Subtypes
Next, we investigated the association of the tHcy-related SNPs with IS subtypes. For SVD, 2 SNPs located near MUT (rs9369898: OR, 1.12; 95% CI, 1.04–1.21; P=0.0022) and FUT1 (rs838133: OR, 1.07; 95% CI, 1.00–1.15; P=0.04), respectively, were nominally associated at P<0.05 (Table IV in the online-only Data Supplement). The variant located near MUT (rs9369898) also passed Bonferroni-corrected value P=0.0027. The major allele A of rs9369898 associated with higher tHcy levels was also associated with increased risk of SVD. There was no evidence of between-study heterogeneity for rs9369898 (I2=7.4%; P-het=0.37). The GRS, including the 18 independent tHcy SNPs, did not show an association with SVD risk (OR, 1.1; 95% CI, 0.85–1.42; P=0.43; Table III in the online-only Data Supplement).
For LVD, 1 SNP (rs838133: OR, 1.08; 95% CI, 1.01–1.16; P=0.018) located near FUT1 gene was associated at a nominal P value (P<0.05) but did not pass Bonferroni correction for multiple testing (Table V in the online-only Data Supplement). The GRS of the 18 independent tHcy SNPs did not show an association with LVD risk (OR, 1.06; 95% CI, 0.82–1.35; P=0.64; Table III in the online-only Data Supplement). None of the tested SNPs were associated with cardioembolic risk even at a nominal P value (P<0.05; Table VI in the online-only Data Supplement). In addition, the GRS of the 18 independent tHcy SNPs did not show an association with cardioembolic risk (OR, 0.9; 95% CI, 0.71–1.14; P=0.4; Table III in the online-only Data Supplement).
Investigation of Associations in Regions Located ±50 kb Around the 18 tHcy-Associated SNPs
As a secondary analysis, we assessed the associations with IS and its subtypes for SNPs located within ±50 kb of the 18 tHcy-associated SNPs.
For overall IS, 3 variants (rs9379800, rs17271121, and rs12664474) located within 50 kb from the tHcy-associated polymorphism, rs548987, were associated with IS at P values lower than the Bonferroni-corrected threshold (P=0.05/166=0.0003; Table). Two SNPs (rs9379800 and rs12664474) located upstream of SLC17A3 were highly correlated (r2=0.766) with each other, and moderately correlated with the third SNP, rs17271121, located in an intron of SLC17A3 (r2 [rs9379800 and rs17271121]=0.306 and r2 [rs17271121 and rs12664474]=0.545). None of the 3 SNPs were in LD with the tHcy-associated polymorphism, rs548987 (r2<0.035).
In addition, another SNP rs2287921 located in an intron of RASIP1 gene, within 50 kb from the FUT1 polymorphism, rs838133, was associated with IS at a P=0.0002 (OR, 0.94; 95% CI, 0.91–0.97), lower than Bonferroni correction for multiple testing (P<0.0003). This SNP was in moderate LD (r2=0.658) with the sentinel SNP rs838133, which may suggest that this could be a broader risk region spanning the 2 neighboring genes RASIP1 and FUT1.
For LVD, 1 SNP (rs1801131), a missense variant located in MTHFR gene, near the 2 sentinel SNPs in this gene, rs1801133 and rs12134663, was associated with LVD with P=1.92×10–4 (OR, 1.15; 95% CI, 1.06–1.23) lower than the Bonferroni correction for multiple testing (P<0.0003). This missense SNP was in low LD with the 2 tHcy sentinel SNPs (r2 [rs1801131 and rs1801133]=0.19 and r2 [rs1801131 and rs12134663]=0.268).
For SVD, 1 SNP (rs566295) located upstream of MUT, 44 kb from the tHcy-associated polymorphism, rs9369898, was associated with SVD with P=2.2×10–4 (OR, 0.87; 95% CI, 0.80–0.93) passing Bonferroni correction for multiple testing. This SNP was in low LD with the MUT polymorphism, rs9369898 (r2=0.264).
For cardioembolic stroke, no significant associations were observed at a threshold exceeding Bonferroni correction for multiple testing (P<0.0003).
This large study of 12 389 IS cases and 62 004 controls has identified several potential novel associations with IS and its subtypes by testing previously reported associations with homocysteine levels in stroke. We found evidence of an association of MUT gene with SVD, an association of MTHFR gene with LVD, and associations of RASIP1 and SLC17A3 with overall IS.
Of the 18 tHcy polymorphisms investigated, 1 polymorphism located upstream of MUT gene was significantly associated with SVD, whereas none of the 18 tHcy-related SNPs were significantly associated with LVD, cardioembolic, or overall IS. The allele correlated with increased tHcy levels at MUT gene showed to be also associated with increased risk of SVD, suggesting a potential small but significant effect on SVD risk.
On a closer inspection of this region, another SNP located 44 kb from the sentinel SNP and in low LD with the sentinel SNP was also associated with SVD. This polymorphism was also significantly associated with homocysteine levels at a genome-wide significance level (P=2.27E–09),11 but conditional analysis has not been conducted to establish if these 2 polymorphisms were independently influencing homocysteine levels. These 2 polymorphisms may thus potentially be correlated with either a single or multiple regulatory variants in this region that modulate both tHcy levels and SVD risk.
The MUT gene is known to encode the mitochondrial enzyme methylmalonyl Coenzyme A mutase, a vitamin B12–dependent enzyme. Considering that vitamin B12 is an important cofactor in homocysteine metabolism, a potential pleiotropic effect of MUT gene on both plasma homocysteine and vitamin B12 levels was suggested previously.11 In addition, vitamin B12 deficiency is highly prevalent in SVD and may underlie blood–brain barrier damage, leading to small-vessel dysfunction, especially periventricular white matter lesions,19 suggesting a potential role of MUT polymorphisms in SVD via mechanisms involving vitamin B12 deficiency.
Moreover, hyperhomocysteinemia as an independent risk factor for SVD may act via endothelial dysfunction,20 suggesting that homocysteine-lowering therapy may be particularly effective in this stroke subgroup. In support of this, the VITAmins TO Prevent Stroke (VITATOPS) trial found that after a 3-year vitamin B supplementation period, the risk of stroke was reduced only in patients with symptomatic SVD.8 Therefore, considering that in our study the only homocysteine significant association was with SVD, brings more evidence to support the hypothesis that homocysteine may be a risk factor in particular to SVD.
Our study also found an association with LVD, a missense variant located in MTHFR (A1298C), 2 kb from the well-studied homocysteine-associated polymorphism MTHFR C677T.2,3 Previous studies showed that A1298C was associated with a decrease in MTHFR activity but was not associated with increased homocysteine levels.21 The functional differences between the 2 polymorphisms could be explained by their location: C677T is located within the N-terminal catalytic region, whereas A1298C is located within the C-terminal regulatory domain, being involved in enzyme regulation.21
Given that high folate levels have been associated with a reduced effect of MTHFR C677T on homocysteine levels and stroke,22 the nonassociation of MTHFR C677T with stroke could be explained by the inclusion in the METASTROKE study of individuals from countries where folic acid fortification has already been implemented.12 The relationship between MTHFR A1298C and folate status has not yet been investigated, and it is not known whether this polymorphism is influenced by folate levels. However, if this SNP is not influenced or influenced at a lesser extent by folate status when compared with C677T, it may explain why A1298C polymorphism, and not C677T, was found to be associated with LVD in our stroke cohort.
For overall IS, 2 highly correlated SNPs located upstream of SLC17A3 and 1 SNP located in an intron of SLC17A3 were found to be associated with IS after correction for multiple testing. None of these SNPs were in LD with the sentinel polymorphism, suggesting the presence of another independent potential risk locus in this region.
SLC17A3 encodes a voltage-driven transporter that excretes intracellular urate and organic anions from the blood into renal tubule cells.23 A significant association between a polymorphism (rs1165205) located in intron 1 of SLC17A3 and serum uric acid concentrations has been found by a GWAS study.24 This SNP was located ≤80 kb from our investigated SNPs but was in low LD (r2<0.2) with all our SLC17A3 SNPs. Therefore, the possibility exists that these polymorphisms are correlated with another variant that has a causal role on influencing urate levels. Alternatively, there may be several independent loci associated with urate levels in this region as it has been suggested previously,24 and our investigated SNPs may be correlated with another polymorphism influencing urate levels independent from the GWAS reported association. Epidemiological studies have shown that elevated serum uric acid is a strong independent risk factor for hypertension,25 and considering that hypertension is a risk factor for stroke, it has been suggested that increased uric acid levels may be involved in predicting stroke risk.26
Another association with IS was a variant located in RASIP1 gene. RASIP1 is required for the proper formation of vascular structures that develop via both vasculogenesis and angiogenesis.27 As it is well known that insufficient vessel growth or maintenance can lead to stroke among other disorders,28 we provide a possible link between RASIP1 and stroke risk. Furthermore, this polymorphism has been reported to be significantly associated with retinal vascular caliber in a previous GWAS.29 Therefore, considering that changes in retinal vascular caliber are associated with cardiovascular diseases, including IS,30 it provides more evidence of a potential role of this gene in stroke.
To date, several GWAS have been conducted on IS and its subtypes, which have identified associations of common polymorphisms specific to each stroke subtype,12,16,31 endorsing the fact that different stroke subtypes have different risk factor profiles and pathophysiological mechanisms. Our study supported this hypothesis and identified a potential association of MUT with SVD and of MTHFR with LVD. In addition, we also identified potential associations of RASIP1 and SLC17A3 with overall IS. The reason why we did not find significant associations of RASIP1 and SLC17A3 with any of the stroke subtypes could be because of the smaller sample sizes and thus reduced power of the stroke subtype cohorts. However, it is important to highlight that our findings need to be validated by replication in independent cohorts to avoid considering spurious results.
In addition, the combined genetic score of the 18 independent tHcy SNPs did not show a significant association with overall IS or its subtypes. The low power of our study to detect small effects, together with the fact that stroke and homocysteine may only partly share their allelic architecture, may explain the lack of association of the tHcy GRS.
However, our results apply to the European population and to countries with established policies of folic acid fortification of grains. A recent study has shown that in these countries, the evidence from genetic studies and from randomized trials with folic acid suggested no benefit from lowering homocysteine levels for stroke prevention.22 Thus, to elucidate the controversial role of homocysteine in stroke, additional studies should be conducted in regions with low folate levels where homocysteine-lowering interventions may have an important effect in reducing stroke risk.
To conclude, our study provides evidence of several potential associations with IS and its subtypes: an association of MUT gene with small-vessel stroke, an association of MTHFR gene with large-vessel stroke, and associations of RASIP1 and SLC17A3 with overall IS, highlighting possible roles of these genes in IS and its subtypes.
We gratefully acknowledge the METASTROKE Collaboration.12
Sources of Funding
This study was supported by the Department of Health (United Kingdom), Henry Smith Charity, British Council, United Kingdom-India Education and Research Institute, and the Qatar National Research Foundation. Genotpying of the cohorts within METASTROKE was supported by the Wellcome Trust, UK Medical Research Council, Australian National and Medical Health Research Council, National Institutes of Health, including National Heart, Lung, and Blood Institute, the National Institute on Aging, the National Human Genome Research Institute, the National Institute of Neurological Disorders and Stroke, and the American Heart Association-Bugher Foundation Centers for Stroke Prevention.
Dr Worrall received support from US National Institute of Neurological Disorders and Stroke and National Human Genome Research Institute. Dr Rosand received support from National Institutes of Health and he is a consultant for Boehringer Ingelheim. Drs Cotlarciuc and Sharma received support from the Qatar National Research Fund NPRP grant no. 4-154-3-050. The other authors report no conflicts.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.005208/-/DC1.
- Received February 17, 2014.
- Revision received March 31, 2014.
- Accepted April 4, 2014.
- © 2014 American Heart Association, Inc.
- Brattström L,
- Wilcken DE
- Saposnik G,
- Ray JG,
- Sheridan P,
- McQueen M,
- Lonn E
- Hankey GJ,
- Eikelboom JW,
- Yi Q,
- Lees KR,
- Chen C,
- Xavier D,
- et al
- Toole JF,
- Malinow MR,
- Chambless LE,
- Spence JD,
- Pettigrew LC,
- Howard VJ,
- et al
- van Meurs JB,
- Pare G,
- Schwartz SM,
- Hazra A,
- Tanaka T,
- Vermeulen SH,
- et al
- Traylor M,
- Farrall M,
- Holliday EG,
- Sudlow C,
- Hopewell JC,
- Cheng YC,
- et al
- Jerrard-Dunne P,
- Cloud G,
- Hassan A,
- Markus HS
- Grau AJ,
- Weimar C,
- Buggle F,
- Heinrich A,
- Goertler M,
- Neumaier S,
- et al
- Pieters B,
- Staals J,
- Knottnerus I,
- Rouhl R,
- Menheere P,
- Kessels A,
- et al
- Hassan A,
- Hunt BJ,
- O’Sullivan M,
- Bell R,
- D’Souza R,
- Jeffery S,
- et al
- Holmes MV,
- Newcombe P,
- Hubacek JA,
- Sofat R,
- Ricketts SL,
- Cooper J,
- et al
- Jutabha P,
- Anzai N,
- Kitamura K,
- Taniguchi A,
- Kaneko S,
- Yan K,
- et al
- Bos MJ,
- Koudstaal PJ,
- Hofman A,
- Witteman JC,
- Breteler MM
- Mitin NY,
- Ramocki MB,
- Zullo AJ,
- Der CJ,
- Konieczny SF,
- Taparowsky EJ
- Ikram MK,
- De Jong FJ,
- Van Dijk EJ,
- Prins ND,
- Hofman A,
- Breteler MM,
- et al
- Holliday EG,
- Maguire JM,
- Evans TJ,
- Koblar SA,
- Jannes J,
- Sturm JW,
- et al