|Home | About | Journals | Submit | Contact Us | Français|
Identifying new risk factors for ischemic stroke occurrence could help improve prevention strategies and identify new therapeutic targets. Genetic risk factors are particularly interesting, because they can offer a direct clue to the biological pathways involved. Ischemic stroke is a heterogeneous disorder, and this must be considered in the search for genetic susceptibility factors. In Western countries, most ischemic strokes can be attributed to large-artery atherosclerosis (atherothrombotic stroke) and small-artery occlusion (lacunar stroke), the rest being due to cardioembolism, rare etiologies such as cervical artery dissection, vasculitis, or monogenic disorders, or to undetermined causes. According to the TOAST criteria an ischemic stroke is classified as atherothrombotic (or due to large-artery atherosclerosis) if it is associated with a >50% stenosis of an appropriate intracranial or extracranial artery,1 and as lacunar (typically due to lipohyalinosis and microatheroma) if a brainstem or subcortical infarct of ≤1.5cm diameter can be demonstrated and large-artery atherosclerosis and cardioembolism can be excluded.1
In this article we reviewed published data on the genetic risk factors underlying atherothrombotic and lacunar strokes. Our reasons for focusing on these two subtypes are that: (i) extensive reviews exploring the genetic risk factors for total ischemic stroke already exist,2-4 (ii) there is some evidence that genetic susceptibility factors may differ according to stroke subtype and that atherothrombotic and lacunar stroke may have a greater genetic component than cardioembolic stroke.5-7
Several monogenic disorders can cause large-artery and small-vessel stroke. Overall, they are responsible for a very small proportion of these strokes, probably less than 1%,8 and should mainly be considered when stroke occurs at a young age in individuals with little or no exposure to conventional vascular risk factors. The mechanisms by which these monogenic disorders result in stroke are varied.
CADASIL (Cerebral Autosomal Dominant Arteriopathy, Subcortical Infarcts and Leukoencephalopathy) is a rare autosomal dominant disease caused by mutations in the Notch3 gene on chromosome 19.9 Notch3 encodes a cell-surface receptor involved in vascular smooth muscle cell survival and vascular remodeling. The prevalence of this disorder has been estimated at about 1/24,000, which is probably an underestimation (www.orpha.net). CADASIL patients suffer from recurrent lacunar strokes. The phenotype also includes progressive cognitive impairment, psychiatric disturbances - especially mood disturbances - and migraine with aura, with an onset between the third and sixth decade. The underlying vascular lesion is a non-arteriosclerotic, amyloid-negative angiopathy involving small arteries and capillaries.10 The definitive diagnostic test is the molecular genetic analysis of the Notch3 gene. Skin biopsy can reveal characteristic granular osmiophilic material within the vascular basal lamina on electron microscopic examination.11 So far only symptomatic treatments are available.
Sickle cell disease is an autosomal recessive disease, which can be due to either a homozygous state for hemoglobin S (HbS) or to the combined heterozygous state with other hemoglobinopathies such as hemoglobin C (HbC) or mild β-thalassemia.12. HbS results from a mutation on the HBB (hemoglobin beta) gene on chromosome 11, which is present on three African haplotypes and a fourth Arab-Indian haplotype.12 The prevalence of the disease at birth varies in affected populations between 0.02 and 3% (www.orpha.net). Stroke affects up to 25% of patients by age 45, and includes both large-artery and small-vessel ischemic stroke (with a peak incidence in early childhood), and hemorrhagic stroke (mainly in adults).13 Large artery strokes are due to intimal thickening, fibroblast and smooth muscle cell proliferation, and thrombus formation, mainly in the distal internal carotid artery and the proximal middle and anterior cerebral arteries.10,12 Small vessel strokes are probably caused by sludging and intravascular sickling in smaller vessels. Other clinical features include vaso-occlusive or painful crisis, retinopathy, chronic leg ulcers, and increased susceptibility to infection.12 Transfusion therapy can reduce the risk of stroke.
Homocysteinuria includes several heritable, mainly autosomal recessive, diseases causing elevated plasma concentrations of homocysteine, and homocysteinuria. The most frequent cause is a mutation in the CBS (cystathionine beta-synthase) gene on chromosome 21. Based on data from newborn screening the rate of occurrence of homocysteinuria is approximately 1/344,000 (www.orpha.net), but prevalence rates of up to 1/20,500 have been described in some European countries.14 Thromboembolism is a key clinical feature, affecting large and small arteries, and veins. The chances of suffering a thromboembolic event were estimated at 25% by age 16 and 50% by age 29, over half these events being venous, and 32% cerebrovascular.15 Other clinical features include ectopia lentis, mental retardation and osteoporosis. Ischemic stroke in homocysteinuria has been primarily attributed to accelerated atherosclerosis.16 However, some pathological studies suggest that arterial damage in homocysteinuria differs from atherosclerosis, most notably by a lack of lipid deposition.17 Other mechanisms could include endothelium-mediated thrombosis through direct toxicity of homocysteine on the endothelium.17 Isolated cases of cervical artery dissection have also been reported.18 Treatments to lower plasma homocysteine levels and stroke risk include pyridoxine, methionine-restricted diet, folate and vitamin B12 supplementation, and betaine.
Fabry's disease is an X-linked disease caused by a mutation in the GLA (galactosidase alpha) gene. Women heterozygous for the mutation can also be symptomatic, but are usually less severely affected than men. The incidence of the disease is estimated at 1/40,000 (www.orpha.net). In a large survey of young patients with cryptogenic ischemic stroke, 4% carried a mutation in GLA.19 Ischemic stroke can be due to both small and large-vessel disease, with a predominance in the vertebro-basilar circulation.19 Deposition of globotriaosylceramide in the vascular wall and dolichectasia of the basilar and vertebral arteries have been described,10 but the mechanisms underlying stroke are still poorly understood. Clinical features include acroparesthesiae, angiokeratomas and hypohidrosis, which often begin in childhood or adolescence, before systemic complications leading to strokes, heart failure and chronic kidney disease occur. The diagnosis is made by measuring α-galactosidase A activity or by mutation screening, the former being less reliable in women. Enzyme-replacement therapy was recently shown to improve some symptoms, but the effect on stroke prevention remains unclear.20
Interestingly, recent studies suggest that polymorphisms in other genes can influence the phenotypic expression of these monogenic diseases. In sickle cell disease in particular, the risk of stroke was shown to be substantially influenced by a variety of modifier genes.21 However there is no data on whether these modifier genes differentially affect the risk of specific subtypes of ischemic stroke. One of the suggested modifier genes of stroke risk in sickle cell disease, selectin P (SELP),21 was also studied in relation with sporadic lacunar and atherothrombotic stroke, but no significant association was observed (supplemental table 12).
There is evidence in the literature, both from twin studies,22 and from studies on the family history of stroke5,23,24 that genetic factors substantially contribute to stroke susceptibility. However, the influence of family history on stroke risk remains controversial, some studies describing no impact of parental history on stroke risk.25 Reasons for these inconsistencies may be: (i) difficulty in reliably assessing parental stroke status, (ii) late occurrence of stroke (parents with a genetic predisposition to stroke may have died of alternative causes before they developed a stroke), (iii) the heterogeneity of stroke, which only some studies have accounted for. Many studies combined ischemic and hemorrhagic stroke, and only a few studies considered ischemic stroke subtypes.25 In four hospital-based series where ischemic stroke patients were compared to healthy controls, family history of stroke was found to be independently associated with an increased risk of large-artery stroke (odds ratio[OR]=1.88[95%CI:1.02-3.44] to 2.24[1.49-3.36]) and small-vessel stroke (OR=1.79[1.13-2.84] to 2.76[1.55-4.91]), the association with cardio-embolic stroke being much weaker.5-7,26 Other studies could not demonstrate any association,27 or only showed a trend towards an increased frequency of positive family history in persons with a lacunar stroke,28 and decreased frequency in cardioembolic stroke,29 but these studies did not include a control population and were probably underpowered.
Assuming that a positive family history is a marker of increased genetic risk, this risk is expected to be higher in stroke subtypes that are more strongly associated with a positive family history. Thus, the sample size needed to unravel genetic effects is probably smaller when focusing on these subtypes.6 Moreover, certain genetic factors probably predispose to specific subtypes of ischemic stroke only, and not to ischemic stroke overall.6 Focusing on ischemic stroke subtype for which a stronger genetic component is suspected may therefore be a key element in elucidating the genetic contribution to ischemic stroke. Genetic variants could act at various levels, e.g. by increasing the risk of and susceptibility to “conventional” stroke risk factors such as hypertension or diabetes, by influencing specific mechanisms underlying stroke, such as the occurrence and progression of atherosclerosis or lipohyalinosis, by predisposing to arterial thrombosis, or by modifying tolerance to brain ischemia.3 The underlying genetic model is expected to be multifactorial, involving several genetic polymorphisms that each confer small increases in risk, with ORs almost always below 1.5 and usually around 1.1-1.3.30,31 A number of approaches have been used to determine the underlying genetic variants contributing to sporadic ischemic stroke risk, including: (i) candidate gene association studies, (ii) genome-wide association studies, and (iii) linkage studies.
References for this second part of the review were identified through searches of PubMed from 1966, to July 1st 2008, with the terms: “gene”[Mesh], “polymorphism, genetic”[Mesh], “haplotypes”[Mesh], “genetic markers”[Mesh], “linkage (genetics)”, “mutation”[Mesh], “genetic predisposition to disease”, “sequence analysis, DNA”[Mesh], “genotype”[Mesh], “alleles”[Mesh], “Brain Infarction”[Mesh], “Stroke”[Mesh]. We selected only articles that had specific data on genetic variation underlying `all stroke' and `ischemic stroke,' after perusal of all titles, abstracts, and, where necessary, the complete article. We further narrowed our selection to include only papers with pertinent information on the atherothrombotic and/or lacunar subtypes of ischemic stroke. Details on the selection criteria are provided in a methodological supplement online.
Candidate gene association studies test the association of a disease with genetic variants selected through a priori hypotheses about the underlying pathophysiology; typically they use a case-control study design. We identified 133 candidate gene association studies that included data on atherothrombotic and lacunar ischemic stroke subtypes and satisfied our inclusion criteria. Data from all these studies is summarized in the online appendix as a series of 12 Supplemental Tables. Each table summarizes data related to a single etiopathologic pathway. For each study we describe the gene and specific SNP or polymorphism studied, the ethnic composition and size of the study sample, the stroke subtype classification system used, the number of atherothrombotic and lacunar stroke events and for studies that found an association, the odds-ratio and 95% confidence intervals reported.
The insertion/deletion polymorphism in intron 16 of the angiotensin-converting enzyme (ACE) gene is one of the most frequently studied genetic variations in relation to atherosclerosis and vascular disease. ACE converts angiotensin I to the vasoconstrictor angiotensin II and inactivates the vasodilator bradykinin, thus contributing to the regulation of vascular tone, endothelial function, and smooth muscle cell proliferation. The insertion/deletion (I/D) polymorphism in the ACE gene was significantly associated with lacunar stroke in 3 independent populations, with an increased risk for D-allele carriers.32-34 However 9 other studies found no association of this variant with lacunar stroke.35-43 None of the studies found an association of the ACE I/D variant with atherothrombotic stroke, and other ACE variants were associated neither with atherothrombotic nor with lacunar stroke.
Inflammation is a key process in atherosclerosis and therefore stroke and coronary heart disease. Leukotrienes are potent proinflammatory molecules, which are involved in leukocyte recruitment and adhesion to the vascular endothelium, as well as in endothelial cell permeability and vascular smooth muscle cell migration.44 One genetic variant (rs730012) in the leukotriene C4 synthase (LTC4S) gene was found to be associated with lacunar stroke in 2 different populations, with an increased risk for A-allele carriers.44 Interleukin-6 (IL-6) is a pleiotropic cytokine that inhibits the production of potent proinflammatory cytokines such as tumor necrosis factor alpha and interleukin-1.45 Increased levels of IL-6 in the plasma and cerebrospinal fluid were shown to correlate with infarct size and functional outcome of stroke patients.45 One genetic variant (rs1800795) in the interleukin-6 (IL6) gene was associated with lacunar stroke in 2 different populations, with an increased risk for CC-homozygotes,45, 46 but it is not clear whether the samples in these 2 studies overlap, and a third study found no association.47 Two studies found an association of atherothrombotic stroke with 2 different variants in IL6 (rs1800796 and rs1800795),47,48 but these associations were not confirmed.46 None of the other associations described in single studies, between lacunar stroke and variants in CD14, Il1B, TGFB1, and TNF, or between atherothrombotic stroke and variants in LTA, and LTC4S, were replicated.
Platelet glycoprotein receptors are members of the integrin family, which, when activated, bind fibrinogen, von Willebrand factor or collagen, and hence promote platelet aggregation and thrombosis.3 An increased risk of atherothrombotic stroke was found in A2-allele carriers of the ITGB3 (integrin, beta 3 or GPIIIa) P1(A1/A2) variant (rs5918) in 3 independent populations.49-51 However, three other studies did not find any association between atherothrombotic stroke and this variant,52-54 or found such an association only in subgroups of patients.52 Factor XIII catalyzes the formation of covalent bounds between fibrin monomers, which stabilizes the fibrin clot, and is also involved in the cross-linking of alpha-2 antiplasmin, fibronectin, and collagen.55 Two studies found an increased risk of atherothrombotic stroke associated with the Val/Val genotype of rs5985 in F13A1 (Coagulation factor XIII, A1 polypeptide),55,56 but two other studies found no association.57,58 None of the other associations noted, between lacunar stroke and variants in F13A1, GP1BA, CPB2, and PLAT, or between atherothrombotic stroke and variants in F5, F12, FGB, ITGA, and CPB2, could be consistently replicated.
Individuals with increased plasma cholesterol, especially low-density lipoprotein (LDL), levels are at higher risk of premature atherosclerosis. Thus genes encoding proteins involved in plasma lipoprotein metabolism are interesting candidates when looking for genetic risk factors of ischemic stroke. The epsilon polymorphism in exon 4 of the gene encoding apolipoprotein E (ApoE), which plays a key role in lipoprotein transport, has been extensively studied in relation with atherosclerosis and vascular events. An increased risk of atherothrombotic stroke was found for APOE ε4 carriers in 5 different studies (compared to non-carriers or to ε3 homozygotes),32,59-62 but in 1 other study the association was in the opposite direction,63 and 5 studies did not find any association.40,48,64-66 One study also observed an increased risk of atherothrombotic stroke in patients with ε2/ε2 or ε2/ε3 genotypes.62 In a meta-analysis including a portion of the data from these studies, no significant association was observed between APOE and either lacunar or atherothrombotic stroke.67 Paraoxonase 1 (PON1) prevents lipid peroxidation of LDLs, an important early step in the atherosclerosis process. Knockout mice lacking PON1 were shown to have an increased susceptibility to atherosclerosis and lowered PON1 activity was observed in vascular disease.68 Two studies found an increased risk of atherothrombotic stroke for carriers of the R allele or the RR genotype of rs662 in PON1,68,69 but 3 other studies did not find any association.48,70,71 The associations, between lacunar stroke and variants in APOA5, or between atherothrombotic stroke and variants in ABCA1, LPA, APOA5, EPHX2, LPL, PON2, and PON3, have not been replicated.
Several studies have suggested that mild to moderate elevations of serum homocysteine are associated with increased risk of vascular disease. Homocysteine levels tend to be higher in individuals homozygous for the thermolabile variant of the methylene tetrahydrofolate reductase (MTHFR) gene, due to a C-to-T transition at position 677. Four studies found an increased risk of atherothrombotic stroke in carriers of the TT genotype of the 677C>T variant in MTHFR,48,72-74 but 7 other studies did not find any association.32,40,72,75-78 In a meta-analysis no association was found with the lacunar and atherothombotic stroke subtypes.79 None of the associations between lacunar stroke and variants in NOS3, MTHFR, or CYBA, or between atherothrombotic stroke and variants in NOS3, and CYBA, were consistently replicated.
The deCODE group in Iceland first demonstrated linkage of stroke with the 5q12 locus using a genome-wide linkage approach,80 and subsequently identified an association of stroke with genetic variants in the PDE4D (phosphodiesterase 4D) gene within this locus.81 The same group also identified significant linkage of myocardial infarction with the 13q12-13 locus, and a significant association of both myocardial infarction and stroke with variants in the ALOX5AP (arachidonate 5-lipoxygenase-activating protein) gene within this second locus.82 Several studies subsequently looked for an association of genetic variants in these two genes with stroke. Two studies found an increased risk of atherothrombotic stroke for carriers of the rare variant of SNP83 (rs966221) in PDE4D,81,83 but 4 other studies could not replicate this finding.84-87 None of the other associations of PDE4D and none of the associations of ALOX5AP with the lacunar and atherothrombotic stroke subtypes have been consistently replicated. In a recent meta-analysis, a borderline significant association of lacunar stroke with PDE4D SNP89 was observed (OR=0.79[0.62-1.00],p=0.05), which became non-significant when the analysis was restricted to whites.88
None of the associations of lacunar stroke with variants in CTSG, PRKCH, and mitochondrial DNA, or of atherothrombotic stroke with variants in NPY, HSPA1B, or mitochondrial DNA, have been replicated.
Other genes that have been tested, but for which no association was found in any of the published studies, are presented in supplemental tables 8-12.
Genome-wide association studies (GWAS) explore the genetic riak factors underlying a condition by genotyping a large number (100,000-1,000,000) of SNPs that are distributed across the chromosomes; they assume no a priori hypothesis regarding genomic loci of interest. They have been applied to a number of complex diseases with several notable successes, e.g. the identification of previously unsuspected genes conferring an increased risk of diabetes,31 and coronary heart disease.89 So far only one GWAS has been published on ischemic stroke.90 This study compared 408,803 single nucleotide polymorphisms in 249 white patients with ischemic stroke and 268 white controls, using the Illumina Infinium Human-1 and HumanHap300 SNP chips (Illumina Inc, San Diego, CA, USA). None of the polymorphisms reached the threshold for genome-wide significance, for ischemic stroke overall and for all ischemic stroke subtypes. However, the statistical power of this pilot study was insufficient to detect small size effects (OR<1.5), especially for rarer genetic variants.90 Noteworthy, in the genome-wide analysis for large-artery and small-artery ischemic stroke none of the SNPs with the lowest p-values (<10-4) was located in any of the previously reported candidate genes.90
Linkage studies are family-based and test whether genetic markers co-segregate with a given disease within families. They have been used successfully to identify genes underlying monogenic forms of ischemic stroke such as CADASIL,9 but they are less powerful when the pattern of inheritance is not mendelian,91 which is true for most ischemic strokes.
Two genome-wide linkage studies on sporadic ischemic stroke have been published.80,92 The first study used a multipoint, affected-only allele-sharing method in 476 stroke patients and 438 of their relatives, and found strong evidence for linkage with a LOD (logarithm of odds) score of 4.86 on chromosome 5q12 for ischemic stroke.80 When ischemic stroke subtypes were investigated, the LOD scores were still positive but smaller than those observed in the overall analysis, consistent with a loss of power due to smaller sample sizes (detailed results were not published). The second study did not identify any new stroke locus and did not report linkage analyses for ischemic stroke subtypes.92 Other studies that attempted to replicate the genetic linkage with the chromosome 5 locus did not report results on ischemic stroke subtypes.83,93
Although they represent only a limited proportion of the published genetic association studies on ischemic stroke, a substantial number of studies have looked for genetic risk factors underlying lacunar and atherothrombotic stroke. A majority of these studies were negative. Several significant associations that were described were not replicated in independent populations. Only a very small number of associations were found in at least 2 independent samples: these are associations of lacunar stroke with ACE,32-34 and LTC4S45,46 variants, and of atherothrombotic stroke with ITGB3,49-51 F13A1,55,56 APOE,32,59-63 PON1,68,69 MTHFR,48,72,73 and PDE4D81,83 variants. For most of these associations there are also negative studies that were published, so that even these remain controversial. Interestingly, in most instances genetic associations differed between lacunar and atherothrombotic stroke, which supports the notion that their genetic backgrounds are distinct.
Published studies on the genetic risk factors for sporadic atherothrombotic and lacunar stroke suffer from limitations that are similar to those of most genetic association studies on stroke,94 and most candidate gene association studies in general.30,94 The two main limitations are: (i) the lack of replication in independent samples of the vast majority of positive associations, and (ii) the insufficient statistical power of most studies to detect modest associations.30,94 The sample sizes needed to achieve a statistical power ≥80% (assuming an additive genetic model, and a type I error rate <0.05) can be estimated at 267 cases and the same number of controls for a SNP with a minor (rare) allele frequency of 20% and an OR≥1.5 (http://pngu.mgh.harvard.edu/~purcell/gpc/). This number increases to 385 and 646 for ORs of 1.3 and 1.1, respectively. If the minor allele frequency is 10%, the numbers become 455, 843 and 2372. Among the studies reviewed here, most had less than 100 subjects per group.
Other limitations are more specific to the study of stroke subtypes. Most studies tested the association of genetic variants with several ischemic stroke subtypes in a secondary analysis and only a few of them adjusted for multiple testing, thus leading potentially to spurious associations. Moreover, the variation in the criteria used to define ischemic stroke subtypes may lead to substantial heterogeneity in the phenotype definition.95 This may have either diluted real effects of genetic variants or may lead to non-specific associations, and thus could explain contradictory findings between studies. Some studies used mechanistic subtyping systems (such as the Classification of Cerebrovascular Diseases III [CCDIII],96 International Classification of Diseases [ICD], or Trial of ORG 10172 in Acute Stroke Treatment [TOAST]1 system), which classify stroke on the basis of inferred origin of cerebrovascular occlusion.95 Other studies used systems based on clinical presentation or location and size of the lesion within the brain (such as the Oxfordshire Community Stroke Project system [OCPS]97), that are practical but may result in less homogenous categorization of stroke subtypes from a mechanistic point of view. In this review we have not addressed genetic variation in copy number, telomeric length or epigenetic mechanisms as possible explanations for the observed heritability of stroke risk. Finally, although efforts were made to be comprehensive in the retrieval of studies, it is possible that relevant studies were inadvertently missed.
There is an urgent need for large genetic association studies designed specifically to look for genetic risk factors underlying ischemic stroke subtypes. The atherothrombotic and lacunar subtypes are particularly important because they are the most frequent cause of ischemic stroke and because there is some evidence that they have a stronger genetic component than other ischemic stroke subtypes.5-7 The genotyping of large numbers of well phenotyped patients, defined according to standardized and validated mechanistic classification systems, will require large collaborative efforts, such as the International Stroke Genetics Consortium (ISGC [www.strokegenetics.org]), the Cohorts for Heart and Aging Research in Genetic Epidemiology (CHARGE) consortium, or the Candidate Gene Association Resource (CARe).98
Genetic marker selection can be improved by studying several SNPs and haplotypes instead of single markers on a given gene to account for possible allelic variation between individuals and ethnic groups, and by simultaneously studying several genes in the same pathway instead of a single gene, with adequate correction for multiple testing. Genome-wide association studies, offer a promising alternative, permitting an unbiased approach, without any a priori assumptions regarding the underlying pathophysiology. Pre-planned replication of positive findings is highly recommended. Emerging statistical and bioinformatic techniques that improve our ability to study gene-gene and gene-environment interactions may assist in uncovering novel genetic variation underlying stroke risk.
Genetic association studies on intermediate phenotypes of atherothrombotic and lacunar stroke, such as carotid intima-media thickness and plaques, or white matter hyperintensities on brain MRI, could also help identify genetic susceptibility factors to these stroke subtypes. Indeed, these intermediate traits are highly heritable, quantitative, and can be assessed non-invasively in large population samples, thus increasing the statistical power to detect associations. Several candidate gene studies on these intermediate phenotypes have been published,99,100 and genome-wide association analyses are under way. Validation of these associations in case-control studies of ischemic stroke patients will however still be required.
Noteworthy, genetic association studies on total ischemic stroke remain scientifically valid as some genes may be associated with pathophysiological mechanisms common to all ischemic strokes, such as susceptibility to ischemia. Thus, large and well designed genetic association studies of ischemic stroke in general and ischemic stroke subtypes should contribute substantially in the near future to a better understanding of the pathophysiology underlying ischemic stroke. These discoveries may permit the identification of new therapeutic targets aimed at stroke prevention and neuroprotection from ischemic injury.
Funding sources: Supported by the National Heart, Lung and Blood Institute's Framingham Heart Study (NIH/NHLBI Contract # N01-HC-25195); grants from the National Institute of Neurological Disorders and Stroke NS17950 and the National Institute of Aging AG08122, AG16495 (PI: Philip A. Wolf) and AG033193 (PI: Sudha Seshadri). Dr. Debette is supported by a Fulbright grant and received an award from the Lilly Institute.
Disclosure statement: None