The genetic basis of haemorrhagic stroke has proved difficult to unravel, partly hampered by the small numbers of subjects in any single study. A meta-analysis of all candidate gene association studies of haemorrhagic stroke (including ruptured subarachnoid haemorrhage and amyloid angiopathy-related haemorrhage) was performed, allowing more reliable estimates of risk.
A systematic review and meta-analysis of all genetic studies in haemorrhagic stroke was conducted. Electronic databases were searched until and including March 2007 for any candidate gene in haemorrhagic stroke. Odds ratio (OR) and 95% confidence intervals (CI) were determined for each gene disease association using fixed and random effect models.
Our meta-analyses included 6,359 cases and 13,805 controls derived from 55 case-control studies, which included 12 genes (13 polymorphisms). Statistically significant associations with haemorrhagic stroke were identified for those homozygous for the ACE/I allele (OR, 1.48; 95% CI, 1.20–1.83; p = 0.0003) and for the 5G allele in the SERPINE1 4G/5G polymorphism (OR, 1.42; 95% CI, 1.03–1.96; p = 0.03). In addition, both &b.epsi;2 and &b.epsi;4 alleles of APOE were significantly associated with lobar haemorrhage (OR, 1.81; 95% CI, 1.26–2.62; p = 0.002 and OR, 1.49; 95% 1.08–2.05; p = 0.01 respectively). Furthermore, a significant protective association against haemorrhagic stroke was found for the factor V Leiden mutation (OR, 0.30; 95% CI, 0.10–0.87; p = 0.03).
Our data suggests a genetic contribution to some types of haemorrhagic stroke, with no overall responsible single gene but rather supporting a polygenic aetiology . However, the evidence base is smaller compared to ischaemic stroke. Importantly, for several alleles previously found to be associated with protection from ischaemic stroke, there was a trend towards an increased risk of haemorrhagic stroke.
Citation: Peck G, Smeeth L, Whittaker J, Casas JP, Hingorani A, Sharma P (2008) The Genetics of Primary Haemorrhagic Stroke, Subarachnoid Haemorrhage and Ruptured Intracranial Aneurysms in Adults. PLoS ONE 3(11): e3691. https://doi.org/10.1371/journal.pone.0003691
Editor: Peter Sandercock, Western General Hospital, United Kingdom
Received: April 28, 2008; Accepted: October 14, 2008; Published: November 14, 2008
Copyright: © 2008 Peck et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Haemorrhagic stroke accounts for approximately 13% of all stroke and is associated with a mortality rate four times higher than ischaemic stroke,,  with only 38% of haemorrhagic stroke patients surviving beyond the first year. Intracranial haemorrhage can be subdivided into extradural, subdural, subarachnoid, intraventricular and intracerebral (with/without subarachnoid and/or intraventricular) haemorrhage. Primary intracerebral haemorrhage (ICH) originates from the spontaneous rupture of small vessels damaged by chronic hypertension or cerebral amyloid angiopathy, while secondary intracerebral haemorrhage is most commonly associated with subarachnoid haemorrhage (SAH)_due to ruptured intracranial aneurysms or ruptured arteriovenous malformations. Haemorrhage from rupture of hypervascular tumors or impaired coagulation occurs in a very small proportion of cases while traumatic related haemorrhage tends to have different characteristics.
Although significant progress has been made towards unraveling the basis of single gene stroke disorders and common ischaemic stroke , , identifying the underlying genes for multifactorial haemorrhagic stroke for which there is no obvious Mendelian pattern of inheritance, has proved difficult. This is despite the evidence for a genetic contribution towards haemorrhagic stroke being considerable (with a likely greater genetic effect for SAH compared to ICH). Stroke cases have also been found to cluster in families. Approximately 10% of patients with intracerebral haemorrhage have a positive family history of haemorrhagic stroke , and having a first-degree relative with intracerebral haemorrhage has been found to be a risk factor for developing the disease. First degree relatives of patients with subarachnoid haemorrhage have up to a seven-fold increased risk of subarachnoid haemorrhage than the general population,,  and about 10% of patients with aneurysmal subarachnoid haemorrhage have a first or second degree relative with subarachnoid haemorrhage or unruptured intracranial aneurysm., 
Genetic association studies in haemorrhagic stroke have identified a considerable number of candidate genes, however, due to lack of reproducibility, uncertainty remains about the nature and number of genes contributing to risk of haemorrhagic stroke. There is concern, on one hand, that positive associations might be spurious and, on the other hand, that the negative findings from some studies might be a consequence of inadequate statistical power from small sample size or methodological shortcomings, such as selection of inappropriate control groups.,  The use of a genetic meta-analysis strategy overcomes some of these limitations, although the results are limited by the available data. Therefore, in this study a comprehensive literature-based meta-analysis of all published association studies in mainly primary intracerebral and aneurysmal subarachnoid forms of haemorrhage was undertaken.
Electronic searches initially using PubMed, Embase and Google Scholar were used to identify published genetic association studies evaluating any candidate gene and any form of haemorrhagic stroke in humans published until and including March 2007. Letters and abstracts were included in the searches.
The Medical Subject Headings and text words used for the search were: haemorrhagic stroke, intracerebral haemorrhage, intracranial haemorrhage, subarachnoid haemorrhage, intracranial aneurysm, aneurysmal h(a)emorrhage, arteriovenous malformation, cerebrovascular malformation, cavernous angioma and cerebral amyloid angiopathy in combination with polymorphism, gene, genotype or mutation. Search results were limited to ‘human’. All languages were searched and translated when necessary. The references of all identified publications were searched for additional studies and the MEDLINE option related articles was used to examine all relevant articles.
Selection criteria included case-control or cohort studies where stroke was analyzed as a dichotomous trait. Studies were selected if neuroimaging (magnetic resonance imaging, computerised tomography or angiography) or autopsy had been used to confirm a diagnosis of intracerebral haemorrhage, subarachnoid haemorrhage, ruptured intracranial aneurysm or cerebral amyloid angiopathy-related haemorrhage. Control populations: The vast majority of the time subjects classified as ‘controls’ had not been subjected to neuro-imaging with investigators assuming that any haemorrhage would be clinically detectable. Different sub-types of haemorrhages were combined on the basis that its aetiology, in the broadest sense, is vessel damage. Moreover, most investigators in our study combined such data. However, haemorrhagic sub-type data was analyzed separately where possible, e.g. Factor V Leiden. Studies of all ethnic backgrounds were included. Genes with three or more publications on each SNP were included in our analyses but, if very large numbers of subjects were identified in only two studies dealing with a solitary SNP these were included. Studies were excluded if (1) the patients were children (aged <18 years), (2) genotype frequency was not adequately reported (and such data could not be obtained from the authors), (3) genotype frequency of haemorrhagic stroke sub-types were not reported independently from ischaemic stroke cases, (4) cases due to trauma, haemorrhagic rupture of tumour or haemorrhagic conversion of ischaemic stroke were included in case genotype frequencies, or (5) quantitative traits or intermediate phenotypes were investigated. Data from SNP investigated in three or more studies are presented.
The first pass data extraction was undertaken by GP. Several subsequent passes were then undertaken by GP and PS to ensure comprehensive inclusion of appropriate studies. Once studies were identified data was extracted initially by GP and then checked by PS. For studies with more than one publication describing results among the same or overlapping groups of patients or controls the largest of the available published data sets was included to avoid double counting. For studies with more than one control group, the most appropriate control group was used. Where neither control group was methodologically superior the largest was used.
Data were analyzed using software for preparing and maintaining Cochrane reviews (Review Manager v4.2.8, Cochrane Collaboration, http://www.cc-ims.net/RevMan) and Comprehensive Meta Analysis v2.2.023 (Biostat, http://www.biostat.org). To determine the strength of genetic association a pooled odds ratio (OR) was calculated for each gene variant using fixed- and random-effects models, in addition to the calculation of 95% confidence intervals (CI). Fixed-effects summary ORs were calculated using the Mantel-Haenszel method,,  and the DerSimonian and Laird method was used to calculate random-effects summary ORs . The results were very similar for both summary OR so only one is presented. The frequency of at-risk genotypes was compared between cases and controls for each single nucleotide polymorphism (SNP) analyzed. For each meta-analysis, the I2 was calculated and a chi-squared test for heterogeneity was performed. The genetic models used reflected those evaluated in the primary publication. For assessment of small-study bias, funnel plots and the Egger regression asymmetry test were conducted for each SNP with four or more publications.
The proportion of haemorrhagic stroke cases in the population that could be attributed to a particular gene variant (population-attributable risk [PAR]) was estimated as follows:For this calculation the fixed effects model was used and the prevalence of exposure was estimated as the genotype frequency among control adults.
Our initial search identified 1,565 studies of which 85 met our inclusion criteria. In total, 40 polymorphisms in 27 genes were identified. Genes with three or more publications on each SNP were included in our analyses, leaving a total of 55 publications addressing 13 polymorphisms in 12 genes (Figure 1).
From the 12 genes analysed in detail (representing 6,367 cases and 13,504 controls), the mean number of studies per candidate gene was five. Six (46%) of the 13 meta-analyses had more than 500 cases, and 8 (62%) had a total participant size of greater than 1000 (Table 1). No single meta-analysis had more than 2000 cases.
Table 2 shows the genotypic odds ratios (ORs) for the 13 polymorphisms evaluated.
The most investigated gene was apolipoprotein E, with 11 studies – that included 1376 cases and 3531 controls. APOE 𝛆2, 𝛆3 and 𝛆4 alleles were analysed in a dominant model (i.e. 𝛆2𝛆4+𝛆3𝛆4+𝛆4𝛆4 vs all other genotypes or 𝛆2𝛆2+𝛆2𝛆3+𝛆2𝛆4 vs all other genotypes). For intracerebral haemorrhage, no association was found for carriers of the 𝛆2 allele (Figure 2a) (OR, 1.09; 95% CI, 0.88–1.35), but an association was observed for carriers of the 𝛆4 allele (Figure 2b) with an OR of 1.17 (95% CI, 0.98–1.40; P = 0.08). No significant interstudy OR heterogeneity was observed for either allele (χ2 = 9.54; PHet = 0.30 and χ2 = 7.48; PHet = 0.49 respectively).
2a. Results for apolipoprotein &b.epsi;2 allele and intracerebral haemorrhage comparing carriers of the &b.epsi;2 allele with those with the &b.epsi;3 and &b.epsi;4 alleles. 2b. Results for apolipoprotein &b.epsi;4 allele and intracerebral haemorrhage comparing carriers of the &b.epsi;4 allele with those with the &b.epsi;2 and &b.epsi;3 alleles. 2c. Results for apolipoprotein &b.epsi;2 allele and subarachnoid haemorrhage comparing carriers of the &b.epsi;2 allele with those with the &b.epsi;3 and &b.epsi;4 alleles. 2d. Results for apolipoprotein &b.epsi;4 allele and subarachnoid haemorrhage comparing carriers of the &b.epsi;4 allele with those with the &b.epsi;2 and &b.epsi;3 alleles. 2e. Results for apolipoprotein &b.epsi;2 allele and cerebral amyloid angiopathy-related haemorrhage comparing carriers of the &b.epsi;2 allele with those with the &b.epsi;3 and &b.epsi;4 alleles. 2f. Results for apolipoprotein &b.epsi;4 allele and cerebral amyloid angiopathy-related haemorrhage comparing carriers of the &b.epsi;4 allele with those with the &b.epsi;2 and &b.epsi;3 alleles. 2g. Results for apolipoprotein &b.epsi;2 allele and lobar or non-lobar haemorrhage comparing carriers of the &b.epsi;2 allele with those with the &b.epsi;3 and &b.epsi;4 alleles. 2h. Results for apolipoprotein &b.epsi;4 allele and lobar or non-lobar haemorrhage comparing carriers of the &b.epsi;4 allele with those with the &b.epsi;2 and &b.epsi;3 alleles.
For subarachnoid haemorrhage, no association was found for carriers of the 𝛆2 (Figure 2c) or 𝛆4 (Figure 2d) allele with ORs of 1.20 (95% CI, 0.85–1.70; p = 0.30) and 1.22 (95% CI, 0.90–1.64; p = 0.19), respectively. No significant interstudy OR heterogeneity was observed for either allele.
Only two publications (64 cases and 430 controls) with genotype frequencies addressed cerebral amyloid angiopathy-related haemorrhage. McCarron et al.  used histology to confirm a diagnosis of CAAH, while Garcia et al.  used neuroimaging and clinical criteria to diagnose ‘probable/possible’ CAAH. Carriers of the 𝛆2 allele demonstrated an OR of 3.36 (95% CI, 1.71–6.59; p = 0.0004) (Figure 2e) and for carriers of 𝛆4 an OR of 2.69 (95% CI, 1.47–4.92; p = 0.001) (Figure 2f) for this condition. No significant interstudy OR heterogeneity was observed for either allele (𝛆2: χ2 = 0.78; PHet = 0.36 and 𝛆4: χ2 = 0.55; PHet = 0.45).
Previous studies investigating apolipoprotein E in haemorrhagic stroke have addressed the association of the 𝛆2 and 𝛆4 alleles in lobar and non-lobar haemorrhage. Four studies , , ,  provided data that appropriately categorized haemorrhage location as lobar or non-lobar. Both 𝛆2 and 𝛆4 alleles were significantly associated with lobar haemorrhage (OR, 1.81; 95% CI, 1.26–2.62; P = 0.002 and OR, 1.49; 95% 1.08–2.05; P = 0.01 respectively) but not with non-lobar haemorrhage (OR, 1.08; 95% CI, 0.80–1.47 and OR, 1.05; 95% CI, 0.81–1.37) (Figures 2g and 2h). No significant interstudy OR heterogeneity was observed for either allele (𝛆2 lobar: χ2 = 6.91; PHet = 0.07; 𝛆2 non-lobar: χ2 = 2.01; PHet = 0.57; 𝛆4 lobar: χ2 = 3.20; PHet = 0.36; 𝛆4 non-lobar: χ2 = 2.88; PHet = 0.41).
A total of five studies – (240 cases and 1233controls) investigated the SERPINE1 (also known as plasminogen activator inhibitor-1) 4G/5G polymorphism. Four addressed intracerebral haemorrhage and one study  addressed intracerebral haemorrhage, subarachnoid haemorrhage and ruptured intracranial aneurysm. For intracerebral haemorrhage alone an association was observed for those homozygous for the 5G allele with an OR of 1.40 (95% CI, 0.98–2.00; P = 0.07) (Figure 3). When subarachnoid haemorrhage and ruptured intracranial aneurysm cases were included in this analysis a significant association was observed with a similar OR of 1.42 (95% CI, 1.03–1.96; P = 0.03). No significant interstudy OR heterogeneity was observed (χ2 = 0.28; PHet = 0.99). The distribution of the OR in relation to its standard deviation in the funnel plot was symmetrical, and the Egger test result was not significant (P = 0.178), suggesting a low probability of small-study bias.
A total of 8 studies – (744 cases and 1289 controls) were identified that evaluated the ACE/ID polymorphism of which five related to intracerebral haemorrhage, one for subarachnoid haemorrhage and two for ruptured intracranial aneurysm.
No significant association for haemorrhagic stroke was found for those homozygous for the D allele (ACE/DD) for intracerebral haemorrhage alone (OR, 1.00; 95% CI, 0.73–1.36; p = 1.00), or with subarachnoid haemorrhage or ruptured intracranial aneurysm (OR, 0.87; 95% CI, 0.65–1.14; p = 0.09). In addition to examining the association of the ACE/DD genotype with haemorrhagic stroke, several studies have suggested that the ACE/II genotype may be associated with haemorrhagic stroke. When all studies were evaluated together, those homozygous for the ACE/I allele demonstrated an overall pooled OR of 1.48 (95% CI, 1.20–1.83; p = 0.0003) (Figure 4). Significant interstudy OR heterogeneity was observed (χ2 = 16.69; PHet = 0.02), however a random-effect model that takes into account the intra- and inter-study variability resulted in a similar overall estimate that remained significant (OR, 1.45; 95% CI, 1.01–2.07; p = 0.04). The distribution of the OR in relation to its standard deviation in the funnel plot was symmetrical, and the Egger test result was not significant (p = 0.667), suggesting a low probability of small-study bias. For intracerebral haemorrhage alone this association did not remain significant (OR, 1.27; 95% CI, 0.90–1.77; P = 0.17). However, for subarachnoid haemorrhage or ruptured intracranial aneurysm, a summary OR, under a fixed-effect model, of 1.64 (95% CI, 1.24–2.17; p = 0.0003) was observed.
Three studies (285 cases, 836 controls) investigated haemorrhagic stroke in the factor V Leiden mutation. One addressed intracerebral haemorrhage  and two addressed both intracerebral haemorrhage and subarachnoid haemorrhage , . When considered in totality, carriers of the adenine allele were less likely to develop haemorrhagic stroke (OR, 0.31; 95% CI, 0.11–0.90; p = 0.03) when compared to wild-type (guanine/guanine) (Figure 5a). No significant interstudy heterogeneity was observed (χ2 = 0.74; PHet = 0.69)
5a. Results for Factor V Leiden mutation and haemorrhagic stroke. 5b. Results for Factor V Leiden mutation and ICH. 5c. Results for Factor V Leiden mutation and SAH.
For intracerebral haemorrhage alone, a summary OR, under the fixed-effects model, of 0.29 (95% CI, 0.08–1.05; p = 0.06) was observed for adenine allele homozygotes (Figure 5b). For subarachnoid haemorrhage alone, a summary OR, under a fixed-effect model, of 0.45 (95% CI, 0.11–1.95; p = 0.29) was observed (Figure 5c). No significant interstudy heterogeneity was observed for intracerebral (χ2 = 1.38; PHet = 0.50) or subarachnoid haemorrhage (χ2 = 0.26; PHet = 0.61).
Three studies (285 cases; 512 controls) investigated risk of haemorrhagic stroke and the exon 7/8 GA insertion of the gene encoding endoglin. One study addressed intracerebral haemorrhage  and two studies addressed subarachnoid haemorrhage , . When considered in totality, those homozygous for the insertion allele demonstrated an OR of 3.47 (95% CI, 1.45–8.30; p = 0.005). No significant interstudy OR heterogeneity was observed (χ2 = 1.15; PHet = 0.56).
A total of 6 studies – (1069 cases and 1294 controls) were identified that evaluated the polymorphism in the gene encoding SERPINA3 (also known as α1-antichymotrypsin) where alanine is replaced by threonine. Four of these addressed intracerebral haemorrhage and two addressed subarachnoid haemorrhage.
For intracerebral haemorrhage, a summary OR, under the fixed-effects model, of 1.28 (95% CI, 0.98–1.67; p = 0.07) was observed for individuals homozygous for the T allele compared with A allele carriers (AT+AA). For subarachnoid haemorrhage alone, a summary OR, under a fixed-effect model, of 0.94 (95% CI, 0.71–1.24; p = 0.65) was observed (Figure 6). No significant interstudy heterogeneity was observed (χ2 = 5.17; PHet = 0.40) and the funnel plot distribution was symmetrical with an insignificant Egger test (P = 0.14), indicating a low probability of small-study bias.
Endothelial nitric oxide synthase:
The eNOS T786C mutation was evaluated in 4 studies –, all of which addressed subarachnoid haemorrhage with a total of 893 cases and 784 controls. The summary OR under a fixed-effect model showed an OR for dominant carriers of the 786C mutation of 1.27 (95% CI, 0.99–1.62; P = 0.06) (Figure 7). No significant interstudy heterogeneity was observed (χ2 = 1.76; PHet = 0.62). The distribution of the ORs from individual studies in relation to their respective standard deviations was symmetrical, and the Egger test result suggested a low probability of small-study bias (p = 0.29).
Five studies – investigated the MTHFR C677T polymorphism, all addressing intracerebral haemorrhage (635 cases; 2275 controls). No significant association for intracerebral haemorrhage was found for 677TT (OR, 1.11; 95% CI, 0.89–1.39; p = 0.35) (Figure 8). No significant interstudy heterogeneity was observed (χ2 = 7.29; PHet = 0.12) and the funnel plot distribution was symmetrical with an insignificant Egger test (p = 0.27), indicating a low probability of small-study bias.
The factor XIII valine/leucine polymorphism was investigated in 5 studies , –. Subarachnoid haemorrhage was addressed in addition to intracerebral haemorrhage in two of these studies (531 cases; 1172 controls). When considered in totality, those homozygous for the leucine allele demonstrated an OR of 1.39 for haemorrhagic stroke (95% CI, 0.89–2.08), although this result was not statistically significant (p = 0.15). No significant interstudy heterogeneity was observed (χ2 = 7.45; PHet = 0.11) and the funnel plot distribution was symmetrical with an nonsignificant Egger test (p = 0.27), indicating a low probability of small-study bias. For intracerebral haemorrhage, a summary OR under a fixed-effect model of 1.32 (95% CI, 0.82–2.06; p = 0.27) was observed for individuals homozygous for the leucine allele compared with valine allele carriers (V/L plus V/V)).
Three studies investigated haemorrhagic stroke in the glycoprotein 1b-α VNTR polymorphism. Two addressed intracerebral haemorrhage ,  (198 cases; 563 controls) and one addressed subarachnoid haemorrhage  (103 cases; 103 controls). When considered in totality carriers of the B group of tandem repeats had no associated risk for haemorrhagic stroke with a summary OR, under a fixed-effect model, of 1.04 (95% CI, 0.66–1.64; p = 0.85).
Glycoprotein Ia and IIIa:
Three studies investigated haemorrhagic stroke in the glycoprotein Ia Glu505Lys polymorphism , ,  (286 cases; 590 controls). No association for haemorrhagic stroke was found for carriers of the lysine allele with a summary OR, under the fixed-effect model, of 0.82 (95% CI, 0.56–1.20). Additionally, the same studies found no association for haemorrhagic stroke with the C807T polymorphism of glycoprotein Ia (OR, 1.15; 95% CI, 0.83–1.59) nor for the glycoprotein IIIa Leu33Pro polymorphism (OR, 0.76; 95% CI, 0.54–1.07).
In this comprehensive meta-analysis, four of the 13 (31%) polymorphisms analyzed significantly increased the risk of haemorrhagic stroke. The mean number of participants in two of these meta-analyses (ACE and SERPINE1) was more than 1850, allowing more precise estimates to be made of the effect of these genes than from any single study. The individual risk provided by these candidate genes was moderate (OR, 1.48; 95% CI: 1.20–1.83 and 1.42, 95% CI: 1.03–1.96, respectively). This effect size is in broad agreement with previous studies in ischaemic stroke  and other complex diseases that are thought to have a polygenic basis, such as ischaemic heart disease , . Further, our study suggests that the hypercoagulable state conferred by the factor V Leiden mutation is protective against haemorrhagic stroke (OR, 0.31; 95% CI, 0.11–0.90; p = 0.03). This is in accordance with previous studies that have suggested a protective role for this polymorphism in other bleeding disorders. ,  However, these results are based on a smaller dataset compared to ischaemic stroke making the point estimates less reliable. The seven gene variants in the six remaining genes – factor XIII, MTHFR, SERPINA3 and glycoproteins Ia, 1b-α, and IIIa – have so far failed to provide evidence to support an increased susceptibility to haemorrhagic stroke.
The majority of candidate genes in haemorrhagic stroke have been investigated initially for their potential role in ischaemic stroke. For APOE, our results are in accordance with previous histological studies suggesting that, while the 𝛆4 allele enhances the deposition of amyloid-β in peripheral (lobar) cerebral vasculature, the 𝛆2 allele predisposes to the rupture of these lobar amyloid-β laden vessels, possibly through the promotion of fibrinoid necrosis –
A meta-analysis of genetic studies in ischaemic stroke by our group in 2004 investigated, among others, 9 (75%) of the 12 candidate genes in our study (Table 3) and involved approximately 3 times as many cases and controls. Although ischaemic and haemorrhagic stroke may share some common risk factors they are thought to have quite distinct molecular pathogeneses. Our current work demonstrates strikingly similar genetic ORs between the two major pathologies supporting a similar aetiological role for both ischaemic and haemorrhagic stroke (Table 3) with a greater patho-aetiological similarity existing between small vessel ischaemic disease and deep cerebral haemorrhage. Interestingly, Factor V was associated with ischaemic stroke and protective for haemorrhagic stroke, providing a biological relevant and important role for its aetiology in cerebrovascular disease.
The population-attributable risks (PARs) for gene variants with the most reliable associations for haemorrhagic stroke in this study were 9.24% for the ACE/ID polymorphism and 7.78% for the SERPINE1 4G/5G. These values are lower than those reported for well-established risk factors for haemorrhagic stroke, such as hypertension  but not surprising, because the genetic contribution of any single gene toward a complex disease is unlikely to act in a simple mendelian fashion but rather with epistatic (gene-gene or gene-environmental interaction) effects. Nevertheless, given the high incidence of haemorrhagic stroke (100,000 per year in the United States), if these estimates hold true they suggest that common variants in the ACE and SERPINE1 genes alone may contribute between 8,000–10,000 haemorrhagic strokes in the United States each year.
The interpretation of any meta-analysis must be made within the context of its limitations, including study selection, publication bias, and variability in the methodological quality of the included studies. A comprehensive search was undertaken for all relevant studies including non-English language , , . Although publication bias can never be completely excluded, many of the individual studies included in our meta-analyses were not statistically significant and were interpreted by their authors as negative studies suggesting a desire by authors to submit, and a willingness by editors to publish such work. In addition, Egger asymmetry tests and funnel plots showed no substantial evidence of publication bias in the nine meta-analyses for which four or more studies had been published (apolipoprotein E ε4, and ε2; SERPINA3 A/T; eNOS T786C; ACE/ID; MTHFR C677T; factor XIII Val/Leu and SERPINE1 4G/5G), although it is acknowledged that at times the study numbers for each gene were small as was the number of subjects for each study making the point estimates less reliable when compared against our previous larger meta-analyses in stroke , . Moreover, rigorous selection criteria (use of neuroimaging or autopsy to diagnose haemorrhagic stroke) enriched the meta-analyses for studies with comparable selection of participants and haemorrhagic stroke sub-types. Notwithstanding these selection criteria, patients with severe ICH from any aetiology who died at onset without gaining access to secondary care could not have entered in any of the included manuscripts.
Only two of our meta-analyses included more than 1000 cases (apolipoprotein E and SERPINA3). Reliable interpretation of the association of candidate genes with haemorrhagic stroke will only come from studies with an order of magnitude larger than those performed to date. Dichgans et al. suggest that to confirm odds ratios between 1.2–1.5 in candidate genes with allele frequencies between 0.1–0.5, sample sizes of 800–20,000 are required to achieve reliable statistical significance at a P value of at least 0.05. The average number of cases and controls in our meta-analysis were just 578 and 1224 respectively. Achieving appropriate sample sizes is only likely to be achieved using a multicenter collaborative approach, with studies using uniform criteria for the selection of cases and controls and submitting both positive and negative data to a common database for continuously updated and cumulative meta-analysis. Using neuroimaging- or autopsy-diagnosed haemorrhagic stroke as a selection criteria may have helped to maintain comparable groups of cases. However the selection of control groups varied considerably between studies with some excluding control subjects with established hypertension, while others argued that as not all patients with haemorrhagic stroke have high blood pressure such exclusion of control population is unnecessary. Statistical methods using marker genotype data may in the future permit the detection and control of confounding due to population stratification and selection bias in genetic association studies. This may reduce the impact of variability within controls groups, although the strategy of using a control group that has not been specifically phenotyped for comparison with each individual disease has recently been successfully employed  suggesting that this is not a major problem.
All studies from different ancestral backgrounds were included. Although there is evidence to suggest that allele frequencies for several candidate genes investigated in our study vary between ethnicity, and epidemiological studies suggest differences in the incidence and prevalence of haemorrhagic stroke in certain ethnic groups with the rate in black Americans twice that of white Americans,,  as well as a greater incidence in Japanese population when compared to Caucasian populations , the majority of our studies were conducted in Caucasians. Indeed, we have recently shown in ischaemic stroke that differences in genetic effects across different ethnic groups may be overstated . However, the effect of heterogeneity by ethnicity cannot be completely dismissed.
Only adults with ruptured intracranial aneurysm were included in our analyses addressing subarachnoid haemorrhage. Many published articles have investigated genetic polymorphisms in subjects with unruptured intracranial aneurysm. While some genes may predispose to the formation of intracranial aneurysms, others may predispose to rupture of aneurysms. By performing a similar meta-analysis of genetic association studies in unruptured intracranial aneurysm it may be possible to further our understanding of the genetic and molecular pathogenesis of subarachnoid haemorrhage. Although attempts were made to analyse sub-types of haemorrhage separately, the majority of the time such data were not readily available. For this, as well as reasons of aetiology where it can be argued that haemorrhagic stroke at its most basic consists of vessel rupture, data were combined. Readers should therefore assess this manuscript with this limitation in mind. Further, even within the category of primary intracerebral haemorrhage, heterogeneity may exist while some patients may not have been subjected to detailed investigations in order to determine secondary underlying causes, thus erroneously labelling them clinically as ‘primary’.
Although it is not possible to exclude the future identification of one or more genes with a more substantial effect on risk of haemorrhagic stroke using different models such as genome wide association searches, our findings based upon candidate gene strategies suggest that several genes, each with a small to moderate effect, are likely to act individually, together, or in combination with environmental determinants to contribute towards haemorrhagic stroke. One implication of these findings is that predictive genetic tests that use any single variant are unlikely individually to have much value. However, tests that combine genotyping for one or more risk alleles and that integrate the results with established risk prediction tools based on acquired risk factors may have greater utility., 
This study may have important implications for those involved in exploring the genetic aetiology of blood pressure because of the relationship between haemorrhagic stroke and hypertension. Few studies (∼50%) reported the prevalence of hypertension among their case and control populations. Of those that did the average incidence of hypertension among cases in the ACE study was 51.7% compared to 29.7% in controls, and for PAI-1 the average for cases was 26.7% and for controls 12.9%. From our study the mechanism by which causative SNPs lead to haemorrhagic stroke cannot be certain, although this could be via known causal pathways such as hypertension especially if those genes are involved in blood pressure regulation. There is, therefore, a clear need for future investigators and editors to ensure that this important variable is documented and published in the final manuscripts.
In summary, this study suggests the existence of a genetic basis for some types of haemorrhagic stroke with no single gene but rather demonstrates a polygenic aetiology. However, the evidence base is smaller than when compared against ischaemic stroke, although the odds ratios are of similar magnitude. An international collaborative approach is more likely to lead to a sufficient number of subjects being recruited for the reliable identification of risk alleles in haemorrhagic stroke.
Conceived and designed the experiments: PS. Analyzed the data: GP LS JW JPC AH PS. Wrote the paper: GP LS JW JPC AH PS.
- 1. El-Saed A, Kuller LH, Newman AB, et al. (2006) Geographic variations in stroke incidence and mortality among older populations in four US communities. Stroke 37(8): 1975–1979.
- 2. Rosamond WD, Folsom AR, Chambless LE, et al. (1999) Stroke incidence and survival among middle-aged adults: 9-year follow-up of the Atherosclerosis Risk in Communities (ARIC) cohort. Stroke 30(4): 736–743.
- 3. Dennis MS, Burn JP, Sandercock PA, Bamford JM, Wade DT, et al. (1993) Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke 24(6): 796–800.
- 4. Sutherland GR, Auer RN (2006) Primary intracerebral hemorrhage. J Clin Neurosci 13(5): 511–517.
- 5. Ariyaratnam R, Casas JP, Whittaker J, Smeeth L, Hingorani AD, Sharma P (2007) Genetics of ischaemic stroke among persons of non-European descent: a meta-analysis of eight genes involving approximately 32,500 individuals. PLoS Med 4(4): e131.
- 6. Casas JP, Hingorani AD, Bautista LE, Sharma P (2004) Meta-analysis of genetic studies in ischemic stroke: thirty-two genes involving approximately 18,000 cases and 58,000 controls. Archives of Neurology 61(11): 1652–1661.
- 7. Alberts MJ, McCarron MO, Hoffmann KL, Graffagnino C (2002) Familial clustering of intracerebral hemorrhage: a prospective study in North Carolina. Neuroepidemiology 21(1): 18–21.
- 8. Bromberg JE, Rinkel GJ, Algra A, et al. (1995) Subarachnoid haemorrhage in first and second degree relatives of patients with subarachnoid haemorrhage. BMJ 311(7000): 288–289.
- 9. Gaist D, Vaeth M, Tsiropoulos I, et al. (2000) Risk of subarachnoid haemorrhage in first degree relatives of patients with subarachnoid haemorrhage: follow up study based on national registries in Denmark. BMJ 320(7228): 141–145.
- 10. Schievink WI, Schaid DJ, Michels VV, Piepgras DG (1995) Familial aneurysmal subarachnoid hemorrhage: a community-based study. J Neurosurg 83(3): 426–429.
- 11. Ronkainen A, Hernesniemi J, Puranen M, et al. (1997) Familial intracranial aneurysms. Lancet 349(9049): 380–384.
- 12. Dichgans M, Markus HS (2005) Genetic association studies in stroke: methodological issues and proposed standard criteria. Stroke 36(9): 2027–2031.
- 13. Ioannidis JP, Gwinn M, Little J, et al. (2006) A road map for efficient and reliable human genome epidemiology. Nature Genet 38(1): 3–5.
- 14. Robins J, Greenland S, Breslow NE (1986) A general estimator for the variance of the Mantel-Haenszel odds ratio. Am J Epidemiol 124(5): 719–723.
- 15. Mantel N, Haenszel W (1959) Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 22(4): 719–748.
- 16. DerSimonian R, Laird N (1986) Meta-analysis in clinical trials. Control Clin Trials 7(3): 177–188.
- 17. Deek JJ, Altman DG, Bradburn MJ (2001) Statistical methods for examining heterogeneity and combining results from several studies in meta-analysis. In: Egger M, Davey SG, Altman DG, editors. Systemic reviews in health care: meta-analysis in context. Annapolis Junction, Md: BMJ Publishing Group.
- 18. Egger M, Davey SG, Schneider M, Minder C (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315(7109): 629–634.
- 19. Yi HA, Lim JG, Kwon OD, et al. (2007) Apolipoprotein E Polymorphism in Ischemic and Hemorrhagic Stroke. Korean J Anat 35: 355–361.
- 20. Woo D, Kaushal R, Chakraborty R, et al. (2005) Association of apolipoprotein E4 and haplotypes of the apolipoprotein E gene with lobar intracerebral hemorrhage. Stroke 36(9): 1874–1879.
- 21. Tang J, Zhao J, Zhao Y, Wang S, Chen B, et al. (2003) Apolipoprotein E epsilon4 and the risk of unfavorable outcome after aneurysmal subarachnoid hemorrhage. Surg Neurol 60(5): 391–396.
- 22. Seifert T, Lechner A, Flooh E, Schmidt H, Schmidt R, et al. (2006) Lack of association of lobar intracerebral hemorrhage with apolipoprotein E genotype in an unselected population. Cerebrovasc Dis 21(4): 266–270.
- 23. McCarron MO, Nicoll JA (1998) High frequency of apolipoprotein E epsilon 2 allele is specific for patients with cerebral amyloid angiopathy-related haemorrhage. Neurosci Lett 247(1): 45–48.
- 24. Kokubo Y, Chowdhury AH, Date C, Yokoyama T, Sobue H, et al. (2000) Age-dependent association of apolipoprotein E genotypes with stroke subtypes in a Japanese rural population. Stroke 31(6): 1299–1306.
- 25. Garcia C, Pinho e Melo , Rocha L, Lechner MC (1999) Cerebral hemorrhage and apoE. J Neurol 246(9): 830–834.
- 26. Fontanella M, Rainero I, Gallone S, et al. (2007) Lack of association between the apolipoprotein E gene and aneurysmal subarachnoid hemorrhage in an Italian population. J Neurosurg 106(2): 245–249.
- 27. Duzenli S, Pirim I, Gepdiremen A, Deniz O (2004) Apolipoprotein E polymorphism and stroke in a population from eastern Turkey. J Neurogenet 18(1): 365–375.
- 28. Chowdhury AH, Yokoyama T, Kokubo Y, Zaman MM, Haque A, et al. (2001) Apolipoprotein E genetic polymorphism and stroke subtypes in a Bangladeshi hospital-based study. J Epidemiol 11(3): 131–138.
- 29. Catto AJ, McCormack LJ, Mansfield MW, et al. (2000) Apolipoprotein E polymorphism in cerebrovascular disease. Acta Neurol Scand 101(6): 399–404.
- 30. Sudlow C, Martinez Gonzalez NA, Kim J, et al. (2006) Does apolipoprotein E genotype influence the risk of ischemic stroke, intracerebral hemorrhage, or subarachnoid hemorrhage? Systematic review and meta-analyses of 31 studies among 5961 cases and 17,965 controls. Stroke 37(2): 364–370.
- 31. Zhang C, Li J, Li L, Luo B (2001) [The study of PAI-1 promotor region gene polymorphism in cerebrovascular disease]. Zhonghua Yi Xue Yi Chuan Xue Za Zhi 18(5): 383–387.
- 32. Wiklund PG, Nilsson L, Ardnor SN, et al. (2005) Plasminogen activator inhibitor-1 4G/5G polymorphism and risk of stroke: replicated findings in two nested case-control studies based on independent cohorts. Stroke 36(8): 1661–1665.
- 33. Reiner AP, Schwartz SM, Frank MB, et al. (2001) Polymorphisms of coagulation factor XIII subunit A and risk of nonfatal hemorrhagic stroke in young white women. Stroke 32(11): 2580–2586.
- 34. Hindorff LA, Schwartz SM, Siscovick DS, Psaty BM, Longstreth WT Jr, et al. (2002) The association of PAI-1 promoter 4G/5G insertion/deletion polymorphism with myocardial infarction and stroke in young women. J Cardiovasc Risk 9(2): 131–137.
- 35. Catto AJ, Carter AM, Stickland M, Bamford JM, Davies JA, et al. (1997) Plasminogen activator inhibitor-1 (PAI-1) 4G/5G promoter polymorphism and levels in subjects with cerebrovascular disease. Thromb Haemost 77(4): 730–734.
- 36. Slowik A, Borratynska A, Pera J, et al. (2004) II genotype of the angiotensin-converting enzyme gene increases the risk for subarachnoid hemorrhage from ruptured aneurysm. Stroke 35(7): 1594–1597.
- 37. Slowik A, Turaj W, Dziedzic T, et al. (2004) DD genotype of ACE gene is a risk factor for intracerebral hemorrhage. Neurology 63(2): 359–361.
- 38. Pannu H, Kim DH, Seaman CR, Van GG, Shete S, et al. (2005) Lack of an association between the angiotensin-converting enzyme insertion/deletion polymorphism and intracranial aneurysms in a Caucasian population in the United States. J Neurosurg 103(1): 92–96.
- 39. Nakata Y, Katsuya T, Rakugi H, et al. (1997) Polymorphism of angiotensin converting enzyme, angiotensinogen, and apolipoprotein E genes in a Japanese population with cerebrovascular disease. Am J Hypertens 10(12 Pt 1): 1391–1395.
- 40. Lin JJ, Yueh KC, Lin GY, et al. (2000) Lack of association between angiotensin I-converting enzyme gene deletion polymorphism and cerebrovascular disease in Taiwanese. J Formos Med Assoc 99(12): 895–901.
- 41. Keramatipour M, McConnell RS, Kirkpatrick P, Tebbs S, Furlong RA, et al. (2000) The ACE I allele is associated with increased risk for ruptured intracranial aneurysms. J Med Genet 37(7): 498–500.
- 42. Catto A, Carter AM, Barrett JH, et al. (1996) Angiotensin-converting enzyme insertion/deletion polymorphism and cerebrovascular disease. Stroke 27(3): 435–440.
- 43. Dikmen M, Gunes HV, Degirmenci I, Ozdemir G, Basaran A (2006) Are the angiotensin-converting enzyme gene and activity risk factors for stroke? Arq Neuropsiquiatr 64(2A): 211–216.
- 44. Catto A, Carter A, Ireland H, et al. (1995) Factor V Leiden gene mutation and thrombin generation in relation to the development of acute stroke. Arterioscler Thromb Vasc Biol 15(6): 783–785.
- 45. Longstreth WT Jr, Rosendaal FR, Siscovick DS, et al. (1998) Risk of stroke in young women and two prothrombotic mutations: factor V Leiden and prothrombin gene variant (G20210A). Stroke 29(3): 577–580.
- 46. Corral J, Iniesta JA, Gonzalez-Conejero R, Villalon M, Vicente V (2001) Polymorphisms of clotting factors modify the risk for primary intracranial hemorrhage. Blood 97(10): 2979–2982.
- 47. Alberts MJ, Davis JP, Graffagnino C, et al. (1997) Endoglin gene polymorphism as a risk factor for sporadic intracerebral hemorrhage. Ann Neurol 41(5): 683–686.
- 48. Peters DG, Kassam AB, Chang YF (2005) A DNA sequence polymorphism in the endoglin gene is not associated with intracranial aneurysm or aneurysmal subarachnoid hemorrhage. Cerebrovasc Dis 20(2): 96–100.
- 49. Pera J, Slowik A, Dziedzic T, et al. (2005) Endoglin gene insertion polymorphism not associated with aneurysmal subarachnoid hemorrhage. J Neurosurg 102(5): 879–881.
- 50. Krischek B, Akagawa H, Tajima A, et al. (2007) The alanine/threonine polymorphism of the alpha-1-antichymotrypsin (SERPINA3) gene and ruptured intracranial aneurysms in the Japanese population. Cerebrovasc Dis 23(1): 46–49.
- 51. Slowik A, Borratynska A, Turaj W, et al. (2005) Alpha1-antichymotrypsin gene (SERPINA3) A/T polymorphism as a risk factor for aneurysmal subarachnoid hemorrhage. Stroke 36(4): 737–740.
- 52. Vila N, Obach V, Revilla M, Oliva R, Chamorro A (2000) Alpha(1)-antichymotrypsin gene polymorphism in patients with stroke. Stroke 31(9): 2103–2105.
- 53. Pera J, Slowik A, Dziedzic T, Szczudlik A (2006) SERPINA3 polymorphism is not associated with primary intracerebral hemorrhage in a Polish population. Stroke 37(3): 906–907.
- 54. Obach V, Revilla M, Vila N, Cervera AA, Chamorro AA (2001) Alpha(1)-antichymotrypsin polymorphism: a risk factor for hemorrhagic stroke in normotensive subjects. Stroke 32(11): 2588–2591.
- 55. Fu Y, Xie R, Wang Y, et al. (2002) [Association between alpha-1-antichymotrypsin gene polymorphism and cerebral hemorrhage]. Zhonghua Yi Xue Za Zhi 82(13): 915–917.
- 56. Song MK, Kim MK, Kim TS, et al. (2006) Endothelial nitric oxide gene T-786C polymorphism and subarachnoid hemorrhage in Korean population. J Korean Med Sci 21(5): 922–926.
- 57. Krischek B, Kasuya H, Akagawa H, et al. (2006) Using endothelial nitric oxide synthase gene polymorphisms to identify intracranial aneurysms more prone to rupture in Japanese patients. J Neurosurg 105(5): 717–722.
- 58. Khurana VG, Meissner I, Sohni YR, et al. (2005) The presence of tandem endothelial nitric oxide synthase gene polymorphisms identifying brain aneurysms more prone to rupture. J Neurosurg 102(3): 526–531.
- 59. Akagawa H, Kasuya H, Onda H, et al. (2005) Influence of endothelial nitric oxide synthase T-786C single nucleotide polymorphism on aneurysm size. J Neurosurg 102(1): 68–71.
- 60. Sazci A, Ergul E, Tuncer N, Akpinar G, Kara I (2006) Methylenetetrahydrofolate reductase gene polymorphisms are associated with ischemic and hemorrhagic stroke: Dual effect of MTHFR polymorphisms C677T and A1298C. Brain Res Bull 71(1–3): 45–50.
- 61. Nakata Y, Katsuya T, Takami S, et al. (1998) Methylenetetrahydrofolate reductase gene polymorphism: relation to blood pressure and cerebrovascular disease. Am J Hypertens 11(8 Pt 1): 1019–1023.
- 62. Li Z, Sun L, Zhang H, et al. (2003) Elevated plasma homocysteine was associated with hemorrhagic and ischemic stroke, but methylenetetrahydrofolate reductase gene C677T polymorphism was a risk factor for thrombotic stroke: a Multicenter Case-Control Study in China. Stroke 34(9): 2085–2090.
- 63. Fang X, Namba H, Akamine S, Sugiyama K (2005) Methylenetetrahydrofolate reductase gene polymorphisms in patients with cerebral hemorrhage. Neurol Res 27(1): 73–76.
- 64. Dikmen M, Ozbabalik D, Gunes HV, et al. (2006) Acute stroke in relation to homocysteine and methylenetetrahydrofolate reductase gene polymorphisms. Acta Neurol Scand 113(5): 307–314.
- 65. Slowik A, Dziedzic T, Pera J, Figlewicz DA, Szczudlik A (2005) Coagulation factor XIII VaI34Leu polymorphism in patients with small vessel disease or primary intracerebral hemorrhage. Cerebrovasc Dis 19(3): 165–170.
- 66. Gemmati D, Serino ML, Ongaro A, et al. (2001) A common mutation in the gene for coagulation factor XIII-A (VAL34Leu): a risk factor for primary intracerebral hemorrhage is protective against atherothrombotic diseases. Am J Hematol 67(3): 183–188.
- 67. Endler G, Funk M, Haering D, et al. (2003) Is the factor XIII 34Val/Leu polymorphism a protective factor for cerebrovascular disease? Br J Haematol 120(2): 310–314.
- 68. Corral J, Iniesta JA, Gonzalez-Conejero R, Villalon M, Rivera J, et al. (2000) Factor XIII Val34Leu polymorphism in primary intracerebral haemorrhage. Hematol J 1(4): 269–273.
- 69. Iniesta JA, Corral J, Gonzalez-Conejero R, Piqueras C, Vicente V (2003) Polymorphisms of platelet adhesive receptors: do they play a role in primary intracerebral hemorrhage? Cerebrovasc Dis 15(1–2): 51–55.
- 70. Carter AM, Catto AJ, Bamford JM, Grant PJ (1998) Platelet GP IIIa PlA and GP Ib variable number tandem repeat polymorphisms and markers of platelet activation in acute stroke. Arterioscler Thromb Vasc Biol 18(7): 1124–1131.
- 71. Iniesta JA, Gonzalez-Conejero R, Piqueras C, Vicente V, Corral J (2004) Platelet GP IIIa polymorphism HPA-1 (PlA) protects against subarachnoid hemorrhage. Stroke 35(10): 2282–2286.
- 72. Reiner AP, Kumar PN, Schwartz SM, et al. (2000) Genetic variants of platelet glycoprotein receptors and risk of stroke in young women. Stroke 31(7): 1628–1633.
- 73. Wald DS, Law M, Morris JK (2002) Homocysteine and cardiovascular disease: evidence on causality from a meta-analysis. BMJ 325(7374): 1202.
- 74. Keavney B, McKenzie C, Parish S, et al. (2000) Large-scale test of hypothesised associations between the angiotensin-converting-enzyme insertion/deletion polymorphism and myocardial infarction in about 5000 cases and 6000 controls. International Studies of Infarct Survival (ISIS) Collaborators. Lancet 355(9202): 434–442.
- 75. Lindqvist PG, Svensson PJ, Dahlback B, Marsal K (1998) Factor V Q506 mutation (activated protein C resistance) associated with reduced intrapartum blood loss–a possible evolutionary selection mechanism. Thromb Haemost 79(1): 69–73.
- 76. Donahue BS, Gailani D, Higgins MS, Drinkwater DC, George AL Jr (2003) Factor V Leiden protects against blood loss and transfusion after cardiac surgery. Circulation 107(7): 1003–1008.
- 77. McCarron MO, Nicoll JA, Stewart J, et al. (1999) The apolipoprotein E epsilon2 allele and the pathological features in cerebral amyloid angiopathy-related hemorrhage. Journal of Neuropathology & Experimental Neurology 58(7): 711–718.
- 78. Vonsattel JP, Myers RH, Hedley-Whyte ET, et al. (1991) Cerebral amyloid angiopathy without and with cerebral hemorrhages: a comparative histological study. Annals of Neurology 30(5): 637–649.
- 79. Nicoll JA, McCarron MO, Nicoll JA, McCarron MO (2001) APOE gene polymorphism as a risk factor for cerebral amyloid angiopathy-related hemorrhage. Amyloid 8: Suppl 151–55.
- 80. Brott TMD, Thalinger KMD, Hertzberg VP (1986) Hypertension as a Risk Factor for Spontaneous Intracerebral Hemorrhage. Stroke 17(6): 1078–1083.
- 81. Zhang C, Li J, Li L, et al. (2001) The study of PAI-1 promotor region gene polymorphism in cerebrovascular disease. Chung-Hua i Hsueh i Chuan Hsueh Tsa Chih 18(5): 383–387.
- 82. Dichgans M, Markus HS (2005) Genetic association studies in stroke: methodological issues and proposed standard criteria. Stroke 36(9): 2027–2031.
- 83. Hoggart CJ, Parra EJ, Shriver MD, et al. (2003) Control of confounding of genetic associations in stratified populations. Am J Hum Genet 72(6): 1492–1504.
- 84. Wellcome Trust Case Control Consortium (2007) Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 447(7145): 661–678.
- 85. Broderick JP, Brott T, Tomsick T, Huster G, Miller R (1992) The risk of subarachnoid and intracerebral hemorrhages in blacks as compared with whites. N Engl J Med 326(11): 733–736.
- 86. Qureshi AI, Suri MA, Safdar K, Ottenlips JR, Janssen RS, et al. (1997) Intracerebral hemorrhage in blacks. Risk factors, subtypes, and outcome. Stroke 28(5): 961–964.
- 87. Suzuki K, Kutsuzawa T, Takita K, et al. (1987) Clinico-epidemiologic study of stroke in Akita, Japan. Stroke 18(2): 402–406.
- 88. Yamada Y (2006) Identification of genetic factors and development of genetic risk diagnosis systems for cardiovascular diseases and stroke. Circ J 70(10): 1240–1248.
- 89. Yang Q, Khoury MJ, Botto L, Friedman JM, Flanders WD (2003) Improving the prediction of complex diseases by testing for multiple disease-susceptibility genes. Am J Hum Genet 72(3): 636–649.