Circulation: Genomic and Precision Medicine ORIGINAL ARTICLE Subtype Specificity of Genetic Loci Associated With Stroke in 16 664 Cases and 32 792 Controls BACKGROUND: Genome-wide association studies have identified Matthew Traylor, PhD multiple loci associated with stroke. However, the specific stroke subtypes et al affected, and whether loci influence both ischemic and hemorrhagic stroke, remains unknown. For loci associated with stroke, we aimed to infer the combination of stroke subtypes likely to be affected, and in doing so assess the extent to which such loci have homogeneous effects across stroke subtypes. METHODS: We performed Bayesian multinomial regression in 16 664 stroke cases and 32 792 controls of European ancestry to determine the most likely combination of stroke subtypes affected for loci with published genome-wide stroke associations, using model selection. Cases were subtyped under 2 commonly used stroke classification systems, TOAST Downloaded from http://ahajournals.org by on February 18, 2022 (Trial of Org 10172 Acute Stroke Treatment) and causative classification of stroke. All individuals had genotypes imputed to the Haplotype Reference Consortium 1.1 Panel. RESULTS: Sixteen loci were considered for analysis. Seven loci influenced both hemorrhagic and ischemic stroke, 3 of which influenced ischemic and hemorrhagic subtypes under both TOAST and causative classification of stroke. Under causative classification of stroke, 4 loci influenced both small vessel stroke and intracerebral hemorrhage. An EDNRA locus demonstrated opposing effects on ischemic and hemorrhagic stroke. No loci were predicted to influence all stroke subtypes in the same direction, and only one locus (12q24) was predicted to influence all ischemic stroke subtypes. The full author list is available on page 313. CONCLUSIONS: Heterogeneity in the influence of stroke-associated Key Words: atherosclerosis loci on stroke subtypes is pervasive, reflecting differing causal pathways. ◼ cerebral hemorrhage ◼ chromosome ◼ dementia ◼ haplotype However, overlap exists between hemorrhagic and ischemic stroke, which may reflect shared pathobiology predisposing to small vessel arteriopathy. © 2019 The Authors. Circulation: Genomic and Precision Medicine is Stroke is a complex, heterogeneous disorder requiring tailored analytic published on behalf of the American strategies to decipher genetic mechanisms. Heart Association, Inc., by Wolters Kluwer Health, Inc. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution, and reproduction in any medium, provided that the original work is properly cited. https://www.ahajournals.org/journal/ circgen Circ Genom Precis Med. 2019;12:e002338. DOI: 10.1161/CIRCGEN.118.002338 July 2019 307 Traylor et al; Subtype Specificity of Stroke Genetic Loci T he burden of stroke on global healthcare and soci- well-defined phenotyping to determine if subtype spec- ety is substantial; it is consistently one of the lead- ificity or cross-subtype associations could be identified. ing causes of death and disability worldwide,1 and a major cause of cognitive impairment and dementia. METHODS However, there exist significant gaps in our understand- To minimize the possibility of unintentionally sharing informa- ing of the pathological processes that underlie the dis- tion that can be used to reidentify private information, a sub- ease. In recent years, GWAS (genome-wide association set of the data generated for this study are available from the studies) have made considerable advances in identifying database of Genotypes and Phenotypes (dbGaP) and can be genetic components underlying complex traits, in many accessed at https://www.ncbi.nlm.nih.gov/projects/gap/cgi- cases identifying novel disease pathways and treatments.2 bin/study.cgi?study_id=phs000615.v1.p1. Trinculo v0.96 is Characterizing the genetic component to stroke has available from https://sourceforge.net/projects/trinculo/files/. been challenging, in part due to clinical heterogeneity, MEGASTROKE data is available from http://megastroke.org. with at least 3 distinct major pathological processes (car- All contributing studies were approved by institutional review committees. All research participants contributing clinical dioembolism, large artery atherosclerosis, and small ves- and genetic samples for analysis in this study provided writ- sel disease) underlying the majority of ischemic strokes; ten informed consent. Full methods are provided in the Data and 2 processes underlying the majority of intracerebral Supplement. hemorrhagic stroke (small vessel disease and cerebral amyloid angiopathy).3,4 However, recent GWAS have made considerable advances; 32 independent genome- RESULTS wide significant loci were identified in the MEGASTROKE project.5 The majority of these loci were identified as After QC, there were up to 16 664 cases and 32 792 con- being associated with inclusive all stroke or ischemic trols remaining for analysis (Table 1). In the merged data stroke categories, rather than specific stroke subtypes. set, a binomial genome-wide analysis of all cases against This is in part due to study design, with much larger controls had a genomic inflation λ=1.09, while the link- samples for these broader categories and only a fraction age disequilibrium score regression (LDSCORE) intercept of stroke cases having detailed phenotyping. Indeed, value was 1.04,10 suggesting that the majority of inflation this finding is in contrast to earlier studies that identified was due to polygenicity and that any bias introduced by loci such as HDAC9, PITX2 as being associated with spe- merging the data sets was minimal. cific subtypes.6,7 To interpret genetic risk associations in Sixteen loci contained SNPs with log (Bayes factors) Downloaded from http://ahajournals.org by on February 18, 2022 the context of biological mechanisms, a pertinent ques- of at least 4 in analyses of alternative stroke classifica- tion is whether the newly identified stroke-associated tion systems: TOAST or CCS system (causative system). loci truly confer risk across all stroke subtypes, or wheth- We took these 16 loci forward for further model selec- er isolated or combinations of subtypes are affected. At tion. Plots for all loci under each classification system least one of the novel variants (on chromosome 1q22) are provided in Figures I through XVI in the Data Sup- shows association with both ischemic and hemorrhagic plement. For each of the 16 loci, we identified the most stroke, which might point to some shared mechanisms likely combination of associated phenotypes at each underlying these clinically distinct entities, which have locus (Figure 1, Table 2) based on model selection. A thus far been separated in genetic studies. comparison of odds ratios (ORs) for analyzed loci from Conventional approaches to GWAS, which employ within study analysis and subsequent meta-analysis Table 1. Sample Sizes across groups, do not enable detailed model compari- TOAST CCS son across different subgroups. In this analysis, we used Age Age multinomial logistic regression on well-characterized (mean Male (mean Male subjects with individual-level data to investigate the N [SD]) (%) N [SD]) (%) association of all identified genetic GWAS loci to date CES 3847 72 (14) 49 2826 75 (12) 44 with all stroke subtypes (cardioembolic [CES], large LAS 2803 68 (12) 65 2204 67 (12) 62 artery stroke [LAS], small vessel stroke [SVS], and intra- SVS 3976 64 (13) 62 3093 63 (13) 62 cerebral hemorrhage [ICH]), determining the most UND 4085 65 (16) 54 4013 65 (15) 53 likely combination of stroke subtypes affected at each locus. We performed our analysis using 2 established ICH 1953 71 (13) 53 1953 71 (13) 53 subtyping approaches: the TOAST (Trial of Org 10172 Controls 32 792 62 (17) 46 28 052 62 (17) 48 in Acute Stroke Treatment),8 and causative classifica- Age available not available for controls from WTCCC2 studies. tion of stroke (CCS) system,9 to provide a comprehen- CCS indicates causative classification of stroke system (causative system); sive account of these loci across available classification CES, cardioembolic stroke; ICH, intracerebral hemorrhage; LAS, large artery atherosclerotic stroke; SVS, small artery occlusion stroke; TOAST, Trial of Org systems. Our overall aim was to evaluate genetic loci 10172 Acute Stroke Treatment Classification System; and UND, stroke of identified in previous studies using stroke data sets with undetermined cause. Circ Genom Precis Med. 2019;12:e002338. DOI: 10.1161/CIRCGEN.118.002338 July 2019 308 Traylor et al; Subtype Specificity of Stroke Genetic Loci Figure 1. Stroke subtypes in best fitting model at each locus, for Causative Classification System (CCS) and TOAST (Trial of Org 10172 Acute Stroke Treatment Classification System) classification systems, with size weighted by association odds ratio. Results are presented for the 16 loci showing log (Bayes factor) >4 in causative classification system of stroke (CCS) or TOAST analyses. Classification/locus combinations in gray indicate that the locus did not reach log (Bayes factor) >4 in that analysis. CES indicates cardioembolic stroke; LAS, large artery stroke; SVS, small vessel stroke; and ICH, intracerebral hemorrhage. MEGASTROKE and the most recent ICH publication FOXF2 were associated with only one phenotype under Downloaded from http://ahajournals.org by on February 18, 2022 with those from our analysis showed high consistency particular classification systems but did not show con- (r2=0.95, Figure XVII in the Data Supplement) despite sistency across TOAST and CCS (Figures II, III, IV, and slightly differing samples. Linkage Disequilibrium (LD) IX in the Data Supplement). For TSPAN2, which was values between our lead and previously published SNPs for the 16 loci in this analysis are provided in Table I in previously identified as being associated with LAS,13 the Data Supplement. the best-fit model also included CES under CCS, albeit For 7 loci, the combination of phenotypes most likely with a much weaker effect than LAS (rs17479660; CES, to be influenced by the lead genetic variant at the loci OR=1.08; LAS, OR=1.19 under CCS). Echoing previous included both ischemic and hemorrhagic stroke sub- results, the locus showed much stronger significance types. These are shown in Figure 2. At four of these loci: under CCS classifications than under TOAST (Figure XV EDNRA, 1q22, MMP12, and SH3PXD2A, the ischemic subtype included SVS, highlighting shared mechanisms in the Data Supplement). underlying ICH, and SVS, likely through predisposition For COL4A2, the strongest association found under to cerebral small vessel disease. At the EDNRA locus, TOAST was for rs9515201. The most likely model con- the direction of association for ICH was opposite to tained ICH (OR=1.14) and SVS (OR=1.13), consistent that for LAS and SVS, pointing to contrasting influence with findings from previous analyses.12 However, under on ischemic and hemorrhagic stroke risk. We explored CCS an alternate SNP, rs1927349, was the strongest whether ICH-associated loci were specific to deep or lobar ICH. As in previous reports,11,12 associations at associated. No association with SVS was observed, and 1q22 and COL4A2 appear to be specific to deep ICH, a weak association with CES was observed instead. with no effect in lobar ICH. For other regions, the evi- Reasons for this discrepancy between CCS and TOAST dence for specificity was more equivocal (Table II in the are not immediately clear, but nonoverlapping samples Data Supplement). between the 2 classification systems are a likely factor. For 4 loci HDAC9, PITX2, ZFHX3, and ANK2, only one The mean (SD) number of stroke subtypes affected phenotype was affected by the lead variant (Figure 1, Figures X, XIII, XVI, and V in the Data Supplement) in at each locus were 1.88 (0.89) under TOAST and 1.69 the most likely configuration across all classification (0.87) under CCS. Under CCS, the most common com- systems. Several other loci 9p21, 12q24, 16q24, and bination of affected subtypes was SVS and ICH (4 loci). Circ Genom Precis Med. 2019;12:e002338. DOI: 10.1161/CIRCGEN.118.002338 July 2019 309 Traylor et al; Subtype Specificity of Stroke Genetic Loci Table 2. Lead SNPs, Association Statistics, and Affected Stroke Table 2. Continued Subtypes for Each Locus Lead SNP Subtypes in Lead SNP Subtypes in [Best Best-Fitting [Best Best-Fitting Locus Model] log OR (SE) log BF Model Locus Model] log OR (SE) log BF Model HDAC9 rs2107595 0.04 (0.04) SVS 19.2 LAS 1q22 rs2758603 0.10 (0.03) SVS 4.0 SVS and ICH [TOAST] −0.08 (0.06) ICH [CCS] 0.11 (0.05) ICH 0.05 (0.04) CES 0.02 (0.03) CES 0.06 (0.03) UNK 0.07 (0.03) UNK 0.27 (0.04) LAS 0.07 (0.03) LAS LINC01492 rs10990643 −0.02 (0.04) SVS 4.1 LAS and ICH 9p21 rs1412830 0.08 (0.03) SVS 5.7 LAS and SVS [TOAST] 0.12 (0.06) ICH [TOAST] 0.07 (0.04) ICH 0.03 (0.04) CES −0.01 (0.03) CES 0.01 (0.03) UNK 0.03 (0.03) UNK 0.17 (0.04) LAS 0.14 (0.03) LAS MMP12 rs470234 0.09 (0.04) SVS 8.7 LAS, SVS, and 12q24 rs10774624 0.10 (0.03) SVS 5.8 CE, LAS, and [CCS] ICH 0.17 (0.06) ICH [TOAST] SVS −0.03 (0.05) ICH 0.04 (0.04) CES 0.07 (0.03) CES 0.03 (0.04) UNK 0.07 (0.03) UNK 0.20 (0.04) LAS 0.10 (0.03) LAS PITX2 rs2723334 0.0 (0.04) SVS 48.0 CES 16q24 rs12445022 0.13 (0.03) SVS 5.8 LAS and SVS [TOAST] 0.08 (0.06) ICH [TOAST] 0.05 (0.05) ICH 0.29 (0.04) CES −0.01 (0.03) CES 0.03 (0.03) UNK 0.07 (0.03) UNK −0.03 (0.04) LAS 0.07 (0.03) LAS SH3PXD2A rs10883922 0.13 (0.03) SVS 6.0 SVS and ICH ANK2 rs149538932 0.07 (0.03) SVS 6.4 ICH [CCS] 0.16 (0.05) ICH [CCS] Downloaded from http://ahajournals.org by on February 18, 2022 0.18 (0.05) ICH 0.02 (0.03) CES 0.04 (0.03) CES 0.02 (0.03) UNK 0.08 (0.03) UNK 0.04 (0.03) LAS 0.02 (0.03) LAS TSPAN2 rs7537796 −0.05 (0.03) SVS 6.8 CES and LAS CDK6 rs4272 [CCS] 0.05 (0.04) SVS 8.5 LAS and ICH [CCS] −0.06 (0.05) ICH 0.10 (0.05) ICH 0.06 (0.03) CES 0.07 (0.03) CES −0.02 (0.03) UNK 0.12 (0.03) UNK 0.14 (0.03) LAS 0.14 (0.04) LAS ZFHX3 rs67329386 −0.02 (0.03) SVS 13.8 CES COL4A2 rs1927349 −0.02 (0.03) SVS 5.0 CES and ICH [TOAST] −0.05 (0.05) ICH [CCS] 0.16 (0.05) ICH 0.20 (0.03) CES 0.08 (0.03) CES 0.02 (0.03) UNK 0.04 (0.03) UNK 0.00 (0.03) LAS 0.02 (0.03) LAS CCS indicates causative classification system of stroke; CES, cardioembolic EDNRA rs17612742 0.09 (0.04) SVS 10.5 CES, LAS, SVS, stroke; ICH, intracerebral hemorrhage; LAS, large artery stroke; UNK, Stroke [CCS] and ICH of undetermined cause; log BF, log transform of Bayes factor; log OR, log −0.23 (0.06) ICH transform of odds ratio; SVS, small vessel stroke; TOAST, Trial of Org 10172 −0.08 (0.04) CES Acute Stroke Treatment Classification System. −0.00 (0.04) UNK 0.13 (0.04) LAS DISCUSSION FOXF2 rs11242678 0.13 (0.03) SVS 7.4 LAS and SVS [CCS] We performed a large-scale genetic analysis, charac- −0.05 (0.05) ICH terizing the effects of established stroke risk loci with 0.07 (0.03) CES ischemic and hemorrhagic stroke subtypes in up to 0.09 (0.03) UNK 16 664 cases and 32 792 controls. Our main findings 0.09 (0.04) LAS are 2-fold. First, for the vast majority of loci studied, (Continued ) multiple but never all stroke subtypes were affected Circ Genom Precis Med. 2019;12:e002338. DOI: 10.1161/CIRCGEN.118.002338 July 2019 310 Traylor et al; Subtype Specificity of Stroke Genetic Loci Downloaded from http://ahajournals.org by on February 18, 2022 Figure 2. Local plots showing associations with regions conferring risk of ischemic and hemorrhagic stroke and odds ratios for all stroke subtypes. A, 1q22 region; (B) Endothelin receptor type A region; (C) Cyclin dependent kinase 6 CDK6 region; (D) Long Intergenic Non-Protein Coding RNA 1492 region; (E) SH3 And PX Domains 2A region; (F) Matrix metalloproteinase 12 region; and (G) COL4A2 collagen type IV alpha 2 chain region. Results are presented for the classification system in which the locus showed strongest significance. Stroke subtypes in bold indicate those included in the best fitting model and, therefore, predicted to be influenced by the lead genetic variant, based on Bayesian model selection. CCS indicates causative classification of stroke system; CES, cardioem- bolic stroke; LAS, large artery atherosclerotic stroke; SVS, small vessel stroke; ICH, intracerebral hemorrhage; TOAST, Trial of Org 10172 Acute Stroke Treatment Classification System; and UND, stroke of undetermined cause. at the locus. Only one locus (12q24) was assumed to well-established loci14), and HDAC9 which is associ- influence all ischemic stroke subtypes. This indicates ated with large vessel stroke. Under TOAST, the FOXF2 that although these loci were identified in analyses of locus was associated solely with SVS. However, under inclusive stroke phenotypes, in the main, their effects CCS, LAS was also implicated. For CCS, the 9p21 locus are specific to particular combinations of stroke sub- was predicted to influence only LAS. However, under types. The mean number of subtypes affected was TOAST, SVS was also implicated. Our analyses sug- 1.88 for TOAST and 1.69 for CCS classification sys- gest that ANK2 confers risk of stroke predominantly tems. Notable exceptions were the PITX2 and ZFHX3 through its influence on ICH. We were unable to iden- loci, which were associated with cardioembolic stroke tify any loci for which the most likely model included most likely through atrial fibrillation (for which they are all stroke phenotypes in the same direction and only Circ Genom Precis Med. 2019;12:e002338. DOI: 10.1161/CIRCGEN.118.002338 July 2019 311 Traylor et al; Subtype Specificity of Stroke Genetic Loci one (12q24) which for which the most likely model or hemorrhagic and as such warrants further detailed included all ischemic stroke subtypes. investigation. Second, we find evidence that several loci influence Some loci were notably more significant when phe- both hemorrhagic and ischemic stroke. This was evident notyped using CCS; SH3XPD2A, MMP12, TSPAN2, for 7 loci in total (1q22, COL4A2, EDNRA, LINC01492, FOXF2, and EDNRA, which might point to CCS having MMP12, SH3PXD2A, and CDK6). Under CCS, 4 loci greater accuracy and, therefore, utility in stroke GWA (SH3PXD2A, MMP12, EDNRA, and 1q22) influenced studies. However, the opposite was also true for others: both SVS and ICH, highlighting shared mechanisms 16q24, HDAC9. We note that some differences may underlying small vessel disease. Previous GWAS analy- be due to the fact that not all individuals were sub- ses have tended to separate ischemic and hemorrhagic typed under both CCS and TOAST; the TOAST cohort stroke on the basis of presumed differing causes. Our was a least 20% larger. A detailed discussion of the results suggest that including hemorrhagic alongside relative merits of TOAST and CCS is beyond the scope ischemic stroke in multiphenotype analyses will provide of this article, but our results highlight that the impor- further insights. tance of collecting individual phenotypic qualities that For one locus EDNRA (endothelin receptor type A), make up the etiologic subtypes in genetic studies of the association with ICH was in the opposite direction stroke so that associated loci can be more systemati- to the ischemic stroke subtypes, suggesting opposing cally examined. risk mechanisms. This locus has previously been associ- Our study has several strengths. The data set was a ated with a variety of vascular phenotypes, including large stroke population, including intracerebral hemor- coronary artery disease, carotid plaques, and peripheral rhage and ischemic stroke cases, the majority of which arterial disease (all in concordant direction with isch- were subtyped under both TOAST and CCS. We had full emic stroke), as well as intracranial aneurysm (in con- access to genotype-level data enabling us full control cordant direction with intracerebral hemorrhage).15–18 over all analyses. The implementation of a multinomial The locus has also been associated with migraine in regression approach enabled us to systematically assess candidate gene studies,19 but this has not been vali- which stroke subtypes were likely to be affected at dated in GWA studies and is likely a false positive.20 ETA each locus, which would not be formally possible under (EDNRA encodes the type A receptor) for ET-1 (endo- standard binomial regression approaches which analyze thelin-1), a potent vasoconstrictor with proinflamma- each stroke subtype separately. Ultimately, mechanistic tory effects. ETA-specific antagonists increase nitric studies will be required to determine the influence of Downloaded from http://ahajournals.org by on February 18, 2022 oxide-mediated endothelium-dependent relaxation, associated genetic variants, but analyses such as this reduce ET-1 levels and inhibit atherosclerosis in mice,21 have utility in directing the focus and model systems suggesting that higher levels of ETA are proatherogenic: suitable for such follow-up studies. consistent with the observation that higher ETA levels Similarly, there are limitations. We present results are observed in atherosclerotic plaques.22 Based on this, for the most likely combination of stroke phenotypes one might expect the EDNRA risk variant (C allele of affected at each locus: the best-fitting model. We had rs17612742 in this study) to lead to increased risk of limited statistical power to determine with statistical ischemic stroke through elevated ETA levels. Indeed, in certainty that this was the correct model; significantly GWA studies of intracranial aneurysm, the susceptibil- larger samples would be required to achieve this. One ity variant (in LD with the T allele of rs17612742 in our consequence of this is that there remains the poten- study) was shown to result in higher transcription factor tial that some associations are due to random varia- binding affinity, likely resulting in repression of the tran- tion rather than true biological differences. It would, scriptional activity of EDNRA.17 This suggests that car- therefore, be prudent to treat some of the findings riers of the C allele have lower levels of EDNRA, which here as preliminary until confirmed in larger samples. consequently higher ET-1 levels and greater suscepti- Due to the challenges of performing these analyses bility to atherosclerosis. The reason why for carriers of across different ancestry populations, and as we only T allele lower levels of ETA might promote intracranial had a small number of non-European ancestry ICH aneurysm and intracerebral hemorrhage is not imme- cases available which could lead to overfitting, we diately obvious, but several mechanisms are possible. performed analyses in European populations only. The Levels of ET-1 have been linked to vascular remodel- results can, therefore, not be generalized to all popu- ing, an important process underlying ICH and IA23,24; lations. Repeating these analyses once sufficient data subtle changes in this process induced by altered avail- from other ancestral groups are available should be ability of ETA is one such mechanism. Deep ICH and highly prioritized to ensure advancements in the field ischemic SVS arise due to the same arteriopathy that are made for all ancestral groups. In all analyses we arises in the deep perforating arteries of the brain. The assume there is a single causal variant at the locus, EDNRA variant in this study points to a mechanism that which may not be true in all cases. Our analyses are influences whether the resulting pathology is ischemic based on use of a default prior, which has been used Circ Genom Precis Med. 2019;12:e002338. DOI: 10.1161/CIRCGEN.118.002338 July 2019 312 Traylor et al; Subtype Specificity of Stroke Genetic Loci in many genetic studies. An alternative is to derive an University of Edinburgh (C.L.M.S.). Usher Institute of Population Health Sciences and Informatics, Nine Bioquarter, Edinburgh, United Kingdom (C.L.M.S.). Divi- empirical prior from associated genetic loci. As more sion of Endocrinology, Diabetes and Nutrition, Department of Medicine (H.X., loci are identified as being associated with stroke, this B.D.M.,) and Department of Neurology (S.J.K., J.W.C), University of Maryland will become a more realistic possibility and should be School of Medicine. Geriatrics Research and Education Clinical Center, Baltimore Veterans Administration Medical Center (B.D.M.). Department of Neurology, Vet- explored in future analyses. erans Affairs Medical Center, Baltimore, MD (J.W.C., S.J.K). Channing Division of Network Medicine and Division of Women’s Health, Department of Medicine, Brigham and Women’s Hospital, Boston, MA (K.R.). Neurovascular Research Unit, CONCLUSIONS Department of Neurology (J.J.-C.), Program in Inflammation and Cardiovascu- lar Disorders (J.J.-C.), Institut Municipal d’Investigacio´ Medica-Hospital del Mar, Our findings suggest that although large scale genome- Universitat Autonoma de Barcelona, Spain. Department of Neurology, Medical University of Graz, Austria (R. Schmidt). Neuroscience Institute, Saint Francis wide studies of broad all stroke or all ischemic stroke Medical Center, School of Health and Medical Sciences, Seton Hall University, phenotypes are able to identify multiple associations; South Orange, NJ (R.P.G.). Department of Neurology, Miller School of Medicine, it should not be assumed that such associations con- University of Miami, FL (R. Sacco, T.R.). Psychiatric Genetics Unit, Group of Psy- chiatry, Mental Health and Addiction, Vall d’Hebron Research Institute (VHIR), fer risk equally across stroke subtypes. Heterogeneity Universitat Autònoma de Barcelona (M.R.). Department of Psychiatry, Hospital in the influence of genetic variants on different stroke Universitari Vall d’Hebron (M.R.). Biomedical Network Research Center on Mental subtypes is the norm, not the exception. The multino- Health (CIBERSAM), Instituto de Salud Carlos III, Barcelona, Spain (M.R.). Institute mial regression approach used here provided insights for Stroke and Dementia Research, Klinikum der Universität München, Ludwig- Maximilians-Universität München (M.D., R.M.). Munich Cluster for Systems Neu- into the etiological stroke subtypes most prominently rology (SyNergy), Germany (M.D.). Department of Neurology, Radiology, and influenced by genetic variants at these loci—a prereq- Biomedical Engineering, Washington University School of Medicine, St Louis, MO uisite to decide on the most appropriate model systems (J.-M.L.). Department of Neurology and Rehabilitation Medicine and Comprehen- sive Stroke Center, University of Cincinnati, OH (D.W.). to choose for further mechanistic studies. Stroke is a complex, heterogeneous disorder: our findings high- Acknowledgments light the ongoing need for large, well-phenotyped case MT and RM designed the experiments. MT and MC performed the imputa- collections and tailored analytic strategies to decipher tions. MT performed the statistical analyses. MT, CDA, LCARJ, HSM, DW, and the underlying genetic mechanisms. RM wrote the first draft of the manuscript. All authors read and approved the final manuscript. ARTICLE INFORMATION Sources of Funding Received August 24, 2018; accepted June 14, 2019. This work was supported by a British Heart Foundation Programme Grant Downloaded from http://ahajournals.org by on February 18, 2022 The Data Supplement is available at https://www.ahajournals.org/doi/ (RG/16/4/32218). The National Institute of Neurological Disorders and Stroke— suppl/10.1161/CIRCGEN.118.002338. Stroke Genetics Network (NINDS-SIGN) study was funded by the US National Institute of Neurological Disorders and Stroke, National Institutes of Health (U01 NS069208 and R01 NS100178). Collection of the UK Young Lacunar Authors Stroke DNA Study (DNA Lacunar) was primarily supported by the Wellcome Matthew Traylor, PhD; Christopher D. Anderson, MD; Loes C.A. Rutten-Jacobs, Trust (WT072952) with additional support from the Stroke Association (TSA PhD; Guido J. Falcone, MD; Mary E. Comeau, MA; Hakan Ay, MD; Cathie 2010/01). Genotyping of the DNA Lacunar samples was supported by a Stroke L.M. Sudlow, MD; Huichun Xu, PhD; Braxton D. Mitchell, PhD; John W. Cole, Association Grant (TSA 2013/01). The principal funding for the WTCCC2 MD; Kathryn Rexrode, MD; Jordi Jimenez-Conde, MD; Reinhold Schmidt, MD; stroke study was provided by the Wellcome Trust, as part of the Wellcome Raji P. Grewal, MD; Ralph Sacco, MD; Marta Ribases, MD; Tatjana Rundek, Trust Case Control Consortium 2 project (085475/B/08/Z and 085475/Z/08/Z MD; Jonathan Rosand, MD; Martin Dichgans, MD; Jin-Moo Lee, MD; Carl D. and WT084724MA). H.S. Markus is supported by a National Institute for Health Langefeld, PhD; Steven J. Kittner, MD; Hugh S. Markus, DM; Daniel Woo, MD; Research (NIHR) Senior Investigator award, and his work is supported by the Rainer Malik, PhD; on behalf of the NINDS Stroke Genetics Network (SiGN) and Cambridge Universities NIHR Comprehensive Biomedical Research Center. Dr International Stroke Genetics Consortium (ISGC) Anderson is supported by NIH R01NS103924 and K23NS086873. Dr. Kittner is supported by VA RR&D N1699-R and NINDS R01NS100178 and R01NS105150. Correspondence Matthew Traylor, PhD, Department of Clinical Neurosciences, University of Disclosures Cambridge, Stroke Research Group, R3, Box 83, Cambridge, CB2 0QQ, United Dr Anderson has consulted for ApoPharma, Inc. The other authors report no Kingdom. Email mt628@medschl.cam.ac.uk conflicts. Affiliations Stroke Research Group, Department of Clinical Neurosciences, University of REFERENCES Cambridge (M.T., H.S.M.). William Harvey Research Institute, Barts and The Lon- 1. GBD 2016 DALYs and HALE Collaborators. Global, regional, and national don School of Medicine and Dentistry, Queen Mary University of London (M.T.). disability-adjusted life-years (DALYs) for 333 diseases and injuries and Center for Genomic Medicine (C.D.A., J.R.), Stroke Service (H.A.), A.A. Martinos healthy life expectancy (HALE) for 195 countries and territories, 1990– Center for Biomedical Imaging, Department of Radiology (H.A.), J. Philip Kistler 2016: a systematic analysis for the Global Burden of Disease Study 2016. Stroke Research Center, Department of Neurology (C.D.A., J.R.), Division of Neu- Lancet. 2017;390:1260–1344. rocritical Care and Emergency Neurology, Department of Neurology (C.D.A., J.R.), 2. Visscher PM, et al. 10 years of GWAS discovery: biology, function, Massachusetts General Hospital, Boston. Program in Medical and Population and translation. Am J Hum Genet. 2017;101:5–22. doi: 10.1016/j. Genetics, Broad Inst, Cambridge, MA (C.D.A., J.R.). German Center for Neuro- ajhg.2017.06.005 degenerative Diseases, Population Health Sciences, Bonn, Germany (L.C.A.R.-J.). 3. Martini SR, et al. Risk factors for intracerebral hemorrhage differ ac- Division of Neurocritical Care and Emergency Neurology, Department of Neurol- cording to hemorrhage location. Neurology. 2012;79:2275–2282. doi: ogy, Yale School of Medicine, New Haven, CT (G.J.F.). Department of Biostatistical 10.1212/WNL.0b013e318276896f Sciences, Division of Public Health Sciences, School of Medicine, Wake Forest 4. Hankey GJ. Stroke. Lancet. 2017;389:641–654. doi: 10.1016/S0140- University, Winston-Salem, NC (M.E.C., C.D.L.). Center for Clinical Brain Sciences, 6736(16)30962-X Circ Genom Precis Med. 2019;12:e002338. DOI: 10.1161/CIRCGEN.118.002338 July 2019 313 Traylor et al; Subtype Specificity of Stroke Genetic Loci 5. Malik R, et al; AFGen Consortium; Cohorts for Heart and Aging Research 13. NINDS Stroke Genetics Network (SiGN); International Stroke Genetics in Genomic Epidemiology (CHARGE) Consortium; International Genomics Consortium (ISGC). Loci associated with ischaemic stroke and its subtypes of Blood Pressure (iGEN-BP) Consortium; INVENT Consortium; STARNET; (SiGN): a genome-wide association study. Lancet Neurol. 2016;15:174–184. BioBank Japan Cooperative Hospital Group; COMPASS Consortium; EPIC- 14. Roselli C, et al. Multi-ethnic genome-wide association study for atrial fibril- CVD Consortium; EPIC-InterAct Consortium; International Stroke Genet- lation. Nat Genet. 2018;50:1225–1233. doi: 10.1038/s41588-018-0133-9 ics Consortium (ISGC); METASTROKE Consortium; Neurology Working 15. Bis JC, et al; CARDIoGRAM Consortium. Meta-analysis of genome-wide Group of the CHARGE Consortium; NINDS Stroke Genetics Network association studies from the CHARGE consortium identifies common (SiGN); UK Young Lacunar DNA Study; MEGASTROKE Consortium. Multi- variants associated with carotid intima media thickness and plaque. Nat ancestry genome-wide association study of 520,000 subjects identifies 32 Genet. 2011;43:940–947. doi: 10.1038/ng.920 loci associated with stroke and stroke subtypes. Nat Genet. 2018;50:524– 16. Howson JMM, et al; CARDIoGRAMplusC4D; EPIC-CVD. Fifteen new risk 537. doi: 10.1038/s41588-018-0058-3 loci for coronary artery disease highlight arterial-wall-specific mecha- 6. Bellenguez C, et al;International Stroke Genetics Consortium (ISGC); Well- nisms. Nat Genet. 2017;49:1113–1119. doi: 10.1038/ng.3874 come Trust Case Control Consortium 2 (WTCCC2). Genome-wide asso- 17. Low SK, et al. Genome-wide association study for intracranial aneurysm ciation study identifies a variant in HDAC9 associated with large vessel in the japanese population identifies three candidate susceptible loci and ischemic stroke. Nat Genet. 2012;44:328–333. a functional genetic variant at EDNRA. Hum Mol Genet. 2012;21:2102– 7. Traylor M, et al; Australian Stroke Genetics Collaborative, Wellcome 2110. doi: 10.1093/hmg/dds020 Trust Case Control Consortium 2 (WTCCC2); International Stroke Ge- 18. Matsukura M, et al. Genome-wide association study of peripheral arte- netics Consortium. Genetic risk factors for ischaemic stroke and its rial disease in a japanese population. PLoS One. 2015;10:e0139262. doi: subtypes (the METASTROKE collaboration): a meta-analysis of ge- 10.1371/journal.pone.0139262 nome-wide association studies. Lancet Neurol. 2012;11:951–962. doi: 19. Tzourio C, et al. Association between migraine and endothelin type A 10.1016/S1474-4422(12)70234-X receptor (ETA -231 A/G) gene polymorphism. Neurology. 2001;56:1273– 8. Adams HP Jr, et al. Classification of subtype of acute ischemic stroke. 1277. doi: 10.1212/wnl.56.10.1273 Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 20. Gormley P, et al; International Headache Genetics Consortium. Meta-anal- in acute stroke treatment. Stroke. 1993;24:35–41. 9. Ay H, et al. A computerized algorithm for etiologic classification of ysis of 375,000 individuals identifies 38 susceptibility loci for migraine. Nat ischemic stroke: the causative classification of stroke system. Stroke. Genet. 2016;48:856–866. doi: 10.1038/ng.3598 2007;38:2979–2984. doi: 10.1161/STROKEAHA.107.490896 21. Barton M, et al. Endothelin ETA receptor blockade restores NO-mediated 10. Bulik-Sullivan BK, et al; Schizophrenia Working Group of the Psychiat- endothelial function and inhibits atherosclerosis in apolipoprotein E- ric Genomics Consortium. LD Score regression distinguishes confound- deficient mice. Proc Natl Acad Sci U S A. 1998;95:14367–14372. doi: ing from polygenicity in genome-wide association studies. Nat Genet. 10.1073/pnas.95.24.14367 2015;47:291–295. doi: 10.1038/ng.3211 22. Ramzy D, et al. Elevated endothelin-1 levels impair nitric oxide homeo- 11. Woo D, et al; International Stroke Genetics Consortium. Meta-analysis of stasis through a PKC-dependent pathway. Circulation. 2006;114(1 genome-wide association studies identifies 1q22 as a susceptibility locus suppl):I319–I326. doi: 10.1161/CIRCULATIONAHA.105.001503 for intracerebral hemorrhage. Am J Hum Genet. 2014;94:511–521. doi: 23. Penn DL, et al. The role of vascular remodeling and inflammation in the 10.1016/j.ajhg.2014.02.012 pathogenesis of intracranial aneurysms. J Clin Neurosci. 2014;21:28–32. 12. Rannikmäe K, et al; Stroke Genetics Network (SiGN), METASTROKE Col- doi: 10.1016/j.jocn.2013.07.004 laboration, and International Stroke Genetics Consortium (ISGC). COL4A2 24. Amiri F, et al. Endothelium-restricted overexpression of human en- is associated with lacunar ischemic stroke and deep ICH: meta-analyses dothelin-1 causes vascular remodeling and endothelial dysfunction. Downloaded from http://ahajournals.org by on February 18, 2022 among 21,500 cases and 40,600 controls. Neurology. 2017;89:1829– Circulation. 2004;110:2233–2240. doi: 10.1161/01.CIR.0000144462. 1839. doi: 10.1212/WNL.0000000000004560 08345.B9 Circ Genom Precis Med. 2019;12:e002338. DOI: 10.1161/CIRCGEN.118.002338 July 2019 314