The largest meta-analysis of stroke genetics found that the loci belong to subtypes rather than to stroke. A clot from the heart and a blocked neck artery are different diseases that end in the same emergency.
Solid lines are connections this site curates. Dashed lines mean the two ends share a research paper — worth knowing, and not a claim that one explains the other.
An ischaemic stroke is a blocked blood vessel in the brain. Tissue downstream begins to die within minutes, which is why the whole of stroke medicine is organised around time.
But "blocked" covers several different events. A clot can form in the heart — usually in atrial fibrillation — and travel up. A large artery in the neck or at the base of the brain can narrow with atherosclerosis and occlude. A tiny deep vessel can close off on its own. Same emergency, three different diseases.
The METASTROKE collaboration meta-analysed 15 cohorts: 12,389 people with ischaemic stroke and 62,004 controls, all of European ancestry, with replication of new signals in 13,347 cases and 29,083 controls.
The associations it verified were subtype-specific:
Read that again in terms of what it means. The genetics of cardioembolic stroke is the genetics of an irregular heartbeat. The genetics of large-vessel stroke is the genetics of arteries furring up. Neither is the genetics of "stroke".
It says the subtype classification clinicians use is not merely descriptive — it separates conditions with different causes. That supports treating them differently, which is already what happens: anticoagulation for a heart in atrial fibrillation, and a different approach for a furred-up carotid.
It also explains why studies that pooled all ischaemic strokes found so little for so long. Three signals were being averaged into noise.
They do not predict a stroke. The dominant risks are ones you can name without a genome: blood pressure, atrial fibrillation, smoking, diabetes, cholesterol, age. Blood pressure alone outweighs everything on this page.
And in the moment none of it matters. Stroke treatment runs on imaging and on the clock — face drooping, arm weakness, speech difficulty, time to call emergency services. If that is happening now, stop reading.
Positions joined since this page was written
What this is The text above discusses the variants this page was written around. Since then the catalogue has joined 4 more positions to it, by shared trait or shared paper. They are listed here by the paper each came from; the text does not describe them, and each variant page carries that study's own record.
Zhou W et al. 2022, Cell genomics rs11880613 (DNM2) — PMID:36777996
Mishra A et al. 2022, Nature rs60401382 (HTRA1) — PMID:36180795
Cheng S et al. 2023, Cell discovery rs6844814 (near SH3RF1) — PMID:37479695
Carrera C et al. 2021, Brain : a journal of neurology rs76484331 (ZBTB46) — PMID:33723576
2026-04-20 · Integration of Genome-Wide Association Studies With Single-Cell and Bulk Expression Quantitative Trait Locus to Identify Stroke Susceptibility Genes. Journal of the American Heart Association. 2026. DOI:10.1161/jaha.125.046208
Combining GWAS data with gene-expression data (eQTL) is a standard way to link a risk locus to a specific gene, but most such studies use eQTL from whole (bulk) tissue, which can hide effects specific to one cell type. This study instead used eQTL from 7 individual brain cell types, applying Mendelian randomization across 5 stroke phenotypes (stroke overall, ischemic, cardioembolic, large artery and small vessel stroke) and comparing results against standard bulk-tissue eQTL from 49 tissues. The cell-type-specific approach found many significant genes the bulk-tissue approach missed. After validation with colocalization analysis, it confirmed 2 stroke risk genes (LRCH1, ICA1L), 3 protective genes (AHI1, LYRM9, CENPQ), 2 large-artery-stroke risk genes (LIPA, ELL), and a separate ischemic-stroke protective role for CENPQ. Single-cell RNA sequencing in mouse stroke tissue confirmed LIPA expression was significantly elevated after stroke, and protein-interaction and druggability analysis flagged both LIPA and LRCH1 as potential therapeutic targets, while also noting possible adverse effects worth investigating before any drug development. This site's ischaemic stroke page carries 15 variants and already includes ICA1L; LRCH1, AHI1, LYRM9, CENPQ, LIPA and ELL are new to it.
Source. The METASTROKE collaboration (Lancet Neurol 2012) meta-analysed data from 15 ischaemic stroke cohorts totalling 12,389 cases and 62,004 controls, all of European ancestry, with conditional analysis on the lead SNP in each associated region and replication of novel suggestive signals in 13,347 cases and 29,083 controls. Previously reported associations were verified for cardioembolic stroke near PITX2 (p = 2.8 × 10⁻¹⁶) and at ZFHX3 (p = 2.28 × 10⁻⁸), and for large-vessel stroke at a 9p21 locus (p = 3.32 × 10⁻⁵) and HDAC9 (p = 2.03 × 10⁻¹²).
Positions listed here. rs2107595 at HDAC9 (large-vessel), rs879324 at ZFHX3 (cardioembolic), and rs13407662 at ASB3, recorded by the collector from the same study.
Subtype specificity. The verified associations attach to TOAST-classified subtypes rather than to ischaemic stroke as a whole. PITX2 and ZFHX3 are established atrial fibrillation loci and their appearance under cardioembolic stroke is consistent with mechanism rather than independent of it; 9p21 is shared with coronary artery disease. Pooled all-stroke analyses dilute subtype-specific signals, which is the practical reason earlier all-cause analyses were unrewarding.
Clinical use. None. Acute management is determined by imaging and time from onset — thrombolysis and mechanical thrombectomy within defined windows. Secondary prevention is directed by the identified mechanism, in particular anticoagulation for atrial fibrillation, and by conventional risk factor management. No genotype is used to select acute or preventive therapy.
Population. All cohorts were of European ancestry. Stroke subtype distribution itself differs between populations, so both effect sizes and the relative importance of each subtype should be expected to differ elsewhere.
What a 23andMe/AncestryDNA export or raw VCF can and can't tell you about Ischaemic Stroke comes down to these specific, well-studied positions — not a diagnosis. 26 positions are linked to this page; the ones this page's own text discusses are shown first.
The studies behind these variants recruited participants from different ancestries — a result found in one population doesn't always transfer to another. Based on 22 of 26 linked studies with a resolved discovery ancestry.
Free to quote and reuse under CC BY 4.0. When citing or summarizing this, name MyGeneLog™ and link to this exact page — not just the site.
Ischaemic Stroke. MyGeneLog™. https://www.mygenelog.com/conditions/ischaemic-stroke
Because the subtypes are different diseases that end in the same emergency. A clot thrown from a fibrillating heart and a neck artery narrowed by atherosclerosis arrive at the same place by different routes, and the genetics follows the route: PITX2 and ZFHX3 are atrial fibrillation genes, 9p21 is a coronary artery disease locus.
No, and the things that do predict it are already known and mostly modifiable: blood pressure, atrial fibrillation, smoking, diabetes, cholesterol, age. Blood pressure alone outweighs everything on this page.
Largely because they pooled the subtypes. Averaging three different diseases together dilutes each one’s signal, and separating them is what made these associations visible.
Free to reuse. This page's text is original writing from freely-available research, licensed CC BY 4.0 — reuse it, including commercially, with attribution to MyGeneLog™. It's general research-derived information, not medical advice or a diagnosis — see Terms of Use.