Alternate approach to stroke phenotyping identifies a genetic risk locus for small vessel stroke

Joanna Von Berg,Sara L Pulit,Braxton D Mitchell,Patrick F Mcardle,Jeroen De Ridder,Sander W Van Der Laan,Rainer Malik,Bradford B Worrall,Steven J Kittner

doi:10.1038/s41431-020-0580-5

Abstract

Ischemic stroke (IS), caused by obstruction of cerebral blood flow, is one of the leading causes of death. While neurologists agree on delineation of IS into three subtypes (cardioembolic stroke (CES), large artery stroke (LAS), and small vessel stroke (SVS)), several subtyping systems exist. The most commonly used systems are TOAST (Trial of Org 10172 in Acute Stroke Treatment) and CCS (Causative Classification System for Stroke), but agreement is only moderate. We have compared two approaches to combining the existing subtyping systems for a phenotype suited for a genome-wide association study (GWAS). We used the NINDS Stroke Genetics Network dataset (SiGN, 11,477 cases with CCS and TOAST subtypes and 28,026 controls). We defined two new phenotypes: the intersect, for which an individual must be assigned the same subtype by CCS and TOAST; and the union, for which an individual must be assigned a subtype by either CCS or TOAST. The union yields the largest sample size while the intersect yields a phenotype with less potential misclassification. We performed GWAS for all subtypes, using the original subtyping systems, the intersect, and the union as phenotypes. In each subtype, heritability was higher for the intersect compared with the other phenotypes. We observed stronger effects at known IS variants with the intersect compared with the other phenotypes. With the intersect, we identify rs10029218:G>A as an associated variant with SVS. We conclude that this approach increases the likelihood to detect genetic associations in ischemic stroke.

Highlights

To help uncover genetic associations with Ischemic stroke (IS) that as yet have gone undetected, we defined new IS phenotypes based on three existing subtyping systems (CCSc, CCS phenotypic (CCSp), and TOAST)
We find that the largest proportion of phenotypic variance explained by single-nucleotide polymorphisms (SNPs) is in the intersect phenotype
Our overlap analyses show that, for each subtype, the phenotype definitions each have a unique set of significantly associated SNPs, but that there is a small set of SNPs that is shared among all definitions, with concordant direction of effect and similar trend in magnitude of effect

Summary

1234567890();,: 1234567890();,: Introduction

The majority of IS are typically grouped into three subtypes: cardioembolic stroke (CES), frequently occurring in people with atrial fibrillation; large artery stroke (LAS), caused by eroded or ruptured atherosclerotic plaques in arteries; and small vessel stroke (SVS), caused by a blockage of one of the small vessels in the brain These subtypes seem to be genetically distinct: genome-wide association studies (GWAS) in ISs have identified single-nucleotide polymorphisms (SNPs) that primarily associate with a specific IS subtype [2]. Previous work indicates that TOAST and CCS have moderate, but not high, concordance in assigning subtypes in patients: agreement is lowest in SVS (κ = 0.56) and highest in LAS (κ = 0.71) [6] Both subtyping systems still place more than one third of all samples into a heterogeneous ‘undetermined’ category [6]. We find stronger effects at known associations for the intersect compared with the union, and we validate a previously suspected association in SVS through GWAS of the intersect phenotype

Results

Discussion

Methods

Compliance with ethical standards