Supplementary Material for: Genetic Susceptibility for Ischemic Infarction and Arteriolosclerosis Based on Neuropathologic Evaluations

<b><i>Background:</i></b> Recent genetic studies of stroke and related risk factors have identified a growing number of susceptibility loci; however, the relationship of these alleles to ischemic stroke is unknown. The challenge in finding reproducible loci of ischemic stroke susceptibility may be in part related to the etiologic heterogeneity in clinically defined stroke subtypes. In this study, we tested whether known single nucleotide polymorphisms (SNPs) associated with stroke or putative stroke risk factors are associated with neuropathologically defined micro- or macroscopic infarcts and with arteriolosclerosis. <b><i>Methods:</i></b> Measures of neuropathology and genotyping were available from 755 deceased participants from the Religious Orders Study and the Rush Memory and Aging Project. All donated brains were examined by a board-certified neuropathologist using standardized protocol for the presence of microscopic infarct, macroscopic infarct and arteriolosclerosis (lipohyalinosis). In primary analysis, 74 candidate SNPs previously associated (p < 5 × 10<sup>-8</sup>) with ischemic stroke or known risk factors, including atrial fibrillation (AF), hypertension, diabetes, low-density lipoprotein (LDL) level and carotid artery stenosis, were evaluated for association with neuropathologic endpoints. We performed a secondary exploratory analysis to include 93 additional SNPs associated with putative ischemic stroke risk factors including SNPs associated with high-density lipoprotein (HDL), triglyceride serum levels, myocardial infarction (MI), coronary artery disease and cerebral white matter disease. Regression models relating SNPs to cerebrovascular neuropathology were adjusted for age at death, gender and cohort membership. <b><i>Results:</i></b> The strongest associations seen for both macroscopic and microscopic infarcts were risk variants associated with diabetes. The diabetes risk variant <i>rs7578326</i> located near the <i>IRS1</i> locus was associated with both macroscopic (OR = 0.73, p = 0.011) and microscopic (OR = 0.71, p = 0.009) infarct pathology. Another diabetes susceptibility locus (<i>rs12779790</i>) located between the <i>calcium/calmodulin-dependent protein kinase ID (CAMK1D)</i> and <i>cell division cycle 123 homolog (CDC123)</i> genes is also associated with both macroscopic (OR = 1.40, p = 0.0292) and microscopic infarcts (OR = 1.43, p = 0.0285). The diabetes risk variant <i>rs864745</i> within <i>JAZF1</i> was associated with arteriolosclerosis (OR = 0.80, p = 0.014). We observed suggestive associations with the diabetes risk variant <i>rs7961581</i> (p = 0.038; between <i>TSPAN8 </i>and <i>LGR5</i>) and rs5215 (p = 0.043; <i>KCNJ11</i>), the LDL risk variant rs11206510 (p = 0.045; <i>PCSK9</i>), as well as the AF risk locus <i>ZFHX3.</i> The <i>CDKN2A/B</i> locus (<i>rs2383207</i>, 9p21), identified initially as a susceptibility allele for MI and recently implicated in large vessel stroke, was associated with macroscopic infarct pathology in our autopsy cohort (OR = 1.26, p = 0.031). <b><i>Conclusion:</i></b> Our results suggest replication of the candidate <i>CDKN2A/B</i> stroke susceptibility locus with directly measured macroscopic stroke neuropathology, and further implicate several diabetes and other risk variants with secondary, pleiotropic associations to stroke-related pathology in our autopsy cohort. When coupled with larger sample sizes, cerebrovascular neuropathologic phenotypes will likely be powerful tools for the genetic dissection of susceptibility for ischemic stroke.