Person:

Rosand, Jonathan

The mechanism for hemorrhage enlargement in the brain, a key determinant of patient outcome following hemorrhagic stroke, is unknown. We performed computer-based stochastic simulation of one proposed mechanism, in which hemorrhages grow in “domino” fashion via secondary shearing of neighboring vessel segments. Hemorrhages were simulated by creating an initial site of primary bleeding and an associated risk of secondary rupture at adjacent sites that decayed over time. Under particular combinations of parameters for likelihood of secondary rupture and time-dependent decay, a subset of lesions expanded, creating a bimodal distribution of microbleeds and macrobleeds. Systematic variation of the model to simulate anticoagulation yielded increases in both macrobleed occurrence (26.9%, 53.2%, and 70.0% of all hemorrhagic events under conditions simulating no, low-level, and high-level anticoagulation) and final hemorrhage size (median volumes 111, 276, and 412 under the same three conditions), consistent with data from patients with anticoagulant-related brain hemorrhages. Reversal from simulated high-level anticoagulation to normal coagulation was able to reduce final hemorrhage size only if applied relatively early in the course of hemorrhage expansion. These findings suggest that a model based on a secondary shearing mechanism can account for some of the clinically observed properties of intracerebral hemorrhage, including the bimodal distribution of volumes and the enhanced hemorrhage growth seen with anticoagulation. Future iterations of this model may be useful for elucidating the effects of hemorrhage growth of factors related to secondary shearing (such as small vessel pathology) or time-dependent decay (such as hemostatic agents).

Genome-wide association studies (GWAS) have begun to identify the common genetic component to ischaemic stroke (IS). However, IS has considerable phenotypic heterogeneity. Where clinical covariates explain a large fraction of disease risk, covariate informed designs can increase power to detect associations. As prevalence rates in IS are markedly affected by age, and younger onset cases may have higher genetic predisposition, we investigated whether an age-at-onset informed approach could detect novel associations with IS and its subtypes; cardioembolic (CE), large artery atherosclerosis (LAA) and small vessel disease (SVD) in 6,778 cases of European ancestry and 12,095 ancestry-matched controls. Regression analysis to identify SNP associations was performed on posterior liabilities after conditioning on age-at-onset and affection status. We sought further evidence of an association with LAA in 1,881 cases and 50,817 controls, and examined mRNA expression levels of the nearby genes in atherosclerotic carotid artery plaques. Secondly, we performed permutation analyses to evaluate the extent to which age-at-onset informed analysis improves significance for novel loci. We identified a novel association with an MMP12 locus in LAA (rs660599; p = 2.5×10−7), with independent replication in a second population (p = 0.0048, OR(95% CI) = 1.18(1.05–1.32); meta-analysis p = 2.6×10−8). The nearby gene, MMP12, was significantly overexpressed in carotid plaques compared to atherosclerosis-free control arteries (p = 1.2×10−15; fold change = 335.6). Permutation analyses demonstrated improved significance for associations when accounting for age-at-onset in all four stroke phenotypes (p<0.001). Our results show that a covariate-informed design, by adjusting for age-at-onset of stroke, can detect variants not identified by conventional GWAS.