Meschia, James F.

Objectives: Identification of gene variants of stroke subtypes is important for the development of tailored ischemic stroke therapies among various ethnic groups. Valid and reliable determination of ischemic stroke subtype is essential for achieving this goal and to standardize a classification scheme across multi-center studies and different populations. Causative Classification System for Ischemic Stroke (CCS) is a novel computerized subclassification tool developed to improve reliability and accuracy of classifying stroke types. The CCS algorithm relies on both phenotypic and causative stroke variables. A Hispanic subset of the SiGN, an important and distinct target population with greater risk of certain stroke subtypes, was evaluated with Trial of Org 10172 in Acute Stroke Treatment (TOAST) and CCS and the agreement between the two classification systems was analyzed. Methods: Over 6000 subjects at 15 sites across US and Europe were enrolled, with TOAST and CCS locally adjudicated. Blood collection and central data quality control (10% central readjudication) were performed on all participants. A subset of Hispanics was analyzed for the purpose of this study and the agreement between the TOAST and CCS were assessed by kappa statistic. Findings: Hispanics (n=595, 10.9%) compared to non-Hispanics (n=5457) were more likely to be younger (63.7 vs. 64.0), male (55% vs. 46%) and have fewer of the traditional stroke risk factors HTN (54% vs. 64%), Afib (11% vs. 14%), DM(23% vs. 25%), CAD(16% vs. 20%) and smoking(19% vs. 22%). While the TOAST showed no differences between stroke subtypes for Hispanic vs. non-Hispanics, in CCS, Hispanics were classified with more of large vessel (22% vs. 20%), cardioembolic (37% vs. 30%) and small vessel strokes (13% vs. 9%) and fewer with undetermined etiology (28% vs. 40%) as compared to non-Hispanics. TOAST and CCS offered moderate correlation across all stroke types in Hispanics: kappa of 0.66 for large artery atherosclerosis, 0.58 for cardioembolic, and 0.58 for small artery occlusion. Conclusion: CCS offers a more sensitive and accurate system for subphenotyping of strokes in Hispanics who tended to have relatively fewer risk factors and unclassified strokes. Further studies correlating the two classification systems and their relation to genotyping data are warranted.

Objectives The Carotid Revascularization Endarterectomy versus Stenting Trial (CREST) showed no difference in the composite endpoint of stroke, myocardial infarction, or death between carotid artery stenting (CAS) and carotid endarterectomy (CEA) in patients with symptomatic or asymptomatic carotid stenosis (CS). We compared restenosis or occlusion, and repeat revascularization, between CAS and CEA over 2 years of follow-up. Methods Restenosis, occlusion, and repeat revascularization were assessed at 1, 6, 12 and 24 months post-randomization. Hemodynamically significant restenosis (≥70% diameter reduction) was defined by a peak systolic velocity (PSV) ≥300 cm/second on standardized duplex ultrasonography (DUS), occlusion by an absence of flow within the target artery on DUS, and repeat revascularization by any additional procedure (CEA, angioplasty or CAS) performed on the index artery. Studies were performed in CREST-certified laboratories and interpreted in the CREST Ultrasound Core Lab. Patients included in this report were those who received their assigned treatment within 30 days of randomization and had an ultrasound reviewed at the Core Lab (n=2191). Treatment differences were assessed using proportional hazards models adjusting for age, sex, and symptomatic status. Results In the analytic cohort, 1086 patients received CAS and 1105 received CEA. Over 2 years 113 patients developed restenosis, 56 (rate 5.8%) in the CAS group and 57 (rate 5.8%) in the CEA group; and 8 developed an occlusion, 3 (rate 0.3 %) in the CAS group and 5 (rate 0.5%) in the CEA group. The combined restenosis-occlusion rate was 6.0% (n=58) for CAS and 6.3% (n=62) for CEA (HR=0.90, 95% CI=0.63-1.29, p=0.58). Forty-three patients of the 2191 underwent repeat revascularization (20 CAS, 23 CEA, p=0.69) of which 28 had restenosis by our criteria and so were included ( Figure ). Stroke occurred in 13 (4 CAS; 9 CEA) of the 120 patients with restenosis or occlusion; 1 of the 4 CAS strokes occurred after restenosis was detected, and 5 of 9 CEA strokes occurred after restenosis was detected. Conclusions This analysis of carotid restenosis is the largest reported to date from any randomized clinical trial. Restenosis and occlusion were infrequent and similar up to 2 years following CAS or CEA among 2191 patients. The rates of revascularization likewise did not differ between CAS and CEA. Follow-up out to 10 years is ongoing. Figure. Kaplan-Meier curves of restenosis and occlusion over 2 years.

Background: The Stroke Genetics Network (SiGN) funded by the NINDS aims to identify genetic risk factors in ischemic stroke using whole-genome association studies (GWAS). High quality phenotyping is crucial to successful application of GWAS. As a heterogenous disorder, stroke poses specific challenges. The Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification is a broadly used, but its validity is challenged especially when performed by multiple investigators with differing interpretations of the system. The Causative Classification System for Ischemic Stroke (CCS) system is a new, web-based, and computerized algorithm that integrates clinical, diagnostic, and etiologic stroke characteristics in an evidence-based manner ( ccs.mgh.harvard.edu ) to generate subtypes. Methods: In planning the SiGN proposal, a sample of 20 coded charts were collected from a subset of participating studies to assess feasibility of central adjudication and comparability to study-specific TOAST. Two central adjudicators reviewed all records and generated TOAST and CCS subtypes. These were compared to study-specific TOAST subtype and the CCS phenotype generated for SiGN by local trained adjudicators. CCS data is now available for 7134 included cases using both a 5 and a 7 category system as defined in the table . Results: All 4 phenotypes were available for 115 ischemic stroke cases from 6 studies in SiGN. Basic demographics were 54% women, 63% white, and median age between 65-74. Table 1 provides the agreement between the various subtypes. Table 2 describes the types of disagreement. Conclusions: Central adjudication with only two adjudicators and curated medical records yielded more consistent subtyping independent of phenotyping system. The agreement for TOAST was higher than published rates by independent groups (∼0.50). In contrast, the agreement for CCS was lower than previously published (0.85-0.95). Site adjudicators' familiarity with TOAST and inexperience with CCS may contribute. Although CCS is an automated algorithm and has a number of user friendly features, our findings suggest that formal training and certification process before starting to use CCS may be worthwhile to achieve optimal benefit from the system.

Background and Purpose: Family-based methods for estimating heritability cannot discriminate between shared genetic and shared environmental exposures. Recently, methods have been developed for estimating heritability in population samples using genome-wide SNPs. We used the approach developed by Visscher and colleagues to estimate the heritability of ischemic stroke in Caucasian subjects. In addition to evaluating the overall heritability of ischemic stroke, we assessed whether stroke heritability varies by age, gender, and stroke subtype. Methods: Using publicly available software (GCTA and PLINK), we estimated ischemic stroke heritability stratified by age and gender using genome-wide association (GWA) data from three Caucasian ischemic stroke studies: Ischemic Stroke Genetics Study (ISGS), Bio-Repository of DNA in Stroke (BRAINS), and Genetics of Early-Onset Stroke (GEOS). Weighted means of site-specific heritability point estimates were combined according to a standard fixed effects model. Results: Conclusions: A SNP-based approach may be useful in discerning differences in ischemic stroke heritability between different cohorts and subtypes. Overall, our analysis estimated ischemic stroke heritability to be 31% (SE = 7%), with a suggestion of higher heritability for younger cases. Small vessel stroke showed the highest heritability (58 ± 19%), with cardioembolic showing the lowest heritability (16 ± 14%). It should be emphasized that heritability estimates are population-specific and that the method used only reflects the heritability captured by common SNP variants measured in GWA studies, and not phenotypic variability explained by rare variants.

Background: Identification of clinical and imaging characteristics of ischemic stroke that can predict the underlying etiology is critical for accurate stroke subtyping. We sought to investigate the relationship between stroke characteristics and stroke etiology in 7,591 patients in the NINDS Stroke Genetics Network Study. Methods: We determined the probability of having a major etiology in the presence of certain stroke characteristics that were previously linked to specific stroke mechanisms. We excluded patients with multiple competing major etiologies. All patients had brain imaging, intra- and extra-cranial vascular imaging, and complete cardiac assessment. We defined major etiologies according to the Causative Classification of Stroke System Criteria without reference to the stroke characteristics under study. Results: History of 1 ipsilateral ischemic events within the last month, the presence of internal watershed infarcts or multiple, ipsilateral, unilateral, and temporally separate infarcts were associated with large artery atherosclerosis (Table). Multiple acute infarcts in different cerebral circulations were more common in patients with major cardiac source of embolism as well as in those with uncommon/cryptogenic causes. Stereotypic lacunar TIAs within the last week and lacunar syndrome were associated with lacunar infarction Discussion: Clinical and imaging stroke characteristics appear to predict stroke etiologies. This information may find utility in developing algorithms for a guided stroke evaluation as well as help identify the most likely causative etiology in the presence of multiple competing etiologies.

Purpose: The safety of revascularization for asymptomatic carotid stenosis, and the efficacy of medical therapy for stroke prevention have improved. Therefore, results of prior randomized trials may not apply to current treatment decisions. The NINDS-funded CREST-2 will compare carotid endarterectomy and intensive medical therapy (IMT) versus IMT alone (n=1240), and carotid stenting and IMT versus IMT alone (n=1240) in asymptomatic patients with≥70% stenosis. Materials & Methods: CREST-2 consists of two parallel randomized clinical trials to be conducted at a target of ≈120 centers, including within NINDS StrokeNet. The composite primary outcome is stroke or death during the peri-procedural period or ipsilateral ischemic stroke thereafter up to 4 years. Blinded assessment of cognition will be done periodically. Centrally directed IMT includes tight control of blood pressure (systolic target <140 mm Hg) and cholesterol (LDL target <70 mg/dl) as well as lifestyle coaching. Results: As of June 12, 2015, 94 centers have been approved by the Site Selection Committee. Credentialing is ongoing, with 198 approved surgeons and 64 approved interventionists; 124 additional conditionally approved interventionists will be able to submit additional cases for review under the CREST-2 Registry. The Centers for Medicare and Medicaid will offer CAS reimbursement for Registry enrollees. As of June 12, 2015, there are 39 actively enrolling centers, and 37 patients have been randomized. Conclusion: CREST-2 is designed to identify the best approach for asymptomatic carotid stenosis. The first patient was randomized in December, 2014. An update will be provided regarding the numbers of patients randomized, centers certified, as well as surgeons and interventionists fully approved. Registration: ClinicalTrials.gov Identifier: NCT02089217

Background: Levels of the pro-inflammatory cytokines interleukin (IL)-6 and IL-8 and the anti-inflammatory cytokine IL-10 are altered in acute stroke patients compared to controls. However, results for IL-10 are inconsistent (higher in stroke patients compared to controls in some studies and decreased in others) and very few studies have examined associations of these biomarkers with risk of incident stroke. Methods: We examined associations of baseline levels of IL-6, IL-8 and IL-10 with stroke risk factors and risk of incident stroke in 1,572 white and black men and women from the REGARDS Study; an observational cohort of 30,239 adults followed for 5 years. Among those without prebaseline stroke, stroke cases (n=592, 53% men, 59% white, mean age 70 years) were compared to a cohort random sample (n=980, 44% men, 58% white, mean age 65 years). We used Cox proportional hazards models to examine associations of biomarkers with incident stroke. Hazard ratios (HR; 95% confidence intervals) for highest compared to lowest quartile for each biomarker are presented. Results: Baseline IL-6 was significantly higher in incident stroke cases compared to the cohort sample (p<0.001) while IL-8 and IL-10 levels did not vary significantly (p>0.05 for both). Adjusting for age, sex and race, IL-6 was higher in blacks compared to whites and higher in current smokers compared to never/former smokers (all p≤0.01). IL-6 was inversely associated with physical activity, alcohol use and education (all p≤0.01). IL-8 and IL-10 did not vary significantly with the risk factors examined. Adjusting for age, sex and race, the HR for risk of incident stroke in those in the highest compared to lowest quartile of IL-6 was 2.4 (1.6-3.4). HRs for IL-8 and IL-10 were 1.5 (1.0-2.1) and 1.4 (1.0-1.9), respectively. Adjusting for Framingham stroke risk factors and history of coronary heart disease, only IL-6 remained significantly associated with stroke risk (HR 2.0; 1.3-3.2). HRs for IL-8 and IL-10 were 1.1 (0.7-1.6) and 1.2 (0.8-1.8), respectively. IL-6 remained significantly associated with stroke risk when C-reactive protein was added to the model (HR 1.7; 1.1-2.7). Associations did not differ by race. Conclusions: In this population-based sample of US black and white adults, IL-6, but not IL-8 or IL-10, was associated with stroke risk factors and risk of incident stroke. Further study is needed on the clinical utility of IL-6 measurement in stroke risk assessment.

Background: Fundamental advances in stroke care will require pooling imaging phenotype data from multiple centers, to complement the current aggregation of genomic, environmental, and clinical information. Sharing clinically acquired MRI data from multiple hospitals is challenging due to inherent heterogeneity of clinical data, where the same MRI series may be labeled differently depending on vendor and hospital. Furthermore, the de-identification process may remove data describing the MRI series, requiring human review. However, manually annotating the MRI series is not only laborious and slow but prone to human error. In this work, we present a recurrent convolutional neural network (RCNN) for automated classification of the MRI series. Methods: We randomly selected 1000 subjects from the MRI-GENetics Interface Exploration study and partitioned them into 800 training, 100 validation and 100 testing subjects. We categorized the MRI series into 24 groups (see Table). The RCNN used a modified AlexNet to extract features from 2D slices. AlexNet was pretrained on ImageNet photographs. Since clinical MRI are 3D and 4D, a gated recurrent unit neural network was used to aggregate information from multiple 2D slices to make the final prediction. Results: We achieved a classification accuracy (correct/total cases) of 99.8%, 98.5% and 97.5% on the training, validation and testing set, respectively. The averaged F1-score (percent overlap between predicted cases and actual cases) over all categories were 99.8% 98.2% and 94.4% on the training, validation and testing set. Conclusion: We showed that automated annotation of MRI series by repurposing deep-learning techniques used for photographic image recognition tasks is feasible. Such methods can be used to facilitate high throughput curation of MRI data acquired across multiple centers and enable scientifically productive collaboration by researchers and, ultimately enhancing big data stroke research.

Introduction: Left ventricular assist device (LVAD) is known to extend survival in patients with advanced heart failure but it is associated with intracranial cranial hemorrhage (ICH). In the last decade, the use of LVAD substantially increased along with advances in LVAD technology, but there is limited data regarding the trends of ICH risk and predictors in patients with LVAD. Methods: We included the patients aged≥18 years with primary diagnosis of LVAD hospitalized in the US from 2005 to 2014 using the National Inpatient Sample. We computed the survey weighted percentages with ICH across the 10-year study period and assessed whether the proportions changed over time. Predictors of ICH were evaluated using the multivariable logistic regression model. All analyses were appropriately adjusted for the survey design variables to account for the complex survey design. Results: Of 33,246 hospitalizations, 568 (1.7%) had ICH. The number of LVAD placement has increased from 873 in year 2005 to 5,175 in year 2014. However, the risk of ICH remained largely unchanged from 1.7% to 2.3% (linear trend: p=0.604). The adjusted odds of ICH was increased with the presence one of the following variables: female sex (odds ratio [OR]: 1.58, 95% confidence interval [CI] 1.03-2.43), history of ischemic stroke (OR: 3.13, 95% CI 1.86-5.28), and Charlson Comorbidity Index (CCI) of ≥3 (OR: 77.40, 95% CI 10.03-597.60). Conclusions: Despite advances in LVAD technology and its increased use in patients with advanced heart failure, the risk of ICH has remained relatively unchanged. High CCI and history of ischemic stroke were associated with higher odds of ICH in patients with LVAD. With the advent of magnetically levitated centrifugal-flow LVAD, future trend of ICH may change and further research efforts are needed to prevent ICH in patients with LVAD.

Introduction: Robust leptomeningeal collaterals protect the brain from ischemia, yet the determinants of poor collaterals have not been established. We aimed to identify predictors of poor collaterals in patients with large vessel occlusive stroke. Methods: Consecutive patients with large vessel occlusive stroke who underwent thrombectomy at 2 academic medical centers (derivation cohort/validation cohort) from 2012-2017 were included. Poor collaterals were defined as ≤50% filling on CT angiogram. Pre-specified predictors of poor collaterals were analyzed with adjusted logistic regression models: demographics (age, sex, race), risk factors (hypertension, diabetes, atrial fibrillation, smoking), hemodynamics (blood pressure, cardiac output, troponin), metabolic (hyperglycemia, hyponatremia, uremia), hematologic (anemia, leukocytosis, thrombocytosis) and neuroimaging evidence of chronic cerebral small-vessel disease and chronic steno-occlusive arterial disease. Results: Of the 248 total patients, poor collaterals were detected in 107 (43%). Mean age was 71±14y vs. 63±16y in patients with poor vs good collaterals, respectively (p<.001). There were 136 patients in the derivation cohort; 54 (48%) patients with poor collaterals. Multivariable modeling identified that older age (OR 1.8, p<.001), hyperglycemia (OR 2.4, p=.029) and white matter hyperintensities (WMH) (OR 6.1, p=.003) were independent predictors of poor collaterals in the derivation cohort. The validation cohort consisted of 136 patients; 53 (39%) had poor collaterals, of which older age (OR 1.4, p=.015), thrombocytosis (OR 24.0, p=.012), and WMH (OR 6.7, p=.018) were significant predictors of poor collaterals. Age and WMH were the only significant predictors of poor collaterals across the derivation and validation cohorts. When pooled the cohorts, older age (OR=1.5, p<.001), thrombocytosis (OR=7.6, p=.001) and WMH (OR=6.4, p<.001) were significant predictors of poor collaterals. Conclusions: We have identified that advanced age, WMH, and thrombocytosis are predictors of poor collaterals. Future studies to explore the effect of vascular aging on collateral therapeutics are indicated.