ORIGINAL RESEARCH published: 25 June 2020 doi: 10.3389/fneur.2020.00577 Diffusion-Weighted Imaging, MR Angiography, and Baseline Data in a Systematic Multicenter Analysis of 3,301 MRI Scans of Ischemic Stroke Edited by: Patients—Neuroradiological Review Emmanuel Carrera, University of Geneva, Switzerland Within the MRI-GENIE Study Reviewed by: Marios K. Georgakis, Mattias Drake 1,2 , Petrea Frid 3,4 , Björn M. Hansen 1,2 , Ona Wu 5 , Anne-Katrin Giese 6,7 , Krankenhaus der Universität Markus D. Schirmer 6,8 , Kathleen Donahue 6 , Lisa Cloonan 6 , Robert E. Irie 5 , München, Germany Mark J. R. J. Bouts 5 , Elissa C. McIntosh 5 , Steven J. T. Mocking 5 , Adrian V. Dalca 5,9 , Fabien Scalzo, Ramesh Sridharan 9 , Huichun Xu 10 , Eva Giralt-Steinhauer 11 , Lukas Holmegaard 12 , University of California, Los Angeles, Katarina Jood 12 , Jaume Roquer 11 , John W. Cole 13 , Patrick F. McArdle 10 , United States Joseph P. Broderick 14 , Jordi Jiménez-Conde 11 , Christina Jern 15 , Brett M. Kissela 14 , *Correspondence: Dawn O. Kleindorfer 14 , Robin Lemmens 16,17 , James F. Meschia 18 , Tatjana Rundek 19 , Johan Wassélius Ralph L. Sacco 19 , Reinhold Schmidt 20 , Pankaj Sharma 21,22 , Agnieszka Slowik 23 , johan.wasselius@med.lu.se Vincent Thijs 24,25 , Daniel Woo 14 , Bradford B. Worrall 26 , Steven J. Kittner 13 , Braxton D. Mitchell 10,27 , Jonathan Rosand 5,6,28 , Polina Golland 9 , Arne Lindgren 29,30 , Specialty section: Natalia S. Rost 6 and Johan Wassélius 1,2* on behalf of the Stroke Genetics This article was submitted to Network (SiGN) the MRI-Genetics Interface Exploration (MRI-GENIE) Study Stroke, 1 a section of the journal Department of Clinical Sciences Lund, Radiology, Lund University, Lund, Sweden, 2 Department of Radiology, Frontiers in Neurology Neuroradiology, Skåne University Hospital, Lund, Sweden, 3 Department of Clinical Sciences Lund, Neurology, Lund University, Lund, Sweden, 4 Department of Neurology and Rehabilitation Medicine, Neurology, Skåne University Hospital, Received: 30 October 2019 Malmö, Sweden, 5 Department of Radiology, Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Accepted: 19 May 2020 Hospital (MGH), Harvard Medical School, Charlestown, MA, United States, 6 Department of Neurology, Massachusetts Published: 25 June 2020 General Hospital, Harvard Medical School, Boston, MA, United States, 7 Program in Medical and Population Genetics, Broad Citation: Institute of MIT and Harvard, Cambridge, MA, United States, 8 Department of Population Health Sciences, German Centre Drake M, Frid P, Hansen BM, Wu O, for Neurodegenerative Diseases (DZNE), Bonn, Germany, 9 Computer Science and Artificial Intelligence Laboratory, MIT, Giese A-K, Schirmer MD, Donahue K, Cambridge, MA, United States, 10 Division of Endocrinology, Diabetes and Nutrition, Department of Medicine, University of Cloonan L, Irie RE, Bouts MJRJ, Maryland School of Medicine, Baltimore, MD, United States, 11 Department of Neurology, Neurovascular Research Group McIntosh EC, Mocking SJT, Dalca AV, (NEUVAS), IMIM-Hospital del Mar (Institut Hospital del Mar d’Investigacions Mèdiques), University at Autonoma de Sridharan R, Xu H, Giralt-Steinhauer E, Barcelona, Barcelona, Spain, 12 Institute of Neuroscience and Physiology, The Sahlgrenska Academy at University of Holmegaard L, Jood K, Roquer J, Gothenburg, Gothenburg, Sweden, 13 Department of Neurology, University of Maryland School of Medicine and Veterans Cole JW, McArdle PF, Broderick JP, Affairs Maryland Health Care System, Baltimore, MD, United States, 14 Department of Neurology and Rehabilitation Medicine, Jiménez-Conde J, Jern C, Kissela BM, University of Cincinnati College of Medicine, Cincinnati, OH, United States, 15 Department of Laboratory Medicine, Institute of Kleindorfer DO, Lemmens R, Biomedicine, The Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden, 16 Department of Neurosciences, Meschia JF, Rundek T, Sacco RL, Experimental Neurology and Leuven Research Institute for Neuroscience and Disease (LIND), Leuven, Belgium, 17 Schmidt R, Sharma P, Slowik A, Department of Neurology, VIB, Vesalius Research Center, Laboratory of Neurobiology, University Hospitals Leuven, Leuven, Thijs V, Woo D, Worrall BB, Kittner SJ, Belgium, 18 Department of Neurology, Mayo Clinic, Jacksonville, FL, United States, 19 Department of Neurology and the Mitchell BD, Rosand J, Golland P, Evelyn F. McKnight Brain Institute, Miller School of Medicine, University of Miami, Miami, FL, United States, 20 Clinical Division Lindgren A, Rost NS and Wassélius J of Neurogeriatrics, Department of Neurology, Medical University Graz, Graz, Austria, 21 Institute of Cardiovascular Research, (2020) Diffusion-Weighted Imaging, Royal Holloway University of London (ICR2UL), Egham, United Kingdom, 22 Ashford and St Peter’s Hospital, Surrey, MR Angiography, and Baseline Data in United Kingdom, 23 Department of Neurology, Jagiellonian University Medical College, Krakow, Poland, 24 Stroke Division, a Systematic Multicenter Analysis of Florey Institute of Neuroscience and Mental Health, Heidelberg, VIC, Australia, 25 Department of Neurology, Austin Health, 3,301 MRI Scans of Ischemic Stroke Heidelberg, VIC, Australia, 26 Departments of Neurology and Public Health Sciences, University of Virginia, Charlottesville, VA, Patients—Neuroradiological Review United States, 27 Geriatric Research and Education Clinical Center, Veterans Administration Medical Center, Baltimore, MD, Within the MRI-GENIE Study. United States, 28 Henry and Allison McCancer Center for Brain Health and Center for Genomic Medicine, Massachusetts Front. Neurol. 11:577. General Hospital, Boston, MA, United States, 29 Department of Clinical Sciences Lund, Neurology, Lund University, Lund, doi: 10.3389/fneur.2020.00577 Sweden, 30 Department of Neurology and Rehabilitation Medicine, Neurology, Skåne University Hospital, Lund, Sweden Frontiers in Neurology | www.frontiersin.org 1 June 2020 | Volume 11 | Article 577 Drake et al. The MRI-GENIE Study—Baseline Radiological Findings 11 Background: Magnetic resonance imaging (MRI) serves as a cornerstone in defining stroke phenotype and etiological subtype through examination of ischemic stroke lesion appearance and is therefore an essential tool in linking genetic traits and stroke. Building on baseline MRI examinations from the centralized and structured radiological assessments of ischemic stroke patients in the Stroke Genetics Network, the results of the MRI-Genetics Interface Exploration (MRI-GENIE) study are described in this work. Methods: The MRI-GENIE study included patients with symptoms caused by ischemic stroke (N = 3,301) from 12 international centers. We established and used a structured reporting protocol for all assessments. Two neuroradiologists, using a blinded evaluation protocol, independently reviewed the baseline diffusion-weighted images (DWIs) and magnetic resonance angiography images to determine acute lesion and vascular occlusion characteristics. Results: In this systematic multicenter radiological analysis of clinical MRI from 3,301 acute ischemic stroke patients according to a structured prespecified protocol, we identified that anterior circulation infarcts were most prevalent (67.4%), that infarcts in the middle cerebral artery (MCA) territory were the most common, and that the majority of large artery occlusions 0 to 48 h from ictus were in the MCA territory. Multiple acute lesions in one or several vascular territories were common (11%). Of 2,238 patients with unilateral DWI lesions, 52.6% had left-sided infarct lateralization (P = 0.013 for χ² test). Conclusions: This large-scale analysis of a multicenter MRI-based cohort of AIS patients presents a unique imaging framework facilitating the relationship between imaging and genetics for advancing the knowledge of genetic traits linked to ischemic stroke. Keywords: stroke, imaging, MRI, phenotype, DWI BACKGROUND insights can be gained in understanding the AIS characteristics on MRI across a multiethnic, hospital-based cohort. Neuroimaging analysis is the cornerstone of modern stroke Building on the extensive data obtained from clinical management, with computed tomography (CT) and magnetic stroke patients within the National Institute of Neurological resonance imaging (MRI) being the main modalities (1). Disorders and Stroke–funded Stroke Genetics Network Compared to CT, MRI has greater sensitivity and versatility and (SiGN), the MRI-Genetics Interface Exploration (MRI-GENIE) provides an earlier and more precise method for delineating study offers such an opportunity by providing clinical MRIs acute ischemic lesions (2) and also a more detailed assessment from 12 hospital-based cohorts of AIS patients enrolled of chronic small vessel disease. internationally. A detailed description of the design and Magnetic resonance imaging differentiates acute ischemic rationale for both these studies has been published previously lesions from subacute or chronic infarcts better than CT (2, 3). (11, 12). The MRI-GENIE initiative has already resulted in Magnetic resonance imaging can aid in the early etiological a number of pioneering studies in automated volumetric assessment of acute ischemic stroke (AIS), by differentiating MRI analysis (13–15), posterior-circulation stroke (16), and among hypoperfusion injury, lacunar infarction, and cortical several ongoing projects using the imaging data to link genetic vessel occlusion (4). The addition of magnetic resonance traits, such as the recently discovered RABEP2-gene (17) with angiography (MRA) further assists in the characterization of the phenotypic outcome. cerebrovascular phenotype (5). These properties have made MRI We capitalized on the robust MRI-GENIE database of a common tool in the acute management of ischemic stroke to 3,301 ischemic stroke cases with genome-wide genotyping and guide acute intervention, optimize secondary prevention, and MRI scans by systematically reviewing and characterizing the improve outcomes prediction. multimodal clinical scans for specific vascular anatomy and Several population-based MRI initiatives, such as the pathology, ischemic lesion location, lateralization, multiplicity, Rotterdam and Framingham studies, have provided valuable and other characteristics according to a predefined structured insights into the prevalence of cerebral infarcts (6, 7), including reporting protocol. In this article, we report the initial findings the distribution with regard to laterality (8, 9). However, there from the structured neuroradiological analysis of MRI scans are few large MRI-based stroke registries (10), and additional within MRI-GENIE. Frontiers in Neurology | www.frontiersin.org 2 June 2020 | Volume 11 | Article 577 Drake et al. The MRI-GENIE Study—Baseline Radiological Findings METHODS major cerebral arterial branches, and angiographic data to categories for predefined vascular segments. When assigning MRI-GENIE is an international multicenter collaborative study lesions to the predefined anatomical categories basal and vascular of ischemic stroke cases for whom clinical MRI scans were segments, a generally accepted map of the major cerebral vascular obtained on admission for AIS, in addition to genome-wide territories was used (Supplemental Figure 1). Ischemic lesions genotyping subsequently obtained. It is a major international from DWI were further assigned as cortical and/or subcortical, initiative to explore the genetic architecture of MRI traits in AIS single/multiple, and—for the supratentorial subcortical lesions— patients. As MRI-GENIE was based on the SiGN collaboration, lacunar or nonlacunar. A single subcortical, supratentorial lesion participating sites within SiGN were invited to participate. smaller than 1.5 cm was further defined as lacunar. Volumetric Participating centers have committed to contribute all available analysis of the DWIs was recently published (14). MRI for the patients included within the SiGN collaboration. To The reviewers determined vessel stenosis (stenosis visually date, 12 of the 24 centers have contributed data to the project gauged to >50%) and occlusion on MRA images. The reviewers (Figure 1 and Supplemental Table 1). A Scientific Steering also determined the proportion of patients with large artery Committee oversees the MRI-GENIE study and critically reviews occlusion (LAO) based on information on occlusion and lesion project proposals facilitating collaborative efforts. location on the DWI sequences. Unexpected findings such as tumors and aneurysms were Patients noted, but not categorized further. The structured reporting All participants provided signed informed consent. Contributed protocol is detailed in Supplemental Table 2. data included demographic and stroke risk factor data, availability of genome-wide genotyping data, and clinical MRI Statistics scan. Stroke subtypes were classified in the parent SiGN Study Data were processed using SPSS Statistics (version 24; IBM according to the Causative Classification System for Ischemic Corporation, Armonk, NY, USA). Descriptive statistics, Stroke (CCS), and Trial of Org 10 172 in Acute Stroke Treatment including means with standard deviations (SDs) and medians (TOAST) classification systems (18). with interquartile range (IQR), were reported where applicable. We deidentified the clinical MRI examinations from The χ² test was used to compare proportions of left- vs. right- contributing centers, uploaded the DICOM format images in the sided cerebral infarcts among all patients, as well as patients with MRI-GENIE image repository, and systematically annotated the anterior circulation infarcts only. Patients with bilateral infarcts imaging sequences (12). or no visible infarct were excluded. Level of significance was set to P < 0.05. Radiological Characteristics Two licensed neuroradiologists (J.W. and M.D.) reviewed all RESULTS MRI examinations via the web interface (12). They remained blinded to clinical information, stroke phenotype data, and any Study Population diagnostic information from the contributing center to avoid Of the 3,301 correctly included patients in the image repository, bias of image interpretation. A structured reporting protocol was 183 were unavailable for analysis due to technical errors. implemented in close collaboration between the coinvestigators Additionally, 18 patients in the image repository had not been and utilized for all reviewed cases. Interrater agreement was correctly included and were therefore excluded from the analysis. measured in a portion of cases (446 cases, 13.5%) reviewed by A flowchart describing the material, including available MRI both assessors. sequences, is shown in Figure 2. Diffusion-weighted image was The protocol included assigning the ischemic lesions available for 2,739 patients (median age = 65 years, IQR observed on the diffusion-weighted images (DWIs) to predefined = 53–75 years; 39.4% female), whereas MRA was available anatomical categories, arterial supply regions, territories of the for 1,597 patients (median age = 64 years, IQR = 52–75 FIGURE 1 | Number of MRI, DWI, and MRA in the MRI-Genie image depository by the contributing center. Frontiers in Neurology | www.frontiersin.org 3 June 2020 | Volume 11 | Article 577 Drake et al. The MRI-GENIE Study—Baseline Radiological Findings TABLE 1 | Diffusion-weighted image lesion characteristics related to time from stroke to MRI (328 patients lacked information about time to DWI). DWI parameter 0–48 h 2–14 d >14 d Number of patients 1,227 1,061 123 Median time to MRI (days) (IQR) 1 (0–1) 3 (2–6) 27 (20–37) Anterior/posterior circulation/both (N) 762/324/39 676/278/31 64/25/1 Supratentorial/infratentorial/both (N) 886/179/63 759/176/50 71/16/3 Right/Left/both (N) 486/539/104 434/464/86 40/45/5 Brainstem, N (%) 90 (7.3) 103 (9.8) 8 (6.5) Cerebellum, N (%) 62 (5.1) 59 (5.6) 6 (4.9) Basal ganglia (%) 189 157 10 Hemisphere (%) 658 563 60 Infarct not seen, N (%) 99 (8.1) 76 (7.2) 33 (26.8) FIGURE 2 | CONSORT diagram of the MRI-GENIE study population. Multiple infarcts, N (%) 129 (10.5) 103 (9.7) 6 (4.9) years; 39.3% female). Both DWI and MRA were available (excellent) and weighted κ = 0.97 (excellent) measured using a for 1,539 patients (median age = 64 years, IQR = 52–75 single review criterion (vascular territory). years; 39.4% female). Technical specifications such as MRI The most common anatomical category based on DWI was manufacturer, model names, and field strengths are shown in the hemispheric (51.7%). The second most common ischemic lesion Supplemental Table 3. location was deep gray matter (defined as basal ganglia and The median time from symptom onset to MRI for the 2,739 thalamus), seen in 14.6% of the examinations. patients with available DWI was 1 day (IQR = 1–4 days). In In 2,044 (77%) of the examinations with lesions on DWIs, addition, 1,227, 1,061, and 123 of these patients had an MRI supratentorial lesions were seen; in 561 (17%), infratentorial performed in the time intervals 0 to 48 h, 2 to 14 days, and more lesions were seen, and in 138 examinations (6%), both than 14 days, respectively, as shown in Table 1. supratentorial and infratentorial lesions were seen. Overall, 67% of the ischemic lesions occurred in the anterior Center Differences circulation, 29% in the posterior circulation, 3.5% to both The 12 participating centers contributed 70–475 patients. Center- anterior and posterior circulation, and 0.1% undetermined. specific demographics and radiological data are included in When further stratified by vascular territory according to the Supplemental Table 1. major arterial branches, the middle cerebral artery (MCA) territory was most common, followed by vertebrobasilar artery Stroke Etiology According to CCS and territory and posterior cerebral artery territory. TOAST Of the 2,238 patients with unilateral infarcts seen on DWI, All 3,301 correctly included subjects in the image repository had 53% had left-sided infarct lateralization (P = 0.013, Figure 3F). a stroke etiology classification according to CCS, and 89.6% had When comparing the distribution of unilateral infarctions for a TOAST assessment (Supplemental Tables 1A–D). supratentorial and infratentorial distribution, a higher frequency of left-sided lesions was seen only for supratentorial infarcts Risk Factor Distribution (53%; P = 0.018). When comparing the distribution of unilateral Hypertension was the most common cardiovascular risk factor, infarctions for the major vascular territories, a significantly present among 65% of the 3,301 cases (0.7% missing data). higher frequency of left-sided infarcts was seen only for the MCA Additionally, 51% were current or former smokers (0% missing), territory (53%; P = 0.032). 23% had diabetes mellitus (0.9% missing), 18% had coronary artery disease (2.3% missing), and 14% had atrial fibrillation MRA Characteristics of Intracranial Vessels (1.1% missing). Among the 1,597 patients with available MRA of the intracranial vessels, moderate to severe vessel stenosis (stenosis visually Radiological Characteristics gauged to >50% luminal diameter reduction) was most common DWI Lesion Characteristics in the basilar artery and the vertebral arteries (4% and 3%, Ischemic lesions on the available DWI (n = 2,739) were assigned respectively). Stenosis was observed in 1% to 2% of the large to predefined anatomical categories based on normal cerebral vessels of the anterior circulation and the posterior cerebral anatomy and vascular territories (Supplemental Figure 1) with artery. Vessel occlusion was most common in the MCA (14%) further classification into solitary or multiple lesions. Ischemic and the vertebral arteries (12%), whereas occlusion of the basilar lesion distributions are described in detail in Figure 3. Interrater artery was least common (2%). agreement assessed in the 443 cases independently reviewed Figure 4 illustrates the distribution of stenosis in the major by both assessors was 98%, with Cohen unweighted κ = 0.96 intracranial arterial branches and LAOs with or without Frontiers in Neurology | www.frontiersin.org 4 June 2020 | Volume 11 | Article 577 Drake et al. The MRI-GENIE Study—Baseline Radiological Findings FIGURE 3 | Charts summarizing DWI lesion distribution according to (A) anatomical region, (B) vascular region, (C) vascular territory, (D) supratentorial and infratentorial location, and (E) lateralization. In chart (C), multiple infarcts were recorded once per vascular territory, and therefore the number of infarcts (2,713) exceeds the total number of patients with visible DWI lesions (2,467). Chart (F) shows the difference between right- and left-sided lesions for all lesions and specified for supratentorial and infratentorial lesions and specified for the major vascular territories [χ2 test after excluding patients with bilateral infarcts and with no infarcts (270 cases)]. FIGURE 4 | Distribution of stenosis and occlusions in the large intracranial vessels on MRA (N = 1,597). (A) shows the distribution of stenosis (visually graded as >50%). (B) shows the distribution of large-vessel occlusions. (C) shows the distribution of large-vessel occlusions with corresponding DWI lesions. (D) shows the distribution of large-vessel occlusions without corresponding DWI lesions. ACA, anterior cerebral artery; MCA, middle cerebral artery; ICA, internal carotid artery; ICA-T, top of internal carotid artery; PCA, posterior cerebral artery; BA, basilar artery; VA, vertebral artery. corresponding DWI lesions, as well as the distribution of LAOs and anterior cerebral artery occlusion were the rarest occlusion in relation to the time from stroke onset to MRI. Of the 1,539 types (4.7% and 4.2% of all LAOs, respectively). patients with available DWI and MRA, 423 (28%) had an large vessel occlusion (LVO) relevant to the localization of an infarct DISCUSSION on DWI (Table 2). The proportion of patients with LAO was higher in MRAs performed within 48 h of stroke onset (37%) In this systematic radiological analysis of clinical MRI scans of compared with MRAs performed later than 48 h (23%) after 3,301 AIS patients enrolled in 12 international hospital-based stroke. Middle cerebral artery occlusions accounted for more cohorts, we identified that anterior circulation infarcts were most than a third (34%) of all LAOs, whereas basilar artery occlusion prevalent (67%), that infarcts in the MCA territory were the most Frontiers in Neurology | www.frontiersin.org 5 June 2020 | Volume 11 | Article 577 Drake et al. The MRI-GENIE Study—Baseline Radiological Findings TABLE 2 | Large artery occlusions (LAO) in the 1,507 patients with DWI and MRA ischemic tissue injury and for developing targets for future (32 patients lacked information about time to DWI). interventions and therapeutics. MRI-GENIE offers a novel Arterial segment 0–48 h 2–14 days >14 days approach to developing such type of data, and systematic (n = 803) (n = 635) (n = 69) review and radiological analysis of the MRI data are the first steps toward detailed characterization of clinical MRI Proximal internal carotid, N (%) 21 (2.6) 20 (3.1) 7 (10.1) phenotypes. We will continue our work by combining this Distal internal carotid, N (%) 25 (3.1) 15 (2.4) 3 (4.3) work with recently developed automated volume segmentation Middle cerebral artery, N (%) 100 (12.4) 41 (6.5) 2 (2.9) algorithms (14) to assess the combined effect of risk factors Anterior cerebral artery, N (%) 11 (1.4) 5 (0.8) 2 (2.9) and vascular anatomy with lesion characteristics. We will also Vertebral artery, N (%) 31 (3.9) 19 (3.0) 0 assess the effect of genetic traits that may alter the MRI stroke Basilar artery, N (%) 11 (1.4) 8 (1.3) 1 (1.4) phenotype in general, or in specific conditions such as in Posterior cerebral artery, N (%) 31 (3.9) 20 (3.1) 0 the presence of certain neurovascular conditions or certain Multiple arteries, N (%) 32 (4.0) 17 (2.7) 1 (1.4) risk factors. No occlusion visible, N (%) 541 (67.3) 490 (77.2) 53 (76.8) Limitations of the current approach include known and potential differences between the individual contributing sites. Because this is a hospital-based study, the use of MRI may have varied with clinical practice over time and common, and that the majority of LAOs 0 to 48 h from ictus were between centers. The low mean patient age at several centers in the MCA territory. It was also observed that the proportion of (Supplemental Table 1A) suggests that MRI may predominantly large-vessel occlusions decreased with increasing time from onset have been used for younger patients. For example, MRI may to imaging, especially for the MCA territory. have been used predominantly for cases where CT had not Multiple acute lesions in one or several vascular territories been conclusive, thereby underrepresenting cases of large were common (11%), and this prevalence was comparable infarcts, or for suspected ischemic strokes in the posterior with other current studies (8.8–9.8%) (19). However, the two fossa where CT is of limited diagnostic value because of major etiologies for multiple acute DWI lesions, large artery artifacts in this region (21). However, the proportion of atherosclerosis and cardioembolic, were notably common in the posterior circulation lesions in our cohort matches previous cohort (36 and 40%, CCS and TOAST), suggesting that most estimates (15–35%) (16, 22). Another limitation is that the MRI patients with stroke attributed to cardioembolic and large-artery scanners and examination protocols varied between centers and atherosclerosis present with a single DWI lesion. over time. We also showed a higher prevalence of left-sided ischemic The strengths of our approach include a large sample of lesions compared to right-sided lesions and that this difference MRI scans from multiple international centers obtained as part in prevalence also could be observed for the largest subgroups of hospital-based care for patients with AIS, thereby likely of patients with supratentorial lesions and lesions in the MCA generalizable to global stroke patient population as compared to territory. Left-sided lesions were also more common on CT in previously published single-center studies. Additional strengths a recent population-based cohort (8, 9). In our material, this include the large number of DWI and MRAs, as well as detailed difference is caused mainly by a difference in supratentorial stroke subtype characterization using validated TOAST and lesions in the MCA territory. CCS classifications. These results highlight how “big data” allow for stratification of patients into smaller but more homogeneous subgroups based on lesion appearance on MRI, an important step to CONCLUSIONS individualizing stroke care. Our structured data on imaging and clinical characteristics provide baseline information to In this article, we report baseline data from the MRI-GENIE assess the feasibility of studies on specific MRI phenotypes study and the initial neuroradiological findings describing such as patients with multiple lesions or lesions in a certain ischemic lesions, vascular anatomy, and vascular lesions vascular territory. Additionally, our data can be used in the according to a structured reporting protocol. further development and validation of automated segmentation This work will provide an imaging framework for genetic software by providing a very large annotated data set based on studies of the MRI-GENIE cohort, facilitating the linkage standardized neuroradiological assessment. between imaging and genetics in advancing the knowledge of This unique database also provides an analytical platform ischemic stroke. It may also serve as a reference material for for novel, large-scale studies of genetic architecture of MRI- automated analysis of ischemic lesions or vascular characteristics. based phenotypes in patients with AIS. To date, there are limited data on the genetics of imaging phenotypes, such as white matter hyperintensities (WMH) and DWI volumes (13, 14). DATA AVAILABILITY STATEMENT Furthermore, there are limited large-scale studies of genetic architecture and MRI-based phenotypes from hospital-based The raw data supporting the conclusions of this article will be multicenter stroke populations (20). These data are critical to made available by the authors, without undue reservation, to any obtain to advance knowledge of underlying mechanisms of qualified researcher. Frontiers in Neurology | www.frontiersin.org 6 June 2020 | Volume 11 | Article 577 Drake et al. The MRI-GENIE Study—Baseline Radiological Findings ETHICS STATEMENT ACKNOWLEDGMENTS The studies involving human participants were reviewed and Dr. Lee Nolan is gratefully acknowledged for professional approved by All 12 sites in the SIGN collaboration obtained language editing. Professor Peter Höglund is gratefully ethics committee approval for its part of the project, as well acknowledged for professional statistical expertise. as signed informed consent from all included patients. The patients/participants provided their written informed consent to participate in this study. SUPPLEMENTARY MATERIAL AUTHOR CONTRIBUTIONS The Supplementary Material for this article can be found All authors listed have made a substantial, direct and intellectual online at: https://www.frontiersin.org/articles/10.3389/fneur. contribution to the work, and approved it for publication. 2020.00577/full#supplementary-material REFERENCES 15. Dubost F, Bruijne M, Nardin M, Dalca AV, Donahue KL, Giese AK, et al. Multi-atlas image registration of clinical data with automated quality 1. Lövblad KO, Altrichter S, Mendes Pereira V, Vargas M, Marcos Gonzalez A, assessment using ventricle segmentation. Med Image Anal. (2020) 63:101698. Haller S, et al. Imaging of acute stroke: CT and/or MRI. J Neuroradiology. doi: 10.1016/j.media.2020.101698 (2015) 42:55–64. doi: 10.1016/j.neurad.2014.10.005 16. Frid P, Drake M, Giese AK, Wasselius J, Schirmer MD, Donahue 2. Vilela P, Rowley HA. Brain ischemia: CT and MRI techniques KL, et al. Detailed phenotyping of posterior vs. anterior circulation in acute ischemic stroke. Eur J Radiol. (2017) 96:162–72. ischemic stroke: a multi-center MRI study. J Neurol. (2019) 267:649–58. doi: 10.1016/j.ejrad.2017.08.014 doi: 10.1007/s00415-019-09613-5 3. Vert C, Parra-Fariñas C, Rovira À. MR imaging in hyperacute ischemic stroke. 17. Lucitti JL, Sealock R, Buckley BK et al. Variants of Rab GTPase-Effector Eur J Radiol. (2017) 96:125–32. doi: 10.1016/j.ejrad.2017.06.013 binding protein-2 cause variation in the collateral circulation and severity of 4. Kim BJ, Kang HG, Kim HJ, Ahn SH, Kim NY, Warach S, et al. Magnetic stroke. Stroke. (2016) 47:3022–1. doi: 10.1161/STROKEAHA.116.014160 resonance imaging in acute ischemic stroke treatment. J Stroke. (2014) 16:131– 18. McArdle PF, Kittner SJ, Ay H, Brown RD Jr, Meschia JF, Rundek 45. doi: 10.5853/jos.2014.16.3.131 T, et al. Agreement between TOAST and CCS ischemic stroke 5. Ko Y, Lee S, Chung JW, Han MK, Park JM, Kang K, et al. MRI-based algorithm classification: the NINDS SiGN study. Neurology. (2014) 83:1653–60. for acute ischemic stroke subtype classification. J Stroke. (2014) 16:161–72. doi: 10.1212/WNL.0000000000000942 doi: 10.5853/jos.2014.16.3.161 19. Novotny V, Thomassen L, Waje-Andreassen U, Naess H. Acute cerebral 6. Ikram MA, van der Lugt A, Niessen WJ, Koudstaal PJ, Krestin GP, Hofman A, infarcts in multiple arterial territories associated with cardioembolism. Acta et al. The rotterdam scan study: design update 2016 and main findings. Eur J Neurol Scand. (2017) 135:346–51. doi: 10.1111/ane.12606 Epidemiol. (2015) 30:1299–315. doi: 10.1007/s10654-015-0105-7 20. Thompson PM, Stein JL, Medland SE, Hibar DP, Vasquez AA, Renteria 7. Seshadri S, Wolf PA, Beiser A, Elias MF, Au R, Kase CS, ME, et al. The ENIGMA Consortium: large-scale collaborative analyses et al. Stroke risk profile, brain volume, and cognitive function: of neuroimaging and genetic data. Brain Imaging Behav. (2014) the Framingham Offspring Study. Neurology. (2004) 63:1591–9. 8:153–82. doi: 10.1212/01.WNL.0000142968.22691.70 21. Barber PA, Hill MD, Eliasziw M, Demchuk AM, Pexman JH, Hudon 8. Foerch C1, Misselwitz B, Sitzer M, Berger K, Steinmetz H, Neumann- ME, et al. Imaging of the brain in acute ischaemic stroke: comparison of Haefelin T. Arbeitsgruppe Schlaganfall Hessen. Difference in recognition computed tomography and magnetic resonance diffusion-weighted imaging. of right and left hemispheric stroke. Lancet. (2005) 366:392–3. J Neurol Neurosurg Psychiatry. (2005) 76:1528–33. doi: 10.1136/jnnp.2004. doi: 10.1016/S0140-6736(05)67024-9 059261 9. Portegies ML, Selwaness M, Hofman A, Koudstaal PJ, Vernooij 22. Flossmann E, Rothwell PM. Prognosis of vertebrobasilar transient MW, Ikram MA. Left-sided strokes are more often recognized than ischaemic attack and minor stroke. Brain. (2003) 126:1940–54. right-sided strokes: the Rotterdam study. Stroke. (2015) 46:252–4. doi: 10.1093/brain/awg197 doi: 10.1161/STROKEAHA.114.007385 10. Zeng Q, Tao W, Lei C, Dong W, Liu M. Etiology and risk Conflict of Interest: OW: Consulting for Penumbra, Inc; Advisory Board factors of posterior circulation infarction compared with anterior Member for Genentech, Inc. JRos: Consulting for Boehringer Ingelheim and New circulation infarction. J Stroke Cerebrovasc Dis. (2015) 24:1614–20. Beta Innovation. AL: Bayer, BMS/Pfizer, Astra Zeneca and Portola, outside the doi: 10.1016/j.jstrokecerebrovasdis.2015.03.033 submitted work. JW: Co-founder and shareholder of UmanSense AB, Sweden. 11. Meschia JF, Arnett DK, Ay H, Brown RD Jr, Benavente OR, Cole JW, et al. Stroke Genetics Network (SiGN) study: design and rationale for a The remaining authors declare that the research was conducted in the absence of genome-wide association study of ischemic stroke subtypes. Stroke. (2013) any commercial or financial relationships that could be construed as a potential 44:2694–702. doi: 10.1161/STROKEAHA.113.001857 conflict of interest. 12. Giese AK, Schirmer MD, Donahue KL, Cloonan L, Irie R, Winzeck S, et al. Design and rationale for examining neuroimaging genetics in Copyright © 2020 Drake, Frid, Hansen, Wu, Giese, Schirmer, Donahue, ischemic stroke: The MRI-GENIE study. Neurol Genet. (2017) 3:e180. Cloonan, Irie, Bouts, McIntosh, Mocking, Dalca, Sridharan, Xu, Giralt-Steinhauer, doi: 10.1212/NXG.0000000000000180 Holmegaard, Jood, Roquer, Cole, McArdle, Broderick, Jiménez-Conde, Jern, Kissela, 13. Schirmer MD, Dalca AV, Sridharan R, Giese AK, Donahue KL, Nardin MJ, Kleindorfer, Lemmens, Meschia, Rundek, Sacco, Schmidt, Sharma, Slowik, Thijs, et al. White matter hyperintensity quantification in large-scale clinical acute Woo, Worrall, Kittner, Mitchell, Rosand, Golland, Lindgren, Rost and Wassélius. ischemic stroke cohorts - the MRI-GENIE study. NeuroImage Clinical. (2019) This is an open-access article distributed under the terms of the Creative Commons 23:101884. doi: 10.1016/j.nicl.2019.101884 Attribution License (CC BY). The use, distribution or reproduction in other forums 14. Wu O, Winzeck S, Giese AK, Hancock BL, Etherton MR, Bouts MJRJ, et al. is permitted, provided the original author(s) and the copyright owner(s) are credited Big data approaches to phenotyping acute ischemic stroke using automated and that the original publication in this journal is cited, in accordance with accepted lesion segmentation of multi-center magnetic resonance imaging data. Stroke. academic practice. No use, distribution or reproduction is permitted which does not (2019) 50:1734–41. doi: 10.1161/STROKEAHA.119.025373 comply with these terms. Frontiers in Neurology | www.frontiersin.org 7 June 2020 | Volume 11 | Article 577