Overlap of Diseases Underlying Ischemic Stroke The ASCOD Phenotyping

Transcription

1 Overlap of Diseases Underlying Ischemic Stroke The ASCOD Phenotyping Gaia Sirimarco, MD; Philippa C. Lavallée, MD; Julien Labreuche, BST; Elena Meseguer, MD; Lucie Cabrejo, MD; Céline Guidoux, MD; Isabelle F. Klein, MD, PhD; Jean-Marc Olivot, MD, PhD; Halim Abboud, MD; Valérie Adraï, MD; Jérôme Kusmierek, MD; Samina Ratani, MD; Pierre-Jean Touboul, MD; Mikael Mazighi, MD, PhD; Philippe Gabriel Steg, MD; Pierre Amarenco, MD Downloaded from by guest on April 9, 2018 Background and Purpose ASCOD phenotyping (A, atherosclerosis; S, small vessel disease; C, cardiac pathology; O, other causes; and D, dissection) assigns a degree of likelihood to every potential cause (1 for potentially causal, 2 for causality is uncertain, 3 for unlikely causal but disease is present, 0 for absence of disease, and 9 for insufficient workup to rule out the disease) commonly encountered in ischemic stroke. We used ASCOD to investigate the overlap of underlying vascular diseases and their prognostic implication. Methods A single rater applied ASCOD in 405 patients enrolled in the Asymptomatic Myocardial Ischemia in Stroke and Atherosclerotic Disease study. Results A was present in 90% of patients (A1=43% and A2=15%), C in 52% (C1=23% and C2=14%), and S in 66% (S1=11% and S2=2%). On the basis of grades 1 and 2, 25% of patients had multiple underlying diseases, and 80% when all 3 grades were considered. The main overlap was found between A and C; among C1 patients, A was present in 92% of cases (A1=28%, A2=20%, and A3=44%). Conversely, among A1 patients, C was present in 47% of cases (C1=15%, C2=15%, and C3=17%). Grades for C were associated with gradual increase in the 3-year risk of vascular events, whereas risks were similar across A grades, meaning that the mere presence of atherosclerotic disease qualifies for high risk, regardless the degree of likelihood for A. Conclusions ASCOD phenotyping shows that the large overlap among the 3 main diseases, and the high prevalence of any form of atherosclerotic disease, reinforces the need to systematically control atherosclerotic risk factors in all ischemic strokes. (Stroke. 2013;44: ) Stroke classification is crucial to perform clinical trials, epidemiological studies, phenotyping for genetic studies, and classify patients to evaluate best treatment strategy. Precise analysis of stroke subtypes requires the integration of clinical features, diagnostic test findings, and knowledge about potential etiologic factors. With improvement of noninvasive brain and cardiovascular imaging, as well as biological tests, the amount of available information is fast increasing. Current classifications only consider the most likely cause, neglecting the overlap between diseases. 1,2 Therefore, a patient classified in the atherosclerosis group is rigidly analyzed with this group even if there are MRI or clinical evidence of presence of small vessel disease (SVD) deemed to be not causally related with the index stroke. 3 In this example, any genetic analysis Key Words: atherosclerosis etiology stroke with this kind of classification would, therefore, be biased by neglecting the SVD presence. Furthermore, current diagnostic tools often identify several coexisting causes resulting in 40% of ischemic strokes classified as undetermined in causative system. 4,5 The ASCO (atherosclerosis, SVD, cardiac pathology, and other uncommon causes) phenotyping assigned a degree of likelihood on potential causality of all 4 main underlying diseases commonly encountered in ischemic stroke patients. 6 Therefore, the purpose of the ASCO phenotyping is not to reduce the rate of undetermined cause as in causative classifications, 7 because there is no undetermined category in ASCO, but to describe the overlap degree between the diseases. We investigated the overlap of underlying vascular diseases of ischemic stroke in the Asymptomatic Myocardial Ischemia Received March 5, 2013; final revision received April 26, 2013; accepted May 7, From the INSERM U698 and Paris-Diderot University, Sorbonne Paris Cité, Paris, France (G.S., P.C.L, J.L., E.M., L.C., C.G., I.F.K., P.-J.T., M.M., P.G.S., P.A.); and Department of Neurology and Stroke Centre (G.S., P.C.L, E.M., L.C., C.G., I.F.K., J.-M.O., H.A., V.A., J.K., S.R., P.-J.T., M.M., P.A.), Department of Radiology (I.F.K.), and Department of Cardiology (P.G.S.), Bichat University Hospital, Paris, France. Ralph L. Sacco, MD, approved the final version of this article. The online-only Data Supplement is available with this article at /-/DC1. Correspondence to Pierre Amarenco, MD, Department of Neurology and Stroke Centre, Bichat University Hospital, 46 rue Henri Huchard, Paris, France. pierre.amarenco@bch.aphp.fr 2013 American Heart Association, Inc. Stroke is available at DOI: /STROKEAHA

2 2428 Stroke September 2013 Downloaded from by guest on April 9, 2018 in Stroke and Atherosclerotic Disease (AMISTAD) cohort and their prognostic implication using a revised ASCOD classification. Material and Methods Study Population We analyzed patients in the AMISTAD study, designed to assess the prevalence and impact of systemic atherosclerosis on short and long-term recurrence risk. 8 Between June 2005 and December 2008, 785 patients were consecutively assessed for eligibility and 405 patients with ischemic stroke documented by neuroimaging were enrolled consecutively within 10 days of symptom onset (Figure I in the online-only Data Supplement). Patients with dissection were excluded. Pregnant women or patients with other nonvascular diseases associated with life expectancy of <30 months were not eligible. Informed consent was obtained from patient or surrogate, and the research protocol was approved by the Ethics Committee of Paris Bichat-Claude Bernard. Investigations and Data Collection All patients had a standardized evaluation, including medical history, physical examination, routine blood biochemistry, cerebral imaging (MRI and/or computed tomographic scan), vascular imaging of cerebral arteries (supra-aortic trunks or transcranial Doppler, MRA, and computed tomography angiography), and cardiac evaluation included a 12-lead ECG (and Holter monitoring) and echocardiography (transthoracic±transesophageal echocardiogram). In addition, all included patients underwent coronary angiography except if they had history of coronary artery disease (defined as acute coronary syndrome, myocardial infarction, or prior coronary revascularization) and underwent a Doppler ultrasonography of abdominal and femoral arteries. Stroke Subtype Classification All cases were classified by a single trained stroke neurologist (G.S.) after consensual review of 30 cases (G.S., P.A.) using the revised version of ASCO classification, named ASCOD. 9 The ASCO classification categorizes 4 predefined phenotypes: atherosclerosis (A), SVD (S), cardiac pathology (C), and other causes (O). Each of the 4 phenotypes is graded according to following categories: 1 when the disease is a potential cause of the index stroke, 2 when causality is uncertain, 3 when the disease is present but is unlikely a direct cause, 0 when the disease is absent, and 9 when the workup is insufficient to rule out the disease. 6 In the revised version, dissection has been separated from the group of other causes because of its high frequency in young stroke patients and diagnostic specificities; therefore, this additional phenotype modified the classification name. In addition, the cerebral artery luminal stenosis cutoff point considered as a potential cause of stroke was 50% rather than the more stringent 70%, in agreement with current recommendations. 10,11 Full details of the revised version are available in the online-only Data Supplement. The same stroke neurologists (G.S.) who had classified the 405 cases with ASCOD performed a second independent classification in 30 randomly cases with sufficient workup to grade all phenotypes. Only 2 discordances (6.7%) were found between the 2 readings: 1 in S phenotype (S0 versus S3) and 1 in C phenotype (C2 versus C3). Follow-up Follow-up visits were scheduled between 3 and 6 months after enrollment and every year thereafter. The current report is based on the database lock in December 2011, when the last patient completed a 3-year follow-up. At each follow-up visit, treatment, blood pressure (BP), lipid profile (except at sixth month), and any occurrence of clinical events or hospitalizations were recorded. The main outcome was the time to the first major vascular event, including vascular death, nonfatal cardiac event, nonfatal stroke, or major peripheral arterial event. Statistical Analysis Continuous variables are expressed as mean (SD), and categorical variables are expressed as frequencies (percentages). The differences in baseline characteristics between ASCOD grades for main phenotypes (atherosclerosis, cardiac pathology, and SVD) were tested using the test for trend in ANOVA for continuous variable and Cochran Mantel Haenszel analysis for categorical variables. Comparison among grades for a given phenotype was adjusted for grades for other main phenotype. All baseline characteristics associated with grade for a given phenotype (with a P value <0.10) were further included in multivariate ordinal logistic regression model. Using grades 1 and 2, patients were divided into 2 groups according to the presence of an overlap between the main phenotype. Comparison in baseline characteristics between the 2 groups was done using Student t test (continuous variable) and χ 2 test (categorical variable). Multivariate analysis was done using logistic regression model. We estimated and compared the 3-year major vascular event rates among ASCOD grades for main phenotype using the Kaplan Meier method and log-rank test. Patients who died from nonvascular causes were censored at the time of death. We estimated the relative risk of vascular events associated with ASCOD grades through the calculation of hazard ratio using grade 0 as reference. Estimates were calculated from a Cox proportional hazards model including the ASCOD grades for 3 main phenotypes together. Regarding the low number of patients with grade 2 evidence for SVD (n=7), the grades 1 and 2 were pooled together in all analysis. Statistical testing was done at the 2-tailed α level of Data were analyzed using SAS version 9.3 (SAS Institute, Cary, NC). Results Of the 405 patients enrolled in AMISTAD, ASCOD could be adjudicated on the basis of case report forms and imaging files in 403 patients. Demographic, risk factors, and investigation profiles of this study sample are available in Table II in the online-only Data Supplement. All patients had undergone 1 cerebral imaging (computed tomographic scan, 84.1%; diffusion-weighted imaging MRI, 93.5%; fluid attenuated inversion recovery MRI, 87.1%; and T2-weighted gradient echo sequence MRI, 83.6%), 1 arterial vascular imaging (supraaortic trunks Doppler, 99.8%; extracranial angiography, 76.7%; transcranial Doppler, 99.0%; and intracranial angiography, 93.8%), and 1 ECG/Holter at hospital admission; 377 (94%) of them had undergone echocardiography, including transesophageal imaging in 317 patients. Of the 377 patients with MRI, 324 (86%) had diffusion-weighted imaging, fluid attenuated inversion recovery, and T2-weighted gradient echo sequences. ASCOD Results Atherosclerosis was the most prevalent underlying disease (grades 1+2+3=90%), as well as the most frequent causal underlying disease, with 43% of patients graded as A1 and 15% as A2 (Table 1). SVD was also highly prevalent (grades 1+2+3=66%) but was considered potentially causal or with uncertain causality in only 13% of patients. Cardiac pathology was present in 52% of patients and was the second most frequent causal underlying disease (C1=23% and C2=14%). Other causes (O1+O2+O3) were found in 14 patients (3.5%). Eight patients (2%) could not be classified into 1 of the 5 categories because of insufficient workup. In 395 patients with sufficient workup to grade all phenotypes, 61 different ASCOD patterns were found. The most frequent ASCOD pattern was A1-S3- C0-O0-D0 (n=54) followed by A1-S0-C0-O0-D0 (n=30).

3 Sirimarco et al ASCOD and Overlap of Disease in Ischemic Stroke 2429 Table 1. Distribution of the Main Diseases Underlying Ischemic Stroke Grades Phenotypes 1 (Potentially Causal) 2 (Uncertain Causality) 3 (Unlikely Causal) 0 (Disease) Not Present 9 (Insufficient Workup) Atherosclerosis (A) 172 (42.7) 61 (15.1) 129 (32.0) 41 (10.2) 0 (0.0) SVD (S) 46 (11.4) 7 (1.8) 212 (52.6) 135 (33.5) 3 (0.7) Cardiac pathology (C) 91 (22.6) 57 (14.1) 61 (15.1) 191 (47.4) 3 (0.7) Dissection or other causes (OD) 3 (0.7) 2 (0.5) 9 (2.2) 386 (95.8) 3 (0.7) Values represent number (%). SVD indicates small vessel disease. Downloaded from by guest on April 9, 2018 Overlap in Main Phenotypes We investigated the overlap among the 3 main phenotypes in 395 patients with sufficient workup to grade all phenotypes (Figure 1). Considering grade 1 (as other classification systems do, such as TOAST [Trial of ORG in acute stroke treatment]), 8% of patients had multiple underlying diseases (25 patients were both A1 and C1, 5 were both A1 and S1, and 3 were C1 and S1). The rate of patients with overlapping underlying disease increased to 25% when grades 1 and 2 were considered and increased to 80% when considering all 3 grades (Figure 1). The main overlap was found between atherosclerosis and cardiac pathology, when only grade 1 was considered (n=25 [6%]), as well as when considering grades 1+2 (n=73 [18%]). Among 90 patients classified C1, atherosclerosis was present in 92% of cases and was considered as potentially causal in 28% and with uncertain causality in another 20% (Figure 2). Conversely, among 168 patients classified A1, cardiac pathology was present in 47% of cases and was considered as potentially causal in 15% and with uncertain causality in another 15%. Figure 2 also shows that C2 and C3 patients have a high prevalence of atherosclerotic disease, being present in almost all patients, and of SVD. Risk Factor Profile of Main Phenotypes Demographics and risk factors by ASCOD grades for main phenotypes are available in Tables III to V in the onlineonly Data Supplement. In multivariate analysis adjusting for cardiac pathology and SVD grades, age, men, sex, hypertension history, and baseline systolic BP (SBP) were independently associated with increased likelihood of causal link for atherosclerosis. In multivariate analysis adjusting for atherosclerosis and SVD grades, age, hypertension history, and current smoking were independently associated with increased likelihood of causal link for cardiac pathology. In multivariate analysis adjusting for cardiac pathology and atherosclerosis grades, hypertension history, baseline SBP, and absence of coronary heart disease history (personal or familial) were independently associated with increased likelihood of causal link for SVD. As shown in Table 2, age, hypertension history, baseline SBP, dyslipidemia, personal stroke history, and coronary heart disease were associated with more as compared with less overlap in the main phenotypes (using grade 1 or 2 for more overlap, and grade 0 or 3 for less overlap). Except age and history of stroke, all of these risk factors remained significantly associated with an overlap in multivariate analysis. Vascular Risk Recurrence of Main Phenotypes Among 392 patients with 1 postbaseline follow-up visit and sufficient workup to grade all phenotypes, 53 patients experienced 1 major vascular events during a 3-year follow-up; there were 14 vascular deaths (including 12 from cardiac disease), 20 first nonfatal coronary events (5 nonfatal myocardial infarctions, 1 resuscitated cardiac arrests, and 14 hospitalizations for unstable angina or heart failure), 15 nonfatal strokes, and 13 peripheral arterial disease events. Events occurred despite a high rate of self-reported medication adherence after discharge (Table VI in the online-only Data Supplement). At discharge, 81% of patients were receiving BP-lowering therapy (57% were treated by dual or triple therapy), 90% were receiving statins, 22% were receiving antidiabetic drugs, and 97% were receiving antithrombotic therapy. In addition, SBP and low-density lipoprotein-cholesterol levels decreased Figure 1. Venn diagram among the 3 main diseases underlying ischemic stroke according to ASCOD (A, atherosclerosis; S, small vessel disease; C, cardiac pathology; O, other causes; and D, dissection) classification.

4 2430 Stroke September 2013 Downloaded from by guest on April 9, 2018 Figure 2. Overlap between each pair of main diseases underlying ischemic stroke according to ASCOD (A, atherosclerosis; S, small vessel disease; C, cardiac pathology; O, other causes; and D, dissection) classification. The distribution of ASCOD grades for the main diseases underlying ischemic stroke are plotted in patients with ASCOD grade 1 (A), ASCOD grade 2 (B), and ASCOD grade 3 (C). significantly during 1-year follow-up period (9 mm Hg and 40 mg/dl, respectively). Regarding the ASCOD phenotyping, a significant difference in SBP change was found among ASCOD grades for cardiac pathology and SVD (Figure II in the online-only Data Supplement). Except lower change in low-density lipoprotein-cholesterol for C1 in comparison with C0, C3, and C2, no difference was found among ASCOD grades. As shown in Figure 3, the risk for combined major vascular events differed significantly among ASCOD grades for cardiac pathology (P<0.001). No such differences were found among ASCOD grades for atherosclerosis or among ASCOD grades for SVD (Table 3). In multivariate analysis, including ASCOD grades for the 3 main phenotypes, the graded increase in the risk observed for cardiac pathology remained significant. Using C0 as reference, the multivariate hazard ratios (95% confidence interval) were 2.80 ( ) for C3, 3.18 ( ) for C2, and 5.32 ( ) for C1. Discussion In this cohort of 403 stroke patients, the ASCOD phenotyping captured and weighted the overlap between diseases underlying ischemic stroke. Being present in 90% of cases, atherosclerotic disease was highly prevalent and considered potentially causal (A1) in 43% of patients. SVD was present in 66% of cases, although it was considered potentially causal (S1) in only 11% of patients, and cardiac pathology was present in 52% of cases, but was potentially causal in 23% of patients. As shown in Figure 2, the overlap is even more important than anticipated, with, for example, 92% of patients C1 having also any form of atherosclerotic disease (A1, A2, or A3). ASCOD phenotyping provided additional information not available using a causative classification 4,12 with a good diagnostic accuracy, as shown by the observed intrarater agreement and the previous published inter-rate agreement on the original ASCO classification. 4 The ASCOD phenotyping takes into account underlying diseases or abnormalities that are not necessarily causally related to the index stroke, such as

5 Sirimarco et al ASCOD and Overlap of Disease in Ischemic Stroke 2431 Downloaded from by guest on April 9, 2018 Table 2. Characteristics of Patients With More (ASCOD 1 or 2) or Less (ASCOD 3 or 0) Overlap in the 3 Main Diseases Underlying Ischemic Stroke Less (n=298) Overlap* More (n=97) P Value Demographics Age, y, mean (SD) 61.3 (13.1) 66.2 (12.4) Men 220 (73.8) 76 (78.4) 0.37 Risk factors Hypertension 232 (77.9) 88 (90.7) Diabetes mellitus 61 (20.5) 25 (25.8) 0.27 Dyslipidemia 111 (37.3) 56 (57.7) Current smoking 115 (38.7) 35 (36.1) 0.64 Personal history of stroke 19 (6.4) 13 (13.4) 0.03 Family history of stroke 65 (21.9) 26 (27.1) 0.29 Personal history of 33 (11.1) 26 (26.8) <0.001 coronary heart disease Family history of coronary 79 (26.6) 22 (23.2) 0.50 heart disease Examination findings Systolic BP, mm Hg, 138 (18) 143 (19) 0.02 mean (SD) Diastolic BP, mm Hg, 79 (11) 79 (11) 0.85 mean (SD) LDL-C, mg/dl, mean (SD) 120 (40) 114 (42) 0.24 Values are percentage (count) unless otherwise indicated. ASCOD indicates A, atherosclerosis; S, small vessel disease; C, cardiac pathology; O, other causes; and D, dissection; BP, blood pressure; and LDL-C, low-density lipoproteincholesterol. *Grades 1 and 2. Student t test or χ 2 test. leukoaraiosis and microbleeds (that witness SVD) or atherosclerotic stenosis in opposite circulation. Retaining all abnormalities, ASCOD is immediately spendable for the purpose of comprehensive phenotyping, whereas using causative classifications restricts the analyses to cases deemed to be causally related to the index stroke neglecting other patients with the same underlying disease for whom the degree of causal relationship with the index stroke is deemed to be low. For instance, when targeting SVD for a phenotype genotype correlation study, should we consider only 13% of patients S1 (as it was done in previous genetic studies) 13 or 66% of patients with any form of SVD? Investigating genetic markers linked to SVD, it may be justified to include patients with severe leukoaraiosis, microbleeds, or both even when the index stroke is not attributable to this pathological mechanism. Detecting more specific subgroups, phenotypic systems such as ASCOD may facilitate studying genetic and environmental risk factors. It would be wrong to multiply combinations of phenotypes that could lead to say that ASCOD has 625 different phenotypes. By phenotype we meant only 5 phenotypes derived from ASCOD: atherosclerotic phenotype, SVD phenotype, cardiac pathology phenotype, other cause phenotype, and dissection phenotype. With ASCOD, each phenotype for a given patient has a weight depending on the likelihood of a causal relationship and is described in every patient. The ASCOD phenotyping no longer has a cryptogenic or unknown cause or undetermined origin group. In causative classifications, this categorization is only reflecting our failure to identify a direct cause among mixed underlying abnormalities or diseases. This group is unlikely to be useful for selecting patients in therapeutic trial, given the difficulty for 1 neurologist to another, or for primary care physicians and regulators to clearly delineate this population of patients and have a common, validated definition. Figure 2 also shows that C2 and C3 patients (those among whom patent foramen ovale closure trials are recruiting) have a high prevalence of atherosclerotic disease, being present in almost all patients, and of SVD. Patients that could also be classified as embolic stroke of unknown source or cryptogenic in causative classification systems are A2, A3, C2, C3, O2, and O3, in ASCOD, meaning that using ASCOD, a patient being cryptogenic in TOAST has in fact atherosclerotic disease (eg, A2 or A3) with ASCOD or another underlying pathology. We move from a negative message to the cryptogenic stroke patient with TOAST (ie, we do not know what you have) to a positive message with ASCOD (ie, you have atherosclerotic disease, and you will be treated as such). ASCOD phenotyping may also improve current methods to select stroke patients for clinical trials targeting a specific Figure 3. Cumulative event curve for the composite end point of vascular events by ASCOD (A, atherosclerosis; S, small vessel disease; C, cardiac pathology; O, other causes; and D, dissection) grades for cardiac pathology.

6 2432 Stroke September 2013 Downloaded from by guest on April 9, 2018 Table 3. Three-Year Major Vascular Rates According to ASCOD Grades for the Main Diseases Underlying Ischemic Stroke ASCOD Grades Event Rates, % (n) P Value* Multivariate HR (95% CI) P Value Atherosclerosis A0 (n=41) 12.6 (5) (ref) A3 (n=125) 13.2 (16) 0.68 ( ) 0.46 A2 (n=59) 12.0 (7) 0.63 ( ) 0.43 A1 (n=167) 15.3 (25) 1.18 ( ) 0.74 SVD S0 (n=132) 13.0 (17) (ref) S3 (n=207) 14.0 (28) 0.89 ( ) 0.71 S1, S2 (n=53) 15.3 (8) 1.74 ( ) 0.22 Cardiac pathology C0 (n=187) 6.6 (12) < (ref) C3 (n=61) 16.6 (10) 2.80 ( ) 0.02 C2 (n=55) 18.5 (10) 3.18 ( ) C1 (n=89) 24.5 (21) 5.32 ( ) <0.001 ASCOD indicates A, atherosclerosis; S, small vessel disease; C, cardiac pathology; O, other causes; D, dissection; CI, confidence interval; and HR, hazard ratio. *Log-rank test for trend. Determined in Cox regression analysis, including the 3 ASCOD grades. disease (eg, selection of patients with atherosclerotic phenotype). It is interesting to note that, during a 3-year follow-up, there was no significant difference in the risk of major vascular events among A1, A2, and A3, or among S1, S2, and S3. This means that in ischemic stroke, patients with any form of underlying atherosclerotic disease, the risk is similar across all grades. Therefore, in clinical trials of antithrombotic agents, we can simply select patients with A1, A2, and A3 if we want to target atherosclerotic disease-related stroke, and we can also exclude S1 and S2 if we want to reduce the bleeding complication risk because SVD is associated with a higher bleeding risk than other stroke causes. 14 Regarding cardiac pathology, we found a significant difference between the risk of major vascular events, with a gradual increase for C0, C3, C2, and C1. This probably reflects the fact that the components of grades C3 and C2 have a much lower risk than the components of grade C1 (eg, the risk associated with strands, C3, is known to be much lower than the risk associated with atrial fibrillation, C1), 15 whereas the global risk associated with carotid plaque or contralateral carotid stenosis, A3, is as high as the global risk associated with severe ipsilateral stenosis, A1. These findings underline the need of investigating what is hidden in the large group elsewhere classified as undetermined cause because a patient classified A3 in ASCOD may require a different management compared with a patient in whom the disease is completely absent. The mere presence of atherosclerotic plaque in an arterial territory represents a marker of atherosclerosis disease per se associated with an increased vascular risk. 16 For instance, statin therapy showed greater benefit in stroke patients with atherosclerosis disease, 17,18 and this has been translated in guidelines. 19 Our study had strengths and limitations. First, our study was not population based, but it was hospital based in one of the largest stroke units in Paris, France, working as primary care referral center for acute stroke admissions with dedicated catchment area. Second, we have excluded the most severe stroke patients (Rankin 5, ie, bedridden patients) who are likely to die from their index stroke. Regarding the main objective of AMISTAD study, we also excluded the patients with carotid or vertebral artery dissection. Therefore, we could not exclude that these selection criteria biased the results somewhat, and we caution that the present study was limited to the main ischemic stroke phenotypes. Finally, the observed 3-year risk for incident major vascular events was lower than expected, likely because of the best medical care applied to these patients, as exemplified by an average 40 mg/ dl low-density lipoprotein-cholesterol reduction after the Stroke-Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL), 20 or by average 9 mm Hg SBP reduction after the Perindopril Protection Against Recurrent Stroke Study (PROGRESS) trial. 21 For this reason, we cannot exclude the possibility of a lack of statistical power to detect difference among ASCOD grades. In conclusion, ASCOD phenotyping shows that, regardless the disease deemed to be actually causal, the large overlap among the 3 main diseases in ischemic stroke patients, and the high prevalence of any form of atherosclerotic disease, reinforces the need to control atherosclerotic risk factors systematically in all ischemic strokes. Acknowledgments We are indebted to all nurses of Bichat Stroke Center and to the interventional cardiology personnel for their continuous support in this research. Aimee Grosz, Nassima Schmoll, Evelyne Herinomenjanahary, Hugo Brandao, and Genevieve Pétré were appointed clinical research assistants and were funded by SOS-ATTAQUE CEREBRALE. None. Disclosures References 1. Adams HP Jr, Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon DL, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org in Acute Stroke Treatment. Stroke. 1993;24: Ay H, Benner T, Arsava EM, Furie KL, Singhal AB, Jensen MB, et al. A computerized algorithm for etiologic classification of ischemic stroke: the Causative Classification of Stroke System. Stroke. 2007;38: Amarenco P, Bogousslavsky J, Caplan LR, Donnan GA, Hennerici MG. Classification of stroke subtypes. Cerebrovasc Dis. 2009;27: Wolf ME, Sauer T, Alonso A, Hennerici MG. Comparison of the new ASCO classification with the TOAST classification in a population with acute ischemic stroke. J Neurol. 2012;259: Ay H. Advances in the diagnosis of etiologic subtypes of ischemic stroke. Curr Neurol Neurosci Rep. 2010;10: Amarenco P, Bogousslavsky J, Caplan LR, Donnan GA, Hennerici MG. New approach to stroke subtyping: the A-S-C-O (phenotypic) classification of stroke. Cerebrovasc Dis. 2009;27: Amort M, Fluri F, Weisskopf F, Gensicke H, Bonati LH, Lyrer PA, et al. Etiological classifications of transient ischemic attacks: subtype classification by TOAST, CCS and ASCO a pilot study. Cerebrovasc Dis. 2012;33: Amarenco P, Lavallée PC, Labreuche J, Ducrocq G, Juliard JM, Feldman L, et al. Prevalence of coronary atherosclerosis in patients with cerebral infarction. Stroke. 2011;42:22 29.

7 Sirimarco et al ASCOD and Overlap of Disease in Ischemic Stroke 2433 Downloaded from by guest on April 9, Amarenco P, Bogousslavsky, Caplan LR, Donnan GA, Wolf ME, Hennerici MG. The A-S-C-O-D phenotyping of ischemic stroke (updated ASCO phenotyping). Cerebrovasc Dis. In press. 10. Cina CS, Clase CM, Haynes RB. Carotid endarterectomy for symptomatic carotid stenosis. Cochrane Database Syst Rev. 2011;4:CD Alamowitch S, Eliasziw M, Algra A, Meldrum H, Barnett HJ; North American Symptomatic Carotid Endarterectomy Trial (NASCET) Group. Risk, causes, and prevention of ischaemic stroke in elderly patients with symptomatic internal-carotid-artery stenosis. Lancet. 2001;357: Marnane M, Duggan CA, Sheehan OC, Merwick A, Hannon N, Curtin D, et al. Stroke subtype classification to mechanism-specific and undetermined categories by TOAST, A-S-C-O, and causative classification system: direct comparison in the North Dublin population stroke study. Stroke. 2010;41: Markus HS, Barley J, Lunt R, Bland JM, Jeffery S, Carter ND, et al. Angiotensin-converting enzyme gene deletion polymorphism. A new risk factor for lacunar stroke but not carotid atheroma. Stroke. 1995;26: Benavente OR, Hart RG, McClure LA, Szychowski JM, Coffey CS, Pearce LA. Effects of clopidogrel added to aspirin in patients with recent lacunar stroke. N Engl J Med. 2012;367: Cohen A, Tzourio C, Chauvel C, Bertrand B, Crassard I, Bernard Y, et al. Mitral valve strands and the risk of ischemic stroke in elderly patients. The French Study of Aortic Plaques in Stroke (FAPS) Investigators. Stroke.1997;28: Polak JF, Pencina MJ, Pencina KM, O Donnell CJ, Wolf PA, D Agostino RB Sr. Carotid-wall intima-media thickness and cardiovascular events. N Engl J Med. 2011;365: Sillesen H, Amarenco P, Hennerici MG, Callahan A, Goldstein LB, Zivin JA, et al. Atorvastatin reduces risk of cardiovascular events in patients with carotid atherosclerosis. Stroke. 2008;39: Amarenco P, Labreuche J. Lipid management in stroke prevention: review and up-dated meta-analysis of statin in stroke prevention. Lancet Neurol. 2009;8: Furie KL, Kasner SE, Adams RJ, Albers GW, Bush RL, Fagan SC, et al.; American Heart Association Stroke Council, Council on Cardiovascular Nursing, Council on Clinical Cardiology, and Interdisciplinary Council on Quality of Care and Outcomes Research. Guidelines for the prevention of stroke in patients with stroke or transient ischemic attack: a guideline for healthcare professionals from the American Heart Association/ American Stroke Association. Stroke. 2011;42: Amarenco P, Bogousslavsky J, Callahan A 3rd, Goldstein LB, Hennerici M, Rudolph AE, et al.; Stroke Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL) Investigators. High-dose atorvastatin after stroke or transient ischemic attack. N Engl J Med. 2006;355: PROGRESS Collaborative Group. Randomised trial of a perindoprilbased blood-pressure-lowering regimen among 6,105 individuals with previous stroke or transient ischaemic attack. PROGRESS Collaborative Group. Lancet. 2001;358:

8 Downloaded from by guest on April 9, 2018 Overlap of Diseases Underlying Ischemic Stroke: The ASCOD Phenotyping Gaia Sirimarco, Philippa C. Lavallée, Julien Labreuche, Elena Meseguer, Lucie Cabrejo, Céline Guidoux, Isabelle F. Klein, Jean-Marc Olivot, Halim Abboud, Valérie Adraï, Jérôme Kusmierek, Samina Ratani, Pierre-Jean Touboul, Mikael Mazighi, Philippe Gabriel Steg and Pierre Amarenco Stroke. 2013;44: ; originally published online July 16, 2013; doi: /STROKEAHA Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX Copyright 2013 American Heart Association, Inc. All rights reserved. Print ISSN: Online ISSN: The online version of this article, along with updated information and services, is located on the World Wide Web at: Data Supplement (unedited) at: Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: Subscriptions: Information about subscribing to Stroke is online at:

9 SUPPLEMENTARY MATERIAL

10 Supplementary Method The methods of the A-S-C-O-D (Revised ASCO classification) : In ASCOD, every patient should be graded in 5 predefined phenotypes : A (atheroclerosis); S (small vessel disease); C (Cardiac pathology); O (Other cause); D (Dissection). As done in the ASCO classification (Amarenco P, Bogousslavsky J, Caplan LR, Donnan GA, Hennerici MG. A new approach to stroke subtyping : the A-S-C-O (phenotypic) classification of stroke. Cerebrovasc Dis ;27 : ), three degree of causality between the index ischemic stroke and each categories are considered : - If the disease is present and can potentially be a cause, grade is 1. - If the disease is present, but the causal link is uncertain, grade is 2. - If the disease is present, but the causal link is unlikely, grade is 3. To determine the degree of causality, a number of diagnostic tests must be performed. These diagnostic tests are described for each category in table S1. If after a minimum work-up has been performed, the disease is not detected, grade is 0. In case of incomplete work-up (e.g, a minimal work-up has not been performed) to rule out the disease, grade is 9 (table S1). Table S1. Causality grades of predefined phenotypes A : Causality grades for Atherothrombosis A1 «Potentially causal» Atherothrombotic stroke defined as: (1) Ipsilateral atherosclerotic stenosis between 50 99% in an intra-/or extracranial artery supplying the ischemic field; or (2) Ipsilateral atherosclerotic stenosis <50% in an intra-/or extracranial artery with an endoluminal thrombus supplying the ischemic field; or (3) Mobile thrombus in the aortic arch; or (4) Ipsilateral arterial occlusion in an intra-/or extracranial artery with evidence of underlying atherosclerotic plaque supplying the ischemic field A2 Causal link is uncertain (1) Atherosclerotic stenosis <50% in an intra-/or extracranial artery supplying the ischemic field; or (2) Aortic plaque 4 mm without mobile lesion. A3 Causal link is unlikely, but the disease is present (1) Plaque with less than 30% stenosis in an intra- or extracranial artery, ipsilateral to the infarct area (2) Aortic plaque <4 mm without mobile thrombus; (3) Stenosis (any degree) in a cerebral artery not supplying the infarct area (e.g., contralateral side or opposite circulation)

11 (4) History of myocardial infarction, coronary revascularization or peripheral arterial disease. A0 Atherosclerosis not detected Ruling out atheroscleosis: (1) Extracranial arterial stenosis: one or several of the following diagnostic tests are performed and are negative: Duplex, CTA, MRA, XRA, autopsy. (2) Intracranial arterial stenosis: one or several of the following diagnostic tests are performed and are negative: TCD, MRA, CTA, XRA, autopsy. (3) Aortic arch atheroma: TEE with specific assessment of the aortic arch (when the probe is pulled back at the end of the cardiac examination, turn the probe counter clockwise and take time to watch the aortic arch). A9 Incomplete work-up Duplex, TCD or CTA, or MRA, or XRA or autopsy not performed. [A minimum work-up is extra and intracranial assessment of cerebral arteries maximum also include transesophageal assessment of the aortic arch (or a default CT angiography of the aortic arch)]. S : Causality grades for Small-Vessel Disease S1 «Potentially causal» Combination of: (1) Lacunar infarction: small deep infarct <15 mm on MRI (or CT) in an area corresponding to the symptoms and at least one of the two following criteria: (2) one or several small deep older infarct of lacunar type in other territories ; and/or (3) severe (confluent) leukoaraiosis, or microbleeds, or severe dilatation of perivascular spaces («état criblé»). (4) Repeated, recent -<1 month-, TIAs attributable to the same territory as the index infarct, increases the prediction of a «lacunar infarction from 57% in their absence to 80% in their presence S2 Causal link is uncertain (1) Only one, recent, lacunar infarction and no other abnormality on MRI (or CT) or (2) Clinical syndrome suggestive of a deep branch artery stroke, without ischemic lesion in the appropriate area seen on MRI or CT [Main clinical syndrome suggesting a deep branch artery lacunar stroke : pure hemiparesis, pure hemi sensory loss, Hemiparesis-ataxy, Dysarthria clumsy hand syndrome, unilateral sensory-motor deficit, others: hemichorea, hemiballism, pure dysarthria, etc.)] S3 Causal link is unlikely, but the disease is present Severe (confluent) leukoaraiosis visible on MRI and/or CT scan, and/or microbleeds visible on T2*-weighted MRI, and/or severe dilatation of perivascular spaces (visible on T2-weighted MRI), and/or one or several old, small deep infarcts of lacunar type S0 Small-vessel disease Ruling out small vessel disease stroke: negative MRI (T2, FLAIR, GRE, DWI) and no

12 not detected appropriate clinical syndrome suggestive of a deep branch artery stroke S9 Incomplete work-up MRI (or CT) not performed C : Causality grades for Cardiac pathology C1 «Potentially causal» Detection of at least one of the following potential cause: (1) Mitral stenosis; (2) Mechanical valve; (3) Myocardial infarction within 4 weeks preceding the cerebral infarction; (4) Mural thrombus mural in the left cavities; (5) Aneurysm of the left ventricle; (6) history of documented atrial fibrillation whether paroxysmal (>60 sec), persistent or permanent- or flutter, with or without left atrial thrombus or spontaneous echo; (7) Atrial disease (tachycardia-bradycardia syndrome); (8) Dilated cardiomyopathy; (9) Left ventricle ejection fraction <35%; (10) Endocarditis; (11) Intracardiac mass; (12) PFO and thrombus in situ; (13) PFO and concomitant pulmonary embolism or proximal DVT preceding the index cerebral infarction C2 Causal link is uncertain (1) PFO + atrial septal aneurysm; (2) PFO and pulmonary embolism or proximal DTV concomitant but NOT preceding the index cerebral infarction; (3) intracradiac spontaneous echocontrast; (4) Apical akinesia of the left ventricle and decreased ejection fraction (but >35%) ; (5) History of Myocardial infarction or palpitation and multiple brain infarction, repeated either bilateral or in two different arterial territories (e.g., both anterior and posterior circulation) : (6) no direct cardiac source identified, but evidence of systemic emboli : renal or splenic or mesenteric infarction (on CT, MRI or autopsy) or embolism in peripheral artery supplying arm or leg. C3 Causal link is unlikely, but the disease is present One of the following abnormalities present in isolation: PFO, ASA, strands, mitral annulus calcification, calcification aortic valve, non apical akinesia of the left ventricle C0 Cardiac pathology not detected or not suspected Ruling out a cardiac source of embolism: minimum is negative ECG and auscultation by a cardiologist; maximum is negative ECG/telemetry/24-hour Holter ECG/long-term ECG recording (implantable device, transtelephonic ECG, loop recorder) and negative TEE for atrium, valves and septal abnormalities, negative TTE for PFO and assessment of left ventricle, negative cardiac CT/MRI, negative abdominal CT/MRI (search for old or simultaneous subdiaphragmatic visceral infarction) C9 Incomplete work-up Minimum is ECG and auscultation by a cardiologist

13 O : Causality grades for Other Causes O1 Potentially causal (1) Dolichoectasia with complicated aneurysm (2) Polycythemia vera or thrombocytemia > /mm 3 (3) Systemic Lupus (4) disseminated intravascular coagulation (5) Antiphospholipid antibody syndrome (including >100 GPL units or lupus anticoagulant) (6) Fabry s disease (7) Coexisting meningitis (8) sickle cell disease (9) Ruptured intracranial aneurysm with or without vasospasm of the artery supplying the infarcted area (10) severe hyperhomocysteinemia; (11) Horton s disease; (12) other cerebral inflammatory or infectious angeitis; (13) etc. O2 Causal link is uncertain (1) Saccular aneurysm (with a suspicion of embolism from it) O3 Causal link is unlikely but the disease is present (1) Arterio-venous malformation; (2) thrombocythosis < /mm 3, (3) antiphospholipid antibody <100 GPL units (4) homocysteinemia <40 µmol/l. O0 No other cause detected Ruling out other causes: negative: cerebrospinal fluid, complete haemostasis, cerebral arterial imaging and cardiac imaging, family history of inherited disease, inflammatory markers (erythrocyte sedimentation rate, C-reactive protein), hematologic tests (platelet, leucocytes, and eosinophilic counts, hematocrit) O9 Incomplete work-up Unable to exclude an other cause based on diagnostic tests and clinical history available D : Causality grades for Dissection D1 Potentially causal (1) Arterial dissection by direct demonstration (evidence of mural hematoma: hypersignal on Fatsaturated MRI or at autopsy or on TOF-MRA or CT on axial sections showing both enlargement of the arterial wall by the hematoma with narrowing of the lumen- or on echography showing an hypoechoic arterial wall with narrowing of the lumen and sudden enlargement of the carotid or vertebral (V2) artery diameter. (2) Arterial dissection by indirect demonstration or by less sensitive or less specific diagnostic test (only long arterial stenosis beyond the carotid bifurcation or in V2, V3 or V4 without demonstration of arterial wall hematoma : on X-ray angiography, and/or echography and/or CTA and/or MRA) D2 Causal link is uncertain (1) arterial dissection by weak evidence (suggestive clinical history, e.g., painful Horner s syndrome or past-history of arterial dissection) (2) Imaging evidence of fibromuscular dysplasia of a cerebral artery supplying the ischemic field

14 D3 Causal link is unlikely but the disease is present (1) Kinking or dolichoectasia without complicated aneurysm or plicature, (2) Fibromuscular dysplasia on arteries not supplying the ischemic field; D0 No dissection detected or suspected Ruling out dissection : negative FAT saturated MRI of suspected artery or good-quality, negative X-ray angiography D9 Incomplete work-up No FAT saturated MRI on the extra or intracranial artery supplying the ischemic field or no X-ray angiography performed within 15 days of symptom onset. [too early FAT saturated MRI (within 3 days of symptom onset) can be falsely negative and then should be repeated].

15 Table S2.Demographics, Risk Factors, and Investigations in AMISTAD cohort Values Demographics Age, y, mean (SD) 62.6 (13.1) Male gender 298 (74.0) Risk factors Hypertension 328 (81.4) Diabetes 88 (21.8) Dyslipidemia 172 (42.7) Current smoking 154 (38.3) Personal history of stroke 34 (8.5) Family history of stroke 92 (22.9) Personal history of coronary heart disease 62 (15.4) Family history of coronary heart disease 104 (26.0) Examination findings Systolic BP, mmhg, mean (SD) 139 (19) Diastolic BP, mmhg, mean (SD) 79 (11) LDL-C, mg/dl mean (SD) 119 (40) Investigations At least 1 cerebral imaging 403 (100.0) CT scan 339 (84.1) DWI-MRI 377 (93.5) FLAIR-MRI 351 (87.1) T2*-weighted GRE-MRI 337(83.6) At least 1 Extracranial vascular imaging 403 (100.0) MRA 292 (72.5) Supra-aortic trunks Doppler 402 (99.8) CT angiography 68 (16.9) At least 1 Intracranial vascular imaging 402 (99.8) MRA 369 (91.6) Transcranial Doppler 399 (99.0) CT angiography 53 (13.2) Transthoracic echocardiography 377 (93.5) Transesophageal echocardiography 317 (78.7) Electrocardiography or holter monitoring 403 (100.0) Values are number (%) unless otherwise indicated. SD indicates standard deviation; BP, blood pressure; LDL-C, low-density lipoprotein cholesterol; CT, computed tomography; MRI, magnetic resonance imaging; DWI, diffusion weighted imaging; FLAIR, fluid attenuated inversion recovery; GRE, gradient echo; MRA, magnetic resonance angiography.

16 Table S3. Demographics and Risk Factors according to ASCOD Grades for atherothrombosis phenotype A0 (n=41) A3 (n=126) A2 (n=60) A1 (n=168) P Values * Demographics Age, y, mean (SD) 51.3 (15.3) 61.9 (12.3) 63.2 (13.0) 65.5 (11.5) <0.001 Male gender 28 (68.3) 90 (71.4) 45 (75.0) 133 (79.2) 0.05 Risk factors Hypertension 26 (63.4) 93 (73.8) 55 (91.7) 146 (86.9) <0.001 Diabetes 6 (14.6) 29 (23.0) 11 (18.3) 40 (23.8) 0.14 Dyslipidemia 10 (24.4) 52 (41.3) 25 (41.7) 80 (47.6) 0.01 Current smoking 15 (37.5) 46 (36.5) 28 (46.7) 61 (36.3) 0.55 Personal history of stroke 2 (4.9) 9 (7.1) 4 (6.7) 17 (10.1) 0.20 Family history of stroke 11 (26.8) 25 (19.8) 21 (35.0) 34 (20.5) 0.66 Personal history of coronary heart disease 2 (4.9) 20 (15.9) 11 (18.3) 26 (15.5) 0.11 Family history of coronary heart disease 9 (22.0) 45 (35.7) 15 (25.4) 32 (19.3) 0.04 Examination findings Systolic BP, mmhg, mean (SD) 132 (17) 136 (18) 141 (18) 143 (19) <0.001 Diastolic BP, mmhg, mean (SD) 79 (10) 79 (11) 79 (12) 79 (11) 0.21 LDL-C, mg/dl mean (SD) 115 (38) 118 (41) 118 (41) 119 (41) 0.99 Values are number (%) unless otherwise indicated.* P values for linear trend adjusted for cardiac pathology and small-vessel disease grades(calculated using the Cochran-Mantel-Haenszel test for categorical data and analysis of variance for continuous data) SD indicates standard deviation; blood pressure; LDL-C, low-density lipoprotein cholesterol.

17 Table S4. Demographics and Risk Factors according to ASCOD Grades for Small-Vessel Disease phenotype S0 (n=134) S3 (n=208) S2, S1 (n=53) P Values* Demographics Age, mean (SD) 58.5 (13.4) 66.5 (12.1) 57.1 (10.7) 0.24 Male gender 104 (77.6) 150 (72.1) 42 (79.3) 0.86 Risk factors Hypertension 97 (72.4) 177 (85.1) 46 (86.8) Diabetes 27 (20.2) 42 (20.2) 17 (32.1) 0.18 Dyslipidemia 56 (41.8) 93 (44.7) 18 (34.0) 0.96 Current smoking 54 (40.3) 65 (31.4) 31 (58.5) 0.65 Personal history of stroke 9 (6.7) 20 (9.7) 3 (5.7) 0.61 Family history of stroke 28 (20.9) 52 (25.1) 11 (21.2) 0.95 Personal history of coronary heart disease 25 (18.7) 31 (14.9) 3 (5.7) 0.07 Family history of coronary heart disease 43 (32.3) 48 (23.2) 10 (19.2) 0.04 Examination findings Systolic BP, mmhg, mean (SD) 135 (17) 141 (19) 146 (18) <0.001 Diastolic BP, mmhg, mean (SD) 78 (10) 79 (11) 83 (11) LDL-C, mg/dl mean (SD) 114 (37) 123 (43) 110 (39) 0.35 Values are number (%) unless otherwise indicated. * P values for linear trend adjusted for adjusted for atherothrombosis and cardiac pathology grades(using the Cochran-Mantel-Haenszel test for categorical data and analysis of variance for continuous data). SD indicates standard deviation; blood pressure; LDL-C, low-density lipoprotein cholesterol.

18 Table S5.Demographics and Risk Factors according to ASCOD Grades for Cardiac Pathology phenotype C0 (n=188) C3 (n=61) C2 (n=56) C1 (n=90) P Values * Demographics Age, mean (SD) 58.5 (12.5) 65.6 (11.6) 59.3 (11.7) 69.4 (12.2) <0.001 Male gender 143 (76.1) 46 (75.4) 43 (76.8) 64 (71.1) 0.85 Risk factors Hypertension 148 (78.7) 51 (83.6) 37 (66.1) 84 (93.3) Diabetes 42 (22.3) 15 (24.6) 9 (16.1) 20 (22.2) 0.95 Dyslipidemia 62 (33.0) 26 (42.6) 29 (51.8) 50 (55.6) <0.001 Current smoking 87 (46.3) 20 (32.8) 24 (42.9) 19 (21.4) Personal history of stroke 11 (5.9) 4 (6.6) 4 (7.1) 13 (14.6) 0.02 Family history of stroke 44 (23.5) 13 (21.7) 14 (25.0) 20 (22.2) 0.68 Personal history of coronary heart disease 11 (5.9) 7 (11.5) 11 (19.6) 30 (33.3) < Family history of coronary heart disease 33 (17.7) 19 (31.7) 15 (26.8) 34 (38.2) Examination findings Systolic BP, mmhg, mean (SD) 141 (18) 139 (19) 137 (17) 137 (22) 0.30 Diastolic BP, mmhg, mean (SD) 81 (11) 76 (10) 79 (10) 78 (12) 0.37 LDL-C, mg/dl mean (SD) 122 (38) 114 (44) 122 (44) 111 (41) 0.05 Values are number (%) unless otherwise indicated. * P values for linear trend adjusted for atherothrombosis and small vessel disease grades (using the Cochran-Mantel-Haenszel test for categorical data and analysis of variance for continuous data). SD indicates standard deviation; blood pressure; LDL-C, low-density lipoprotein cholesterol.