Cerebrovascular Risk Factors and Clinical Classification of Strokes

Cerebrovascular Risk Factors and Clinical Classification of Strokes Antonio Pinto, M.D., 1 Antonino Tuttolomondo, M.D., 1 Domenico Di Raimondo, M.D., 1 Paola Fernandez, M.D., 1 and Giuseppe Licata, M.D. 1 ABSTRACT Cerebrovascular risk represents a progressive and evolving concept owing to the particular distribution of risk factors in patients with ischemic stroke and in light of the newest stroke subtype classifications that account for pathophysiological, instrumental, and clinical criteria. Age represents the strongest nonmodifiable risk factor associated with ischemic stroke, while hypertension constitutes the most important modifiable cerebrovascular risk factor, confirmed by a host of epidemiological data and by more recent intervention trials of primary (HOT, Syst-Eur, LIFE) and secondary (PROGRESS) prevention of stroke in hypertensive patients. To be sure, a curious relationship exists between stroke and diabetes. Although the Framingham Study, The Honolulu Heart Program, and a series of Finnish studies reported a linear relationship between improved glucose metabolism and cerebral ischemia, the clinical and prognostic profile of diabetic patients with ischemic stroke remains to be fully understood. Our group, on the basis of TOAST classification a diagnostic classification of ischemic stroke developed in 1993 that distinguishes five different clinical subtypes of ischemic stroke: large-artery atherosclerosis (LAAS), cardioembolic infarct (CEI), lacunar infarct (LAC), stroke of other determined origin (ODE), and stroke of undetermined origin (UDE), and now extensively used in clinical and scientific context analysed the prevalence of cerebrovascular risk factors and the distribution of TOAST subtypes in more 300 patients with acute ischemic stroke in two consecutives studies that reported the significant association between diabetes and the lacunar subtype and a better clinical outcome for diabetic patients, most likely related to the higher prevalence of the lacunar subtype. Well-confirmed are the roles of cigarette smoking, atrial fibrillation, and asymptomatic carotid stenosis as cerebrovascular risk factors. Particularly interesting seems to be the function of inflammation markers (CRP, TNF-a, IL-1b, ISPs) as potential risk factors. Still elusive remains the association between cholesterol serum levels and stroke, on the basis of the epidemiological data regarding this causative relationship, confirmed only by the results of intervention trials (4S, LIPID, CARE, HPS, ASCOT). Ultimately, cerebrovascular risk appears peculiar owing to the unique relationship between some modifiable risk factors (mainly diabetes and cholesterol) and the possible preferential association with stroke subtypes and specific cerebrovascular risks. Dyslipidemias and Atherosclerotic Thrombotic Disease; Editor in Chief, Jan Jacques Michiels, M.D., Ph.D.; Guest Editor, Joep C. Defesche, Ph.D. Seminars in Vascular Medicine, Volume 4, Number 3, 2004. Address for correspondence and reprint requests: Antonio Pinto, M.D., Istituto di Clinica Medica Policlinico di Palermo, Piazza delle Cliniche n 2, 90127 Palermo, Italia. 1 Department of Internal Medicine and Cardioangiology, University of Palermo, Palermo, Italy. Copyright # 2004 by Thieme Medical Publishers, Inc., 333 Seventh Avenue, New York, NY 10001 USA. Tel: +1(212)584-4662. 1528-9648,p;2004,04,03,287,303,ftx,en;svm00197x. 287

288 SEMINARS IN VASCULAR MEDICINE/VOLUME 4, NUMBER 3 2004 KEYWORDS: Stroke, cerebrovascular risk, TOAST classification Educational Objectives Upon completion of this article, the reader should (1) appreciate the prevalence of stroke risk factors; and (2) recognize the peculiarities of cerebrovascular risk in relation to subtypes of stroke. Recent discussions of the real meaning of cerebrovascular risk consider not only the classical risk factors of stroke but also vascular dementia and cognitive impairment secondary to cerebral vasculopathy. In Italy and Europe, strokes are the third most common cause of death and resulting disability. In Italy, 130,000 new stroke cases (48,000 of them lethal) occur in a population of 57 million inhabitants every year. If the actual incidence stays constant, this will mean more than 170,000 new cases each year in Italy and a prevalence of 1 million people affected by stroke in 2010 because of progressive population aging. Overall, ischemic stroke is a problem of older people; in the ever increasing population of those 80 years old and over, strokes become more serious because of the clinical presentation during the acute phase and the 10 times higher mortality; also, the resulting disability is twice as high as for younger subjects. 1 The evidence of risk factors for stroke has been derived from analyses of data collected during prospective and case-control studies of men and women. The main well-known predisposing factors for ischemic stroke include the nonmodifiable risk factors sex, age, and race. Age older than 65 years is associated with a significantly higher risk of stroke. Four studies, Framingham, 2 Cardiovascular Health Study, 3 Rochester Epidemiology Project, 4 and Western Australia, 5 have clearly demonstrated the positive correlation between stroke incidence and increasing age. Male sex seems to be associated with a higher incidence in several trials. 6 The incidences of stroke and stroke-associated mortality are higher in blacks than in other races. Several of the known risk factors for stroke, such as hypertension, diabetes, and obesity, are more common in blacks than whites. 7 African Americans also have a high risk of stroke and high prevalence of cardiovascular risk factors. The cardiovascular and cerebrovascular profile appears different between hypertensive and nonhypertensive patients; African Americans without hypertension may have lower stroke recurrence rates. 8 The metabolic syndrome is a clinical entity of associated occurrence of insulin resistance (related to obesity), impaired glucose tolerance or non insulindependent diabetes mellitus (NIDDM), hypertension, hypercholesterolemia (dyslipidemia), increased fibrinogen, and hyperuricemia 9 that is associated with an increased vascular risk. The present review is focused on a detailed description of the cerebrovascular risk profile. HYPERTENSION The association of blood pressure (BP) with stroke was investigated in several prospective observational studies; for the diastolic blood pressure (DBP) the results demonstrate positive, continuous, and apparently independent association, with no significant heterogeneity of effect among different studies. Within the range of DBP studied (70 110 mm Hg), there was no evidence of any threshold below which lower levels of DBP were not associated with lower risks of stroke. 10 An elevated number of randomized trials of antihypertensive drugs (chiefly diuretics or b-blockers) with a mean treatment duration of 5 years and mean DBP difference of 5 6 mm Hg in usual DBP showed 35 40% less stroke. For those dying in the trials, the DBP difference had persisted only 2 3 years, yet an overview showed that vascular mortality was significantly reduced (2p < 0.0002); nonvascular mortality appeared unchanged. Stroke was reduced by 42% (95% confidence interval 33 50%, 289 versus 484 events, 2p < 0.0001), suggesting that virtually all the epidemiologically expected stroke reduction appeared rapidly. 11 Concerning the relation between systolic blood pressure and stroke, the Syst-Eur Study (Systolic Hypertension in Europe Trial) evaluated risk factors for stroke in elderly patients with isolated systolic hypertension. It was a double-blind, randomized, placebocontrolled trial involving 4736 persons older than 60 years with systolic blood pressure 160 to 219 mm Hg and DBP < 95 mm Hg randomly assigned, during a follow-up of 4.5 years, to active treatment with nitrendipine (a dihydropyridine calcium channel blocker), with the possible addition of enalapril and hydrochlorothiazide, or placebo. The trial showed a 42% reduction in incidence of fatal and nonfatal stroke in the treated group; a multivariate analysis demonstrated that older age, smoking, history of diabetes, higher systolic blood pressure, lower high-density lipoprotein (HDL) cholesterol and electrocardiographic abnormalities were significantly associated (p < 0.05) with an increased incidence of stroke or transient ischemic attack (TIA). The benefits of active treatment were not independently related to sex, and treatment was more effective in patients with diabetes at entry. 12 The benefit of implementation in the treatment of hypertension of a low dose of acetylsalicylic acid in patients with elevated DBP was assessed in the HOT (Hypertension Optimal Treatment) study. A total of 18,790 patients aged 50 80 years with hypertension and

CEREBROVASCULAR RISK FACTORS AND CLINICAL CLASSIFICATION OF STROKES/PINTO ET AL 289 DBP between 100 and 115 mm Hg were randomly assigned a target DBP (< 90 mm Hg, < 85 mm Hg, < 80 mm Hg). In addition, 9399 patients were randomly assigned to acetylsalicylic acid at 75 mg/day; the other group received placebo. DBP was reduced by 20.3, 22.3, and 24.3 mm Hg in the < 90, < 85, and < 80 mm Hg groups, respectively. The lowest risk for stroke was associated with DBP below 80 mm Hg and systolic blood pressure below 142.2 mm Hg. The implementation of therapy with acetylsalicylic acid had no effect on stroke. 13 A new approach to the relationship between hypertension and stroke was offered by the HOPE (Heart Outcomes Prevention Evaluation) study, which was designed to assess the utility of the angiotensinconverting enzyme (ACE) inhibitor ramipril in diabetic patients with a history of cardiovascular events plus one or more demonstrated cardiovascular risk factors (one of these was hypertension) to prevent cardiovascular and renal disease. In this study 3577 patients were enrolled (37.5% had diabetes) and randomly assigned to receive ramipril (10 mg/day) or placebo and vitamin E or only placebo. Primary outcomes were myocardial infarction, stroke, and cardiovascular death. Ramipril reduced the risk of stroke by 33% (95% confidence interval [CI], 10 50) and total mortality by 24%. The benefit in cardiovascular effect was more relevant than expected from the reduction of blood pressure (reduction of 2.2 mm Hg of SBP and 1.4 mm Hg of DBP); there was a protective vascular and renal effect. 14 Another intervention study designed to determine the effects of a blood pressure lowering regimen with another ACE inhibitor (perindopril) in hypertensive and nonhypertensive patients with a history of stroke or TIA is PROGRESS (Perindopril Protection Against Recurrent Stroke Study). In this trial 6105 individuals were randomly assigned active treatment (n ¼ 3.051) based on the ACE inhibitor perindopril with the diuretic indapamide or placebo (n ¼ 3.054). The primary outcome was total stroke (fatal or nonfatal). In over 4 years of follow-up the active treatment reduced systolic/dbp by 9/4 mm Hg. A total of 307 individuals assigned active treatment suffered from stroke compared with 420 individuals assigned placebo with a relative risk reduction of 28% (95% CI, 17 38). There were similar reductions in the risk of stroke in hypertensive and nonhypertensive subgroups. Combination therapy with perindopril plus indapamide reduced blood pressure by 12/ 5 mm Hg and stroke risk by 43%. Single drug therapy with perindopril reduced blood pressure by 5/3 mm Hg and produced no discernable reduction in the risk of stroke. The results of this study show that the treatment with perindopril and indapamide is useful in prevention of stroke in patients with a history of stroke or TIA, irrespective of their blood pressure. 15 The LIFE (Losartan Intervention For Endpoint reduction in hypertension) study is a double-masked, randomized, parallel-group trial in which 9193 participants with essential hypertension and left ventricular hypertrophy ascertained by electrocardiography were randomly assigned to once daily losartan-based or to atenolol-based antihypertensive treatment for at least 4 years. The primary endpoints were death, myocardial infarction, and stroke. The two groups of subjects, treated with losartan or with atenolol, had a similar reduction of the systolic and diastolic pressure (30.2 and 16.6% reduction with losartan and 29.1 and 16.8% reduction with atenolol). The stroke outcome was highly in favor of losartan, showing a 24.9% relative risk reduction compared with atenolol. Losartan, an angiotensin II type 1 receptor antagonist, induced a reduction of the blood pressure similar to that with atenolol but was associated with a 25% further reduction of stroke; losartan could have a significant additive effect on stroke over and above blood pressure. 16 Further, these favorable effects of losartan on stroke are derived from a Canadian study. In this study, patients with untreated mild essential hypertension were randomly assigned in double-blind fashion to losartan or atenolol treatment for 1 year. Both treatments reduced blood pressure to a comparable degree, but the angiotensin II type 1 receptor antagonist losartan corrected the altered structure and endothelial dysfunction of resistance arteries and normalized endothelium-dependent relaxation (acetylcholine induced) whereas the b-blocker atenolol had no effect. 17 DIABETES The increased risk of stroke in diabetic patients has been established in several studies. This increase has been connected with the pathophysiological changes seen in the cerebral vessels of patients with diabetes. Increased mortality from stroke in diabetic patients has also been reported. The reason for this excess in mortality is unknown. It is clear that diabetic angiopathy is different from atherosclerotic angiopathy. The Framingham Study reported the association between diabetes and stroke with a relative risk (RR) for the diabetic population of 2.6 in males and 3.8 in females. 2 The Honolulu Heart Program was another of the first studies conducted to determine diabetes as independent risk factor for thromboembolic, hemorrhagic, and total stroke and included 7549 Japanese-American men aged 45 to 68 years during a follow-up of 4 years. These 7549 subjects were divided into four glucose intolerance categories by analyzing the history of diabetes, diabetic medication, and nonfasting glucose 1 hour after a 50-g load: group 1, low-normal (glucose < 151 mg/ dl); group 2, high-normal (glucose 151 to 224 mg/dl); group 3, asymptomatic high (glucose > 225 mg/dl); and group 4, known diabetes. The incidence of thromboembolic but not hemorrhagic

290 SEMINARS IN VASCULAR MEDICINE/VOLUME 4, NUMBER 3 2004 stroke increased with worsening glucose tolerance category; all four groups had significantly elevated age-adjusted relative risks of thromboembolic stroke. Associations were the same in hypertensive and nonhypertensive subjects and similar but slightly stronger in younger (aged 45 to 54 years) than in older (aged 55 to 68 years) subjects. This association appears largely independent of other cardiovascular disease risk factors. 18 Whether stroke is different in patients with and without diabetes (stroke subtype, stroke severity, prognosis, the relation between admission glucose levels and stroke severity or mortality) was addressed in the Copenhagen Stroke Study, which included 1135 acute stroke patients (233 had diabetes). All patients were evaluated until the end of rehabilitation by weekly assessment of neurological deficits (Scandinavian Stroke Scale) and functional disabilities. The diabetic stroke patient was 3.2 years younger than the nondiabetic patient (p < 0.01) and had hypertension more frequently (48% versus 30%). Initial stroke severity, lesion size, and site were comparable between the two groups. However, mortality was higher in the diabetic patients (24% versus 17%). Outcome was comparable in surviving patients with and without diabetes. The conclusions are that diabetes influences stroke in several aspects: in age, in subtype, in speed of recovery, and in mortality. Increased glucose levels on admission independently increased mortality from stroke in nondiabetic but not in diabetic patients. 19 A study involving a Finnish cohort of 1298 subjects aged 65 to 74 years investigated the impact of several major risk factors for stroke on the incidence of both fatal and nonfatal stroke during 3.5 years of followup. Interestingly, this prospective study evaluated metabolic control and duration of diabetes as predictors of stroke in elderly diabetic subjects. Of 1298 subjects participating in the baseline study, 1069 did not have diabetes and 229 had NIDDM. During follow-up, 3.4% of the nondiabetic subjects and 6.1% of the NIDDM subjects had a nonfatal or fatal stroke. The incidence of stroke was significantly higher in diabetic women compared with nondiabetic women (odds ratio [OR] 2.25; 95% CI, 1.65 to 3.06). In contrast, the risk of stroke was not significantly higher in diabetic men than in nondiabetic men (OR 1.36). In multivariate logistic regression analyses including all study subjects, the conditions that predicted stroke events in diabetic subjects were fasting and 2-hour glucose, glycosylated hemoglobin A 1c, the duration of diabetes, and atrial fibrillation (AF). Conditions that predicted stroke in nondiabetic subjects were fasting insulin, hypertension, and previous stroke. This study provides evidence that NIDDM, its metabolic control, and the duration of diabetes are important predictors of stroke in elderly subjects, particularly in women. 20 Another prospective study of 8077 men and 8572 women with an average follow-up of 16.4 years was performed in eastern Finland with the aim of assessing the independent effect of diabetes as a risk factor for stroke. Diabetes mellitus was the stronger risk factor for death from stroke among both men and women in univariate and multivariate analyses. In addiction, smoking and systolic blood pressure appeared to be independent risk factors in both sexes, as did serum total cholesterol in men. Men with diabetes at baseline appeared to be at a sixfold increased risk of death from stroke, and the RR for men who developed diabetes during the follow-up was 1.7. In women, those who were diabetic at baseline were at higher risk of stroke than women who developed diabetes later (8.2 and 3.7, respectively). 21 To confirm diabetes mellitus (NIDDM) as a risk factor for stroke in middle-aged patients, another Finnish study involved 1059 NIDDM patients and 1373 nondiabetics as controls. The patients were 45 to 64 years of age, had a follow-up of 7 years, and the primary outcome was stroke. The results confirm previous data in similar studies: the risk of stroke in NIDDM men was about threefold and in NIDDM women fivefold higher than in corresponding nondiabetic subjects. Previous history of stroke increased the risk of a new stroke event by threefold. Patients with hyperglycemia (plasma glucose > 13.4 mmol/l) and high hemoglobin A 1 (> 10.7%) had about a twofold higher risk of stroke than patients with better glycemic control. Low levels of HDL cholesterol (<0.90 mmol/ L), high levels of total triglyceride (> 2.30 mmol/ L), and the presence of hypertension were associated with a twofold increase in the risk of stroke mortality or morbidity. 22 A step forward is the UKPDS 36 study. The aim of this study was to determine the association between systolic blood pressure and macrovascular and microvascular complications of type 2 diabetes The clinical outcome endpoints of this prospective observational study involving 4801 patients were any complications or deaths related to diabetes and all-cause mortality. The incidence of clinical complications in patients with type 2 diabetes was significantly associated with systolic blood pressure; each 10 mm Hg decrease in updated mean systolic blood pressure was associated with a reduction in risk of 12% for any complications related to diabetes (95% CI, 10 to 14%, p < 0.0001), 15% for death related to diabetes (12 to 18%, p < 0.0001), 11% for myocardial infarction (7 to 14%, p < 0.0001), and 13% for microvascular complications (10 to 16%, p < 0.0001). The results of this study clearly show that the presence of more than one cerebrovascular risk factor increases the risk of the event and the importance of early assessment of blood pressure, especially in the course of diabetes. 23 The relationship between two main stroke risk factors, diabetes and carotid bruit, was evaluated in the

CEREBROVASCULAR RISK FACTORS AND CLINICAL CLASSIFICATION OF STROKES/PINTO ET AL 291 Fremantle Diabetes Study. This prospective study evaluated 1181 diabetic patients without a history of cerebrovascular disease during a 6.5-year mean follow-up. During the follow-up period, 11.3% of the patients suffered from a first stroke. The patients found to have an incidental carotid bruit have a greater than six times higher risk of first stroke in the first 2 years than diabetic patients without carotid bruit. These data confirm the importance of screening of the risk factors to obtain a complete risk profile for every patient. 24 In 1993, the Trial of Org 10172 in Acute Stroke Treatment (TOAST) group studied the relation between ischemic stroke, diabetes, and other risk factors; this group developed a new system to classify the subtypes of ischemic stroke to evaluate outcome and factors associated with each stroke subtype. Our group, on the basis of this classification, developed a study including patients with a diagnosis of acute ischemic stroke consecutively admitted to the Department of Internal Medicine and Cardioangiology of the University of Palermo. According to the TOAST classification, patients were subdivided into five different groups: (1) large artery atherosclerosis (LAAS), (2) cardioembolic infarct (CEI), (3) lacunar infarct (LAC), (4) stroke of other determined etiology (ODE), and (5) stroke of undetermined etiology (UDE). The neurological deficit at admission was scored using the Scandinavian Stroke Scale (SSS). The disability degree on discharge and after 6 months was longitudinally evaluated with the Rankin disability scale. We included 159 consecutive patients affected by an acute ischemic stroke; their demographic and clinical features according to the TOAST classification are summarized in Table 1. The analyses of risk factors showed a different profile according to the stroke subtype at univariate and multivariate levels (Tables 2 and 3). Hypercholesterolemia and cigarette smoking were more frequent among patients with LAAS. Hypertension and smoking were associated with CEI on univariate analysis alone. Diabetes, hypertension, and a history of TIA were more frequent among patients with LAC. Patients affected by ODE showed a infrequent history of diabetes and patients affected by UDE a rare history of hypercholesterolemia. The mean SSS score at admission was significantly different in the five subtypes of stroke (Table 4). CEI and UDE were the subtypes of stroke with the higher disability at admission, LAC that with the lower disability (Table 4). CEI had the worst prognosis and LAC the best prognosis as shown by Rankin disability scale at discharge and after 6 months (Table 5). Finally, survival at 30 days was lower for CEI (84%) and for UDE (86%); however, because of the low number of deaths, no significant difference was evident (Table 6). 25,26 Our group has conducted another study, a casecontrol study, to confirm the interesting results of our previous study. The aim of this study was to compare diabetic versus nondiabetic subjects affected by acute ischemic stroke: (1) to evaluate whether diabetics have a different frequency of stroke subtypes classified according to the TOAST classification, (2) to determine whether diabetics and nondiabetics have different severities at onset and functional outcomes, and (3) to search for a different profile of associated risk factors between the two study groups. We included 102 diabetic patients and 204 controls matched by age and sex, admitted to the hospital for acute ischemic stroke. Forty-four cases (43%) were women and 58 (57%) were men; the median age at inclusion was 66 years (range 49 88 years) for diabetics. Age and sex distribution, by means of a matching procedure, was similar for patients and controls. We studied the association of diabetes with some clinical variables (SSS at admission, Rankin disability score at discharge and after 6 months of follow-up, 30-day survival), ischemic stroke subtype Table 1 Variable Demographic and Clinical Parameters in Relation to the TOAST Subtypes TOAST Clinical Subtypes LAAS CEI LAC ODE UDE Total Sex M 37 12 22 5 14 90 F 31 7 16 7 8 69 Age* 69, 46 89 68, 42 79 67, 49 78 66, 47 80 660.5, 41 80 61, 41 89 SSS (mean SD) 30.9 2.3 25.9 3.4 43.9 2.9 32.8 3.5 28.9 3.2 33.3 þ 6.8 Rankin at discharge* 1, 0 4 2, 0 4 0, 0 3 0, 0 3 1.5, 0 4 1, 0 4 Rankin after 6 months* 1, 0 4 2, 0 4 0, 0 2 0.5, 0 4 1, 0 4 0.5, 0 4 Diabetes y 41.2 31.6 81.6 16.7 54.6 49.7 Hypercholesterolemia y 67.1 15.8 26.3 16.7 4.6 30.2 Hypertension y 65.8 31.6 78.9 41.7 45.5 57.2 Cigarette smoking y 58.8 21.1 47.4 41.7 31.8 46.5 Previous TIA y 30.9 31.6 47.4 9.0 0.0 28.9 *Median values and range are reported. y The number reported indicates the percent value.

292 SEMINARS IN VASCULAR MEDICINE/VOLUME 4, NUMBER 3 2004 Table 2 Univariate Analysis for the Association between Stroke Subtypes, According to the TOAST Classification, and Some Risk Factors* Variable LAAS CEI LAC ODE UDE Yes No RR (95% CI) Yes No RR (95% CI) Yes No RR (95% CI) Yes No RR (95% CI) Yes No RR (95% CI) Sex 1.2 (0.6 2.2) 0.7(0.3 2.0) 0.9 (0.4 1.9) 1.9 (0.5 6.3) 0.7 (0.3 1.8) M 37 53 12 78 22 68 5 85 14 76 F 31 38 7 62 16 53 7 62 8 61 Diabetes 0.5 (0.3 1.1) 0.4(0.1 1.2) 6.7 y (2.7 16.5) 0.2 z (0.04 0.9) 1.2 (0.5 3.1) No 40 40 13 67 7 73 10 70 10 70 Yes 28 51 6 73 31 48 2 77 12 67 Hypercholesterolemia 4.2 y (2.0 8.6) 0.4 (0.1 1.4) 0.8 (0.3 1.8) 0.4 (0.1 2.1) 0.1 y (0.0 0.7) No 36 75 16 95 28 83 10 101 21 90 Yes 32 16 3 45 10 38 2 46 1 47 Hypertension 1.1 (0.6 2.1) 0.3 z (0.1 0.8) 3.7 y (1.5 8.7) 0.5 (0.1 1.7) 0.6 (0.2 1.4) No 28 40 13 55 8 60 7 61 12 56 Yes 40 51 6 85 30 61 5 86 10 81 Cigarette smoking 2.4 z (1.2 4.6) 0.3 z (0.1 0.8) 1.0 (0.5 2.2) 0.8 (0.2 2.7) 0.5 (0.2 1.3) No 28 57 15 70 20 65 7 78 15 70 Yes 40 34 4 70 18 56 5 69 7 67 Previous stroke 1.9 (0.8 4.9) 0.3 (0.04 2.6) 2.2 (0.8 5.8) Undefined Undefined No 56 82 18 120 30 108 12 126 22 116 Yes 12 9 1 20 8 13 0 21 0 21 Previous TIA 1.2 (0.6 2.3) 1.1 (0.4 3.2) 3.0 y (1.4 6.4) 0.2 (0.0 1.6) Undefined No 47 66 13 100 20 93 11 102 22 91 Yes 21 25 6 40 18 28 1 45 0 46 *LAAS indicates large-artery atherosclerosis; CEI, cardioembolic infarct; LAC, lacunar infarct; ODE, Stroke of other determined etiology; UDE, stroke of undetermined etiology; TIA, transient ischemic attack; RR, relative risk; CI, confidence interval. y <.01. z <.05.

Table 3 Multivariate Analysis for the Association between Stroke Subtypes, According to the TOAST Classification, and Some Risk Factors* Variable LAAS CEI LAC ODE UDE Adjusted RR (95% CI) Adjusted RR (95% CI) Adjusted RR (95% CI) Adjusted RR (95% CI) Adjusted RR (95%CI) Sex 1.0 (0.5 2.1) 0.7 (0.3 2.2) 0.8 (0.4 1.9) 2.8 (0.8 10.2) 0.6 (0.2 1.8) Diabetes 0.6 (0.3 1.2) 0.4 (0.1 1.2) 7.5 y (2.7 20.9) 0.1 z (0.03 0.7) 1.5 (0.6 4.2) Hypercholesterolemia 3.4 y (1.6 7.2) 0.4 (0.1 1.7) 0.9 (0.3 2.4) 0.3 (0.1 1.8) 0.1 z (0.01 0.9) Hypertension 1.2 (0.6 2.5) 0.4 (0.1 1.2) 2.8 z (1.1 7.6) 0.8 (0.2 3.0) 0.7 (0.3 2.0) Cigarette smoking 2.0 z (1.0 3.9) 0.3 (0.1 1.0) 0.8 (0.3 1.9) 0.9 (0.2 3.2) 0.8 (0.3 2.4) Previous stroke 1.6 (0.5 4.5) 0.3 (0.03 2.7) 2.3 (0.6 7.9) Undefined Undefined Previous TIA 1.0 (0.5 2.3) 1.3 (0.4 4.0) 3.9 y (1.5 10) 0.2 (0.02 1.7) Undefined *LAAS indicates large-artery atherosclerosis; CEI, cardioembolic infarct; LAC, lacunar infarct; ODE, stroke of other determined etiology; UDE, stroke of undetermined etiology; TIA, transient ischemic attack; SSS, Scandinavian Stroke Scale; RR, relative risk; CI, confidence interval. y <.01. z <.05. CEREBROVASCULAR RISK FACTORS AND CLINICAL CLASSIFICATION OF STROKES/PINTO ET AL 293 according to the TOAST classification (LAAS, CEI, LAC, ODE, UDE), and some risk factors for stroke (hypertension, hypercholesterolemia, history of TIA, previous stroke). Table 7 shows the frequency and the distribution of exposure for each variable according to the matched triplets and the results of univariate and multivariate conditional logistic regression analyses. Diabetes was associated with LAC ischemic stroke subtype (OR 3.89, 95% CI, 2.23 6.80), with a history of hypertension (OR 2.53, 95% CI, 1.48 4.32), and inversely with a higher SSS at admission (OR 0.58, 95% CI, 0.36 0.96). The association of diabetes with LAC also remains significant after adjustment for hypertension (adjusted Table 4 One-Way ANOVA of SSS Scores at Inclusion between the Different Stroke Subtypes of the TOAST Classification* Stroke Subtype Mean SD F value P LAAS y 30.9 2.3 188.4 0.0001 CEI z 25.9 3.4 LAC 43.9 2.9 ODE yy 32.8 3.5 UDE zz 28.9 3.2 *SSS indicates Scandinavian Stroke Scale; LAAS, large-artery atherosclerosis; CEI, cardioembolic infarct; LAC, lacunar infarct; ODE, stroke of other determined etiology; UDE, stroke of undetermined etiology; and SD, standard deviation. y Bonferroni correction indicates that the mean score of SSS among patients with LAAS was different from those with CEI (p ¼ 0.0001), LAC (p ¼ 0.0001), ODE (p ¼ 0.01), and UDE (p ¼ 0.003). z Bonferroni correction indicates that the mean score of SSS among patients with CEI was different from those with LAAS (p ¼ 0.0001), LAC (p ¼ 0.0001), ODE (p ¼ 0.0001), and UDE (p ¼ 0.0005). Bonferroni correction indicates that the mean score of SSS among patients with LAC was different from those with LAAS (p ¼ 0.0001), CEI (p ¼ 0.0001), ODE (p ¼ 0.0001), and UDE (p ¼ 0.0001). yy Bonferroni correction indicates that the mean score of SSS among patients with ODE was different from those with LAAS (p ¼ 0.01), CEI (p ¼ 0.0001), LAC (p ¼ 0.0001), and UDE (p ¼ 0.0001). zz Bonferroni correction indicates that the mean score of SSS among patients with UDE was different from those with LAAS (p ¼ 0.003), CEI (p ¼ 0.0005), LAC (p ¼ 0.0001), and ODE (p ¼ 0.0001). OR 3.37, 95% CI, 1.90 5.99) or for LAAS and CEI (adjusted OR 2.69, 95% CI, 1.08 6.69). 26,27 In the period 1977 2001, the association between insulin resistance and risk for stroke has been examined in four case-control studies and five prospective observational cohort studies. Six of the nine studies are methodologically sound and provide evidence that insulin resistance is associated with risk for stroke. Naturally, drugs can reduce insulin resistance may have a role in stroke prevention. 28 The Finnish cohort trial evaluated the incidence of stroke in NIDDM and nondiabetic subjects and also measured other cardiovascular risk factors, metabolic control, and their impact on incidence in cerebrovascular risk. 20 In the nondiabetic population, hyperinsulinemia predicted stroke after univariate logistic regression analysis. A hypothesis of the investigators to explain this association involved the atherothrombotic origin of several strokes, where hyperinsulinemia and insulin resistance are risk factors for atherosclerosis. 20 Hyperinsulinemia was associated with the risk of stroke in the Helsinki Policemen Study, a trial based on a cohort of 640 men aged 34 to 64 years, free of cerebrovascular disease, other cardiovascular disease, or diabetes, in which the association of hyperinsulinemia with the risk of stroke during a 22-year follow-up was investigated. Risk factor measurements at baseline included an oral glucose tolerance test and plasma insulin measurement. During the follow-up, 70 men had a fatal or nonfatal stroke. The association between hyperinsulinemia and the risk of stroke (age-adjusted hazard ratio, 2.12; 95% CI, 1.28 to 3.49) was not independent of other risk factors, particularly obesity. Of other risk factors, upper body obesity, blood pressure, and smoking were each independent predictors for increased risk of stroke. 29 MICROALBUMINURIA AND PROTEINURIA A prospective case-control study involving 186 older men and women (individuals with recent ischemic stroke

294 SEMINARS IN VASCULAR MEDICINE/VOLUME 4, NUMBER 3 2004 Table 5 Rankin Disability Scale Score at Discharge and 6 Months after Discharge According to the Stroke Subtypes of the TOAST Classification* Variable Rankin Disability Scale at Discharge Rankin Disability Scale 6 Months after Discharge Univariate Analysis Univariate Analysis Score Multivariate Analysis Score Multivariate Analysis 1 3 4 6 RR (95%CI) Adjusted RR (95%CI) 1 3 4 6 RR (95%CI) Adjusted RR (95%CI) LAAS 0.9 (0.4 2.3) 1.5 (0.2 10.4) 0.6 (0.2 1.5) 0.9 (0.1 6.1) No 78 13 70 15 Yes 59 9 58 7 CEI 3.6 z (1.2 10.7) 4.5 (0.6 31.6) 6.7 y (2.3 19.9) 7.1 z (1.0 50.3) No 124 16 118 14 Yes 13 6 10 8 LAC Undetermined 0.005 y (0.0007 0.03) Undetermined 0.07 y (0.01 0.5) No 99 22 90 22 Yes 38 0 38 0 ODE 0.5 (0.1 4.4) 0.9 (0.1 6.2) Undetermined 0.07 y (0.01 0.5) No 126 21 118 22 Yes 11 1 10 0 UDE 2.8 (1.0 8.3) 3.6 (0.5 25.7) 4.5 y (1.5 13.0) 4.1 (0.6 29.4) No 121 16 116 15 Yes 16 6 12 7 *LAAS indicates large-artery atherosclerosis; CEI, cardioembolic infarct; LAC, Lacunar infarct; ODE, stroke of other determined etiology; UDE, stroke of undetermined etiology; RR, relative risk; CI, confidence interval. z <.01. z <.05. or TIA or having similar clinical risk factors for stroke) was performed to determine the incidence of microalbuminuria, its relationship to risk factors for stroke, its prevalence in the major subtypes of ischemic stroke, and Table 6 Thirty-Day Survival According to the Different Stroke Subtypes of the TOAST Classification* Variable Alive Subjects Death RR (95% CI) LAAS 0.4 (0.1 2.2) No 85 6 Yes 66 2 CEI 5.1 (1.1 23.2) No 135 5 Yes 16 3 LAC Undetermined No 113 8 Yes 38 0 ODE Undetermined No 139 8 Yes 12 0 UDE 4.2 (0.9 18.9) No 132 5 Yes 19 3 *LAAS indicates large-artery atherosclerosis; CEI, cardioembolic infarct; LAC, lacunar infarct; ODE, stroke of other determined etiology; UDE, stroke of undetermined etiology; RR, relative risk; CI, confidence interval. its potential for identifying patients at increased risk for recurrent stroke, myocardial infarction, or vascular death. Microalbuminuria was three times more prevalent in patients with recent stroke (29%) than in those with a clinical risk factor for stroke (10%). Among patients with recent stroke, the prevalence of microalbuminuria did not differ among major ischemic stroke subtypes. During a mean 1.5 0.9 years of follow-up 20% of patients with recent stroke, 14% with risk factors for stroke, and 0% of healthy elderly volunteers had vascular endpoints. After controlling for major clinical risk factors, microalbuminuria remained an independent significant predictor of future stroke in combined recent stroke and remote stroke or TIA. These data suggest that microalbuminuria merits further examination as a potentially inexpensive and easily measured marker of increased risk for stroke. 30 To assess the cerebrovascular risk in patients with NIDDM and increased urinary protein excretion rate and to determine whether the proteinuria is an independent risk factor for cardiovascular disease in patients with NIDDM, a case-control study was performed involving 59 diabetic patients with first-ever ischemic stroke due to thrombotic arterial occlusion and 180 diabetic patients without stroke as a control group. In a multivariate logistic regression analysis, the ORs and 95% CIs for the variables identified as risk factors for stroke were as follows: systolic pressure, diastolic pressure, fasting glucose > 11.1 mmol, HbA 1c > 9.5%, and proteinuria.

CEREBROVASCULAR RISK FACTORS AND CLINICAL CLASSIFICATION OF STROKES/PINTO ET AL 295 Table 7 Diabetes Mellitus and Ischemic Stroke: Case-Control Analysis Exposure Frequency (%) Discordant Triplets Concordant Triplets Variables Cases Controls þþ þ þ þ þ þ þþ þþþ OR (95% CI) p Large artery atherosclerotic stroke 35 50 17 12 35 18 7 13 0.38 (0.18 0.81) 0.01* Cardioembolic 10 22 2 7 28 6 1 58 0.57 (0.35 0.92) 0.02* Lacunar 46 17 16 31 15 2 0 38 3.89 (2.23 6.80) < 0.01* y Other determined etiology 1 7 1 0 13 0 0 88 0.13 (0.02 1.02) 0.05 Undetermined etiology 8 4 0 8 8 0 0 86 2.00 (0.75 5.33) 0.17 Hypercholesterolemia 45 50 13 14 24 14 19 18 0.77 (0.46 1.30) 0.33 Hypertension 85 54 42 14 13 7 21 5 2.53 (1.48 4.32) < 0.01 y Previous stroke 25 22 10 15 25 4 1 47 1.28 (0.73 2.24) 0.39 Previous TIA 31 27 14 16 29 4 2 37 1.24 (0.74 2.07) 0.42 SSS at admission ( 34 vs < 34) 40 17 11 29 21 0 1 40 0.58 (0.36 0.96) 0.03 Rankin at admission (1 3 vs 4 6) 11 17 4 7 27 2 0 62 0.59 (0.29 1.21) 0.15 Rankin after 6 months (1 3 vs 4 6) 13 15 5 8 22 2 0 65 0.81 (0.40 1.64) 0.56 Survival after 30 days 2 3 1 1 6 0 0 94 0.54 (0.10 2.79) 0.46 *In a conditional logistic regression analysis the risk of lacunar stroke subtype was adjusted by large artery ischemic and cardioembolic stroke subtypes (lacunar: adjusted OR ¼ 2.69, 95% CI ¼ 1.08 6.69, p ¼ 0.03; large artery ischemic stroke: adjusted OR ¼ 0.71, 95% CI ¼ 0.28 1.79, p ¼ 0.47; cardioembolic: adjusted OR ¼ 0.38, 95% CI ¼ 0.12 1.22, p ¼ 0.10). y In a conditional logistic regression analysis the risk of lacunar stroke subtype was adjusted by hypertension (lacunar: adjusted OR ¼ 3.37, 95% CI ¼ 1.90 5.99, p ¼ 0.0001; hypertension: adjusted OR ¼ 1.96; insulin resistance is associated with metabolic and cellular events that promote atherosclerosis. Resistance to insulin-mediated glucose uptake by peripheral tissues is a cardinal defect in type 2 diabetes mellitus, is also common in nondiabetic individuals, and may be an important risk factor for stroke.

296 SEMINARS IN VASCULAR MEDICINE/VOLUME 4, NUMBER 3 2004 Proteinuria was identified in 70 subjects, 45 with ischemic stroke and 25 of the control group. In both groups, all the subjects with proteinuria had a positive microalbumin test, which was negative in all the subjects without proteinuria. Therefore, proteinuria can be used as an independent risk factor for stroke in patients with NIDDM. The mechanisms of the association between proteinuria and cardio- and cerebrovascular disease are poorly understood. It has been proposed that albuminuria is associated with the increase of both albumin and fibrinogen transcapillary escape rate, which reflects widespread vascular damage, or endothelial dysfunction. Albuminuria has been shown to be related to increased extravascular coagulation, which leads to increased release of von Willebrand factor, contributing to the formation of microthrombi and platelet plugs, followed by areas of nonperfusion. 31 CHOLESTEROL The relation between serum total cholesterol concentration and coronary heart disease has been well established. The association between serum cholesterol and stroke is less well studied and unresolved. The first major study performed to collect data on the association of serum total cholesterol, different types of stroke diagnosed by autopsy, and distribution of stenosis in cerebral arteries is the Akita Pathology Study, which provides a database on 750 autopsied men aged 30 years and older who were admitted to a local hospital in northeast Japan between 1966 and 1984. The ageadjusted mean value of serum total cholesterol concentration was 164 mg/ dl for cerebral hemorrhage, 177 mg/ dl for infarction in penetrating artery regions (lacunar infarction), and 200 mg/ dl for infarction in cortical artery regions (embolic infarction). There were positive associations of serum cholesterol with stenosis of basal and penetrating arteries, indicating that high cholesterol levels contribute to the development of atherosclerosis in the cerebra arteries as in coronary arteries. The study shows a distinct difference in both pathology and serum total cholesterol levels among stroke types. 32 The Multiple Risk Factor Intervention Trial (MRFIT) is a large trial (350,977 men) that examined the relation between the serum total cholesterol level and the risk of death from stroke during 6 years of follow-up. The population studied, aged 35 to 37 years, had no history of heart attack and were not currently being treated for diabetes mellitus. The diagnosis was not based on autopsy but obtained from death certificates. Using proportional-hazards regression to control for age, cigarette smoking, DBP, and race or ethnic group, the study found that there was a striking difference in the relation of the serum cholesterol level to death from intracranial hemorrhage as compared with its relation to death from nonhemorrhagic stroke; the association with intracranial hemorrhage was inverse, and that with nonhemorrhagic stroke was positive. The inverse association of the serum cholesterol level with the risk of death from intracranial hemorrhage was confined to men with DBP > 90 mm Hg in whom death from intracranial hemorrhage is relatively common. 33 A similarly designed trial but involving females is the Women s Pooling Project; this trial evaluated the risk of death caused by total stroke, nonhemorrhagic stroke, and hemorrhagic stroke by race, age, and cholesterol quintile in 24,343 women with no previous cardiovascular disease combining data from eight long-term prospective studies with long-term follow-up. After a multivariate analysis, black women younger than 55 years had a 76% increased risk of death caused by stroke compared with white women (RR, 1.76; 95% CI, 1.10 to 2.8). For black women younger than 55 years, the top compared with the lowest cholesterol quintile remained an independent predictor of stroke mortality (RR 2.58; 95% CI, 1.05 to 6.32). For white women younger than 55 years, the top compared with the lowest cholesterol quintile did not predict stroke mortality with significance (RR 1.47; 95% CI, 0.57 to 3.76). In analogous multivariate models, a positive relation was found between continuous cholesterol and nonhemorrhagic stroke death in women younger than 55 years (RR 1.23; 95% CI, 1.02 to 1.49). 34 From 1989 through the 1990s, clinical intervention trials concerning b-hydroxy-b-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors (statins) with the aim of evaluating the statin-related reduction of lowdensity lipoprotein cholesterol, cardiovascular events, and cerebrovascular events resulted in a highly significant decrease in the occurrence of fatal and nonfatal cerebrovascular events. In the 4S (Scandinavian Simvastatin Survival Study), 35 CARE (Cholesterol And Recurrent Event), 36 and LIPID (Long-term Intervention with Pravastatin in Ischemic Disease) 37 studies the cerebrovascular event relative risk reduction from statins was 28, 32, and 19%, respectively, in the treated and control groups. In the WOSCOP (West of Scotland Coronary Prevention) trial, pravastatin caused a nonsignificant decrease of stroke incidence (6%). 38 In the AFCAPS/TEXCAPS (Air Force/Texas Coronary Atherosclerosis Prevention Study), involving subjects with a normal lipid profile except for low HDL cholesterol, an analogous treatment seemed to be more effective ( 22%, but the population involved in this study was too small). 39 The HPS (Heart Protection Study) enrolled 20,536 subjects between 40 and 80 years, with history of coronary disease, diabetes, or other vascular peripheral disease, treated with 40 mg of simvastatin or placebo. During the 5 years of follow-up, considering the

CEREBROVASCULAR RISK FACTORS AND CLINICAL CLASSIFICATION OF STROKES/PINTO ET AL 297 endpoint stroke, the authors observed a mean reduction of 25% (4.3% in the simvastatin group versus 5.7% in the placebo group). 40 The latest of these, the ASCOT (Anglo-Scandinavian Cardiac Outcomes Trial) study, was designed to demonstrate the utility in primary prevention of cardiovascular and cerebrovascular disease of statins in patients with high blood pressure and at least three other cardiovascular risk factors but not dyslipidemia. The 19,342 patients enrolled were treated with atorvastatin 10 mg/day or placebo. The study was interrupted with a mean follow-up of 3.3 years of the 5 years previsioned. A significant reduction of fatal myocardial infarction was observed in the atorvastatin group versus the placebo group (RR 0.64; 95% IC, 0.50 0.83). Also, fatal and nonfatal stroke was significantly reduced (RR 0.73; 95% IC, 0.56 0.96), a 27% reduction between the two groups, and total cardiovascular events. 41 OBESITY The best universally accepted method to evaluate obesity is the calculation of body mass index (BMI), defined as body weight in kg divided by the square of height in m 2 ; a value over 29.9 indicates obesity. The Nurse s Health Study examined the associations of BMI and weight change with risk of stroke in women and included 116,759 women aged 30 to 55 years free from diagnosed coronary heart disease or stroke. Primary outcomes were ischemic stroke and hemorrhagic stroke. During 16 years of follow-up, the documented stroke, adjusted for age, smoking, postmenopausal hormone use, and menopausal status, showed a significantly increased risk of ischemic stroke in women with increased BMI ( 27 kg/m 2 ) with an RR of 1.75 (95% CI, 1.17 2.59) for BMI of 27 to 28.9; 1.90 (95% CI, 1.28 2.82) for BMI of 29 to 31.9; 2.37 (95% CI, 1.60 3.50) for BMI of 32 or more, as compared with those with a BMI of less than 21 kg/m 2. The association with an increased risk of ischemic stroke was also observed analyzing weight gain: RR of 2.52 (95% CI, 1.80 3.52) for a gain of 20 kg or more. For hemorrhagic stroke there was a no significant inverse relation. 42 More generally, the ARIC (Atherosclerosis Risk In Communities) study tested the hypothesis that diabetes, body fat distribution, and (in nondiabetic subjects) fasting insulin levels are positively associated with ischemic stroke incidence in the general population. The study included a cohort of 15,792 adults aged 14 64 years who had no cardiovascular disease at baseline, observing them for 6 8 years for ischemic stroke occurrence. After adjustment for age, sex, race, smoking, and educational level, the RR of ischemic stroke was 3.70 (95% CI, 2.7 5); for diabetes, 1.74 (1.4 2.2) for a 0.11 increment of waist-hip ratio and 1.19 (1.1 1.3) for a 50 pmol/ L increment of fasting insulin among nondiabetic subjects. Ischemic stroke incidence was not statistically significantly associated with BMI (comparably adjusted RR ¼ 1.15, 95% CI, 0.97 1.36). The conclusions of this trial confirm, after adjustments for other stroke risk factors (some of which may mediate the effects of diabetes), fat distribution, and hyperinsulinemia, that diabetes is a strong risk factor for ischemic stroke. Insulin resistance, as reflected by elevated waisthip ratios (measure of fat distribution) and elevated fasting insulin levels, may also contribute to a greater risk of ischemic stroke. 43 Another prospective cohort study involving 21,414 U.S. male physicians, the Physicians Health Study, evaluated the association of BMI with risk of ischemic and hemorrhagic stroke. During the 12.5 years of follow-up, 747 strokes occurred. Compared with participants with BMIs less than 23, when BMI was evaluated as a continuous variable, each unit increase of BMI was associated with a significant 6% increase in the adjusted relative risks of total, ischemic, and hemorrhagic stroke. The increase of the relative risk appeared to be independent of the effects of other major risk factors for stroke such as hypertension, diabetes, or hypercholesterolemia. 44 HYPERURICEMIA The significance of hyperuricemia as an independent risk factor for cardio- and cerebrovascular disease has remained uncertain, even when it is a common finding in patients with NIDDM. Letho et al investigated serum urate as a predictor of stroke in 1017 NIDDM patients free of clinical nephropathy. During the follow-up period of 7 years, the incidence of stroke increased significantly by quartiles of serum uric acid levels (p < 0.01). High uric acid level (above the median value of > 295 mmol/l) was significantly associated with the risk of fatal and nonfatal stroke by Cox regression analysis (RR 1.93, p ¼ 0.001), indicating that hyperuricemia is a predictor of stroke events in middle-aged patients with NIDDM independently of other weak cardiovascular risk factors. 45 Another study developed in the United Kingdom in 2003 studied 3731 patients with a first ischemic stroke, determining the association of urate level with 90-day clinical outcome, recurrence of acute stroke, new occurrence of myocardial infarction, or vascular death. In this trial elevated urate levels predicted a lower chance of a good clinical outcome (OR 0.78 per additional 0.1 mmol/ L) independently of stroke severity and other prognostic factors. Vascular event risk increased with urate level (RR 1.27 per additional 0.1 mmol/l), particularly in the presence of diabetes (additional RR 1.22 per additional 0.1 mmol/l). 46

298 SEMINARS IN VASCULAR MEDICINE/VOLUME 4, NUMBER 3 2004 The role of urate in stroke pathophysiology remains uncertain; is possible that serum uric acid is a factor not strongly linked to the development of ischemic stroke but is a secondary less important phenomenon than other risk factors of stroke (NIDDM, obesity, hypertension, nephropathy). ATRIAL FIBRILLATION Nonvalvular AF, the most common sustained cardiac arrhythmia, is associated with a four- to sixfold increased risk of ischemic stroke because of the embolization of thrombus formed within the left atrial appendage as shown in several large-scale randomized clinical trials for prevention of thromboembolism in nonrheumatic AF performed during the 1990s. The first five major trials Copenhagen Atrial Fibrillation Aspirin Anticoagulant Trial (AFASAK), 47 Stroke Prevention in Atrial Fibrillation (SPAF I-III), 48 Boston Area Anticoagulation Trial for Atrial Fibrillation (BAATAF), 49 Canadian Atrial Fibrillation Anticoagulation (CAFA) Study, 50 and Stroke Prevention in Non-rheumatic Atrial Fibrillation (SPINAF) study 51 for primary prevention of stroke clearly showed the increased risk of stroke in the patients with AF. These five major trials have also demonstrated a significant risk reduction (68%) for stroke with oral anticoagulation without any significant increase in major hemorrhage. The incidence of stroke in nonvalvular AF patients is similar in paroxysmal AF and in chronic AF. Epidemiological studies have identified clinical and echocardiographic factors associated with increased stroke risk in AF (older age, prior stroke or TIA, hypertension, diabetes mellitus, and the presence of left ventricular systolic dysfunction), but mechanisms linking these factors to stroke in AF are incompletely understood. The SPAF III study evaluated von Willebrand factor and soluble P-selectin; these factors are hypothesized to be associated with increased endothelial damage or dysfunction and platelet activation among patients with AF. In this study, 1321 AF patients were studied for the presence of stroke risk factors and cardiovascular disease, relating these to plasma levels of vwf and soluble P-selectin. Increased levels of vwf were independently associated with age, prior cerebral ischemia, recent heart failure, diabetes, and BMI; increased plasma levels of P-selectin were associated with diabetes, peripheral vascular disease, and smoking. The results show that four well-recognized risk factors for stroke in AF (advancing age, prior cerebral ischemia, recent heart failure, diabetes) were independently associated with raised plasma levels of vwf whereas only diabetes was associated with increased soluble P-selectin. Further studies are needed to establish the true prognostic importance of vwf (endothelial damage or dysfunction) and P-selectin (platelet activation) as potential contributors to thromboembolic stroke in AF. 52 Stroke recurrence and mortality in patients with AF and a first ischemic stroke were evaluated in Finland in a study of 2635 patients aged 75 years and older. After 1 year of follow-up the mortality was higher in the AF group than the non-af group (1.24: 95% CI, 1.10 1.39 p < 0.01). The strongest risk factor predicting 1-year mortality was recent myocardial infarction (RR 1.90, 95% CI, 1.49 2.42). The 1-year recurrence rate among those alive after the first stroke event was 11.5% in the AF group and 9.4% in the non-af group. The authors concluded that recent myocardial infarction and AF are independent negative prognostic factors in older patients with stroke. Although the relative risk estimates attributable to AF are of the same magnitude in older as in middle-aged stroke patients, the much higher prevalence of AF in the older patients emphasizes its absolute impact on the mortality and recurrence after the first ischemic stroke in the age group 75 years and older. 53 Another study to determine stroke recurrence in patients suffering from AF, particularly elderly patients, was performed in a retrospective cohort study of 915 patients aged 50 to 94 years admitted to the Internal Medicine Department of the Valdecilla Hospital of Santander (Spain) for ischemic stroke. Of the 829 stroke patients who survived the initial hospitalization, 163 had a stroke during the follow-up. Among the patients with AF not anticoagulated, 27% had recurrent strokes, compared with 18% among those with AF who were anticoagulated and 17% among those without AF. The age-adjusted hazard ratio for recurrent stroke among those with AF who were not treated with anticoagulants was 2.1 (95% CI, 1.4 to 2.9, p < 0.0001). The increased risk was observed even in patients older than 80 years and persisted during the follow-up for more than 5 years. 54 A similar prospective study, enrolling 2101 older patients (mean age 81 years) of whom 13% suffered from AF, was performed to evaluate, after a mean follow-up of 42 months, the association between AF and new incidence of stroke in patients with and without prior thromboembolic stroke. Previous thromboembolic stroke occurred in 43% of AF patients and 24% of the sinus rhythm patients. New stroke occurred in 46% of patients with AF and in 17% of the nonarrhythmic patients. A multivariate Cox regression model confirmed that independent risk factors for new thromboembolic stroke were male sex (RR 1.3), prior stroke (RR 3.1), and AF (RR 3.3). 55 LEFT VENTRICULAR HYPERTROPHY Left ventricular hypertrophy is a risk factor for cardiovascular events, but its effect on ischemic stroke risk is established mainly in whites. A population-based case-control study to evaluate whether left ventricular