Silent brain infarctions and high white matter grade

Transcription

1 Hyperintensity on Cranial Magnetic Resonance Imaging A Predictor of Stroke Lewis H. Kuller, MD, DrPH; W.T. Longstreth Jr, MD, MPH; Alice M. Arnold, PhD; Charles Bernick, MD; R. Nick Bryan, MD, PhD; Norman J. Beauchamp Jr, MD, MHS; for the Cardiovascular Health Study Collaborative Research Group Background and Purpose We have previously reported that several silent infarcts found on magnetic resonance imaging (MRI) were a risk factor for stroke. Several recent reports have shown that high white matter grade (WMG) and increasing WMG over time were risk factors for stroke. We tested the hypothesis that high WMG 2 was a predictor of risk for stroke, independent of other risk factors. Methods We examined the extent of white matter hyperintensity on cranial MRI of 3293 participants from the Cardiovascular Health Study (CHS). The degree of white matter hyperintensity was graded from least severe (grade 0) to most severe (grade 9). Participants were followed-up for an average of 7 years for the occurrence of a stroke. Clinical stroke diagnoses were based on hospital records reviewed by an adjudication committee expert in stroke diagnosis. During this period, 278 strokes occurred. Results The relative risk of stroke increased significantly as the WMG increased. The risk of stroke was 2.8% per year for participants with high WMG (grades 5), compared with only 0.6% for participants with grades 0 to 1. Conclusions The risk of stroke with high WMG is independent of traditional stroke risk factors and persists when controlling for MRI infarcts, another subclinical imaging marker of cerebrovascular disease. Assessment of white matter disease may be valuable in assessing future risk of stroke. (Stroke. 2004;35: ) Key Words: stroke magnetic resonance imaging white matter infarcts, silent hypertension diabetes mellitus Silent brain infarctions and high white matter grade (WMG), referred to as leukoaraiosis, are frequently observed on magnetic resonance imaging (MRI) scans. 1,2 These findings are associated with gait and cognitive and neurobehavioral disturbances, 3,4 and are likely of vascular origin. 5 8 High WMG is a stronger predictor of lacunar infarcts and is most likely associated with small-vessel pathology. 9 Progression of leukoaraiosis over a 6-year follow-up was associated with increased risk of stroke. 10 The Rotterdam Scan Study has recently reported that participants in the highest tertile of white matter lesion distribution had a 4.7-fold increased risk of stroke for periventricular and 3.6-fold for subcortical white matter lesions. 11 Cerebral MRI examinations for 3660 participants were performed in the Cardiovascular Health Study (CHS) MRIdetected infarcts and higher WMG were strongly associated with carotid intimal medial thickness and stenosis even after adjustment for age and sex (P 0.02). 13 In the multivariate model, age, infarcts on MRI, higher systolic blood pressure, low forced expiratory volume in 1 second, and lower income were associated with higher WMG. 12 Silent predominantly subcortical brain infarcts were associated with a 1.9-increased (1.2 to 2.8) risk of stroke over a 4.2-year follow-up. Risk of stroke increased with number of silent infarcts. 17 The current report explores the relationship between white matter findings on MRI, other stroke risk factors, including subcortical brain infarcts, and subsequent incidence of stroke, with an average follow-up of 7 years. 18 Materials and Methods Participant Selection Participants in the CHS study were recruited in 4 communities from Medicare eligibility lists. There were 5201 participants enrolled from 1989 to 1990, and an additional 687 black participants were added in 1992 to 1993, bringing the total to Details of CHS design and recruitment have been described elsewhere. 19 Received December 17, 2003; final revision received April 21, 2004; accepted April 22, From the Department of Epidemiology (L.H.K.), University of Pittsburgh, Pittsburgh, Pa; the Departments of Epidemiology and Neurology (W.T.L.), Biostatistics (A.M.A.), and Radiology (N.J.B.), University of Washington, Seattle, Wash; the Division of Neurology (C.B.), University of Nevada, Las Vegas, Nev; and the Department of Radiology (R.N.B.), University of Pennsylvania, Philadelphia, Pa. Correspondence to Dr Lewis H. Kuller, Department of Epidemiology, University of Pittsburgh, 130 N. Bellefield Avenue, Room 550, Pittsburgh, PA KullerL@edc.pitt.edu 2004 American Heart Association, Inc. Stroke is available at DOI: /01.STR

2 1822 Stroke August 2004 Of the total cohort of 5888, 3660 participants without contraindication who consented underwent MRI in a standard fashion between 1991 and To determine the role of WMG as a predictor of incident stroke, 338 patients with stroke or transient ischemic attack (TIA) before the baseline MRI were excluded. Twenty-nine individuals were also excluded because of incomplete MRI data, resulting in 3293 participants included in this analysis. MRI MRI was performed on General Electric or Picker 1.5-tesla scanners at 3 field centers and on a 0.35-tesla Toshiba scanner at the fourth center. The scanning protocol included standard sagittal and axial spin-echo T1-weighted images (TR/TE 500/15 to 25) and axial spin-density and T2-weighted images (TR/TE 3000/20 to 35/70 to 100), all with 5-mm thickness and no interslice gap. 13,14 Axial scans were aligned parallel to a line to the anterior and posterior commissures (AC PC line). Scans were all performed without contrast administration. The white matter signal changes of each individual were assessed on a semi-quantitative 10-point WMG (0 to 9) scale using predefined visual standards of 8 reference cases. WMG was estimated as the total extent of periventricular and subcortical white matter signal abnormality on spin-density weighted axial images that successively increased from no or barely detectable changes (grades 0 and 1, respectively) to almost all white matter involved (grade 9). 16 This scale has an interreader reliability agreement within 1 grade of 92.1%, with relaxed kappa of 0.81; intrareader reliability for agreement within 1 grade is 94.5%, with relaxed kappa of Volumetric analytic validation of the visual scale corresponded to a rank increase in white matter hyperintensity normalized for cerebral parenchymal volume. Abnormalities interpreted as representing areas of large-vessel cerebral infarction or small-vessel lacunar infarction were coded separately in the database as infarct-like lesions 12 (Appendix and Table 1). Event Ascertainment Event ascertainment followed a detailed protocol at each of the field centers. 20 Incident stroke was the outcome of interest in the analysis used in this article. 21 Participants without stroke or TIA at the time of the MRI scan were considered to be at risk for stroke. The CHS Events Stroke Subcommittee adjudicated all of the events based on defined criteria. 21 The stroke events committee decided by consensus not only by stroke occurrence but also by stroke type and subtype. 21 The type of stroke was classified by history and clinical observations as well as available MRI or computed tomography obtained at the time of clinical evaluation. Adjudicated incident strokes occurring after the baseline MRI (1992 to 1994) and before July 1, 2000, TABLE 1. Risk Factor and Outcome Comparison for Participants in Study and Those Excluded Risk Factor In Analysis (3293) Excluded* (1498) P Black Male Hypertensive History of CVD Diabetes [American Diabetes Association (ADA)] Normal Impaired fasting glucose Diabetic Creatinine Atrial fibrillation Left ventricular hypertrophy Systolic blood pressure Age, y Internal carotid wall thickness, mm Incident strokes, N per 1000 person-years *Participants with a contact in 1992 to 1993 who had no history of stroke or TIA at the time and were excluded from analysis because white matter grade was not measured. Some participants were contacted by phone or proxy and did not have a clinic visit, resulting in missing data. Total Ns are given for variables with missing data. Impaired fasting glucose: 110 mg/dl glucose 126 mg/dl. Adjusted for age, race, and sex. are included in this analysis. The median follow-up time was 7 years; interquartile range was 6.2 to 7.5 years, with a maximum of 8.6 years. Statistical Methods All analyses were performed using SPSS for Windows, version 10. Values of risk factors for stroke were determined by using participant information obtained from the annual examination closest in time and before the MRI. 19 Values of risk factors for participants in the analysis were compared against those of participants not in the analysis for Distribution of white matter grade and number of infarcts on MRI 3 mm.

3 Kuller et al Hyperintensity and Risk of Stroke 1823 TABLE 2. Incidence and Relative Risk of Stroke by # Strokes/ # At Risk Incidence Rate* Hazard Ratio Hazard Ratio Hazard Ratio 0 or 1 48/ (reference) 1.0 (reference) 1.0 (reference) 2 69/ (1.1, 2.3) 1.5 (1.0, 2.2) 1.4 (0.9, 2.0) 3 83/ (2.1, 4.3) 2.8 (1.9, 4.0) 2.4 (1.6, 3.4) 4 44/ (2.7, 6.2) 3.8 (2.5, 5.8) 2.9 (1.9, 4.6) 5 34/ (2.1, 5.3) 3.0 (1.9, 4.7) 2.2 (1.4, 3.7) 5 19/ / / / /7 0.0 *Unadjusted rates per 1000 person-years. Adjusted for clinic and age at the time of the MRI. Adjusted for clinic, age, gender, race, systolic blood pressure, diabetes (ADA definition), CVD (angina, MI, claudication, or CHF), and atrial fibrillation (by ECG). Further adjusted for number of large infarcts on MRI, left ventricular hypertrophy by ECG, internal carotid wall thickness, and creatinine 1.25 mg/dl. whom data were collected from 1992 to 1993 and who had no history of stroke or TIA. A t test was used to compare differences in means and a 2 test was used for differences in proportions. WMG was scored on a scale of 0 to 9, with few participants scoring 0or 5. To obtain reliable estimates of the relative risk of stroke for each WMG, the upper and lower grades were collapsed. Incidence rates of stroke for the resultant WMG categories were computed by dividing the number of strokes in each category by the person-years at risk. Hazard ratios (HR) were determined from Cox proportional hazards models, with 3 stages of adjustment. Interactions between WMG and TABLE 3. Risk of Stroke for Subgroups of Participants Defined by Pairs of Risk Factors each risk factor were tested by likelihood ratio test for nested models. We computed stroke rates and HR for WMG 2 in combination with each of 5 known risk factors for stroke: hypertension, diabetes, history of myocardial infarction, coronary heart failure, and atrial fibrillation. The cutoff point of 2 was chosen to ensure adequate numbers in each cell. There was little change in risk at 2 in the analysis. We examined the risk associated with the presence of an MRIdetected infarct in combination with WMG scores by computing incidence rates and HR for each WMG category separately for those with and without infarcts. Strokes/1000 Person-Years Hazard Ratio for Incident Stroke for Paired Risk Factor Risk Factor Risk Factor-Adjusted* 2 2 Total 2 2 Hypertension No ( ) Yes ( ) 3.9 ( ) Diabetes (ADA) Normal ( ) IFG ( ) 2.4 ( ) Diabetic ( ) 2.7 ( ) History of MI No ( ) Yes ( ) 4.4 ( ) History of CHF No ( ) Yes ( ) 2.7 ( ) Atrial Fibrillation on ECG No ( ) Yes ( ) 9.3 ( ) *Risk factors include age, gender, race, CVD (MI, angina, claudication, CHF), diabetes, atrial fibrillation, and systolic blood pressure unless one of those is the variable of interest. When modeling hypertension, systolic blood pressure was not in the model. IFG indicates impaired fasting glucose. When modeling MI or CHF, adjusted for remaining 3 cardiovascular diseases.

4 1824 Stroke August 2004 TABLE 4. MRI Infarct and and Risk of Stroke Strokes/1000 Person-Years (Number at Risk) Hazard Ratio for Incident Stroke for Paired Risk Factors MRI Infarct Risk Factor-Adjusted* MRI Infarct No Yes No Yes (1005) 7.2 (170) 1.0 (reference) 1.1 (0.53, 2.5) (753) 14.6 (269) 1.4 (0.92, 2.2) 2.1 (1.3, 3.6) (414) 31.0 (222) 2.4 (1.6, 3.8) 4.2 (2.7, 6.5) (116) 34.1 (124) 3.7 (2.1, 6.5) 4.4 (2.6, 7.4) (92) 28.9 (128) 2.7 (1.4, 5.2) 3.7 (2.1, 6.5) Total 10.5 (2380) 21.4 (913) *Risk factors include age, gender, race, CVD (MI, angina, claudication, CHF), diabetes, atrial fibrillation, and systolic blood pressure. Results Participants who underwent MRI tended to be younger and healthier than those who did not (Table 1). WMG increased with age, with a trend toward greater severity in women. 3 Strokes occurred in 278 of the 3293 participants in the analysis, with an incidence of 13.3 per 1000 person-years. (Table 1) The distribution of WMG and number of MRI infarcts by WMG are shown in the Figure. The number of participants with WMG 2 is low. There is a strong association of WMG and number of infarcts. We adjusted the analysis for clinic in CHS because of the difference in scanners at the Johns Hopkins University clinic. There was no interaction between scanner type and WMG, or with risk of stroke. The incidence of stroke was 6.0 per 1000 person-years for grades 0 to 1, to 27.6 per 1000 person-years for grades 5, with an age-adjusted HR of 3.3 (Table 2). The HR was slightly attenuated to 3.0 after adjusting for known stroke risk factors and remained significant at 2.2 after additional adjustment for subclinical risk factors and infarcts (Table 2). For all levels of adjustment, P value for a linear trend in WMG was Risk of stroke was associated with WMG 2 whether a participant had hypertension, diabetes, history of myocardial infarction or congestive heart failure, atrial fibrillation, or none of these on electrocardiogram (Table 3). Risk of stroke was highest for participants with both higher WMG and a traditional risk factor. The combination conferring the greatest risk of stroke was WMG 2 and atrial fibrillation on electrocardiogram. The HR was 9.7 (CI, 5.5 to 17.1), with the wide confidence interval reflecting the fact that the estimate was based on only 36 participants with this combination of risk factors. There were 913 participants who had a silent brain infarct on the MRI (Figure). For each level of WMG, the risk of stroke was higher for participants with an infarct than for those with no infarct (Table 4). Similarly, for participants with an infarct, risk of stroke was greater in the presence of white matter disease. There was no significant interaction between WMG and presence of an infarct. The risk of stroke was increased for participants with high WMG, independent of the time from MRI to clinical stroke. Most of the clinical strokes in CHS were ischemic; 226 (81%) of 278 incident strokes were ischemic. The ischemic strokes were further subclassified into lacunar 33 (14.6%), cardioembolic 61 (27%), atherosclerotic 12 (5.3%), unknown 105 (46.5%), mixed 14 (6.2%), or other. 1 Higher WMG ( 5) was related to total ischemic stroke (HR, 2.86; CI, 1.70 to 4.80), cardioembolic stroke (HR, 4.82; CI, 1.68 to 13.8), and unknown ischemic stroke (HR, 2.84; CI, 1.35 to 5.97). There was an increased risk of lacunar strokes with higher WMG, but the number of cases was small (33) and rates were very unstable (HR, 1.99; CI, 0.59 to 6.73) for WMG of 4 versus 0 to 1. Discussion Our current results are consistent with a recent report from the Rotterdam group. 11 Increased WMG is an independent predictor of stroke. This study clarifies the potential value of this marker to future stroke risk. We have previously reported that high WMG was strongly related to vascular risk factors. 13 It is likely that high WMG is determined, in part, by smaller-vessel disease in the brain as reflected by the strong association of infarcts (mostly subcortical) 3 mm on MRI examination. The high WMG may also be a marker of inflammation secondary to ischemic injury. High WMG can be used to identify participants at high risk for stroke, especially in combination with other cardiovascular risk factors such as hypertension and diabetes. Better characterization of the relationship between pathology and MRI findings would be very useful. Improved methods of imaging may better-determine white matter pathophysiology. Postmortem comparison with premortem MRI is also needed to interpret WMG. Better determinants of the characteristics of WMG (phenotype) could result in better genetic and risk factor profiles. 22 It may be useful to measure progression of WMG as a marker of the efficacy of various pharmacologic and nonpharmacologic therapies. We need to determine: (1) age at which differences in extent of WMG are first identified between those with or without vascular risk factors; (2) rates of progression of WMG changes; (3) efficacy of specific therapies to slow WMG change; and (4) possible host susceptibility or genetic determinants of extent of WMG in relation to risk factor levels. Further evaluation of specific localization of WMG in the brain to risk factors and outcomes are also important.

5 Kuller et al Hyperintensity and Risk of Stroke 1825 Appendix Quartiles of and Brain Volume by MRI Semiquantitative Volume (cm 3 ) Brain Volume (cm 3 ) Volume (%) The semiquantitative assessment scale used in CHS was volumetrically validated comparing the semiquantitative grade to the ratio of white matter abnormality normalized to brain volume. Acknowledgments The research reported in this article was supported by contracts N01-HC through N01-HC-85086, N01-HC-35129, and N01 HC from the National Heart, Lung, and Blood Institute. For a full list of participating CHS investigators and institutions, see About CHS: Principal Investigators and Study Sites online at References 1. Hachinski VC, Merskey H. Leukoaraiosis. Arch Neurol. 1987;44: Clarke R, Joachim C, Esiri M, Morris JG, Bungay H, Molyneux A, Budge M, Frost C, King E, Barnetson L, Smith AD. Leukoaraiosis at presentation and disease progression during follow-up in histologically confirmed cases of dementia. In: Kalaria RN, Ince P, eds. Vascular Factors in Alzheimer s Disease. New York: New York Academy of Sciences; 2000: Longstreth WT Jr, Manolio TA, Arnold A, Burke GL, Bryan N, Jungreis CA, Enright PL, O Leary D, Fried L, for the Cardiovascular Health study Collaborative Research Group. Clinical correlates of white matter findings on cranial magnetic resonance imaging of 3301 elderly people. The Cardiovascular Health Study. Stroke. 1996;27: Ylikoski R, Ylikoski A, Erkinjuntti T, Sulkava R, Raininko R, Tilvis R. White matter changes in healthy elderly persons correlate with attention and speed of mental processing. Arch Neurol. 1993;50: Munoz D, Hastak SM, Harper B, Lee D, Hachinski VC. Pathological correlates of increased signals of the centrum semiovale on magnetic resonance imaging. Arch Neurol. 1993;50: Pantoni L, Garcia JH. Pathogenesis of leukoaraiosis. Stroke. 1997;28: Fazekas F, Englund E. White matter lesions. In: Erkinjuntti T, Gauthier S, eds. Vascular Cognitive Impairment. London: Martin Dunitz Ltd; 2002: Breteler MM, van Swieten JC, Bots ML, Grobbee DE, Claus JJ, van den Hout JH, van Harskamp F, Tanghe HL, de Jong PT, van Gijn J. Cerebral white matter lesions, vascular risk factors, and cognitive function in a population-based study: the Rotterdam Study. Neurology. 1994;44: Inzitari D. Leukoaraiosis: an independent risk factor for stroke? Stroke. 2003;34: Streifler JY, Eliasziw M, Benavente OR, Alamowitch S, Fox AJ, Hachinski V, Barnett HJM, for the North American Symptomatic Carotid Endarterectomy Trial Group. Development and progression of leukoaraiosis in patients with brain ischemia and carotid artery disease. Stroke. 2003;34: Vermeer SE, Hollander M, van Dijk EJ, Hofman A, Koudstaal PJ, Breteler MMB. Silent brain infarcts and white matter lesions increase stroke risk in the general population. The Rotterdam Scan Study. Stroke. 2003;34: Longstreth WT Jr, Bernick C, Manolio TA, Bryan N, Jungreis CA, Price TR, for the Cardiovascular Health Study Collaborative Research Group. Lacunar infarcts defined by magnetic resonance imaging of 3660 elderly people: the Cardiovascular Health Study. Arch Neurol. 1998;55: Manolio T, Burke GL, O Leary DH, Evans G, Beauchamp N, Knepper L, Ward B, for the CHS Collaborative Research Group. Relationship of cerebral MRI findings to ultrasonographic carotid atherosclerosis in older adults: the Cardiovascular Health Study. Arterioscler Thromb Vasc Biol. 1999;19: Bryan RN, Manolio TA, Schertz LD, Jungreis C, Poirier VC, Elster AD, Kronmal RA. A method for using MR to evaluate the effects of cardiovascular disease on the brain: the Cardiovascular Health Study. Am J Neuroradiol. 1994;15: Manolio TA, Kronmal RA, Burke GL, Poirier V, O Leary DH, Gardin JM, Fried LP, Steinberg EP, Bryan RN, for the Cardiovascular Health Study Collaborative Research Group. Magnetic resonance abnormalities and cardiovascular disease in older adults. The Cardiovascular Health Study. Stroke. 1994;25: Yue NC, Arnold AM, Longstreth WT Jr, Elster AD, Jungreis CA, O Leary DH, Poirier VC, Bryan RN. Sulcal, ventricular, and white matter changes at MR imaging in the aging brain: data from the Cardiovascular Health Study. Radiology. 1997;202: Bernick CB, Kuller L, Dulberg C, Longstreth WT Jr, Manolio T, Beauchamp N, Price T, for the Cardiovascular Health Study Collaborative Research Group. Silent MRI infarcts and the risk of future stroke: the Cardiovascular Health Study. Neurology. 2001;57: Longstreth WT Jr, Diehr P, Beauchamp NJ, Manolio TA. Patterns on cranial magnetic resonance imaging in elderly people and vascular disease outcomes. Arch Neurol. 2001;58: Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A, O Leary DH, Psaty B, Rautaharju P, Tracy RP, Weiler PG, for the Cardiovascular Health Study Research Group. The Cardiovascular Health Study: design and rationale. Ann Epidemiol. 1991;1: Ives DG, Bild DE, Psaty BM, Kuller LH, Crowley PM, Cruise RG, Theroux S. Surveillance and ascertainment of cardiovascular events. The Cardiovascular Health Study. Ann Epidemiol. 1995;5: Price TR, Psaty B, O Leary D, Burke G, Gardin J, for the Cardiovascular Health Study Research Group. Assessment of cerebrovascular disease in the Cardiovascular Health Study. Ann Epidemiol. 1993;3: Bowler JV. The progression of leukoaraiosis [editorial comment]. Stroke. 2003;34: