Association of White Matter Lesions and Lacunar Infarcts With Executive Functioning

American Journal of Epidemiology Advance Access published September 25, 2009 American Journal of Epidemiology ª The Author 2009. Published by the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org. DOI: 10.1093/aje/kwp256 Original Contribution Association of White Matter Lesions and Lacunar Infarcts With Executive Functioning The SMART-MR Study Mirjam I. Geerlings, Auke P. A. Appelman, Koen L. Vincken, Willem P. T. M. Mali, and Yolanda van der Graaf for the SMART Study Group Initially submitted January 14, 2009; accepted for publication July 24, 2009. The authors investigated the association of white matter lesions and with cognitive performance and whether brain atrophy mediates these associations. Within the Second Manifestations of Arterial Disease- Magnetic Resonance study (2001 2005, the Netherlands), cross-sectional analyses of 522 patients were performed (mean age, 57 years (standard deviation, 10); 76% male). Brain segmentation was used to quantify volumes of brain tissue, cerebrospinal fluid, and white matter lesions. Infarcts were rated visually. Brain volume, ventricular volume, and gray matter volume were divided by intracranial volume to obtain indicators of brain atrophy. Neuropsychological tests assessing executive functioning and memory were performed, and scores were transformed into z scores. The authors used linear regression analyses, adjusted for age, sex, education, intelligence, and vascular risk factors, to investigate the association of white matter lesions and number of lacunar infarcts with cognitive performance. A 1-standard-deviation higher volume of white matter lesions (b ¼ 0.12, 95% confidence interval: 0.20, 0.04) and the presence of 2 (b ¼ 0.48, 95% confidence interval: 0.87, 0.09) were associated with worse executive functioning. These associations remained after adjusting for brain atrophy. Both were not associated with worse memory. Results suggest that subcortical ischemic vascular lesions are associated with decreased executive functioning, but not with memory functioning, independent of brain atrophy. brain infarction; cognition; leukoaraiosis; magnetic resonance imaging Abbreviations: CI, confidence interval; LI, ; MRI, magnetic resonance imaging; SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance; WML, white matter lesions. Vascular cognitive impairment is used to describe a range of cognitive disorders related to vascular disease (1). A major subtype of vascular cognitive impairment is subcortical ischemic vascular disease, which is characterized by white matter lesions (WML) and (LI) on magnetic resonance imaging (MRI) and is associated with a specific cognitive profile involving impairments in attentional and executive functioning (2). Although some aspects of new learning and memory are better preserved in patients with subcortical vascular disease than in patients with Alzheimer s disease, they are nonetheless not typically entirely preserved or normal (3). Neurodegenerative processes, characterized on MRI as widening of the sulci, narrowing of the gyri, and enlargement of the ventricles, are also involved in the pathogenesis of cognitive deterioration. In nondemented persons, the extent and rate of progression of global, cortical, and subcortical brain atrophy are associated with cognitive deterioration and conversion to Alzheimer s disease (4, 5). Cognitive impairment due to neurodegeneration and cognitive impairment due to vascular disease were traditionally regarded as separate clinical and pathophysiologic Correspondence to Dr. Mirjam I. Geerlings, University Medical Center Utrecht, Julius Center for Health Sciences and Primary Care (Str.6.131), P.O. Box 85500, 3508 GA Utrecht, the Netherlands (e-mail: m.geerlings@umcutrecht.nl). 1

2 Geerlings et al. entities. However, evidence is increasing that there is an overlap in the pathways by which vascular pathology and neurodegenerative processes contribute to cognitive decline (6, 7). Several studies observed associations between WML and progression of brain atrophy (8 10), suggesting that cerebral small-vessel disease may lead to cognitive impairment through brain atrophy. The objective of our study was to investigate whether WML and LI are associated with executive functioning and memory performance in patients with atherosclerotic disease. Furthermore, we investigated whether global, cortical, or subcortical brain atrophy mediates these associations. MATERIALS AND METHODS We used cross-sectional data from the Second Manifestations of ARTerial disease-magnetic Resonance (SMART-MR) study, a prospective cohort study within the SMART study, with the objective to investigate brain changes on MRI in patients with atherosclerotic disease. The rationale and design of the SMART study, including inclusion criteria, have been described in detail elsewhere (11). Between May 2001 and December 2005, 1,309 patients newly referred to the University Medical Center Utrecht with manifest coronary artery disease, cerebrovascular disease, peripheral arterial disease, or an abdominal aortic aneurysm, and without magnetic resonance contraindications, were included. Our study capacity was limited, so not all eligible patients were included. Detailed information about otherwise-eligible patients is therefore not available. During a 1-day visit to our medical center, an MRI of the brain was performed, in addition to a physical examination, ultrasonography of the carotid arteries, and blood and urine sampling. Risk factors, medical history, and functioning were assessed with questionnaires that the patients completed before their visit to the medical center (7, 12). Neuropsychological testing was introduced in the SMART-MR study starting in January 2003 and was performed on the same day as magnetic resonance and other investigations. The SMART study and SMART-MR study were approved by the ethics committee of our institution, and written informed consent was obtained from all participants. MRI protocol The magnetic resonance investigations were performed on a 1.5-T whole-body system (Gyroscan ACS-NT; Philips Medical Systems, Best, the Netherlands). The protocol consisted of a transversal T1-weighted gradient-echo sequence (repetition time/echo time: 235 ms/2 ms; flip angle, 80 ), a transversal T2-weighted turbo spin-echo sequence (repetition time/echo time: 2,200 ms/11 ms and 2,200 ms/100 ms; turbo factor 12), a transversal T2-weighted fluid attenuating inverse recovery sequence (repetition time/echo time/inversion time: 6,000 ms/100 ms/2,000 ms), and a transversal inversion recovery sequence (repetition time/echo time/ inversion time: 2,900 ms/22 ms/410 ms) (field of view 230 mm 3 230 mm; matrix size, 180 3 256; slice thickness, 4.0 mm; slice gap, 0.0 mm; 38 slices). Brain segmentation We used the T1-weighted gradient-echo, inversion recovery sequence and fluid attenuating inverse recovery sequence for brain segmentation. The probabilistic segmentation technique was performed by k-nearest neighbor classification and has been described elsewhere (13). The result of the classification method is a probability value for each voxel that quantifies the amount of a specific tissue type contained in that voxel. The segmentation program distinguishes cortical gray matter, white matter, sulcal and ventricular cerebrospinal fluid, and lesions. Subcortical gray matter was not segmented separately but was included in the white matter volume. The results of the segmentation analysis were visually checked for the presence of infarcts and adapted if necessary to make a distinction between WML volumes and infarct volumes. Total brain volume, including the cerebrum, brain stem, and cerebellum, was calculated by summing the volumes of gray and white matter and, if present, the volumes of WML and infarcts. Total intracranial volume was calculated by summing the total brain volume and the volumes of the sulcal and ventricular cerebrospinal fluid. Brain atrophy Brain parenchymal fraction, an indicator for global brain atrophy, represents the percentage of the intracranial volume occupied by brain tissue. Ventricular enlargement, an indicator for subcortical brain atrophy, was assessed with the ventricular fraction and was calculated as the percentage ventricular volume of the total intracranial volume. Cortical atrophy was assessed with the cortical gray matter fraction and was calculated as the percentage cortical gray matter volume of the total intracranial volume. lesions The volumes of WML obtained with the segmentation program were summed to obtain the total volume of WML. We did not distinguish between deep and periventricular WML because it has been shown that deep, periventricular, and total WML are highly correlated with each other, and it has been suggested that categorical distinctions between periventricular and deep WML are arbitrary (14). We normalized WML volumes for intracranial volume to correct for differences in head size (15) by dividing total WML volume by intracranial volume and multiplying this value by the average intracranial volume of the study population. Infarcts The whole brain, including cortex, brain stem, and cerebellum, was visually searched for infarcts by 2 investigators and a neuroradiologist. Discrepancies in rating were reevaluated in a consensus meeting. All raters were blinded

Cerebral Small Vessel Disease and Cognition 3 regarding the history and diagnosis of the patient. Infarcts were defined as focal hyperintensities on T2-weighted images at least 3 mm in diameter. Infarcts located in the white matter also had to be hypointense on T1-weighted and fluid attenuating inverse recovery images to distinguish them from WML. Dilated perivascular spaces were distinguished from infarcts on the basis of their location (along perforating or medullary arteries, often symmetric bilaterally, usually in the lower third of the basal ganglia or in the centrum semiovale), form (round/oval), and absence of gliosis (16). We defined LI as infarcts of 3 15 mm and located in the subcortical white matter, thalamus, or basal ganglia. Neuropsychological assessment Cognitive performance was assessed with a set of standard neuropsychological tests, sensitive to mild impairments. Verbal memory was assessed with 5 consecutive trials of the 15-word learning test (a modification of the Rey Auditory Verbal Learning test (17)). Immediate recall (range, 0 15) and delayed recall (range, 0 15) were assessed, and a retention score was calculated by dividing the number of words recalled after 25 minutes by the maximum number of words recalled during the immediate recall. A composite score for memory performance was calculated by averaging the z scores (individual test score minus mean test score divided by the standard deviation of that score) for the mean score of the 5 trials of the immediate recall, the z score for the delayed recall, and the z score for the retention score. Executive functioning was assessed with 3 tests. First, the Visual Elevator test (subtest of the Test of Everyday Attention) was used, a timed test of 10 trials that measures mental flexibility (18). The timing score of the Visual Elevator test is equivalent to time per switch for correct items (seconds per switch). Second, the Brixton Spatial Anticipation test was used to assess the capacity to discover logical rules (19). The total number of errors was scored. Third, the letter Verbal Fluency test (letter N) was used to assess mental flexibility, shifting of attention, and employment of strategies (20). A composite score for executive functioning was estimated by averaging the z scores for the Visual Elevator test, the Brixton Spatial Anticipation test, and the Verbal Fluency test. Premorbid intellectual functioning was assessed by using the Dutch Adult Reading Test, in which patients read aloud a list of words with irregular pronunciation (21). Educational level was divided into 7 categories, graded from primary school to academic degree, according to the Dutch educational system. Vascular risk factors During the patient s visit to the medical center, an overnight fasting venous blood sample was taken to determine glucose and lipid levels. Height and weight were measured without shoes and heavy clothing, and body mass index was calculated (weight (kg)/height (m) 2 ). Systolic and diastolic blood pressures (mm Hg) were measured twice with a sphygmomanometer, and the average was obtained. Hypertension was defined as mean systolic blood pressure 140 mm Hg Table 1. Patient Characteristics (n ¼ 522), the SMART-MR Study, the Netherlands, 2001 2005 or mean diastolic blood pressure 95 mm Hg or use of antihypertensive drugs. Diabetes mellitus was defined as a glucose level of 7.0 mmol/l or use of oral antidiabetic drugs or insulin. Hyperlipidemia was defined as total cholesterol >5.0 mmol/l, low density lipoprotein cholesterol >3.2 mmol/l, or use of lipid-lowering drugs. Smoking habits and alcohol intake were assessed by using questionnaires. Pack-years of smoking were calculated as the average number of cigarettes smoked per day divided by 20 and then multiplied by the number of years of cigarette smoking. Alcohol intake was divided into 3 categories: never, former, and current. Patients who had quit drinking during the past year were assigned to the category current alcohol intake. Ultrasonography was performed to measure the intima-media thickness (millimeters) in the left and right common carotid arteries, represented by the mean value of 6 measurements. Intima-media thickness is a marker for the extent of subclinical atherosclerosis. Study sample Characteristic Value Age, years, mean (SD) 57 (10) Male gender, % 76 Educational level, mean (SD) a 3.6 (1.8) Dutch Adult Reading Test score, mean (SD) 79 (17) Hypertension, % 54 Diabetes mellitus, % 19 Hyperlipidemia, % 79 Body mass index, kg/m 2 (SD) 27 (4) Smoking, pack-years, median 18 (6, 32) (25th percentile, 75th percentile) Alcohol consumption, % Never 17 Former 9 Current 74 Intima-media thickness, mm, mean (SD) 0.9 (0.2) Normalized white matter lesions 1.5 (0.9, 2.9) volume, ml, median (25th, 75th percentile) Any lacunar infarct, % 10 One lacunar infarct 6 Two or more 4 Brain parenchymal fraction, mean (SD) 79 (3) Ventricular fraction, mean (SD) 2.0 (0.9) Cortical gray matter fraction, mean (SD) 37 (3) Abbreviations: SD, standard deviation; SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance. a Educational level was divided into 7 categories, graded from primary school to academic degree, according to the Dutch educational system. A total of 1,309 patients were investigated in the SMART- MR study. Since neuropsychological testing was not

4 Geerlings et al. introduced until 2003, data on cognitive performance were present for 831 patients. Of these 831 patients, segmentation data were missing for 226 (missing magnetic resonance sequences needed for the segmentation analyses because of a temporary change in magnetic resonance protocol (n ¼ 179), irretrievable magnetic resonance data (n ¼ 11), missing fluid attenuating inverse recovery sequence (n ¼ 8), or motion or other artifacts (n ¼ 28)). Patients without available segmentation data were more often male (82% vs. 76%) and more often had diabetes (25% vs. 19%). Age and other vascular risk factors were comparable between patients with and without segmentation results. We excluded from our study sample 83 patients with cortical, large subcortical, or infratentorial infarcts on MRI because we were interested in the relation between small-vessel disease pathology on MRI and cognition. Thus, the analyses included 522 patients. Statistical analysis We used multiple linear regression analyses to investigate the associations among WML, LI, and measures of cognitive function. Composite z scores for executive functioning and memory were used as outcome variables. First, we assessed the association among WML, LI, and cognitive performance. For this purpose, normalized WML volume was entered in a first model, adjusted for age, sex, education, and Dutch Adult Reading Test score. Because WML and LI are strongly associated with each other, we included presence of LI as a covariate in a second model to investigate whether the relation between WML and cognitive performance could be explained by the concomitant presence of LI. The association between LI and cognition was investigated similarly. We categorized patients into 3 groups: no LI (n ¼ 472), 1 LI (n ¼ 30), and 2 or more LI (n ¼ 20). In the first model, we included 1 LI and 2 or more LI as dummy variables; in the second model, WML volume was entered as a covariate. The regression coefficients of 1 LI and 2 or more LI represent the differences in cognitive performance compared with patients without LI. We repeated the analyses using the individual cognitive test scores as the outcome. The raw scores of the tests were made comparable by transforming them into standardized z scores. In addition, since our study population was relatively young and the extent of vascular brain lesions may reflect relatively early changes, we also investigated the association of WML and LI with the composite z scores for executive functioning and memory in a subgroup of patients with more severe disease, defined as those in the top quartile of WML volume. Finally, we investigated to what extent the associations of WML and LI with cognition were mediated by global, subcortical, or cortical brain atrophy by adding these variables to the model. We repeated all analyses with additional adjustments for hypertension, diabetes mellitus, hyperlipidemia, body mass index, pack-years of smoking, alcohol intake, and intimamedia thickness to investigate whether the associations of WML and LI with cognitive performance were independent of vascular risk factors and extent of subclinical atherosclerosis. In all analyses, we report 95% confidence intervals. Table 2. Raw Scores for Neuropsychological Tests a (n ¼ 522), the SMART-MR Study, the Netherlands, 2001 2005 SPSS version 14.0 software (SPSS, Inc., Chicago, Illinois) was used to analyze our data. RESULTS Test Score Memory 15-Word learning task Immediate recall, mean 7.6 (2.0) (no. of words) Delayed recall (no. of words) 7.4 (2.9) Retention score (%) 72.1 (19.3) Executive functioning Visual Elevator test, timing score 5.1 (2.2) (seconds per switch) Brixton Spatial Anticipation test 18.2 (6.1) (total no. of errors) Verbal Fluency test (words with 10.4 (4.2) the letter N, during 1 minute) Abbreviation: SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance. a All scores are expressed as mean (standard deviation). A higher score on the 15-word learning task and the Verbal Fluency test indicates a better performance. A higher score on the Visual Elevator test and Brixton Spatial Anticipation test indicates a worse performance. The mean age of the study sample was 57 years (standard deviation, 10), and the majority (76%) were men. LI on MRI were present in 50 patients (10%); 30 had 1 LI and 20 had 2 or more LI (Table 1). Table 2 shows the raw scores for the neuropsychological tests. Table 3 shows the associations of normalized WML volume with executive functioning and memory. Linear Table 3. Association of White Matter Lesion Volume With z Scores for Executive Functioning and Memory (n ¼ 522), the SMART-MR Study, the Netherlands, 2001 2005 a Executive Functioning Memory b 95% CI b 95% CI 0.14 0.22, 0.06 0.06 0.14, 0.03 0.12 0.20, 0.04 0.05 0.13, 0.04 Abbreviations: CI, confidence interval; SD, standard deviation; SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance. a b represents the difference (95% CI) in z scores for executive functioning and memory per standard deviation (3.4 ml) higher white matter lesion volume. A lower z score denotes a worse performance. b Adjusted for age, sex, educational level, and Dutch Adult Reading Test score. c Model I additionally adjusted for.

Cerebral Small Vessel Disease and Cognition 5 Table 4. Association of Number of Lacunar Infarcts With z Scores for Executive Functioning and Memory (n ¼ 522), the SMART-MR Study, the Netherlands, 2001 2005 a Executive Functioning regression analysis adjusted for age, sex, educational level, and Dutch Adult Reading Test score showed that an increase of 1 standard deviation (3.4 ml) in WML volume was associated with worse executive functioning performance (b ¼ 0.14, 95% confidence interval (CI): 0.22, 0.06). Additional adjustment for presence of LI attenuated the association with executive functioning slightly, but the relation remained statistically significant. WML were not significantly associated with memory. Table 4 shows the associations of LI with executive functioning and memory. Presence of a single LI was not associated with worse executive functioning performance, but 2 or more LI were associated with a lower score for executive functioning (b ¼ 0.61, 95% CI: 0.99, 0.23) after adjustment for age, sex, educational level, and Dutch Adult Reading Test score. This association was also attenuated after adjustment for WML. LI were not associated with a lower memory score. When we restricted our analyses to the patients with severe WML (highest quartile, n ¼ 132; mean age, 64 years; 9% with 1 LI and 11% with 2 or more LI), WML and LI were also not associated with memory (data not shown). The association of WML (b ¼ 0.16, 95% CI: 0.29, 0.03) and 2 or more LI (b ¼ 0.80, 95% CI: 1.36, 0.24) with executive functioning remained. Tables 5 and 6 show the associations of WML and LI with z scores for all individual neuropsychological tests. WML volume was associated with lower scores on all executive functioning tests in model I, and these associations remained (borderline) significant after adjustment for LI in model II (Table 5). WML were not associated with delayed recall or retention score but were borderline significantly Table 5. Association of White Matter Lesion Volume With z Scores for Individual Executive Functioning and Memory Tests (n ¼ 522), the SMART-MR Study, the Netherlands, 2001 2005 a Memory b 95% CI b 95% CI No Reference Reference One lacunar infarct 0.05 0.27, 0.37 0.16 0.18, 0.50 Two or more 0.61 0.99, 0.23 0.30 0.70, 0.11 No Reference Reference One lacunar infarct 0.09 0.23, 0.42 0.18 0.16, 0.52 Two or more 0.48 0.87, 0.09 0.25 0.66, 0.17 Abbreviations: CI, confidence interval; SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance. a b represents the difference (95% CI) in z scores for executive functioning and memory compared with patients without. A lower z score denotes a worse performance. b Adjusted for age, sex, educational level, and Dutch Adult Reading Test score. c Model I additionally adjusted for white matter lesion volume. Visual Elevator Test Individual Executive Functioning Tests Brixton Spatial Anticipation Test Verbal Fluency Test b 95% CI b 95% CI b 95% CI 0.09 0.18, 0.01 0.09 0.19, 0.00 0.10 0.19, 0.02 0.06 0.14, 0.01 0.09 0.18, 0.00 0.10 0.18, 0.01 Individual Memory Tests Immediate Recall Delayed Recall Retention Score b 95% CI b 95% CI b 95% CI 0.09 0.18, 0.01 0.02 0.11, 0.07 0.01 0.09, 0.11 0.07 0.16, 0.02 0.01 0.10, 0.08 0.01 0.09, 0.11 Abbreviations: CI, confidence interval; SD, standard deviation; SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance. a b represents the difference (95% CI) in z scores for individual executive functioning and memory tests per standard deviation (3.4 ml) higher white matter lesion volume. A lower z score denotes a worse performance. b Adjusted for age, sex, educational level, and Dutch Adult Reading Test score. c Model I additionally adjusted for.

6 Geerlings et al. Table 6. Association of Number of Lacunar Infarcts With z Scores for Individual Executive Functioning and Memory Tests (n ¼ 522), the SMART-MR Study, the Netherlands, 2001 2005 a Individual Executive Functioning Tests Visual Elevator Test Brixton Spatial Anticipation Test associated with poorer immediate recall. LI were associated with a lower score on the Visual Elevator test but were not associated with the other executive functioning tests, and they were also not significantly associated with any of the memory tests (Table 6). Next, we investigated whether the associations of WML and LI with executive functioning could be explained by global, cortical, or subcortical brain atrophy. Additional adjustment for brain parenchymal fraction or gray matter fraction did not change the association of WML with executive functioning (Table 7). However, after adjustment for ventricular fraction, the association between WML and executive functioning attenuated, but an increase in WML volume of 1 standard deviation remained significantly associated with a lower executive functioning score (b ¼ 0.09, 95% CI: 0.18, 0.01). The association between LI and executive functioning did not materially change after adjustment for measures of brain atrophy (Table 8). Finally, additional adjustment for vascular risk factors and intima-media thickness did not materially change the associations among WML, LI, and executive functioning (data not shown). DISCUSSION Verbal Fluency Test b 95% CI b 95% CI b 95% CI No Reference Reference Reference One lacunar infarct 0.06 0.29, 0.40 0.07 0.31, 0.46 0.02 0.37, 0.34 Two or more 0.88 1.29, 0.46 0.29 0.74, 0.17 0.29 0.71, 0.13 No Reference Reference Reference One lacunar infarct 0.08 0.26, 0.42 0.11 0.28, 0.49 0.02 0.34, 0.37 Two or more 0.81 1.24, 0.39 0.19 0.66, 0.28 0.19 0.62, 0.24 Individual Memory Tests Immediate Recall Delayed Recall Retention Score b 95% CI b 95% CI b 95% CI No Reference Reference Reference One lacunar infarct 0.01 0.31, 0.34 0.16 0.19, 0.52 0.29 0.10, 0.67 Two or more 0.36 0,75, 0.02 0.30 0.71, 0.12 0.18 0.64, 0.27 No Reference Reference Reference One lacunar infarct 0.04 0.28, 0.37 0.17 0.19, 0.52 0.28 0.11, 0.67 Two or more 0.33 0.72, 0.07 0.28 0.71, 0.15 0.19 0.66, 0.27 Abbreviations: CI, confidence interval; SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance. a b represents the difference (95% CI) in z scores for individual executive functioning and memory tests compared with patients without. A lower z score denotes a worse performance. b Adjusted for age, sex, educational level, and Dutch Adult Reading Test score. c Model I additionally adjusted for white matter lesion volume. We found that larger WML volume and the presence of multiple LI were associated with poorer executive functioning in a relatively young population of patients with atherosclerotic disease. WML and LI were not associated with poorer overall memory performance. The association between WML and decreased executive performance attenuated after adjustment for ventricular enlargement but remained statistically significant. Ventricular enlargement is most likely the result of ex vacuo enlargement due to subcortical brain atrophy. Subcortical brain atrophy may be an intermediate factor in the relation

Cerebral Small Vessel Disease and Cognition 7 Table 7. Association of White Matter Lesion Volume With z Scores for Executive Functioning, Adjusted for Measures of Brain Atrophy (n ¼ 522), the SMART-MR Study, the Netherlands, 2001 2005 a Executive Functioning b 95% CI 0.12 0.20, 0.04 0.12 0.20, 0.04 Model III d 0.09 0.18, 0.01 Model IV e 0.12 0.21, 0.04 Abbreviations: CI, confidence interval; SD, standard deviation; SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance. a b represents the difference (95% CI) in z scores for executive functioning per standard deviation (3.4 ml) higher white matter lesion volume. A lower z score denotes a worse performance. b Adjusted for age, sex, educational level, Dutch Adult Reading Test score, and. c Model I additionally adjusted for brain parenchymal fraction. d Model I additionally adjusted for ventricular fraction. e Model I additionally adjusted for cortical gray matter fraction. between WML and cognitive dysfunction because ventricular enlargement is associated with executive impairment (22) and associations between WML and ventricular enlargement have been found (23), which suggests that ischemic damage due to small-vessel pathology may be involved in the pathogenesis of subcortical brain atrophy (24). Because of the cross-sectional design of our study, we cannot exclude the possibility that subcortical brain atrophy also is a confounder in the relation between WML and cognitive functioning and thus an independent risk factor for cognitive impairment. Nevertheless, the association between WML and executive functioning was still present after adjustment for ventricular enlargement. This finding suggests that WML have an independent effect on cognitive performance. We also observed that WML were related to worse executive functioning independent of the presence of LI. Associations between WML and impairment in executive functioning have been observed in other studies (3, 25, 26). Executive functioning is primarily controlled by the prefrontal cortex. The frontal lobes are extensively connected to other parts of the brain through axons that lie in the subcortical matter (27, 28). It is thought that WML interrupt these prefrontal-subcortical connections, which leads to impaired prefrontal lobe functioning characterized by executive dysfunction (2). A larger WML volume was associated with lower scores on all individual executive functioning tests, suggesting that WML impair multiple executive functions, including mental flexibility, shifting of attention, and strategy formation. Since 2 of the tests were timed, our results may to some extent reflect cognitive slowing or information processing speed. Associations between WML and cognitive slowing have been Table 8. Association of Number of Lacunar Infarcts With z Scores for Executive Functioning, Adjusted for Measures of Brain Atrophy (n ¼ 522), the SMART-MR Study, the Netherlands, 2001 2005 a Executive Functioning b 95% CI No Reference One lacunar infarct 0.09 0.23, 0.42 Two or more 0.48 0.87, 0.09 No Reference One lacunar infarct 0.09 0.23, 0.42 Two or more 0.44 0.83, 0.05 Model III d No Reference One lacunar infarct 0.13 0.20, 0.46 Two or more 0.47 0.86, 0.08 Model IV e No Reference One lacunar infarct 0.13 0.21, 0.47 Two or more 0.44 0.85, 0.04 Abbreviations: CI, confidence interval; SMART-MR, Second Manifestations of ARTerial disease-magnetic Resonance. a b represents the difference (95% CI) in z scores for executive functioning compared with patients without. A lower z score denotes a worse performance. b Adjusted for age, sex, educational level, Dutch Adult Reading Test score, and white matter lesion volume. c Model I additionally adjusted for brain parenchymal fraction. d Model I additionally adjusted for ventricular fraction. e Model I additionally adjusted for cortical gray matter fraction. described previously and may be the result of inefficient neural activity following damage to white matter tracts connecting distant cortical areas (29). However, because WML were also associated with lower scores on the nontimed Brixton Spatial Anticipation test, mental slowing alone cannot explain our findings. The presence of multiple LI was also associated with poorer executive functioning, of which the strongest association was with the Visual Elevator test. The association between presence of multiple LI and worse executive functioning was independent of brain atrophy. Because strategically located infarcts are strongly associated with cognitive impairment, it is possible that they already impair cognitive functioning in the absence of detectable brain atrophy. WML and LI were not associated with overall memory performance. Our study population was relatively young, and it is possible that the executive functioning tests were more sensitive than the memory tests to detect early changes in this vascular population. However, the crude scores for immediate and delayed recall of our study population are comparable to the crude scores reported in another study (30) that included a population-based sample of men whose mean age and age distribution were similar to our

8 Geerlings et al. population. In this respect, our population had, on average, normal age-adjusted memory performance. In addition, when we restricted our analyses to patients with severe WML, we still did not find an association between WML or LI and memory. Furthermore, memory performance as assessed with the 15-word learning test is mainly a function of the hippocampus and to a lesser extent a distinctive feature of subcortical ischemic vascular disease (1, 3). When we analyzed the associations of WML and LI with the individual memory tests, we observed that WML were borderline significantly associated with immediate recall but not with delayed recall or retention. Delayed recall and retention rely to a greater extent on hippocampal functions, while immediate recall, of which attention is an important aspect, relies mainly on involvement of the prefrontal cortex and may be more vulnerable to subcortical vascular disease. Although an association between WML and worse memory has been found in other studies (31 33), this association disappeared after adjustment for medial temporal lobe atrophy (32, 33), suggesting that the memory impairment was caused by hippocampal atrophy. Strengths of our study include the large number of patients investigated and the volumetric assessment of WML and brain atrophy measures, which made it possible to obtain precise estimates of the relation between brain changes on MRI and cognitive functioning and resulted in a large power to detect associations. Furthermore, we could perform subgroup analyses using a sample of older patients with more severe vascular brain changes. Second, since WML and LI were assessed simultaneously in our study, it was possible to investigate their independent impact on executive functioning and memory. Furthermore, the segmentation of different brain tissue types and cerebrospinal fluid spaces enabled us not only to investigate the role of global brain atrophy but also to differentiate between cortical and subcortical brain atrophy. In summary, WML and multiple LI were associated with decreased executive functioning in a relatively young population of patients with atherosclerotic disease, independent of brain atrophy. WML and LI were not associated with overall memory performance. These findings highlight the importance of acknowledging that WML and LI should be considered risk factors for vascular cognitive impairment independent of neurodegenerative processes. ACKNOWLEDGMENTS Author affiliations: Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, the Netherlands (Mirjam I. Geerlings, Auke P. A. Appelman, Yolanda van der Graaf); Department of Radiology, University Medical Center Utrecht, Utrecht, the Netherlands (Auke P. A. Appelman, Willem P. T. M. Mali); and Image Sciences Institute, University Medical Center Utrecht, Utrecht, the Netherlands (Koen L. Vincken). This study was supported by a program grant from the Netherlands Organization for Scientific Research-Medical Sciences (NWO-MW: project no. 904-65-095) and a grant from the Netherlands Organization for Scientific Research (NWO: project no. 917-66-311). The funding sources had no involvement in the writing of this article or in the decision to submit it for publication. The authors gratefully acknowledge the members of the SMART Study Group at the University Medical Center Utrecht: Dr. A. Algra, Julius Center for Health Sciences and Primary Care and Rudolf Magnus Institute for Neurosciences, Department of Neurology; Dr. P. A. Doevendans, Department of Cardiology; Dr. Y. van der Graaf, Dr. D. E. Grobbee, and Dr. G. E. H. M. Rutten, Julius Center for Health Sciences and Primary Care; Dr. L. J. Kappelle, Department of Neurology; Dr. W. P. Th. M. Mali, Department of Radiology; Dr. F. L. Moll, Department of Vascular Surgery; and Dr. F. L. J. Visseren, Department of Vascular Medicine. Conflict of interest: none declared. REFERENCES 1. O Brien JT, Erkinjuntti T, Reisberg B, et al. Vascular cognitive impairment. Lancet Neurol. 2003;2(2):89 98. 2. Román GC, Erkinjuntti T, Wallin A, et al. Subcortical ischaemic vascular dementia. Lancet Neurol. 2002;1(7): 426 436. 3. Chui HC. Subcortical ischemic vascular dementia. Neurol Clin. 2007;25(3):717 740, vi. 4. Mungas D, Reed BR, Jagust WJ, et al. Volumetric MRI predicts rate of cognitive decline related to AD and cerebrovascular disease. Neurology. 2002;59(6):867 873. 5. Jack CR Jr, Shiung MM, Weigand SD, et al. Brain atrophy rates predict subsequent clinical conversion in normal elderly and amnestic MCI. Neurology. 2005;65(8):1227 1231. 6. Iadecola C, Gorelick PB. Converging pathogenic mechanisms in vascular and neurodegenerative dementia. Stroke. 2003;34(2): 335 337. 7. Muller M, Appelman AP, van der Graaf Y, et al. Brain atrophy and cognition: interaction with cerebrovascular pathology? Neurobiol Aging. Advance Access: June 9, 2009. (DOI: 10.1016/j.neurobiolaging.2009.05.005). 8. Schmidt R, Ropele S, Enzinger C, et al. lesion progression, brain atrophy, and cognitive decline: the Austrian stroke prevention study. Ann Neurol. 2005;58(4):610 616. 9. Walters RJ, Fox NC, Schott JM, et al. Transient ischaemic attacks are associated with increased rates of global cerebral atrophy. J Neurol Neurosurg Psychiatry. 2003;74(2): 213 216. 10. Enzinger C, Fazekas F, Matthews PM, et al. Risk factors for progression of brain atrophy in aging: six-year follow-up of normal subjects. Neurology. 2005;64(10):1704 1711. 11. Simons PC, Algra A, van de Laak MF, et al. Second Manifestations of ARTerial disease (SMART) study: rationale and design. Eur J Epidemiol. 1999;15(9):773 781. 12. Appelman AP, van der Graaf Y, Vincken KL, et al. Total cerebral blood flow, white matter lesions and brain atrophy: the SMART-MR study. J Cereb Blood Flow Metab. 2008;28(3): 633 639. 13. Anbeek P, Vincken KL, van Bochove GS, et al. Probabilistic segmentation of brain tissue in MR imaging. Neuroimage. 2005;27(4):795 804. 14. DeCarli C, Fletcher E, Ramey V, et al. Anatomical mapping of white matter hyperintensities (WMH): exploring the

Cerebral Small Vessel Disease and Cognition 9 relationships between periventricular WMH, deep WMH, and total WMH burden. Stroke. 2005;36(1):50 55. 15. Hachinski V, Iadecola C, Petersen RC, et al. National Institute of Neurological Disorders and Stroke-Canadian Stroke Network vascular cognitive impairment harmonization standards. Stroke. 2006;37(9):2220 2241. 16. Giele JL, Witkamp TD, Mali WP, et al. Silent brain infarcts in patients with manifest vascular disease. Stroke. 2004;35(3): 742 746. 17. Lezak MD. Neuropsychological Assessment. 3rd ed. Oxford, United Kingdom: Oxford University Press; 1995. 18. van Raamt AF, Kalmijn S, Mali WP, et al. Homocysteine level and cognitive function in patients with arterial disease: the second manifestations of ARTerial Disease Study. J Am Geriatr Soc. 2006;54(4):575 579. 19. Burgess PW, Shallice T. Bizarre responses, rule detection and frontal lobe lesions. Cortex. 1996;32(2):241 259. 20. Wilkins AJ, Shallice T, McCarthy R. Frontal lesions and sustained attention. Neuropsychologia. 1987;25(2):359 365. 21. Schmand B, Geerlings MI, Jonker C, et al. Reading ability as an estimator of premorbid intelligence: does it remain stable in emergent dementia? J Clin Exp Neuropsychol. 1998;20(1):42 51. 22. Breteler MM, van Amerongen NM, van Swieten JC, et al. Cognitive correlates of ventricular enlargement and cerebral white matter lesions on magnetic resonance imaging. The Rotterdam Study. Stroke. 1994;25(6):1109 1115. 23. Longstreth WT Jr, Arnold AM, Manolio TA, et al. Clinical correlates of ventricular and sulcal size on cranial magnetic resonance imaging of 3,301 elderly people. The Cardiovascular Health Study. Collaborative Research Group. Neuroepidemiology. 2000;19(1):30 42. 24. Pantoni L, Garcia JH. Pathogenesis of leukoaraiosis: a review. Stroke. 1997;28(3):652 659. 25. Pantoni L, Poggesi A, Inzitari D. The relation between whitematter lesions and cognition. Curr Opin Neurol. 2007;20(4): 390 397. 26. Carey CL, Kramer JH, Josephson SA, et al. Subcortical lacunes are associated with executive dysfunction in cognitively normal elderly. Stroke. 2008;39(2):397 402. 27. Cummings JL. Anatomic and behavioral aspects of frontalsubcortical circuits. Ann N Y Acad Sci. 1995;769:1 13. 28. Tekin S, Cummings JL. Frontal-subcortical neuronal circuits and clinical neuropsychiatry: an update. J Psychosom Res. 2002;53(2):647 654. 29. van den Heuvel DM, ten Dam VH, de Craen AJ, et al. Increase in periventricular white matter hyperintensities parallels decline in mental processing speed in a non-demented elderly population. J Neurol Neurosurg Psychiatry. 2006;77(2):149 153. 30. Muller M, Grobbee DE, Aleman A, et al. Cardiovascular disease and cognitive performance in middle-aged and elderly men. Atherosclerosis. 2007;190(1):143 149. 31. de Groot JC, de Leeuw FE, Oudkerk M, et al. Cerebral white matter lesions and cognitive function: the Rotterdam Scan Study. Ann Neurol. 2000;47(2):145 151. 32. O Brien JT, Desmond P, Ames D, et al. Magnetic resonance imaging correlates of memory impairment in the healthy elderly: association with medial temporal lobe atrophy but not white matter lesions. Int J Geriatr Psychiatry. 1997;12(3): 369 374. 33. Jokinen H, Kalska H, Mäntylä R, et al. Cognitive profile of subcortical ischaemic vascular disease. J Neurol Neurosurg Psychiatry. 2006;77(1):28 33.