Cortical Hubs and Subcortical Cholinergic Pathways as Neural Substrates of Poststroke Dementia

Transcription

1 Cortical Hubs and Subcortical Cholinergic Pathways as Neural Substrates of Poststroke Dementia Jae-Sung Lim, MD, MSc; Nayoung Kim, MD, PhD; Min Uk Jang, MD; Moon-Ku Han, MD, PhD; SangYun Kim, MD, PhD; Min Jae Baek, MSc; Myung Suk Jang, BS; Byeolnim Ban, BS; Yeonwook Kang, MD, PhD; Dong-Eog Kim, MD, PhD; Ji Sung Lee, PhD; Juneyoung Lee, PhD; Byung-Chul Lee, MD, PhD; Kyung-Ho Yu, MD, PhD; Sandra E. Black, MD, PhD; Hee-Joon Bae, MD, PhD Background and Purpose A role of neural networks in the development of poststroke dementia has not been clearly established. We hypothesized that stroke-mediated disruption of subcortical cholinergic pathway or large-scale neural networks contributes to poststroke dementia. Methods A matched case control study was conducted in a predetermined cohort with acute ischemic stroke. Cases were defined as newly developed dementia diagnosed >3 months after stroke using the Korean Vascular Cognitive Impairment Harmonization Standards. Each case was matched to 2 controls for age, education, and initial stroke severity. The Cholinergic Pathways HyperIntensities Scale was applied with some modifications to characterize disruption of cholinergic pathways by acute stroke lesions. Involvement of major cortical hub locations of the default mode network, central executive network, and salience network was also investigated. Results After matching, 38 cases and 66 matched controls were included. Cholinergic Pathways HyperIntensities Scale scores were significantly higher in cases than in controls (2.2±2.9 versus 0.9±1.4). Acute ischemic lesions affecting the default mode and central executive networks were more frequently observed in cases compared with controls (36.8% versus 7.6% and 26.3% versus 6.1%, respectively). These findings remained significant in the multiple logistic regression models adjusted for various sets of potential confounders. Lesion location analysis revealed that cases were more likely to have acute lesions in the left corona radiata, hippocampal formation, and posterior parietal cortex. Conclusions Disruption of cholinergic pathways and major hubs of large-scale neural networks might contribute to newly developed dementia after acute ischemic stroke. (Stroke. 2014;45: ) Key Words: cholinergic fibers dementia neural networks neural pathways stroke About 10% of patients with first-ever stroke and 30% of those with recurrent stroke eventually develop poststroke dementia (PSD). 1 Among the known risk factors of PSD, those related to lesion characteristics, such as volume, multiplicity, and location, are drawing increasing attention because of their strong associations with PSD development. 1 However, the characterization of lesion location in previous studies was empirically based on gross anatomic demarcations. Previous research of structural or functional neural networks underscores the importance of those networks for normal cognition. 2,3 Cholinergic pathway dysfunction is known to adversely affect short-term episodic memory, attention, and executive functions, 3,4 and cholinesterase inhibitors could stabilize or improve symptoms of vascular dementia 5 as well as Alzheimer disease (AD). 6 Resting-state functional MRI studies have suggested that the default mode network (DMN), central executive network (CEN), and salience network (SN) are important large-scale networks for proper processing of various stimuli to maintain normal cognitive function. 7 Thus, it can be predicted that disruption of the nodes and connections in these networks provides a neural basis for PSD. 1,8 We hypothesized that stroke-mediated disruption of subcortical cholinergic pathways or major cortical hubs of large-scale neural networks might contribute to the development of PSD. Received November 14, 2013; final revision received January 23, 2014; accepted February 4, From the Department of Neurology, Seoul National University Boramae Hospital, Seoul, Korea (J.-S.L.); Department of Neurology and Clinical Neuroscience Center, Seoul National University Bundang Hospital, Seongnam, Korea (N.K., M.U.J., M.-K.H., S.Y.K., M.J.B., M.S.J., B.B., H.-J.B.); Department of Psychology, Hallym University, Chuncheon, Korea (Y.K.); Department of Neurology, Dongguk University Ilsan Hospital, Ilsan, Korea (D.-E.K.); Biostatistical Consulting Unit, Soonchunhyang University Medical Center, Seoul, Korea (J.S.L.); Department of Biostatistics, Korea University College of Medicine, Seoul, Korea (J.L.); Department of Neurology, Hallym University Sacred Heart Hospital, Anyang, Korea (B.-C.L., K.-H.Y.); and Department of Medicine (Neurology), Brain Sciences Research Program, Heart and Stroke Foundation Centre for Stroke Recovery at Sunnybrook Research Institute, University of Toronto, Toronto, Ontario, Canada (S.E.B.). The online-only Data Supplement is available with this article at /-/DC1. Correspondence to Hee-Joon Bae, MD, PhD, Cerebrovascular Center, Seoul National University Bundang Hospital, Department of Neurology, Seoul National University College of Medicine, 82 Gumi-ro 173 beon-gil, Bundang-gu, Seongnam , Korea. braindoc@snu.ac.kr 2014 American Heart Association, Inc. Stroke is available at DOI: /STROKEAHA

2 1070 Stroke April 2014 Methods Study Design and Eligibility Criteria A matched case control design was applied to a predetermined stroke cohort admitted to a university hospital. Patients with ischemic stroke who were hospitalized within 1 week of onset between May 2007 and March 2011 and had acute ischemic lesions on diffusion-weighted imaging were identified from a prospective stroke registry database. Those who underwent neuropsychological tests (Korean Vascular Cognitive Impairment Harmonization Standards: Neuropsychology Protocol [K-VCIHS-NP]) 9 3 months after stroke onset were included. Patients who had prestroke cognitive impairment, a history of dementia or major depression based on medical records, or hearing difficulty, poor cooperation, or neurological deficits such as severe aphasia or motor weakness to a degree that would preclude neuropsychological testing, were excluded (Figure 1; Methods in the online-only Data Supplement). The premorbid cognitive impairment was defined as 3.6 in the Korean version of Informant Questionnaire on Cognitive Decline in the Elderly score, 10 which was performed as a part of VCIHS-NP. Subject Selection and Matching Among patients who met the above criteria, those who showed cognitive impairment in 2 domains of the K-VCIHS-NP including memory, and whose Korean Instrumental Activities of Daily Living score was 0.43, 11 were defined as PSD cases. 12 Cognitive impairment was defined as a score of less than the seventh percentile (approximately mean 1.5 SD) for age and education-corrected reference scores in each domain-specific test. Controls were selected from those who met all the same eligibility criteria as cases but did not show cognitive impairment in any domain of the K-VCIHS-NP and had a Korean Instrumental Activities of Daily Living score <0.43. Each PSD case was matched to 2 controls with the SAS 9.3 macro %match for age (±3 years), level of education (±3 years), and National Institutes of Health Stroke Scale (NIHSS) score on admission (±1). The local institutional review board approved the study protocol and allowed it to be conducted without patients consent because of its retrospective nature and minimal risk to participants. Neuropsychological Assessment The K-VCIHS-NP was adapted from the 60-minute neuropsychology protocol of the VCIHS proposed by the National Institute of Neurological Disorders and Stroke and the Canadian Stroke Network. 13 It assesses 4 cognitive domains (frontal executive, language, visuospatial, and memory) with 8 cognitive tests. All tests and scales were validated and standardized for the Korean language. Detailed information on methods and validation of the Korean versions is presented in the Methods in the online-only Data Supplement. Data Collection Information on demographics, vascular risk factors, stroke characteristics, and other clinical and neuroimaging-related factors was obtained from the stroke registry database 14 or by reviewing the electronic records of our institution and the Picture Archiving and Communication System (Infinite G3, Infinite Healthcare; Seoul, Republic of Korea). Detailed protocols and acquisition parameters for magnetic resonance images are described in the Methods in the online-only Data Supplement. Fluid-attenuated inversion recovery imaging sequence MRI was additionally performed between 5 and 7 days of hospitalization after initial diffusion-weighted imaging and fluid-attenuated inversion recovery imaging to confirm the final infarct size and location. Figure 1. Enrollment flow chart. ADL indicates activities of daily living; DWI, diffusion-weighted imaging; and VCIHS-NP, Vascular Cognitive Impairment Harmonization Standards: Neuropsychology Protocol.

3 Lim et al Neural Substrates of Poststroke Dementia 1071 Image Preprocessing and Cumulative Lesion Maps To generate cumulative lesion maps, we used the custom-written software Imaging software for Quantitative Neurovascular lesion Assessment (Image_QNA)15 for semiautomatic segmentation and transfer of stroke lesions onto a common brain template (ch2better). Acute stroke lesions captured using follow-up fluid-attenuated inversion recovery imaging sequences in reference to initial diffusion-weighted imaging were segmented and registered on the templates. Detailed information on the cumulative lesion map is presented in the Methods in the online-only Data Supplement. Assessment of Acute Ischemic Lesions We characterized acute ischemic lesions according to 6 measures: involvement of (1) cholinergic pathways and (2) hubs of the DMN, CEN, or SN; (3) cortical versus subcortical location; (4) lesion volume (cubic centimeter); (5) laterality; and (6) multiplicity. Assessment of Cholinergic Pathway Involvement Cholinergic pathway involvement was determined using the established Cholinergic Pathways HyperIntensities Scale (CHIPS)3 with some modification. CHIPS were rated in the 4 index magnetic resonance axial images, which were termed low external capsule, high external capsule, corona radiata, and centrum semiovale according to the major anatomic landmarks of each index magnetic resonance image (Figure I in the online-only Data Supplement and Figure 2). On those reference images, the cholinergic pathways were visualized based on the original immunohistochemistry study in reference to surrounding anatomic landmarks such as the external capsule, claustrum, and cingulate cortex (Methods in the online-only Data Supplement). We simplified CHIPS scoring by dichotomizing involvement of each predetermined region of the cholinergic pathway by the acute ischemic lesions instead of using severity scores of the original CHIPS. Each patient s lesion map was overlaid with templates for CHIPS with 50% transparency using Adobe Photoshop CS5 (Adobe Systems Inc; San Jose, CA) to identify the involvement of CHIPS by stroke lesions. We coded one if there were any stroke lesions in the prespecified cholinergic pathway of each index slice, and weighted numbers for each index image were given as the original CHIPS3 considering the cholinergic fiber density: weighted number 4 for the most densely packed fibers of low external capsule and 1 for the panningout fibers of centrum semiovale. Total scores were summed for both sides, resulting in scores ranging from 0 to 30 (Table I in the onlineonly Data Supplement). Assessment of Neural Network Involvement For the functional neural networks, we investigated the DMN, CEN, and SN.7 The involvement of each network was coded as 0 or 1: 0 for none of the corresponding hubs, 1 for any of these locations. The involvement of each hub was decided based on the lesion maps according to the corresponding Montreal Neurological Institute coordinates in the published studies, as well as anatomic landmarks such as characteristic cortical or subcortical structures (Table II in the online-only Data Supplement). We drew 10-mm diameter regions of interest for each hub region. Some regions, such as the hippocampal formation and insular region, could be identified easily. For the inferior parietal lobule or posterior parietal cortex, we identified each region using prespecified Montreal Neurological Institute coordinates from the literature and neighboring gyri and sulci in the common templates and superimposed them onto each patient s MRI (Methods in the online-only Data Supplement). For better visualization, cutaway images for the 3 representative network hubs in Figure 3 were generated by MRIcron software ( Assessment of Other Neuroimaging Variables Other measures for acute stroke lesions (cortical versus subcortical locations, lesion volume, laterality, and multiplicity) were also investigated. Characteristics of chronic lesions (white matter hyperintensities, chronic cortical infarcts, lacunar infarcts, microbleeds, medial temporal lobe atrophy, and cerebral atrophy index) were coded as covariates. Details are described in the Methods in the online-only Data Supplement. Inter-Rater Reliability Two independent raters (J.-S.L. and M.U.J.) who were blinded to case allocation coded all neuroimaging-related variables, and intrarater Figure 2. Cumulative lesion maps of cases and controls for Cholinergic Pathways HyperIntensities Scale. Axial images are common templates for T1 structural magnetic resonance embedded in Image_QNA software. Light magenta colored lines on the templates indicate medial and lateral cholinergic pathways. The red thin arrow points to the left corona radiata, and the white thick arrow indicates the left centrum semiovale. Colored bars and numbers next to the maps denote lesion frequencies. Specifically, 4/19 indicates that 19 subjects had stroke lesions in the corresponding templates, and 4 subjects had lesions in the areas with the corresponding color.

4 1072 Stroke April 2014 Figure 3. Cumulative lesion maps for cortical hub regions. The hub cortical regions of the default mode network (DMN), central executive network (CEN), and salience network (SN) are denoted as colored circles on the 3-dimensional brain images. We then constructed cumulative lesion maps on representative axial images for cases (middle column) and controls (right column). Colored bars next to the maps denote lesion frequency. The red thin arrows indicate the hippocampal formation, and the white thick arrow denotes the posterior parietal cortex. and inter-rater reliability were determined. κ and intraclass correlation coefficients were between 0.80 and 1.00 for each variable. Statistical Analysis Before matching, Student t tests were used for continuous variables, and χ 2 tests and Fisher exact tests were used for categorical variables. After matching, conditional logistic and exact conditional logistic regression analyses were performed. Variables for adjustments in regression models were chosen based on their P values (<0.20) in comparisons of the cases and matched controls and the current evidence about their associations with PSD in the literature. Because of the limitation of a small sample size, variables for adjustments were grouped according to their clinical meanings, and 5 models were developed with each group of covariates (Methods in the online-only Data Supplement). Statistical analyses were performed with SAS version 9.3 (SAS Institute Inc; Cary, NC), and a 2-sided P value <0.05 was considered as the minimum level of statistical significance.

5 Lim et al Neural Substrates of Poststroke Dementia 1073 Results Among the patients with acute ischemic stroke who were hospitalized during the study period, 428 patients underwent VCIHS-NP at 3 months or later from symptom onset. Patients without VCIHS-NP were more likely to have higher baseline NIHSS score, hyperlipidemia, and history of stroke compared with those with VCIHS-NP (Table III in the online-only Data Supplement). Among 428 patients, 76 cases and 173 controls met the study criteria (Figure 1). Comparisons of the unmatched cases and controls are presented in Table IV in the online-only Data Supplement. Compared with controls, cases were more likely to be older and have atrial fibrillation, history of stroke, and more severe stroke on admission and less likely to be educated. After matching for age, education, and NIHSS score on admission, 38 cases and 66 controls were ultimately enrolled. Using a prespecified matching algorithm, 28 cases were matched with 2 controls, and 10 cases were matched with 1 control. Although the imbalance of matching variables between cases and controls was mitigated, there were significant differences in clinical and neuroimaging variables between cases and matched controls (Table 1). Multiple lesions and left hemisphere involvement were more frequently observed in cases compared with controls, and lesion volume was larger in cases than in controls. CHIPS scores were significantly higher in cases than in controls (2.2±2.9 versus 0.9±1.4; P=0.01). The cholinergic pathway, DMN, and CEN were more frequently involved in cases, but the SN was not (Table 1). Those associations remained significant after adjustments for each group of confounders (Methods and Table V in the online-only Data Supplement; Table 2). Among the clinical and neuroimaging variables, history of stroke, lacunes, deep white matter hyperintensities, left hemisphere involvement, and lesion volumes were associated with PSD in the various multivariable models. As for the detailed lesion locations in the cholinergic pathway, the left corona radiata was the most commonly involved location in cases, followed by the left centrum semiovale (Table VI in the online-only Data Supplement). Among major hub locations, hippocampal formation and posterior parietal cortex involvement were more common in cases than in controls. Cumulative lesion maps supported those observations (Figures 2 and 3). For the neuropsychological construct of the cases after matching, memory was impaired (less than the seventh percentile) in all patients, frontal executive function in 89.5%, visuospatial function in 78.9%, and language in 63.2%. About 47.4% of cases showed impairments in all 4 cognitive domains, 36.8% showed impairments in 3 domains, and 15.8% showed impairments in 2 domains. Discussion Our findings suggest that the disruption of neural networks by acute ischemic stroke may contribute to the development of PSD. The involvement of cholinergic pathways and major hubs of neural networks were more frequent in the PSD group than in the control group, even after adjusting for potential confounders. The major differences in lesion locations between cases and controls occurred in the hippocampal formation, posterior parietal cortex, and left corona radiata. This exploratory work is novel in that it examined strategic neural correlates of PSD in the light of recent developments of brain behavior relationships, highlighting the importance of the cholinergic pathways and functional network hubs discovered with resting-state functional MRI. Mounting evidence suggests that human cognitive functions result from interactions and communications among brain regions comprising large-scale networks. 16 However, these approaches were largely focused on various neurodegenerative dementias. 7,16 Although it is reasonable to explain the pathogenesis of PSD with disruption of large-scale networks by stroke lesions, there are few studies about the integrity of these networks in PSD. Our results showed that stroke lesions occurred in the cholinergic pathway in 60.5% of cases compared with 39.4% of controls, and this difference was reflected by the difference in the CHIPS scores. Among various regions of CHIPS scoring, the differences between cases and matched controls were prominent in the left corona radiata, which might constitute the capsular divisions of the lateral cholinergic pathway. 2 Although the corona radiata is not the conventional location for strategic infarction, stroke lesions of the corona radiata were suggested to interrupt the connections between the basal forebrain and remote cortical areas, and ultimately lead to the development of multi-infarct dementia. 2 In fact, cortical cholinergic denervation has been observed in multi-infarct dementia 17 and was also suggested as a pathophysiology of other neurodegenerative dementias, such as AD or Parkinson disease dementia. 6,18 Our results showed that cases had stroke lesions in the DMN 5-fold more frequently than matched controls. The DMN is known to be involved in self-referential mental activity, including autobiographical memory, 7 and overlapped with networks active in episodic memory tasks. 19 Reduced connectivity in the DMN has been reported in AD and mild cognitive impairment. 19,20 No direct evidence for the role of DMN in PSD has been published. Only a few studies have suggested that the DMN is disrupted in patients with vascular cognitive impairment 21 and subcortical vascular mild cognitive impairment. 22 Among the various hubs of the DMN, both hippocampal formations were frequently involved. On the contrary, damage to the posterior cingulate cortex, which is the main hub of the DMN, was rare in both cases and controls; it may be explained in part by that the posterior cingulate cortex is supplied by the anterior cerebral artery, and its occlusion is infrequently observed in patients with stroke (Table VI in the online-only Data Supplement). 23 Studies of AD or mild cognitive impairment showed that the frontal executive network is altered in AD, and this change is associated with clinically observed executive dysfunctions. 24 However, the exact role of the CEN in PSD had not yet been described. In the present study, 90% of cases showed frontal dysfunction on neuropsychological tests, and 26% of cases had stroke lesions in the CEN. Only 6% of controls had stroke lesions in the CEN. Among various cortical hub regions of the CEN, we found that damage to the posterior parietal cortex was significantly associated with PSD. The posterior parietal cortex is mainly related to spatial perception 25 and selective attention during multiple lines of sensory input, together with prefrontal

6 1074 Stroke April 2014 Table 1. Comparisons Between Cases and Matched Controls Cases (n=38) Matched Controls (n=66) P Value* Matching variables Age, y 69.8± ± Education, y 9.1± ± NIHSS score on admission 2.8± ± Clinical variables Male 27 (71.1) 37 (56.1) 0.08 Hypertension 27 (71.1) 40 (60.6) 0.35 Diabetes mellitus 13 (34.2) 14 (21.2) 0.16 Hyperlipidemia 7 (18.4) 11 (16.7) 0.78 Atrial fibrillation 5 (13.2) 4 (6.1) 0.37 Exposure to smoking 16 (42.1) 20 (30.3) 0.16 History of stroke 9 (23.7) 6 (9.1) 0.04 Time interval between stroke onset and K-VCIHS-NP, d 491± ± Neuroimaging variables chronic Microbleeds 16 (42.1) 11 (17.2) 0.01 Microbleeds count 2.1± ± Lobar microbleeds 0.8± ± Deep microbleeds 1.2± ± Lacune 14 (36.8) 6 (9.1) <0.01 Lacune count 0.8± ± Chronic cortical lesions 3 (7.9) 4 (6.1) 0.79 Periventricular WMHs (range, 0 3) 2.0± ± Deep WMHs (range, 0 3) 1.5± ±0.7 <0.01 MTA (range, 0 4) 1.2± ± Dichotomized MTA 14 (36.8) 11 (16.7) 0.04 Cerebral atrophy index 0.17± ± Neuroimaging variables acute Lesion volume, cm ± ± Left hemisphere involvement 30 (78.9) 32 (48.5) <0.01 Multiple lesions 19 (50.0) 19 (28.8) 0.02 Acute cortical lesions 21 (55.3) 24 (36.4) 0.06 Disruption of networks Cholinergic pathway 23 (60.5) 26 (39.4) 0.03 CHIPS 2.2± ± DMN 14 (36.8) 5 (7.6) <0.01 CEN 10 (26.3) 4 (6.1) 0.01 SN 3 (7.9) 6 (9.1) 0.88 Values are presented as mean±sd for continuous variables or as number (%) of subjects for categorical variables. CEN indicates central executive network; CHIPS, Cholinergic Pathways HyperIntensities Scale; DMN, default mode network; K-VCIHS- NP, Korean Vascular Cognitive Impairment Harmonization Standards: Neuropsychology Protocol; MTA, medial temporal lobe atrophy; NIHSS, National Institutes of Health Stroke Scale; SN, salience network; and WMHs, white matter hyperintensities. *By conditional logistic regression analysis only considering the matching stratum. cortex. 26 However, we did not find an association between PSD and prefrontal cortex damage. Infrequent involvement of the prefrontal cortex was observed in both cases and controls (Table VI in the online-only Data Supplement). The SN is essential for dissociating the most pertinent stimuli among internal and external stimuli for deciding behavior. 7 Involvement of the SN may be useful for differentiating among various neurodegenerative dementia syndromes. 16 However, in our results, the frequencies of stroke lesions in the hubs of the SN were not significantly different between cases and matched controls. Our study differed from previous ones in terms of the classification of lesion locations and the cognitive assessment tools. The usual methods of classifying lesion locations used in previous studies were cortical versus subcortical, laterality, multiplicity, and number and size of lesions. 8,27 A recent

7 Lim et al Neural Substrates of Poststroke Dementia 1075 Table 2. Odds Ratios (95% Confidence Intervals) for PSD With Adjustments for Clinical or Neuroimaging Variables Clinical Variables (Sex, Diabetes Mellitus, Smoking, History of Stroke) Premorbid Cerebrovascular Changes (Microbleeds, Lacunes, Periventricular WMHs, Deep WMHs) Adjusted for Premorbid Neurodegenerative Changes (Dichotomized MTA, Cerebral Atrophy Index) Lesion Characteristics (Left Hemisphere, Cortical, Multiple) Lesion Characteristics (Lesion Volume) CHIPS 1.83 ( ) 1.68 ( ) 2.33 ( ) 1.98 ( ) 1.67 ( ) Cholinergic pathway 3.52 ( ) 1.82 ( ) 5.15 ( ) 6.50 ( ) 3.36 ( ) DMN ( ) ( ) ( ) 7.00 ( ) 7.12 ( ) CEN 7.58 ( ) ( ) ( ) 8.27 ( ) 7.08 ( ) SN 3.33 ( ) 2.61 ( ) 0.59 ( ) 0.60 ( ) 0.29 ( ) CEN indicates central executive network; CHIPS, Cholinergic Pathways HyperIntensities Scale; DMN, default mode network; MTA, medial temporal lobe atrophy; PSD, poststroke dementia; SN, salience network; and WMHs, white matter hyperintensities. comprehensive review based on 73 eligible articles showed that left hemisphere stroke conferred 1.4 times the risk of PSD, and brain stem infarction was less likely to result in PSD. 1 However, those studies did not use sophisticated methods to delineate and classify the location of acute ischemic lesions is in this study. Because of recent advances of neuroimaging analysis techniques such as lesion registration and cumulative mapping, we could analyze the effects of specific lesion locations on the development of PSD. Furthermore, we adopted the recently proposed VCIHS-NP for the evaluation of cognitive function. 13 Previous studies about PSD have used cognitive assessment tools with a limited number of domains. 8,27 VCIHS-NP was proposed to systematically evaluate the cognitive function of patients with vascular cognitive impairment. This also could be another reason that our study showed different results from the previous ones. This study has limitations. First, the small sample size did not allow us to simultaneously adjust all the potential confounders, and the possibility of inadequate adjustment or residual confounding remains. We matched for NIHSS score on admission to overcome the imbalance in stroke severity between cases and controls, but it should be noted that NIHSS score is weighted more for sensorimotor severity than cognitive deficits. Furthermore, the imbalance in lesion volume continued despite severity matching. Although the associations between neural network disruptions and PSD remained significant after adjusting for lesion volume, we could not exclude the possibility that larger lesions disrupt more network nodes and interconnections, thus posing a greater risk for dementia. Lesion multiplicity might have an additive volume effect and contribute to the development of PSD through multiple network disruption. The possibility of residual confounding or inadequate adjustments for other pre-existing brain injury shown by brain imaging should also be noted. Second, we presented cumulative lesion maps rather than functional MRI data. For this reason, we should be cautious about using the term, disrupted networks. We inferred functional network disruption based on the existence of acute ischemic lesions in the constituents of CHIPS or the major hub cortical regions of each network. Hence, it may be more reasonable to refer to disrupted nodes or connectivity than disrupted networks. Nevertheless, we hope this will stimulate future research into network disruption using resting-state functional MRI and diffusion tensor imaging. Last, because of the limitation of visual measurement, our methods may not be specific to cholinergic pathway. Neural pathways might be necessarily intermingled with other corticocortical and projection fibers. Thus, the exact scoring of involvement of specific pathways may be challenging. Although we strictly adhered to the previous methods designed to preclude any arbitrary delineation of cholinergic pathways, 2,3 the possibility of overlapping with other multimodal association pathways should be also considered. In conclusion, our results suggest that acute ischemic stroke mediated disruption of cholinergic pathways and major hubs of large-scale networks may result in the development of clinically significant dementia after stroke. Sources of Funding This study was partly supported by a grant from the Korea Healthcare Technology R&D Project; Ministry of Health, Welfare, and Family Affairs; Republic of Korea (HI10C2020); and by a grant from SK Chemicals. The funding organizations did not participate in the design, conduct, or analysis of the study or in the preparation of this report. Dr Bae had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Disclosures Dr Bae is a principal investigator, a member of the steering committee, and a site investigator of multicenter clinical trials or clinical studies sponsored by Otsuka Korea, Bayer Korea, Handok Pahrmaceutical Company, SK Chemicals, ESAI-Korea, Daewoong Pharmaceutical Co. Ltd, Daichi Sankyo, Pfizer, Sanofi-Aventis Korea, Dong-A Pharmaceutical, and Yuhan Corporation; served as the consultant or scientific advisory board for Bayer Korea, Boehringer Ingelheim Korea, YuYu Pharmaceutical Company, BMS Korea, and Pfizer Korea; and received lecture honoraria from MSD Korea, AstraZeneca Korea, BMS Korea, Novartis Korea, Otsuka Korea, Pfizer Korea, Daichi Sankyo Korea, and Handok Pharmaceutical Company (modest). Dr D.-E. Kim is a site investigator of multicenter clinical trials sponsored by Otsuka Korea and Sanofi-Aventis Korea, has served as a consultant for Otsuka Korea and Sanofi-Aventis Korea, and received lecture honoraria from Otsuka Korea and Boehringer Ingelheim Korea. Dr Black has been an ad hoc consultant to Novartis, Bristol Meyers Squibb, Roche, GlaxoSmithKline, Pfizer, and Elan and has received speaker honoraria from Novartis, Pfizer, and Eisai. She has been a principal investigator on clinical trials sponsored by Novartis, Pfizer, Roche, and Elan (no personal remuneration). The other authors report no conflicts.

8 1076 Stroke April 2014 References 1. Pendlebury ST, Rothwell PM. Prevalence, incidence, and factors associated with pre-stroke and post-stroke dementia: a systematic review and meta-analysis. Lancet Neurol. 2009;8: Selden NR, Gitelman DR, Salamon-Murayama N, Parrish TB, Mesulam MM. Trajectories of cholinergic pathways within the cerebral hemispheres of the human brain. Brain. 1998;121(pt 12): Bocti C, Swartz RH, Gao FQ, Sahlas DJ, Behl P, Black SE. A new visual rating scale to assess strategic white matter hyperintensities within cholinergic pathways in dementia. Stroke. 2005;36: Swartz RH, Sahlas DJ, Black SE. Strategic involvement of cholinergic pathways and executive dysfunction: Does location of white matter signal hyperintensities matter? J Stroke Cerebrovasc Dis. 2003;12: Román GC, Salloway S, Black SE, Royall DR, Decarli C, Weiner MW, et al. Randomized, placebo-controlled, clinical trial of donepezil in vascular dementia: differential effects by hippocampal size. Stroke. 2010;41: Smith CM, Swash M. Possible biochemical basis of memory disorder in Alzheimer disease. Ann Neurol. 1978;3: Bressler SL, Menon V. Large-scale brain networks in cognition: emerging methods and principles. Trends Cogn Sci. 2010;14: Saczynski JS, Sigurdsson S, Jonsdottir MK, Eiriksdottir G, Jonsson PV, Garcia ME, et al. Cerebral infarcts and cognitive performance: importance of location and number of infarcts. Stroke. 2009;40: Yu KH, Cho SJ, Oh MS, Jung S, Lee JH, Shin JH, et al; Korean-Vascular Cognitive Impairment Harmonization Standards Study Group. Cognitive impairment evaluated with Vascular Cognitive Impairment Harmonization Standards in a multicenter prospective stroke cohort in Korea. Stroke. 2013;44: Lee DW, Lee JY, Ryu SG, Cho SJ, Hong CH, Lee JH, et al. Validity of the Korean version of informant questionnaire on cognitive decline in the elderly (IQCODE). J Korean Geriatr Soc. 2005;9: Kang SJ, Choi SH, Lee BH, Kwon JC, Na DL, Han SH; Korean Dementia Research Group. The reliability and validity of the Korean instrumental activities of daily living (K-IADL). J Korean Neurol Assoc. 2002;20: American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Text Revision. Washington, DC: American Psychiatric Association; Hachinski V, Iadecola C, Petersen RC, Breteler MM, Nyenhuis DL, Black SE, et al. National Institute of Neurological Disorders and Stroke-Canadian Stroke Network vascular cognitive impairment harmonization standards. Stroke. 2006;37: Lee BC, Roh JK; Korean Stroke Registry. International experience in stroke registries: Korean Stroke Registry. Am J Prev Med. 2006;31(6 suppl 2):S243 S Kim DE, Park KJ, Schellingerhout D, Jeong SW, Ji MG, Choi WJ, et al. A new image-based stroke registry containing quantitative magnetic resonance imaging data. Cerebrovasc Dis. 2011;32: Pievani M, de Haan W, Wu T, Seeley WW, Frisoni GB. Functional network disruption in the degenerative dementias. Lancet Neurol. 2011;10: Coricelli G, Nagel R. Neural correlates of depth of strategic reasoning in medial prefrontal cortex. Proc Natl Acad Sci U S A. 2009;106: Klein JC, Eggers C, Kalbe E, Weisenbach S, Hohmann C, Vollmar S, et al. Neurotransmitter changes in dementia with Lewy bodies and Parkinson disease dementia in vivo. Neurology. 2010;74: Greicius MD, Srivastava G, Reiss AL, Menon V. Default-mode network activity distinguishes Alzheimer s disease from healthy aging: evidence from functional MRI. Proc Natl Acad Sci U S A. 2004;101: Petrella JR, Sheldon FC, Prince SE, Calhoun VD, Doraiswamy PM. Default mode network connectivity in stable vs progressive mild cognitive impairment. Neurology. 2011;76: Sheorajpanday RV, Marien P, Weeren AJ, Nagels G, Saerens J, van Putten MJ, et al. EEG in silent small vessel disease: sloreta mapping reveals cortical sources of vascular cognitive impairment no dementia in the default mode network. J Clin Neurophysiol. 2013;30: Sridharan D, Levitin DJ, Menon V. A critical role for the right fronto-insular cortex in switching between central-executive and default-mode networks. Proc Natl Acad Sci U S A. 2008;105: Brust JCM, Chamorro A. Anterior cerebral artery disease. In: Mohr JP, Wolf PA, Grotta JC, Moskowitz MA, Mayberg MR, Kummer R, eds. Stroke: Pathophysiology, Diagnosis, and Management. Philadelphia, PA: Elsevier Saunders; 2011: Agosta F, Pievani M, Geroldi C, Copetti M, Frisoni GB, Filippi M. Resting state fmri in Alzheimer s disease: beyond the default mode network. Neurobiol Aging. 2012;33: Nachev P, Husain M. Disorders of visual attention and the posterior parietal cortex. Cortex. 2006;42: Shomstein S, Kravitz DJ, Behrmann M. Attentional control: temporal relationships within the fronto-parietal network. Neuropsychologia. 2012;50: Troncoso JC, Zonderman AB, Resnick SM, Crain B, Pletnikova O, O Brien RJ. Effect of infarcts on dementia in the Baltimore longitudinal study of aging. Ann Neurol. 2008;64: