Brain Advance Access published November 2, 2014 doi:10.1093/brain/awu316

BRAIN 2014: Page 1 of 10

| 1


Structural network alterations and neurological dysfunction in cerebral amyloid angiopathy Yael D. Reijmer,1 Panagiotis Fotiadis,1 Sergi Martinez-Ramirez,1 David H. Salat,2 Aaron Schultz,2 Ashkan Shoamanesh,1 Alison M. Ayres,1 Anastasia Vashkevich,1 Diana Rosas,3 Kristin Schwab,1 Alexander Leemans,4 Geert-Jan Biessels,5 Jonathan Rosand,1 Keith A. Johnson,3,6 Anand Viswanathan,1 M. Edip Gurol1 and Steven M. Greenberg1

1 2 3 4 5 6

Stroke Research Center, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA Athinoula A. Martinos Centre for Biomedical Imaging, Massachusetts General Hospital, Boston, MA, USA Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA Image Sciences Institute, University Medical Centre Utrecht, Utrecht, The Netherlands Department of Neurology, Brain Centre Rudolf Magnus, University Medical Centre Utrecht, Utrecht, The Netherlands Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA

Correspondence to: Steven M. Greenberg, MD, PhD, J.P. Kistler Stroke Research Centre, 175 Cambridge Street, Suite 300, Boston, MA, USA, 02114 E-mail: [email protected]

Received July 28, 2014. Revised August 29, 2014. Accepted September 21, 2014. ß The Author (2014). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]

Downloaded from by guest on November 8, 2014

Cerebral amyloid angiopathy is a common form of small-vessel disease and an important risk factor for cognitive impairment. The mechanisms linking small-vessel disease to cognitive impairment are not well understood. We hypothesized that in patients with cerebral amyloid angiopathy, multiple small spatially distributed lesions affect cognition through disruption of brain connectivity. We therefore compared the structural brain network in patients with cerebral amyloid angiopathy to healthy control subjects and examined the relationship between markers of cerebral amyloid angiopathy-related brain injury, network efficiency, and potential clinical consequences. Structural brain networks were reconstructed from diffusion-weighted magnetic resonance imaging in 38 non-demented patients with probable cerebral amyloid angiopathy (69  10 years) and 29 similar aged control participants. The efficiency of the brain network was characterized using graph theory and brain amyloid deposition was quantified by Pittsburgh compound B retention on positron emission tomography imaging. Global efficiency of the brain network was reduced in patients compared to controls (0.187  0.018 and 0.201  0.015, respectively, P 5 0.001). Network disturbances were most pronounced in the occipital, parietal, and posterior temporal lobes. Among patients, lower global network efficiency was related to higher cortical amyloid load (r = 0.52; P = 0.004), and to magnetic resonance imaging markers of small-vessel disease including increased white matter hyperintensity volume (P 5 0.001), lower total brain volume (P = 0.02), and number of microbleeds (trend P = 0.06). Lower global network efficiency was also related to worse performance on tests of processing speed (r = 0.58, P 5 0.001), executive functioning (r = 0.54, P = 0.001), gait velocity (r = 0.41, P = 0.02), but not memory. Correlations with cognition were independent of age, sex, education level, and other magnetic resonance imaging markers of small-vessel disease. These findings suggest that reduced structural brain network efficiency might mediate the relationship between advanced cerebral amyloid angiopathy and neurologic dysfunction and that such large-scale brain network measures may represent useful outcome markers for tracking disease progression.


| BRAIN 2014: Page 2 of 10

Y. D. Reijmer et al.

Keywords: small vessel disease; mild cognitive impairment; stroke: imaging; amyloid imaging; executive function; tractography Abbreviations: CAA = cerebral amyloid angiopathy; ICH = intracerebral haemorrhage; PiB = Pittburgh compound B


Materials and methods Study design This study is a cross-sectional analysis of data from an ongoing single-centre longitudinal cohort study on the natural history of CAA. The Institutional Review Board approved the study and informed consent was obtained from all participants or their surrogates.

Study participants Thirty-eight non-demented patients with probable (n = 34) or definite (n = 4) CAA defined by the Boston criteria (Knudsen et al., 2001) who underwent research MRI between May 2006 and April 2013 were included in this study. Seventeen patients had a symptomatic ICH before enrolment, but all research imaging was performed at least 6 months after the ICH in these patients. CAA participants underwent a clinical evaluation, extensive cognitive testing (described below) and research 1.5 T MRI scan including a T1-weighted 3D spoiled gradient recalled-echo (SPGR), fluid attenuated inversion recovery (FLAIR), susceptibility-weighted imaging and diffusion-weighted imaging sequence (Auriel et al., 2014). Twenty-nine control subjects without CAA, with similar ages, were selected from among cognitively normal participants in the National Alzheimer’s Coordinating Centre-based Longitudinal Cohort (Dumas et al., 2012) and the Harvard Cooperative Program on Ageing at MGH (Salat et al., 2006) and underwent the same clinical evaluation and MRI protocol as the patient group. Exclusion criteria of both groups were

Downloaded from by guest on November 8, 2014

Vascular cognitive impairment due to cerebral small vessel disease represents a major public health threat to rapidly ageing populations (Savva et al., 2010; Arvanitakis et al., 2011). However, the development of effective treatment to modify or prevent vascular cognitive impairment is hampered by the absence of a clear understanding of how individual small brain lesions produce cognitive impairments and how to measure brain injury in a clinically meaningful way (Greenberg et al., 2014). Current MRI markers of small vessel disease-related brain injury include white matter T2 hyperintensities, lacunes, global atrophy, microbleeds, and microinfarcts (Wardlaw et al., 2013). Although these markers have been related to cognitive outcome in population-based (Prins et al., 2005) and stroke cohorts (Gregoire et al., 2011; Patel et al., 2013), correlations are generally relatively weak or inconsistent across studies (van der Vlies et al., 2012; Kloppenborg et al., 2014). One explanation for the small effect sizes might be that the cumulative effect of these small, spatially distributed lesions, rather than the individual lesions themselves, determines their impact on cognition. This is consistent with findings from autopsy studies showing multiple concurrent pathologies in cases with dementia (Schneider et al., 2007). Whole-brain connectivity might be particularly vulnerable to the effects of multiple spatially distributed brain lesions. Brain connectivity has particular relevance to neurological functions that are preferentially affected in patients with small vessel disease, such as information processing speed, executive functioning and gait, which depend on the rapid communication between widespread brain regions (Hachinski et al., 2006; Viswanathan and Sudarsky, 2012; Duering et al., 2014). In the present study we aimed to address these possibilities by analysing large-scale structural brain network measures reconstructed from diffusion-weighted images as a marker of small vessel disease-related brain injury. We applied this approach to cerebral amyloid angiopathy (CAA), a form of small vessel disease associated with intracerebral haemorrhage (ICH) and an important contributor to vascular cognitive impairment independent of ICH (Viswanathan and Greenberg, 2011). A CAA cohort has several inherent advantages for this analysis, including its characteristic pattern of posterior-predominant small vessel disease lesions such as lobar microbleeds, white matter hyperintensities, and small asymptomatic infarcts (Rosand et al., 2005; Auriel et al., 2012; Zhu et al., 2012; Thanprasertsuk et al., 2014), and the ability to detect the vascular disease itself by molecular amyloid imaging.

Although Pittburgh compound B (PiB) cannot differentiate between vascular amyloid and parenchymal amyloid plaques, several PiB PET studies have provided convincing evidence that PiB retention in patients with CAA is indeed largely driven by vascular amyloid burden (Johnson et al., 2007; Greenberg et al., 2008; Gurol et al., 2012). Our unifying hypothesis was that structural connectivity measures are reduced in patients with CAA and that they constitute a pathway between vascular amyloid and clinical dysfunction. To that end, our specific aims were to show that: (i) the efficiency of the network is reduced in individuals with CAA compared to similar aged controls; (ii) that the posterior predominance of CAA-related vascular pathology (Nelson et al., 2013) is reflected in the pattern of local network disturbances; (iii) that individual MRI and PET markers of CAA-related brain injury would be associated with reduced global network efficiency; and (iv) that global network efficiency correlates with neurological function measured by tests of processing speed, executive functioning, and gait velocity.

Brain network alterations in CAA

dementia, a diagnosis of cerebrovascular disease other than CAA, and contraindication to functional MRI (metallic implant/devices, claustrophobia or seizure history). Controls were excluded if they had a history of neurological or psychiatric disorder, a history of stroke or haemorrhage, or serious cardiovascular disease.

Diffusion tensor imaging processing and network reconstruction High angular resolution diffusion imaging (HARDI) scans (60 directions, b-value 700, 10 b0 images, voxel size 2  2  2 mm3) were obtained on a 1.5 T MRI scanner (Siemens) and were analysed and processed in ExploreDTI ( The HARDI scans were corrected for subject motion and eddy current induced geometric distortions as described previously (Leemans and Jones, 2009). The diffusion tensors were calculated using the RESTORE approach (Chang et al., 2012). For each data set, whole-brain white matter tractography was performed using deterministic streamline constrained spherical deconvolution (CSD)-based tractography (Jeurissen

BRAIN 2014: Page 3 of 10

| 3

et al., 2011). The CSD method can model multiple fibre populations within a voxel and therefore allows fibre tracking to proceed through crossing fibre regions (Jeurissen et al., 2012; Reijmer et al., 2012). Fibres were reconstructed by starting seed samples uniformly throughout the white matter at 2 mm isotropic resolution with a maximum deflection angle of 45 and a fibre orientation distribution threshold of 0.1. The whole-brain fibre tract reconstructions were parcellated into 90 cortical and subcortical grey matter regions using the automated anatomical labelling atlas (AAL; Tzourio-Mazoyer et al., 2002). Two brain regions or nodes were considered to be connected if a fibre bundle was present with two end points located in these regions, resulting in a 90  90 binary connectivity matrix. A weighted connectivity matrix was obtained by multiplying each connection by the mean fractional anisotropy of that connection (Fig. 1). Fractional anisotropy is a commonly used metric to examine the microstructural aspects of brain connectivity (van den Heuvel and Sporns, 2011) and has been shown to be sensitive to CAA-related white matter injury (Salat et al., 2006). To account for the effects of ICH on structural connectivity, we also performed reconstruction limited to the ICH-free

Downloaded from by guest on November 8, 2014

Figure 1 Flowchart of the brain network reconstruction. For each subject, whole brain fibre tract reconstructions (A) are parcellated into 90 cortical and subcortical regions (B) resulting in a 90  90 whole-brain network (C). Connections were weighted by the mean fractional anisotropy of the fibre tract pathways. In addition, we calculated for each subject the network of a single hemisphere (D). In CAA patients with ICH (as indicated on the T2* weighted MRI) the non-affected hemisphere was selected. In patients without ICH and controls one hemisphere was also excluded such that half of the patient and control group had a network based on the right hemisphere and half on the left hemisphere. The global efficiency of the network was computed from the whole-brain and hemispheric connectivity matrices using graph theory (Rubinov and Sporns, 2010).


| BRAIN 2014: Page 4 of 10

hemisphere, resulting in a 45  45 connectivity matrix for each subject. In this analysis, we followed the same procedure for patients with CAA without ICH and control subjects by excluding one hemisphere such that half of the patient and control group had a network based on the right hemisphere and half on the left hemisphere.

Network characteristics

Neuroimaging markers of amyloid load and CAA-related brain injury Microbleeds were identified on axial susceptibility-weighted imaging sequences by an experienced rater (S.M.R.) as described previously (Martinez-Ramirez et al., 2013). White matter hyperintensity segmentation on FLAIR MRI scans was performed using a semi-automated method (Gurol et al., 2006). We have previously reported a high interrater concordance for microbleeds (Greenberg et al., 1999), and white matter hyperintensity (Gurol et al., 2006) measurements. Total brain volumes were obtained from the T1-weighted SPGR images using Freesurfer (http://surfer.nmr.mgh.harvard. edu) (Fischl and Dale, 2000) and included the brain tissue within the cranium, excluding the brainstem, ventricles, CSF and choroid plexus. All brain volumes were visually inspected for segmentation errors. For individuals with an ICH, the volume estimates of the ICH-free hemispheres were used and multiplied by 2. To account for between-subject differences in head size, brain volumes were expressed as percentage of intracranial volume. We also computed the median fractional anisotropy of the total white matter of each subject using a normalized white matter atlas (Mori et al., 2008) to evaluate whether the correlations with network efficiency added explained variance on top of median white matter fractional anisotropy. Voxels were selected with a minimal fractional anisotropy threshold of 0.2. In individuals with an ICH, the median fractional anisotropy of the ICH-free hemisphere was computed.

A subsample of the patient group (n = 29, 76%) had PiB PET imaging. PiB PET acquisitions and processing were performed as described (Gurol et al., 2013). PET data were reconstructed with ordered set expectation maximization, corrected for attenuation, and each frame was evaluated to verify absence of head motion. The distribution volume ratio was calculated to express specific PiB retention with cerebellar cortex as the reference tissue. To obtain the distribution volume ratio in specific cortical regions, each subject’s PiB PET data were normalized and parcellated into four different lobes (frontal, parietal, temporal, and occipital) (Johnson et al., 2007; Gurol et al., 2012). Total distribution volume ratio was defined as the mean of all four cortical regions of interest. In cases with ICH, distribution volume ratio estimates from the ICH-free hemisphere were used.

Cognitive and gait testing Measures of cognitive functioning among the subjects with probable CAA were assessed using a standardized test battery. Tests include verbal memory (immediate and delayed memory score of the Hopkins Verbal Learning Test; Brandt, 1991), processing speed (Trail Making Test A: Corrigan and Hinkeldey, 1987; Symbol Substitution Test: Wechsler, 1997), and executive functioning (Trail Making Test B, Digit Span Test backwards: Wechsler, 1987; Verbal Fluency Test: Tombaugh et al., 1999). Each cognitive test score was transformed into z-scores using the mean and standard deviation (SD) of the whole population and averaged across tests to obtain one average z-score per cognitive domain. Cognitive data were not available in three patients: one who declined cognitive testing, one a non-native English speaker, and one who had aphasia as a result of the ICH. Raw cognitive test scores and corresponding age-, sex-, and education-adjusted normative scores are presented in Supplementary Table 2. Gait was assessed using the speed velocity measure of the Timed Get Up and Go Test (Podsiadlo and Richardson, 1991). Data on gait velocity were available from 31 patients with CAA (82%).

Statistical analysis Patient characteristics were compared between individuals with CAA and controls using independent samples t-test for continuous variables and Fisher’s exact test for proportions. The number of microbleeds and white matter hyperintensities volume were non-normally distributed and therefore logtransformed. Measures of brain network organization and network efficiency were compared between CAA patients and controls using ANOVA, adjusted for age. To evaluate whether the posterior predominant pattern of CAA pathology is reflected in the pattern of local network disturbances in directly connected tracts, we compared the mean connectivity strength of each individual node between patients and controls. Nodes within ICH-containing hemispheres were excluded from this analysis. P-values of nodal comparisons were adjusted for multiple hypotheses testing using the false discovery rate (FDR) correction (Benjamini and Hochberg, 1995). To test whether structural network alterations form a mechanistic link between vascular amyloid and clinical manifestations, we analysed the relationship between global network

Downloaded from by guest on November 8, 2014

Organizational properties of the brain networks were calculated using the brain connectivity toolbox (https// com/site/bcnet), (Rubinov and Sporns, 2010). Global network efficiency was chosen as the primary outcome measure in this study as it has shown the most robust relationships with cognition in previous diffusion tensor imaging network studies (Wen et al., 2011; Reijmer et al., 2013a, b). Global efficiency is calculated as the inverse of the shortest path lengths (i.e. the minimum number of fractional anisotropy-weighted connections between each pair of brain regions) (Rubinov and Sporns, 2010) and quantifies how efficiently information is exchanged over the network. Local differences in network connectivity were assessed by the mean nodal strength (i.e. the mean fractional anisotropy of the connections projecting to each individual node). To evaluate overall network topology independent of network strength, we also calculated network properties from the unweighted connectivity matrices. Unweighted measures included network density, i.e. the observed number of connections divided by the number of possible connections, and the betweenness centrality of each node. Nodes with a high betweenness centrality are central nodes that participate in many short paths and indicate nodal ‘hubs’.

Y. D. Reijmer et al.

Brain network alterations in CAA

BRAIN 2014: Page 5 of 10

efficiency and (i) measures of CAA-related brain injury (i.e. MRI markers of small vessel disease burden and brain amyloid on PiB PET imaging); and (ii) cognitive and gait performance within the CAA group. The relationship between network connectivity and PiB uptake was further examined by testing whether patients with predominantly posterior PiB-uptake (posterior distribution volume ratio 4 frontal distribution volume ratio) also showed predominantly posterior network disturbances. Posterior was defined as the mean of the occipital, parietal and temporal lobe. The mean strength of posterior and frontal nodes was compared between patients with and without predominantly posterior PiB-uptake and controls using an independent samples t-test. Lastly, we examined whether global network efficiency explained variance in cognition/gait on top of other MRI markers with stepwise linear regression analyses. Potential predictors including age, sex, education level, white matter hyperintensities volume, median fractional anisotropy, microbleeds, total brain volume, and global efficiency were sequentially entered in the model and kept if they explained additional variance compared to the previous model (P-value change R2 5 0.05).

Results We analysed structural network measures in 38 patients with CAA and 29 similar aged control subjects (Table 1). Overall, CAA patients had greater white matter hyperintensity volume and lower fractional anisotropy than controls (P 5 0.05). Age- and sex-adjusted comparison in network parameters showed no substantial difference in network density and the location of major network hubs (Table 2 and Supplementary Table 1), but global network efficiency was reduced in patients with CAA compared to controls (P 5 0.001, Table 2). The between-group difference in global efficiency remained present when restricting the analysis to ICH-free hemispheres (P = 0.01, Table 2). We also observed no difference in global efficiency of the ICH-free hemispheric network between CAA patients with and without ICH (0.195  0.022 and 0.196  0.024, respectively

P = 0.94), indicating that the ICH had no major effects on efficiency in the contralateral hemisphere. Figure 2 shows network nodes with lower nodal strength in CAA compared to controls (P 5 0.05, FDR corrected). The strongest differences in connectivity of nodes appeared in occipital, posterior parietal, and posterior temporal cortex. There were no nodes with a significantly higher global efficiency in patients with CAA compared with control subjects. Affected nodes in patients with CAA did not differ from unaffected nodes with respect to average number of connections (P = 0.89) or total connectivity strength (P = 0.56).

Relationship between network measures and neuroimaging markers of CAA-related brain injury Within the CAA group, lower global efficiency was correlated to increased age (r = 0.63, P 5 0.001), greater white matter hyperintensity volume (r = 0.57, P 5 0.001), and lower total brain volume (r = 0.39, P = 0.02). When restricting our analyses to the ICH-free hemispheres, similar correlations were observed with age (r = 0.64, P 5 0.001), white matter hyperintensity volume (r = 0.68, P 5 0.001) and total brain volume (r = 0.39, P = 0.02). Global efficiency tended to decline with increasing number of microbleeds (r = 0.31, P = 0.06). Among the 29 patients with CAA who received PiB PET imaging, global network efficiency correlated with global mean distribution volume ratio in analysis of whole-brain efficiency (r = 0.43; P = 0.02; Fig. 3) or when restricting to the ICH-free hemispheres (r = 0.52; P = 0.004). CAA patients with predominantly posterior PiB-uptake (posterior distribution volume ratio 4 frontal distribution volume ratio, n = 14) also demonstrated predominantly posterior network disturbances: this subgroup of patients showed lower strength of posterior nodes compared to controls (mean difference  SEM: 0.16  0.007, P = 0.02), whereas no difference was found in frontal node strength ( 0.005  0.007, P = 0.46). The remaining subgroup of CAA patients (n = 15) showed greater frontal as well as posterior PiB retention compared to patients with

Table 1 Characteristics of the study sample

Age Sex, % male ICH, % Microbleeds, n White matter hyperintensity volume, % ICV Total brain volume, % ICV Fractional anisotropy white matter

Controls n = 29

CAA n = 38


71.3  7.1 69 0.1 (0–2.1)

68.9  9.9 82 45 35 (1–729) 1.3 (0.1–6.8)

40.1 40.1 50.001

66  3 0.39  0.03

62  4 0.36  0.03

40.1 50.001

Data are given as mean  SD, percentages or as median (range). ICV = intracranial volume.

Table 2 Between-group differences in brain network parameters Controls



Network density (%) 24.6  1.6 24.0  2.9 40.1 Average network weight 0.286  0.017 0.304  0.017 50.001 Global efficiency whole brain 0.201  0.015 0.187  0.018 50.001 Global efficiency ICH-free 0.205  0.018 0.194  0.023 0.012 hemisphere Global network efficiency measures are given as mean  SD of age- and sex-adjusted zscores. Network density of the whole brain network indicates the total number of connections divided by the total number of possible connections within the network.

Downloaded from by guest on November 8, 2014

Between-group comparisons

| 5


| BRAIN 2014: Page 6 of 10

Y. D. Reijmer et al.

Figure 2 Axial and sagittal projections of grey matter nodes with reduced connectivity strength in CAA compared to controls based on the whole-brain networks. Nodes with an FDR corrected thresholds of P 5 0.05 are indicated in orange and P 5 0.005 are indicated in red. Nodes in ICH containing hemispheres were excluded from the analysis. The cortical and subcortical brain regions corresponding to each node are given in Supplementary Fig. 1.

Relationship between network measures and neurological function in CAA Whole-brain global efficiency was correlated with worse performance on cognitive tests measuring processing speed (r = 0.58; P 5 0.001) and executive functioning (r = 0.54; P = 0.001), but not memory (r = 0.18, P = 0.31; Fig. 4). The associations with speed and executive functioning remained independent after controlling for potential confounders as detailed above (P 5 0.05; Supplementary Table 3), indicating that the network-derived global efficiency measure carried additional information beyond these conventional structural small vessel disease measures. The correlation with speed also remained independent after controlling for average network strength (P = 0.03), while the relationship with executive functioning became a trend (P = 0.14). The strength of the correlations with cognition did not change when restricting the analyses to global efficiency of the ICH-free hemispheres (Supplementary Table 3). Gait velocity was measured in a subsample of 31 subjects with CAA and correlated with global efficiency of the ICHfree hemispheres (r = 0.41, P = 0.02), though not wholebrain global efficiency (r = 0.30; P = 0.11). The relationship between gait velocity and efficiency of the ICH-free hemisphere was not independent of median fractional anisotropy (P = 0.193; Supplementary Table 3).

and the distribution volume ratio obtained from PiB PET imaging in patients with CAA. High distribution volume ratio (DVR) indicates high global amyloid burden.

Discussion The present study shows reduced global efficiency of structural brain networks in patients with CAA compared with controls, independent of ICH. CAA-related reductions in global network efficiency were related to cortical amyloid burden, MRI markers of small vessel disease, and worse cognitive and gait performance, supporting our view that microstructural alterations can be a link between vascular pathology and clinical manifestations. Importantly, the associations with cognition remained independent after controlling for white matter hyperintensity load, microbleeds, brain volume, or median fractional anisotropy, suggesting that network measures can explain additional variance not captured by these currently used MRI markers. Our findings of lower structural network efficiency and its association with neurological outcome are consistent with previous reports in patients with Alzheimer’s disease (Lo et al., 2010), diabetes (Reijmer et al., 2013a) and small vessel disease (Lawrence et al., 2014). The consistency of these findings across different populations, despite subtle differences in image processing, suggests that diffusionbased network measures are indeed sensitive to clinically relevant brain injury. The current results support a model in which CAA leads to multiple types of spatially distributed brain lesions which together impact the strength and efficiency of the brain network. Although the exact pathophysiological mechanism underlying structural network disturbances in these patients remain to be established, several of our findings point towards an important role of CAA-related pathology. First, reductions in brain connectivity were most pronounced in tracts projecting onto the occipital and posterior temporal cortex, whereas subcortical tracts were less affected. This pattern of brain injury is consistent

Downloaded from by guest on November 8, 2014

predominantly posterior PiB (P 5 0.01), suggestive of more severe global CAA pathology. In line with the higher global PiB uptake, those patients also showed lower frontal and posterior node strength than controls (posterior: 0.023  0.007, frontal: 0.018  0.006, P 5 0.009).

Figure 3 Relationship between global network efficiency

Brain network alterations in CAA

BRAIN 2014: Page 7 of 10

| 7

(z-scores) (A–C) and gait velocity (m/s) (D) in patients with CAA.

with reports from post-mortem and imaging studies showing a posterior lobar predominance of vascular amyloid deposition (Nelson et al., 2013), white matter hyperintensity burden (Zhu et al., 2012; Thanprasertsuk et al., 2014), and PiB uptake (Johnson et al., 2007) in patients with CAA. Secondly, we found a correlation between global network efficiency and amyloid burden on PiB PET imaging, suggesting that the vascular disease itself contributes to the observed network disturbances. The greater tendency for reduced frontal node strength in brains with relatively greater frontal PiB retention supports the possibility of regional association between worsened CAA burden and damage to the local structural network. PiB cannot differentiate between vascular amyloid and parenchymal amyloid plaques, thus we do not know to what extent parenchymal amyloid contributed to the PiB PET signal. However, the fact that global network efficiency was related to markers of small vessel disease on MRI (i.e. white matter hyperintensity volume, total brain volume, microbleeds) strengthens our view that the observed network alterations are a result of CAA-related brain injury.

Finally, global network efficiency was correlated with reduced processing speed, executive functioning, and gait velocity, but not with memory performance. This clinical profile is typically observed in patients with vascular cognitive impairment (Hachinski et al., 2006) and to a lesser degree in Alzheimer’s disease, supporting a primarily vascular mechanism underlying the network alterations. Importantly, global network efficiency explained changes in cognition in beyond individual markers of small vessel disease, indicating that network measures provide information that is complementary to widely used conventional MRI biomarkers. One explanation for the relatively strong correlation with cognition is that diffusion-based network measures account for the total burden of brain injury, including microstructural tissue abnormalities such as microinfarcts not visible to the naked eye (Smith et al., 2012). Measures of global network efficiency also take into account the position of the connection within the network: central connections that participate in many shortest paths have a greater contribution to the overall efficiency (Alstott et al., 2009; van den Heuvel et al., 2012). These central connections may

Downloaded from by guest on November 8, 2014

Figure 4 Correlation plots. Correlation plots showing the relationship between global network efficiency and cognitive performance


| BRAIN 2014: Page 8 of 10

Funding This work is supported by grants from the National Institutes of Health [R01 AG26484 and R01 NS070834, K23 NS083711 and R01NR010827], from the American Heart Association [14POST20010031], and from the Netherlands Organization for Scientific Research (NWO) [VIDI grant 639.072.411].

Conflict of interest Dr Rosand serves as a consultant for Boehringer Ingelheim. Drs Rosand, Greenberg, and Gurol receive research support from National Institutes of Health. The other authors report no conflicts.

Supplementary material Supplementary material is available at Brain online.

References Alstott J, Breakspear M, Hagmann P, Cammoun L, Sporns O. Modeling the impact of lesions in the human brain. PLoS Comput Biol 2009; 5: e1000408. Arvanitakis Z, Leurgans SE, Barnes LL, Bennett DA, Schneider JA. Microinfarct pathology, dementia, and cognitive systems. Stroke 2011; 42: 722–7. Auriel E, Gurol ME, Ayres A, Dumas AP, Schwab KM, Vashkevich A, et al. Characteristic distributions of intracerebral hemorrhage-associated diffusion-weighted lesions. Neurology 2012; 79: 2335–41. Auriel E, Edlow BL, Reijmer YD, Fotiadis P, Ramirez-Martinez S, Ni J, et al. Microinfarct disruption of white matter structure: a longitudinal diffusion tensor analysis. Neurology 2014; 83: 182–8. Bastiani M, Shah NJ, Goebel R, Roebroeck A. Human cortical connectome reconstruction from diffusion weighted MRI: the effect of tractography algorithm. Neuroimage 2012; 62: 1732–49. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B 1995; 57: 289–300. Brandt J. The hopkins verbal learning test: development of a new memory test with six equivalent forms. Clin Neurophysiol 1991; 5: 125–42. Buckner RL, Snyder AZ, Shannon BJ, LaRossa G, Sachs R, Fotenos AF, et al. Molecular, structural, and functional characterization of Alzheimer’s disease: evidence for a relationship between default activity, amyloid, and memory. J Neurosci 2005; 25: 7709–17. Chang LC, Walker L, Pierpaoli C. Informed RESTORE: a method for robust estimation of diffusion tensor from low redundancy datasets in the presence of physiological noise artifacts. Magn Reson Med 2012; 68: 1654–63. Corder EH, Woodbury MA, Volkmann I, Madsen DK, Bogdanovic N, Winblad B. Density profiles of Alzheimer disease regional brain pathology for the huddinge brain bank: pattern recognition emulates and expands upon braak staging. Exp Gerontol 2000; 35: 851–64. Corrigan J, Hinkeldey N. Relationships between parts A and B of the train making test. J Clin Psychol 1987; 43: 402–9. Duering M, Gesierich B, Seiler S, Pirpamer L, Gonik M, Hofer E, et al. Strategic white matter tracts for processing speed deficits in agerelated small vessel disease. Neurology 2014; 82: 1946–50.

Downloaded from by guest on November 8, 2014

be specifically relevant for cognitive functions such as speed and executive functioning that rely on the global transfer and integration of information between distant brain regions. Indeed, a recent DTI network study showed that long-range association fibres connecting parietal, temporal, and frontal regions are primarily affected in small vessel disease patients with speed and executive deficits, but no memory impairment (Lawrence et al., 2014). Strengths of our study include the collection of multimodal imaging in a well-described cohort of CAA patients in combination with the detailed assessment of cognitive functioning. Our study is limited, however, by the crosssectional design and by the lack of PiB and cognitive data from our control group. Therefore we were not able to examine whether the observed correlations between network efficiency and cognition are specific to CAA. Another limitation is the possible confounding effect of Alzheimer pathology on our results. However, the pattern of local network disturbances in our sample differed from those found in Alzheimer’s disease. In patients with Alzheimer’s disease and amnestic mild cognitive impairment, alterations in network connectivity are primarily seen in temporal and frontal regions, whereas the occipital lobe is relatively spared (Lo et al., 2010; Shu et al., 2012), resembling the progression of Alzheimer-related pathology (Corder et al., 2000). Hub regions have also been shown to be particular vulnerable to Alzheimer’s disease (Buckner et al., 2005), but were not primarily affected in this sample of patients with CAA. These discrepancies suggest that the contribution of Alzheimer pathology to our results is limited. Finally, the optimal protocol for reconstructing network connectivity is still a matter of debate. Different grey matter parcellation methods (Reus, 2013), fibre tractography methods (Bastiani et al., 2012), and measures of connectivity strength (Fornito et al., 2013) have been suggested. The high correlation among weighted network measures makes it difficult to determine whether global efficiency is truly preferred as an outcome marker above other global network measures. Larger study samples are necessary to identify which network parameter is most sensitive to clinically relevant brain injury. Other than comparing different metrics and analytic techniques, future studies should examine whether our findings are generalizable to other vascular cognitive impairment populations and whether global network efficiency can predict cognitive decline. In conclusion, this study supports the view that brain network alterations detected with diffusion-based methods can be one of the mechanistic links between vascular amyloid and neurological dysfunction in patients with CAA. In addition to providing new information about the biological basis of vascular cognitive impairment, diffusion-based network analysis may also prove to be a useful tool for measuring clinically meaningful disease progression in clinical trials or practice.

Y. D. Reijmer et al.

Brain network alterations in CAA

| 9

Mori S, Oishi K, Jiang H, Jiang L, Li X, Akhter K, et al. Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template. Neuroimage 2008; 40: 570–82. Nelson PT, Pious NM, Jicha GA, Wilcock DM, Fardo DW, Estus S, et al. APOE-epsilon2 and APOE-epsilon4 correlate with increased amyloid accumulation in cerebral vasculature. J Neuropathol Exp Neurol 2013; 72: 708–15. Patel B, Lawrence AJ, Chung AW, Rich P, Mackinnon AD, Morris RG, et al. Cerebral microbleeds and cognition in patients with symptomatic small vessel disease. Stroke 2013; 44: 356–61. Podsiadlo D, Richardson S. The timed “up & go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc 1991; 39: 142–8. Prins ND, van Dijk EJ, den HT, Vermeer SE, Jolles J, Koudstaal PJ, et al. Cerebral small-vessel disease and decline in information processing speed, executive function and memory. Brain 2005; 128: 2034–41. Reijmer YD, Leemans A, Brundel M, Kappelle LJ, Biessels GJ. Utrecht Vascular Cognitive Impairment (VCI) Study GroupDisruption of the cerebral white matter network is related to slowing of information processing speed in patients with type 2 diabetes. Diabetes 2013a; 62: 2112–5. Reijmer YD, Leemans A, Caeyenberghs K, Heringa SM, Koek HL, Biessels GJ, et al. Disruption of cerebral networks and cognitive impairment in Alzheimer disease. Neurology 2013b; 80: 1370–7. Reijmer YD, Leemans A, Heringa SM, Wielaard I, Jeurissen B, Koek HL, et al. Improved sensitivity to cerebral white matter abnormalities in Alzheimer’s disease with spherical deconvolution based tractography2012; 7: e44074. Rosand J, Muzikansky A, Kumar A, Wisco JJ, Smith EE, Betensky RA, et al. Spatial clustering of hemorrhages in probable cerebral amyloid angiopathy. Ann Neurol 2005; 58: 459–62. Rubinov M, Sporns O. Complex network measures of brain connectivity: uses and interpretations. Neuroimage 2010; 52: 1059–69. Salat DH, Smith EE, Tuch DS, Benner T, Pappu V, Schwab KM, et al. White matter alterations in cerebral amyloid angiopathy measured by diffusion tensor imaging. Stroke 2006; 37: 1759–64. Savva GM, Stephan BC. Alzheimer’s Society Vascular Dementia Systematic Review GroupEpidemiological studies of the effect of stroke on incident dementia: a systematic review. Stroke 2010; 41: e41–6. Schneider JA, Arvanitakis Z, Bang W, Bennett DA. Mixed brain pathologies account for most dementia cases in community-dwelling older persons. Neurology 2007; 69: 2197–204. Shu N, Liang Y, Li H, Zhang J, Li X, Wang L, et al. Disrupted topological organization in white matter structural networks in amnestic mild cognitive impairment: relationship to subtype. Radiology 2012; 265: 518–27. Smith EE, Schneider JA, Wardlaw JM, Greenberg SM. Cerebral microinfarcts: the invisible lesions. Lancet Neurol 2012; 11: 272–82. Thanprasertsuk S, Martinez-Ramirez S, Pontes-Neto OM, Ni J, Ayres A, Reed A, et al. Posterior white matter disease distribution as a predictor of amyloid angiopathy. Neurology 2014; 83: 1–7. Tombaugh T, Kozak J, Rees L. Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Arch Clin Neuropsychol 1999; 14: 167–17. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Delcroix N, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 2002; 15: 273–89. van den Heuvel MP, Sporns O. Rich-club organization of the human connectome. J Neurosci 2011; 31: 15775–86. van den Heuvel MP, Kahn RS, Goni J, Sporns O. High-cost, highcapacity backbone for global brain communication. Proc Natl Acad Sci USA 2012; 109: 11372–7.

Downloaded from by guest on November 8, 2014

Dumas A, Dierksen GA, Gurol ME, Halpin A, Martinez-Ramirez S, Schwab K, et al. Functional magnetic resonance imaging detection of vascular reactivity in cerebral amyloid angiopathy. Ann Neurol 2012; 72: 76–81. Fischl B, Dale AM. Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proc Natl Acad Sci USA 2000; 97: 11050–5. Fornito A, Zalesky A, Breakspear M. Graph analysis of the human connectome: promise, progress, and pitfalls. Neuroimage 2013; 80: 426–44. Greenberg SM, O’Donnell HC, Schaefer PW, Kraft E. MRI detection of new hemorrhages: potential marker of progression in cerebral amyloid angiopathy. Neurology 1999; 53: 1135–8. Greenberg SM, Al-Shahi Salman R, Biessels GJ, van Buchem M, Cordonnier C, Lee JM, et al. Outcome markers for clinical trials in cerebral amyloid angiopathy. Lancet Neurol 2014; 13: 419–28. Greenberg SM, Grabowski T, Gurol ME, Skehan ME, Nandigam RN, Becker JA, et al. Detection of isolated cerebrovascular beta-amyloid with pittsburgh compound B. Ann Neurol 2008; 64: 587–91. Gregoire SM, Charidimou A, Gadapa N, Dolan E, Antoun N, Peeters A, et al. Acute ischaemic brain lesions in intracerebral haemorrhage: multicentre cross-sectional magnetic resonance imaging study. Brain 2011; 134: 2376–86. Gurol ME, Irizarry MC, Smith EE, Raju S, Diaz-Arrastia R, Bottiglieri T, et al. Plasma beta-amyloid and white matter lesions in AD, MCI, and cerebral amyloid angiopathy. Neurology 2006; 66: 23–9. Gurol ME, Dierksen G, Betensky R, Gidicsin C, Halpin A, Becker A, et al. Predicting sites of new hemorrhage with amyloid imaging in cerebral amyloid angiopathy. Neurology 2012; 79: 320–6. Gurol ME, Viswanathan A, Gidicsin C, Hedden T, MartinezRamirez S, Dumas A, et al. Cerebral amyloid angiopathy burden associated with leukoaraiosis: a positron emission tomography/magnetic resonance imaging study. Ann Neurol 2013; 73: 529–36. 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: 2220–41. Jeurissen B, Leemans A, Jones DK, Tournier JD, Sijbers J. Probabilistic fiber tracking using the residual bootstrap with constrained spherical deconvolution. Hum Brain Mapp 2011; 32: 461–79. Jeurissen B, Leemans A, Tournier JD, Jones DK, Sijbers J. Investigating the prevalence of complex fiber configurations in white matter tissue with diffusion magnetic resonance imaging. Hum Brain Mapp 2012; 34: 2747–66. Johnson KA, Gregas M, Becker JA, Kinnecom C, Salat DH, Moran EK, et al. Imaging of amyloid burden and distribution in cerebral amyloid angiopathy. Ann Neurol 2007; 62: 229–34. Kloppenborg RP, Nederkoorn PJ, Geerlings MI, van den Berg E. Presence and progression of white matter hyperintensities and cognition: a meta-analysis. Neurology 2014; 82: 2127–38. Knudsen KA, Rosand J, Karluk D, Greenberg SM. Clinical diagnosis of cerebral amyloid angiopathy: validation of the boston criteria. Neurology 2001; 56: 537–9. Lawrence AJ, Chung AW, Morris RG, Markus HS, Barrick TR. Structural network efficiency is associated with cognitive impairment in small-vessel disease. Neurology 2014; 83: 304–11. Leemans A, Jones DK. The B-matrix must be rotated when correcting for subject motion in DTI data. Magn Reson Med 2009; 61: 1336–49. Lo CY, Wang PN, Chou KH, Wang J, He Y, Lin CP. Diffusion tensor tractography reveals abnormal topological organization in structural cortical networks in Alzheimer’s disease. J Neurosci 2010; 30: 16876–85. Martinez-Ramirez S, Pontes-Neto OM, Dumas AP, Auriel E, Halpin A, Quimby M, et al. Topography of dilated perivascular spaces in subjects from a memory clinic cohort. Neurology 2013; 80: 1551–6.

BRAIN 2014: Page 9 of 10


| BRAIN 2014: Page 10 of 10

van der Vlies AE, Goos JD, Barkhof F, Scheltens P, van der Flier WM. Microbleeds do not affect rate of cognitive decline in Alzheimer disease. Neurology 2012; 79: 763–9. Viswanathan A, Sudarsky L. Balance and gait problems in the elderly. Handb Clin Neurol 2012; 103: 623–34. Viswanathan A, Greenberg SM. Cerebral amyloid angiopathy in the elderly. Ann Neurol 2011; 70: 871–80. Wardlaw JM, Smith EE, Biessels GJ, Cordonnier C, Fazekas F, Frayne R, et al. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 2013; 12: 822–38.

Y. D. Reijmer et al. Wechsler D. Wechsler adult intelligence scale - III. San Antonio, TX: Psychological Corporation; 1997. Wechsler D. Wechsler memory scale-revised manual. San Antonio, TX: The Psychological Corporation; 1987. Wen W, Zhu W, He Y, Kochan NA, Reppermund S, Slavin MJ, et al. Discrete neuroanatomical networks are associated with specific cognitive abilities in old age. J Neurosci 2011; 31: 1204–12. Zhu YC, Chabriat H, Godin O, Dufouil C, Rosand J, Greenberg SM, et al. Distribution of white matter hyperintensity in cerebral hemorrhage and healthy aging. J Neurol 2012; 259: 530–6.

Downloaded from by guest on November 8, 2014

Structural network alterations and neurological dysfunction in cerebral amyloid angiopathy.

Cerebral amyloid angiopathy is a common form of small-vessel disease and an important risk factor for cognitive impairment. The mechanisms linking sma...
669KB Sizes 0 Downloads 13 Views