Sage Journals: Discover world-class research

Abstract

Cognitive impairment is an inevitable feature of cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), affecting executive function, attention and processing speed from an early stage. Impairment is associated with structural markers such as lacunes, but associations with functional connectivity have not yet been reported. Twenty-two adults with genetically-confirmed CADASIL (11 male; aged 49.8 ± 11.2 years) underwent functional magnetic resonance imaging at rest. Intrinsic attentional/executive networks were identified using group independent components analysis. A linear regression model tested voxel-wise associations between cognitive measures and component spatial maps, and Pearson correlations were performed with mean intra-component connectivity z-scores. Two frontoparietal components were associated with cognitive performance. Voxel-wise analyses showed an association between one component cluster and processing speed (left middle temporal gyrus; peak −48, −18, −14; Z_E = 5.65, p_FWE_corr = 0.001). Mean connectivity in both components correlated with processing speed (r = 0.45, p = 0.043; r = 0.56, p = 0.008). Mean connectivity in one component correlated with faster Trailmaking B minus A time (r = −0.77, p < 0.001) and better executive performance (r = 0.56, p = 0.011). This preliminary study provides evidence for associations between cognitive performance and attentional network connectivity in CADASIL. Functional connectivity may be a useful biomarker of cognitive performance in this population.

Keywords

CADASIL cognition functional magnetic resonance imaging resting state small vessel disease

Introduction

Cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), the most common monogenic cerebral small vessel disease (SVD), is a systemic vasculopathy caused by mutations of the NOTCH3 gene, which encodes a receptor expressed in vascular smooth muscle cells and in pericytes of brain arteries and capillaries.¹ The disorder manifests clinically as a diffuse leukoencephalopathy with liability to recurrent stroke and mood disturbance. CADASIL is associated with cognitive impairment, principally in the domains of processing speed, executive function and attention.^2–4 Cognitive deficits in CADASIL have been shown to correlate with structural brain measures including lacunes and atrophy,^5–8 echoing findings in patients with sporadic SVD.^9–11 White matter tract integrity on diffusion tensor imaging is associated with cognitive function in CADASIL¹² and other diffuse white matter pathologies including sporadic SVD and multiple sclerosis (MS),^11,13–15 and it has been proposed that this reflects ‘disconnection’ of cognitive networks.^13,16

Speed, attention and executive function depend on subcortical-frontal and frontoparietal networks.^17,18 With the advent of connectivity analysis methods being applied to resting state functional magnetic resonance imaging (fMRI) data, it has been possible to delineate a number of intrinsic functional networks that are found consistently in healthy subjects at rest.^19,20 Connectivity within these networks correlates with behavioural measures obtained outside the scanning environment.^21–23

There are, however, few studies of functional connectivity and cognition in patients with white matter disease. Functional connectivity has potential to be an informative biomarker for cognitive function in CADASIL, since structural MRI changes evolve slowly.^24–26 CADASIL may also serve as a model to understand the relationship between cerebrovascular pathology and cognitive impairment: in patients with sporadic SVD, this association is likely to be confounded by age and comorbid cerebral pathology (e.g. of Alzheimer’s type),²⁷ whereas patients with CADASIL show cognitive impairment in middle adulthood even in the absence of other risk factors. Research in the CADASIL population may enable clearer conclusions to be drawn about the direct effects of SVD on cognitive function, thereby elucidating causal mechanisms and potential targets for prevention or treatment.

Our research questions were:

In patients with CADASIL, is it possible to identify resting state functional networks that correspond to those associated with attention and executive function in the healthy population?

Is there an association between functional connectivity in these networks and performance on measures of speed, attention and executive function obtained outside the scanner?

Materials and methods

Participants

Participants were adults with CADASIL who were taking part in an ongoing 3-year longitudinal study of cerebral and peripheral haemodynamics (University of Glasgow, Scotland; UKCRN reference 13794). Inclusion criteria were (a) 18 years of age or older, (b) genetic diagnosis of CADASIL and (c) able to give informed consent. Exclusion criteria were (a) co-existent severe untreated depression or degenerative neurological disease unrelated to CADASIL, (b) contraindications to MRI and (c) treatment with angiotensin-converting enzyme inhibitors or calcium channel blockers (due to vasoreactivity studies in the protocol). The study was conducted in accordance with the Declaration of Helsinki and was approved by the West of Scotland Research Ethics Service (reference WS/12/0295). All participants gave written informed consent.

Clinical assessment

Clinical information, including medical history, history of stroke, cardiovascular risk factors and medication, was recorded on a standardised case report form. Structured neurological examination in the form of the National Institutes of Health Stroke Scale²⁸ was conducted by a clinical research fellow in neurology (F.C.M.), along with an assessment of disability (modified Rankin scale)²⁹ and the Hospital Anxiety and Depression Scale (HADS; GL Assessment Ltd, London).³⁰

Neuropsychological assessment was carried out over one or two sessions by a specialist clinical neuropsychologist (B.C.), and included standardised tests of estimated premorbid ability, global cognitive function, processing speed, verbal and visual attention and working memory, verbal and visual episodic memory, visuospatial function, verbal fluency, verbal reasoning and executive function. Because the present study was focused on speed, attention/working memory and executive function, these domains were chosen a priori for analysis. Composite domain scores were derived from individual tests by converting raw scores to standardised scores (corrected for age, or age and education, with reference to published normative tables) and calculating the mean of scores within a particular cognitive domain. Higher composite scores represent better performance.

The processing speed composite score was the mean of z-scores on the Symbol-Digit Modalities Test (SDMT, written version)³¹ and Trailmaking Test Part A.³² The attention/working memory composite score was the mean of scale scores on the forward and backward Digit Span sub-tests from the Wechsler Adult Intelligence Scale (WAIS-IV)³³ and the Symbol Span sub-test from the Wechsler Memory Scale (WMS-IV).³⁴ The executive function composite score was the mean of z-scores on the Similarities sub-test from the WAIS-IV,³³ FAS letter fluency test,³⁵ Trailmaking Test Part B³² and Stroop Neuropsychological Screening Test.³⁶ Trailmaking B minus A time was also analysed separately (measured in seconds, with higher values indicating worse performance), as this has been shown to correlate with frontoparietal network connectivity in the healthy population.²³

MRI protocol and quantitative measures

Participants underwent MRI on a 3T Signa scanner (GE Healthcare; Waukesha, WI, USA) within several weeks of the neuropsychological assessment (mean interval = 45 days, SD = 30).

A 10-minute T2*-weighted EPI BOLD scan was acquired at rest (repetition time = 2000 ms, echo time = 30 ms, flip angle = 78°, slice thickness = 3.5 mm, interslice gap = 1.0 mm, matrix = 64 × 64 × 40; FOV =224 × 224 × 179 mm³), with participants instructed to close their eyes and stay awake.

Structural scans performed as part of the main study protocol were used to quantify lesion load and brain volume. These included a three-dimensional T1-weighted BRAVO sequence (repetition time = 9000 ms, echo time = 3.6 ms, inversion time = 450 ms, flip angle = 12°, matrix = 320 × 320 × 156, FOV = 240 ×240 × 156 mm³; acquisition time = 4 min 28 s) and fluid-attenuated inversion recovery sequence (T2-FLAIR; repetition time = 10 s, echo time = 140 ms, inversion time = 2250 ms, flip angle 90°, slice thickness = 5 mm, interslice gap = 1.5 mm, matrix 384 ×256, FOV = 240 × 240 mm²; acquisition time = 3 min 20 s).

Lacunes were defined radiologically as hypointense areas between 3 and 15 mm on T1-weighted and FLAIR sequences.³⁷ Enlarged perivascular spaces, identified as fluid-filled cavities which were dot-like (<3 mm) when perpendicular and linear when parallel to the imaging plane, were excluded.³⁸ Lacunes were identified and their volume calculated on T1-weighted images using 3D thresholding tools (Analyze v11.0; http://analyzedirect.com/); this was performed by one rater (F.C.M.) on two occasions (intraclass correlation (ICC) = 0.94; 95% confidence interval (CI) 0.85, 0.97) and the second rating was used in the analysis. Lacune volume was normalised to intracranial cavity volume (nLV, %), which was calculated using tools from FMRIB Software Library (FSL v5.0; http://fsl.fmrib.ox.ac.uk/fsl/).³⁹ Signal abnormalities in white matter, grey matter and brainstem showing hyperintensity on FLAIR were termed subcortical hyperintensities (SH). A skull-stripped FLAIR image was created,^39–41 and the mode intensity of this image, multiplied by 1.3, was used to threshold the image, with manual editing if required⁴² to create SH maps. This was performed by one rater (F.C.M.); reliability of repeat measurements on a sub-sample of five scans was high (ICC = 0.99; 95% CI 0.99, 1.00). SH volume was normalised (nSH, %) to parenchymal volume. Normalised brain volume (nBV), normalised for subject head size, was calculated on T1-weighted images using SIENAX,⁴³ part of FSL.³⁹

Independent components analysis

The resting state fMRI scans were pre-processed using SPM8 (http://www.fil.ion.ucl.ac.uk/spm/). For each participant, the functional images were realigned, co-registered to the participant’s 3D T1-weighted image, segmented and spatially normalised to 2 mm cubic voxels in MNI space, and smoothed with an 8 mm full-width at half-maximum Gaussian kernel.

Spatial independent components analysis (ICA) was then performed on the group data; this identifies temporally coherent functional networks by decomposing linearly mixed signal timecourses into maximally independent spatial sources. As a purely data-driven method, ICA does not depend on prior knowledge about expected signal timecourse⁴⁴ and it avoids the need for a priori selection of regions of interest, as required in seed-based methods.⁴⁵ Group ICA was conducted using the GIFT toolbox (http://mialab.mrn.org/software/gift/); further details are provided in the Supplementary Methods. This analysis produced individual subject spatial maps and timecourses for each independent component (IC), as well as mean maps and timecourses for the whole sample.

For identification, the mean IC maps were visually compared with components identified in a previous study by Allen et al. of a large sample of healthy subjects,⁴⁶ to identify components of interest within attentional and executive networks. This was done by two authors independently (B.C. and F.C.M.), followed by consensus discussion with a third author (R.K.). Spatial correlations were also computed in GIFT between our components of interest and unthresholded t-maps of the corresponding components from the Allen et al. study⁴⁶ to confirm the visual identifications.

Voxel-wise analyses

The individual subject spatial maps for each component of interest were further analysed using SPM8. Each voxel in the spatial map has a z-score value, which represents the strength of that voxel’s contribution to the timecourse of the component; voxel values can be taken as indicating the strength of connectivity within a particular component. Voxel-wise regression analyses were performed between each component map of interest (unthresholded) and each cognitive score (processing speed, attention/working memory, executive function and Trailmaking B minus A). Although the SPM8 regression model assigns the spatial map data as the dependent variable, bivariate analyses can be interpreted as correlations. The threshold for statistical significance was p < 0.05 with family-wise error (FWE) correction and a minimum extent of five contiguous voxels.

Region of interest analyses

For each attentional/executive component of interest, a one-sample t-test was carried out in SPM8 to determine the threshold that was statistically significant (FWE-corrected p < 0.05) across all individual subject component maps. For all voxels that were above this threshold, the mean voxel z-score value was then computed for each subject, using a program implemented in MATLAB v2014a (http://uk.mathworks.com/products/matlab/). This provided a single overall measure of connectivity within a particular component for each participant, allowing us to treat the component as a region of interest (ROI) in subsequent analyses. Pearson correlations were computed between these mean connectivity scores and each cognitive measure. Correlations were also computed among the structural MRI measures (nLV, nSH, nBV), mean connectivity scores and cognitive scores. The threshold for statistical significance was p < 0.05. These analyses were conducted in Stata 13 (http://www.stata.com/stata13/). Exploratory mediation analyses were conducted to analyse relationships between structural, functional and cognitive measures, using the PROCESS macro (http://www.processmacro.org/) for SPSS v22 (https://www-01.ibm.com/software/analytics/spss/products/statistics/features.html); further details are provided in the Supplementary Methods.

Effects of motion artefact

Framewise displacement (FD)⁴⁷ was calculated using a program implemented in MATLAB (http://becs.aalto.fi/∼eglerean/bramila.html). FD is a time series of the sum of the absolute values of the derivatives of the six realignment parameters that were generated during fMRI image pre-processing, and provides an overall indication of head motion during scanning for each participant. No participant had a mean FD value exceeding 0.25. No significant correlations were observed between mean FD and any of the cognitive measures, or between mean FD and mean connectivity values in the components of interest. Motion parameters were therefore not included in analyses involving the mean connectivity values (i.e. ROI analyses). For some individual participants, significant correlations were seen between the FD time series and the timecourse of some components of interest. Significant voxel-wise analyses were therefore repeated with mean FD included as a regressor of no interest.

Results

Sample characteristics

Participants were 22 adults with CADASIL, from 19 pedigrees. Table 1 summarises their demographic and clinical characteristics. Eleven had a history of stroke or transient ischaemic attack, but most participants had no or minor disability. Median anxiety and depression symptom scores were in the normal range. Two participants were unable to complete some of the cognitive tests, leading to missing data on the speed, executive and Trailmaking B minus A scores (n with missing data is given in Table 1). All participants completed the MRI protocol.

Table 1.

Demographic and clinical characteristics of the sample.

	Participants with CADASIL (n = 22)
Age, mean (SD), years	49.8 (11.2)
Sex, male n (%)	11 (50.0)
Education, median (25th, 75th percentile), years	12 (11, 15)
NIHSS score, median (25th, 75th percentile)	0 (0, 0)
Modified Rankin Scale score, n (%)
0 No symptoms	14 (63.6)
1 No significant disability	1 (4.6)
2 Slight disability	4 (18.2)
3 Moderate disability	3 (13.6)
Previous stroke or TIA, yes n (%)	11 (50.0)
Age at first stroke or TIA, mean (SD), years	47.3 (9.3)
Normalised lacune volume, mean (SD), %	0.04 (0.05)
Normalised subcortical hyperintensity volume, mean (SD), %	6.92 (4.39)
Normalised brain volume, mean (SD), 10⁶ mm³	1.54 (0.08)
HADS Anxiety, median (25th, 75th percentile)^a	6.0 (4.5, 14.0)
HADS Depression, median (25th, 75th percentile)^a	5.5 (2.0, 10.0)
ToPF estimated IQ, mean (SD)	97.2 (12.7)
Processing speed, mean (SD), composite z-score^b	−0.5 (1.7)
Attention/Working memory, mean (SD), composite scale score^c	9.3 (2.7)
Executive function, mean (SD), composite z-score^a	−0.7 (1.7)
Trailmaking B minus A, median (25th, 75th percentile), seconds^b	35 (26, 76)

n = 2 missing.

n = 1 missing.

Standard scale with mean = 10 and SD = 3.

CADASIL: cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy; HADS: Hospital Anxiety and Depression Scale; IQ: intelligence quotient; NIHSS: National Institutes of Health Stroke Scale; SD: standard deviation; TIA: transient ischaemic attack; ToPF: Test of Premorbid Functioning.

Since cognitive scores were already adjusted for education and/or age, and scores did not differ by sex, no statistical adjustment for these variables was necessary in the analyses reported below.

Independent components analysis

Based on estimation from the data of the first participant, the group ICA was specified to produce 38 components. Visual identifications were concordant between two independent raters for 32 of these, with reference to components from the Allen et al. study.⁴⁶ Four components belonged to Allen et al.’s attentional or frontal/executive component groups. These were IC 2 (right frontoparietal; Allen component 60), IC 4 (left frontoparietal; Allen component 34), IC 25 (bilateral frontoparietal; Allen component 72) and IC 31 (bilateral dorsolateral prefrontal; Allen component 47). Spatial correlation coefficients between these components and the corresponding Allen et al.⁴⁶ t-maps were high: r values were 0.42 to 0.61, exceeding the threshold of 0.4 that has previously been applied in research using the same reference maps.²¹ Correlation coefficients were computed between our four components of interest and all of the Allen et al. t-maps, and in each case, the highest coefficient value was with the Allen et al. t-map that had been identified a priori as the closest visual match. Figure 1 shows spatial maps of the four components of interest and Table 2 lists their peak activations.

Table 2.

Peak activations of component spatial maps.

	BA	Voxel extent	Z _E	Coordinates
IC 2
R middle frontal gyrus	8	5021	7.2	32, 16, 44
R inferior parietal lobule	40	2482	7.4	42, −56, 44
R precuneus	7	335	6.1	6, −66, 30
L cerebellum posterior lobe		223	6.6	−38, −68, −32
R middle temporal gyrus	21	219	6.2	68, −40, −4
L inferior parietal lobule	40	85	5.8	−52, −60, 42
L middle frontal gyrus		34	5.2	−42, 18, 36
IC 4
L angular gyrus	39	2784	7.7	−40, −64, 32
L precuneus		545	6.2	−6, −58, 34
L middle temporal gyrus	21	460	7.0	−60, −34, −8
R superior temporal gyrus	39	429	6.4	56, −64, 28
L middle frontal gyrus	8	387	5.8	−28, 20, 48
L middle frontal gyrus		319	6.3	−38, 56, −8
R middle temporal gyrus		45	5.9	60, −36, −8
R middle frontal gyrus		29	5.9	40, 52, −8
L superior parietal lobule		22	5.1	−10, −66, 58
IC 25
Bilateral precuneus	7	4431	6.6	6, −76, 42
R middle temporal gyrus		263	6.2	40, −74, 24
R superior frontal gyrus	6	235	6.1	22, 8, 60
Bilateral lingual gyrus		229	6.1	0, −76, −12
L middle frontal gyrus	6	199	6.2	−26, 4, 60
L middle frontal gyrus		161	5.6	−30, 32, 30
R superior frontal gyrus		79	6.1	30, 34, 34
L angular gyrus	39	13	5.4	−40, −78, 28
IC 31
L inferior frontal gyrus		5030	7.8	−52, 14, 20
R inferior frontal gyrus		2543	7.2	48, 28, 12
Bilateral posterior cingulate	29	121	5.8	−4, −54, 8
L fusiform gyrus	37	103	6.0	−38, −46, −24
L inferior parietal lobule		57	5.5	−42, −46, 42
L middle temporal gyrus	37	24	5.4	−56, −66, 0
L medial frontal gyrus		21	5.4	−8, 6, 56
L middle temporal gyrus	37	15	5.3	−60, −48, −12

Coordinates are given in Montreal Neurological Institute space. Results are taken from one-sample t-tests in SPM8.

BA: Brodmann area; IC: independent component; L: left hemisphere; R: right hemisphere; Z_E: peak equivalent Z.

Figure 1.

Independent component spatial maps. Images are t-maps, thresholded at the statistically significant value from the one-sample t-test.

Voxel-wise analyses

Regression analyses of subject spatial maps for each component with each cognitive measure yielded no significant associations for IC 2, IC 25 or IC 31. IC 4 connectivity was associated with faster processing speed (left middle temporal gyrus; peak coordinates −48, −18, −14; Z_E = 5.65, p_FWE_corr = 0.001), and this result remained significant after motion (mean FD) was added to the model. IC4 connectivity was also associated with lower Trailmaking B minus A time (right inferior parietal lobule; peak coordinates 56, −42, 24; Z_E = 4.95, p_FWE_corr = 0.041), but this was no longer significant when motion was included in the model.

Region of interest analyses

Correlations between each of the four component mean z-scores and each of the four cognitive measures were computed. IC 2 and IC 31 did not correlate significantly with any cognitive measure. As shown in Table 3 and Figure 2, IC 4 connectivity was significantly associated with faster processing speed; IC 25 connectivity was significantly associated with faster processing speed, better executive function, and lower Trailmaking B minus A time. Significant correlations were also evident between structural measures and cognitive performance (see Table 3).

Table 3.

Correlations between structural, connectivity and cognitive measures.

	nLV	nSH	nBV	IC 2	IC 4	IC 25	IC 31	Processing speed	Attention/ working memory	Executive function	Trailmaking B–A
nLV	–
nSH	0.313	–
	0.156
nBV	−0.447	0.073	–
	0.037	0.748
IC 2	−0.242	−0.301	0.174	–
	0.278	0.173	0.439
IC 4	−0.384	−0.176	0.401	0.631	–
	0.078	0.433	0.065	0.002
IC 25	−0.378	−0.313	0.533	0.499	0.679	–
	0.083	0.156	0.011	0.018	0.001
IC 31	−0.038	−0.386	0.000	0.267	0.586	0.418	–
	0.867	0.076	0.999	0.230	0.004	0.053
Processing speed	−0.559	0.053	0.558	0.378	0.446	0.564	0.096	–
	0.008	0.820	0.009	0.091	0.043	0.008	0.679
Attention/Working memory	−0.243	0.073	0.194	0.291	0.312	0.337	0.279	0.710	–
	0.276	0.747	0.388	0.189	0.157	0.125	0.209	<0.001
Executive function	−0.409	−0.116	0.512	0.261	0.173	0.557	0.012	0.847	0.776	–
	0.073	0.625	0.021	0.266	0.467	0.011	0.961	<0.001	<0.001
Trailmaking B–A	0.421	0.122	−0.428	−0.360	−0.347	−0.773	−0.156	−0.791	−0.621	−0.908	–
	0.058	0.599	0.053	0.109	0.123	<0.001	0.500	<0.001	0.003	<0.001

Values shown are Pearson r and p value.

IC: independent component; nBV: normalised brain volume; nLV: normalised lacune volume; nSH: normalised subcortical hyperintensity volume.

Figure 2.

Significant correlations between mean connectivity z-scores and cognitive measures. (a) IC 4 and processing speed; r = 0.45, n = 21, p = 0.043. (b) IC 25 and processing speed; r = 0.56, n = 21, p = 0.008. (c) IC 25 and executive function; r = 0.56, n = 20, p = 0.011. (d) IC 25 and Trailmaking B minus A time; r = − 0.77, n = 21, p < 0.001.

Mediation analysis was used to explore the nature of these relationships, by modelling a path between the structural measure and cognitive performance, with functional connectivity as the intermediate variable. In light of the small sample size, these analyses were exploratory in nature. One model provided evidence of a mediation effect: the association between normalised lacune volume and Trailmaking B minus A time was attenuated after controlling for IC 25 connectivity (Figure 3).

Figure 3.

Association between normalised lacune volume and Trailmaking B minus A time, as mediated by IC 25 mean connectivity. Coefficients are standardised beta values. Coefficient c is the total effect of normalised lacune volume on Trailmaking B minus A time, and c’ is the direct effect after controlling for IC 25 mean connectivity. The indirect effect (a*b) is 0.33 (bootstrapped standard error = 0.23).

Discussion

This is the first study to investigate the relationship between resting state functional connectivity and cognitive performance in CADASIL. Associations between cognitive measures and frontoparietal network connectivity in our sample were consistent with previous studies in healthy adults and those with sporadic SVD. Seeley et al.²³ showed that in healthy adults, connectivity within a frontoparietal network at rest was significantly associated with Trailmaking test performance outside the scanner, and Reineberg et al.²² demonstrated distinct patterns of connectivity associated with overall and set-shifting aspects of executive function. Research in patients with cerebral SVD has indicated that reduced functional connectivity in frontoparietal and cingulate regions correlates with neuropsychological dysfunction.^48,49 Our findings demonstrate that intrinsic frontoparietal functional networks are also detectable in patients with CADASIL and are associated with similar cognitive measures, particularly tests of speed and executive ability.

We also found that structural brain changes were correlated with cognitive performance. In line with the previous literature, greater lacune volume and smaller brain volume were associated with impaired processing speed and executive function in our sample. It has been postulated that ‘disconnection’ underlies the cognitive impairment commonly observed in diffuse white matter disease,^13,16 and we would suggest that reduced functional connectivity can be conceptualised as an intermediate stage on the causal pathway between structural lesion development and cognitive dysfunction. Our study lacked power to investigate these relationships robustly, but we did find preliminary support for a mediating effect of connectivity in a bilateral frontoparietal network (IC 25) on the association between lacune volume and Trailmaking performance. Modelling the proposed pathway from structural pathology to functional disconnection to cognitive impairment will be an important goal for future studies. Our results indicate that informative measures can be derived for each stage of the pathway, and the addition of other intermediate measures such as diffusion tensor imaging and tractography will allow the relationship between structural and functional connectivity to be better delineated.

We note that the results of our voxel-wise and ROI analyses differed from each other in some respects. This may be a consequence of the thresholds applied: the mean component z-scores for the ROI analyses were thresholded with the one-sample t-test result (i.e. strictly defined), whereas the whole-brain voxel-wise analyses were not. Relatively weaker associations on ROI analysis may indicate that effects are confined to certain small clusters only, and hence are only detectable in voxel-wise analyses. Conversely, relatively stronger results on ROI analysis (as seen for IC 25) may suggest a widely dispersed effect across the component but without attaining significance at the required level in any single cluster. We also note that confounding effects of motion artefact need to be considered, particularly in research with clinical populations. Motion may be associated with clinical indices (e.g. if patients who are more disabled are less able to tolerate scanning). This may potentially confound associations between functional MRI data and outcome measures of interest, even when data have been collected at different times, as was the case in the present study.

There were a number of limitations to this study. The absence of a healthy comparison group meant we were unable to assess whether functional connectivity measures were abnormally reduced in the patient sample. The small sample size reduced the power to detect significant effects, and precluded robust multivariate analyses. Multiple statistical tests were conducted between the various structural, functional and cognitive measures, although we aimed to reduce the likelihood of spurious results by choosing cognitive scores a priori and limiting the analyses to only those networks that clearly matched attention or executive networks reported in a previous large study.⁴⁶ Reduced dexterity may have affected participants’ performance on the processing speed tests—thereby reflecting physical rather than cognitive dysfunction—but written tasks were not administered to participants who had clear problems using a pen. The use of a z-score transformation to produce composite cognitive scores has limitations where the original score distribution was skewed. The MRI and neuropsychological assessments were conducted several weeks apart, but progression of CADASIL over such a short interval is highly unlikely in the absence of clinical stroke or hospital admission, which affected none of our patients between the two visits. Our findings will not necessarily be replicable in other CADASIL patients, e.g. those with more severe structural brain pathology where ICA methods may not be successful.

Future studies in larger samples will be able to use multivariate methods to model the proposed causal pathway from structural integrity to functional connectivity to cognitive outcome. It will also be possible to investigate changes in these measures longitudinally, which we aim to do by following up this sample annually. If future research shows functional connectivity to be predictive of cognitive status, then resting state connectivity may be a useful biomarker of neuropsychological outcome in CADASIL; not all patients with CADASIL will be capable of undertaking cognitive assessment, but resting state fMRI could be a quick and potentially informative way to measure the integrity of distributed brain systems that underpin cognitive performance.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the Scottish Executive Health Department (Chief Scientist Office), reference ETM/244.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors' contributions

Study conception and design: Cullen, Moreton, Krishnadas, López-González, Santosh, Schwarzbauer, Muir; acquisition of data: Cullen, Moreton, López-González, Santosh; analysis and interpretation of data: Cullen, Moreton, Stringer, Krishnadas, Kalladka, Schwarzbauer, Muir; drafting the article and revising it critically for important intellectual content: all authors; final approval of the version to be published: all authors.

Supplementary material

Supplementary material for this paper can be found at

References

Chabriat

Joutel

Dichgans

. CADASIL. Lancet Neurol 2009; 8: 643–653.

Buffon

Porcher

Hernandez

. Cognitive profile in CADASIL. J Neurol Neurosurg Psychiatry 2006; 77: 175–180.

Dichgans

. Cognition in CADASIL. Stroke 2009; 40: S45–S47.

Peters

Opherk

Danek

. The pattern of cognitive performance in CADASIL: A monogenic condition leading to subcortical ischemic vascular dementia. Am J Psychiatry 2005; 162: 2078–2085.

Duering

Gonik

Malik

. Identification of a strategic brain network underlying processing speed deficits in vascular cognitive impairment. Neuroimage 2013; 66: 177–183.

Lee

Choi

Kang

. Effects of lacunar infarctions on cognitive impairment in patients with cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy. J Clin Neurol 2011; 7: 210–214.

Liem

Oberstein

SAJL

Haan

. MRI correlates of cognitive decline in CADASIL A 7-year follow-up study. Neurology 2009; 72: 143–148.

Zieren

Duering

Peters

. Education modifies the relation of vascular pathology to cognitive function: cognitive reserve in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy. Neurobiol Aging 2013; 34: 400–407.

Benjamin

Lawrence

Lambert

. Strategic lacunes and their relationship to cognitive impairment in cerebral small vessel disease. Neuroimage-Clin 2014; 4: 828–837.

10.

Biesbroek

Kuijf

van der Graaf

. Association between subcortical vascular lesion location and cognition: a voxel-based and tract-based lesion-symptom mapping study: The SMART-MR study. Plos One 2013; 8: e60541–e60541.

11.

Lawrence

Patel

Morris

. Mechanisms of cognitive impairment in cerebral small vessel disease: multimodal MRI results from the St George’s Cognition and Neuroimaging in Stroke (SCANS) study. Plos One 2013; 8: e61014–e61014.

12.

O’Sullivan

Barrick

Morris

. Damage within a network of white matter regions underlies executive dysfunction in CADASIL. Neurology 2005; 65: 1584–1590.

13.

Lawrence

Chung

Morris

. Structural network efficiency is associated with cognitive impairment in small-vessel disease. Neurology 2014; 83: 304–311.

14.

Mesaros

Rocca

Kacar

. Diffusion tensor MRI tractography and cognitive impairment in multiple sclerosis. Neurology 2012; 78: 969–975.

15.

Patel

Markus

. Magnetic resonance imaging in cerebral small vessel disease and its use as a surrogate disease marker. Int J Stroke 2011; 6: 47–59.

16.

Kleinschmidt

Vuilleumier

. Disconnecting cognition. Curr Opin Neurol 2013; 26: 333–338.

17.

Petersen

Posner

. The attention system of the human brain: 20 years after. Annu Rev Neurosci 2012; 35: 73–89.

18.

Stuss

. Frontal lobes and attention: Processes and networks, fractionation and integration. J Int Neuropsych Soc 2006; 12: 261–271.

19.

Damoiseaux

Rombouts

SARB

Barkhof

. Consistent resting-state networks across healthy subjects. Proc Natl Acad Sci USA 2006; 103: 13848–13853.

20.

Deco

Corbetta

. The dynamical balance of the brain at rest. Neuroscientist 2011; 17: 107–123.

21.

Meier

Wildenberg

Liu

. Parallel ICA identifies sub-components of resting state networks that covary with behavioral indices. Front Hum Neurosci 2012; 6: 1–14.

22.

Reineberg

Andrews-Hanna

Depue

. Resting-state networks predict individual differences in common and specific aspects of executive function. Neuroimage 2015; 104: 69–78.

23.

Seeley

Menon

Schatzberg

. Dissociable intrinsic connectivity networks for salience processing and executive control. J Neurosci 2007; 27: 2349–2356.

24.

Holtmannspotter

Peters

Opherk

. Diffusion magnetic resonance histograms as a surrogate marker and predictor of disease progression in CADASIL – A two-year follow-up study. Stroke 2005; 36: 2559–2565.

25.

Liem

Oberstein

SAJL

Haan

. Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy: Progression of MR abnormalities in prospective 7-year follow-up study. Radiology 2008; 249: 964–971.

26.

Peters

Herzog

Opherk

. A two-year clinical follow-up study in 80 CADASIL subjects – Progression patterns and implications for clinical trials. Stroke 2004; 35: 1603–1608.

27.

Charlton

Morris

Nitkunan

. The cognitive profiles of CADASIL and sporadic small vessel disease. Neurology 2006; 66: 1523–1526.

28.

Brott

Adams

Olinger

. Measurements of acute cerebral infarction – A clinical examination scale. Stroke 1989; 20: 864–870.

29.

Rankin

. Cerebral vascular accidents in patients over the age of 60. II. Prognosis. Scottish Med J 1957; 2: 200–215.

30.

Zigmond

Snaith

. The Hospital Anxiety and Depression Scale. Acta Psychiatrica Scand 1983; 67: 361–370.

31.

Smith

. Symbol Digit Modalities Test (SDMT), Torrance, CA: WPS, 1982.

32.

Tombaugh

. Trail making Test A and B: Normative data stratified by age and education. Archiv Clin Neuropsychol 2004; 19: 203–214.

33.

Wechsler

. Wechsler Adult Intelligence Scale – Fourth UK edition (WAIS-IV UK), London: Pearson Assessment, 2011.

34.

Wechsler

. Wechsler Memory Scale – Fourth UK edition (WMS-IV UK), London: Pearson Assessment, 2011.

35.

Tombaugh

Kozak

Rees

. Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Archiv Clin Neuropsychol 1999; 14: 167–177.

36.

Trenerry

Crosson

DeBoe

. Stroop Neuropsychological Screening Test (SNST), Lutz, FL: PAR, Inc, 1989.

37.

Wardlaw

Smith

Biessels

. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 2013; 12: 822–838.

38.

Doubal

MacLullich

AMJ

Ferguson

. Enlarged perivascular spaces on MRI are a feature of cerebral small vessel disease. Stroke 2010; 41: 450–454.

39.

Smith

Jenkinson

Woolrich

. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004; 23: S208–S219.

40.

Smith

. Fast robust automated brain extraction. Human Brain Mapp 2002; 17: 143–155.

41.

Zhang

Brady

Smith

. Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm. IEEE Trans Med Imag 2001; 20: 45–57.

42.

Smart

Firbank

O’Brien

. Validation of automated white matter hyperintensity segmentation. J Aging Res 2011; 2011: 391783–391783.

43.

Smith

Zhang

Jenkinson

. Accurate, robust, and automated longitudinal and cross-sectional brain change analysis. Neuroimage 2002; 17: 479–489.

44.

McKeown

Hansen

Sejnowski

. Independent component analysis of functional MRI: what is signal and what is noise? Curr Opin Neurobiol 2003; 13: 620–629.

45.

Lee

Smyser

Shimony

. Resting-state fMRI: A review of methods and clinical applications. Am J Neuroradiol 2013; 34: 1866–1872.

46.

Allen

Erhardt

Damaraju

. A baseline for the multivariate comparison of resting-state networks. Front Syst Neurosci 2011; 5: 1–23.

47.

Power

Barnes

Snyder

. Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. Neuroimage 2012; 59: 2142–2154.

48.

Schaefer

Quinque

Kipping

. Early small vessel disease affects frontoparietal and cerebellar hubs in close correlation with clinical symptoms-a resting-state fMRI study. J Cerebr Blood Flow Metab 2014; 34: 1091–1095.

49.

Liang

Liu

. Disrupted topological organization of resting-state functional brain network in subcortical vascular mild cognitive impairment. CNS Neurosci Ther 2015; 21: 846–854.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB

Resting state connectivity and cognitive performance in adults with cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy

Abstract

Keywords

Introduction

Materials and methods

Participants

Clinical assessment

MRI protocol and quantitative measures

Independent components analysis

Voxel-wise analyses

Region of interest analyses

Effects of motion artefact

Results

Sample characteristics

Independent components analysis

Voxel-wise analyses

Region of interest analyses

Discussion

Footnotes

Funding

Declaration of conflicting interests

Authors' contributions

Supplementary material

References

Supplementary Material