Frequency of cognitive impairment in multiple sclerosis english.pdf · answers to the following...

The objective of this thesis was to better understand how the accumulation of different types of

pathology during the course of multiple sclerosis ultimately drive cognitive decline. We aimed to find

answers to the following questions:

• How does the accumulation of grey matter atrophy relate to changes in brain function,

protective capacity and, ultimately, cognitive decline during the course of multiple sclerosis?

• Can the use of a network approach enable a further unravelling of the underlying mechanisms

of cognitive decline in multiple sclerosis?

• What are the main MR imaging predictors of future cognitive decline and do these predictors

differ according to disease stage?

In the current section, our main findings regarding these questions will be discussed and suggestions

for future research will be provided.

Frequency of cognitive impairment in multiple sclerosis

As this thesis investigates determinants, mechanisms and predictors of cognitive decline in multiple

sclerosis, a first step was to find ways to best characterize “impairment” and “decline”. Multiple

sclerosis patients frequently suffer from cognitive deficits, which can have a severe impact on daily

functioning.3, 4 Estimations in literature of the prevalence of cognitive impairment in multiple sclerosis

range from 40-70%, depending on patient populations and types of neuropsychological tests as well

as the choice of threshold for “impairment”.3 To measure cognitive function in the current thesis, all

reported studies used the same extensive neuropsychological evaluation, based on an expanded Brief

Repeatable Battery of Neuropsychological tests.8 Executive functioning was assessed using the concept

shifting test,9 verbal memory using the selective reminding test,10 verbal fluency using the word list

generation test,12 information processing speed using the symbol digit modalities test,14 visuospatial

memory using the spatial recall test,12 attention using the Stroop colour-word test16 and working

memory using the memory comparison test.17 Using these cognitive tests, the 332 patients that were

part of the baseline measurement of our Amsterdam multiple sclerosis cohort showed a moderate

global impairment in cognitive function of Z = -0.80 compared to a group of 96 matched healthy

controls, at an average disease duration of 15 years. The most severely affected cognitive domain in

the entire patient group was information processing speed (Z = -1.18 compared to the healthy control

group), which is often reported to be one of the most common and worst affected cognitive domains

in multiple sclerosis.18, 19 Other severely affected cognitive domains in our cohort were working

memory (Z = -1.02 compared to the healthy control group) and executive function (Z = -1.18 compared

to the healthy control group).

Summary and general discussion

Nonetheless, there was considerable heterogeneity in cognitive performance in our patient

sample, as is common in multiple sclerosis. Therefore, for part of the studies performed in this thesis,

the patient group was subdivided into subgroups based on the severity of cognitive impairment, as

was done previously.20, 21. Patients who scored one and a half standard deviations (i.e., Z ≤ -1.5) below

the average of the healthy control group on at least two cognitive domains were classified as

cognitively impaired, classifying 46% of our sample (N=152) as impaired. In chapter 3, we further

subdivided this cognitively impaired patient group into a mildly cognitive impaired group (i.e., Z ≤ -1.5)

and a more severely cognitively impaired group (i.e., Z ≤ -2.0) to be able to analyse underlying

mechanisms in these groups separately. Using these criteria, 65 patients were labelled as mildly

cognitively impaired and 87 patients as severely cognitively impaired, percentages comparable to

previous literature.3, 22 While the neuropsychological tests used in the studies part of this thesis are

commonly applied in multiple sclerosis, there are still cognitive domains that are underrepresented,

like multi-tasking and theory of mind.19, 23 Another concern is that this battery is most likely too long

for clinical implementation. Therefore, future studies could consider using either the long but

potentially more sensitive Minimal Assessment of Cognitive Function In Multiple Sclerosis (MACFIMS)24

or the shorter but extensively validated Brief International Cognitive Assessment for Multiple Sclerosis

(BICAMS).25

Acceleration of cognitive decline during the course of multiple sclerosis

While cognitive impairment can already occur in early disease, the frequency tends to increase with

disease progression.3, 4, 22, 26 This was also seen in our patient sample at baseline, where 39% of

relapsing-remitting patients and 64% of progressive patients were cognitively impaired. Before the

initiation of this research, however, it was unclear whether the rate of cognitive decline also differed

between early and progressive stages of multiple sclerosis, or whether differences between these

groups were simply due to a longer average disease duration in progressive patients with a steady rate

of decline during the disease. For instance, the high prevalence of cognitive deficits that has been

reported in some studies during the clinically isolated syndrome (i.e., the very first stage of multiple

sclerosis) or around the time of clinically definite multiple sclerosis diagnosis has sometimes been

interpreted as an indication that cognitive decline could demonstrate an initial drop, followed by a

subsequent slowing of the rate of decline or even stabilization.22, 26-28 Since some of these (cross-

sectional) studies relied on relaxed cognitive impairment classification (e.g., one standard deviation on

one test)28 and since the duration between disease onset and appearance of first symptoms is

unclear,29, 30 these conclusions of an initial drop in cognitive function should be interpreted with

caution until longitudinally confirmed.

It is therefore essential that such research questions are addressed using longitudinal study

designs that preferably also include longitudinal control samples. We therefore addressed this

question in chapter 2.1 by looking at the longitudinal data of our aforementioned cohort, of whom

230 patients and 59 healthy controls had returned after five years with identical cognitive evaluations.

To assess cognitive decline in patients with multiple sclerosis during the follow-up period, one major

challenge was how to deal with normal aging and learning effects. We decided to deal with these

effects by computing the modified practice adjusted reliable change index31 for each cognitive domain

separately. This method corrects for practice effects based on longitudinal changes observed in the

healthy control group. Individual cognitive domain reliable change index scores were then divided by

each individual subject’s time interval, to remove effects of variability in follow-up duration, and

averaged across domains to obtain a yearly rate of cognitive decline. We observed that the yearly

cognitive decline rate was around three times higher in progressive compared to relapsing-remitting

patients, which indicates that cognitive impairment is not only more severe in progressive patients due

to a longer average disease duration, but also due to a higher yearly rate of cognitive decline.

The cognitive decline rates that we observed were highly similar between primary and

secondary progressive patients as shown in chapter 4.2. While an early study reported more severe

cognitive deficits in secondary compared to primary progressive patients,32 other studies have shown

similar levels of cognitive impairment, which suggests that observed differences could be due to study-

specific heterogeneity regarding disease duration or overall disease burden.33, 34 The distinction

between a primary and a secondary form of disease progression has become somewhat less relevant

in recent years, as recent clinical guidelines moved towards a new model in which inflammation and

progression are classified separately.35, 36 Pathologically, this seems supported by accumulating

evidence that suggests at least partly similar underlying disease processes, such as the presence of

activated microglia and oxidative stress.37, 38 Given these pathological similarities and clinical

guidelines, we analysed the accelerated cognitive decline in primary and secondary progressive

patients together as a single group in chapter 2.1 to enable more robust statistical testing. On the

other side of the spectrum, however, we were unable to include patients with very early disease, i.e.

around or even before diagnosis, as the earliest patients in our sample had a disease duration of

around four years. We were therefore not able to assess the rate of cognitive decline at this earliest

stage of the disease and we encourage future longitudinal studies to investigate early inception

cohorts to address this remaining question.

Longitudinal structural pathology assessment using MR imaging – effects of scanner upgrade

What pathological process could be driving the continuous and even accelerated cognitive decline in

progressive multiple sclerosis is still unclear. The fact that the rate of cognitive decline varies according


to disease stage, may even suggest that these underlying pathological substrates may change over

time.39-42 Several distinct cross-sectional pathological correlates of cognitive impairment have been

discovered, such as the amount of white matter lesions, diffuse white matter integrity loss and

especially the severity of grey matter atrophy.43-45 Using such cross-sectional study designs, it is,

however, difficult to assess shifts in the pathological drivers of cognitive decline with disease

progression since information on the ordering of pathology and accompanying changes in cognitive

function is missing. To longitudinally study pathological correlates of cognitive decline at different

stages of multiple sclerosis, we also obtained MR imaging in our cohort at both time points using a

General Electric 3-Tesla system. This enabled us to evaluate the rates of grey matter atrophy and white

matter lesion accumulation in relapsing and progressive patients separately as described in chapter

2.1. However, between the initial and follow-up visit, a scanner upgrade took place which is known to

have effects on volumetric measurements.46 To be able to correct for this effect, we employed a novel

approach in which we modelled the upgrade effect based on the volumetric changes observed in the

healthy control group. Normal aging-related atrophy was modelled using both the cross sectional

volumes at each time point and volumetric changes during follow-up in the healthy controls assuming

a non-linear and sex dependent relationship between age and volume.47-49 The differences between

volumes at initial visit and corrected follow-up volumes were then used to derive annual atrophy rates

in both the healthy control subjects and patients with multiple sclerosis. Although not without

limitations, such as the use of a single correction factor, we think this innovative approach could

provide a means of effectively dealing with scanner upgrade challenges and could also possibly enable

the longitudinal study of additional existing datasets.

Progression-related faster cortical atrophy associated with accelerated cognitive decline

The obtained cortical and deep grey matter atrophy rates were then compared between relapsing-

remitting and progressive groups. This analysis showed that while deep grey matter regions

demonstrated a stable (but high) atrophy rate in both relapsing-remitting and progressive groups,

cortical atrophy rates showed a clear acceleration with an about 1.8 times faster atrophy rate in

progressive compared to relapsing-remitting patients. What is causing this pattern of stable deep, but

accelerating cortical atrophy is still unclear, but it has been hypothesized that the close proximity of

the deep grey matter regions to the lesion-prone periventricular white matter could make them more

vulnerable for the impact of white matter lesions through axonal damage and Wallerian or dying-back

degeneration.50, 51 This would leave the deep grey matter regions vulnerable for the bouts of

inflammation typical for the relapsing-remitting stage of the disease.37 The acceleration in cortical

atrophy in the more neurodegenerative progressive phase of multiple sclerosis could then possibly be

explained by pathological processes within the (cortical) grey matter itself such as local microglial

activation,52 the presence of B cell follicles in the inner meningeal layers53 and the formation of cortical

lesions.54 However, given the absence of an association between cortical lesions and neuroaxonal

loss,55 an alternative explanation could be that initial white matter tract damage and neuroaxonal loss

in deep grey matter regions could lead to a ‘second-order network-mediated effect’ on connected

cortical regions.56, 57 To further investigate the impact of deep and cortical atrophy, as well as lesions

on cognitive decline, we correlated longitudinal changes in these imaging measures to longitudinal

rates of cognitive decline. Interestingly, longitudinal correlates of cognitive decline shifted from the

rate of lesion volume increase in stable relapsing-remitting patients, to the rate of deep grey matter

atrophy in converting relapsing-remitting patients to the rate of cortical atrophy in progressive

patients. With the transition from a more inflammatory towards a more neurodegenerative disease,37,

38 the substrate of cognitive decline, therefor, also seems to shift from white matter lesions and deep

grey matter in earlier disease stages towards cortical atrophy in late stage disease.

The aforementioned findings point towards an acceleration of cortical atrophy, with severe

cognitive implications. It is now of the utmost importance to further investigate, which of the proposed

or other underlying mechanisms primarily explains the accelerating cortical atrophy. These insights

could potentially delineate new neuroprotective treatment targets, which are unfortunately largely

lacking at the moment. Such neuroprotective strategies could include the stimulation of axonal

remyelination, restoration of trophic support, reduction of oxidative stress and the prevention of

sodium accumulation.58 In addition, future studies should aim to longitudinally acquire high quality

conventional as well as advanced MR imaging data such as diffusion-weighted, magnetic-transfer

weighted as well as double/phase-sensitive inversion recovery images to measure additional

pathological substrates such as white matter integrity, myelin content and cortical lesions

respectively.59 Finally, the inclusion of additional imaging time points (at least three) would enable a

more sophisticated measurement and modelling of within subject acceleration and deceleration in the

accumulation of different types of pathology, thereby reducing potential sources of bias and enabling

a more in-depth study of pathological correlates.

Depletion of cognitive reserve in patients with grey matter atrophy

In chapter 2.2 we used a different approach to unravel underlying mechanisms of cognitive

impairment in early and advanced stages of multiple sclerosis by using a radiological instead of a clinical

classification. Using the cross-sectional measurements of the first time-point of our cohort of 332

patients, we divided patients into either having atrophy or not within the deep and cortical grey matter

separately. Within each subgroup we investigated the frequency and correlates of cognitive

impairment. In patients without any type of grey matter atrophy (N=132) cognitive impairment was

still common (32%) and best explained by a lower level of education. In patients with deep as well as


cortical grey matter atrophy (N=65), cognitive impairment (75%) was explained very differently,

namely by the severity of accompanying white matter damage, such as white matter lesion volume

and white matter integrity damage, but not by level of education. Interestingly, there were almost no

patients with isolated cortical atrophy (N=6), and patients with isolated deep grey matter atrophy

(N=125) showed a frequency of cognitive impairment that was in between those without any atrophy

and those with wide-spread atrophy. These cross sectional findings further substantiate the hypothesis

of an order of events, as was already indicated in chapter 2.1, where patients move from no atrophy

towards deep grey matter atrophy, and then towards atrophy in both deep and cortical grey matter.

The protective effect of higher level of education against cognitive impairment that we

observed has been observed previously in multiple sclerosis,60 and a similar protective effect has been

observed for intellectual enrichment. This effect has also been observed in other neurological

disorders,61 and is commonly referred to as cognitive reserve, which seemingly serves to protect

against cognitive decline in the face of accumulating disease-related pathology.62, 63 Our finding that

this protective effect of cognitive reserve is primarily relevant in early stages of multiple sclerosis has

been confirmed recently using a disease duration based classification.64 This depletion of cognitive

reserve could perhaps, in conjunction with the acceleration in cortical atrophy described in chapter

2.1, at least partly explain the acceleration in cognitive decline observed in progressive multiple

sclerosis, where this buffer is seemingly lost. This protective effect of cognitive reserve and how it

could relate to the accumulation of different types of pathology during different disease stages is

illustrated in Figure 1.

Regional disturbance in brain function as an imaging marker of cognitive impairment

In addition to examining how structural pathology and protective mechanisms such as cognitive

reserve could determine cognitive decline at different disease stages, we also explored disturbances

in brain function in chapter 2.2, which will be more thoroughly discussed in chapter 3. By comparing

cognitively impaired and preserved patients in the groups without and with atrophy separately, we

showed that a disturbance in the connectedness of the posterior cingulate cortex characterized both

cognitively impaired patient groups. The presence of this functional disturbance in cognitively impaired

patients without grey matter atrophy was an interesting findings since white matter damage (i.e.

lesions, integrity damage) was also still limited in this group and did not differ between cognitively

impaired and preserved patients. Based on these findings we hypothesized that individuals with a low

cognitive reserve might be more vulnerable to develop such a functional disturbance in the presence

of only limited structural pathology. In this theoretical model, the accumulating pathology is buffered

by cognitive reserve and when this pathology exceeds an individual’s cognitive reserve buffer, both

functional network disturbances and cognitive decline could set in. Future research should now further

investigate whether such functional measures are indeed more closely related to clinical dysfunction

compared to structural measures (which are to a variable extent buffered by cognitive reserve) and

could serve as more specific imaging markers of cognitive impairment.

Figure 1. Theoretical model of the evolution of imaging measures of pathology and cognitive decline.

Patients with low cognitive reserve (red) and high cognitive reserve (green) have different trajectories, but in both cognitive

decline is driven by the accumulation of different types of pathology; white matter lesions (orange), deep grey matter atrophy

(light blue) and cortical grey matter atrophy (dark blue). Note: atrophy measures are expressed as the inverse of deep and

cortical grey matter volumes and indicate atrophy severity. Abbreviations: GM = grey matter, WM = white matter, IS =

clinically isolated syndrome, RRMS = relapsing-remitting multiple sclerosis, SPMS = secondary progressive multiple sclerosis,

PPMS = primary progressive multiple sclerosis.

Disrupted deep - cortical functional connectivity balance with disease progression

In chapter 2.1 and chapter 2.2 we identified cross-sectional and longitudinal atrophy patterns related

to cognitive impairment and decline. These studies seem to indicate a transition from deep towards

cortical grey matter atrophy. Still unclear, however, is how this progressively accumulating volume loss

disturbs normal brain function (or vice versa), and whether these functional disturbances could shed

additional light on the rapid clinical deterioration during the progressive phase of multiple sclerosis. In

Chapter 2.2 we already briefly investigated disturbances in brain function, while in Chapter 2.3 these


concepts were explored further, by studying patterns of functional connectivity changes within and

between the deep and cortical grey matter structures in early and late relapsing-remitting, as well as

secondary progressive multiple sclerosis. The main findings were that in early relapsing-remitting

multiple sclerosis, functional changes were not yet seen, despite mild atrophy in the deep grey matter.

However, in late relapsing-remitting patients there was an increase in deep grey matter connectivity

(both between individual deep grey matter regions as well as between deep and cortical grey matter

regions), while within cortex connectivity remained normal. In progressive patients there was a further

increase in deep grey matter connectivity, but now also accompanied by a decrease in connectivity

between cortical regions. This shift in the subcortical – cortical communication therefore appears to

follow the same pattern previously observed for atrophy with deep grey matter functional

abnormalities already present in late relapsing-remitting patients and cortical functional abnormalities

becoming most apparent in progressive multiple sclerosis.

Only few studies have investigated functional connectivity abnormalities across multiple

sclerosis disease stages, suggesting primarily increases in functional connectivity in early multiple

sclerosis and a broad decrease in connectivity in late stage disease.21, 65-70 Our results extend these

findings by showing that the pattern of functional disturbances not only follows an initial increase >

late decrease pattern, but also seems to spread from the deep grey matter towards the cortical grey

matter. Longitudinal studies should now further assess what could be driving these shifting patterns

of functional connectivity and investigate the temporal ordering within individual patients to

determine whether the atrophy shift indeed precedes the functional connectivity shift. A recent

modelling study already showed that white matter lesions in thalamocortical connections (modelled

by removal of connections) resulted in an initial increase in connectivity, followed by a global

decrease.71 How these patterns of atrophy accumulation and disturbed functional connectivity relate

to clinical functioning is also still poorly understood. Early interpretations that increases in functional

connectivity likely reflected compensatory changes seem to be overly simplistic, especially since such

changes were often related to worse cognitive functioning, as was also found in this study.20, 66, 67, 72, 73

However, such correlations are also difficult to interpret, since patients with a more severe cognitive

dysfunction, could theoretically also simply require more compensation. To further unravel these,

sometimes confusing, functional connectivity patterns and to elucidate how these patterns relate to

clinical and especially cognitive functioning, a move towards a more sophisticated network approach

was needed.

Shifts in the functional importance of brain regions

In chapter 3 we, therefore, entirely focused on functional mechanisms that could explain cognitive

dysfunction in multiple sclerosis. Our previous results indicated that there is a spread of functional

abnormalities, but that we may need to more holistically study brain function. Therefore, we adopted

a more in-depth connectomics approach that allowed us to study how regions are embedded within

the entire functional brain network.74

Previous studies in multiple sclerosis already reported changes in global network organization such as

a lower global efficiency and a more modular architecture, both related to lower cognitive

functioning.75, 76 While studies performed in other neurological disorders have shown a common

pattern of disturbances to be located within the highly central hub regions, such hub-related regional

changes have not been extensively studied in multiple sclerosis.15, 77, 78 In chapter 3 we therefore

investigated changes in the functional connectedness or ‘hubness’ of regions within the functional

brain network.15 To specifically investigate changes related to cognitive status, our cohort of 332

Box A. Connectomics

In recent decades, it has become increasingly clear that if we want to better understand

complex systems, such as the cognitive functions performed by the human brain, only

studying individual brain regions or individual connections in isolation provides an

incomplete picture. This realization that the behaviour of complex systems is shaped by

the interactions among their constituent elements has led to the ascendance of network

neuroscience.5 This research field studies the brain as a network and such brain networks

can be constructed using functional MR imaging data, by computing functional connectivity

strengths between all pairs of brain regions.5 This functional network can then be analysed using

graph theoretical approaches to reveal global and regional network properties.1 One of the key

global (i.e., whole-brain) network properties is the presence of a small-world network type of

organization.11 This organization is characterized by the combination of a high local clustering of

connections between brain regions (likely facilitating local processing) and a short path length

between brain regions (likely facilitating the efficient global integration of information).11 The

amount of segregation of the brain network into smaller clusters of brain regions or modules is

captured by the network measure modularity.1, 13 In addition to investigating such global

(summary) measures of network organization, a large body of research is dedicated to unravelling

the role that individual brain regions play within this network. One group of brain regions in

particular, the so-called hub-regions, is receiving a lot of attention.15 These highly central (i.e.,

highly connected) brain regions are thought to be important for the neural integration of

information and to play a vital role in a diverse set of cognitive functions.15


patients with multiple sclerosis was divided into a group of patients without cognitive impairment

(N=180), a group with mild cognitive impairment (N=65) and a group with severe cognitive impairment

(N=87). In chapter 3.1, we evaluated differences in functional network architecture between

cognitively impaired and preserved patients on a voxel-wise level using the network measure

‘eigenvector centrality’, which denotes regional functional importance or ‘hubness’.79 This analysis

featured a statistical comparison of each individual voxel between groups, which (after multiple

comparison correction) showed a shift towards a higher centrality (i.e., a higher functional importance)

of brain regions that are commonly seen as part of the default-mode network and fronto-parietal

networks (as also shown in chapter 3.3) in cognitively impaired compared to preserved patients

during the resting state.

The default-mode network was originally discovered as a group of brain regions exhibiting

suppressed activity during goal-directed behaviour, which led to the term “default-mode”, as this was

interpreted as a sort of baseline activity of the brain.80 This network was initially thought to be mainly

involved in introspective and self-referential thought,81 but recent studies have also pointed out its

role as a structural core network of brain hubs facilitating cognitive processing.82 A previous task-

related functional MR imaging study in multiple sclerosis already demonstrated a reduction in the

normal deactivation of the default-mode network in cognitively impaired compared to preserved

patients during a working memory task, indicating a loss of the normal dynamic behavior.83 Based on

these as well as our findings we hypothesized that the abnormally strong connectedness of the default-

mode network (detectable during rest) could interfere with adequate deactivation during cognitive

processing.84, 85 A possible mechanisms explaining the importance of inadequate default-mode

suppression is postulated by the default-mode interference hypothesis which states that impaired

default-mode network attenuation may result in lapses of attention due to the intrusion of

introspective thought, thereby reducing cognitive task performance.86 This may also relate to the

trouble focusing on specific cognitive tasks as commonly described by patients with multiple sclerosis,

although such anecdotal reports still need to be verified in future studies. In addition to a network-

mediated explanation for inadequate default-mode suppression, it could also be explained by local

pathological processes. A recent study, for example, used a combined MR spectroscopy and functional

MR imaging approach to show that healthy individuals with a higher glutamate/GABA ratio in the

posterior cingulate cortex (the central core hub of the default-mode network) had a lower task-

induced deactivation of this same region.87 It could therefore be hypothesized that a selective loss of

GABAergic (inhibitory) interneurons, for example due to higher sensitivity to multiple sclerosis

pathology, could result in such reduced task-induced deactivation, which now needs further

investigation.

The analyses performed in this thesis mostly focussed on ‘general’ cognitive impairment, but did not

take into account that underlying mechanisms could differ depending on the affected cognitive

domain. We, therefore, already briefly explored functional network changes in patients with isolated

cognitive deficits. To study isolated deficits, we classified patients in our cohort as having an isolated

cognitive deficit when scoring Z<-2 compared to the healthy controls on the most affected cognitive

domain, while all other domains scored at least 25% higher. Using this classification, a total of 80

patients (24%) displayed isolated cognitive deficits, including a visuospatial memory deficit for 18

patients (5%, visuospatial memory Z = -3.80), an executive functioning deficit for 16 patients (5%,

executive functioning Z = -4.93), an information processing speed deficit for 13 patients (4%,

information processing speed Z = -2.96), and a working memory deficit for 18 patients (5%, working

memory Z = -3.31). Other isolated cognitive domain deficit groups included less than 10 patients with

multiple sclerosis and were not considered for subsequent analyses. The functional network

comparison on eigenvector centrality differences between these isolated cognitive deficit groups and

healthy controls showed some remarkable heterogeneity between isolated deficit groups (Fig. 2).

While the information processing speed impaired and working memory impaired groups showed the

familiar pattern with increased posterior cingulate centrality (Fig. 2A-B), the executive function group

showed a cluster of decreased centrality in the frontal lobe (Fig. 2C) and the visuospatial memory

impaired group showed a cluster of decreased centrality in the sensorimotor cortex (Fig. 2D). The

pattern of decreased visual lobe centrality why, however, shared by all groups. These results highlight

that underlying mechanisms indeed appear to differ as expected and future studies should more

precisely delineate such patterns.

Increased connectivity of network hubs

In chapter 3.2 we explored hub-connectivity further, by exploring what could be driving the observed

shift in hub-connectivity towards more central (i.e., highly connected) default-mode and frontoparietal

network regions. In contrast to chapter 3.1, we now used a top-down approach, by initially defining

brain hubs as regions belonging to the default-mode and frontoparietal networks,88, 89 and we then

studied alterations in their connectivity patterns in cognitively impaired patients. We evaluated

connections between hub-regions within these networks, connections between hub-regions across

these networks and between hubs and the rest of the brain. This analysis showed that the higher

connectedness was not driven by stronger connections within or between these networks, but by a

stronger connectivity between these hub-networks and the rest of the brain. Regionally, we observed

that these connectivity increases were driven mostly by the inferior parietal, angular and posterior

cingulate cortices, but not the frontal components of these networks. Interestingly this latter region,

the posterior cingulate, is one of the most strongly connected hubs in the brain and forms the core of


the default-mode network.90 The results presented in chapter 2.2 also pointed towards a functional

disturbance in this posterior cingulate cortex in cognitively impaired patients and this region has been

implicated in a range of neurological diseases characterized by cognitive dysfunction.90

Figure 2. Centrality changes in isolated cognitive groups compared to healthy controls.

A. Clusters with significantly altered centrality in patients with isolated information processing speed deficits compared to

healthy controls. B. Clusters with significantly altered centrality in patients with isolated working memory deficits compared

to healthy controls. C. Clusters with significantly altered centrality in patients with isolated executive functioning deficits

compared to healthy controls. D. Clusters with significantly altered centrality in patients with isolated visuospatial memory

deficits compared to healthy controls. Red/yellow: voxels with significantly higher centrality. Blue/light blue: voxels with

significantly lower centrality. Significant clusters are projected on Freesurfer’s fsaverage standard brain for display purposes.

This finding across neurological disorders has led to the theory that there may be a generic hub-specific

vulnerability within the brain.77, 78 It is, however, still unclear what could be driving such high

connectivity of hub regions, i.e., whether it could result from a dysfunction originating within these

hub regions themselves or reflect an increased information load directed towards hub regions.77, 78

Such a hub-specific vulnerability has been postulated as a final common pathway in a range of

disorders in which nodes in a hierarchically organized network start to fail, leading to a rewiring of

information to nodes higher up in the hierarchy.77 While the aforementioned transferral of information

to higher order hubs could represent an active process that initially serves to preserve information

flow in the system, these hubs could eventually become overloaded which could result in hub failure

and structural damage due to excitotoxicity.77 This could also possibly explain how a network-guided

process with increased processing demands placed upon brain hubs could lead to a regional

acceleration of neurodegeneration as hypothesized in chapter 2.1. More work needs to be done,

however, to confirm whether such a mechanism is at play in multiple sclerosis, but several recent

studies have demonstrated the early occurrence of atrophy in the posterior cingulate cortex and its

relevance for cognitive dysfunction.56, 91 In future studies, it would be highly interesting to

longitudinally study such changes in brain function and regional atrophy simultaneously, perhaps even

in combination with other local measures of regional (metabolic) changes, such as arterial spin labelling

and MR spectroscopy. Such studies might be able to distil the order of events and unravel whether the

high posterior cingulate centrality indeed precedes neurodegeneration in individual patients, which

could be an indication of hub failure.

Increased network rigidity in cognitively impaired patients with multiple sclerosis

In the previous chapters we assessed changes in functional connectivity between brain regions based

on average signal coherence, whereas recent evidence indicates the non-stationary nature of

functional connectivity between brain regions over time (Box B).1, 2 Given the likelihood of missing

relevant information by only studying average connectivity strength between rain regions and the

possible relevance of disturbances in connectivity dynamics in explaining cognitive deficits in multiple

sclerosis, we shifted to a dynamic network approach in chapter 3.3.6 These dynamic functional

Box B. Moving from connectomics to chronnectomics

Most studies investigating the functional connectome so far have looked at the average functional

brain network organization, by computing T2*-based blood-oxygen-level-dependent (BOLD) signal

coherence between brain regions across the duration of an entire functional scan, also referred to

as stationary functional connectivity. However, accumulating evidence indicates that functional

connectivity between brain regions is non-stationary and alternates between periods of low and

high functional coupling over time.1, 2 These fluctuating coupling strengths could reflect ongoing

changes in processing requirements to enable a multitude of cognitive functions.2 To better

understand cognitive impairment, it is therefore essential to not only study the average functional

connectivity strengths between brain regions, but also incorporate the time-varying nature of

these coupling strengths. Or, in network terms, it is essential to not only study the average network

organization or average functional connectome, but also asses functional connectome dynamics

or the functional ‘chronnectome’.6 An early study in patients with schizophrenia already indicated

the potential utility of such approach by reporting shorter dwell times and abnormal functional

connectivity in some, but not all, dynamic brain states.7


connectivity and dynamic network analyses, however, also comes with some methodological concerns,

one of which is the strong need for applying effective motion reduction strategies.92, 93 The field has

progressed considerably on the topic of motion regression strategies, with clarity on the effectiveness

of different procedures provided by recent papers.93-95 Before performing the dynamic analyses, an

independent component analysis based strategy for Automatic Removal of Motion Artefacts (ICA-

AROMA)96 was, therefore, added to our pre-processing pipeline. In this pipeline, part of the

independent components in an individual functional dataset are classified as motion and specifically

these components are regressed out, leaving a denoised functional MR imaging dataset. Using this

corrected dataset, we started by repeating the stationary centrality comparison between cognitively

impaired and preserved patients presented in chapter 3.1. This repeated analyses confirmed our

previous findings but even expanded upon them, as now an even clearer shift in the functional balance

between brain networks could be seen, with an additional decreased centrality in the primary sensory

and visual networks and increased centrality in the default-mode and frontoparietal networks (chapter

3.3) as well as thalamus, which suggests that applying motion correction strategies could reduce type

I but also type II errors. After this methodological improvement, we proceeded with the investigation

of functional network dynamics.

To assess functional network dynamics, we adopted a dynamic read-out of the eigenvector

centrality metric used in chapter 3.1, by separating the individual functional images into 181 smaller

20 time point sliding windows and computing the functional network topology for each window

separately. This enabled the investigation of dynamically changing patterns in the functional

importance of brain regions. In the healthy control group, highest regional dynamics were observed in

the default-mode and frontoparietal networks, extending recent insights from dynamic functional

connectivity studies that showed higher variability in coupling strength between nodes of higher order

cognitive network.97, 98 It has been hypothesized that this could enable a broader set of connectivity

configurations required for more complex cognitive functions.97, 98 The comparison between

cognitively impaired and preserved patients on regional centrality dynamics showed a marked

reduction of dynamics in regions part of the default-mode network (i.e., posterior cingulate cortex and

right angular gyrus), frontoparietal network (i.e., middle frontal gyrus and superior parietal gyrus),

visual network and thalamus. While the interpretation of this increased resting-state network rigidity

is still unclear, we hypothesized that these networks could have become ‘stuck’ in their respective high

and low centrality states, possibly interfering with their normally alternating behaviour.

Altered between-network coherence in cognitively impaired patients with multiple sclerosis

To be able to assess whether this loss of dynamics of specific networks also impacted their normally

alternating behaviour, we first used an independent component analysis on the dynamic centrality

maps to show that fluctuations in the functional importance of brain regions over time indeed occurred

in such functional network patterns. Given the normally strong anti-correlated behaviour of networks

such as the default-mode network with other networks99, 100 we next investigated whether the

aforementioned network rigidity could have affected this aspect of normal network dynamics. Before

investigating such disturbances, we first used a correlation and hierarchical clustering analysis to show

that individual resting-state network fluctuations in functional importance are organized into two main

multi-network clusters; a cluster that included default-mode and frontoparietal networks and a cluster

that included sensorimotor and visual networks. Networks within the same cluster mostly showed

positively correlated fluctuations over time, whereas networks in opposing clusters mostly

demonstrated anti-correlated behaviour. A similar hierarchical organization has been previously

shown based on activity fluctuation patterns99, 101 and a widely held view is that these anti-correlated

groups of networks process opposing functions and compete for processing resources.102, 103

The investigation of how the observed network rigidity could have affected normal network

dynamics in multiple sclerosis showed a loss in the normal anti-correlation between the default-mode

and visual networks in cognitively impaired compared to preserved patients with multiple sclerosis.

While previous studies in healthy controls already demonstrated the cognitive relevance of a stronger

anti-correlation between the default-mode and the dorsal attention networks101 this was not yet

known for the anti-correlation between the default-mode and visual networks. While still unclear, it

could be hypothesized that this loss of anti-correlation, combined with the more central default-mode

at rest (chapter 3.1) and less suppressed default-mode during cognitive processing,83 could represents

a disruption in the intrinsic wiring between internally and externally focused systems, ultimately

resulting in cognitive disturbances. While several studies have already shown reduced co-fluctuations

after lesional damage due to stroke104 and traumatic brain injury,105 future studies now need to further

investigate whether lesional disconnection or other pathological processes could result in such a

reduced anti-correlation in multiple sclerosis.

Functional network dynamics during cognitive task states

The functional MR imaging studies described in this thesis were all performed on functional data

acquired during the resting state, a situation in which no particular cognitive task is presented. This

state is thought to mimic conditions in which subjects are required to engage in internally directed

mental operations.106, 107 While it could be argued that this state represents just another cognitive

state,107 explicit cognitive task paradigms perhaps better mimic the daily difficulties experienced by

patients with multiple sclerosis. In addition, the simultaneous acquisition of performance metrics such

as reaction times and error rates during task-based functional MR imaging enables the coupling

between measures of network functioning and desired performance outcomes. To better understand


cognitive functioning and its disturbances under pressure of disease, it is therefore important to also

investigate the functional network (dynamics) during explicit cognitive task conditions. Since our

multiple sclerosis cohort dataset did not feature functional MR imaging during cognitive task

conditions, we started by exploring functional network dynamics during cognitive task states in a group

of 100 healthy controls part of the Washington University–Minnesota Consortium Human Connectome

Project.108 In this freely available dataset, 3-Tesla MR imaging was acquired during rest as well as six

different cognitive task paradigms.109-112 We adopted the same pipeline as the one applied in chapter

3.3 and computed average centrality and centrality dynamics within resting-state networks during rest

and six cognitive task states.

After ordering the resting-state networks based on their respective centrality and dynamics

during rest (Fig. 3, top panels), we zoomed in on the default-mode network and compared its centrality

and dynamics during the cognitive task states to the resting condition. During cognitive task states, the

default-mode network became less central (significant for all tasks), but more dynamic (significant for

three out of six tasks, Fig. 3, bottom panels).

Figure 3. Default-mode network centrality changes during cognitive tasks compared to the resting state.

Using data from the Human Connectome Project,108 average centrality and centrality dynamics were computed within

resting-state network masks during the resting state (top panels). In addition, default-mode network centrality/dynamics

were computed during six different cognitive task states and then compared to default-mode network centrality/dynamics

during the resting state, expressed as Z-scores compared to the resting-state. Bars reflect group means and whiskers reflect

standard error of the mean. Abbreviations: vs = versus, DMN = default-mode network.

This indicates that the default-mode network is less important during active cognitive processing, as

was expected, but the higher dynamics could indicate that this default-mode suppression is not a

constant, but a highly dynamic process. It is now essential to further investigate whether the more

central but rigid default-mode network in cognitively impaired patients with multiple sclerosis during

rest hampers its dynamic attenuation when switching to externally oriented cognitive processing. In

addition, there still is an ongoing debate regarding to what extent functional connectivity dynamics

represent changes in cognitive content in contrast to processes concerned with maintenance of the

stability of the brain’s functional organization and this also needs further clarification.113, 114 Finally, a

large number of methodological approaches to analyse and model dynamic functional connectivity

data is currently being used and developed.113, 114 The challenge in the years to come will be to

integrate findings from all these different approaches to derive an overarching understanding of how

functional network dynamics contribute to the elucidation of cognitive decline in multiple sclerosis.

MR imaging predictors of physical disability progression and cognitive decline

The previous chapters largely focused on increasing our mechanistic understanding of how

accumulating pathology, counterbalanced by protective mechanisms such as cognitive reserve, could

result in disturbed brain function and cognitive decline. In chapter 4, we shifted to a more clinically

oriented purpose by assessing whether we could use MR imaging measures to predict future cognitive

decline. There is an urgent clinical need to be able to accurately predict such future decline in order to

optimize disease management and treatment strategies. In chapter 4.1 we, therefore, assessed

whether MR imaging measures acquired around the time of diagnose, as well as changes on MR

imaging in the first two years after diagnose, were able to predict physical disability and cognitive

functioning six and twelve years later. For this study, we included a cohort of 115 patients with multiple

sclerosis, part of our larger sample of 332 patients of whom MR imaging was obtained around the time

of diagnose and two years later. This study showed that patients with higher T1-hypointense lesion

volumes at baseline and early whole-brain atrophy were more likely to suffer from both disability

progression and cognitive dysfunction. While previous studies already indicated the predictive value

of lesions115 and atrophy116, 117 in smaller samples and using shorter follow-up durations, we showed

that especially the volume of the more severe T1-hypointense lesions around diagnose is predictive

for long-term clinical outcome.

In addition to MR imaging measures, the predictive value of demographic factors was also

assessed and this analysis showed that cognitive dysfunction, but not physical disability progression,

was also predicted by male sex and lower level of education. The effect of level of education likely

reflects the protective effect of the aforementioned cognitive reserve against the accumulation of

disease-related pathology.62, 63 While its value as a predictor of future cognitive decline was to our

knowledge not previously shown in multiple sclerosis, the finding was highly expected considering that

level of education is commonly reported as a cross sectional correlate of cognitive impairment as we


also observed in chapter 2.2.60, 118, 119 Similarly, previous cross sectional studies have already shown

that male patients tend to be more severely affected by multiple sclerosis and the current study also

showed the predictive value of male sex on future cognitive dysfunction.40, 120, 121

Advanced structural MR imaging predictors of cognitive decline

Our study described in chapter 4.1 and other studies that assessed cognitive decline predictors,115-117

provided a useful set of MR imaging-based and demographic predictors of future cognitive decline that

can be assessed at the time or in the first years after the diagnose of multiple sclerosis is established.

However, these studies only included relatively early multiple sclerosis cohorts, MR imaging measures

acquired at relatively low field strength (1.0 or 1.5 Tesla) and did not include regional MR imaging

measures.115-117, 122 In addition, our previous paper described in chapter 2.1 indicated that there may

be a change in the primary drivers of cognitive decline in different stages of the disease. The MR

imaging measures that are primarily predictive at time of diagnose therefore do not need to be the

same as predictors in late relapsing-remitting and progressive stages of multiple sclerosis.

In chapter 4.2 we, therefore, investigated MR imaging predictors of subsequent cognitive

decline in our heterogeneous longitudinal cohort of 234 patients with relapsing-remitting, secondary

progressive as well as primary progressive multiple sclerosis described in chapter 2.1. In this study,

advanced 3-Tesla MR imaging and identical neuropsychological evaluations were acquired at two time

points with a five year interval. Since neuropsychological data was available for both time points, we

were able to investigate predictors of cognitive change over five years. Changes in cognitive function

were assessed by computing the aforementioned practice adjusted reliable change index for each

cognitive domain, which also enabled us to correct for learning effects based on changes in the healthy

control group.31 By then dividing these individual reliable change index scores by the interval duration,

we obtained yearly cognitive decline rates for each individual patient. To evaluate MR imaging

predictors of cognitive decline, we first separated patients into cognitively stable and cognitively

declining groups. To enable this classification, a new approach was developed, which showed optimal

sensitivity and specificity for the criterion of a yearly rate of cognitive decline < -0.25 on two or more

cognitive domains.

Based on this criterion, approximately 28% of patients demonstrated cognitive decline over a

period of five years, which was more prevalent in progressive patients (54%) than relapsing-remitting

patients (21%). To determine whether patients that demonstrated cognitive decline during follow-up

already differed from patients that remained stable at the time of initial visit, we compared MR

imaging measures between these groups at time of initial visit. This comparison showed that patients

that demonstrated cognitive decline during follow-up already showed more severe structural

pathology at initial visit, including higher white matter lesion volume, lower white matter integrity,

lower deep and cortical grey matter volumes. We then evaluated which of the MR imaging measures

at initial visit best predicted subsequent cognitive decline at different disease stages separately; early

relapsing-remitting multiple sclerosis (<10 years symptom duration), late relapsing-remitting (>10

years symptom duration) and progressive multiple sclerosis. This analysis showed that white matter

integrity damage and deep grey matter atrophy were the main MR imaging predictors of future

cognitive decline (general and information processing speed decline respectively) in early relapsing-

remitting multiple sclerosis, whereas cortical atrophy was the main MR imaging predictor of future

cognitive decline in both late relapsing-remitting as well as progressive multiple sclerosis (Fig. 4).

Figure 4. Predictors of cognitive decline at different disease stages.

In early relapsing-remitting multiple sclerosis, general cognitive decline was predicted by the severity of white matter

integrity damage and information processing speed decline by deep grey matter atrophy. In both late relapsing-remitting and

progressive multiple sclerosis, cognitive decline and information processing speed decline were predicted by cortical grey

matter atrophy. Abbreviations: MS = multiple sclerosis, RRMS = relapsing-remitting multiple sclerosis, WM = white matter,

GM = grey matter.

These results indicate that the most relevant MR imaging predictors of cognitive decline indeed

depend on the disease stage. When trying to determine the likelihood of cognitive decline in patients

with early relapsing-remitting multiple sclerosis, assessing the severity of white matter integrity

damage and deep grey matter atrophy seems most relevant, whereas in later disease stages assessing

cortical atrophy appears most relevant. This shift in relevant predictors for future cognitive decline is

in line with the longitudinal MR imaging substrates of cognitive decline presented in chapter 2.1. In

this previous chapter we showed a progressive shift in the underlying pathology of cognitive decline,

which seemed to move from white matter lesions in the relapsing-remitting stage towards deep grey

matter atrophy around the time of conversion to a progressive phenotype and towards cortical

atrophy in the more progressive stage of the disease.37, 38 With initial disease mainly characterized by

an increase in inflammatory lesions and deep grey matter atrophy,37, 123 it seems logical that

quantifying inter individual differences in the severity of these process is most predictive of future

clinical functioning, since cortical atrophy is yet to start during this early disease stage. After conversion


to progressive phenotype, the rate of cortical atrophy, on average, increases, possibly accompanied

by an acceleration of cognitive decline (chapter 2.1). At this stage assessment of the presence and/or

severity of this cortical atrophy becomes the most relevant predictor of future cognitive decline in

patients with multiple sclerosis.

An ongoing topic of debate in the field of cognition research in multiple sclerosis is whether

focusing on aggregate levels of tissue damage or regional damage better explains the severity of

cognitive impairment, with several studies highlighting the relevance of damage to the thalamus,124

hippocampus125 and cortical regions such as the posterior cingulate cortex in particular.56 We therefore

also investigated such regional damage measures as possible predictors for future cognitive decline in

chapter 4.2. This regional analysis showed that the combination of integrity damage of the anterior

thalamic radiation, the percentage of lesions in the superior longitudinal fasciculus and atrophy of the

temporal lobe best predicted future cognitive decline. This regional analysis explained a substantially

larger proportion of variance compared to the model that included whole-brain (i.e., aggregate) MR

imaging measures only, which suggests that the assessment of regional damage patterns could further

increase the accuracy of predicting future cognitive deterioration.

Functional MR imaging predictors of cognitive decline

After evaluating the predictive value of structural MR imaging measures for cognitive decline in

chapter 4.1 and chapter 4.2, the predictive value of functional MR imaging measures also needs

further evaluation, especially since these could potentially precede certain structural processes as

hypothesized above. While functional MR imaging was not acquired at baseline in the cohort studied

in chapter 4.1, it was obtained at the initial visit of the cohort studied in chapter 4.2. We, therefore,

already briefly explored whether we could detect differences in brain function at initial visit between

patients that would demonstrate cognitive decline during subsequent follow-up compared to patients

that would remain cognitively stable. We compared these groups using the network measure

eigenvector centrality, which we also used in chapter 3, and observed that patients that showed

cognitive decline during follow-up already demonstrated a higher centrality in both the left and right

putamen at initial visit (fig. 5). The predictive value and potential mechanistic insights that such

functional MR imaging predictors of future cognitive decline could provide should now be further

investigated in future studies.

Figure 5. The prediction of future cognitive decline using baseline functional MRI characteristics.

Functional network differences at initial visit between patients that demonstrated cognitive decline (N=66) and patients that

remained cognitively stable (N=168) in the following five year. A. When cognitively declining patients were compared to

cognitively stable patients, a significantly increased centrality in both the left and right putamen was observed in patients

that would demonstrate cognitively decline in the following five years. B. Patients were split into four groups based on the

centrality in the left putamen at initial visit and the average yearly cognitive decline rate during the following five years was

then computed for each group. C. Patients were split into four groups based on the centrality in the right putamen at initial

visit and the average yearly cognitive decline rate during the following five years was then computed for each group. Bars

reflect group means and whiskers reflect standard error of the mean. Abbreviations: L = left, R = right, RCI = reliable change

index.


Conclusions

1. How does the accumulation of grey matter atrophy relate to changes in brain function,

reserve capacity and, ultimately, cognitive decline during the course of multiple sclerosis?

o Cognitive decline is not constant, but seems to accelerate in progressive multiple

sclerosis, as does cortical atrophy.

o The underlying correlate of cognitive decline shifts from white matter lesions and deep

grey matter atrophy in relapsing-remitting multiple sclerosis towards cortical atrophy

in progressive multiple sclerosis.

o Similarly, functional connectivity disturbances seem to transition from deep towards

cortical grey matter areas.

o Cognitive reserve is important, but seems to buffer against cognitive decline mainly

before the manifestation of widespread grey matter atrophy.

2. Can the use of a network approach enable a further unravelling of the underlying

mechanisms of cognitive decline in multiple sclerosis?

o Cognitive decline characterized by a disturbed functional network balance, with more

strongly connected default-mode and frontoparietal networks.

o Especially an increased connectedness of the posterior cingulate hub, which was

already detectable in the presence of limited structural pathology in patients with low

cognitive reserve, signalled cognitive impairment.

o Dynamically, cognitive impairment was characterized by an increased rigidity of core

resting-state networks.

3. What are the main MR imaging predictors of future cognitive decline at different stages

during the course of multiple sclerosis?

o The presence of the more severe T1-hypointense lesions around diagnose and early

atrophy are predictive for worse clinical outcomes after up to twelve years.

o MR imaging predictors of cognitive decline are not constant during the disease course,

but shift from white matter integrity damage and deep grey matter atrophy in early

relapsing-remitting multiple sclerosis towards cortical atrophy in later disease stages.

Future studies are now needed to further investigate

1. Pathological processes underlying cognitive decline

o Is the acceleration of cortical atrophy primarily driven by a network-guided process

that spreads between (cortical) brain regions or by local pathological processes

occurring within the cortex?

o Can we detect novel neuroprotective treatment targets that slow the progression of

cortical atrophy and thereby preserve cognitive function?

o Can we delineate additional relevant pathological processes of cognitive decline by

longitudinally acquiring more advanced MR imaging measures?

o Can we further elucidate the neural mechanisms that protect patients with a high

cognitive reserve from cognitive decline in early, but not late, multiple sclerosis?

2. The relation between functional network disturbances and cognitive decline

o Can we unravel whether the ‘stuck’ default-mode network in a more central position

during rest and its impaired attenuation during task, are primarily due to a local or

network-mediated process?

o Can we delineate whether the functional network disturbances in the posterior

cingulate cortex indeed precede the strong atrophy in this region, possibly suggesting

hub failure?

o Can we more precisely unravel the mechanisms underlying specific cognitive deficits

by studying functional network dynamics during these cognitive task states?

3. MR imaging predictors of future cognitive decline

o Can we further improve the prediction of future cognitive decline by incorporating

functional network disturbances in addition to structural pathology?

o Can we integrate information provided by independent predictors of future cognitive

decline in multiple sclerosis and their interactions to predict an individual patient’s

cognitive trajectory?


References 1. Bassett DS, Sporns O. Network neuroscience. Nat Neurosci 2017;20:353-364. 2. Hutchison RM, Womelsdorf T, Allen EA, et al. Dynamic functional connectivity: promise, issues, and interpretations. Neuroimage 2013;80:360-378. 3. Chiaravalloti ND, DeLuca J. Cognitive impairment in multiple sclerosis. Lancet Neurol 2008;7:1139-1151. 4. Benedict RHB, DeLuca J, Enzinger C, Geurts JJG, Krupp LB, Rao SM. Neuropsychology of Multiple Sclerosis: Looking Back and Moving Forward. J Int Neuropsychol Soc 2017;23:832-842. 5. Bullmore E, Sporns O. Complex brain networks: graph theoretical analysis of structural and functional systems. Nat Rev Neurosci 2009;10:186-198. 6. Calhoun VD, Miller R, Pearlson G, Adali T. The chronnectome: time-varying connectivity networks as the next frontier in fMRI data discovery. Neuron 2014;84:262-274. 7. Damaraju E, Allen EA, Belger A, et al. Dynamic functional connectivity analysis reveals transient states of dysconnectivity in schizophrenia. Neuroimage Clin 2014;5:298-308. 8. Rao SM. A Manual for the Brief Repeatable Battery of Neuropsychological Tests in Multiple Sclerosis: Medical College of Wisconsin. Milwaukee: Medical College of Wisconsin. 1990. 9. Van der Elst W, Van Boxtel MP, Van Breukelen GJ, Jolles J. The Concept Shifting Test: adult normative data. Psychol Assess 2006;18:424-432. 10. Buschke H. Selective Reminding for Analysis of Memory and Learning. J Verb Learn Verb Be 1973;12:543-550. 11. Bassett DS, Bullmore ET. Small-World Brain Networks Revisited. Neuroscientist 2017;23:499-516. 12. Boringa JB, Lazeron RH, Reuling IE, et al. The brief repeatable battery of neuropsychological tests: normative values allow application in multiple sclerosis clinical practice. Mult Scler 2001;7:263-267. 13. Sporns O, Betzel RF. Modular Brain Networks. Annu Rev Psychol 2016;67:613-640. 14. Smith A. The Symbol Digits Modalities Test Manual, revised. Los Angeles: Western Psychological Services, 1982. 15. van den Heuvel MP, Sporns O. Network hubs in the human brain. Trends Cogn Sci 2013;17:683-696. 16. Stroop JR. Studies of Interference in Serial Verbal Reactions (Reprinted from Journal Experimental-Psychology, Vol 18, Pg 643-662, 1935). J Exp Psychol Gen 1992;121:15-23. 17. Brand N, Jolles J. Information processing in depression and anxiety. Psychol Med 1987;17:145-153. 18. Benedict RH, DeLuca J, Phillips G, et al. Validity of the Symbol Digit Modalities Test as a cognition performance outcome measure for multiple sclerosis. Mult Scler 2017;23:721-733. 19. Sumowski JF, Benedict R, Enzinger C, et al. Cognition in multiple sclerosis: State of the field and priorities for the future. Neurology 2018;90:278-288. 20. Louapre C, Perlbarg V, Garcia-Lorenzo D, et al. Brain networks disconnection in early multiple sclerosis

cognitive deficits: an anatomofunctional study. Hum Brain Mapp 2014;35:4706-4717. 21. Schoonheim MM, Hulst HE, Brandt RB, et al. Thalamus structure and function determine severity of cognitive impairment in multiple sclerosis. Neurology 2015;84:776-783. 22. Amato MP, Zipoli V, Portaccio E. Multiple sclerosis-related cognitive changes: a review of cross-sectional and longitudinal studies. J Neurol Sci 2006;245:41-46. 23. Banati M, Sandor J, Mike A, et al. Social cognition and Theory of Mind in patients with relapsing-remitting multiple sclerosis. Eur J Neurol 2010;17:426-433. 24. Benedict RH, Cookfair D, Gavett R, et al. Validity of the minimal assessment of cognitive function in multiple sclerosis (MACFIMS). J Int Neuropsychol Soc 2006;12:549-558. 25. Langdon DW, Amato MP, Boringa J, et al. Recommendations for a Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS). Mult Scler 2012;18:891-898. 26. Amato MP, Ponziani G, Siracusa G, Sorbi S. Cognitive dysfunction in early-onset multiple sclerosis: a reappraisal after 10 years. Arch Neurol 2001;58:1602-1606. 27. Amato MP, Ponziani G, Pracucci G, Bracco L, Siracusa G, Amaducci L. Cognitive impairment in early-onset multiple sclerosis. Pattern, predictors, and impact on everyday life in a 4-year follow-up. Arch Neurol 1995;52:168-172. 28. Achiron A, Barak Y. Cognitive impairment in probable multiple sclerosis. J Neurol Neurosurg Psychiatry 2003;74:443-446. 29. Cortese M, Riise T, Bjornevik K, et al. Preclinical disease activity in multiple sclerosis: A prospective study of cognitive performance prior to first symptom. Ann Neurol 2016;80:616-624. 30. Wijnands JM, Zhu F, Kingwell E, et al. Five years before multiple sclerosis onset: Phenotyping the prodrome. Mult Scler 2018:1352458518783662. 31. Iverson GL. Interpreting change on the WAIS-III/WMS-III in clinical samples. Arch Clin Neuropsychol 2001;16:183-191. 32. Comi G, Filippi M, Martinelli V, et al. Brain MRI correlates of cognitive impairment in primary and secondary progressive multiple sclerosis. J Neurol Sci 1995;132:222-227. 33. Foong J, Rozewicz L, Chong WK, Thompson AJ, Miller DH, Ron MA. A comparison of neuropsychological deficits in primary and secondary progressive multiple sclerosis. J Neurol 2000;247:97-101. 34. Camp SJ, Stevenson VL, Thompson AJ, et al. Cognitive function in primary progressive and transitional progressive multiple sclerosis: a controlled study with MRI correlates. Brain 1999;122 ( Pt 7):1341-1348. 35. Lublin FD, Reingold SC, Cohen JA, et al. Defining the clinical course of multiple sclerosis The 2013 revisions. Neurology 2014;83:278-286.

36. Thompson AJ, Banwell BL, Barkhof F, et al. Diagnosis of multiple sclerosis: 2017 revisions of the McDonald criteria. Lancet Neurol 2018;17:162-173. 37. Lassmann H, van Horssen J, Mahad D. Progressive multiple sclerosis: pathology and pathogenesis. Nat Rev Neurol 2012;8:647-656. 38. Mahad DH, Trapp BD, Lassmann H. Pathological mechanisms in progressive multiple sclerosis. Lancet Neurol 2015;14:183-193. 39. Benedict RH, Weinstock-Guttman B, Fishman I, Sharma J, Tjoa CW, Bakshi R. Prediction of neuropsychological impairment in multiple sclerosis: comparison of conventional magnetic resonance imaging measures of atrophy and lesion burden. Arch Neurol 2004;61:226-230. 40. Benedict RH, Zivadinov R. Risk factors for and management of cognitive dysfunction in multiple sclerosis. Nat Rev Neurol 2011;7:332-342. 41. Barkhof F. The clinico-radiological paradox in multiple sclerosis revisited. Curr Opin Neurol 2002;15:239-245. 42. Chard D, Trip SA. Resolving the clinico-radiological paradox in multiple sclerosis. F1000Res 2017;6:1828. 43. Benedict RH, Weinstock-Guttman B, Fishman I, Sharma J, Tjoa CW, Bakshi R. Prediction of neuropsychological impairment in multiple sclerosis: comparison of conventional magnetic resonance imaging measures of atrophy and lesion burden. Arch Neurol 2004;61:226-230. 44. Hulst HE, Steenwijk MD, Versteeg A, et al. Cognitive impairment in MS: impact of white matter integrity, gray matter volume, and lesions. Neurology 2013;80:1025-1032. 45. Rocca MA, Amato MP, De Stefano N, et al. Clinical and imaging assessment of cognitive dysfunction in multiple sclerosis. Lancet Neurol 2015;14:302-317. 46. Steenwijk MD, Amiri H, Schoonheim MM, et al. Agreement of MSmetrix with established methods for measuring cross-sectional and longitudinal brain atrophy. Neuroimage Clin 2017;15:843-853. 47. Hedman AM, van Haren NE, Schnack HG, Kahn RS, Hulshoff Pol HE. Human brain changes across the life span: a review of 56 longitudinal magnetic resonance imaging studies. Hum Brain Mapp 2012;33:1987-2002. 48. Fotenos AF, Snyder AZ, Girton LE, Morris JC, Buckner RL. Normative estimates of cross-sectional and longitudinal brain volume decline in aging and AD. Neurology 2005;64:1032-1039. 49. Ziegler G, Dahnke R, Jancke L, Yotter RA, May A, Gaser C. Brain structural trajectories over the adult lifespan. Hum Brain Mapp 2012;33:2377-2389. 50. Steenwijk MD, Daams M, Pouwels PJ, et al. Unraveling the relationship between regional gray matter atrophy and pathology in connected white matter tracts in long-standing multiple sclerosis. Hum Brain Mapp 2015;36:1796-1807. 51. Azevedo CJ, Cen SY, Khadka S, et al. Thalamic atrophy in multiple sclerosis: A magnetic resonance imaging marker of neurodegeneration throughout disease. Ann Neurol 2018;83:223-234.

52. Prineas JW, Kwon EE, Cho ES, et al. Immunopathology of secondary-progressive multiple sclerosis. Ann Neurol 2001;50:646-657. 53. Franciotta D, Salvetti M, Lolli F, Serafini B, Aloisi F. B cells and multiple sclerosis. Lancet Neurol 2008;7:852-858. 54. Calabrese M, Filippi M, Gallo P. Cortical lesions in multiple sclerosis. Nat Rev Neurol 2010;6:438-444. 55. Klaver R, Popescu V, Voorn P, et al. Neuronal and axonal loss in normal-appearing gray matter and subpial lesions in multiple sclerosis. J Neuropathol Exp Neurol 2015;74:453-458. 56. Steenwijk MD, Geurts JJ, Daams M, et al. Cortical atrophy patterns in multiple sclerosis are non-random and clinically relevant. Brain 2016;139:115-126. 57. Calabrese M, Magliozzi R, Ciccarelli O, Geurts JJ, Reynolds R, Martin R. Exploring the origins of grey matter damage in multiple sclerosis. Nat Rev Neurosci 2015;16:147-158. 58. Franklin RJ, ffrench-Constant C, Edgar JM, Smith KJ. Neuroprotection and repair in multiple sclerosis. Nat Rev Neurol 2012;8:624-634. 59. Giorgio A, De Stefano N. Advanced Structural and Functional Brain MRI in Multiple Sclerosis. Semin Neurol 2016;36:163-176. 60. Sumowski JF, Chiaravalloti N, Wylie G, Deluca J. Cognitive reserve moderates the negative effect of brain atrophy on cognitive efficiency in multiple sclerosis. J Int Neuropsychol Soc 2009;15:606-612. 61. Stern Y. Cognitive reserve in ageing and Alzheimer's disease. Lancet Neurol 2012;11:1006-1012. 62. Benedict RH, Morrow SA, Weinstock Guttman B, Cookfair D, Schretlen DJ. Cognitive reserve moderates decline in information processing speed in multiple sclerosis patients. J Int Neuropsychol Soc 2010;16:829-835. 63. Sumowski JF, Rocca MA, Leavitt VM, et al. Brain reserve and cognitive reserve in multiple sclerosis: what you've got and how you use it. Neurology 2013;80:2186-2193. 64. Rimkus CM, Avolio IMB, Miotto EC, et al. The protective effects of high-education levels on cognition in different stages of multiple sclerosis. Mult Scler Relat Disord 2018;22:41-48. 65. Roosendaal SD, Schoonheim MM, Hulst HE, et al. Resting state networks change in clinically isolated syndrome. Brain 2010;133:1612-1621. 66. Rocca MA, Valsasina P, Absinta M, et al. Default-mode network dysfunction and cognitive impairment in progressive MS. Neurology 2010;74:1252-1259. 67. Hawellek DJ, Hipp JF, Lewis CM, Corbetta M, Engel AK. Increased functional connectivity indicates the severity of cognitive impairment in multiple sclerosis. Proc Natl Acad Sci U S A 2011;108:19066-19071. 68. Schoonheim MM, Meijer KA, Geurts JJ. Network collapse and cognitive impairment in multiple sclerosis. Front Neurol 2015;6:82. 69. Tona F, Petsas N, Sbardella E, et al. Multiple sclerosis: altered thalamic resting-state functional connectivity and its effect on cognitive function. Radiology 2014;271:814-821. 70. Hulst HE, Schoonheim MM, Roosendaal SD, et al. Functional adaptive changes within the hippocampal


memory system of patients with multiple sclerosis. Hum Brain Mapp 2012;33:2268-2280. 71. Tewarie P, Steenwijk MD, Brookes MJ, et al. Explaining the heterogeneity of functional connectivity findings in multiple sclerosis: An empirically informed modeling study. Hum Brain Mapp 2018;39:2541-2548. 72. Leavitt VM, Paxton J, Sumowski JF. Default network connectivity is linked to memory status in multiple sclerosis. J Int Neuropsychol Soc 2014;20:937-944. 73. Faivre A, Rico A, Zaaraoui W, et al. Assessing brain connectivity at rest is clinically relevant in early multiple sclerosis. Mult Scler 2012;18:1251-1258. 74. Sporns O. The human connectome: origins and challenges. Neuroimage 2013;80:53-61. 75. Gamboa OL, Tagliazucchi E, von Wegner F, et al. Working memory performance of early MS patients correlates inversely with modularity increases in resting state functional connectivity networks. Neuroimage 2014;94:385-395. 76. Rocca MA, Valsasina P, Meani A, Falini A, Comi G, Filippi M. Impaired functional integration in multiple sclerosis: a graph theory study. Brain Struct Funct 2016;221:115-131. 77. Stam CJ. Modern network science of neurological disorders. Nat Rev Neurosci 2014;15:683-695. 78. Crossley NA, Mechelli A, Scott J, et al. The hubs of the human connectome are generally implicated in the anatomy of brain disorders. Brain 2014;137:2382-2395. 79. Lohmann G, Margulies DS, Horstmann A, et al. Eigenvector centrality mapping for analyzing connectivity patterns in fMRI data of the human brain. PLoS One 2010;5:e10232. 80. Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. A default mode of brain function. Proc Natl Acad Sci U S A 2001;98:676-682. 81. Gusnard DA, Akbudak E, Shulman GL, Raichle ME. Medial prefrontal cortex and self-referential mental activity: relation to a default mode of brain function. Proc Natl Acad Sci U S A 2001;98:4259-4264. 82. de Haan W, Mott K, van Straaten EC, Scheltens P, Stam CJ. Activity dependent degeneration explains hub vulnerability in Alzheimer's disease. PLoS Comput Biol 2012;8:e1002582. 83. Rocca MA, Valsasina P, Hulst HE, et al. Functional correlates of cognitive dysfunction in multiple sclerosis: A multicenter fMRI Study. Hum Brain Mapp 2014;35:5799-5814. 84. Raichle ME. The brain's default mode network. Annu Rev Neurosci 2015;38:433-447. 85. Anticevic A, Cole MW, Murray JD, Corlett PR, Wang XJ, Krystal JH. The role of default network deactivation in cognition and disease. Trends Cogn Sci 2012;16:584-592. 86. Broyd SJ, Demanuele C, Debener S, Helps SK, James CJ, Sonuga-Barke EJ. Default-mode brain dysfunction in mental disorders: a systematic review. Neurosci Biobehav Rev 2009;33:279-296. 87. Gu H, Hu Y, Chen X, He Y, Yang Y. Regional excitation-inhibition balance predicts default-mode network deactivation via functional connectivity. Neuroimage 2018;185:388-397.

88. Buckner RL, Sepulcre J, Talukdar T, et al. Cortical hubs revealed by intrinsic functional connectivity: mapping, assessment of stability, and relation to Alzheimer's disease. J Neurosci 2009;29:1860-1873. 89. Sporns O. Contributions and challenges for network models in cognitive neuroscience. Nat Neurosci 2014;17:652-660. 90. Leech R, Sharp DJ. The role of the posterior cingulate cortex in cognition and disease. Brain 2014;137:12-32. 91. Eshaghi A, Marinescu RV, Young AL, et al. Progression of regional grey matter atrophy in multiple sclerosis. Brain 2018;141:1665-1677. 92. Power JD, Schlaggar BL, Petersen SE. Recent progress and outstanding issues in motion correction in resting state fMRI. Neuroimage 2015;105:536-551. 93. Ciric R, Wolf DH, Power JD, et al. Benchmarking of participant-level confound regression strategies for the control of motion artifact in studies of functional connectivity. Neuroimage 2017;154:174-187. 94. Power JD, Mitra A, Laumann TO, Snyder AZ, Schlaggar BL, Petersen SE. Methods to detect, characterize, and remove motion artifact in resting state fMRI. Neuroimage 2014;84:320-341. 95. Parkes L, Fulcher B, Yucel M, Fornito A. An evaluation of the efficacy, reliability, and sensitivity of motion correction strategies for resting-state functional MRI. Neuroimage 2018;171:415-436. 96. Pruim RHR, Mennes M, van Rooij D, Llera A, Buitelaar JK, Beckmann CF. ICA-AROMA: A robust ICA-based strategy for removing motion artifacts from fMRI data. Neuroimage 2015;112:267-277. 97. Gonzalez-Castillo J, Handwerker DA, Robinson ME, et al. The spatial structure of resting state connectivity stability on the scale of minutes. Front Neurosci 2014;8:138. 98. Shen K, Hutchison RM, Bezgin G, Everling S, McIntosh AR. Network structure shapes spontaneous functional connectivity dynamics. J Neurosci 2015;35:5579-5588. 99. Fox MD, Snyder AZ, Vincent JL, Corbetta M, Van Essen DC, Raichle ME. The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proc Natl Acad Sci U S A 2005;102:9673-9678. 100. Uddin LQ, Kelly AM, Biswal BB, Castellanos FX, Milham MP. Functional connectivity of default mode network components: correlation, anticorrelation, and causality. Hum Brain Mapp 2009;30:625-637. 101. Kelly AM, Uddin LQ, Biswal BB, Castellanos FX, Milham MP. Competition between functional brain networks mediates behavioral variability. Neuroimage 2008;39:527-537. 102. Gao W, Lin W. Frontal parietal control network regulates the anti-correlated default and dorsal attention networks. Hum Brain Mapp 2012;33:192-202. 103. Dixon ML, Andrews-Hanna JR, Spreng RN, et al. Interactions between the default network and dorsal attention network vary across default subsystems, time, and cognitive states. Neuroimage 2017;147:632-649. 104. Carter AR, Astafiev SV, Lang CE, et al. Resting interhemispheric functional magnetic resonance imaging connectivity predicts performance after stroke. Ann Neurol 2010;67:365-375.

105. MacDonald CL, Schwarze N, Vaishnavi SN, et al. Verbal memory deficit following traumatic brain injury: assessment using advanced MRI methods. Neurology 2008;71:1199-1201. 106. Andrews-Hanna JR. The brain's default network and its adaptive role in internal mentation. Neuroscientist 2012;18:251-270. 107. Buckner RL, Krienen FM, Yeo BT. Opportunities and limitations of intrinsic functional connectivity MRI. Nat Neurosci 2013;16:832-837. 108. Van Essen DC, Ugurbil K, Auerbach E, et al. The Human Connectome Project: a data acquisition perspective. Neuroimage 2012;62:2222-2231. 109. Glasser MF, Sotiropoulos SN, Wilson JA, et al. The minimal preprocessing pipelines for the Human Connectome Project. Neuroimage 2013;80:105-124. 110. Smith SM, Beckmann CF, Andersson J, et al. Resting-state fMRI in the Human Connectome Project. Neuroimage 2013;80:144-168. 111. Smith SM, Vidaurre D, Beckmann CF, et al. Functional connectomics from resting-state fMRI. Trends Cogn Sci 2013;17:666-682. 112. Barch DM, Burgess GC, Harms MP, et al. Function in the human connectome: task-fMRI and individual differences in behavior. Neuroimage 2013;80:169-189. 113. Liegeois R, Laumann TO, Snyder AZ, Zhou J, Yeo BTT. Interpreting temporal fluctuations in resting-state functional connectivity MRI. Neuroimage 2017;163:437-455. 114. Laumann TO, Snyder AZ, Mitra A, et al. On the Stability of BOLD fMRI Correlations. Cereb Cortex 2017;27:4719-4732. 115. Summers M, Swanton J, Fernando K, et al. Cognitive impairment in multiple sclerosis can be predicted by imaging early in the disease. J Neurol Neurosurg Psychiatry 2008;79:955-958.

116. Summers M, Fisniku L, Anderson V, Miller D, Cipolotti L, Ron M. Cognitive impairment in relapsing-remitting multiple sclerosis can be predicted by imaging performed several years earlier. Mult Scler 2008;14:197-204. 117. Deloire MS, Ruet A, Hamel D, Bonnet M, Dousset V, Brochet B. MRI predictors of cognitive outcome in early multiple sclerosis. Neurology 2011;76:1161-1167. 118. Rimkus CM, Schoonheim MM, Steenwijk MD, et al. Gray matter networks and cognitive impairment in multiple sclerosis. Mult Scler 2018:1352458517751650. 119. Schoonheim MM, Popescu V, Rueda Lopes FC, et al. Subcortical atrophy and cognition: sex effects in multiple sclerosis. Neurology 2012;79:1754-1761. 120. Schoonheim MM, Popescu V, Rueda Lopes FC, et al. Subcortical atrophy and cognition: sex effects in multiple sclerosis. Neurology 2012;79:1754-1761. 121. Beatty WW, Aupperle RL. Sex differences in cognitive impairment in multiple sclerosis. Clin Neuropsychol 2002;16:472-480. 122. Filippi M, Preziosa P, Copetti M, et al. Gray matter damage predicts the accumulation of disability 13 years later in MS. Neurology 2013;81:1759-1767. 123. Bergsland N, Horakova D, Dwyer MG, et al. Subcortical and cortical gray matter atrophy in a large sample of patients with clinically isolated syndrome and early relapsing-remitting multiple sclerosis. AJNR Am J Neuroradiol 2012;33:1573-1578. 124. Minagar A, Barnett MH, Benedict RH, et al. The thalamus and multiple sclerosis: modern views on pathologic, imaging, and clinical aspects. Neurology 2013;80:210-219. 125. Damjanovic D, Valsasina P, Rocca MA, et al. Hippocampal and Deep Gray Matter Nuclei Atrophy Is Relevant for Explaining Cognitive Impairment in MS: A Multicenter Study. AJNR Am J Neuroradiol 2017;38:18-24.

Frequency of cognitive impairment in multiple sclerosis english.pdf · answers to the following...

Documents

Transcript of Frequency of cognitive impairment in multiple sclerosis english.pdf · answers to the following...