fornito2015

The ancient Roman physician Galen was one of the first to propose that pathology in one part of the nervous system could affect other regions when he posited that animal spirits could flow through interconnecting neural path-ways1,2. This hypothesis was revisited nearly two millen-nia later by Brown-Squard, who suggested that the effects of focal brain damage on remote regions resulted from actions at a distance3. von Monakow extended the concept, and coined the term diaschisis (derived from Greek and meaning shocked throughout) to describe the depression of function that can arise in undamaged brain regions that are connected to a lesioned site2,4 (reviewed in REF.5). At a similar time, Wernicke proposed an associative theory of brain function, in which higher-order cognitive processes arose from the integration of multiple, spatially distrib-uted neural systems and in which disorders as diverse as aphasia and schizophrenia resulted from the disruption of specific associative pathways6 (see also REF.7). Wernickes contemporaries, including Hughlings Jackson, Meynert, Flechsig, Dejerine and Lichtheim, also viewed connectiv-ity to be central to any understanding of CNS pathology. Their work paved the way for Geschwinds introduction of the disconnexion syndrome and the concomitant expan-sion of the range of clinical symptoms that may now be attributed to disordered brain connectivity8,9. These develop ments complemented the emergence of neural-systems-based accounts of perception, cognition and emotion10,11 (for a detailed review, see REF.7).

Despite these early insights, a major focus of twenti-eth-century clinical neuroscience was the localization of psychological processes to specific areas of the brain.

From this perspective, the behavioural impairments that arise from neural insult were thought to emerge from damage to discrete and specialized brain regions, as exemplified by famous clinical case studies such as Brocas Leborgne12, Harlows Phineas Gage13, and Scoville and Milners H.M.14. However, this view provides only a partial account of brain function. The brain is a highly complex, interconnected network that balances regional segregation and specialization of function with strong integration15,16, a balance that gives rise to complex and precisely coordinated dynamics across multiple spatio-temporal scales17. A generic property of any network is that dysfunction can spread easily between linked ele-ments, leading to pathological cascades that can encom-pass large swathes of the system18. In the brain, axonal and synaptic contacts can act as conduits for the propaga-tion of disease processes. This is exemplified by the rapid spread of focal epileptogenic activity into generalized seizures19, the spatially distributed activation changes resulting from regionally localized ischaemic insults20 and the gradual progression of pathology in degenera-tive diseases that are thought to have focal onset, such as Huntington disease, Parkinson disease and other forms of neurodegeneration21,22.

An important first step in understanding how neural network organization influences the onset, expression and course of disease is the generation of a compre-hensive map a connectome23 of the connectivity architecture of the brain (BOX1). This goal has driven several recent, large-scale collaborative endeavours that are unprecedented in neuroscience2426, and has led to

1Monash Clinical and Imaging Neuroscience, School of Psychological Sciences and Monash Biomedical Imaging, Monash University, Clayton, Victoria, Australia 3168.2Melbourne Neuropsychiatry Centre and Melbourne School of Engineering, The University of Melbourne, Parkville, Victoria, Australia 3053.3Systems Neuroscience Group, QIMR Berghofer Medical Research Institute, Herston, Queensland, Australia 4029.4Metro North Mental Health Service, The Royal Brisbane and Womens Hospital, Herston, Queensland, Australia 4029.Correspondence to A.F. email: [email protected]:10.1038/nrn3901

The connectomics of brain disordersAlex Fornito1, Andrew Zalesky2 and Michael Breakspear3,4

Abstract | Pathological perturbations of the brain are rarely confined to a single locus; instead, they often spread via axonal pathways to influence other regions. Patterns of such disease propagation are constrained by the extraordinarily complex, yet highly organized, topology of the underlying neural architecture; the socalled connectome. Thus, network organization fundamentally influences brain disease, and a connectomic approach grounded in network science is integral to understanding neuropathology. Here, we consider how brainnetwork topology shapes neural responses to damage, highlighting key maladaptive processes (such as diaschisis, transneuronal degeneration and dedifferentiation), and the resources (including degeneracy and reserve) and processes (such as compensation) that enable adaptation. We then show how knowledge of network topology allows us not only to describe pathological processes but also to generate predictive models of the spread and functional consequences of brain disease.

R E V I E W S

NATURE REVIEWS | NEUROSCIENCE VOLUME 16 | MARCH 2015 | 159

2015 Macmillan Publishers Limited. All rights reserved

Diffusion tractographyAn MRI technique for reconstructing large-scale white-matter fibres based on the preferential diffusion of water along the axes of these fibres.

Hierarchical modularityThe nested organization of highly interconnected subsets, or modules, of nodes within a network, such that modules are contained within modules and so on, across multiple scales of organization.

Box 1 | Mapping the human connectome

MRI is the tool most widely used to map structural and functional properties of the human connectome. MRI connectomics involves three main steps. First, network nodes are defined, commonly using one of many heuristic approaches (see REF.148 for a discussion of these methods; see the brain below figure parts a, b on which the green spheres depict an example of apriori-defined regions of interest). Second measures of structural, functional or effective connectivity between nodes are quantified. Both functional connectivity and effective connectivity can be measured during active task performance119,151153 or task-free (resting) states142,154156 using functional MRI (see the figure, part a). Structural connectivity can be measured using diffusion tractography (see REFS148,150 for limitations; see the figure, part b149).

The final step involves analysis of either network connectivity or topology77. Connectivity analyses examine variations in the type and strength of connectivity between brain regions, and can be carried out at the level of candidate neural systems (candidate systems analysis) or across the entire brain (connectome-wide analysis)157. In the latter case, whole-brain connectivity can be succinctly represented as a matrix in which each row and column represents a different region, and each matrix element [i, j] encodes the type and strength of connectivity between region pairs (see the figure, parts c, d). The network can also be represented as a graph that comprises nodes (which represent regions) connected by edges (which represent connections; see the figure, parts e, f, in which line thickness represents strength of connectivity). In MRI analyses, connectivity matrices and graphs are often undirected (that is, they represent the presence but not direction of a connection) and weighted to reflect variations of inter-regional connectivity strength (see the figure, parts c, e). Models of effective connectivity can resolve directionality (see the figure, parts d, f).

Topological analyses are grounded in graph theory15 and have uncovered several non-trivial characteristic properties of the brain. These properties include a short average path-length between nodes, which enables efficient communication (that is, a parallel and integrated topology), coupled with low wiring cost158,159; a heavy-tailed distribution of regional connectivity, which implies the existence of highly connected hub regions27,111; strong interconnectivity of hub regions, defining a topologically central core, or rich club113,126; and high clustering and hierarchical modularity across multiple resolution scales, which supports functional specialization160. Part g of the figure illustrates some of these key topological properties of brain networks; namely, the shortest path length between two nodes at opposite ends of the network (left; path length shown in orange), clustered connectivity (middle; orange lines represent edges linking a cluster of nodes), modular organization (right; modules outlined by grey shading) and rich-club organization (right; rich-club edges in orange). Image in part b is adapted, with permission, from REF.149, PLoS http://creativecommons.org/licenses/by/3.0/.

Nature Reviews | Neuroscience

a Functional connectivity b Structural connectivity d Directed matrix

i

j

NN

123...

1 2 3 . . .

f Directed graphe Undirected graph

g Topological analysis

c Undirected matrix

i

j

NN

123...

1 2 3 . . .

R E V I E W S

160 | MARCH 2015 | VOLUME 16 www.nature.com/reviews/neuro


Graph theoryA branch of mathematics concerned with studying networks of connected elements. With graph theory, a brain network can be modelled as a graph of nodes (depicting single neurons, neuronal populations or macroscopic brain regions) linked by edges (depicting inter-regional structural or functional interactions).

the construction of increasingly detailed maps of brain connectivity at various resolutions in diverse species2730. However, the sheer scale of the data sets involved poses difficulties for analysis and interpretation. The human brain comprises an estimated 8.6 1011 neurons and approximately 1014 synapses31,32, a digital atlas of which would require more memory than is required to store all the written information present in the world today33. Representing, analysing and interpreting these data is challenging even at the macro-scale resolutions that are accessible with current invivo MRI; on the order of ~102 to ~104 nodes (that is, brain regions) and ~5,000 to ~5 106 connections.

Network science and graph theory offer powerful tools for overcoming these challenges to map, track and predict patterns of disease spread (BOX1; FIG.1). The application of these techniques has enabled detailed descriptions of how disease affects the brain, and has uncovered new insights into the shared characteristics of diverse disease processes based on fundamental prop-erties of brain network organization34,35. More recent conceptual and technical advances mean that the field is now ready to move beyond mere descriptions of dis-ease processes to generate hypotheses about underlying pathophysiological mechanisms and clinically useful predictions concerning key prognostic indicators.


1

2

a b c d e

3

4

f g Group A Group B Group A Group B

Normal Abnormal At risk

Eective connectivity Quantitative dierence Qualitative dierence

Figure 1 | Connectomics can track and predict patterns of disease spread. a|Typically, a brain-imaging experiment may compare a patient with a control population using a measure of neural structural or functional integrity across many locations in the brain. In this example, an analysis identifies two brain regions, 1 and 2, that exhibit significant groupdependent differences. These differences can be presented as a map that localizes the abnormalities but offers no information regarding the interplay between these putative pathophysiological markers. b|Mapping the connectivity of these regions reveals that the abnormalities occur within a broader network context. In this example, there is no direct connection between regions 1 and 2, suggesting that pathology may affect these two areas independently; parts ce depict scenarios in which regions 1 and 2 are directly connected. c|Regions 1 and 2 are abnormal, but the connection between them is intact. In this case, pathology may have originated in one area and affected the other via aberrant signalling along the intact pathway. Identification of the primary abnormality here would require either longitudinal data or an investigation of effective connectivity (f) (see also BOX1). The connection linking regions 1 and 2 is considered to be at risk of deterioration because it is interposed between two dysfunctional regions. d|Regions 1 and 2, and the connection linking them, are abnormal, suggesting that there may be a direct association between the two regional abnormalities. In this case, a primary pathology in either region may have resulted in secondary deterioration of their connecting pathway and, subsequently, the other area. Alternatively, the pathology may have originated in the axonal tracts that link regions 1 and 2, and subsequently caused dysfunction in both regions. Again, longitudinal analysis would be required to distinguish between these possibilities. e|Dysfunction in regions 1 and/or 2 may alter their connectivity with other regions (red connections linking to regions 3 and 4). In this case, regions 3 and 4 are considered to be at risk of impairment given prolonged exposure to the aberrant signals emanating from regions 1 and 2. f|Study of effective connectivity can provide further clues regarding the primary source of pathology. For example, if region 2 influences 1 but not vice versa (left), then pathology is more likely to have originated in region 2. g | Casecontrol differences in connectivity can be quantitative (left) or qualitative (right). Aquantitative difference occurs when patients and controls share the same underlying connectivity architecture, but show a difference in the strength of connectivity (represented by line thickness) between specific pairs of brain regions. A qualitative difference refers to a distinct pattern of connectivity in the patient group; for instance, a fibre bundle may be present in patients but not controls. Such differences can result from abnormal wiring of the connectome (BOX2).

R E V I E W S



Network topologyThe way in which the connections of a network are organized with respect to each other.

Functional connectivityA statistical dependence (such as a correlation) between neurophysiological recordings acquired from distinct brain regions.

In this Review, we consider how these goals can be achieved by understanding the various responses of the brain to pathological perturbation, and how neural network topology constrains these responses (BOX1). We examine well-known examples of adaptive and maladap-tive neural responses to insult, and illustrate how brain connectivity shapes their expression at the level of large-scale neural systems. We then show how knowledge of network topology can be used to characterize and model vulnerability and resilience to disease and dysfunction.

Maladaptive responses and pathological spreadConnectomics offers a powerful analytic framework for localizing pathology, tracking patterns of disease spread and predicting which areas will be affected next (FIG.1). However, simply tracking the spread of a disease will not necessarily elucidate the mechanisms through which this spread occurs. Such mechanisms may be construed as maladaptive, as they compound the degree of functional compromise that results from the insult. Here, we con-sider three major types of maladaptive response that can mediate the spread of pathology throughout the con-nectome: diaschisis, transneuronal degeneration and dedifferentiation (FIG.2).

Diaschisis. von Monakow defined diaschisis as a tempo-rary interruption of function in regions that are remote from an injured site4 (FIG.2a). He attributed this interrup-tion to a deafferentation of excitatory input to the remote area. Although his definition identifies the changes as transient, von Monakow subsequently acknowledged that more persistent forms of diaschisis are also possi-ble2,5. Diaschisis is now a well-established phenomenon, particularly following stroke. It has been observed in the forebrain after damage to the brainstem or cerebellum, in cortical regions following subcortical infarction and in contralesional cortex following focal cortical insult20 (reviewed in REF.5; see also REF.20). These distributed changes seem to be circuit-selective. For example, one study found that lesions to either the fronto-parietal or the cingulo-opercular network two dissociable neural systems involved in cognitive control affected con-nected areas within the same system but not the func-tions of the other network36. Animal models of crossed cerebellar diaschisis, in which cerebellar function is depressed after a focal lesion in the contralateral cor-tex, have shown that cerebellar neurons do indeed show reduced spiking output after the insult but maintain normal levels of excitability37. This result suggests that the functional depression of the cerebellum is caused by reduced excitatory drive from the damaged cortex, afinding that is consistent with von Monakows original deafferentation hypothesis. Whole-brain computational modelling has also suggested that focal lesions can have adiffuse effect on inter-regional synchronization dynam-ics that extend well beyond the affected site, and in a way that critically depends on the connection topology of the damaged region38,39 (FIG.3). Accordingly, the concept of diaschisis has recently been extended to include altera-tions of functional connectivity between areas that may not even be directly linked to the lesionedarea40.

Studies of patients who have suffered stroke suggest that the severity of behavioural impairment that follows focal neural damage often correlates with the extent of activation and connectivity changes in regions remote from the injured site41,42. These associations between behaviour and altered network functional connectivity occur even if anatomical connectivity between damaged and undamaged regions is intact43. This finding suggests that a functional deafferentation of remote sites may be sufficient to impair behaviour. Nonetheless, dam-age to anatomical pathways linking the lesioned area to unaffected regions seems to compound the severity of behavioural impairment42,44.

Diaschisis may manifest only during the performance of certain tasks (that is, it may be context-dependent)45 and may vary as a function of the type of lesion sustained46. In some cases, the correlation between the severity of behavioural impairment and the extent of insult-related functional connectivity changes in remote regions can-not be explained by the characteristics of the lesion to the damaged area alone47 and performance improvements on key behavioural tasks correlate with the return of normal function in the affected network4244. Together, these find-ings indicate that the behavioural impairments that arise from damage to the CNS may often be the result of how the insult affects distributed neural dynamics, rather than of its impact on the lesioned sitealone.

Transneuronal degeneration. In contrast to diaschi-sis, which is an interruption of function in a region that is remote from a lesion, transneuronal degenera-tion is astructural deterioration of areas remote from the initial insult. It is a process that evolves over time, and therefore can only be characterized longitudinally (BOX2; FIG.2b). Transneuronal degeneration can be either anterograde (whereby damage or dysfunction of one neuron causes the degeneration of its postsynaptic target) or retrograde (whereby a presynaptic neuron deteriorates as a result of reduced trophic support from an injured or necrotic postsynaptic target)48. The form of degeneration can vary, and encompasses changes such as: neuronal shrinkage; reductions in dendrite and synapse number; alterations of axonal myelin content and fibre number; and neuronal death48.

Both anterograde and retrograde degeneration have been identified in numerous neural circuits4850. For example, studies of the visual system in humans and ani-mals have shown anterograde changes in the optic tract, lateral geniculate nucleus and striate cortex following surgical damage of the retina or optic nerve, or following surgical closure of the eyes5153. By contrast, retrograde degeneration in the lateral geniculate and retina has been reported in monkeys following lesions of the visual cor-tex54,55. Animal models also indicate that transneuronal degeneration is generally more severe when damage is sustained at an early age51,54, highlighting how the stage of development may influence neural responses to insult (BOX2).

Several mechanisms can mediate transneuronal degeneration. As a general rule, pathology in any single area can disrupt interactions with other regions, causing

R E V I E W S



irregular firing and metabolic stress in the connected site56. Degeneration of remote regions may also result from diminished excitatory input or a loss of trophic sup-port from damaged presynaptic neurons57. For example, in Alzheimer disease, the accumulation of amyloid- in specific brain regions reduces their functional connectiv-ity with other areas58 and may cause hypometabolism in the distal sites59, rendering them likely targets for disease propagation60. Thus, diaschisis may precede transneu-ronal degeneration in some disorders.

It is also possible for focal pathology to disinhibit activity and cause cell death or damage in remote sites owing to excess neuronal stimulation. Such excitotoxic-ity plays a central part in the damage sustained to remote areas following focal cerebral ischaemia61 and may underlie the distributed degenerative changes seen in focal epilepsy19 and amyotrophic lateral sclerosis (ALS). In the specific case of ALS, hyperexcitability of cortical motor neurons early in disease progression is thought to damage monosynaptically connected neurons in the

Figure 2 | Differentiating major classes of adaptive and maladaptive neural responses to pathological perturbation. The expected pattern of systemslevel changes associated with maladaptive (parts ac) and adaptive (parts df) responses, shown in a simple network of four nodes that is specialized for a particular behavioural output. Each panel illustrates what happens when one node is damaged as the result of some generic insult (black). The text below each illustration lists some candidate mechanisms that may cause, facilitate or exacerbate each response. We focus here on pathology of nodes and ignore changes in connectivity, for simplicity. Note that the spatial location of nodes is arbitrary and the effect of internode anatomical distances on these responses remains unclear. In a maladaptive response (parts ac), the output of the network (for instance, behaviour) is compromised. a | Diaschisis occurs when a focal lesion depresses the function of remote, connected sites (red nodes). b | Transneuronal degeneration occurs when, over time (arrow) there is a structural deterioration of areas connected to the affected site (additional black nodes). This deterioration can only be identified via longitudinal analysis. Depicted here is a case in which initial diaschisis evolves into degeneration of the previously dysfunctional nodes. c | Dedifferentiation must be understood by considering how one specialized system that is damaged interacts with other areas. The characteristic pattern will involve reduced function of the neural system that supports the impaired behaviour (red nodes), and a diffuse increase of activity in other neural systems that are not typically associated with that behaviour (blue nodes), thus reflecting a breakdown of normal functional specialization. In an adaptive response (parts df), behaviour or task performance is compensated for or preserved. d|Compensation occurs when either undamaged nodes within the impaired system, or nodes of other systems, increase their activity or connectivity to preserve behaviour (the former scenario is depicted here). e|Neural reserve is evident when the activity in remaining elements of the affected system is unchanged and behavioural performance is intact. f | Degeneracy is evident when a second system can support the behaviour that is normally mediated by the dysfunctional network, without any substantial change in the activity of this second system (see also REF.104 for a discussion of other manifestations of degeneracy). Note that complex variations of these responses are possible. For example: compensation may be incomplete, only partially restoring function85; more than one of these responses may occur simultaneously; and one response may evolve into another over time, underscoring the importance of a longitudinal perspective (BOX2).


a DiaschisisDysfunction in remote regions

b Transneuronal degenerationDegeneration of remote regions

c DedierentiationDesegregated, non-specialized activity

Deaerentation Aberrant synchronization

Excitotoxicity Axonal transort Prion-like spread

Altered neurodevelopment Impaired neuromodulation Decient plasticity

Maladaptation

e Neural reserveIntact tissue sucient to support function

f DegeneracyOther systems assume function

d CompensationIncreased function in unaected areas

Limited pathology High reserve

Neurodevelopment Experience-dependent plasticity

Plasticity Cognitive exibility

Adaptation

Normal Lesioned Reduced activity Increased activity

R E V I E W S



Effective connectivityThe causal influence that one neuronal system exerts on another. Its measurement often requires a model of the neuronal dynamics causing variations in measured neural signals.

anterior horn of the spinal column62. This hypothesis is supported by longitudinal studies of patients with ALS that show that the brain regions at the highest risk of degeneration are connected to areas already showing degenerative changes63.

Fast axonal transport is another contributor to transneuronal degeneration, and has been implicated in the pathogenesis of several neurodevelopmental and neu-rodegenerative disorders, including ALS, Alzheimer dis-ease, Huntington disease, hereditary motor neuropathy and hereditary spastic paraplegia, multiple sclerosis and CharcotMarieTooth disease57,64,65. Molecular motors continually shuttle organelles, lipids, mitochondria, neurotrophins and other molecules via microtubules and neurofilaments that link the soma and distal segments of the axon. This axonal transport is necessary to sus-tain neuronal activity and integrity, to meet metabolic demands and to enable the clearance of misfolded and/or aggregated proteins57,65. Thus, neuronal cell bodies and axons depend on each other for survival. Pathology at the soma can disrupt anterograde transport of cargo that is necessary for the maintenance and repair of the axonal cytoskeleton and surrounding myelin sheath. Conversely, primary pathology of white matter can inhibit the retro-grade transport of trophic factors that are essential for neuronal survival64,66. Transport mechanisms may also

aid the suggested prion-like spread of tau and other pathologies in certain neurodegenerative diseases67. Such a mechanism may explain the spatial colocaliza-tion of regional atrophy in neurodegenerative disease with the topography of structural and functional brain networks68,69 (FIG.4).

Dedifferentiation. Dedifferentiation is the diffuse, non-specific recruitment of brain regions to perform a task (FIG.2c), and is thought to result from a break-down of usually specialized and segregated neural activ-ity70,71. Dedifferentiation may be caused by aberrant neural plasticity or by a focal cortical pathology that disrupts the balance between excitation and inhibition within discrete neural systems. For instance, analysis of effective connectivity inferred from functional MRI (fMRI) data suggests that the increased activation of the contralesional primary motor cortex (M1) that is often observed in patients with subcortical stroke20 arises from complex alterations of excitatory and inhibitory interactions between the left and right cortical motor systems72. Persistence of this dedifferentiated state is associated with poorer recovery of motor function fol-lowing stroke72, and recovery can be improved through the use of inhibitory repetitive transcranial magnetic stimulation (TMS) of the contralesional M1 (REF.73).

Figure 3 | Network topology constrains the distributed effects of focal lesions on brain dynamics. A model of the widespread effects that focal lesions can have on brain functional connectivity. An average connectome comprising 988 regions of interest was generated from diffusion tractography in five healthy individuals173. Lesions were simulated by deleting all of the structural connections that linked 50 contiguous regions that were centred on an area with high topological centrality (green nodes, part a) or 50 contiguous regions that were centred on an area with low topological centrality (green nodes, part b). A neural mass model and haemodynamic response function174 were used to simulate regional fluctuations in BOLD (blood-oxygenation-level-dependent) signal arising from the structural network before and after each lesion. Edges between nodes depict signficicant reductions (shown in red) and increases (shown in blue) of interregional functional connectivity resulting from each lesion. The lesion affecting the topologically central posterior cingulate precuneus region in part a caused widespread changes of functional connectivity that were characterized by a complex pattern of increases and decreases, unlike the lesion to the less central temporal pole depicted in part b. To determine whether the functional connectivity increases reflect diaschisis or dedifferentiation of neural activity requires an understanding of how the brain changes relate to behavioural impairment (see FIG.2). Left hemisphere is depicted on left. Figures were generated using methods and structural connectivity data from REF.39.


a

b

Posterior cingulate and precuneus

Temporal pole

R E V I E W S



NeuromodulationThe regulation of neuronal activity by ascending neurotransmitter systems.

Another possible cause of dedifferentiation is the disruption of ascending neuromodulatory systems that tune the signal-to-noise ratio of neural information pro-cessing70. Such a mechanism may explain the diffuse patterns of task-related activation that are often seen in ageing populations compared with those observed in younger volunteers71, and may also underlie the distrib-uted changes in neural activation and connectivity that have been reported in neurodevelopmental disorders such as schizophrenia74. For example, one meta-analysis of fMRI activation changes in the brains of people with schizophrenia who were performing executive function tasks found reduced activity in cortico-subcortical net-works that are typically associated with executive abili-ties, and increased activation of regions outside these canonical neural systems75. This pattern is consistent

with a dedifferentiation hypothesis (FIG.2c). In support of this view, abnormal neuromodulation and an altered signal-to-noise ratio of cortical function are well-described characteristics of schizophrenia76 and may arise from neurodevelopmental abnormalities of the wiring of the connectome in that disorder77,78.

Adaptive responses to neural insultThe brain can also respond to pathological perturba-tion in an adaptive manner to maintain homeostasis and performance where possible79. Here, we review three concepts that are crucial for the ability of the brain to respond adaptively to pathology: compensation, degeneracy and reserve (FIG.2).

Compensation. Increases in activity or functional con-nectivity following a pathological insult that preserve behavioural output are commonly attributed to neu-ral compensation (FIG.2d). For example, in patients with stroke a focal ischaemic insult often results in the extensive recruitment of unaffected, remote brain areas8083. Similar effects can be observed in healthy volunteers after inhibitory stimulation of focal cortical areas with TMS84. In general, the extent of focal neural damage and the severity of behavioural impairment correlate with greater compensatory recruitment81,85, functional reorganization86 and altered functional connectivity82 of remote areas. Accordingly, func-tional recovery is compromised if there is damage to the axonal tracts that link damaged regions to unaf-fected areas with the capacity for compensation42. Compensatory adaptations can persist for long peri-ods of time and can preserve behaviour to varying degrees80,84. The causal role of compensatory activation in preserving behaviour was shown by one study that found that inhibitory TMS of the right dorsal premo-tor cortex could disrupt right-handed motor function in patients with ischaemic damage to the left motor system85. In healthy controls, right dorsal premotor cortex stimulation did not affect motor performance, suggesting that this region activates to support motor function only if the left premotor cortex is damaged. This effect has been mimicked in healthy individu-als after repetitive TMS was used to inhibit activity in the left dorsal premotor cortex. Following this virtual lesion, a second stimulation pulse applied to the right premotor cortex as these participants undertook a motor task was found to disrupt performance84.

Progressive normalization towards pre-injury activity patterns in affected networks also predicts behavioural recovery72,80,87, suggesting that optimal recovery may depend on a gradual return to baseline network dynamics. Indeed, persistent hyperactivation in some regions may place neurons under undue metabolic stress, reducing their viability and rendering them susceptible to degen-eration56,88. This mechanism may explain the consistent reports of increased task-evoked activation or functional connectivity in the brains of patients with early-stage neu-rodegenerative disease whether it is Alzheimer disease, Huntington disease or multiple sclerosis followed by declines in these measures at later disease stages8992.

Box 2 | Timing is everything

The age at which an insult to the brain occurs critically influences the outcome of the injury. In some cases, the enhanced plasticity of the developing brain affords greater capacity for recovery from injuries sustained earlier in life than from those sustained later in life161. For example, children with congenital damage to the left hemisphere may show age-appropriate language development and a relatively lower incidence of the aphasic symptoms that are seen in adults who have sustained damage to similar brain areas later in life162. In other cases, certain types of insult that occur prenatally or in the first few years of life can result in more severe functional impairment than do later injuries (reviewed in REF.163).

This variability may be partly explained by the timing of the insult relative to developmentally critical periods164,165; highly regulated, circuit-specific maturational periods that are characterized by exquisite sensitivity to environmental inputs167,168. These periods coincide with the activity-dependent elimination of excess synapses and the consolidation of long-range axonal projections. These processes may underlie the dynamic changes in the topological organization of the connectome that occur throughout childhood and adolescence. Such changes include a reduction in modularity, an increase in topological integration (that is, a reduction in the average minimum path-length; BOX1), an increase in the number of connector hubs (FIG.5b), and a shift from functional modules that comprise spatially adjacent regions to interconnected systems that extend over long distances166168. The developmental trajectories of these topological modifications are system-specific: sensorimotor and limbic systems develop adult-like topological properties by late childhood, whereas the connection topology of associative areas continues to mature throughout adolescence167. The increased capacity for topological integration during critical periods may promote enhanced degeneracy in the face of insult, as such integration facilitates the formation and recruitment of alternative networks to sustain a given function169. However, this degeneracy can be realized only if appropriate environmental inputs are available; if these inputs are not available owing to deprivation or disease degeneracy will be limited, resulting in marked and persistent impairments51,165,170.

Early damage to the brain may also interfere with subsequent maturational processes, derailing the development of a normal brain-network topology. For example, in rodent neurodevelopmental models of schizophrenia, early focal damage of the hippocampus causes changes in the structure and function of the prefrontal cortex that only emerge in adolescence171, which in humans is when the disease usually first presents. Thus, the distributed impact of an early focal insult may manifest only as connected regions mature or come online, making the timing of clinical symptoms difficult to predict.

Early insults may also provide a disordered foundation for subsequent neurodevelopment, resulting in a mis-wiring of the connectome. For example, abnormal development of the corpus callosum in humans often results in the emergence of longitudinal fibres (such as Probust bundles) that run parallel to the inter-hemispheric fissure and that are not seen in healthy controls172. Such mis-wiring will result in a higher frequency of qualitative differences in brain connectivity (that is, the presence of a set of connections in patients that is not apparent in healthy individuals; see FIG.1). The functional significance of such qualitative changes remains an important topic for further investigation.

R E V I E W S



This pattern of increased activity followed by subsequent degeneration has been mimicked in computational mod-els88 and may reflect evidence of an early, adaptive and plastic response that is gradually overwhelmed as patho-logical burden increases91. In this regard, compensation could precede subsequent decline and transneuronal degeneration in somecases.

Structural plasticity is an important neural substrate for compensation93. Animal models94 and human studies95 have shown that focal ischaemic damage can cause wide-spread depolarization of connected regions particularly homotopic contralateral areas resulting in persistent hyperexcitability or disinhibition of functionally related, but spatially distributed, networks94,96. Hyper excitability in these connected areas can be accompanied by increased synaptogenesis, and axonal and dendritic sprouting of undamaged axons97100. This remodelling can take place over long distances, is activity-dependent99,100 and can cause volumetric changes detectable with MRI101. These plastic changes may afford greater flexibility in adopt-ing alternative strategies to preserve behaviour as much as possible.

Degeneracy and reserve. Degeneracy is the capacity of structurally distinct elements of a system to carry out the same function102; in other words, it describes the ability of distinct neuronal systems to make overlapping contribu-tions to the same output, offering both functional adapt-ability and robustness to damage102 (see also REF.103). Degeneracy is a necessary condition for compensation: compensatory activity is simply not possible if other neu-ral systems cannot assume the functions of a compro-mised network (that is, if degeneracy is low). Degeneracy may manifest in different ways. For instance, for any given task, multiple neural systems may become activated in a parallel and redundant way to support performance; only one of several degenerate systems may become activated at any one time (reflecting the use of different strategies); or only one system may be consistently activated while other systems remain latent until they are unmasked by dysfunction of the canonical network104.

Degeneracy is apparent at multiple levels of the neural hierarchy, from the scale of individual neurons and microcircuits (reviewed in REF.79) to large-scale macroscopic systems. An example that is evident at the level of macroscopic systems is given by two function-ally segregated systems that are known to support read-ing ability. One network involves left inferior frontal and anterior occipito-temporal regions and the other comprises right inferior parietal and left posterior occipito-temporal areas105. Reading performance is impaired following damage to elements of both net-works, but not after damage to either network alone106. Therefore, each system is sufficient, but neither is nec-essary, for reading. This implies that the two systems are degenerate, with each being capable of achieving comparable behavioural output. It remains unclear whether, in such circumstances, reliance on just one of the degenerate systems has secondary consequences for other functions.

Degeneracy provides a neural network basis for cog-nitive reserve the ability to flexibly engage alterna-tive cognitive or compensatory strategies to deal with a behavioural impairment caused by neural insult107,108. The ability to engage such alternative strategies is an important component of compensation, and the extent to which such strategies are possible depends on degen-eracy104. In general, we can expect a higher level of neural degeneracy to result in a greater cognitive reserve and thus a greater capacity for compensation107. It is there-fore expected that patients with higher degeneracy and reserve will be better positioned to adapt to the functional impairments arising from a cerebral insult107.

Cognitive reserve also depends on neural reserve, which refers to the amount of remaining intact brain tissue that can still carry out a given task. Generally, a brain with high neural reserve will be able to withstand greater damage before cognitive or behavioural deficits manifest107 (FIG.2e). Although degeneracy, compensa-tion and reserve are closely related, degeneracy does not necessarily imply that compensatory (increased) activ-ity of unaffected regions will occur following an insult (FIG.2f). As the example of the reading system shows, each of two distinct neural systems may be sufficient to

Figure 4 | Transneuronal degeneration in brain networks. Neurodegeneration spreads throughout functionally connected neural systems in Alzheimer disease (AD), behavioural-variant frontotemporal dementia (bvFTD), semantic dementia (SD), progressive non-fluent aphasia (PNFA) and corticobasal syndrome (CBS)69. a|Spatial patterns of diseasespecific greymatter atrophy (shown in blue) in each disease. b|Resting-state networks (shown in yellow), mapped in healthy individuals, that are functionally connected to the seed regions located in areas of maximal atrophy for each disease (represented by the red circles in part a). c| Regional grey-matter volume covariance (shown in green) with these same seed regions. The spatial correspondence between these networks suggests that degeneration occurs within functionally and structurally connected networks. Lighter colour shades indicate higher statistical significance of the depicted association. LIFG, left inferior frontal gyrus; LTPL, left temporal pole; RANG, right angular gyrus; RFI, right frontoinsular cortex; RPMC, right premotor cortex. Reprinted from Neuron, 62, Seeley, W. W., Crawford, R. K., Zhou, J., Miller, B. L. & Greicius, M. D. Neurodegenerative diseases target large-scale human brain networks. 4252, Copyright (2009), with permission from Elsevier.


a Syndrome-specic grey-matter atrophy

b Spontaneous functional connectivity in controls

c Structural covariance in controls

AD bvFTD SD PNFA CBS

RANG RFI LTPL

LIFGRPMC

R E V I E W S



Degree distributionThe distribution of degree values obtained across network nodes.

Network fragmentationThe splitting of a network into disconnected subsets of nodes. The lack of connectivity between these subsets precludes any communication between them, meaning the nodes no longer function as an integrated system.

Structural connectivityThe physical connections (that is, axonal fibres) between brain regions.

carry out a task, and failure of one need not necessarily increase activation of the other. Increased or atypical activation will occur only when degeneracy is partial or incomplete.

From description to prediction via topologyWe have so far illustrated how an understanding of the brains varied responses to insult can augment the descrip-tion of critical pathophysiological processes and facilitate the generation of hypotheses concerning underlying mechanistic causes (FIGS1,2). However, to translate these descriptions into tools that can improve prognostic evalu-ations and treatment planning, it is imperative to develop formal, computational models that allow testable predic-tions to be made about the specific profile of neural or behavioural changes that are expected to occur following an insult. In this section, we consider how an understand-ing of connectome topology can guide the development of such models. Specifically, we consider progress in modelling the maladaptive responses that spread pathol-ogy throughout the brain, and models of how network organization might constrain the adaptive processes that promote recovery.

Centrality and maladaptation. Mapping inter-regional connectivity allows simple predictions to be made about how pathology in one region might affect the structure and function of other brain areas (FIG.1). More nuanced predictions of disease spread are possible if we consider higher-order aspects of connectome topology. To this end, graph theory (BOX1) can be used to identify spe-cific regions that represent critical vulnerability points putative Achilles heels in the brain. Such research suggests that not all brain regions are equal; rather, the functional impact of damage to any single network ele-ment strongly depends on the connection topology of that region38,39 (FIG.3).

The best-studied topological dimension in this context is centrality: the influence that a node has on other network elements. Most simply, centrality can be quantified using node degree (the number of connec-tions attached to a given node), although other meas-ures of centrality have been proposed109. Structural and functional brain networks, like many other complex systems110, are characterized by a heavy-tailed degree distribution27,111; that is, they have many low-degree nodes and a small number of putative hub nodes, which have a very high degree. Such networks are robust to random node failures because the probability of affecting a hub node is low. However, they are highly vulnerable to a tar-geted hub attack, as damage to high-degree nodes affects a disproportionate number of connections and can result in rapid network fragmentation112.

In the brain, high-degree and topologically central hub regions are highly interconnected, forming a rich club a central core of hubs that facilitates efficient communication between disparate network elements113 (BOX1). These central hub nodes are concentrated in heteromodal association cortices. By contrast, primary sensory cortices tend to have low topological central-ity113115. Computational studies have shown that damage

to highly central regions and/or to connections between rich-club members has a more diffuse effect on brain network structure and function than does damage to topologically peripheral nodes or to the connections between them38,39,113.

Distinct hub types can be defined according to the modular organization of a network (BOX1). Network analyses of structural connectivity and functional con-nectivity commonly identify four to ten large-scale canonical modules in the brain, each of which can be linked to a broad behavioural domain113,116,117. Hubs are typically distinguished from non-hubs as they are more highly connected to other nodes in the same module (known as strong inter-modular connectivity). Further, the presumed functional role of these hubs can be determined by their pattern of connectivity with other modules (known as inter-module connectivity)118120. Provincial hubs link primarily to other nodes in the same module and have an important role in functional specialization, whereas putative connector hubs have links that are distributed across multiple different modules, and thereby have a central role in functional integration (FIG.5a,b). Consistent with this organization, computational studies suggest that damage to connector hubs has a more widespread effect on network dynam-ics, whereas lesions to provincial hubs exert a more profound effect on local subsystems38. Thus, we would predict that damage to provincial hubs should yield spe-cific clinical deficits, whereas damage to connector hubs will result in more complex and pervasive dysfunction. This hypothesis is supported by recent evidence that patients who have sustained damage to regions such as the dorsomedial prefrontal cortex and anterior insular cortex regions that are functionally connected to many diverse modules display a pervasive profile of neuro psychological impairment that extends across several cognitive domains121. These findings do not discount the disability caused by specific deficits (for instance, a lesion to primary visual cortex can lead to the specific yet disabling impairment of blindness); rather, they suggest that impairments will manifest across a broader range of functional domains following damage to topologically centralareas.

Collectively, these considerations indicate that mal-adaptive responses to insult such as diaschisis and dedifferentiation should occur more frequently and should be more widespread following damage to topo-logically central regions than after damage to regions with a more peripheral role in the network. Moreover, topologically central hubs can act as conduits for the rapid spread and progression of transneuronal degen-eration122. This assertion is supported by evidence that the brain regions that are most vulnerable to deteriora-tion are those that are linked or topologically proximal to sites that are already affected by the disease5860,63. In this context, greater functional compromise is expected once the disease process has encroached on hub nodes. Indeed, the differential involvement of topologically central versus peripheral regions at varying stages of illness may explain the punctuated and nonlinear pat-tern of functional decline that is often associated with

R E V I E W S




Expe

cted

impa

irm

ent

Degeneracy Centrality

Poorer prognosis

Betterprognosis

a b

c d

0% 80%

Intr

a-m

odul

e co

nnec

tivi

ty

Inter-module connectivity

Provincial hub

Non-hub bridge

Connector hub

Bridge hub

Non

-hub

sH

ubs

Non-hub connectorPeripheral

Module identityNode role Inter-module

edge

Intra-module edge

Local module density

Non-bridges Bridges

Figure 5 | Modules, hubs and the topological characteristics of vulnerability and resilience. a | Schematic of the modular organization of the brain, depicting overlapping module structure. In networks with overlapping modules, nodes can belong to more than one module. This structure has not been studied extensively in brain networks, as most graph theorybased methods for module decomposition that have been used so far separate nodes into unique, nonoverlapping modules. However, a model of neural architecture that allows for overlapping modules offers a more realistic model of brainnetwork organization (for instance, cortical association areas are known to have a role in multiple networks). Internal node colours denote the module to which each node belongs. Intramodule edges are coloured accordingly; intermodule edges are grey. Colours of node outlines denote node roles, as defined in part b. The central module represents a core of interconnected, highdegree nodes, many of which are also involved in other modules114,136. For simplicity, we ignore hierarchy in modular organization (that is, modules within modules)160. b | The roles of nodes are often assessed using the withinmodule degree zscore (a measure of intramodule connectivity) and the participation coefficient (a measure of intermodule connectivity)118. Hubs typically have high intramodular connectivity. Connector hubs also have many links to different modules, whereas bridge nodes represent the more extreme case of nodes, with a relatively equal distribution of connectivity to different modules. It is thus difficult to assign these nodes to any single module; rather, they belong to multiple modules. Bridgeness is best quantified using methods that allow analysis of overlapping module structure134. The colours representing each role are the same as those used to distinguish different types of nodes in part a. c | A map of regions that are thought to contribute to the functions of diverse modules, representing candidate bridge hubs. These regions (indicated by red and warmer colours) provide a topological substrate for degeneracy, and their dysfunction is likely to have a major impact on brainnetwork integrity and on behaviour. The colour scale indexes the local module density; regions with high module density are likely to participate in multiple systems120. The medial wall is depicted in grey. d | The hypothetical relationship between the centrality of a node subjected to a pathological perturbation, the degeneracy of that node, the expected level of functional impairment following node damage and prospects for recovery following the insult. The extent of network dysfunction, and the behavioural impairment associated with it, is expected to increase with greater centrality of the affected region. This relationship is moderated by the degeneracy of the damaged system; higher degeneracy of the affected node affords a better prognosis. Parts b and c reprinted from Neuron, 79, Power, J. D., Schlaggar, B. L., Lessov-Schlaggar, C. N. & Petersen, S. E. Evidence for hubs in human functional brain networks. 798813, Copyright (2013), with permission from Elsevier120.

R E V I E W S



MotifsSimple, recurring patterns or subgraphs that represent building blocks of a larger network.

the progression of neurodegenerative disorders123, such that the largest declines in function may occur when dis-ease impinges on a hub region. Encouragingly, one study showed that these topological constraints on degeneration can be exploited to predict disease spread. Specifically, the spatial distribution of grey-matter atrophy seen in patients with Alzheimer disease and behavioural variant fronto-temporal dementia could be reproduced by a simple com-putational model of disease diffusion that was simulated on empirically derived connectomic maps124. This finding demonstrates how knowledge of network topology can be used to generate prognostically useful models of disease progression.

Several converging lines of evidence also indicate that central hub regions have an increased susceptibility to the effects of brain disease. First, empirical findings suggest that many brain disorders disproportionately affect hub regions35,125. Second, a high proportion of the shortest top-ological paths (BOX1) between brain regions pass through hub regions126, suggesting that these nodes can be easily reached by trans-synaptic pathological processes that orig-inate elsewhere in the brain60. Third, many of the connec-tions to and from hubs extend across long distances115,126 and are thus more susceptible to white-matter injury (for instance, axonal shearing in traumatic brain injury) or dis-ease (for example, demyelinating lesions in multiple scle-rosis)34. Finally, the high baseline activity and metabolic requirements of hub regions127,128 may render their con-stituent neurons particularly vulnerable to metabolic stress or activity-dependent degeneration, especially if activity levels increase beyond this high baseline (for instance, if these regions are recruited for compensation)34,56,88.

Degeneracy and adaptation. The capacity of the brain for resilience, compensation and functional restitution following insult is closely tied to its degeneracy. The degeneracy of a brain network can be quantified directly, using metrics from information theory102, or indirectly ,using measures of topological overlap between nodal-connectivity profiles114 and relatively simple indices such as the clustering coefficient (the probability that two nodes connected to a third node are also connected to each other) (BOX1). Indeed, a defining characteristic of some hub regions is low clustering113,129, and many rich-club nodes tend to occupy the apex within open three-node motifs; that is, they often act as bridges that connect otherwise unconnected pairs of nodes115. This property compounds the deleterious effects of rich-club damage, as removal of the apex node will prevent com-munication between the remaining regions. By contrast, areas that are embedded within a specific module are likely to display higher topological degeneracy, as they form part of a tightly interconnected clique of nodes120.

In addition to supporting degeneracy within specific subsystems, the interconnectivity of nodes within a topo-logical module (BOX1) affords the brain added resilience to disease because this interconnectivity can entrap a patho-logical process, preventing it from spreading to other parts of the network130. Accordingly, computational studies suggest that too much integration between systems that should normally be segregated a shift in organization

that reflects a breakdown of modularity can facilitate the spread of network failures131, reduce functional spe-cialization and result in dedifferentiated neural activity70 (FIG.2c). In more extreme cases, such a break-down may engender a propensity for hypersynchronized, seizure-like activity132,133.

Another topological substrate for degeneracy is pro-vided by bridge nodes regions that are involved in multiple modules134. These areas act as convergence zones (REFS120,135) that allow the integration of special-ized processes between distinct neural systems (FIG.5ac). Preliminary evidence suggests that high-degree nodes in association cortices tend to show high bridgeness (that is, they are involved in multiple modules)136, consistent with their role in supporting the integration of other-wise segregated processes. They are thus well-positioned to support the engagement of alternative systems that can promote functional compensation and recovery after insult. As acorollary, damage to bridge nodes is expected to have amarked effect on network integrity, as these nodes are both topologically central and integral to degeneracy. Consistent with this hypothesis, recent work in patients with diverse brain lesions showed that damage to putative bridge hubs often results in pervasive cognitive impairment121.

In summary, empirical and computational studies suggest that damage to topologically central brain regions is associated with widespread effects on network func-tion. The expression of these effects will vary depend-ing on the topological role of the affected region, such that pathology of provincial hubs is expected to produce specific deficits, whereas dysfunction of connector hubs is proposed to impair multiple behavioural domains. Furthermore, these effects will be modified by the degeneracy of the affected neural network. Specifically, recovery of function may be more probable following damage to regions with a high clustering coefficient or a high degree of topological overlap with other nodes, and/or to regions that are deeply embedded within mod-ules. Conversely, recovery is less likely following damage to topologically central areas, or to regions that support degeneracy, such as bridge nodes. Following this logic, we can use the topological dimensions of centrality and degeneracy to define a broad parameter space that ena-bles us to make testable predictions regarding the extent of functional compromise, and prognosis for recovery, following insult (FIG.5d).

Conclusions and future directionsWe have considered here how connectomics can improve the description and prediction of the clini-cal expression, course and functional impact of brain disease. The description of disease processes can be enhanced by understanding how adaptive and maladap-tive neural responses to insult are expressed at the level of large-scale networks. In particular, this knowledge can be used to move beyond traditional characteriza-tions of case-control differences in neurobiological measures as abnormal increases or decreases to infer underlying pathophysiological mechanisms (see, for example, FIGS1,2).

R E V I E W S



Non-stationarityThe tendency of some time series to show fluctuations in their mean, covariance and other descriptive measures over time. Non-stationary activity in the brain means that long-term temporal averages of neural activity may not accurately summarize dynamics over shorter timescales.

1. Siegel, R. E. Galen on the Affected Parts (Karger Publisher, 1976).

2. Finger,S., Koehler,P.J. & Jagella,C. The von Monakow concept of diaschisis: origins and perspectives. Arch. Neurol. 61, 283288 (2004).

3. Koehler,P.J. Brown-Squard and cerebral localization as illustrated by his ideas on aphasia. J.Hist. Neurosci. 5, 2633 (1996).

4. Monakow, von,C. in Brain and Behaviour Vol.I: Mood, States and Mind (ed. Pribram,K. H.) 2736 (Penguin books, 1969).

5. Feeney,D.M. & Baron,J.C. Diaschisis. Stroke 17, 817830 (1986).

6. Wernicke,C. Some new studies on aphasie. Fortschr. Med. 824830 (1885).

7. Catani,M. & ffytche,D.H. The rises and falls of disconnection syndromes. Brain 128, 22242239 (2005).This paper provides a thorough review of the role of connectivity changes in neurological disorders.

8. Geschwind,N. Disconnexion syndromes in animals and man. PartI. Brain 88, 237294 (1965).

9. Geschwind,N. Disconnexion syndromes in animals and man. PartII. Brain 88, 585644 (1965).

10. Mesulam,M.M. Large-scale neurocognitive networks and distributed processing for attention, language, and memory. Ann. Neurol. 28, 597613 (1990).

11. Damasio,A.R. Time-locked multiregional retroactivation: a systems-level proposal for the neural substrates of recall and recognition. Cognition 33, 2562 (1989).

12. Broca,P. Perte de la parole: ramollissement chronique et destruction partielle du lobe anterieur gauche du cerveau. Bulletins Soc. Anthropolgie 2, 235238 (in French) (1861).

13. Harlow,J.M. Passage of an iron rod through the head. Boston Med. Surg. J. 39, 389393 (1848).

14. Scoville,W.B. & Milner,B. Loss of recent memory after bilateral hippocampal lesions. J.Neurol, Neurosurg. Psychiatry 20, 1121 (1957).

15. Bullmore,E. & Sporns,O. Complex brain networks: graph theoretical analysis of structural and functional systems. Nature Rev. Neurosci. 10, 186198 (2009).This review gives an introduction to the application of graph theory and network science to neuroscientific data.

16. Tononi,G., Sporns,O. & Edelman,G.M. A measure for brain complexity: relating functional segregation and integration in the nervous system. Proc. Natl Acad. Sci. USA 91, 50335037 (1994).

17. Breakspear,M. & Stam,C.J. Dynamics of a neural system with a multiscale architecture. Phil. Trans. R.Soc. Lond. B 360, 10511074 (2005).

18. Buldyrev,S.V., Parshani,R., Paul,G., Stanley,H.E. & Havlin,S. Catastrophic cascade of failures in interdependent networks. Nature 464, 10251028 (2011).

19. Coan,A.C. etal. Frequent seizures are associated with a network of gray matter atrophy in temporal lobe epilepsy with or without hippocampal sclerosis. PLoS ONE 9, e85843 (2014).

20. Rehme,A.K. & Grefkes,C. Cerebral network disorders after stroke: evidence from imaging-based connectivity analyses of active and resting brain states in humans. J.Physiol. 591, 1731 (2012).

21. Tabrizi,S.J. etal. Biological and clinical manifestations of Huntingtons disease in the longitudinal TRACK-HD study: cross-sectional analysis of baseline data. Lancet Neurol. 8, 791801 (2009).

22. Goedert,M., Spillantini,M.G., Del Tredici,K. & Braak,H. 100years of Lewy pathology. Nature Rev. Neurol. 9, 1324 (2012).

23. Sporns,O. The human connectome: a structural description of the human brain. PLoS Comput. Biol. 1, e42 (2005).This paper introduces the idea of a connectome and outlines strategies for connectome mapping.

24. Bohland,J.W. etal. A proposal for a coordinated effort for the determination of brainwide neuroanatomical connectivity in model organisms at a mesoscopic scale. PLoS Comp. Biol. 5, e1000334 (2009).

25. Kandel,E.R., Markram,H., Matthews,P.M., Yuste,R. & Koch,C. Neuroscience thinks big (and collaboratively). Nature Rev. Neurosci. 14, 659664 (2013).

26. Van Essen,D.C. & Ugurbil,K. The future of the human connectome. NeuroImage 62, 12991310 (2012).

27. Oh,S.W. etal. A mesoscale connectome of the mouse brain. Nature 508, 207214 (2014).

28. Chiang,A.S. etal. Three-dimensional reconstruction of brain-wide wiring networks in Drosophila at single-cell resolution. Curr. Biol. 21, 111 (2011).

29. Markov,N.T. etal. A weighted and directed interareal connectivity matrix for macaque cerebral cortex. Cereb. Cortex 24, 1736 (2013).

30. White,J.G., Southgate,E., Thomson,J.N. & Brenner,S. The structure of the nervous system of the nematode Caenorhabditis elegans. Phil. Trans. R.Soc. Lond. B 314, 1340 (1986).

31. Herculano-Houzel,S. The human brain in numbers: a linearly scaled-up primate brain. Front. Hum. Neurosci. 9, 31 (2009).

However, the responses considered here are not exhaustive; cataloguing other possible responses and their underlying mechanisms will provide a more com-plete picture of how disease affects the brain. Large-scale longitudinal studies that track the progression of brain network changes over time will also be crucial to elu-cidate how pathology dynamically evolves in the brain. Improvements in imaging technologies and network-mapping techniques will enhance the precision with which we are able to track these processes in patients. For structural connectivity, this will involve developing more accurate fibre-reconstruction algorithms, measures of connectivity that have a clear physiological interpreta-tion137,138, and the capacity to resolve the source and target of a projection (see BOX1). For analysis of brain network function, the scaling of effective connectivity models to deal with large-scale brain networks139 and diverse exper-imental paradigms and empirical phenomena140,141 will be particularly important. Indeed, such models will enable us to track the mechanisms and direction of pathological spread throughout the brain with much greater certainty than is afforded by undirected measures of connectivity (BOX1; FIG.1). Moreover, accounting for the known non-stationarity of brain dynamics142 will be crucial for more precisely mapping fluctuations in psychological states to variations in neural states. Multi-modal validation of imaging measures143, combined with detailed biophysical models144, may help to mitigate the limitations of existing approaches145.

We have also illustrated how knowledge of network topology can be used to generate simple, testable pre-dictions about the spread, clinical manifestations and prognosis of brain disease. Indeed, a key advantage of graph theory is that it readily enables the integra-tion of experimental and theoretical neuroscience. For

example, computational models of neural or disease dynamics can be simulated on empirically mapped network structures38,88,124, and the effects of various structural lesions on network activity may be simulated insilico39. This integrated approach enables rigorous and testable predictions about the effects of disease on the brain. In this context, open-source platforms (for example, The Virtual Brain; see Further information) for integrating computational modelling with empirical data146,147 offer considerable promise for researchers and cliniciansalike.

Our understanding of functional specialization in the brain is sufficiently advanced to allow general predictions to be made about the pattern of behavioural deficits expected to result from focal pathology. For example, we expect memory problems following hippocampal dam-age, executive deficits following prefrontal insults, and visuospatial impairments following parietal dysfunc-tion. However, our understanding of how changes in large-scale integrated network structure and dynamics relate to behaviour is relatively immature, and few stud-ies have attempted to link variations in network proper-ties to measures of behavioural performance. Such work will inform the mapping of specific network changes to distinct behavioural outputs. One possible approach would involve the development of computational mod-els of putative disease mechanisms that are informed by, and validated against, experimental connectomic data. In turn, model predictions and experimental measures should be tested for their ability to predict changes in behaviour following an insult. Such an integrated frame-work will be necessary to formulate a coherent and com-prehensive understanding of how neural connectivity mediates and constrains the phenotypic expression of brain disease.

R E V I E W S



32. Pakkenberg,B. etal. Aging and the human neocortex. Exp. Gerontol. 38, 9599 (2003).

33. Kasthuri,N. & Lichtman,J.W. Neurocartography. Neuropsychopharmacology 35, 342343 (2010).

34. Bullmore,E. & Sporns,O. The economy of brain network organization. Nature Rev. Neurosci. 13, 336349 (2012).

35. Crossley,N.A. etal. The hubs of the human connectome are generally implicated in the anatomy of brain disorders. Brain 137, 23822395 (2014).This study uses meta-analyses of voxel-based morphometry studies in a wide array of brain disorders to show that pathological changes tend to occur in network hub regions of the brain.

36. Nomura,E.M., Gratton,C. & Visser,R.M. Double dissociation of two cognitive control networks in patients with focal brain lesions. Proc. Natl Acad. Sci. USA 107, 1201712022 (2010).

37. Gold,L. & Lauritzen,M. Neuronal deactivation explains decreased cerebellar blood flow in response to focal cerebral ischemia or suppressed neocortical function. Proc. Natl Acad. Sci. USA 99, 76997704 (2002).

38. Honey,C.J. & Sporns,O. Dynamical consequences of lesions in cortical networks. Hum. Brain Mapp. 29, 802809 (2008).

39. Alstott,J., Breakspear,M., Hagmann,P., Cammoun,L. & Sporns,O. Modeling the impact of lesions in the human brain. PLoS Comp. Biol. 5, e1000408 (2009).This study uses a neural mass model of brain dynamics, simulated on empirically derived structural connectivity networks, to model changes in functional connectivity arising from different types of structural lesions to the human connectome.

40. Carrera,E. & Tononi,G. Diaschisis: past, present, future. Brain 137, 24082422 (2014).

41. Corbetta,M., Kincade,M.J., Lewis,C., Snyder,A.Z. & Sapir,A. Neural basis and recovery of spatial attention deficits in spatial neglect. Nature Neurosci. 8, 16031610 (2005).

42. He,B.J. etal. Breakdown of functional connectivity in frontoparietal networks underlies behavioral deficits in spatial neglect. Neuron 53, 905918 (2007).

43. van Meer,M.P.A. etal. Recovery of sensorimotor function after experimental stroke correlates with restoration of resting-state interhemispheric functional connectivity. J.Neurosci. 30, 39643972 (2010).

44. van Meer,M.P.A. etal. Extent of bilateral neuronal network reorganization and functional recovery in relation to stroke severity. J.Neurosci. 32, 44954507 (2012).

45. Price,C.J., Warburton,E.A., Moore,C.J., Frackowiak,R.S. & Friston,K.J. Dynamic diaschisis: anatomically remote and context-sensitive human brain lesions. J.Cogn. Neurosci. 13, 419429 (2001).

46. Napieralski,J.A., Butler,A.K. & Chesselet,M.F. Anatomical and functional evidence for lesion-specific sprouting of corticostriatal input in the adult rat. J.Comp. Neurol. 373, 484497 (1996).

47. Carter,A.R. etal. Resting interhemispheric functional magnetic resonance imaging connectivity predicts performance after stroke. Ann. Neurol. 67, 365375 (2010).

48. Cowan,W.M. Contemporary Research Methods in Neuroanatomy (Springer,1970).This review provides a thorough overview of various types of transneuronal degeneration.

49. Deller,T., Del Turco,D., Rappert,A. & Bechmann,I. in The Dentate Gyrus: A Comprehensive Guide to Structure, Function, and Clinical Implications (ed. Scharfman, H.) 501528 (Elsevier, 2007).

50. DeGiorgio,L.A., Dibinis,C., Milner,T.A., Saji,M. & Volpe,B.T. Histological and temporal characteristics of nigral transneuronal degeneration after striatal injury. Brain Res. 795, 19 (1998).

51. Hubel,D.H. & Wiesel,T.N. The period of susceptibility to the physiological effects of unilateral eye closure in kittens. J.Physiol. 206, 419436 (1970).

52. Beatty,R.M., Sadun,A.A. & Smith,L. Direct demonstration of transsynaptic degeneration in the human visual system: a comparison of retrograde and anterograde changes. J.Neurol. 45, 143146 (1982).

53. You,Y., Gupta,V.K., Graham,S.L. & Klistorner,A. Anterograde degeneration along the visual pathway after optic nerve injury. PLoS ONE 7, e52061 (2012).

54. Herbin,M., Boire,D., Thoret,H. & Ptito,M. Transneuronal degeneration of retinal ganglion cells in early hemispherectomized monkeys. Neuroreport 10, 14471452 (1999).

55. Cowey,A., Stoerig,P. & Williams,C. Variance in transneuronal retrograde ganglion cell degeneration in monkeys after removal of striate cortex: effects of size of the cortical lesion. Vision Res. 39, 36423652 (1999).

56. Saxena,S. & Caroni,P. Selective neuronal vulnerability in neurodegenerative diseases: from stressor thresholds to degeneration. Neuron 71, 3548 (2011).

57. Perlson,E., Maday,S., Fu,M. M., Moughamian,A.J. & Holzbaur,E.L.F. Retrograde axonal transport: pathways to cell death? Trends Neurosci. 33, 335344 (2010).

58. Myers,N. etal. Within-patient correspondence of amyloid- and intrinsic network connectivity in Alzheimers disease. Brain 137, 20522064 (2014).

59. Klupp,E. etal. In Alzheimers disease, hypometabolism in low-amyloid brain regions may be a functional consequence of pathologies in connected brain regions. 4, 371383 (2014).

60. Zhou,J., Gennatas,E.D., Kramer,J.H., Miller,B.L. & Seeley,W.W. Predicting regional neurodegeneration from the healthy brain functional connectome. Neuron 73, 12161227 (2012).

61. Ross,D.T. & Ebner,F.F. Thalamic retrograde degeneration following cortical injury: an excitotoxic process? Neuroscience 35, 525550 (1990).

62. Vucic,S., Ziemann,U., Eisen,A., Hallett,M. & Kiernan,M.C. Transcranial magnetic stimulation and amyotrophic lateral sclerosis: pathophysiological insights. J.Neurol. Neurosurg. Psychiatry 84, 11611170 (2013).

63. Verstraete,E., Veldink,J.H., van den Berg,L.H. & van den Heuvel,M.P. Structural brain network imaging shows expanding disconnection of the motor system in amyotrophic lateral sclerosis. Hum. Brain Mapp. 35, 13511361 (2013).

64. Bartzokis,G. Alzheimers disease as homeostatic responses to age-related myelin breakdown. Neurobiol. Aging 32, 13411371 (2011).

65. Hirokawa,N., Niwa,S. & Tanaka,Y. Molecular motors in neurons: transport mechanisms and roles in brain function, development, and disease. Neuron 68, 610638 (2010).

66. Nave,K.-A. Myelination and the trophic support of long axons. Nature Rev. Neurosci. 11, 275283 (2010).

67. Frost,B. & Diamond,M.I. Prion-like mechanisms in neurodegenerative diseases. Nature Rev. Neurosci. 11, 155159 (2009).

68. Buckner,R.L. etal. Molecular, structural, and functional characterization of Alzheimers disease: evidence for a relationship between default activity, amyloid, and memory. J.Neurosci. 25, 77097717 (2005).

69. Seeley,W.W., Crawford,R.K., Zhou,J., Miller,B.L. & Greicius,M.D. Neurodegenerative diseases target large-scale human brain networks. Neuron 62, 4252 (2009).This study reports evidence that grey-matter atrophy in a range of neurodegenerative diseases occurs in functionally and possibly structurally connected neural systems.

70. Li,S.C., Lindenberger,U. & Sikstrm,S. Aging cognition: from neuromodulation to representation. Trends Cognitive Sci. 5, 479486 (2001).

71. Rajah,M.N. Region-specific changes in prefrontal function with age: a review of PET and fMRI studies on working and episodic memory. Brain 128, 19641983 (2005).

72. Rehme,A.K., Eickhoff,S.B., Wang,L.E., Fink,G.R. & Grefkes,C. Dynamic causal modeling of cortical activity from the acute to the chronic stage after stroke. NeuroImage 55, 11471158 (2011).

73. Grefkes,C. etal. Modulating cortical connectivity in stroke patients by rTMS assessed with fMRI and dynamic causal modeling. NeuroImage 50, 233242 (2010).

74. Fornito,A. & Bullmore,E.T. Reconciling abnormalities of brain network structure and function in schizophrenia. Current Opin. Neurobiol. 30C, 4450 (2014).

75. Minzenberg,M.J., Laird,A.R., Thelen,S., Carter,C.S. & Glahn,D.C. Meta-analysis of 41 functional neuroimaging studies of executive function in schizophrenia. Arch. Gen. Psychiatry 66, 811822 (2009).

76. Winterer,G. & Weinberger,D.R. Genes, dopamine and cortical signal-to-noise ratio in schizophrenia. Trends Neurosci. 27, 683690 (2004).

77. Fornito,A., Zalesky,A., Pantelis,C. & Bullmore,E.T. Schizophrenia, neuroimaging and connectomics. NeuroImage 62, 22962314 (2012).

78. Weinberger,D.R. Implications of normal brain development for the pathogenesis of schizophrenia. Arch. Gen. Psychiatry 44, 660669 (1987).

79. Marder,E. & Goaillard,J.-M. Variability, compensation and homeostasis in neuron and network function. Nature Rev. Neurosci. 7, 563574 (2006).

80. Saur,D. Dynamics of language reorganization after stroke. Brain 129, 13711384 (2006).

81. Bestmann,S. etal. The role of contralesional dorsal premotor cortex after stroke as studied with concurrent TMSfMRI. J.Neurosci. 30, 1192611937 (2010).

82. Riecker,A. etal. The role of the unaffected hemisphere in motor recovery after stroke. Hum. Brain Mapp. 31, 10171029 (2010).

83. Rehme,A.K., Eickhoff,S.B., Rottschy,C., Fink,G.R. & Grefkes,C. Activation likelihood estimation meta-analysis of motor-related neural activity after stroke. NeuroImage 59, 27712782 (2012).

84. OShea,J., Johansen-Berg,H., Trief,D., Gobel,S. & Rushworth,M.F. Functionally specific reorganization in human premotor cortex. Neuron 54, 479490 (2007).In this study, the authors used TMS and fMRI to show that motor performance is preserved following inhibitory TMS of left dorsal premotor cortex, and that this preservation is due to compensatory recruitment of contralateral (right) premotor cortex.

85. Johansen-Berg,H. etal. The role of ipsilateral premotor cortex in hand movement after stroke. Proc. Natl Acad. Sci. USA 99, 1451814523 (2002).

86. Frost,S.B. Reorganization of remote cortical regions after ischemic brain injury: a potential substrate for stroke recovery. J.Neurophysiol. 89, 32053214 (2003).

87. Ward,N.S. Neural correlates of motor recovery after stroke: a longitudinal fMRI study. Brain 126, 24762496 (2003).

88. de Haan,W., Mott,K., van Straaten,E.C.W., Scheltens,P. & Stam,C.J. Activity dependent degeneration explains hub vulnerability in Alzheimers disease. PLoS Comp. Biol. 8, e1002582 (2012).The authors of this paper simulated a computational model of activity-dependent degeneration on diffusion MRI-derived structural connectivity networks to demonstrate that transient increases in activity preceded gradual declines, and that hub regions were particularly vulnerable to degeneration.

89. Sperling,R. Functional MRI studies of associative encoding in normal aging, mild cognitive impairment, and Alzheimers disease. Ann. NY Acad. Sci. 1097, 146155 (2007).

90. Poudel,G.R., Egan,G.F. & Churchyard,A. Abnormal synchrony of resting state networks in premanifest and symptomatic Huntington disease: the IMAGE-HD study. J.Psychiatry Neurosci. 39, 8796 (2014).

91. Schoonheim,M.M. & Filippi,M. Functional plasticity in MS: friend or foe? Neurology 79, 14181419 (2012).

92. Grady,C.L. etal. Evidence from functional neuroimaging of a compensatory prefrontal network in Alzheimers disease. J.Neurosci. 23, 986993 (2003).

93. Kolb,B. & Teskey,G.C. Age, experience, injury, and the changing brain. Dev. Psychobiol. 54, 311325 (2010).

94. Buchkremer-Ratzmann,I. & Witte,O.W. Extended brain disinhibition following small photothrombotic lesions in rat frontal cortex. Neuroreport 8, 519522 (1997).

95. Butefisch,C.M. Remote changes in cortical excitability after stroke. Brain 126, 470481 (2003).

96. Liepert,J., Hamzei,F. & Weiller,C. Motor cortex disinhibition of the unaffected hemisphere after acute stroke. Muscle Nerve 23, 17611763 (2000).

97. Marik,S.A., Yamahachi,H., Meyer zum Alten Borgloh,S. & Gilbert,C.D. Large-scale axonal reorganization of inhibitory neurons following retinal lesions. J.Neurosci. 34, 16251632 (2014).

98. Gonzalez,C.L.R. & Kolb,B. A comparison of different models of stroke on behaviour and brain morphology. Eur. J.Neurosci. 18, 19501962 (2003).

99. Jones,T.A. & Schallert,T. Use-dependent growth of pyramidal neurons after neocortical damage. J.Neurosci. 14, 21402152 (1994).

100. Carmichael,S.T. & Chesselet,M. F. Synchronous neuronal activity is a signal for axonal sprouting after cortical lesions in the adult. J.Neurosci. 22, 60626070 (2002).

R E V I E W S



FURTHER INFORMATION The Virtual Brain website: http://www.thevirtualbrain.org/tvb/zwei

ALL LINKS ARE ACTIVE IN THE ONLINE PDF

101. Schaechter,J.D., Moore,C.I., Connell,B.D., Rosen,B.R. & Dijkhuizen,R.M. Structural and functional plasticity in the somatosensory cortex of chronic stroke patients. Brain 129, 27222733 (2006).

102. Tononi,G., Sporns,O. & Edelman,G.M. Measures of degeneracy and redundancy in biological networks. Proc. Natl Acad. Sci. USA 96, 32573262 (1999).

103. Kitano,H. Biological robustness. Nature Rev. Genet. 5, 826837 (2004).

104. Noppeney,U., Friston,K.J. & Price,C.J. Degenerate neuronal systems sustaining cognitive functions. J.Anat. 205, 433442 (2004).This paper reviews different forms of neural degeneracy and how they support cognition following brain lesions.

105. Seghier,M.L., Lee,H.L., Schofield,T., Ellis,C.L. & Price,C.J. Inter-subject variability in the use of two different neuronal networks for reading aloud familiar words. NeuroImage 42, 12261236 (2008).

106. Friston,K.J. & Price,C.J. Modules and brain mapping. Cognitive Neuropsychol. 28, 241250 (2011).

107. Barulli,D. & Stern,Y. Efficiency, capacity, compensation, maintenance, plasticity: emerging concepts in cognitive reserve. Trends Cognitive Sci. 17, 502509 (2013).

108. Valenzuela,M.J., Breakspear,M. & Sachdev,P. Complex mental activity and the aging brain: molecular, cellular and cortical network mechanisms. Brain Res. Rev. 56, 198213 (2007).

109. Borgatti,S.P. Centrality and network flow. Social Networks 27, 5571 (2005).

110. Barabasi,A.L. & Albert,R. Emergence of scaling in random networks. Science 286, 509512 (1999).

111. Hagmann,P. etal. Mapping human whole-brain structural networks with diffusion MRI. PLoS ONE 2, e597 (2007).

112. Albert,R., Jeong,H. & Barabasi,A.L. Error and attack tolerance of complex networks. Nature 406, 378382 (2000).

113. van den Heuvel,M.P. & Sporns,O. Rich-club organization of the human connectome. J.Neurosci. 31, 1577515786 (2011).

114. Zamora-Lpez, G. Cortical hubs form a module for multisensory integration on top of the hierarchy of cortical networks. Front. Neuroinform. 19, 4 (2010).

115. Harriger,L., van den Heuvel,M.P. & Sporns,O. Rich club organization of macaque cerebral cortex and its role in network communication. PLoS ONE 7, e46497 (2012).

116. Crossley,N.A., Mechelli,A. & Vertes,P.E. Cognitive relevance of the community structure of the human brain functional coactivation network. Proc. Natl Acad. Sci. USA 110, 1158311588 (2013).

117. Smith,S.M. etal. Correspondence of the brains functional architecture during activation and rest. Proc. Natl Acad. Sci. USA 106, 1304013045 (2009).

118. Guimera,R. & Nunes Amaral,L.A. Functional cartography of complex metabolic networks. Nature 433, 895900 (2005).

119. Fornito,A., Harrison,B.J., Zalesky,A. & Simons,J.S. Competitive and cooperative dynamics of large-scale brain functional networks supporting recollection. Proc. Natl Acad. Sci. USA 109, 1278812793 (2012).

120. Power,J.D., Schlaggar,B.L., Lessov-Schlaggar,C.N. & Petersen,S.E. Evidence for hubs in human functional brain networks. Neuron 79, 798813 (2013).

121. Warren,D.E. etal. Network measures predict neuropsychological outcome after brain injury. Proc. Natl Acad. Sci. USA 111, 1424714252 (2014).This was the first demonstration in human patients that damage to putative bridge hubs of the brain results in widespread neuropsychological impairment.

122. Kitsak,M., Gallos,L.K., Havlin,S., Liljeros,F. & Muchnik,L. Identification of influential spreaders in complex networks. Nature Phys. 6, 888893 (2010).

123. Palop,J.J., Chin,J. & Mucke,L. A network dysfunction perspective on neurodegenerative diseases. Nature 443, 768773 (2006).

124. Raj,A., Kuceyeski,A. & Weiner,M. A network diffusion model of disease progression in dementia. Neuron 73, 12041215 (2012).This was the first demonstration that a relatively simple model of disease diffusion, which was simulated on structural network architectures that were derived empirically with diffusion MRI, can reproduce the spatial topography of atrophic changes seen in Alzheimer disease and frontotemporal dementia.

125. Buckner,R.L. etal. Cortical hubs revealed by intrinsic functional connectivity: mapping, assessment of

stability, and relation to Alzheimers disease. J.Neurosci. 29, 18601873 (2009).

126. van den Heuvel,M.P., Kahn,R.S., Goni,J. & Sporns,O. High-cost, high-capacity backbone for global brain communication. Proc. Natl Acad. Sci. USA 109, 1137211377 (2012).

127. Liang,X., Zou,Q., He,Y. & Yang,Y. Coupling of functional connectivity and regional cerebral blood flow re