From Lichtheim to Rich Club

Abstract

As van den Heuvel et al1 point out in their article on rich club organization of brain networks in schizophrenia, the idea of brain network abnormality goes back a long way in psychiatry. Early (mostly German-speaking) pioneers of neuroanatomy and neurology or psychiatry, such as Meynert, Wernicke, Lichtheim, Dejerine, and Freud, appreciated the functional importance of the brain’s central whitematter and its synaptic connectivity. They understood that whitematter was organized as tracts that interconnected regions of cortex andsubcorticalnuclei.Theyreasoned,mainlyby linkingsymptoms to lesions identified by dissection, that disorders could be caused by focal lesions to cortical network nodes or the white matter links between them. Themostexplicitandclinicallysuccessful 19th-centurynetworkmodelwasprobablytheLichtheimlanguagemodel,which locatednodes for sensory,motor, andassociationcomponents of languageinmoreor lessspecificcorticalareas, linkedbywhite matter tracts, eg, the arcuate fasciculus between the Broca areaof left inferior frontal cortex and the Wernicke area of left superior temporal gyrus. This model, although not perfectly complete, has been a useful and enduring road map to guide clinical treatment of aphasic language disorders.2 It certainlyoccurred to theearlynetworkpsychiatrists that psychosis, being also associatedwithvariousdisorders of language,might equallybeexplicable in termsofdisrupted largescale brainnetworkorganization.However, therewasnoclear empiricalverificationof thishypothesis in the19thorearly20th century.Theavailablemethods formedicalbrain researchwere largely limited to clinical examination of patients in life followed by postmortem examination of their brains. This was too blunt an instrument to convincingly map the brain network diagram for psychosis, as Lichtheim had managed for language. However, methods for clinical brain mapping have vastly improved in the last 30 years, as new imaging technologies such as positron emission tomography, magnetic resonance imaging (MRI), functional MRI (fMRI), diffusion tensor imaging (DTI), and magnetoencephalography have been developed and translated to studies of schizophrenia. This has allowed us to assemble a much more convincing evidence base for brain disorder in schizophrenia. It is now beyond doubt that schizophrenia is typically associated with an anatomically distributed set of regional nodes that have locally abnormal structure or function.3 There is also very strong prior evidence for abnormal white matter organization and for structural or functional dysconnectivity between brain regions in schizophrenia.4 Broadly speaking, 20th-century neuroimaging has definitively confirmed the speculative network hypotheses of a few outstanding 19thcentury clinical scientists. In the last decade, there has been a surge of interest and progress in mapping brain network organization at both microscopicandmacroscopic scales.This ispartlya resultof technological developments in brain-measuring instruments, like MRI scanners, but mainly a result of more conceptual developments inhow these brainmeasurements are analyzed.One major recent conceptual trend has been the increasing use of graph theory.5 In this approach, brain networks are representedabstractlyasa setof regionalnodesconnectedbyedges, and their topological properties—or patterns of connectivity— aremeasured. It hasbeen found thatnormalbraingraphshave complex (not random and not regular) topological properties, such as small-worldness, modules, and highly connected hubs. Topological abnormalities of brain graphs have also been demonstrated in case-control neuroimaging studies of patients with schizophrenia.6 In this context, van den Heuvel and colleagues have first focused their attentionon the topologyofbrain structural networks measured using DTI (acquired at 3 T) in a sample of 48 patients with chronic schizophrenia compared with a group of 45 healthy volunteers. In particular, they have asked an innovativequestion: is the rich clubof thehumanbrain connectome abnormal in schizophrenia? Rich club is a technical term innetwork theory that can be written down as a mathematical formula, but its meaning is also quite immediately intuitive. In many networks, not just brain networks, there will be a few nodes that are exceptionally well connected to other nodes. They are the high-degree hubsof thenetwork.The rich clubmetric simplymeasures the tendency of this elite group of hubs to be more densely connected to each other than would be expected by chance. As van den Heuvel and colleagues have previously shown7,8 and have replicated here, humanDTI networks normally have a rich club comprising regions of superior frontal and parietal cortex, precuneus, and thalamus that are efficiently interconnected to each other, often over long distances in anatomical space. To put it another way, the brain’s rich club has a high capacity for integrative information processingbutalsoahighphysical cost in termsof connectiondistance between club hubs. In people with schizophrenia, van den Heuvel and colleagues have clearly demonstrated that rich club organizaRelated article page 783 Opinion