Callosal morphology in schizophrenia

With the advent of modern neuroimaging techniques, the callosum became an attractive candidate for analysis via structural imaging: an obviously bright structure on sagittal images, it is readily segmented from surrounding structures; its unique shape profile lends itself to a range of morphological descriptors, such as area, thickness and curvature. Unlike other brain structures, the callosum does not lend itself to anatomically meaningful volume estimations because of the absence of clear lateral boundaries; researchers have thus relied on estimates of area or shape. Its topographical fibre organisation - where anterior segments connect anterior cortical regions and posterior segments connect posterior cortical regions^{Reference Pandya, Seltzer, Lepore, Pitto and Jasper8} - suggests that regional cortical alterations may be detectable as subtle regional changes in callosal shape or thickness. In this way, the callosum potentially provides a window to cortical changes in brain disorders, and such changes have been of clinical utility in the differentiation of dementia subtypes.^{Reference Yamauchi, Fukuyama, Nagahama, Katsumi, Hayashi and Oyanagi9} The shape of other topographically organised and projecting structures, such as the basal ganglia, has similarly been shown to differentiate between subtypes of neurodegenerative disorders.^{Reference Looi and Walterfang10}

The first magnetic resonance imaging (MRI) study suggested alterations in corpus callosum size and shape.^{Reference Nasrallah, Andreasen, Coffman, Olson, Dunn and Ehrhardt11} Since this initial study, the widespread availability of MRI scanning has resulted in a range of studies examining the callosum from a variety of morphological perspectives, revealing a number of ways in which the callosum in people with schizophrenia appears to differ from healthy controls, and across different stages of illness (Fig. 1).^{Reference Walterfang, Wood, Reutens, Wood, Chen and Velakoulis12-Reference Walterfang, Wood, Reutens, Wood, Chen and Velakoulis14} Two meta-analyses of these studies, almost 15 years apart, are consistent with the finding that the callosum is smaller in schizophrenia,^{Reference Woodruff, McManus and David15,Reference Arnone, McIntosh, Tan and Ebmeier16} although the latter meta-analysis suggests that this effect may be greatest in patients with first-episode schizophrenia. The difficulty in extracting the nature of the true signal regarding callosal morphology in schizophrenia across these studies is the significant variety of methodological approaches used (with manifold ways to account for head size, choose a consistent mid-sagittal image across participants and parcellate the callosum in anatomically meaningful ways), and the way that descriptors of the callosum are used in these studies: ‘shape’, ‘size’ and ‘reduction’ are terms that have different meanings according to the differences across methodological approaches and statistical models. This heterogeneity may thus dilute or alter the findings from meta-analyses that attempt to combine studies with varying methodology; thus large multistage cohorts who are examined across different phases of psychosis utilising the same sequence parameters and image analysis^{Reference Walterfang, Wood, Reutens, Wood, Chen and Velakoulis12-Reference Walterfang, Wood, Reutens, Wood, Chen and Velakoulis14} may be preferable.

Given that evidence suggests that brain changes may occur prior to illness onset, and may progress with transition to psychosis and/or established illness, a true understanding of shape changes as described by any particular method would be applied across different stages of illness: pre-psychotic, first-episode and established illness.^{Reference Walterfang, Wood, Reutens, Wood, Chen and Velakoulis12,Reference Walterfang, Yung, Wood, Reutens, Phillips and Woods13} Additionally, volumetric imaging - from which callosal shape can be derived - is limited in its capacity to inform on microstructural changes within brain regions. Recent white matter imaging techniques such as diffusion tensor imaging (DTI) suggest that some of the callosal shape alterations in schizophrenia are driven by alterations to white matter microstructure, either through disrupted myelination or altered axonal structure or organisation.^{Reference Yao, Lui, Liao, Du, Hu and Thomas17} Finally, although white and grey matter structures are arbitrarily separated based on their macrostructural appearance and intensity on T ₁-weighted MRI sequences, they remain different components of groups of the same functional unit: the neuron and its axonal process. When white matter change is seen in any illness, the interdependent nature of grey and white matter compartments means that change may be driven by a mixture of primary white matter pathology, or alterations that occur secondary to grey matter change.^{Reference Walterfang, Wood, Velakoulis and Pantelis18,Reference Walterfang, Velakoulis, Whitford and Pantelis19} As a result, a true determination of the role of white matter change in any cohort of patients with schizophrenia will necessarily also analyse grey matter findings in the same cohort.

Fig. 1

Callosal significance maps of regional thickness reductions shown previously in individuals who are pre-psychotic (top), with first-episode psychosis (middle) and with established schizophrenia (bottom).^{Reference Walterfang, Wood, Reutens, Wood, Chen and Velakoulis12-Reference Walterfang, Wood, Reutens, Wood, Chen and Velakoulis14}

Each significance map is projected onto a callosal skeleton in blue; regions of significance change are marked in colour on the skeleton. Patients who are pre-psychotic (top) show changes in the anterior genu, connecting frontal cortical regions, which are also seen in patients with first-episode psychosis (middle) across a larger region of the genu; established patients (bottom) show more significant reductions in this anterior region, but then show reductions in the body and isthmus of the callosum, connecting temporal cortical regions.

The study described by Collinson et al in this issue addresses some, albeit not all, of these issues within one large cohort of 120 patients (with first-episode and established schizophrenia) and 75 controls.^{Reference Collinson, Gan, Woon, Kuswanto, Sum and Yang20} This study utilised both high-resolution volumetric T ₁-weighted images and diffusion-tensor images acquired in the same session on all participants. This well-designed study has some compelling advantages: a large sample size recruited from one geographical region; patients at different illness stages; and an imaging protocol examining both volumetric and microstructural measures of the callosum. It showed that the differences were detectable only between patients with established schizophrenia and controls, but individuals with first-episode schizophrenia showed no significant difference. Additionally, the DTI measures, which could be expected to potentially underpin volumetric change, showed no difference on any group analysis. It adds to the heterogeneity of findings in patients with schizophrenia: the most recent structural imaging meta-analysis suggests that volumetric reductions may be greatest in first-episode rather than established illness,^{Reference Arnone, McIntosh, Tan and Ebmeier16} and a meta-analysis of DTI findings suggests significant reductions in microstructural integrity of the splenium across studies.^{Reference Patel, Mahon, Wellington, Zhang, Chaplin and Szeszko21} The Collinson et al study also illustrates some of the potential pitfalls of studying any structure in schizophrenia, where the choice of methodology may have an impact on the capacity to detect change. One key example is in accounting for head size; many callosal studies have covaried for intracranial volume to account for this, although using a linear transformation and scaling has been shown to be superior in detecting true between-group differences;^{Reference Bermudez and Zatorre22} transformation is part of the FreeSurfer pipeline used by Collinson and colleagues in their analysis. Furthermore, most studies of the callosum attempt to find local regional differences, thus moving from a pure area/volume measure to a de facto metric of shape. Choice of shape metric and method may similarly affect the capacity to detect results, with sophisticated, sensitive and anatomically informed shape analysis methods for the callosum now being described.^{Reference Herron, Kang and Woods23,Reference Adamson, Wood, Chen, Barton, Reutens and Pantelis24} A move to these more sophisticated methods, combined with the microstructural methods as utilised in the Collinson et al study, are - at least in part - the way to move the field forward.

Callosum and corollary discharges

How can an understanding of changes in the callosum inform us about the neurobiology of schizophrenia? As the brain’s largest white matter fibre bundle, if there are diffuse white matter changes in schizophrenia, callosal interhemispheric fibres - alongside long association tracts - are most likely to have both their structure and function disrupted. Consequent abnormalities in neural timing and synchrony between distal brain regions may thus produce the characteristic symptoms of schizophrenia. One particular function is highly dependent on accurate timing in these distal brain regions: corollary discharges, which are neural signals that are initiated in frontal cortical regions coincident with willed actions and fed posteriorly to sensory regions to suppress the sensory consequences of these actions, thereby allowing the brain to recognise these actions as ‘self’-generated.^{Reference Crapse and Sommer25} With subtle changes to white matter integrity, an introduced delay in this system may result in ‘self’-generated phenomena being erroneously experienced as ‘other’: internal speech becomes external auditory hallucinations, internal images or thoughts are experienced as projected from an external source and somatic experiences as passivity phenomena.^{Reference Whitford, Ford, Mathalon, Kubicki and Shenton26} Callosal microstructural abnormalities in schizophrenia are known to be accompanied by abnormal neural timing^{Reference Whitford, Kubicki, Ghorashi, Schneiderman, Hawley and McCarley27} and correlate with the severity of these psychotic symptoms.^{Reference Whitford, Kubicki, Schneiderman, O'Donnell, King and Alvarado28} Thus, alterations in the callosal macro- and microstructure may be an index in people with schizophrenia of the severity of neural timing abnormalities, and thus the likelihood of the pathognomonic reality distortions in these patients whereby the distinction between ‘self’ and ‘other’ breaks down during episodes of psychosis. The Collinson et al ^{Reference Collinson, Gan, Woon, Kuswanto, Sum and Yang20} findings do not completely align with this hypothesis however, as changes would be expected to be detectable during both first-episode and established illness, as individuals are likely to experience these symptoms across illness stages. As the role of the callosum in schizophrenia pathology remains far from settled, the dissonance between the findings in the Collinson et al study and the corollary discharge hypothesis, and other models of schizophrenia that focus on the role of white matter,^{Reference Walterfang, Wood, Velakoulis and Pantelis18,Reference Walterfang, Velakoulis, Whitford and Pantelis19} is likely to remain irresolvable for some time.