ABNORMALITIES IN TENSOR MORPHOLOGY IN PATIENTS WITH SCHIZOPHRENIA: A DTI STUDY OF THE CORPUS CALLOSUM

1`

SUMMARY

The corpus callosum (CC) is the largest white matter (WM) fiber bundle in the brain, consisting of approximately 200 million commissural axons. The CC primarily connects homologous regions of the two cerebral hemispheres. Several 'connectivity' models of schizophrenia have hypothesized the existence of WM abnormalities generally1, and CC abnormalities specifically2. Diffusion Tensor Imaging (DTI) enables the testing these models in vivo, as it provides information as to the integrity and microstructure of WM fiber bundles.

BACKGROUND

1Psychiatry Neuroimaging Laboratory, Department of Psychiatry, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; 2Melbourne Neuropsychiatry Centre, Department of Psychiatry, University of Melbourne, and Melbourne Health, Melbourne, Australia; 3Clinical Neuroscience Laboratory, Department of Psychiatry, VA Boston Healthcare System, Harvard Medical School, Brockton, MA, USA

Thomas J. Whitford1,2, Marek Kubicki1,3, Rebecca King1, Usman Khan1, Douglas Markant1, Jorge Alvarado1, Robert W. McCarley3, Martha E. Shenton1,3

• 19 SZ patients (mean age = 44 yrs) and 19 healthy controls (mean age = 40 years) participated in the study. Patients' symptoms were assessed using the Positive and Negative Syndrome Scale (PANSS)• Participants all underwent a Diffusion-Weighted MR scan on a 3T GE system (EPI sequence, b=700, 51 independent directions, 8 baseline images, 1.7mm3 voxels). These scans were converted into Diffusion-Tensor Images (DTIs).• Whole-Brain Fiber Tractography (Fig.1) was performed on each DTI. The resultant tracts were grouped automatically into 400 clusters, based on a pairwise similarity function that takes into account tracts’ shapes and spatial locations3.• Clusters constituting the CC were manually identified on a single participant (Fig.2), and automatically extracted for all other participants. • The CC clusters were subdivided into 6 segments on the basis of their cortical projections (Fig.3). The 6 CC segments were: Frontal (red), Premotor (blue), Sensorimotor (green), Parietal (pink), Visual (orange) and Temporal (yellow) fibers. For each participant, the 6 CC segments were transformed into binary images and used to mask their DTI.• Mean Fractional Anisotropy (FA), Trace and Mode4 (Fig.5) were calculated for each of the 6 CC segments, for each participant. Axial and Radial Diffusivity were also calculated for any CC segment in which between-group differences in FA were observed

METHODS

Within-group differences (Fig.4) • There were significant differences between the 6 CC segments in terms of FA and Mode (Fig.4).In general, FA and Mode were observed to increase from the genu to the splenium. The highest FA and most cylindrical Mode was observed in CC5 (Visual segment) and CC6 (Temporal segment).Between-group differences (Figs. 4, 5 and 6) • SZ patients showed significantly reduced FA in CC1 (Frontal segment; t

1,36=2.1, p=.

046). Supplementary analyses revealed abnormally increased Radial Diffusivity (t

1,36=2.2, p=0.037) but unchanged Axial Diffusivity (t

1,36=0.5, p=.9) in the SZ patients

in CC1.•Correlation analysis revealed significant positive correlations b/w CC1 FA and patients' PANSS-Delusions (r(16)=.56, p=.015) and Hallucinations (r(16)=.53, p=0.025) scores. Conversely, there were significant negative correlations b/w CC1 Radial Diffusivity and PANSS-Delusions (r(16)=-.558, p=0.016). • SZ patients showed significantly increased Mode (i.e. more cylindrical ellipsoids) in CC4 (Parietal fibers; t

1,36=2.3, p=.025).

RESULTS

•The SZ patients showed FA reductions in the frontal (CC1) fibers. Post-hoc analyses revealed these fibers to be abnormally increased in terms of their Radial Diffusivity (a putative measure of myelination) but unchanged in terms of their Axial Diffusivity (a putative measure of axon damage)5. •Given the role that myelin is known to play in increasing the conduction velocity of action potentials, such a increase in Radial Diffusivity would be expected to result in delayed inter-hemispheric transmission. In this context, it was notable that SZ patients with the most severe FA reductions exhibited the least severe levels of Hallucinations and Delusions. These results suggest that while mild timing abnormalities between discrete brain regions could underpin psychosis, major timing abnormalities could potentially lead to dysynchronized signals which could not be integrated - even pathologically, as in the case of an hallucination or delusion. •SZ patients showed increased Mode (i.e., more cylindrical ellipsoids) in the parietal segment (CC4). This may be the result of a reduction in the number of parietal fiber crossings. A reduction in fiber crossings would be expected to result in an elevation of parietal FA8, which is consistent with the normative FA of CC4 in the SZ patients.•It has been suggested that fiber bundles consisting of lightly-myelinated, thin-fibers (e.g., frontal, parietal – see Fig.86) will show higher FA than heavily-myelinated, thick-fibers (e.g., sensorimotor – Fig.8), because of the associated increase in axonal packing density7. While CC2 (Premotor) and CC3 (Sensorimotor) did show the lowest FA values, the thin-fibers of CC1 (Frontal) showed substantially lower FA than did the thick-fibers of CC5 (Visual). •Future work will focus on the relationship between DTI measures of WM integrity and EEG measures of inter-hemispheric transfer.

DISCUSSION

Figure 2: Fiber clusters constituting the corpus callosum (in brown)

Figure 3:Subdivision of corpus fibers into 6 segments: Frontal (CC1; red), Premotor (CC2; blue), Sensorimotor (CC3; green), Parietal (CC4; pink), Visual (CC5; orange) and Temporal (CC6; yellow)

Figure 8: ↓Schematic illustrating the spatial distribution of corpus fibers according to their diameter5

• Bartzokis. 2002. Neuropsychopharmacology, 27, 672-683. • Crow. 1998. Schizophrenia Research, 30, 111-114.• O'Donnell et al. 2006. American Journal of Neuroradiology, 27, 1032-1036.• Ennis and Kindlmann. 2006. Magnetic Resonance in Medicine, 55, 136-146. • Song et al., 2002. Neuroimage, 17, 1429-1426.• Aboitiz et al. 2003. Biological Research, 36, 89-99.• Hofer et al. 2006. Neuroimage, 32, 989-994.• Kubicki et al. 2007. Journal of Psychiatric Research, 41, 15-30.

REFERENCES

The authors gratefully acknowledge the support of grants from the National Health and Medical Research Council of Australia (NHMRC 520627 to TW), the National Institutes of Health (R03 MH068464-01 to MK, K02 K05 MH070047 and R01 MH 50747 to MES, R01 MH 40799 to RWM), the Department of Veterans Affairs (a VA Merit Award to MES and RWM, a Research Enhancement Award Program to RWM and MES, a VA Schizophrenia Research Center Grant to RWM and MES), the Center for Intervention Development and Applied Research (CIDAR) funding mechanism (P50 to RWM and MES), the Harvard Medical School (Milton Award to MK), the National Alliance for Research on Schizophrenia and Depression (NARSAD: MK). This work is also part of the National Alliance for Medical Image Computing (NAMIC), funded by the National Institutes of Health through the NIH Roadmap for Medical Research, Grant U54 EB005149: MK and MES).

ACKNOWLEDGEMENTS

← Figure 4: Scatterplots illustrating the differences in FA, Mode and Trace between the SZ patients (red dots) and control subjects (blue dots), across the 6 corpus segments. Asterisks indicate significant results

•The white matter (WM) generally, and the corpus callosum specifically, have been implicated as abnormal in schizophrenia (SZ)•19 SZ patients and 19 matched controls underwent a Diffusion-Tensor Imaging (DTI) scan on a 3T system•Following whole-brain tractography, the corpus callosum (CC) was extracted from all subjects' DT images. FA, Mode and Trace were calculated for 6 CC segments, for all participants•The SZ patients exhibited increased Mode (i.e., more cylindrical diffusion ellipsoids) in the Parietal CC fibers, and reduced FA in the Frontal CC fibers, relative to controls. Post-hoc analyses suggested that the FA reductions were underpinned by myelin abnormalities •Significant correlations were observed between patients' FA scores and their Hallucination and Delusion scores. Specifically, the results indicated that patients with the least abnormal FA exhibited the most severe psychotic symptoms•In light of the role that WM is known to play in modulating the transmission velocity of neural signals, the results of this study are considered in the context of the prevailing 'connectivity' theories of SZ

Figure 7: →Mode

describes the shape of the

diffusion ellipsoid

(image created by Gordon

Kindlmann)

*

*

*

Figure 5: → Scatterplots illustrating

between-group differences in

Axial and Radial Diffusivity in CC1

(Frontal fibers)

Figure 6: → Scatterplot

illustrating the observed positive

correlation between patients'

FA in CC1 and their PANSS-

Delusion score. A similar

correlation was seen with PANSS-

Hallucinations

Figure 1: Whole-brain tractography