Microstructural abnormalities in callosal fibers and their relationship with cognitive function in schizophrenia: A tract‐specific analysis study

Abstract Introduction The corpus callosum serves the essential role of relaying cognitive information between the homologous regions in the left and the right hemispheres of the brain. Cognitive impairment is a core dysfunction of schizophrenia, but much of its pathophysiology is unknown. The aim of this study was to elucidate the association between microstructural abnormalities of the corpus callosum and cognitive dysfunction in schizophrenia. Methods We examined stepwise multiple regression analysis to investigate the relationship of the fractional anisotropy (FA) of callosal fibers in each segment with z‐scores of each brief assessment of cognition in schizophrenia subtest and cognitive composite score in all subjects (19 patients with schizophrenia [SZ group] and 19 healthy controls [HC group]). Callosal fibers were separated into seven segments based on their cortical projection using tract‐specific analysis of diffusion tensor imaging. Results The FA of callosal fibers in the temporal segment was significantly associated with z‐scores of token motor test, Tower of London test, and the composite score. In the SZ group, the FA of callosal fibers in the temporal segment was significantly associated with the z‐score of the Tower of London test. In addition, the FA of callosal fibers in temporal segment showed significant negative association with the positive and negative syndrome scale negative score in the SZ group. Compared to the HC group, the FA in temporal segment was significantly decreased in the SZ group. Conclusion Our results suggest that microstructural abnormalities in the callosal white matter fibers connecting bilateral temporal lobe cortices contribute to poor executive function and severe negative symptom in patients with schizophrenia.


| INTRODUC TI ON
The disconnection hypothesis in schizophrenia (Friston, 1999) is supported by repeated reports of abnormalities of white matter (WM) fibers that connect brain regions (Kubicki & Shenton, 2014;Samartzis, Dima, Fusar-Poli, & Kyriakopoulos, 2014;Wheeler & Voineskos, 2014). The corpus callosum (CC), the largest commissural fiber bundle in the brain, connects the left and the right hemispheres and serves an essential role of relaying sensory, motor, and cognitive information between the homologous regions (Huang et al., 2005;Ribolsi, Daskalakis, Siracusano, & Koch, 2014). Many studies have investigated abnormalities of the CC in schizophrenia with an aim of studying the disconnection between the two hemispheres (Isobe et al., 2016;Ribolsi et al., 2014).
Advances in diffusion tensor imaging (DTI) have enabled capture of microstructural WM abnormality. Techniques to build three-dimensional fiber tracts based on DTI provide opportunities to investigate how specific fiber tracts may affect disorders, by visualizing the trajectories of specific WM fiber bundles and by quantitatively characterizing each fiber tract (Lee et al., 2005;Wakana et al., 2007).
Cognitive impairment is a core dysfunction of schizophrenia that is associated with functional prognosis (Green & Harvey, 2014), but much of its pathophysiology is unknown. Cognitive performance is strongly associated with communication between multiple brain regions (Fox et al., 2005). Development of CC in childhood is correlated with intelligence, processing speed, and problem-solving ability, and is thought to play a fundamental role in cognitive functioning (Hinkley et al., 2012). However, few studies have performed detailed investigation of the relationship between microstructural abnormalities of CC and cognitive functioning in schizophrenia.
In the current study, we examined the association between microstructural abnormalities of CC fibers and cognitive dysfunction in schizophrenia by segmenting the CC fibers based on their cortical projection regions using DTI tract-specific analysis (TSA). We hypothesized that patients with schizophrenia show microstructural abnormalities in CC fibers and these abnormalities are related to their cognitive impairment.

| Subjects
The subjects were 19 patients with schizophrenia (SZ group) and 19 healthy controls (HC group; Table 1). The subjects were diagnosed by two independent well-trained psychiatrists based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (APA, 1994), and were recruited from Wakayama Medical University Hospital. Patients with comorbid psychiatric, neurological, or medical illness, or those with substance or alcohol abuse were excluded from the study. All patients were on antipsychotic medication. Equivalent doses of antipsychotics were calculated using the equivalent conversion Inada (Inada & Inagaki, 2015). This study was approved by the Ethics Committee of Wakayama Medical University, and written informed consent was obtained from all subjects.

| Neuropsychological measurements
The severity of clinical symptoms was assessed using the Positive and Negative Syndrome Scale (PANSS). Neurocognitive function was tested by experienced psychologists using the Brief Assessment of Cognition in Schizophrenia (BACS) Japanese version (Kaneda et al., 2007), which is widely used in Japan (Ikebuchi et al., 2017;Itakura et al., 2017;Satogami, Takahashi, Yamada, Ukai, & Shinosaki, 2017;Sawada et al., 2017;Takahashi et al., 2013). This battery includes  (Kaneda et al., 2013). The composite score was calculated by averaging all z-scores of six subcomponents.

| DTI data processing
Some DTI-derived data, such as fractional anisotropy (FA) and the apparent diffusion coefficient (ADC), provide information on WM diffusion (Basser & Pierpaoli, 1996). FA is a composite measure of three eigenvalues (λ 1 , λ 2 , and λ 3 ). λ 1 , the largest in three, which is called the axial diffusivity (AD), is the component parallel to, and λ 2 and λ 3 , whose average is called the radial diffusivity (RD), are components perpendicular to the axonal fibers (Basser, 1995;Wozniak & Lim, 2006). We selected FA as a main index because it measures the degree of water diffusion anisotropy on a scale from zero to one and characterizes WM microstructural abnormalities (Basser & Pierpaoli, 1996). In addition, FA is the most widely used anisotropy measure (O'Donnell & Westin, 2011). We used Philips Extended Workspace (EWS, Release 2.6.3.1; Philips) to analyze DTI data. FA threshold for line tracking was set to 0.2. The maximum angle threshold was 50°. We performed tractography using the two-regions-of-interest (ROIs) approach. In recent years, analyses have been carried out by segmenting CC fibers using a two-ROIs approach into multiple regions based on the cortical regions that the CC projects to (Huang et al., 2005;Lebel, Caverhill-Godkewitsch, & Beaulieu, 2010). One of our previous studies applied that technique in mood disorder investigations (Yamada et al., 2015). The first reference ROI was focused on the CC in a midsagittal slice, and secondary ROI was seven separate cortices spanning both sides of the midline (Appendix S1).
As seen in Figure 1, callosal fibers were separated into seven segments based on their cortical projection zones. Ordered from front to back, the seven sections were as follows: orbital frontal, anterior frontal, superior frontal, superior parietal, posterior parietal, temporal, and occipital. All ROIs were drawn in accordance with specific anatomical landmarks and guidelines that were followed carefully and consistently for all patients ( Figure 2). Same as the previous studies (Brandstack, Kurki, Laalo, Kauko, & Tenovuo, 2016;Huang et al., 2005;Lebel et al., 2010), fibers that were clearly not part of the anatomical connectivity of the tracking were manually removed with exclusion ROIs to include only fibers within the desired tract.

| Statistics
The differences between the SZ and HC groups in age and z-scores of each neurocognitive test were examined by independent-samples F I G U R E 1 Segmentation of corpus callosum by tractography.
The corpus callosum was subdivided into seven separate segments using a two-regions-of-interest (ROIs) fiber tracking approach in accordance with a determined rule and specific anatomical landmarks. The seven segments are, in order from most front to most back, as follows: orbital frontal (OF), anterior frontal (AF), superior frontal (SF), superior parietal (SP), posterior parietal (PP), temporal (Temp), and occipital (

| Demographic and clinical characteristics
There were no differences in age and gender between the SZ and HC groups (Table 1). In the BACS, the SZ group showed significantly lower z-scores in six subtests and composite scores when compared to those in the HC groups (Table 1).

| Correlation of the FA of callosal fibers with age and equivalent dose of antipsychotics
The FA of callosal fibers significantly correlated with age in the su-

| Correlation of the z-score of neuropsychological tests and equivalent doses of antipsychotics
There were no significant correlations between the z-score of neu-

| Stepwise multiple regression analysis of the FA
In all subjects, the FA of callosal fibers in the temporal segment was significantly associated with the z-scores of token motor test, Tower of

| Differences in the FA of callosal fiber in the segment which identified the regression in the SZ group between the SZ and HC groups
Independent-samples t test revealed significant differences in FA between the SZ and HC groups in temporal segment (t = 2.80, p = 0.008) but not in posterior parietal segments (t = 1.75, p = 0.088; Figure 4 and Table 2).

| Statistical analysis on the ADC, AD, and RD
In all subjects, the regression for verbal memory identified the ADC   In the AD, independent-samples t test revealed significant differences in orbital frontal segment (t = −3.06, p = 0.004) but not in occipital segment (t = −1.05, p = 0.300). In the ADC and RD, independent-samples t test revealed no significance.

| D ISCUSS I ON
In the current study, we extracted and divided the CC fibers into seven regions based on the cortical projection regions, and investigated the relationship between the FA values of the CC fibers and cognitive function in schizophrenia. In the SZ group, FA values of the CC that connects the bilateral temporal lobe cortices associated with executive function scores, and FA value of this tract was significantly decreased compared to the HC group. These results indicate an association between microstructural abnormalities of the CC white matter and cognitive dysfunction in schizophrenia.
In the current study, we found a statistically significant association between white matter abnormalities and executive function impairment. Impairment in executive function is one of the most common dysfunctions observed in disease courses of schizophrenia (Orellana & Slachevsky, 2013). We assessed executive function using the Tower of London test. The Tower of London test requires several cognitive processes including working memory (Elliott, 2003), processing speed, response inhibition (Asato, Sweeney, & Luna, 2006;Zook, Davalos, Delosh, & Davis, 2004), and visuospatial processing (Newman, Carpenter, Varma, & Just, 2003), necessitating functional coordination among multiple cortical and subcortical regions (Unterrainer & Owen, 2006). As white matter fibers connect brain regions, many studies have reported a relationship between white matter abnormalities and cognitive function in schizophrenia (Canu, Agosta, & Filippi, 2015). Executive dysfunction of schizophrenia has been reported to be associated with white matter abnormalities in major fiber bundles that connect frontal and temporal lobes, such as superior longitudinal fasciculus and uncinate fasciculus (Kubicki et al., 2002(Kubicki et al., , 2003Nestor et al., 2004;Pérez-Iglesias et al., 2010).
In schizophrenia, associations also have been reported between superior longitudinal fasciculus and working memory (Karlsgodt et al., 2008), uncinate fasciculus and verbal memory (Nestor et al., 2004;Szeszko et al., 2008), inferior longitudinal and inferior frontooccipital fasciculi and processing speed, verbal learning, and visual learning , diffuse white matter abnormalities and processing speed (Karbasforoushan, Duffy, Blackford, & Woodward, 2015;Rigucci et al., 2013), visual memory (Rigucci et al., 2013), and social cognition (Rigucci et al., 2013). In the current study, we found the significant association between FA values of the CC fibers connecting bilateral temporal lobe cortices and executive function scores.
Studies reported association between bilateral cortical thickness reductions in the temporal lobe and executive dysfunction in schizophrenia (Hartberg et al., 2010), as well as an association between white matter volume reductions in the temporal lobe and verbal memory, attention, problem solving, and working memory dysfunctions in a follow-up study of early-onset schizophrenia (Andreasen et al., 2011). The current results suggest that the disconnection be-  the DTI whole-brain analysis in schizophrenia have been repeatedly reported in the rostrum (Ellison-Wright et al., 2014;Fujino et al., 2014;Gu et al., 2016;Hummer et al., 2016;Kochunov et al., 2014;Kong et al., 2011;Lener et al., 2015;Melicher et al., 2015;Pérez-Iglesias et al., 2010;Pomarol-Clotet et al., 2010;Roalf et al., 2013;Spalletta et al., 2015;Zhang et al., 2014Zhang et al., , 2016, body (Fujino et al., 2014;Melicher et al., 2015;Pérez-Iglesias et al., 2010;Roalf et al., 2013;Zhang et al., 2014Zhang et al., , 2016, and splenium of the CC (Cheung et al., 2008;Ellison-Wright et al., 2014;Fujino et al., 2014;Gasparotti et al., 2009;Melicher et al., 2015;Zhang et al., 2014).  Knöchel et al., 2012;Li et al., 2014;Rotarska-Jagiela et al., 2008). Balevich et al. (2015) divided the CC into five anteroposterior segments, but did not find statistically significant decrease of FA values in any specific segments. In studies with segmentation of the CC into nine regions, statistically significant FA reductions were observed in the inferior and superior genu, isthmus (Knöchel et al., 2012), anterior, middle, posterior genu, posterior body, anterior splenium (Li et al., 2014), inferior and superior genu, and splenium (Rotarska-Jagiela et al., 2008). Whitford et al. parcellated the CC fibers into six segments based on the cortical regions they projected, and examined the difference between patients with schizophrenia and healthy participants (Whitford et al., 2010). They found a statistically significant decrease in FA of the frontal fibers in the patient group. FA values of the temporal fibers were reduced in the schizophrenia group but not statistically significantly so. The discrepancy between the current study and Whitford's study may be partially explained by differences in the DTI methods of analysis and in the study participants. In the current study, we divided the CC fibers based on their cortical projection regions using the two-ROIs approach. We followed the procedures used in relevant previous studies to determine the location of ROIs and exclusion criteria (Huang et al., 2005;Lebel et al., 2010;Yamada et al., 2015), and we achieved tractography with high anatomical accuracy in each region. On the other hand, Whitford et al. segmented the CC into clusters after whole-brain tractography. Our study participants included both men and women; however, Whitford's study included only men. TSA is useful as it delineates how fiber tracts connect functional brain regions, providing information on structural connectivity. Our results suggest microstructural abnormalities in the temporal regions of the CC white matter fibers connecting the two hemispheres.
In our SZ group, we observed statistically significant relation between FA values and PANSS scores in the temporal and posterior parietal segments of CC white matter fibers connecting the two hemispheres. The FA of callosal fibers in the temporal segment was significantly negatively associated with the score of the PANSS negative.
Microstructural abnormalities of callosal fiber in the temporal segment may be associated with negative symptom in schizophrenia.
In our SZ group, the AD value of callosal fibers in the orbital frontal segment significantly associated with attention scores, and the AD value of this tract was significantly increased compared to the HC group. Histological information indicates the AD assesses axonal function (Mac Donald, Dikranian, Bayly, Holtzman, & Brody, 2007). Some previous studies reported no significant difference in AD value of the CC between patients with schizophrenia and healthy controls (Hummer et al., 2016;Kochunov et al., 2014;Liu et al., 2013;Spalletta et al., 2015;Whitford et al., 2010). The largest coordinated meta-analysis on DTI data showed significantly higher AD value in schizophrenia patients compared with healthy controls in the fornix but showed no significant differences in the AD value in the CC (Kelly et al., 2018). To our knowledge, there is no study reported relationship of AD value with cognitive function in schizophrenia. The studies on patients with essential tremor showed positive correlation between the AD value and cognitive function (Julian et al., 2017;Bhalsing et al., 2015). Same as these previous studies on essential tremor, our SZ group showed significant relationship of increased AD with better attention scores, while AD value in the SZ group was significantly increased compared to the HC group. The further studies are needed to interpret this paradoxical relation of AD value with cognitive function in the SZ group.
The current study has some limitations. First, the gender composition of our HC and SZ groups was not fully matched. All participants in the SZ group were taking antipsychotic medications at the time of the MRI examination. Gender ratios were not statistically different between the two groups, and no correlation was found between FA values and antipsychotic medication dosage in the SZ group. However, gender difference and antipsychotic medication use are potential confounding factors for diffusion changes. Second, the sample size was relatively small. Last, no distortion correction was performed in DTI analysis. We offered new insight into relationship between microstructural abnormalities in callosal fibers and cognitive function in schizophrenia, but the results of our study should be confirmed in future studies using more subject with an appropriate control over FA value-affecting factors.
In summary, we found microstructural abnormalities in the CC white matter fibers connecting bilateral temporal lobe cortices contribute to poor executive function and severe negative symptom in patients with schizophrenia. Nakao for their contributions to this study.

CO N FLI C T O F I NTE R E S T
None of the authors have potential conflict of interests to be disclosed.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared.