SCHRES-06676; No of Pages 9 Schizophrenia Research xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Schizophrenia Research journal homepage: www.elsevier.com/locate/schres

Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study Yu Sun a,⁎, Yu Chen a, Renick Lee b, Anastasios Bezerianos a, Simon L. Collinson c, Kang Sim d,e a

Singapore Institute for Neurotechnology (SINAPSE), Centre for Life Sciences, National University of Singapore, Singapore Department of Bioengineering, National University of Singapore, Singapore Department of Psychology, National University of Singapore, Singapore d Department of General Psychiatry, Institute of Mental Health (IMH), Singapore e Department of Research, Institute of Mental Health (IMH), Singapore b c

a r t i c l e

i n f o

Article history: Received 7 March 2015 Received in revised form 8 January 2016 Accepted 12 January 2016 Available online xxxx Keywords: Anatomical networks Connectome Diffusion tensor imaging (DTI) Graph theory Longitudinal Schizophrenia

a b s t r a c t Despite convergent neuroimaging evidence indicating a wide range of brain abnormalities in schizophrenia, our understanding of alterations in the topological architecture of brain anatomical networks and how they are modulated over time, is still rudimentary. Here, we employed graph theoretical analysis of longitudinal diffusion tensor imaging data (DTI) over a 5-year period to investigate brain network topology in schizophrenia and its relationship with clinical manifestations of the illness. Using deterministic tractography, weighted brain anatomical networks were constructed from 31 patients experiencing schizophrenia and 28 age- and gender-matched healthy control subjects. Although the overall small-world characteristics were observed at both baseline and follow-up, a scan-point independent significant deficit of global integration was found in patients compared to controls, suggesting dysfunctional integration of the brain and supporting the notion of schizophrenia as a disconnection syndrome. Specifically, several brain regions (e.g., the inferior frontal gyrus and the bilateral insula) that are crucial for cognitive and emotional integration were aberrant. Furthermore, a significant group-by-longitudinal scan interaction was revealed in the characteristic path length and global efficiency, attributing to a progressive aberration of global integration in patients compared to healthy controls. Moreover, the progressive disruptions of the brain anatomical network topology were associated with the clinical symptoms of the patients. Together, our findings provide insights into the substrates of anatomical dysconnectivity patterns for schizophrenia and highlight the potential for connectome-based metrics as neural markers of illness progression and clinical change with treatment. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Schizophrenia is a complex neuropsychiatric disorder with a myriad of clinical manifestations (Howes and Murray, 2014). Whilst the precise neural substrates underpinning the clinical manifestations of schizophrenia are far from understood, the disorder is thought to stem from neurodevelopmental abnormalities of brain structure and function. Using neuroimaging techniques, convergent evidence has revealed a wide range of brain abnormalities, including a general reduction of whole brain volume, increases in ventricular volume (McDonald et al., 2006), and volume reductions in frontal, temporal, limbic, parietal, thalamic gray matter (GM) (Douaud et al., 2007; Ellison-Wright and Bullmore, 2010). More recently, aberrations of white matter (WM) involving frontal and temporal cortices (Kuswanto et al., 2012; Kyriakopoulos and Frangou, 2009; Szeszko et al., 2005), corpus ⁎ Corresponding author at: Centre for Life Sciences, National University of Singapore, Singapore. E-mail address: [email protected] (Y. Sun).

callosum (Collinson et al., 2014), and cingulum (Abdul-Rahman et al., 2011) have been observed. A recent conceptualization suggests that the human brain forms a large-scale network of interconnected regions within the human connectome that provides the anatomical substrate for neural communication. Accumulated studies have shown that healthy brain networks have special topological organizations, including small-worldness (high local clustering and short paths between nodes), as well as highly connected network regions (hubs), and modular structure (for reviews, see (Boccaletti et al., 2006; Bullmore and Sporns, 2009)). Changes in topology have been related to normal cognitive development and to a wide range of brain diseases, including schizophrenia. The current pathophysiological theories of schizophrenia suggests that the clinical emergence of the disorder represents a failure of integration of functional and anatomical brain connectivity because the heterogeneous presentation of schizophrenia (i.e., disorganized, positive, and negative symptoms) may arise from variability in abnormalities of interregional interactions rather than from abnormality in a specific regions (Fitzsimmons et al., 2013; Friston, 1998; Konrad and Winterer, 2008;

http://dx.doi.org/10.1016/j.schres.2016.01.025 0920-9964/© 2016 Elsevier B.V. All rights reserved.

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025

2

Y. Sun et al. / Schizophrenia Research xxx (2016) xxx–xxx

Pettersson-Yeo et al., 2011; Uhlhaas, 2013; van den Heuvel and Fornito, 2014). Recent advances in non-invasive neuroimaging techniques such as diffusion tensor imaging (DTI) and graph theoretical analysis have enabled quantitative mapping of brain anatomical networks in unprecedented detail. Studies of structural brain networks in schizophrenia have found the presence of small-world properties in these individuals albeit that there is reduction of local network connectivity (Zalesky et al., 2011), increases of minimum path length and network robustness (Zhang et al., 2012), loss of hubs in frontal regions (Bassett et al., 2008; van den Heuvel et al., 2010), abnormal rich club organization (highly interconnected hubs) (van den Heuvel et al., 2013). Notwithstanding the significance of these findings, evidence pertaining to the intactness of overall brain anatomical connectivity has not been entirely consistent (for reviews, see (Fornito et al., 2012; Griffa et al., 2013)). It is also worth noting that reported aberrations in structural brain networks are found exclusively in cross sectional studies. As structural changes can manifest and alter at various stages throughout life, longitudinal studies are crucial if a more comprehensive understanding of brain architecture differences and their implications is to be achieved (Pfefferbaum et al., 2013). Although several longitudinal volumetric studies have been successful in shedding light on important focal changes in GM and WM of patients with schizophrenia (Andreasen et al., 2011; Asami et al., 2012; Whitford et al., 2007), the question of how network properties in schizophrenia are conserved or affected over time is still largely unexplored. To the best of our knowledge, this is the first study employing graph theory analysis for investigating longitudinal effects of schizophrenia on structural brain networks. By applying a longitudinal design over 5 years, we recorded repeated DTI images in 31 patients with schizophrenia and 28 age- and gender-matched healthy individuals. Wholebrain anatomical networks were constructed using the commonly used deterministic tractography approach. We calculated several network measures to assess small-world properties (e.g., clustering coefficient, path length, and small-worldness), global and local efficiencies, and relative nodal characteristics. In the context of significance of investigation structural brain network topological changes in schizophrenia and paucity of longitudinal data, we set out to assess: 1) how network architecture is aberrant in schizophrenia, 2) how these disruptions change over time, and 3) whether there is any longitudinal association between the disrupted network topology and clinical variables. 2. Methods and materials 2.1. Participants In this study, thirty-one patients experiencing schizophrenia and twenty-eight matched healthy comparison subjects were recruited at baseline from the Institute of Mental Health (IMH), Singapore, and the local community by advertisements respectively. All the subjects participated in the follow up study with a mean gap of around 5 years. Scan intervals of each participant were shown in Fig. 1. Diagnostic evaluation was performed by a board-certificated psychiatrist (K. S.). The inclusion and exclusion criterial are detailed in the Supplementary materials. This study was approved by the Institutional Review Boards of the IMH, Singapore, as well as the National Neuroscience Institute (NNI), Singapore, and informed consent was obtained from each participant. Antipsychotic medication dosage was recorded at baseline and mean dose at follow-up was calculated by averaging the cumulative received antipsychotic dose over the period of treatment. The socio-demographic and clinical features of the subjects are shown in Table 1. 2.2. Data acquisition Structural magnetic resonance images with consistent high signalto-noise ratio were recorded using a 3-Tesla whole body scanner (Philips Achieva, Philips, Medical System, Eindhoven, The

Fig. 1. Age at scan for longitudinal study. Each subject is shown in a different row, with their scans connected by a straight line. Healthy participants (blue) and patients with schizophrenia (red) are marked separately. Most subjects received two scans approximately 5 years apart. There was no statistical (p N 0.05) difference in scan intervals between both groups. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Netherlands) using an eight-element SENSE receiver head-coil at the National Neuroscience Institute, Singapore. A T1-weighted Magnetization Prepared Rapid Gradient Recalled Echo sequence (repetition time [TR] = 7.2 ms; echo time [TE] = 3.3 ms; flip angle = 8°) was utilized to obtain high-resolution T1-weighted MRI volume images (each volume contains 180 gapless axial slices of 0.9 mm thickness, field of view [FOV] = 230 × 230 mm2; acquisition matrix = 256 × 256; inplane resolution: 0.9 × 0.9 mm2) in the direction of the anterior–posterior commissures (AC–PC). A single-shot echo-planar sequence (TR = 3275 ms; TE = 56 ms; flip angle = 90°; b-factor = 800 s/mm2; 1 baseline image with b = 0 s/mm2) from 15 separate non-parallel directions was utilized to obtain diffusion encoded images (each volume containing 42 slices, 3.0 mm with no gap; FOV = 230 × 230 mm2; acquisition matrix = 112 × 109, reconstructed to 256 × 256). For each participant, the diffusion sequences were scanned three times to improve the signal-to-noise ratios. During the scanning, head motion was minimized using restraining foam pads provided by the manufacturer. The same scanner was used for both the baseline (software version R2.6) and follow-up scans (software version R3.2). The scanning settings were maintained for both baseline and follow-up studies. 2.3. Data preprocessing and structural brain network construction Data preprocessing and structural brain network construction were conducted using FSL (Smith et al., 2004), diffusion toolkit (Wang et al., 2007), and PANDA (Cui et al., 2013), and had been described in detail previously (Sun et al., 2015). In short, preprocessing approaches included correction for head motion and eddy current distortions through registering the DW images to the b0 image with an affine transformation. The gradient direction of each DWI volume was rotated according to the resultant affine transformations to further reduce the influence of motion artifacts (Leemans and Jones, 2009). Six elements of the diffusion tensor were then estimated from which fractional anisotropy (FA) was calculated. Whole-brain fiber tractography was subsequently performed using fiber assignment by continuous tracking (FACT) algorithm (Mori et al., 1999). This algorithm computes fiber trajectories starting from the deep WM regions and terminating at a voxel with a

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025

Y. Sun et al. / Schizophrenia Research xxx (2016) xxx–xxx

3

Table 1 Demographic and clinical features of the samples.a Characteristic

Group (patients/controls = 31/28)

Statistical

Patients with schizophrenia

Healthy controls

t57

p

Baseline scan Age (years) Gender: male/female WRAT scoresd Handedness: right/left Education (years) Age of onset Duration of illness (years) Medication dosage (mg/day)e Antipsychotic medication type: typical/atypical PANSS positive symptomsf PANSS negative symptoms PANSS general symptoms PANSS total

19–54 (31.9 ± 9.7) 14/17 33–57 (50.2 ± 5.8) 29/2 9–16 (12.0 ± 2.0) 15–47 (24.8 ± 7.4) 3.4–40.4 (11.8 ± 8.6) 0–700 (234.7 ± 184.7) 6/25 7–20 (9.9 ± 3.5) 7–23 (9.2 ± 3.9) 16–32 (20.0 ± 3.4) 30–68 (39.2 ± 8.3)

22–54 (31.8 ± 9.3) 13/15 40–57 (49.9 ± 4.6) 25/3 10–18 (14.3 ± 2.1) – – – – – – – –

0.033 – 0.270 – 4.338 – – – – – – – –

0.974b 0.922c 0.788b 0.557c b0.001b – – – – – – – –

Second scang Follow-up interval (months)h Medication dosage (mg/day)e Antipsychotic medication type: typical/atypical PANSS positive symptoms PANSS negative symptoms PANSS general symptoms PANSS total

26–83 (54.5 ± 14.9) 0–550 (221.3 ± 164.7) 6/25 7–14 (8.3 ± 2.2) 7–15 (9.4 ± 2.6) 16–28 (19.1 ± 3.4) 30–55 (36.8 ± 6.3)

38–83 (61.3 ± 12.5) – – – – – –

−1.870 – – – – – –

0.067b – – – – – –

a b c d e f g h

Data are expressed as the range of minimum–maximum (mean ± standard deviation). The p-value was obtained using a two-sample two-tailed t-test. The p-value was obtained using a two-tailed Pearson χ2 test. Wide Range Achievement Test (WRAT) reading subscales was estimated according to (Stone et al., 1995) for the IQ evaluation. Chlorpromazine (CPZ) equivalents doses were calculated using conversion rates according to (Woods, 2003). The positive and negative symptoms scale (PANSS) (Kay et al., 1987) was used to assess the psychopathology and symptom severity. Comparisons of clinical metrics between baseline and follow-up were shown in supplementary Fig. 1. Scan intervals of each participant were shown in Fig. 1.

turning angle greater than 45° or at a voxel with FA less than 0.15. For each subject, the structural brain network was then constructed by combining the individual parcellation map with the WM tractography. Here the widely used automated anatomical labeling (AAL) parcellation scheme (90 regions in total) was used (TzourioMazoyer et al., 2002). These regions served as nodes within the structural brain network. The parcellation process was conducted in the native DTI space for each subject. Edge weight (wij) was computed as the multiplication of fiber number (FN) by the mean FA along the fiber bundles between a pair of cortical regions, w ij = FAij × FNij (Betzel et al., 2014; Lo et al., 2010). As a result, we constructed the weighted structural brain network (90 × 90) for each participant at baseline and follow-up. Details about the network construction can be found in the Supplementary materials (Supplementary Fig. 2).

2.4. Graph theoretical analysis of structural brain networks Graph theoretical analysis was adopted to provide quantitative metrics to examine any difference in topological organization of structural brain network between schizophrenia and healthy comparison controls. In this study, we investigated the network architecture at both global and regional levels for the constructed structural brain networks. Global network architecture was quantified in terms of small-world properties (weighted clustering coefficient, Cw; weighted characteristic path length, Lw; and small-worldness, σ) and efficiency (global efficiency, Eglobal; and local efficiency, Elocal). An exploratory analysis of the behavior of each node was described in terms of nodal efficiency (Enodal(i)). Here we provide brief, formal definitions of each of the metrics used in this study (Table 2). Greater details of the formations and interpretations of the graph theoretical parameters can be found in the Supplementary materials and reviews of this topic (Boccaletti et al., 2006; Bullmore and Sporns, 2009; Rubinov and Sporns, 2010).

2.5. Statistical analysis 2.5.1. Longitudinal group differences Separate two-sample two-tailed t-test was used to test the group differences in baseline age, WRAT scores, years of education, and follow-up intervals. The gender and handedness data were analyzed using a two-tailed Pearson χ2 test. To assess longitudinal effects on network attributes between patients and normal controls, a general linear

Table 2 Introduction of topological measurements (including five global network metrics and one nodal metric) and their meaning in structural brain networks. Network properties Global characteristics Weighted clustering coefficient (Cw) Weighted characteristic path length (Lw)

Measurement and meaning Cw measures the extent of a local density or cliquishness of the network. Lw measures the average minimal travel distance between nodes in the network. It represents the global integration of the network.

Small-worldness (σ)

σ¼

Local efficiency (Elocal)

and Lrand denote property of a network, where Crand w w the average Cw and Lw of an ensemble of 100 surrogate random networks. A small-world network has high local clustering and short paths between brain regions. Elocal is a measure of the information exchange at the clustering level. Eglobal is a measure of the global efficiency of parallel information transfer in the network. It is inversely related to Lw.

Global efficiency (Eglobal)

Nodal characteristics Nodal efficiency (Enodal)

C w =C rand w Lw =Lrand w

is a scalar measurement of the small-world

Enodal(i) is the inverse of the harmonic mean of the shortest path length between node i and all other nodes. A region with high Enodal indicates great interconnectivity with other regions in the network.

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025

4

Y. Sun et al. / Schizophrenia Research xxx (2016) xxx–xxx

model (GLM) comprising of longitudinal scan point (baseline vs. followup) as a within-subject factor, group (NC vs. SCZ) as a between-subject factor, and the scan point by group as an interaction, was performed on the obtained network metrics. Gender, baseline age, and gender-by-age interaction were set as covariates. If any main effect was found to be significant, further post-hoc t-tests were performed (paired t-test for longitudinal scan point effect and two-sample two-tailed t-test for group effect). All statistical analyses were performed using SPSS 17 software. A value of p b 0.05 was considered significant for the global properties analysis. An uncorrected p-value of 0.01 was considered for establishment of a significant difference for the regional characteristics. To address the problem of multiple comparisons, effects were also tested on whether they survived a false discovery rate (FDR) threshold of q = 0.05. 2.5.2. Relationship between network metrics and clinical variables Relationships between the global network measures and clinical variables were also explored in the patient group. Specifically, a generalized estimated equation (GEE) model (Zeger et al., 1988) with exchangeable working correlation matrix was employed and implemented in the statistical package, R (http://www.r-project.org/) in the current study. Effects of gender, baseline age, and age-by-gender interaction were controlled for the correlation analyses. The threshold value for establishment of a significant relationship was set at p b 0.05. 3. Results 3.1. Clinical measurements There were no statistical differences (all p N 0.05) between the healthy participants and patients with schizophrenia with respect to age at scan, gender, baseline WRAT scores, handedness, or the time interval between scans (follow-up interval) (Table 1). For the clinical measurements of patients between the baseline and follow-up study, a significant decrease (t30 = 2.188, p = 0.037) was found in the PANSS positive symptoms, whereas the rest of the clinical metrics (medication dosage, PANSS negative symptoms, PANSS general symptoms, PANSS overall scores) did not show any significant change (all p N 0.05) over time (Supplementary Fig. 1). 3.2. Group differences of global topological organization over time The brain networks of both groups exhibited typical features of small-world topology at both scans (Supplementary Fig. 3). Additional quantitative statistical analyses revealed significant changes in global network metrics between both groups over time (Table 3). Specifically,

Table 3 Comparison of longitudinal brain network topographical changes between patients with schizophrenia and healthy participants. Network metrics

Cw Lw σ Eglobal Elocal

a main effect of longitudinal scan was observed in the small-worldness, σ, (follow-up N baseline, F1, 57 = 4.991, p = 0.029). Group main effect was highly significant in Lw (NC b SCZ, F1, 54 = 9.920, p = 0.003) and Eglobal (NC N SCZ, F1, 54 = 9.863, p = 0.003), suggesting a reduced global integration of the brain network in patients with schizophrenia. Interestingly, significant interaction was revealed in Lw (F1, 57 = 7.833, p = 0.007) and Eglobal (F1, 57 = 4.787, p = 0.033). Post hoc analysis shows that this significant interaction resulted from different development trends in normal controls and patients with schizophrenia (Fig. 2). i.e., compared to healthy controls who exhibit progressive improvements in global integration (Lw: baseline N follow-up, t27 = 2.531, p = 0.018; Eglobal: baseline b follow-up, t27 = − 1.795, p = 0.084), patients with schizophrenia showed insignificant worsen global integration in the follow-up scan (Lw: baseline b follow-up, t30 = − 1.482, p = 0.149; Eglobal: baseline N follow-up, t30 = 1.182, p = 0.246).

3.3. Group differences of regional topological organization over time Significant group effect was revealed on the nodal characteristic of 6 regions (Fig. 3(A), Supplementary Table 2), where four of them (the left inferior frontal gyrus, triangular part, [IFGtriang.L] (p = 0.004), the right paracentral lobule, [PCL.R] (p = 0.002), and the bilateral insula, [INS] (INS.L, p = 0.004; INS.R, p = 0.002)) exhibited normal controls advantage, i.e., NC N SCZ. Regions with significant ‘NC b SCZ’ effect were the left superior frontal gyrus, medial part, [SFGmed.L] (p = 0.007) and the left supramarginal gyrus, [SMG.L] (p = 0.007). In addition, both positive and negative time effects were found at 11 regions in total (Fig. 3 (B), Supplementary Table 2), where half of these regions (6 among 11), predominantly located in the left temporal and right parietal areas, including the left fusiform gyrus, [FFG.L] (p = 0.0003 *, * indicates region survived FDR threshold at q b 0.05); the left Heschl gyrus, [HES.L] (p = 0.0006 *); the left lingual gyrus, [LING.L] (p b 0.0001 *); the left caudate nucleus, [CAU.L] (p = 0.001 *) and the right caudate nucleus, [CAU.R] (p = 0.009), showed decreased nodal efficiency in the followup scan. Regions with significant increased nodal efficiency were mainly resided in the left occipital and right temporal areas, including the left superior occipital gyrus, [SOG.L] (p = 0.005); the right superior temporal gyrus, [STG.R] (p = 0.004); the right temporal pole, superior part, [TPOsup.R] (p = 0.0010 *); the bilateral temporal pole, middle part, [TPOmid] (TPOmid.L, p = 0.0002 * and TPOmid.R p = 0.0008 *). More interesting, a significant interaction effect was observed in the left inferior frontal gyrus, opercula part, [IFGoperc.L] (p = 0.010) and right thalamus, [THA.R] (p = 0.009) (Fig. 3(C), Supplementary Table 2). The posthoc analysis with regard to the interaction revealed that this significant interaction effect was attributed to the significant progressive increase of nodal efficiency in healthy volunteers and a non-significant decrease of nodal characteristics over time in patients with schizophrenia.

3.4. Relationship between network topology and clinical features

General linear model (GLM) Group F1, 54 (p-value)

Scan-point F1, 57 (p-value)

Interaction F1, 57 (p-value)

2.487 (0.121) 9.920 (0.003)▲ 0.074 (0.787) 9.863 (0.003)▼ 1.534 (0.221)

0.078 (0.782) 0.428 (0.516) 4.991 (0.029)↑ 0.325 (0.571) 0.064 (0.801)

0.372 (0.545) 7.833 (0.007) 0.354 (0.554) 4.787 (0.033) 0.970 (0.329)

Global network metrics are expressed as mean ± standard deviation. The statistical results were computed with a general linear model (GLM) with longitudinal scan point as a within-subject fact, group as between-subject factor, and longitudinal scan point by group as interaction. The effect of age at baseline, gender, and age-by-gender interaction were adjusted for all of these analyses. Bold indicates variables that are statistically significant (p b 0.05). Note: ▼, patients b controls, ▲, patients N controls, ↑, follow-up N baseline.

We found a longitudinal association between reducing PANSS general scores and increasing weighted characteristic path length, Lw, (standardized coefficient, β = − 2.535, p = 0.046) and decreasing global efficiency, Eglobal, (β = 3.211, p = 0.013). Of note, Lw measures the overall routing efficiency of the network and is inversely related to Eglobal, hence leading to converse correlations for both. Specifically, a development trend of insignificantly reduced PANSS general scores significantly correlates with a progression toward longer characteristics path length or smaller global efficiency. More interestingly, significantly reduced PANSS positive scores was found to be correlated with the smallworldness (β = −3.520, p = 0.036) in patients with schizophrenia, revealing different development trends between lower PANSS positive scores and higher small-worldness.

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025

Y. Sun et al. / Schizophrenia Research xxx (2016) xxx–xxx

5

Fig. 2. Post-hoc statistical analysis of global network metrics. Normal controls and patients with schizophrenia are marked separately: NC1 = normal controls at baseline (unfilled blue triangle), NC2 = normal controls at follow-up (filled blue triangle), SCZ1 = patients at baseline (unfilled red circle), and SCZ2 = patients at follow-up (filled red circle). Median values of the network metrics are marked with a horizontal line for each group. Each number represents the p-value of a t-test (paired t-test for longitudinal scan point effect and two-sample two-tailed t-test for group effect). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

4. Discussion By applying a longitudinal design over 5 years, this study investigated the longitudinal changes in structural brain networks in schizophrenia for the first time. The significant findings are as follows: first, a scan-point independent significant deficit of global integration was found in patients with schizophrenia; second, compared with healthy controls who exhibited an improvement in global integration, patients with schizophrenia showed a decrease of global integration over time; third, two specific brain regions (the left inferior frontal gyrus, opercula part, and the right thalamus) showed a significant group-by-longitudinal scan interaction effect, attributing to a progressive increase of

nodal efficiency in healthy controls compared to patients; and fourth, the disruption of the brain anatomical network was longitudinally correlated with the clinical symptom ratings on the PANSS in patients with schizophrenia. The identification of small-world architecture has made a great impact on our understanding the topological organization of brain networks (Bullmore and Sporns, 2009). In particular, small-world architecture is characterized by high local clustering of connections between neighboring brain regions but with short path length. This characteristic is thought to provide the brain with an optimal structure to simultaneously support locally segregated and globally integrated processing (Sporns, 2011). Moreover, this neural architecture has the

Fig. 3. The spatial distribution of cortical regions showing significant effect of (A) group, (B) scan point, and (C) group-by-scan point interaction on the nodal characteristics and the posthoc statistical analysis for the significant interaction effect. The color bar represents F values. Significant (gray background, p b 0.05, FDR-corrected; no-background, p b 0.01, uncorrected) regions are overlaid on inflated surface maps at the Medium view with BrainNet Viewer software (Xia et al., 2013). The nodal regions are located according to their centroid stereotaxic coordinates. For the abbreviations of the cortical regions, see Supplementary Table 1. Of note, the subcortical regions including the bilateral CAU in (B) were not presented in the surface spatial distribution here. L = left, R = right. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025

6

Y. Sun et al. / Schizophrenia Research xxx (2016) xxx–xxx

capacity to process information in parallel, which is computationally much more efficient than serial or hierarchical processing. Here we found that the brain structural networks of both groups exhibited typical features of small-world topology at both scans. These findings were consistent with prior observations in both healthy participants (IturriaMedina et al., 2008) and patients with schizophrenia (van den Heuvel et al., 2010; Zalesky et al., 2011). Given that the small-world topology is robust to random and targeted disruptions, the presence of typical small-world architecture in patients with schizophrenia at both scan points may suggest compensatory reactions to developmental neuropathology. In addition, we found a significant deficit of global integration independent of scan points in patients with schizophrenia. This observation was consistent with prior findings from functional (Liu et al., 2008) and structural network studies (Ottet et al., 2013; Wang et al., 2012; Zhang et al., 2012), which have shown increased connection distance within brain networks in schizophrenia in comparable age groups. Therefore, our finding adds to earlier work and provides further evidence supporting the notion of schizophrenia as a disconnection syndrome. More importantly, we found a significant interaction effect in the global integration, i.e., patients with schizophrenia exhibited a development trajectory heading toward less global efficiency of brain anatomical networks while healthy controls showed an improvement in global integration. In accordance with our observation in healthy subjects, Wu and colleagues found in their longitudinal study of healthy adults that improved brain topological architecture related to structural brain network changes occurring from young to middle age (Wu et al., 2013). Similar observations of improved network integration were also found in a recent cross sectional study of the structural connectome in adolescents and adults (Dennis et al., 2013). Previous evidence suggested that high global integration assures effective integrity or rapid transfers of information across remote regions that are believed to constitute the basis of cognitive processing (Sporns and Zwi, 2004). More recently, Li et al. revealed a positive correlation between the global integration of anatomical brain networks and intellectual performance (Li et al., 2009). Therefore, our finding of improvement in global integration in normal controls may reflect a maturation process in the network. Whilst there is no comparable longitudinal anatomical data involving smallworld properties in schizophrenia, these trends were consistent with most cross sectional observations to data across different age spans, suggesting that disruption of brain topological networks and impaired higher order neural function in schizophrenia occurs over time (for reviews, see (Fornito et al., 2012; van den Heuvel and Fornito, 2014)). In one meta-analysis of longitudinal volumetric deficits in schizophrenia investigating 27 studies with a total of 928 patients, Olabi et al. found progressive brain volume reduction (Olabi et al., 2011). More recently, Asami and colleagues showed in their longitudinal neuroimaging volumetric study that patients with schizophrenia showed widespread GM volume reductions in brain neocortical regions including frontal, parietal, and limbic regions when compared to heathy participants and these volumetric reductions were not correlated with medication dosage (Asami et al., 2012). Cerebral GM contains neuronal cell bodies, dendrites, and short protrusions which play an important role in information processing. Our observation of a progressive trend toward less economical topology of brain networks might be attributed to the significant concomitant loss of GM in patients with schizophrenia occurring over time. Aberrations of nodal characteristics are increasingly implicated in normal aging (Wu et al., 2012) or various brain diseases (Crossley et al., 2014), and may underlie the neuropsychiatric basis for schizophrenia symptoms (for a review, see (Rubinov and Bullmore, 2013)). In line with previous studies, we found that schizophrenia was associated with reduced regional efficiency in the left inferior frontal gyrus, right paracentral lobule as well as the bilateral insula and increased nodal efficiency in the left superior frontal gyrus and the left supramarginal gyrus. Insula, a cortical structure with extensive

connections to many areas of the cortex and limbic system, has repeatedly showed functional and structural deficits in patients with schizophrenia (for a review, see (Wylie and Tregellas, 2010)). Moreover, in healthy subjects, the insula and inferior frontal gyrus were typically identified as hubs, which played a central role in receiving convergent inputs from multiple cortical regions (He et al., 2007; Iturria-Medina et al., 2008; Wu et al., 2012). Taken together, we speculate that the profoundly affected regional centrality in these hub regions may indicate more isolated network architecture in schizophrenia, leading toward the observed deficits in global integration. In addition, both positive and negative longitudinal time effects on regional efficiency were found across cerebral cortex, suggesting a putative compensatory mechanism of cortical network reorganization over time (Park and Reuter-Lorenz, 2009). The results of this longitudinal study extend the findings of aging modulated regional efficiency in several crosssectional connectivity studies (Dennis et al., 2013; Gong et al., 2009b) and support the view that longitudinal changes are primary characteristics of the association and paralimbic cortex as opposed to primary cortex (Albert and Knoefel, 2011). Of note, several brain regions, including the superior occipital gyrus, the superior temporal gyrus, the fusiform gyrus, and the lingual gyrus, consistently exhibited an alteration of regional efficiency with normal aging (Gong et al., 2009b; Wu et al., 2012). Furthermore, we found a significant interaction effect on nodal efficiency of the left inferior frontal gyrus, opercula part, [IFGoperc.L] and the right thalamus, [THA.R] attributed to a deficit of improvement in patients. In previous longitudinal volumetric neuroimaging studies, considerable loss of both GW and WM in these regions was consistently revealed in schizophrenia (Andreasen et al., 2011; Asami et al., 2012). Another interesting finding of the current study is that within patients, the longitudinal course of the alterations of the network topological properties was associated with clinical symptom progression on the PANSS. Specifically, we found different progression trends between significantly reduced PANSS positive scores and the increased smallworldness. Since the small-worldness represents the balance between local clustering and global integration; the smaller PANSS positive scores over time, the higher the small-worldness, the correlation was in the expected direction. When combined with the observations of small-world characteristics across two scan points, this finding might suggest that maintenance of the optimal small-world properties is related to improvement in positive symptoms. Moreover, a longitudinal association between the reduced PANSS general symptoms and worsening global integration was also revealed, suggesting some degree of specificity in the way that distinct symptoms correlate with overall connectivity variations in different neural circuits (Fornito et al., 2012; Meyer-Lindenberg and Weinberger, 2006). As revealed by van den Heuvel and Fornito in their recent review paper of connectomic findings in schizophrenia, a robust relationship between abnormal network organization and schizophrenia clinical symptoms has emerged (van den Heuvel and Fornito, 2014). Our findings therefore support a general trend in the literature showing a direct correspondence between network measurements and clinical symptomatology and suggest the usefulness of brain network properties as potential biomarkers for evaluation the severity and progression of the disease. Longitudinal investigations of the disruptions of structural brain networks in schizophrenia, as in this study, are significant for several reasons. First, they allow better appreciation of the extent of brain network properties that are affected over time in schizophrenia which may suggest specific biomarkers of illness progression. Second, how these brain network properties relate to specific brain regions over time can highlight underlying neural pathways disruptions in these complex brain networks. Third, prospective anatomical connectivity disturbances can complement extant and future functional connectivity evaluation in order to better understand the circuitry disturbances in schizophrenia. Fourth, examining changes in the network properties over time can suggest mechanisms being exerted to maintain the efficiency of brain networks in illness. Taken together, we believe that

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025

Y. Sun et al. / Schizophrenia Research xxx (2016) xxx–xxx

employing longitudinal design in addition to cross sectional studies is important for more comprehensive understanding of the neural mechanisms of various brain diseases and how they manifest and evolve throughout life. Several issues need to be further addressed. First, the effect of different medication dosage among patients across both scan points might potentially be a confounding factor. For instance, previous neuroimaging studies of schizophrenia have reported pharmacological changes in localized brain regions and connections (Andreasen et al., 2011; Kanaan et al., 2009; Navari and Dazzan, 2009). Nonetheless, evidence pertaining to the association between network measurements and medication dosage are not entirely consistent (Liu et al., 2008; Micheloyannis et al., 2006). Some have suggested that medication is unlikely to be a confounding factor and may, on the contrary, exert to a normalizing influence (Rubinov et al., 2009). Moreover, the significant difference between groups in years of education may also be a potential confounding factor. To address these issues, we performed separate statistical analysis with medication dosage and years of education as additional covariates and found the key results of the study remain unchanged (data not shown). As such, we consider that the observed longitudinal effect reflects the intrinsic disease process rather than the effect of direct pharmacological treatment or group differences in years of education. Second, in this study, structure brain network edge weight was constructed with considering the fiber number (FN) and FA values. FA value is an important index to evaluate fiber integrity (Beaulieu, 2002) and exhibits a high correlation with conductivity (Tuch et al., 2001); and FN may reflect the white matter structure (Hagmann et al., 2007). Through calculating the multiplication of FN and FA, the edge weight would provide a comprehensive view of white matter anatomical structure. To obtain the FN and FA values, a computationally inexpensive deterministic local tensor based tractography method was used. However, due to the fiber crossing problem, the deterministic method always terminates the fiber tracking when it reaches regions with fiber crossing and low FA values, which may lead to the loss of some existing fibers between brain regions or to the inclusion of some non-existent fibers (Li et al., 2009). In the Supplementary materials (Supplementary Fig. 4), we assessed the credibility of our tracking results through showing seven well-known WM fiber bundles (including 2 short white matter tracts and 5 major tracts) from four randomly selected subjects at baseline. Consequently, the reconstructed fiber bundles are faithful to the human WM anatomy from previous studies (Gong et al., 2009a; Li et al., 2009). Furthermore, structural connectivity networks reconstructed in this study exhibited attributes that are consistent with previous cross-sectional brain connectome studies of schizophrenia (Griffa et al., 2013; Uhlhaas, 2013; van den Heuvel and Fornito, 2014). Nonetheless, a probabilistic tractography may be a better solution for future studies as recent studies have demonstrated the better performance in overcoming the fiber crossings and robustness to the image noise (Buchanan et al., 2014). Third, the widely used AAL template (Tzourio-Mazoyer et al., 2002) was used to define the nodes of the brain anatomical networks. Recent studies have revealed that different parcellation scales might result in different properties of brain networks (Fornito et al., 2010; Zalesky et al., 2010). Furthermore, regions on the AAL template differ in size, which may have a confounding effect on the link weight of the network nodes (van den Heuvel et al., 2010; Wang et al., 2012). To verify this, we performed a post-hoc analysis on the region size difference between both groups at both baseline and follow-up and found no significant result (data not shown), suggesting an equivalent effect of region size on the network metrics in both groups. Although several network edge weighting methods (e.g., streamline density and streamline density with fiber length correction) have been introduced to compensate the ROI size effect (Buchanan et al., 2014; Hagmann et al., 2008), the choice of the most accurate representation of the underlying neurobiological connectivity remains an open question (Jones et al., 2013). The primary focus of the current work is to investigate the longitudinal disruption of

7

the brain anatomical network in schizophrenia. We believe that graph theoretical analyses with different spatial resolutions and more accurate edge weighting method in future studies would be important for better understanding the etiology and its progression of schizophrenia. Finally, as our study is one of the first exploratory investigations of longitudinal effects of schizophrenia on the structural brain network, an uncorrected p-value of 0.01 was employed for establishing the significance and presenting the regional results. It is possible that some of the nodal results may have occurred by chance and some caution is needed when interpreting these results. In the current study, we focused primarily on the interpretation of the general pattern of the findings and highlighted those survived multiple comparisons correction for the reader's interpretation. Nonetheless, future studies using a larger independent longitudinal study sample with multiple intervals and hypothesis-driven study design are expected to confirm our observations. In conclusion, in this first-ever brain connectome study of longitudinal changes in brain structural networks in schizophrenia, we found progressive disruption of brain anatomical networks in patients with schizophrenia, which correlated with the clinical symptoms longitudinally. These findings highlight the potential of brain network measures as neural biomarkers for clinical presentation, illness progress as well as response to treatment. Role of funding source This work was supported by the National Healthcare Group (NHG 11003 & NHG 12003) awarded to Kang Sim, and the Agency for Science, Technology, Research/ Singapore BioImaging Consortium (ASTART/SBIC009/2006) awarded to Kang Sim. The authors thank the National University of Singapore for supporting the Cognitive Engineering Group at the Singapore Institute for Neurotechnology (SINAPSE) under grant number R719-001-102-232. The sponsors of the study had no role in study design, data collection, data analysis, results interpretation, writing the paper, and the decision to submit the paper for publication. Contributors Kang Sim conceived, designed and performed the experiments. Yu Sun., Renick Lee, and Yu Chen undertook the data analysis, performed the statistical analysis and literature search. Yu Sun., Anastasios Bezerianos, Simon Collinson, and Kang Sim interpreted the results and wrote the first draft of the manuscript. All authors contributed and approved the final manuscript for publication. Kang Sim had full access to all of the data in the study. Conflict of interest All authors have reported no known biomedical financial interests or other potential conflicts of interest. All the grant and financial support as well as technical support has been listed. Acknowledgements The authors are very grateful to Dr. Chan Yiong Huak for his insightful advices in relation to the statistical aspects of this work. The authors would also like to thank all patients and controls for their participation.

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.schres.2016.01.025. References Abdul-Rahman, M.F., Qiu, A., Sim, K., 2011. Regionally specific white matter disruptions of fornix and cingulum in schizophrenia. PLoS One 6 (4), e18652. Albert, M.L., Knoefel, J.E., 2011. Clinical Neurology of Aging. Oxford University Press. Andreasen, N.C., Nopoulos, P., Magnotta, V., Pierson, R., Ziebell, S., Ho, B.C., 2011. Progressive brain change in schizophrenia: a prospective longitudinal study of first-episode schizophrenia. Biol. Psychiatry 70 (7), 672–679. Asami, T., Bouix, S., Whitford, T.J., Shenton, M.E., Salisbury, D.F., McCarley, R.W., 2012. Longitudinal loss of gray matter volume in patients with first-episode schizophrenia: DARTEL automated analysis and ROI validation. NeuroImage 59 (2), 986–996. Bassett, D.S., Bullmore, E., Verchinski, B.A., Mattay, V.S., Weinberger, D.R., MeyerLindenberg, A., 2008. Hierarchical organization of human cortical networks in health and schizophrenia. J. Neurosci. 28 (37), 9239–9248. Beaulieu, C., 2002. The basis of anisotropic water diffusion in the nervous system — a technical review. NMR Biomed. 15 (7-8), 435–455.

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025

8

Y. Sun et al. / Schizophrenia Research xxx (2016) xxx–xxx

Betzel, R.F., Byrge, L., He, Y., Goni, J., Zuo, X.N., Sporns, O., 2014. Changes in structural and functional connectivity among resting-state networks across the human lifespan. NeuroImage 102, 345–357. Boccaletti, S., Latora, V., Moreno, Y., Chavez, M., Hwang, D.-U., 2006. Complex networks: structure and dynamics. Phys. Rep. 424 (4), 175–308. Buchanan, C.R., Pernet, C.R., Gorgolewski, K.J., Storkey, A.J., Bastin, M.E., 2014. Test-retest reliability of structural brain networks from diffusion MRI. NeuroImage 86, 231–243. Bullmore, E., Sporns, O., 2009. Complex brain networks: graph theoretical analysis of structural and functional systems. Nat. Rev. Neurosci. 10 (3), 186–198. Collinson, S.L., Gan, S.C., Woon, P.S., Kuswanto, C., Sum, M.Y., Yang, G.L., Lui, J.M., Sitoh, Y.Y., Nowinski, W.L., Sim, K., 2014. Corpus callosum morphology in first-episode and chronic schizophrenia: combined magnetic resonance and diffusion tensor imaging study of Chinese Singaporean patients. Br. J. Psychiatry 204 (1), 55–60. Crossley, N.A., Mechelli, A., Scott, J., Carletti, F., Fox, P.T., McGuire, P., Bullmore, E.T., 2014. The hubs of the human connectome are generally implicated in the anatomy of brain disorders. Brain 137 (Pt 8), 2382–2395. Cui, Z., Zhong, S., Xu, P., He, Y., Gong, G., 2013. PANDA: a pipeline toolbox for analyzing brain diffusion images. Front. Hum. Neurosci. 7, 42. Dennis, E.L., Jahanshad, N., McMahon, K.L., de Zubicaray, G.I., Martin, N.G., Hickie, I.B., Toga, A.W., Wright, M.J., Thompson, P.M., 2013. Development of brain structural connectivity between ages 12 and 30: a 4-Tesla diffusion imaging study in 439 adolescents and adults. NeuroImage 64, 671–684. Douaud, G., Smith, S., Jenkinson, M., Behrens, T., Johansen-Berg, H., Vickers, J., James, S., Voets, N., Watkins, K., Matthews, P.M., James, A., 2007. Anatomically related grey and white matter abnormalities in adolescent-onset schizophrenia. Brain 130 (Pt 9), 2375–2386. Ellison-Wright, I., Bullmore, E., 2010. Anatomy of bipolar disorder and schizophrenia: a meta-analysis. Schizophr. Res. 117 (1), 1–12. Fitzsimmons, J., Kubicki, M., Shenton, M.E., 2013. Review of functional and anatomical brain connectivity findings in schizophrenia. Curr. Opin. Psychiatry 26 (2), 172–187. Fornito, A., Zalesky, A., Bullmore, E.T., 2010. Network scaling effects in graph analytic studies of human resting-state FMRI data. Front. Syst. Neurosci. 4, 22. Fornito, A., Zalesky, A., Pantelis, C., Bullmore, E.T., 2012. Schizophrenia, neuroimaging and connectomics. NeuroImage 62 (4), 2296–2314. Friston, K.J., 1998. The disconnection hypothesis. Schizophr. Res. 30 (2), 115–125. Gong, G., He, Y., Concha, L., Lebel, C., Gross, D.W., Evans, A.C., Beaulieu, C., 2009a. Mapping anatomical connectivity patterns of human cerebral cortex using in vivo diffusion tensor imaging tractography. Cereb. Cortex 19 (3), 524–536. Gong, G., Rosa-Neto, P., Carbonell, F., Chen, Z.J., He, Y., Evans, A.C., 2009b. Age- and genderrelated differences in the cortical anatomical network. J. Neurosci. 29 (50), 15684–15693. Griffa, A., Baumann, P.S., Thiran, J.P., Hagmann, P., 2013. Structural connectomics in brain diseases. NeuroImage 80, 515–526. Hagmann, P., Kurant, M., Gigandet, X., Thiran, P., Wedeen, V.J., Meuli, R., Thiran, J.P., 2007. Mapping human whole-brain structural networks with diffusion MRI. PLoS One 2 (7), e597. Hagmann, P., Cammoun, L., Gigandet, X., Meuli, R., Honey, C.J., Wedeen, V.J., Sporns, O., 2008. Mapping the structural core of human cerebral cortex. PLoS Biol. 6 (7), e159. He, Y., Chen, Z.J., Evans, A.C., 2007. Small-world anatomical networks in the human brain revealed by cortical thickness from MRI. Cereb. Cortex 17 (10), 2407–2419. Howes, O.D., Murray, R.M., 2014. Schizophrenia: an integrated sociodevelopmentalcognitive model. Lancet 383 (9929), 1677–1687. Iturria-Medina, Y., Sotero, R.C., Canales-Rodriguez, E.J., Aleman-Gomez, Y., Melie-Garcia, L., 2008. Studying the human brain anatomical network via diffusion-weighted MRI and Graph Theory. NeuroImage 40 (3), 1064–1076. Jones, D.K., Knosche, T.R., Turner, R., 2013. White matter integrity, fiber count, and other fallacies: the do's and don'ts of diffusion MRI. NeuroImage 73, 239–254. Kanaan, R., Barker, G., Brammer, M., Giampietro, V., Shergill, S., Woolley, J., Picchioni, M., Toulopoulou, T., McGuire, P., 2009. White matter microstructure in schizophrenia: effects of disorder, duration and medication. Br. J. Psychiatry 194 (3), 236–242. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 13 (2), 261–276. Konrad, A., Winterer, G., 2008. Disturbed structural connectivity in schizophrenia primary factor in pathology or epiphenomenon? Schizophr. Bull. 34 (1), 72–92. Kuswanto, C.N., Teh, I., Lee, T.S., Sim, K., 2012. Diffusion tensor imaging findings of white matter changes in first episode schizophrenia: a systematic review. Clin. Psychopharmacol. Neurosci. 10 (1), 13–24. Kyriakopoulos, M., Frangou, S., 2009. Recent diffusion tensor imaging findings in early stages of schizophrenia. Curr. Opin. Psychiatry 22 (2), 168–176. Leemans, A., Jones, D.K., 2009. The B-matrix must be rotated when correcting for subject motion in DTI data. Magn. Reson. Med. 61 (6), 1336–1349. Li, Y., Liu, Y., Li, J., Qin, W., Li, K., Yu, C., Jiang, T., 2009. Brain anatomical network and intelligence. PLoS Comput. Biol. 5 (5), e1000395. Liu, Y., Liang, M., Zhou, Y., He, Y., Hao, Y., Song, M., Yu, C., Liu, H., Liu, Z., Jiang, T., 2008. Disrupted small-world networks in schizophrenia. Brain 131 (Pt 4), 945–961. Lo, C.Y., Wang, P.N., Chou, K.H., Wang, J., He, Y., Lin, C.P., 2010. Diffusion tensor tractography reveals abnormal topological organization in structural cortical networks in Alzheimer's disease. J. Neurosci. 30 (50), 16876–16885. McDonald, C., Marshall, N., Sham, P.C., Bullmore, E.T., Schulze, K., Chapple, B., Bramon, E., Filbey, F., Quraishi, S., Walshe, M., Murray, R.M., 2006. Regional brain morphometry in patients with schizophrenia or bipolar disorder and their unaffected relatives. Am. J. Psychiatry 163 (3), 478–487. Meyer-Lindenberg, A., Weinberger, D.R., 2006. Intermediate phenotypes and genetic mechanisms of psychiatric disorders. Nat. Rev. Neurosci. 7 (10), 818–827. Micheloyannis, S., Pachou, E., Stam, C.J., Breakspear, M., Bitsios, P., Vourkas, M., Erimaki, S., Zervakis, M., 2006. Small-world networks and disturbed functional connectivity in schizophrenia. Schizophr. Res. 87 (1-3), 60–66.

Mori, S., Crain, B.J., Chacko, V.P., van Zijl, P.C., 1999. Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann. Neurol. 45 (2), 265–269. Navari, S., Dazzan, P., 2009. Do antipsychotic drugs affect brain structure? A systematic and critical review of MRI findings. Psychol. Med. 39 (11), 1763–1777. Olabi, B., Ellison-Wright, I., McIntosh, A.M., Wood, S.J., Bullmore, E., Lawrie, S.M., 2011. Are there progressive brain changes in schizophrenia? A meta-analysis of structural magnetic resonance imaging studies. Biol. Psychiatry 70 (1), 88–96. Ottet, M.C., Schaer, M., Debbane, M., Cammoun, L., Thiran, J.P., Eliez, S., 2013. Graph theory reveals dysconnected hubs in 22q11DS and altered nodal efficiency in patients with hallucinations. Front. Hum. Neurosci. 7, 402. Park, D.C., Reuter-Lorenz, P., 2009. The adaptive brain: aging and neurocognitive scaffolding. Annu. Rev. Psychol. 60, 173–196. Pettersson-Yeo, W., Allen, P., Benetti, S., McGuire, P., Mechelli, A., 2011. Dysconnectivity in schizophrenia: where are we now? Neurosci. Biobehav. Rev. 35 (5), 1110–1124. Pfefferbaum, A., Rohlfing, T., Rosenbloom, M.J., Chu, W., Colrain, I.M., Sullivan, E.V., 2013. Variation in longitudinal trajectories of regional brain volumes of healthy men and women (ages 10 to 85 years) measured with atlas-based parcellation of MRI. NeuroImage 65, 176–193. Rubinov, M., Bullmore, E., 2013. Schizophrenia and abnormal brain network hubs. Dialogues Clin. Neurosci. 15 (3), 339–349. Rubinov, M., Sporns, O., 2010. Complex network measures of brain connectivity: uses and interpretations. NeuroImage 52 (3), 1059–1069. Rubinov, M., Knock, S.A., Stam, C.J., Micheloyannis, S., Harris, A.W., Williams, L.M., Breakspear, M., 2009. Small-world properties of nonlinear brain activity in schizophrenia. Hum. Brain Mapp. 30 (2), 403–416. Smith, S.M., Jenkinson, M., Woolrich, M.W., Beckmann, C.F., Behrens, T.E., Johansen-Berg, H., Bannister, P.R., De Luca, M., Drobnjak, I., Flitney, D.E., Niazy, R.K., Saunders, J., Vickers, J., Zhang, Y., De Stefano, N., Brady, J.M., Matthews, P.M., 2004. Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage 23 (Suppl. 1), S208–S219. Sporns, O., 2011. The human connectome: a complex network. Ann. N. Y. Acad. Sci. 1224, 109–125. Sporns, O., Zwi, J.D., 2004. The small world of the cerebral cortex. Neuroinformatics 2 (2), 145–162. Stone, M.H., Jastak, S., Wilkinson, G., 1995. Wide Range Achievement Test (WRAT-3). Rasch Meas. Trans. 403. Sun, Y., Chen, Y., Collinson, S.L., Bezerianos, A., Sim, K., 2015. Reduced hemispheric asymmetry of brain anatomical networks is linked to schizophrenia: a connectome study. Cereb. Cortex [Epub ahead of print]. Szeszko, P.R., Ardekani, B.A., Ashtari, M., Kumra, S., Robinson, D.G., Sevy, S., Gunduz-Bruce, H., Malhotra, A.K., Kane, J.M., Bilder, R.M., Lim, K.O., 2005. White matter abnormalities in first-episode schizophrenia or schizoaffective disorder: a diffusion tensor imaging study. Am. J. Psychiatry 162 (3), 602–605. Tuch, D.S., Wedeen, V.J., Dale, A.M., George, J.S., Belliveau, J.W., 2001. Conductivity tensor mapping of the human brain using diffusion tensor MRI. Proc. Natl. Acad. Sci. U. S. A. 98 (20), 11697–11701. Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N., Mazoyer, B., Joliot, M., 2002. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. NeuroImage 15 (1), 273–289. Uhlhaas, P.J., 2013. Dysconnectivity, large-scale networks and neuronal dynamics in schizophrenia. Curr. Opin. Neurobiol. 23 (2), 283–290. van den Heuvel, M.P., Fornito, A., 2014. Brain networks in schizophrenia. Neuropsychol. Rev. 24 (1), 32–48. van den Heuvel, M.P., Mandl, R.C., Stam, C.J., Kahn, R.S., Hulshoff Pol, H.E., 2010. Aberrant frontal and temporal complex network structure in schizophrenia: a graph theoretical analysis. J. Neurosci. 30 (47), 15915–15926. van den Heuvel, M.P., Sporns, O., Collin, G., Scheewe, T., Mandl, R.C., Cahn, W., Goni, J., Hulshoff Pol, H.E., Kahn, R.S., 2013. Abnormal rich club organization and functional brain dynamics in schizophrenia. JAMA psychiatry 70 (8), 783–792. Wang, R., Benner, T., Sorensen, A., Wedeen, V., 2007. Diffusion toolkit: a software package for diffusion imaging data processing and tractography. Proc. Intl. Soc. Mag. Reson. Med., p. 3720 Wang, Q., Su, T.P., Zhou, Y., Chou, K.H., Chen, I.Y., Jiang, T., Lin, C.P., 2012. Anatomical insights into disrupted small-world networks in schizophrenia. NeuroImage 59 (2), 1085–1093. Whitford, T.J., Grieve, S.M., Farrow, T.F., Gomes, L., Brennan, J., Harris, A.W., Gordon, E., Williams, L.M., 2007. Volumetric white matter abnormalities in first-episode schizophrenia: a longitudinal, tensor-based morphometry study. Am. J. Psychiatry 164 (7), 1082–1089. Woods, S.W., 2003. Chlorpromazine equivalent doses for the newer atypical antipsychotics. J. Clin. Psychiatry 64 (6), 663–667. Wu, K., Taki, Y., Sato, K., Kinomura, S., Goto, R., Okada, K., Kawashima, R., He, Y., Evans, A.C., Fukuda, H., 2012. Age-related changes in topological organization of structural brain networks in healthy individuals. Hum. Brain Mapp. 33 (3), 552–568. Wu, K., Taki, Y., Sato, K., Qi, H., Kawashima, R., Fukuda, H., 2013. A longitudinal study of structural brain network changes with normal aging. Front. Hum. Neurosci. 7, 113. Wylie, K.P., Tregellas, J.R., 2010. The role of the insula in schizophrenia. Schizophr. Res. 123 (2-3), 93–104. Xia, M., Wang, J., He, Y., 2013. BrainNet Viewer: a network visualization tool for human brain connectomics. PLoS One 8 (7), e68910.

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025

Y. Sun et al. / Schizophrenia Research xxx (2016) xxx–xxx Zalesky, A., Fornito, A., Harding, I.H., Cocchi, L., Yucel, M., Pantelis, C., Bullmore, E.T., 2010. Whole-brain anatomical networks: does the choice of nodes matter? NeuroImage 50 (3), 970–983. Zalesky, A., Fornito, A., Seal, M.L., Cocchi, L., Westin, C.F., Bullmore, E.T., Egan, G.F., Pantelis, C., 2011. Disrupted axonal fiber connectivity in schizophrenia. Biol. Psychiatry 69 (1), 80–89.

9

Zeger, S.L., Liang, K.Y., Albert, P.S., 1988. Models for longitudinal data: a generalized estimating equation approach. Biometrics 44 (4), 1049–1060. Zhang, Y., Lin, L., Lin, C.P., Zhou, Y., Chou, K.H., Lo, C.Y., Su, T.P., Jiang, T., 2012. Abnormal topological organization of structural brain networks in schizophrenia. Schizophr. Res. 141 (2-3), 109–118.

Please cite this article as: Sun, Y., et al., Disruption of brain anatomical networks in schizophrenia: A longitudinal, diffusion tensor imaging based study, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.01.025