Abstract Analyzing the schizophrenia connectome can identify illness-related alterations in connectivity across the brain. An important question that remains unanswered is whether connectivity alterations are already evident at the onset of illness, before treatment with antipsychotic medication and possible influences of neuroprogressive or secondary alterations related to chronic illness duration. In the present study, diffusion tensor imaging and deterministic fiber tractography were performed with 137 antipsychotic-naive first-episode schizophrenia patients and 113 matched healthy controls. Using graph theoretic analysis, groups were compared in global and regional measurements and modularity of white matter connectivity. Compared with controls, the patients showed significantly decreased total connection strength. Furthermore, patients demonstrated significantly decreased connections within and between brain modules. Several local brain regions within association cortex exhibited reduced nodal centralities and abnormal participant coefficient or intra-module degree, some of which were correlated with illness duration and overall functional disability. In never-treated schizophrenia patients, networks showed a less effective organizational pattern of white matter pathways. White matter disconnectivity occurred not only within but also between multiple modules, shedding light on the deficits of anatomical network organization early in the course of schizophrenia. schizophrenia, white matter connectivity, diffusion tensor imaging, graph theory, small-world, module Introduction Schizophrenia is increasingly being conceptualized as a disorder that results from altered interactions between brain regions that cause psychotic symptoms and abnormal cognitive and emotional function.1 This model has been supported by structural and functional psychoradiology findings.2–4 However, there have been conflicting reports in the localization of white matter alterations across previous studies, which may result from small samples, variable illness duration and antipsychotic and other drug treatment history.5–9 Another potential reason for inconsistent findings in diffusion tensor imaging (DTI) studies of white matter pathways is that widespread and subtle pathology across the brain might be imperfectly detected by traditional regional or voxel-based analysis. In recent years, connectomics has been developed as an analytic strategy that considers the whole brain as a set of interconnected networks.10 Analyzing the human connectome permits the calculation of connectivity parameters that can identify alterations that occur across the brain and illustrate the complete set of inter-regional interactions that comprise the brain’s intricate anatomic networks.11 Network analyses have shown that the human brain is organized into segregated modules, which represent subcomponents of the brain and facilitate the construction of a complex system.12 Graph analysis of functional imaging data has shown reduced global efficiency and increased clustering coefficients in un-medicated schizophrenia patients,13 and lower connectivity strength, reduced clustering, and longer path length14 suggesting less optimal functional networks, and increased15 or unchanged14 global efficiency in medicated patients. Previous studies of white matter networks in schizophrenia have yielded inconsistent findings including elevated clustering16 and reduced global efficiency.17 Furthermore, modularity findings have led to the proposal that abnormal modularity contributes to the breakdown of mechanisms for information processing in functional brain networks in schizophrenia.18 The nature of structural connectome and modularity alterations early in the course of illness before potential drug, neuroprogressive and other secondary factors that can influence brain networks over the course of illness has not yet been systematically examined. Analysis of white matter pathways using connectomic approaches in never treated first episode patients may provide important insights into the alterations of brain structural networks in schizophrenia. Therefore, the present study recruited a large sample of antipsychotic-naive first-episode schizophrenia patients, and sought to identify changes of structural networks using DTI and graph theoretic analysis. We hypothesized that: (1) patients would show abnormalities of global and regional topological measurements and modularity of white matter connectivity networks; (2) the identified topological alterations would be related to clinical features of schizophrenia. Methods Subjects A total of 137 drug-naive first-episode schizophrenia patients (55 males and 82 females; mean age [±SD], 24.1 y ± 7.6, range, 16–46 y ; mean education 12.5 y ± 3.1, range 3–20 y; mean duration of illness, 7.4 mo ± 8.6, range 0.03–36 mo) were recruited. The diagnosis of schizophrenia was made using the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID), and symptom severity was assessed using the Positive and Negative Syndrome Scale (PANSS) and the Global Assessment Function (GAF) scale (table 1). Duration of untreated illness was evaluated by the Nottingham Onset Schedule19 (with information provided by patients, guardians and family members), which was defined as the time period from the onset of first psychotic symptom to the date of research assessment. No one in the patients group was receiving treatment with antipsychotics and any other psychiatric drugs at the time of MRI scanning. The study was approved by the local research ethics committee. Written informed consent was obtained from all participants, and for minors, parents also provided written informed consent. Table 1. Demographic and Clinical Characteristics of Antipsychotic-Naive First-Episode Schizophrenia Patients and Control Subjects Patients Controls P Value Number 137 133 Gender (male/female) 55/82 66/67 .117 Age (range), y 24.1 ± 7.6 (16–46) 26.5 ± 9.1 (16–50) .092 Education (range), y 12.5 ± 3.1 (3–20) 12.9 ± 3.2 (2–24) .304 Illness duration (range), mo 7.4 ± 8.6 (0.03–36) GAF scores (range) 29.9 ± 10.0 (12–65) PANSS Total (range) 93.2 ± 20.9 (34–136) Positive symptoms (range) 24.6 ± 6.9 (7–38) Negative symptoms (range) 17.5 ± 7.3(7–38) General psychopathology (range) 45.4 ± 10.5 (18–71) Patients Controls P Value Number 137 133 Gender (male/female) 55/82 66/67 .117 Age (range), y 24.1 ± 7.6 (16–46) 26.5 ± 9.1 (16–50) .092 Education (range), y 12.5 ± 3.1 (3–20) 12.9 ± 3.2 (2–24) .304 Illness duration (range), mo 7.4 ± 8.6 (0.03–36) GAF scores (range) 29.9 ± 10.0 (12–65) PANSS Total (range) 93.2 ± 20.9 (34–136) Positive symptoms (range) 24.6 ± 6.9 (7–38) Negative symptoms (range) 17.5 ± 7.3(7–38) General psychopathology (range) 45.4 ± 10.5 (18–71) Note: PANSS, Positive and Negative Syndrome Scale; GAF, Global Assessment Function. View Large A total of 133 healthy controls were recruited (66 males and 67 females; mean age, 26.5 y ± 9.1, range, 16–50 y; mean education, 12.9 y ± 3.2, range 2–24 y). They were evaluated using the SCID Non-Patient version to rule out current or past psychiatric illness. Also, none had a known history of psychiatric illness in first-degree relatives. Exclusion criteria for both groups included the following: left handed, any neurological disorder, neurosurgery, current or past substance abuse or dependence, pregnancy, significant systemic illness and MR contraindications including cardiac pacemakers and other metallic implants. Diagnostic quality MR images were inspected by 2 experienced neuroradiologists to exclude subjects with gross abnormalities. Schizophrenia patients and healthy controls did not differ with respect to age (P = .092), gender (P = .117), years of education (P = .304) or handedness (all right handed) (table 1). Data Acquisition All subjects were scanned using a 3-T magnetic resonance imaging system (EXCITE; General Electric). In this study, we used 2 DTI scanning protocols, which we refer to as MR1 (56 patients and 86 controls) and MR2 (81 patients and 47 controls) (supplementary table S1). The 2 similar sequences differ only in that the second MR protocol used with later-recruited participants increased the slice number to consistently include the whole cerebellum (number of slices: MR1, 42; MR2, 50), with a necessary increase in repetition time (TR) to obtain the additional slices (TR: MR1, 10 000 milliseconds [ms]; MR2 12 000 ms). High-resolution T1-weighted images were acquired using a 3-dimensional spoiled gradient recalled sequence. Details of scanning parameters are provided in supplementary materials. There were no significant differences in age or sex in either participant group across the 2 MRI protocols (supplementary table S1). Furthermore, comparison of patients and controls examined with the 2 DTI protocols showed no significant differences in primary graph characteristics (total connection strength and characteristic shortest path length) (details in supplementary materials) (supplementary figure S1). Hence, data from MR1 and MR2 scanning protocols were pooled as planned for statistical analysis. To further ensure that scan sequence variation would not confound group comparisons, we included scan sequence as a covariate in all statistical analyses. Data Preprocessing and Network Construction With the quality control procedures for DTI data, patient and control samples did not differ in head motion during scans (see supplementary materials). For each subject, the eddy-current induced distortion and head-motion of diffusion weighted images were corrected by registering them to the b0 image with an affine transformation using the PANDA20 pipeline tool and FMRIB Software Library (FSL, http://www.fmrib.ox.ac.uk/fsl). Deterministic fiber tractography proceeded until either it turned an angle greater than 45° or the fractional anisotropy (FA) was less than 0.221 using the Fiber Assignment by Continuous Tracking algorithm.22 We used the atlas of Automated Anatomical Labeling23 to parcellate the gray matter into 90 non-cerebellar anatomical regions of interest. In native diffusion space, 2 regions were considered structurally connected if at least 1 fiber bundle with 2 endpoints was located in the 2 regions.24 We calculated the FA values between the end-nodes as its weight and obtained a weighted symmetrical anatomical 90 × 90 matrix for each subject. Details are provided in supplementary materials. Network Analysis Threshold Selection. We defined a network cost threshold for which each graph had the same number of edges. Consistent with previous studies,25,26 we selected a range of cost thresholds for the white matter connectivity network based on the following criteria: (1) the averaged degree (the degree of a node is the number of connections linked to the node) over all nodes of each thresholded network was larger than 2 × log(90)27; and (2) the small-worldness of the thresholded networks was larger than 1.1 for all participants.25 Based on these criteria, we defined thresholds ranging from 0.1 to 0.19 (with an interval of 0.01). Network Metrics. Graph theoretical analyses were carried out on white matter connectivity networks using the Brain Connectivity Toolbox (http://www.brain-connectivity-toolbox.net).28 The overall clustering coefficient (Cnet) is a measure of the local interconnectivity of the network, and the characteristic shortest path length (Lnet) quantifies the ability for information propagation in the network. To estimate small-world properties, we compared the Cnet and Lnet of the brain network with those of random networks (Crandom and Lrandom). A small-world network has a similar shortest path length, but higher clustering coefficient than a random network, ie, normalized clustering coefficient (gamma, γ) = Cnet/Crandom > 1, normalized shortest path length (lambda, λ) = Lnet/Lrandom ≈ 1, and the small-worldness scalar (sigma, σ) = γ/λ will be larger than 1 in the case of small-world organization. The regional characteristics included the nodal degree (Si), the nodal efficiency (Ei) and the nodal clustering coefficient (Ci) (details in the supplementary materials). The area under the curve (AUC) for each network metric was calculated to provide an overall value for the topological characterization of brain networks independent of any specific cost threshold.25 Modular Architecture. We computed a modularity measure (Newman’s measure of modularity Q)29 to evaluate the degree to which the brain connectivity network is subdivided into specific modules.30 A module in the network can be defined as a subdivision that has more connections within the module than outside the module.31 Using all the participants as the modular segment distribution, we calculated the mean within and between module connectivity, participation coefficient (PC) and intra-module degree (MD) to measure the regional role of each node in the modular architecture (details in supplementary materials). Statistical Analysis We found high correlations among total connection strength (Snet) of the network, Cnet, and Lnet (supplementary table S2), indicating possible redundancy of information provided by these parameters. To address the issue of redundancy, we retained the γ, λ and σ parameters, because those 3 measures are indispensable to measure whether a network has a small-world organization or not, and kept Snet in the primary analysis, which can measure the total connection strength of a network and is independent from γ, λ, and σ (supplementary table S2), as previous studies have demonstrated.24 Therefore, we restricted primary analyses to parameters γ, λ, σ, and Snet. For a full presentation of the data, we performed the secondary analysis to compare the Cnet and Lnet between groups, which are used to calculate γ and λ. We used a nonparametric permutation approach (1000 iterations)32 to test for group differences in the AUC of each network metric (γ, λ, σ, and Snet), modularity (Q) and modular characteristics (within and between module connectivity) with Bonferroni correction for multiple comparisons. We also compared the nodal properties (Si, Ei, and Ci) of the whole brain network, and the PC and MD of each node in the modular architecture between patients and controls using permutation tests. Statistical significance was corrected for multiple comparisons using Bonferroni correction for Si, Ei, and Ci (P < .05/90) and false-positive adjustment for PC and MD (P < 1/90),15,33 respectively. Exploratory 2-tailed Pearson’s correlation analyses were conducted between clinical variables (PANSS and GAF), duration of illness and all altered topological characteristics in the patient group. Age, gender, and years of education were treated as covariates in all analyses. Results Overall Network Topology Both schizophrenia patients and healthy controls showed a small-world organization of white matter networks expressed by a σ > 1 (mean of AUC: patients, 1.89 ± 0.17; controls, 1.88 ± 0.15) with a γ > 1 (mean of AUC: patients, 1.81 ± 0.17; controls: 1.81 ± 0.16) and λ ≈ 1 (mean of AUC: patients, 0.96 ± 0.02; controls, 0.99 ± 0.16). These overall organization characteristics of structural networks did not differ between patients and controls (for σ, P = .35; for γ, P = .36; and for λ, P = .37) (figure 1A), suggesting an intact overall organization of white matter networks in schizophrenia consistent with previous studies.32 However, with Bonferroni correction (P < .05/4), patients showed significantly decreased Snet (P = .0008) compared with controls (figure 1A). For a full presentation of the data, we also performed the secondary analysis to compare the Cnet and Lnet between groups using permutation test, and found decreased Cnet (P = .0034) and increased Lnet (P = .0038) in patients relative to controls (figure 1A). Fig. 1. View large Download slide Overall white matter connectivity topological and nodal characteristics (details in supplementary materials). Fig. 1. View large Download slide Overall white matter connectivity topological and nodal characteristics (details in supplementary materials). Nodal Characteristics With Bonferroni correction (P < .05/90), decreased Ci in schizophrenia patients was found in the right orbitofrontal cortex (OFC) inferior to the superior frontal gyrus, bilateral hippocampus, right putamen, left superior temporal gyrus, and temporal pole of the right superior temporal gyrus. Decreased Ei in patients was found in the bilateral OFC inferior to middle frontal gyrus (MFG), the orbital part of the left inferior frontal gyrus (IFG), left amygdala and postcentral gyrus, bilateral superior parietal lobule (SPL), right supramarginal gyrus and left angular gyrus. Decreased Si in patients was found in the OFC inferior to the right MFG, the orbital part of bilateral IFG, left rolandic operculum, bilateral amygdala and SPL, and pole of right middle temporal gyrus (table 2A and figure 1B). The Ci of left superior temporal gyrus (r = −.195, P = .022) and Si of left rolandic operculum (r = −.176, P = .04) were both negatively correlated with illness duration (figure 1C). Table 2. (A) Brain Areas Showing Decreased Nodal Clustering Coefficient (Ci), Efficiency (Ei) and Degree (Si) in Patients Compared to Controls (P < .05/90 With Bonferroni Correction for the Multiple Comparisons That Were Shown in Bold Font); (B) Brain Areas Showing Altered Module Degree (MD) and Participation Coefficient (PC) in Their Modular Architecture (P < 1/90 With False-Positive Adjustment for the Multiple Comparisons Shown in Bold Font) Side Brain Region AAL Index P Value A Ci Ei Si Reductions in patients relative to controls Right OFC inferior to superior frontal gyrus 6 0.000133 0.001381 0.010983 Left OFC inferior to middle frontal gyrus 9 0.027336 0.000266 0.001704 Right OFC inferior to middle frontal gyrus 10 0.020697 0.000231 0.000163 Left Orbital part of inferior frontal gyrus 15 0.299003 0.000078 0.00029 Right Orbital part of inferior frontal gyrus 16 0.049903 0.000941 0.000157 Left Rolandic operculum 17 0.001106 0.000677 0.000523 Left Hippocampus 37 0.00003 0.001041 0.030347 Right Hippocampus 38 0.000127 0.008631 0.100645 Left Amygdala 41 0.026358 0.000151 0.00016 Right Amygdala 42 0.214891 0.001039 0.000278 Left Postcentral gyrus 57 0.0071 0.000159 0.001734 Left Superior parietal gyrus 59 0.149272 0.000511 0.000399 Right Superior parietal gyrus 60 0.263701 0.000165 0.000082 Right Supramarginal gyrus 64 0.3436 0.000149 0.05721 Left Angular gyrus 65 0.002616 0.000423 0.001735 Right Lenticular nucleus, putamen 74 0.00025 0.000644 0.004501 Left Superior temporal gyrus 81 0.000396 0.005271 0.085166 Right Temporal pole of superior temporal gyrus 84 0.000241 0.008475 0.052387 Right Temporal pole of middle temporal gyrus 88 0.153782 0.000588 0.000153 B MD PC Increases in patients relative to controls Left Middle frontal gyrus 7 0.0023 0.3383 Right Calcarine fissure and surrounding cortex 44 0.0076 0.1758 Right Caudate nucleus 72 0.0107 0.0413 Right Inferior temporal gyrus 90 0.0031 0.0341 Right Precentral gyrus 2 0.4518 0.001 Left Middle occipital gyrus 51 0.066 0.0003 Left Caudate nucleus 71 0.1254 0.0071 Reductions in patients relative to controls Left Cuneus 45 0.01 0.4573 Right Middle frontal gyrus 8 0.1163 0.0079 Right Cuneus 46 0.0265 0.0105 Side Brain Region AAL Index P Value A Ci Ei Si Reductions in patients relative to controls Right OFC inferior to superior frontal gyrus 6 0.000133 0.001381 0.010983 Left OFC inferior to middle frontal gyrus 9 0.027336 0.000266 0.001704 Right OFC inferior to middle frontal gyrus 10 0.020697 0.000231 0.000163 Left Orbital part of inferior frontal gyrus 15 0.299003 0.000078 0.00029 Right Orbital part of inferior frontal gyrus 16 0.049903 0.000941 0.000157 Left Rolandic operculum 17 0.001106 0.000677 0.000523 Left Hippocampus 37 0.00003 0.001041 0.030347 Right Hippocampus 38 0.000127 0.008631 0.100645 Left Amygdala 41 0.026358 0.000151 0.00016 Right Amygdala 42 0.214891 0.001039 0.000278 Left Postcentral gyrus 57 0.0071 0.000159 0.001734 Left Superior parietal gyrus 59 0.149272 0.000511 0.000399 Right Superior parietal gyrus 60 0.263701 0.000165 0.000082 Right Supramarginal gyrus 64 0.3436 0.000149 0.05721 Left Angular gyrus 65 0.002616 0.000423 0.001735 Right Lenticular nucleus, putamen 74 0.00025 0.000644 0.004501 Left Superior temporal gyrus 81 0.000396 0.005271 0.085166 Right Temporal pole of superior temporal gyrus 84 0.000241 0.008475 0.052387 Right Temporal pole of middle temporal gyrus 88 0.153782 0.000588 0.000153 B MD PC Increases in patients relative to controls Left Middle frontal gyrus 7 0.0023 0.3383 Right Calcarine fissure and surrounding cortex 44 0.0076 0.1758 Right Caudate nucleus 72 0.0107 0.0413 Right Inferior temporal gyrus 90 0.0031 0.0341 Right Precentral gyrus 2 0.4518 0.001 Left Middle occipital gyrus 51 0.066 0.0003 Left Caudate nucleus 71 0.1254 0.0071 Reductions in patients relative to controls Left Cuneus 45 0.01 0.4573 Right Middle frontal gyrus 8 0.1163 0.0079 Right Cuneus 46 0.0265 0.0105 Note: OFC, orbitofrontal cortex; AAL, automated anatomical labeling. View Large Modular Organizations The modularity quotient Q monotonically decreased as a function of increasing white matter connectivity cost (figure 1A), consistent with previous findings,34 the AUC of which did not differ between patients and controls. Derived from all participants, identified modules included a left fronto-striato-tempo-parietal network (module I); right ventral orbitofrontal-temporal network (module II); bilateral medial frontal network (module III); right dorsal fronto-striato-parietal network related to attention and spatial functions (module IV); bilateral visual cortex and posterior components of the default-mode network (DMN; module V); and bilateral sensory-motor function network (module VI) (details in supplementary materials) (figure 2A). Fig. 2. View large Download slide Modular organization of the white matter connectivity as derived from all participants, and within and between module connectivity and nodal characteristics (details in supplementary materials). Fig. 2. View large Download slide Modular organization of the white matter connectivity as derived from all participants, and within and between module connectivity and nodal characteristics (details in supplementary materials). Within and Between Module Connectivity Relative to controls, schizophrenia patients demonstrated decreased within module connectivity for modules I (P = .0004), II (P = .0027), III (P = .0046) and IV (P = .0034) (figure 2B). There were no significant differences in the within module connectivity for modules V (P = .1720) and VI (P = .1763) between patients and controls. Decreased between-module connections relative to control subjects were found in schizophrenia patients, including modules I–III (P = .0158), I–V (P = .0096), II–III (P = .0011), II–IV (P = .0005), II–V (P = .0147), II–VI (P = .011), III–IV (P = .0262), III–VI (P = .0009) and V–VI (P = .0002) (figure 2C). However, only 5 differences between the 2 groups survived Bonferroni correction (P < .05/21), with patients showing decreased within module connectivity in modules I, and decreased between-module connections, including modules II–III, II–IV, III–VI, and V–VI. Nodal Alterations in Modules We then explored the nodal alterations of the within and between module networks. With false-positive adjustment (P < 1/90), the MD of left middle frontal gyrus, right calcarine fissure and surrounding cortex, right caudate nucleus, and right inferior temporal gyrus were increased, and the MD of left cuneus was decreased in schizophrenia patients relative to controls. The PC of right precentral gyrus, left middle occipital gyrus, and left caudate nucleus were increased, and the PC of right middle frontal gyrus and right cuneus were decreased in patients relative to controls (table 2B and figure 2D). The MD of the right calcarine fissure and surrounding cortex was negatively correlated with the GAF scores of patients (r = −.217, P = .011) (figure 2E). Discussion Findings from the present study provide insights into the altered topological organization of white matter networks in schizophrenia. There were several specific new findings from this study. First, schizophrenia patients exhibited decreased total connection strength, indicating a reduced capacity for efficient information processing in white matter networks. Second, schizophrenia patients demonstrated decreased within and between module connectivity, especially within and between the bilateral medial frontal module and the right ventral orbitofrontal-temporal module, which may contribute to the diverse alterations of cognitive, affective and behavioral functions in schizophrenia. Third, multiple local brain regions within fronto-temporo-parietal circuitry exhibited reduced nodal centralities and abnormal participant coefficient or intra-module degree, some of which were correlated with illness duration and overall functional disability of the patients. This pattern of findings suggests a decreased hub role of OFC, superior temporal gyrus and parietal regions in the brain networks of schizophrenia patients. Lower total connection strength indicated a decreased ability for parallel information transmission across distributed brain regions, indicating a nonoptimal topological organization with the dysfunctional integration of impaired white matter networks. This observation supports the model of schizophrenia as a dysconnectivity syndrome. Although the neurobiological causes of network alterations of topological properties in schizophrenia is unclear, we speculate that it is relevant to the widely reported abnormal FA of white matter in schizophrenia.2 Genetic studies have shown associations with global reductions in white matter integrity, suggesting an integral role for abnormal FA in the etiology of schizophrenia.35 Previous studies have reported reduced FA in the cingulum bundle tracts, interhemispheric fibers, and the inferior fronto-occipital fasciculus.36 The present FA-weighted white matter networks revealed decreases in the global connectivity profile, suggesting a wide-ranging reduction of the FA of white matter tracts. Since the FA could primarily reflect the myelin integrity,37 this suggests a clinically relevant alteration in myelination of widespread tracts early in the course of schizophrenia.38 In order to explore which modularity alterations contributed to the global reductions in white matter integrity, we conducted a modularity analysis. Modules are parts of a network with many connections within them and fewer connections to other brain regions.12 Schizophrenia patients showed reduced within module connectivity mainly in the left hemisphere (module I in the present study). Previous DTI studies found lower FA of left-hemisphere white matter in first-episode schizophrenia.39,40 Results from the present study replicate and extend reports of decreased connectivity within left hemisphere networks in schizophrenia. This complements previous findings indicating a significant reduction of leftward asymmetry in some key white-matter tracts in schizophrenia, and that attenuated left lateralization is related to greater positive symptoms.41 In addition to analyses of within module pathways, we also characterized between module connectivity and found robust attenuation of between-module connectivity between right hemisphere modules and between the DMN and the sensory-motor areas. Previous studies have found that the functional connectivity of the temporo-parietal network was reduced and associated with auditory hallucinations in schizophrenia,42 as was functional connectivity of the fronto-temporal network,2 and between the DMN and cognitive networks.43 Structural connectivity via white matter fiber tracts provides an anatomical substrate for precise inter-regional synchronization dynamics that are fundamentally important in higher brain functions. Models of developmental miswiring relevant to between-module connectivity have been proposed in schizophrenia, though the developmental ontogeny of such alterations in at-risk populations are needed to test this hypothesis.44 We also explored the role of individual nodes in white matter connectivity. Decreases in nodal centralities among schizophrenia patients were mainly found in the OFC, temporal pole, superior temporal gyrus and adjacent rolandic operculum, and superior and inferior parietal lobules. The decreased nodal centralities of orbitofrontal and parietal areas might contribute to the total connection strength loss observed in patients. Further, since a reduction of gray matter volume and altered neural activity have been reported in the temporal lobe,1 our data indicating aberrant nodal centralities of superior temporal gyrus and adjacent rolandic operculum suggest a reduced capacity for the efficient integration of speech production with auditory processing, an alteration that might be relevant for auditory hallucinations.45,46 Our observation that nodal centralities of the superior temporal gyrus and rolandic operculum were negatively related to the duration of schizophrenia suggests that this impairment may progress over the early course of illness. In modularity analyses, the PC value is close to 1 if the node is extensively linked to all other modules and 0 if it is linked exclusively to other nodes in its own module, and a high value of MD of a node indicates strong within-module connectivity.47 We found decreased PC in right middle frontal gyrus and cuneus in patients, which might be related to the reduced inter-connectivity between module IV, V, and other modules, and to visual working memory deficits.48 We also observed increased PC and MD in nodes of the fronto-striato-temporo-occipital network, suggesting a potential compensatory process associated with attenuated module organization in schizophrenia. However, this result should be interpreted with caution, since there was a negative correlation between the MD of the right calcarine fissure and surrounding cortex and patients’ GAF scores. Our study has limitations that need to be considered. First, given that we acquired a limited number of diffusion directions starting acquisition long before newer diffusion-weighted schemes (eg, Q-ball and diffusion spectrum imaging) were available for more accurate fiber tractography, future research might refine our conclusions. Second, as there is no widely agreed-upon approach for calculating network metrics, and we used the mean FA values of the fibers as the weighting factor in the construction of the graphs. Other measures such as the number of fibers or mean diffusivity could also be considered as weighting factors. Third, we applied a range of thresholds on the correlation matrices so that each graph would have the same number of edges under investigation, which can limit direct comparison to previous studies.17,32,49 In conclusion, using DTI and graph theoretical analysis, we provided evidence for a reduction in small-world brain network properties of schizophrenia patients at the onset of illness. Alterations in white matter connectivity occurred not only within but also between multiple brain modules, which may be relevant to understanding mechanisms of the complex cognitive, affective and sensorimotor alterations associated with schizophrenia. Further, findings from the present investigation of the whole brain connectome indicate that the diverse functional network alterations previously demonstrated in schizophrenia3,50 likely have a basis in the anatomic connectivity of nodes. These observations are consistent with the idea that widespread neocortical alterations are involved in illness pathogenesis, and that this widespread pattern of abnormalities is evident in white matter tracts at the onset of illness. Additionally, the present study adds to the developing psychoradiology (https://radiopaedia.org/articles/psychoradiology), a new field of radiology, which seems primed to play a major clinical role in guiding diagnostic and treatment planning decisions in patients with psychiatric disorders.51,52 Further research is needed to evaluate dynamic brain changes over the course of treatment, and the neurophysiological and neurocognitive correlates of white matter network alterations in patients with schizophrenia. Supplementary Material Supplementary data are available at Schizophrenia Bulletin online. Funding This study was supported by the National Natural Science Foundation (grant nos. 81621003, 81761128023, 81220108013, 81227002 and 81030027 to Q.G.; grant no. 81671664 and 81371527 to S.L.; grant no. 81401396 to F.L.), and Program for Changjiang Scholars and Innovative Research Team in University (PCSIRT, grant no. IRT16R52) of China. Q.G. would also like to acknowledge the support from his Changjiang Scholar Professorship Award (award no. T2014190) of China and American CMB Distinguished Professorship Award (award no. F510000/ G16916411) administered by the Institute of International Education, United States. S.L. would also like to acknowledge the National Youth Top-notch Talent Support Program of China (W02070140) and Chang Jiang Scholars Program (Young Scholars, Q2015154) of China. Acknowledgments J.A.S. has consulted to Takeda. The other authors report no financial relationships with commercial interests. References 1. Owen MJ Sawa A Mortensen PB. Schizophrenia. Lancet . 2016; 388: 86– 97. Google Scholar CrossRef Search ADS PubMed 2. Pettersson-Yeo W Allen P Benetti S McGuire P Mechelli A. Dysconnectivity in schizophrenia: where are we now? Neurosci Biobehav Rev . 2011; 35: 1110– 1124. Google Scholar CrossRef Search ADS PubMed 3. Gong Q Lui S Sweeney JA. A selective review of cerebral abnormalities in patients with first-episode schizophrenia before and after treatment. Am J Psychiatry . 2015; 173: 232– 243. Google Scholar CrossRef Search ADS PubMed 4. Lui S Zhou XJ Sweeney JA Gong Q. Psychoradiology: the frontier of neuroimaging in psychiatry. Radiology . 2016; 281: 357– 372. Google Scholar CrossRef Search ADS PubMed 5. Kuswanto CN Teh I Lee TS Sim K. Diffusion tensor imaging findings of white matter changes in first episode schizophrenia: a systematic review. Clin Psychopharmacol Neurosci . 2012; 10: 13– 24. Google Scholar CrossRef Search ADS PubMed 6. Domen P Peeters S Michielse Set al. . Differential time course of microstructural white matter in patients with psychotic disorder and individuals at risk: a 3-year follow-up study. Schizophr Bull . 2016; 43: 160– 170. Google Scholar CrossRef Search ADS PubMed 7. Zhang XY Fan FM Chen DCet al. . Extensive white matter abnormalities and clinical symptoms in drug-naive patients with first-episode schizophrenia: a voxel-based diffusion tensor imaging study. J Clin Psychiatry . 2016; 77: 205– 211. Google Scholar CrossRef Search ADS PubMed 8. Asami T Hyuk Lee S Bouix Set al. . Cerebral white matter abnormalities and their associations with negative but not positive symptoms of schizophrenia. Psychiatry Res . 2014; 222: 52– 59. Google Scholar CrossRef Search ADS PubMed 9. Lu LH Zhou XJ Keedy SK Reilly JL Sweeney JA. White matter microstructure in untreated first episode bipolar disorder with psychosis: comparison with schizophrenia. Bipolar Disord . 2011; 13: 604– 613. Google Scholar CrossRef Search ADS PubMed 10. Hagmann P Cammoun L Gigandet Xet al. . Mapping the structural core of human cerebral cortex. PLoS Biol . 2008; 6: e159. Google Scholar CrossRef Search ADS PubMed 11. Fornito A Zalesky A Pantelis C Bullmore ET. Schizophrenia, neuroimaging and connectomics. Neuroimage . 2012; 62: 2296– 2314. Google Scholar CrossRef Search ADS PubMed 12. Sporns O. Network attributes for segregation and integration in the human brain. Curr Opin Neurobiol . 2013; 23: 162– 171. Google Scholar CrossRef Search ADS PubMed 13. Hadley JA Kraguljac NV White DM Ver Hoef L Tabora J Lahti AC. Change in brain network topology as a function of treatment response in schizophrenia: a longitudinal resting-state fMRI study using graph theory. NPJ Schizophr . 2016; 2: 16014. Google Scholar CrossRef Search ADS PubMed 14. Liu Y Liang M Zhou Yet al. . Disrupted small-world networks in schizophrenia. Brain . 2008; 131: 945– 961. Google Scholar CrossRef Search ADS PubMed 15. Lynall ME Bassett DS Kerwin Ret al. . Functional connectivity and brain networks in schizophrenia. J Neurosci . 2010; 30: 9477– 9487. Google Scholar CrossRef Search ADS PubMed 16. Zalesky A Fornito A Seal MLet al. . Disrupted axonal fiber connectivity in schizophrenia. Biol Psychiatry . 2011; 69: 80– 89. Google Scholar CrossRef Search ADS PubMed 17. Wang Q Su TP Zhou Yet al. . Anatomical insights into disrupted small-world networks in schizophrenia. Neuroimage . 2012; 59: 1085– 1093. Google Scholar CrossRef Search ADS PubMed 18. Alexander-Bloch AF Gogtay N Meunier Det al. . Disrupted modularity and local connectivity of brain functional networks in childhood-onset schizophrenia. Front Syst Neurosci . 2010; 4: 147. Google Scholar CrossRef Search ADS PubMed 19. Singh SP Cooper JE Fisher HLet al. . Determining the chronology and components of psychosis onset: the Nottingham Onset Schedule (NOS). Schizophr Res . 2005; 80: 117– 130. Google Scholar CrossRef Search ADS PubMed 20. Cui Z Zhong S Xu P He Y Gong G. PANDA: a pipeline toolbox for analyzing brain diffusion images. Front Hum Neurosci . 2013; 7: 42. Google Scholar PubMed 21. Mori S Kaufmann WE Davatzikos Cet al. . Imaging cortical association tracts in the human brain using diffusion-tensor-based axonal tracking. Magn Reson Med . 2002; 47: 215– 223. Google Scholar CrossRef Search ADS PubMed 22. Mori S Crain BJ Chacko VP van Zijl PC. Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol . 1999; 45: 265– 269. Google Scholar CrossRef Search ADS PubMed 23. Tzourio-Mazoyer N Landeau B Papathanassiou Det al. . Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage . 2002; 15: 273– 289. Google Scholar CrossRef Search ADS PubMed 24. Zhang Z Liao W Chen Het al. . Altered functional-structural coupling of large-scale brain networks in idiopathic generalized epilepsy. Brain . 2011; 134: 2912– 2928. Google Scholar CrossRef Search ADS PubMed 25. Zhang J Wang J Wu Qet al. . Disrupted brain connectivity networks in drug-naive, first-episode major depressive disorder. Biol Psychiatry . 2011; 70: 334– 342. Google Scholar CrossRef Search ADS PubMed 26. Liao W Zhang Z Mantini Det al. . Relationship between large-scale functional and structural covariance networks in idiopathic generalized epilepsy. Brain Connect . 2013; 3: 240– 254. Google Scholar CrossRef Search ADS PubMed 27. He Y Chen ZJ Evans AC. Small-world anatomical networks in the human brain revealed by cortical thickness from MRI. Cereb Cortex . 2007; 17: 2407– 2419. Google Scholar CrossRef Search ADS PubMed 28. Rubinov M Sporns O. Complex network measures of brain connectivity: uses and interpretations. Neuroimage . 2010; 52: 1059– 1069. Google Scholar CrossRef Search ADS PubMed 29. Newman ME. Modularity and community structure in networks. Proc Natl Acad Sci U S A . 2006; 103: 8577– 8582. Google Scholar CrossRef Search ADS PubMed 30. Newman ME. Finding community structure in networks using the eigenvectors of matrices. Phys Rev E Stat Nonlin Soft Matter Phys . 2006; 74: 036104. Google Scholar CrossRef Search ADS PubMed 31. Radicchi F Castellano C Cecconi F Loreto V Parisi D. Defining and identifying communities in networks. Proc Natl Acad Sci U S A . 2004; 101: 2658– 2663. Google Scholar CrossRef Search ADS PubMed 32. van den Heuvel MP Mandl RC Stam CJ Kahn RS Hulshoff Pol HE. Aberrant frontal and temporal complex network structure in schizophrenia: a graph theoretical analysis. J Neurosci . 2010; 30: 15915– 15926. Google Scholar CrossRef Search ADS PubMed 33. Fornito A Yoon J Zalesky A Bullmore ET Carter CS. General and specific functional connectivity disturbances in first-episode schizophrenia during cognitive control performance. Biol Psychiatry . 2011; 70: 64– 72. Google Scholar CrossRef Search ADS PubMed 34. Meunier D Achard S Morcom A Bullmore E. Age-related changes in modular organization of human brain functional networks. Neuroimage . 2009; 44: 715– 723. Google Scholar CrossRef Search ADS PubMed 35. Bohlken MM Brouwer RM Mandl RCet al. . Structural brain connectivity as a genetic marker for schizophrenia. JAMA Psychiatry . 2016; 73: 11– 19. Google Scholar CrossRef Search ADS PubMed 36. Yao L Lui S Liao Yet al. . White matter deficits in first episode schizophrenia: an activation likelihood estimation meta-analysis. Prog Neuropsychopharmacol Biol Psychiatry . 2013; 45: 100– 106. Google Scholar CrossRef Search ADS PubMed 37. Beaulieu C. The basis of anisotropic water diffusion in the nervous system - a technical review. NMR Biomed . 2002; 15: 435– 455. Google Scholar CrossRef Search ADS PubMed 38. Walterfang M Wood SJ Velakoulis D Pantelis C. Neuropathological, neurogenetic and neuroimaging evidence for white matter pathology in schizophrenia. Neurosci Biobehav Rev . 2006; 30: 918– 948. Google Scholar CrossRef Search ADS PubMed 39. Szeszko PR Ardekani BA Ashtari Met al. . White matter abnormalities in first-episode schizophrenia or schizoaffective disorder: a diffusion tensor imaging study. Am J Psychiatry . 2005; 162: 602– 605. Google Scholar CrossRef Search ADS PubMed 40. Hao Y Yan Q Liu Het al. . Schizophrenia patients and their healthy siblings share disruption of white matter integrity in the left prefrontal cortex and the hippocampus but not the anterior cingulate cortex. Schizophr Res . 2009; 114: 128– 135. Google Scholar CrossRef Search ADS PubMed 41. Ribolsi M Daskalakis ZJ Siracusano A Koch G. Abnormal asymmetry of brain connectivity in schizophrenia. Front Hum Neurosci . 2014; 8: 1010. Google Scholar CrossRef Search ADS PubMed 42. Vercammen A Knegtering H den Boer JA Liemburg EJ Aleman A. Auditory hallucinations in schizophrenia are associated with reduced functional connectivity of the temporo-parietal area. Biol Psychiatry . 2010; 67: 912– 918. Google Scholar CrossRef Search ADS PubMed 43. Li M Deng W He Zet al. . A splitting brain: imbalanced neural networks in schizophrenia. Psychiatry Res . 2015; 232: 145– 153. Google Scholar CrossRef Search ADS PubMed 44. Fornito A Bullmore ET. Reconciling abnormalities of brain network structure and function in schizophrenia. Curr Opin Neurobiol . 2015; 30: 44– 50. Google Scholar CrossRef Search ADS PubMed 45. Wang J Mathalon DH Roach BJet al. . Action planning and predictive coding when speaking. Neuroimage . 2014; 91: 91– 98. Google Scholar CrossRef Search ADS PubMed 46. Ford JM Mathalon DH Roach BJet al. . Neurophysiological evidence of corollary discharge function during vocalization in psychotic patients and their nonpsychotic first-degree relatives. Schizophr Bull . 2012; 39: 1272– 1280. Google Scholar CrossRef Search ADS PubMed 47. Guimerà R Nunes Amaral LA. Functional cartography of complex metabolic networks. Nature . 2005; 433: 895– 900. Google Scholar CrossRef Search ADS PubMed 48. Reilly JL Harris MS Keshavan MS Sweeney JA. Adverse effects of risperidone on spatial working memory in first-episode schizophrenia. Arch Gen Psychiatry . 2006; 63: 1189– 1197. Google Scholar CrossRef Search ADS PubMed 49. Zhang R Wei Q Kang Zet al. . Disrupted brain anatomical connectivity in medication-naïve patients with first-episode schizophrenia. Brain Struct Funct . 2015; 220: 1145– 1159. Google Scholar CrossRef Search ADS PubMed 50. Ren W Lui S Deng Wet al. . Anatomical and functional brain abnormalities in drug-naive first-episode schizophrenia. Am J Psychiatry . 2013; 170: 1308– 1316. Google Scholar CrossRef Search ADS PubMed 51. Lui S Zhou XJ Sweeney J Gong Q. Psychoradiology: the frontier of neuroimaging in psychiatry. Radiology . 2016;281:357–372. 52. Kressel HY. Setting sail: 2017. Radiology . 2017;282:4–6. © The Author(s) 2017. Published by Oxford University Press on behalf of the Maryland Psychiatric Research Center. All rights reserved. For permissions, please email: email@example.com
Schizophrenia Bulletin – Oxford University Press
Published: Mar 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera