- Email: [email protected]
Network modeling of resting state connectivity points towards the bottom up theories of schizophrenia
Psychiatry Research: Neuroimaging 266 (2017) 19–26
Contents lists available at ScienceDirect
Psychiatry Research: Neuroimaging journal homepage: www.elsevier.com/locate/psychresns
Network modeling of resting state connectivity points towards the bottom up theories of schizophrenia
MARK
François Orliacb, Pascal Delamillieurea,b, Nicolas Delcroixc, Mikael Naveaud, Perrine Brazoa,b, ⁎ Annick Razafimandimbyb, Sonia Dollfusa,b, Marc Joliote,f,g, a
CHU de Caen, Department of Psychiatry, Caen F-14000, France Université de Caen Basse-Normandie, UFR de Médecine, UMR 6301 ISTCT, ISTS group, Caen F-14000, France c GIP CYCERON, UMS 3408, Caen F-14000, France d INSERM UMR-S U919 SP2U, Université Caen Basse-Normandie, Caen F-14000, France e GIN, University of Bordeaux, Institut des Maladies Neurodégénératives, UMR 5293, Bordeaux F-33000, France f GIN, CNRS, Institut des Maladies Neurodégénératives, UMR 5293, Bordeaux F-33000, France g GIN, CEA, Institut des Maladies Neurodégénératives, UMR 5293, Bordeaux F-33000, France b
A R T I C L E I N F O
A B S T R A C T
Keywords: Functional magnetic resonance imaging Hallucinations Neural networks
The dysconnectivity theory of schizophrenia proposes that schizophrenia symptoms arise from abnormalities in neuronal synchrony. Resting-state Functional Connectivity (FC) techniques allow us to highlight synchronization of large-scale networks, the Resting-state Networks (RNs). A large body of work suggests that disruption of RN synchronization could give rise to specific schizophrenia symptoms. The present study aimed to explore withinand between-network FC strength of 34 RNs in 29 patients suffering from schizophrenia, and their relationships with schizophrenia symptoms. Resting-state data were analyzed using independent component analysis and dual-regression techniques. Our results showed that both within-RN and between-RN FC were disrupted in patients with schizophrenia, with a global trend toward weaker FC. This decrease affected more particularly visual, auditory and crossmodal binding networks. These alterations were correlated with negative symptoms, positive symptoms and hallucinations, indicating abnormalities in visual processing and crossmodal binding in schizophrenia. Moreover, we stressed an anomalous synchronization between a visual network and a network thought to be engaged in mental imaging processes, correlated with delusions and hallucinations. Altogether, our results supported the assumption that some schizophrenia symptoms may be related to low-order sensory alterations impacting higher-order cognitive processes, i.e. the “bottom-up” hypothesis of schizophrenia symptoms.
1. Introduction The dysconnectivity theory of schizophrenia proposes that many symptoms may be related to a failure to integrate the activity of local and distributed neural circuits (van den Heuvel and Fornito, 2014). One way of assessing brain connectivity is to study how multiple brain regions functionally interact while an individual is not engaged in a specific task, i.e. using resting-state blood oxygen level-dependent Functional Connectivity (FC) (Rogers et al., 2007).
Numerous FC studies conducted on healthy volunteers reported that the brain exhibited a structured neural activity during the resting state (Bressler and Menon, 2010; Doucet et al., 2011). This can be observed at different scales: the system scale, the module scale (Fig. S1 Supplementary data) and the Resting-state Network (RN) scale (Fig. S2 Supplementary data). Among these scales, the partition into RNs appears to be the most relevant in the context of schizophrenia research. RNs have been suggested to overlap with the networks subtending the brain in action (Smith et al., 2009). In other words,
Abbreviations: AH, Auditory Hallucinations in medical history; ANCOVA, Analysis Of COVAriance; ANOVA, ANalysis Of Variance; AVH, Auditory and Visual Hallucinations in medical history; BNFC, Between-Network Functional Connectivity; C-FD, Cumulated Framewise Displacement; Cpz-eq, average Chlorpromazine equivalent; DMN, Default-Mode Network; FC, Functional Connectivity; FD, Framewise Displacement; HC, Healthy Controls; ICA, Independent Component Analysis; MANOVA, Multivariate ANalysis Of Variance; MICCA, Multi-scale Individual Component Clustering Analysis; MINI, Mini International Neuropsychiatric Interview; MRI, Magnetic Resonance Imaging; NBS, Network-Based Statistics; NH, No Hallucinations in medical history; PANSS, Positive And Negative Syndrome Scale; PANSS-N, Negative symptoms subscale of the PANSS; PANSS-P, Positive symptoms subscale of the PANSS; RN, Resting-state Network; PS, Patients with Schizophrenia; WNFC, Within-Network Functional Connectivity ⁎ Corresponding author at: GIN, UMR5293, Université de Bordeaux, 146, rue Léo Saignat, CS 61292, 33076 Bordeaux Cedex, France. E-mail address: [email protected] (M. Joliot). http://dx.doi.org/10.1016/j.pscychresns.2017.04.003 Received 22 October 2016; Received in revised form 15 March 2017; Accepted 7 April 2017 Available online 12 April 2017 0925-4927/ © 2017 Published by Elsevier Ireland Ltd.
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
functional networks seem to be continuously and dynamically “active” even when the brain is “at rest”. RNs have already been shown to be of potential clinical value as rich and sensitive markers of disease (Menon, 2011). They offer several advantages: the full repertoire of RNs can be tested in a single scanning session without having to decide a priori what functional paradigm is most likely to be useful, and minimal individual participation is required during scanning. This latter point can be critical in clinical settings (e.g. Autism Spectrum Disorders, severe schizophrenia or Alzheimer Disease). As a consequence, the study of RNs allows exploring cognitive processes that are either understudied or examined under simplifying assumptions due to scanning constraints (Papo, 2013; Smith et al., 2009). Moreover, disruptions of RNs may contribute to specific patterns of cognitive and behavioral impairments, providing new insights into aberrant brain organization in several psychiatric and neurological disorders (Menon, 2011; Papo, 2013). Several authors reported within-RN connectivity disruptions in schizophrenia during the resting state. Almost all of these studies were hypothesis-driven, hence the authors selected a few RNs on the basis of their relevance regarding the physiopathology of schizophrenia. As schizophrenia is usually seen to be related to higher-order cognitive dysfunctions, higher order RNs were chosen: DMN, central executive, attentional, language or salience networks (Orliac et al., 2013; Rotarska-Jagiela et al., 2010; Wolf et al., 2011; Woodward et al., 2011). A few studies examined between-RN connectivity in schizophrenia during the resting state (Arbabshirani et al., 2013; Jafri et al., 2008; Khadka et al., 2013; Mamah et al., 2013; Meda et al., 2012; Yu et al., 2012). It should be noted that, apart from Yu et al. (2012) who studied 57 RNs, these studies have focused on less than 20 RNs. All of these reported reduced between-RN connectivity in patients with schizophrenia, except Jafri et al. (2008) and Khadka et al. (2013) who reported both connectivity increases and decreases. In the light of this literature, and by contrast with our previous hypothesis-driven work (Orliac et al., 2013), it seemed relevant to explore RN functional connectivity in schizophrenia as exhaustively as possible. To achieve this, we studied both within-RN and between-RN functional connectivity strength of 34 networks encompassing 98% of the cerebral gray matter, including low-order and high-order networks, and their relationships with schizophrenia symptoms.
in medical history (AH group), or Auditory and Visual Hallucinations in medical history (AVH group). The local ethics committee (CPP de Basse-Normandie, France) approved the study. All participants gave written informed consent. 2.2. Image acquisition Data acquisition was performed on a 3T Philips Achieva MRI scanner. Structural data were acquired using a high-resolution, threedimensional T1-weighted volume (repetition time (TR)=20 ms; echo time (TE)=4.6 ms; flip angle=10°; inversion time=800 ms; turbo field echo factor=65; sense factor=2; field of view=256×256×180 mm; 1×1×1 mm3 isotropic voxel size), and a T2*-weighted, multi-slice acquisition (T2*-weighted fast-field echo; TR=3500 ms; TE=35 ms; flip angle=90°; sense factor=2; 70 axial slices; 2×2×2 mm3 isotropic voxel size). Spontaneous brain activity was monitored using BOLD fMRI while the participants performed a resting-state condition for 8 min (T2*-echo planar imaging; 240 volumes; TR=2 s; TE =35 ms; flip angle=80°; 31 axial slices; 3.75×3.75×3.75 mm3 isotropic voxel size). Immediately before fMRI scanning, participants were instructed to “keep their eyes closed, to relax, to refrain from moving, to stay awake, and to let their thoughts come and go.” 2.3. Image processing Pre-processing of the functional data was based on the methods described in Naveau et al. (2012). Briefly, it included slice-timing correction, motion correction, co-registration to structural scan, spatial normalization to the MNI template and spatial smoothing (6 mm Gaussian kernel). Each individual's structural scan was segmented into gray matter, white matter, and cerebrospinal fluid using the unified segmentation approach (Statistical Parametric Mapping 5; www.fil.ion. ucl.ac.uk/spm5). Time series for white matter, cerebrospinal fluid, and the six motion parameters were regressed out of the data. Finally, fMRI data were temporally filtered using band-pass filtering (0.01 Hz < f < 0.1 Hz). 2.4. Motion control Head motion correction has become a prominent concern in the field of resting state FC, especially during clinical studies. As a matter of facts, several authors have reported that small amounts of movement can produce substantial changes in the timecourses of resting state data, and cause spurious but spatially structured patterns in FC (Power et al., 2012; Satterthwaite et al., 2012). As a consequence, an index of quality control was computed for each participant: the Framewise Displacement (FD) of head position, calculated as the sum of the absolute values of the 6 translational and rotational realignment parameters (Power et al., 2012). Every participant exhibiting FD higher than 0.5 mm in more than 10% of the 240 volumes was excluded from the study. Then, a single motion index was calculated for each participant: the Cumulated Framewise Displacement (C-FD), defined as the sum of FDs over all 240 images. C-FD values were tested for group differences and added as covariate of no interest in our connectivity analyses (see 2.9.) to control for this potentially confounding factor.
2. Methods 2.1. Participants Twenty-nine Patients with Schizophrenia (PS group) attending the Department of Psychiatry of Caen University Hospital and twenty-nine matched healthy controls (HC group) were included in the study. We used a sample expanded from a previous study (Orliac et al., 2013). PS and HC groups were matched for age, sex, handedness, and educational level on a one-to-one basis. All participants were required to be between 18 and 60 years of age. All were screened for magnetic resonance imaging (MRI) contraindications, and participants with a history of a major medical condition, neurological disease, or substance abuse were excluded from the study. PS group participants were diagnosed by an experienced clinician using the Mini International Neuropsychiatric Interview (MINIplus v.4.5). They were required to have been stable on antipsychotic medication for at least four months prior to the study. The Positive and Negative Syndrome Scale (PANSS) (Kay et al., 1987) was used to assess positive (PANSS-P) and negative (PANSS-N) symptoms. Daily antipsychotic medication dosage at the time of inclusion was recorded and converted into chlorpromazine equivalents (mg/d). Our findings pointed towards a need for post-hoc analyses. For this purpose, PS patients were classified into three subgroups according to their hallucinatory status (items A6b and A7b of the MINIplus): No Hallucinations in medical history (NH group), Auditory Hallucinations
2.5. Reference maps of the RNs Reference maps for RNs were estimated from BIL & GIN cohort (Mazoyer et al., 2016) resting-state datasets (n=282), using a group ICA approach based on multi-scale individual component clustering (MICCA, see Naveau et al. (2012)). The MICCA analysis retained 34 non-artifactual RNs. The positive map of each RN was thresholded using a mixture model (p > 0.95) (Beckmann and Smith, 2004). Additionally, voxels exhibiting a probability of being included in the 20
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
Table 1 Population demographics and clinical characteristics. Group
p value for Group comparison
N
PS 28
HC 29
Sex (F/M) Age (years; mean ± SD) Handedness (left/right) Education (years; mean ± SD)
(7/21) 36.1 ± 8 (4/24) 12 ± 1.6
(7/22) 34.5 ± 8.9 (4/25) 12.8 ± 2.4
0.94 0.75 0.96 0.08
Illness duration (years; mean ± SD) Medication (Cpz-eq; mg/d; mean ± SD) PANSS-P (mean ± SD) PANSS-N (mean ± SD) Hallucinatory status (NH/AH/AVH)
12.3 ± 7.9 339.6 ± 255.7
n/a n/a
n/a n/a
12.9 ± 5 12.8 ± 4.3 11 / 8 / 9
n/a n/a n/a
n/a n/a n/a
AH: Auditory Hallucinations in medical history; AVH: Auditory and Visual Hallucinations in medical history; Cpz-eq: average Chlorpromazine equivalent; HC: Healthy Controls; n/a: not applicable; NH: No Hallucinations in medical history; PANSS-N: Negative subscale of the Positive and Negative Syndrome Scale; PANSS-P: Positive subscale of the Positive and Negative Syndrome Scale; PS: Patients with Schizophrenia group.
SPM gray matter canonical map below 0.3 were filtered out. The resultant spatial map of each RN is displayed in Fig. S2 Supplementary data. The numbering of the RNs refers to the hierarchical functional clustering of RNs proposed by Doucet et al. (2011), see Fig. S1.
et al. (2010) Network-Based Statistics (NBS), a new approach designed to identify significant differences between graphs. Briefly, NBS is intended to check for the family-wise error rate using the principles of cluster-based thresholding of statistical parametric maps. We used the NBS toolbox with default settings (T=3.1, p < 0.05 NBS, 5000 permutations), and C-FD was added as covariate of no interest in the model. Second, ANCOVA was conducted to test for a group difference on each BNFC value with the more stringent Bonferroni-Holm correction, still controlling for C-FD (df1=1/df2=53, p < 0.05 BonferroniHolm corrected, n=561).
2.6. Dual regression The dual regression technique (Filippini et al., 2009) was used to build for each individual the level spatial maps corresponding to the 34 RNs, with associated time series. We used the dedicated FSL tool (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/DualRegression). Each voxel value of these maps was a Z-normalized regression beta coefficient, interpreted as a measure of the strength of the association between the voxel time series and the whole network time series, i.e., functional connectivity within this network.
2.9.3. Correlation with clinical variables When a significant group difference was highlighted for a FC variable, a two-tailed non-parametric Spearman test was used to assess the correlation between PANSS subscales and the FC values of this variable within the PS group, controlling for C-FD. Post-hoc analyses were conducted regarding the hallucinatory status: when a correlation was highlighted between an FC variable and PANSS-P score, ANCOVA was conducted to test for an effect of the hallucinatory status on FC, controlling for C-FD (df1:2/df2:22, p < 0.05). When statistically significant, ANCOVA was followed by a two-tailed Tukey's Honest Significant Difference test (HSD; p < 0.05). To check for a possible confounding effect of antipsychotic medication on FC, complementary regression analyses were carried out adding medication dosage as a covariate.
2.7. Within-network functional connectivity For each RN and each individual, the Within-Network Functional Connectivity (WNFC) strength was computed. The WNFC score was defined as the mean Z-score of the voxels located within the corresponding reference RN mask. 2.8. Between-network functional connectivity FC between each possible pair of RNs, annotated Between-Network Functional Connectivity (BNFC), was computed using Pearson's correlation, resulting in a 34×34 connectivity matrix for each participant. This connectivity matrix allowed us to build a weighted connectivity graph for each participant, composed of 34 nodes (representing the RNs) and 561 edges (whose weights represented Pearson's r between two time series, i.e. FC between two RNs). Correlation coefficients were transformed using Fisher's r to z.
3. Results 3.1. Motion control One participant (belonging to the PS group) was excluded from the study due to significant head motion. There was no difference between the two groups of remaining participants regarding C-FD (two-tailed Ttest, T(55)=0.41, p=0.68).
2.9. Statistical analysis 2.9.1. WNFC WNFC values were tested for group difference. First, we tested for a group effect using a Multivariate ANalysis Of VAriance (MANOVA, p < 0.05) including all 34 RNs as dependent variables. Then, ANalysis of COVAriance (ANCOVA) was conducted to test for a group difference on each RN considered individually, controlling for C-FD (df1=1/ df2=53, p < 0.05 Bonferroni-Holm corrected, n=34).
3.2. Clinical data Table 1 provides detailed demographic and clinical data for the remaining participants. None of the matching criteria were statistically different between the groups. With regard more specifically to the PS group, there was no difference between the three “Hallucinatory status” subgroups concerning the PANSS-P score (Analysis Of VAriance, F(2,25)=1.33, p=0.28), the PANSS-N score (ANOVA, F(2,25)=0.64, p=0.54), or medication dosage (ANOVA, F(2,25)=1.14, p=0.33).
2.9.2. BNFC BNFC values were tested for group difference. First, we used Zalesky 21
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
Fig. 1. Between network connectivity is altered in schizophrenia. Connectivity graph is displayed using the hierarchical functional model from Doucet et al. (2011). Gray edges represent the most significant edges in our control group (r > 0.4). Bonf: Bonferroni-Holm correction; HC: Healthy Controls; NBS: Network Based Statistics; RN: Resting-state Network; PS: Patients with Schizophrenia.
3.3. Group comparisons
3.4. Correlation with clinical variables
3.3.1. WNFC MANOVA showed weaker WNFC (Wilks' λ=0.23, χ2(34) =55.7, p=0.01) in the RNs of the PS group. Considered individually, 5 RNs showed significantly reduced WNFC in PS (Fig. S3 Supplementary data): a right parieto-frontal RN (RN10, p < 0.001), a temporo-insular RN (RN26, p < 0.001) and 3 occipital RNs: RN31 (p < 0.001); RN32 (p=0.001); and RN33 (p < 0.001).
3.4.1. WNFC WNFC of RN32 (occipital RN) was negatively correlated with the PANSS-P score in the PS group (rs=−0.43, p=0.02), controlling for CFD. There was no effect of the hallucinatory status on this connectivity variable. This correlation remained significant (rs=−0.41, p=0.03) after adding medication dosage as a covariate. 3.4.2. BNFC Controlling for C-FD, BNFC was negatively correlated with the PANSS-P score in four edges (Fig. 2A): RN24-RN33 (rs=−0.44, p=0.02), RN25-RN33 (rs=−0.43, p=0.02), RN26-RN32 (rs=−0.43, p=0.03), and RN26-RN34 (rs=−0.50, p=0.008). ANCOVA revealed a trend toward an effect of hallucinatory status on the BNFC of RN26RN34, controlling for C-FD (F(2,22)=3.23, p=0.06). In this edge, posthoc Tukey's HSD highlighted that BNFC was significantly lower in AVH patients compared to NH patients. Controlling for C-FD, BNFC was positively correlated (Fig. 2B) with the PANSS-P score in RN6-RN31, an edge linking an occipital RN to the posterior cingulate/precuneus RN (rs=0.42, p=0.03). ANCOVA showed an effect of the hallucinatory status on BNFC (F(2,22)=5.29, p=0.02), and post-hoc Tukey's HSD highlighted that BNFC was significantly higher in AVH patients compared to both NH and AH patients. Correlations between BNFC values and PANSS-P score were no longer significant after adding medication dosage as a covariate. Posthoc analyses were carried out to disentangle the interaction between BNFC, PANSS-P score and medication dosage. First, it appeared that PANSS-P score was strongly correlated with medication dosage
3.3.2. BNFC The NBS analysis uncovered 19 edges with between-group BNFC differences (Fig. 1). Fifteen edges showed reduced BNFC in the PS group, and all but one (RN26-RN28) involved an occipital RN. Most of these (n=13) were positive edges in HC that exhibited lower positive (n=8) or negative (n=5) values in the PS group. The two remaining edges, both involving RN18 (a parieto-frontal RN), were negative edges exhibiting lower negative values in the PS group. This difference remained significant using Bonferroni-Holm correction for four edges. All of these were between an occipital RN (RN31 or RN33) and a temporal RN (RN24, RN25 or RN26). Four edges showed a BNFC increase in the PS group, and all involved RN6, which is an RN encompassing mainly precuneus and posterior cingulate. Three of these are negative edges in HC that become closer to zero in PS. The remaining edge (RN6-RN31) is positive, exhibiting higher values in the PS group. None of these differences remained significant using Bonferroni-Holm correction.
22
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
Fig. 2. Correlation between edge functional connectivity, PANSS subscales and hallucinatory status. Blue edge: connectivity decrease related to greater symptomatology; Red edge: connectivity increase related to greater symptomatology. (A) negative correlations with the positive subscale. (B) positive correlations with the positive subscale. (C) negative correlations with the negative subscale. AH: Auditory Hallucinations; AVH: Auditory and Visual Hallucinations; BNFC: Between Network Functional Connectivity; NH: No Hallucinations; PANSS-N: Negative subscale of the Positive and Negative Syndrome Scale; PANSS-P: Positive subscale of the Positive and Negative Syndrome Scale; RN: Resting-state Network (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).
cant after adding medication dosage as a covariate for both RN26-RN28 (rs=−0.39, p=0.04) and RN26-RN31 (rs=−0.43, p=0.02).
(rs=0.56, p=0.002). Then, among all BNFC values exhibiting group differences, two edges were strongly correlated with medication dosage: RN26-RN32 (rs=−0.60, p < 0.001), and RN26-RN34 (rs=−0.65, p < 0.001). The effect of the hallucinatory status on BNFC remained significant after adding medication dosage as a covariate, for the two edges: RN26RN34 (F(2,22)=3.46, p=0.05) and RN6-RN31 (F(2,22)=5.35, p=0.01). Controlling for C-FD, BNFC was negatively correlated (Fig. 2C) with the PANSS-N score in two edges involving RN26, a temporo-insular RN: RN26-RN28 (rs=−0.39, p=0.04) and RN26-RN31 (rs=−0.46, p=0.01). Correlations between BNFC and PANSS-N score remained signifi-
4. Discussion Our results show that both within-RN and between-RN FC are disrupted in schizophrenia, with a global trend toward weaker FC. Within-RN connectivity (WNFC) was reduced in all the RNs, and mainly in visual networks. Between-RN connectivity (BNFC) analyses revealed that almost all the edges showing weaker connectivity in patients with schizophrenia involved an occipital visual network. Moreover, the most significant differences were found in between two 23
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
Although our results showed a trend toward weaker FC in patients with schizophrenia, between-network connectivity was increased in four edges in this group. All these edges were between RN6 (precuneus and posterior cingulate), on one hand, and an RN belonging to the extrinsic system, on the other. Increased connectivity resulted in diminished anticorrelation between RN6, whose regions are considered to be major components of the intrinsic system (Golland et al., 2008), and three sensory-motor networks (RN20, RN21 and RN22) belonging to the extrinsic system. Diminished anticorrelation between these two systems has already been reported in schizophrenia patients (Abbott et al., 2010; Hasenkamp et al., 2011; Whitfield-Gabrieli et al., 2009). According to Wotruba et al. (2014), diminished anticorrelation between the intrinsic and extrinsic system may lead to confusion between internal information and external stimuli processing. In our study, the FC increase between RN6 and an associative visual network (RN31) was correlated with positive symptoms, and this increase was significantly more pronounced in the AVH group. The precuneus, and more specifically its dorsal regions, seems to be engaged in visual mental imagery processes (Mellet et al., 2000; Zhang and Li, 2012). Thus, we suggest that this aberrant coupling may lead to confusion between the processing of internal visual mental imagery (RN6) and the processing of external visual stimuli (RN31), resulting in delusions and hallucinations. In our study, no patient exhibited visual hallucinations only (i.e. without auditory hallucinations), which prevents a firm conclusion about a specific relationship between visual hallucinations and the FC of RN6-RN31. However, two arguments point in that direction: FC of RN6-RN31 was significantly higher in the AVH group compared to the AH group; and the putative role of RN6 and RN31 itself points towards a dysfunction specifically affecting the visual modality. Interestingly, Jardri et al. (2013) reported increased blood oxygen level-dependent signal in associative visual regions resembling RN31 during visual hallucinations in adolescents experiencing a brief psychotic disorder. In this study temporal and spatial instability of the DMN was correlated with the severity of hallucinations. Moreover the authors reported a negative correlation between associative visual regions and the DMN time course during visual hallucinations. This apparent discrepancy with our results may be surprising, as RN6 is usually regarded as a DMN subcomponent. It should be noted that, in our study, patients in the AVH group have experienced visual hallucination in their medical history, but they were not necessarily hallucinating at the time of scanning. This suggests a different connectivity pattern between DMNrelated regions and visual associative regions, depending on the occurrence of visual hallucinations: anticorrelation during the hallucinatory experience (state marker), and diminished anticorrelation between the hallucination periods (trait marker). This is in line with a recent study exploring the networks dynamics during the different stages of hallucinations in PS (Lefebvre et al., 2016), which reported such a sequential pattern between the DMN and hallucination-related components. This study had several limitations. First, the significance of our results must be discussed with caution. RNs were elicited during rest, and the determination of their putative role relies on their close correspondence with the networks elicited by task-based studies (Smith et al., 2009). Another limitation of this study was the fact that the analyses regarding the hallucinatory status were conducted using a non-specific global severity score (MINIplus), and that the occurrence of hallucination wasn’t assessed during scanning. As a matter of fact, the present study wasn’t initially designed to specifically investigate the pathophysiology of hallucinations in PS, but when discovering our preliminary results we were struck by the fact that sensory networks appeared to be the more affected. This pointed towards a need for posthoc analyses probing the possible relationship between the alteration of these RN and the occurrence of auditory and/or visual hallucinations. The only clinical data available distinguishing between auditory and verbal hallucinations was from MINIplus, which reports the occurrence of hallucinations in medical history, not only at the time of scanning.
medial occipital networks (RN31 and RN33) and three closely linked sensory networks (RN24, RN25 and RN26). RN24 is a low-level auditory network. RN 25 and RN26 encompass several associative regions from different sensory modalities: auditory (Heschl's gyrus for RN25), visual (cuneus for RN25 and lingual gyrus for RN26), auditoryvisual integration (bilateral superior temporal sulcus for RN26) and somato-sensory integration (bilateral insula for RN25 and RN26). On this basis, we assume that FC of these two RNs may be related to how the brain binds signals from multiple sensory modalities together to produce unified percepts of objects and events, i.e. crossmodal binding (Bushara et al., 2003; Senkowski et al., 2008). Altogether, our results suggest that FC disruption in schizophrenia affects visual, auditory and crossmodal binding networks more particularly. These results are in line with two studies (Arbabshirani et al., 2013; Meda et al., 2012) reporting between-RN connectivity disruption involving sensory-motor and visual RNs in schizophrenia. Interestingly, Su et al. (2016), in a recent study using dynamic functional connectivity techniques, reported shared transient dysconnectivity states within visual pathways between schizophrenia patients and their unaffected siblings. Structural connectivity of the visual cortex also seems to be disrupted in schizophrenia. As a matter of facts, Palaniyappan et al. (2013) reported a dysmyelination in visual processing regions, which was most pronounced in patients with the greatest cognitive and functional impairment. These results may be surprising, as schizophrenia is usually seen as a disease mainly involving higher-order cognitive processes. However, a substantial body of literature composed of electrophysiological (Doniger et al., 2002), fMRI (Martínez et al., 2008) and behavioral (Coleman et al., 2009) studies supports the hypothesis of an early visual dysfunction in patients with schizophrenia, affecting more particularly magnocellular processing. Early auditory dysfunction has been reported as well in schizophrenia. The most replicated result is a deficit of the generation of a specific auditory Event-Related Potential component called mismatch negativity (Javitt and Sweet, 2015). This component reflects automatic sensory discrimination, and is associated with cognitive and psychosocial functioning in healthy adults (Light et al., 2007). According to Javitt (2009), several schizophrenia symptoms could be driven by impairments in basic perceptual processes impacting higher-order processes (i.e. "bottom-up" phenomenon). In our study, connectivity disruptions between visual (RN31, RN32, RN33 and RN34) and auditory (RN24) or crossmodal binding (RN25, RN26 and RN28) networks were linked with positive symptoms, negative symptoms and hallucinations. These results suggest that, in schizophrenia, some positive and negative symptoms may be linked with multisensory integration impairment. Interestingly, Williams et al. (2010) yielded similar results using a behavioral paradigm. Concerning hallucinations, connectivity disruption between RN26 and RN34 seemed more pronounced in the AVH group. As stated above, RN26 encompass several associative regions: medial temporal cortex, superior temporal cortex and insula. Several connectivity studies stressed the involvement of these regions in the pathophysiology of hallucinations (Amad et al., 2014; Hare et al., 2016; Lefebvre et al., 2016; Rolland et al., 2015). More specifically, hippocampal cortex seems to be specifically involved in visual hallucinations (Amad et al., 2014). In this seed-based study, the authors reported decreased connectivity between the hippocampal cortex, on the one hand, the left lenticular nucleus, the right thalamus and the superior temporal gyri on the other hand in AVH patients. RN26 encompass (at least a part of) these three regions. Interestingly, in our study WNFC of RN26 was reduced in PS, but there was no effect of PANSS-P score or hallucinatory status on this FC variable. Moreover, Amad et al. (2014) reported increased connectivity between the hippocampus, the medial prefrontal cortex and the caudate nuclei in AVH patients. This discrepancy with our results may be explained by different analysis techniques: the FC of a large scale network like RN26 can’t only reflect the FC pattern of one of its subregions, the hippocampus, like a seed-based analysis can do. 24
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
multiscale hierarchical functional organization. J. Neurophysiol. 105, 2753–2763. http://dx.doi.org/10.1152/jn.00895.2010. Filippini, N., MacIntosh, B.J., Hough, M.G., Goodwin, G.M., Frisoni, G.B., Smith, S.M., Matthews, P.M., Beckmann, C.F., Mackay, C.E., 2009. Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc. Natl. Acad. Sci. USA 106, 7209–7214. http://dx.doi.org/10.1073/pnas.0811879106. Golland, Y., Golland, P., Bentin, S., Malach, R., 2008. Data-driven clustering reveals a fundamental subdivision of the human cortex into two global systems. Neuropsychologia 46, 540–553. http://dx.doi.org/10.1016/j.neuropsychologia. 2007.10.003. Hare, S.M., Ford, J.M., Ahmadi, A., Damaraju, E., Belger, A., Bustillo, J., Lee, H.J., Mathalon, D.H., Mueller, B.A., Preda, A., van Erp, T.G.M., Potkin, S.G., Calhoun, V. D., Turner, J.A., Functional Imaging Biomedical Informatics Research Network, 2016. Modality-Dependent Impact of Hallucinations on Low-Frequency Fluctuations in Schizophrenia. Schizophr. Bull. 〈http://dx.doi.org/10.1093/schbul/sbw093〉. Hasenkamp, W., James, G.A., Boshoven, W., Duncan, E., 2011. Altered engagement of attention and default networks during target detection in schizophrenia. Schizophr. Res. 125, 169–173. http://dx.doi.org/10.1016/j.schres.2010.08.041. Jafri, M.J., Pearlson, G.D., Stevens, M., Calhoun, V.D., 2008. A method for functional network connectivity among spatially independent resting-state components in schizophrenia. NeuroImage 39, 1666–1681. http://dx.doi.org/10.1016/j. neuroimage.2007.11.001. Jardri, R., Thomas, P., Delmaire, C., Delion, P., Pins, D., 2013. The neurodynamic organization of modality-dependent hallucinations. Cereb. Cortex 1991 (23), 1108–1117. http://dx.doi.org/10.1093/cercor/bhs082. Javitt, D.C., 2009. When doors of perception close: bottom-up models of disrupted cognition in schizophrenia. Annu. Rev. Clin. Psychol. 5, 249–275. http://dx.doi.org/ 10.1146/annurev.clinpsy.032408.153502. Javitt, D.C., Sweet, R.A., 2015. Auditory dysfunction in schizophrenia: integrating clinical and basic features. Nat. Rev. Neurosci. 16, 535–550. http://dx.doi.org/10.1038/ nrn4002. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 13, 261–276. Khadka, S., Meda, S.A., Stevens, M.C., Glahn, D.C., Calhoun, V.D., Sweeney, J.A., Tamminga, C.A., Keshavan, M.S., O’Neil, K., Schretlen, D., Pearlson, G.D., 2013. Is aberrant functional connectivity a psychosis endophenotype? A resting state functional magnetic resonance imaging study. Biol. Psychiatry 74, 458–466. http:// dx.doi.org/10.1016/j.biopsych.2013.04.024. Lefebvre, S., Demeulemeester, M., Leroy, A., Delmaire, C., Lopes, R., Pins, D., Thomas, P., Jardri, R., 2016. Network dynamics during the different stages of hallucinations in schizophrenia. Hum. Brain Mapp. 37, 2571–2586. http://dx.doi.org/10.1002/hbm. 23197. Li, H.-J., Xu, Y., Zhang, K.-R., Hoptman, M.J., Zuo, X.-N., 2014. Homotopic connectivity in drug-naïve, first-episode, early-onset schizophrenia. J. Child Psychol. Psychiatry. http://dx.doi.org/10.1111/jcpp.12307. Light, G.A., Swerdlow, N.R., Braff, D.L., 2007. Preattentive sensory processing as indexed by the MMN and P3a brain responses is associated with cognitive and psychosocial functioning in healthy adults. J. Cogn. Neurosci. 19, 1624–1632. http://dx.doi.org/ 10.1162/jocn.2007.19.10.1624. Mamah, D., Barch, D.M., Repovš, G., 2013. Resting state functional connectivity of five neural networks in bipolar disorder and schizophrenia. J. Affect. Disord. 150, 601–609. http://dx.doi.org/10.1016/j.jad.2013.01.051. Martínez, A., Hillyard, S.A., Dias, E.C., Hagler, D.J., Butler, P.D., Guilfoyle, D.N., Jalbrzikowski, M., Silipo, G., Javitt, D.C., 2008. Magnocellular pathway impairment in schizophrenia: evidence from functional magnetic resonance imaging. J. Neurosci. 28, 7492–7500. http://dx.doi.org/10.1523/JNEUROSCI.1852-08.2008. Mazoyer, B., Mellet, E., Perchey, G., Zago, L., Crivello, F., Jobard, G., Delcroix, N., Vigneau, M., Leroux, G., Petit, L., Joliot, M., Tzourio-Mazoyer, N., 2016. BIL & GIN: a neuroimaging, cognitive, behavioral, and genetic database for the study of human brain lateralization. NeuroImage 124, 1225–1231. http://dx.doi.org/10.1016/j. neuroimage.2015.02.071. Meda, S.A., Gill, A., Stevens, M.C., Lorenzoni, R.P., Glahn, D.C., Calhoun, V.D., Sweeney, J.A., Tamminga, C.A., Keshavan, M.S., Thaker, G., Pearlson, G.D., 2012. Differences in resting-state functional magnetic resonance imaging functional network connectivity between schizophrenia and psychotic bipolar probands and their unaffected first-degree relatives. Biol. Psychiatry 71, 881–889. http://dx.doi.org/10. 1016/j.biopsych.2012.01.025. Mellet, E., Tzourio-Mazoyer, N., Bricogne, S., Mazoyer, B., Kosslyn, S.M., Denis, M., 2000. Functional anatomy of high-resolution visual mental imagery. J. Cogn. Neurosci. 12, 98–109. Menon, V., 2011. Large-scale brain networks and psychopathology: a unifying triple network model. Trends Cogn. Sci. 15, 483–506. http://dx.doi.org/10.1016/j.tics. 2011.08.003. Naveau, M., Doucet, G., Delcroix, N., Petit, L., Zago, L., Crivello, F., Jobard, G., Mellet, E., Tzourio-Mazoyer, N., Mazoyer, B., Joliot, M., 2012. A novel group ICA approach based on multi-scale individual component clustering. Application to a large sample of fMRI data. Neuroinformatics 10, 269–285. http://dx.doi.org/10.1007/s12021012-9145-2. Orliac, F., Naveau, M., Joliot, M., Delcroix, N., Razafimandimby, A., Brazo, P., Dollfus, S., Delamillieure, P., 2013. Links among resting-state default-mode network, salience network, and symptomatology in schizophrenia. Schizophr. Res. 148, 74–80. http:// dx.doi.org/10.1016/j.schres.2013.05.007. Palaniyappan, L., Al-Radaideh, A., Mougin, O., Gowland, P., Liddle, P.F., 2013. Combined white matter imaging suggests myelination defects in visual processing regions in schizophrenia. Neuropsychopharmacology 38, 1808–1815. http://dx.doi.org/10. 1038/npp.2013.80.
Finally, we must consider the fact that patients were on antipsychotic medication, which may have an effect on FC (Cole et al., 2013). However, Li et al. (2014) reported FC disruptions in drug-naïve schizophrenia patients, and Sarpal et al. (2015) reported an increase in cortico-subcortical connectivity associated with antipsychotic treatment. According to the latter, FC disruption in schizophrenia may be a state-dependent phenomenon correlated with the severity of the symptoms. In our study, the FC of two edges, both linked to RN26, seemed strongly impacted by medication dosage. In conclusion, this study highlighted within-network and betweennetwork FC disruptions in patients with schizophrenia, with a clear trend toward weaker FC. Our exploratory approach allowed us to point out the fact that sensory networks appeared more functionally altered than high-level cognitive networks. These alterations were correlated with negative symptoms, positive symptoms and hallucinations, indicating abnormalities in visual processing and crossmodal binding in schizophrenia. Moreover, an anomalous synchronization between a visual network and the precuneus was linked with delusions and hallucinations. Altogether, our results support the assumption that some schizophrenia symptoms may be linked with low-order sensory alterations impacting reverberating on higher-order cognitive processes, i.e. the “bottom-up” hypothesis of schizophrenia symptoms. Acknowledgment The authors thank the members of the Cyceron imaging platform for their assistance. This work was supported by the French Health and Research Ministries in a Programme Hospitalier de Recherche Clinique (n° 2003-R06-10). The funding source had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.pscychresns.2017.04. 003. References Abbott, C.C., Kim, D., Sponheim, S.R., Bustillo, J., Calhoun, V.D., 2010. Decreased default mode neural modulation with age in schizophrenia. Am. J. Geriatr. Psychiatry 18, 897–907. http://dx.doi.org/10.1097/JGP.0b013e3181e9b9d9. Amad, A., Cachia, A., Gorwood, P., Pins, D., Delmaire, C., Rolland, B., Mondino, M., Thomas, P., Jardri, R., 2014. The multimodal connectivity of the hippocampal complex in auditory and visual hallucinations. Mol. Psychiatry 19, 184–191. http:// dx.doi.org/10.1038/mp.2012.181. Arbabshirani, M.R., Kiehl, K.A., Pearlson, G.D., Calhoun, V.D., 2013. Classification of schizophrenia patients based on resting-state functional network connectivity. Front. Neurosci. 7, 133. http://dx.doi.org/10.3389/fnins.2013.00133. Beckmann, C.F., Smith, S.M., 2004. Probabilistic independent component analysis for functional magnetic resonance imaging. IEEE Trans. Med. Imaging 23, 137–152. http://dx.doi.org/10.1109/TMI.2003.822821. Bressler, S.L., Menon, V., 2010. Large-scale brain networks in cognition: emerging methods and principles. Trends Cogn. Sci. 14, 277–290. http://dx.doi.org/10.1016/j. tics.2010.04.004. Bushara, K.O., Hanakawa, T., Immisch, I., Toma, K., Kansaku, K., Hallett, M., 2003. Neural correlates of cross-modal binding. Nat. Neurosci. 6, 190–195. http://dx.doi. org/10.1038/nn993. Cole, D.M., Oei, N.Y.L., Soeter, R.P., Both, S., van Gerven, J.M.A., Rombouts, S.A.R.B., Beckmann, C.F., 2013. Dopamine-dependent architecture of cortico-subcortical network connectivity. Cereb. Cortex 1991 (23), 1509–1516. http://dx.doi.org/10. 1093/cercor/bhs136. Coleman, M.J., Cestnick, L., Krastoshevsky, O., Krause, V., Huang, Z., Mendell, N.R., Levy, D.L., 2009. Schizophrenia patients show deficits in shifts of attention to different levels of global-local stimuli: evidence for magnocellular dysfunction. Schizophr. Bull. 35, 1108–1116. http://dx.doi.org/10.1093/schbul/sbp090. Doniger, G.M., Foxe, J.J., Murray, M.M., Higgins, B.A., Javitt, D.C., 2002. Impaired visual object recognition and dorsal/ventral stream interaction in schizophrenia. Arch. Gen. Psychiatry 59, 1011–1020. Doucet, G., Naveau, M., Petit, L., Delcroix, N., Zago, L., Crivello, F., Jobard, G., TzourioMazoyer, N., Mazoyer, B., Mellet, E., Joliot, M., 2011. Brain activity at rest: a
25
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
WNR.0000000000000622. van den Heuvel, M.P., Fornito, A., 2014. Brain networks in schizophrenia. Neuropsychol. Rev. 24, 32–48. http://dx.doi.org/10.1007/s11065-014-9248-7. Whitfield-Gabrieli, S., Thermenos, H.W., Milanovic, S., Tsuang, M.T., Faraone, S.V., McCarley, R.W., Shenton, M.E., Green, A.I., Nieto-Castanon, A., LaViolette, P., Wojcik, J., Gabrieli, J.D.E., Seidman, L.J., 2009. Hyperactivity and hyperconnectivity of the default network in schizophrenia and in first-degree relatives of persons with schizophrenia. Proc. Natl. Acad. Sci. USA 106, 1279–1284. http://dx.doi.org/10. 1073/pnas.0809141106. Williams, L.E., Light, G.A., Braff, D.L., Ramachandran, V.S., 2010. Reduced multisensory integration in patients with schizophrenia on a target detection task. Neuropsychologia 48, 3128–3136. http://dx.doi.org/10.1016/j.neuropsychologia. 2010.06.028. Wolf, N.D., Sambataro, F., Vasic, N., Frasch, K., Schmid, M., Schönfeldt-Lecuona, C., Thomann, P.A., Wolf, R.C., 2011. Dysconnectivity of multiple resting-state networks in patients with schizophrenia who have persistent auditory verbal hallucinations. J. Psychiatry Neurosci. 36, 366–374. http://dx.doi.org/10.1503/jpn.110008. Woodward, N.D., Rogers, B., Heckers, S., 2011. Functional resting-state networks are differentially affected in schizophrenia. Schizophr. Res. 130, 86–93. http://dx.doi. org/10.1016/j.schres.2011.03.010. Wotruba, D., Michels, L., Buechler, R., Metzler, S., Theodoridou, A., Gerstenberg, M., Walitza, S., Kollias, S., Rössler, W., Heekeren, K., 2014. Aberrant coupling within and across the default mode, task-positive, and salience network in subjects at risk for psychosis. Schizophr. Bull. 40, 1095–1104. http://dx.doi.org/10.1093/schbul/ sbt161. Yu, Q., Plis, S.M., Erhardt, E.B., Allen, E.A., Sui, J., Kiehl, K.A., Pearlson, G., Calhoun, V.D., 2012. Modular organization of functional network connectivity in healthy controls and patients with schizophrenia during the resting state. Front. Syst. Neurosci. 5. http://dx.doi.org/10.3389/fnsys.2011.00103. Zalesky, A., Fornito, A., Bullmore, E.T., 2010. Network-based statistic: identifying differences in brain networks. NeuroImage 53, 1197–1207. http://dx.doi.org/10. 1016/j.neuroimage.2010.06.041. Zhang, S., Li, C.R., 2012. Functional connectivity mapping of the human precuneus by resting state fMRI. NeuroImage 59, 3548–3562. http://dx.doi.org/10.1016/j. neuroimage.2011.11.023.
Papo, D., 2013. Why should cognitive neuroscientists study the brain's resting state? Front. Hum. Neurosci. 7, 45. http://dx.doi.org/10.3389/fnhum.2013.00045. Power, J.D., Barnes, K.A., Snyder, A.Z., Schlaggar, B.L., Petersen, S.E., 2012. Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. Neuroimage 59, 2142–2154. http://dx.doi.org/10.1016/j.neuroimage.2011. 10.018. Rogers, B.P., Morgan, V.L., Newton, A.T., Gore, J.C., 2007. Assessing functional connectivity in the human brain by fMRI. Magn. Reson. Imaging 25, 1347–1357. http://dx.doi.org/10.1016/j.mri.2007.03.007. Rolland, B., Amad, A., Poulet, E., Bordet, R., Vignaud, A., Bation, R., Delmaire, C., Thomas, P., Cottencin, O., Jardri, R., 2015. Resting-state functional connectivity of the nucleus accumbens in auditory and visual hallucinations in schizophrenia. Schizophr. Bull. 41, 291–299. http://dx.doi.org/10.1093/schbul/sbu097. Rotarska-Jagiela, A., van de Ven, V., Oertel-Knöchel, V., Uhlhaas, P.J., Vogeley, K., Linden, D.E.J., 2010. Resting-state functional network correlates of psychotic symptoms in schizophrenia. Schizophr. Res. 117, 21–30. http://dx.doi.org/10.1016/ j.schres.2010.01.001. Sarpal, D.K., Robinson, D.G., Lencz, T., Argyelan, M., Ikuta, T., Karlsgodt, K., Gallego, J.A., Kane, J.M., Szeszko, P.R., Malhotra, A.K., 2015. Antipsychotic treatment and functional connectivity of the striatum in first-episode schizophrenia. JAMA Psychiatry 72, 5–13. http://dx.doi.org/10.1001/jamapsychiatry.2014.1734. Satterthwaite, T.D., Wolf, D.H., Loughead, J., Ruparel, K., Elliott, M.A., Hakon, H., Gur, R.C., Gur, R.E., 2012. Impact of in-scanner head motion on multiple measures of functional connectivity: relevance for studies of neurodevelopment in youth. NeuroImage 60, 623–632. http://dx.doi.org/10.1016/j.neuroimage.2011.12.063. Senkowski, D., Schneider, T.R., Foxe, J.J., Engel, A.K., 2008. Crossmodal binding through neural coherence: implications for multisensory processing. Trends Neurosci. 31, 401–409. http://dx.doi.org/10.1016/j.tins.2008.05.002. Smith, S.M., Fox, P.T., Miller, K.L., Glahn, D.C., Fox, P.M., Mackay, C.E., Filippini, N., Watkins, K.E., Toro, R., Laird, A.R., Beckmann, C.F., 2009. Correspondence of the brain's functional architecture during activation and rest. Proc. Natl. Acad. Sci. USA 106, 13040–13045. http://dx.doi.org/10.1073/pnas.0905267106. Su, J., Shen, H., Zeng, L.-L., Qin, J., Liu, Z., Hu, D., 2016. Heredity characteristics of schizophrenia shown by dynamic functional connectivity analysis of resting-state functional MRI scans of unaffected siblings. Neuroreport. http://dx.doi.org/10.1097/
26
Contents lists available at ScienceDirect
Psychiatry Research: Neuroimaging journal homepage: www.elsevier.com/locate/psychresns
Network modeling of resting state connectivity points towards the bottom up theories of schizophrenia
MARK
François Orliacb, Pascal Delamillieurea,b, Nicolas Delcroixc, Mikael Naveaud, Perrine Brazoa,b, ⁎ Annick Razafimandimbyb, Sonia Dollfusa,b, Marc Joliote,f,g, a
CHU de Caen, Department of Psychiatry, Caen F-14000, France Université de Caen Basse-Normandie, UFR de Médecine, UMR 6301 ISTCT, ISTS group, Caen F-14000, France c GIP CYCERON, UMS 3408, Caen F-14000, France d INSERM UMR-S U919 SP2U, Université Caen Basse-Normandie, Caen F-14000, France e GIN, University of Bordeaux, Institut des Maladies Neurodégénératives, UMR 5293, Bordeaux F-33000, France f GIN, CNRS, Institut des Maladies Neurodégénératives, UMR 5293, Bordeaux F-33000, France g GIN, CEA, Institut des Maladies Neurodégénératives, UMR 5293, Bordeaux F-33000, France b
A R T I C L E I N F O
A B S T R A C T
Keywords: Functional magnetic resonance imaging Hallucinations Neural networks
The dysconnectivity theory of schizophrenia proposes that schizophrenia symptoms arise from abnormalities in neuronal synchrony. Resting-state Functional Connectivity (FC) techniques allow us to highlight synchronization of large-scale networks, the Resting-state Networks (RNs). A large body of work suggests that disruption of RN synchronization could give rise to specific schizophrenia symptoms. The present study aimed to explore withinand between-network FC strength of 34 RNs in 29 patients suffering from schizophrenia, and their relationships with schizophrenia symptoms. Resting-state data were analyzed using independent component analysis and dual-regression techniques. Our results showed that both within-RN and between-RN FC were disrupted in patients with schizophrenia, with a global trend toward weaker FC. This decrease affected more particularly visual, auditory and crossmodal binding networks. These alterations were correlated with negative symptoms, positive symptoms and hallucinations, indicating abnormalities in visual processing and crossmodal binding in schizophrenia. Moreover, we stressed an anomalous synchronization between a visual network and a network thought to be engaged in mental imaging processes, correlated with delusions and hallucinations. Altogether, our results supported the assumption that some schizophrenia symptoms may be related to low-order sensory alterations impacting higher-order cognitive processes, i.e. the “bottom-up” hypothesis of schizophrenia symptoms.
1. Introduction The dysconnectivity theory of schizophrenia proposes that many symptoms may be related to a failure to integrate the activity of local and distributed neural circuits (van den Heuvel and Fornito, 2014). One way of assessing brain connectivity is to study how multiple brain regions functionally interact while an individual is not engaged in a specific task, i.e. using resting-state blood oxygen level-dependent Functional Connectivity (FC) (Rogers et al., 2007).
Numerous FC studies conducted on healthy volunteers reported that the brain exhibited a structured neural activity during the resting state (Bressler and Menon, 2010; Doucet et al., 2011). This can be observed at different scales: the system scale, the module scale (Fig. S1 Supplementary data) and the Resting-state Network (RN) scale (Fig. S2 Supplementary data). Among these scales, the partition into RNs appears to be the most relevant in the context of schizophrenia research. RNs have been suggested to overlap with the networks subtending the brain in action (Smith et al., 2009). In other words,
Abbreviations: AH, Auditory Hallucinations in medical history; ANCOVA, Analysis Of COVAriance; ANOVA, ANalysis Of Variance; AVH, Auditory and Visual Hallucinations in medical history; BNFC, Between-Network Functional Connectivity; C-FD, Cumulated Framewise Displacement; Cpz-eq, average Chlorpromazine equivalent; DMN, Default-Mode Network; FC, Functional Connectivity; FD, Framewise Displacement; HC, Healthy Controls; ICA, Independent Component Analysis; MANOVA, Multivariate ANalysis Of Variance; MICCA, Multi-scale Individual Component Clustering Analysis; MINI, Mini International Neuropsychiatric Interview; MRI, Magnetic Resonance Imaging; NBS, Network-Based Statistics; NH, No Hallucinations in medical history; PANSS, Positive And Negative Syndrome Scale; PANSS-N, Negative symptoms subscale of the PANSS; PANSS-P, Positive symptoms subscale of the PANSS; RN, Resting-state Network; PS, Patients with Schizophrenia; WNFC, Within-Network Functional Connectivity ⁎ Corresponding author at: GIN, UMR5293, Université de Bordeaux, 146, rue Léo Saignat, CS 61292, 33076 Bordeaux Cedex, France. E-mail address: [email protected] (M. Joliot). http://dx.doi.org/10.1016/j.pscychresns.2017.04.003 Received 22 October 2016; Received in revised form 15 March 2017; Accepted 7 April 2017 Available online 12 April 2017 0925-4927/ © 2017 Published by Elsevier Ireland Ltd.
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
functional networks seem to be continuously and dynamically “active” even when the brain is “at rest”. RNs have already been shown to be of potential clinical value as rich and sensitive markers of disease (Menon, 2011). They offer several advantages: the full repertoire of RNs can be tested in a single scanning session without having to decide a priori what functional paradigm is most likely to be useful, and minimal individual participation is required during scanning. This latter point can be critical in clinical settings (e.g. Autism Spectrum Disorders, severe schizophrenia or Alzheimer Disease). As a consequence, the study of RNs allows exploring cognitive processes that are either understudied or examined under simplifying assumptions due to scanning constraints (Papo, 2013; Smith et al., 2009). Moreover, disruptions of RNs may contribute to specific patterns of cognitive and behavioral impairments, providing new insights into aberrant brain organization in several psychiatric and neurological disorders (Menon, 2011; Papo, 2013). Several authors reported within-RN connectivity disruptions in schizophrenia during the resting state. Almost all of these studies were hypothesis-driven, hence the authors selected a few RNs on the basis of their relevance regarding the physiopathology of schizophrenia. As schizophrenia is usually seen to be related to higher-order cognitive dysfunctions, higher order RNs were chosen: DMN, central executive, attentional, language or salience networks (Orliac et al., 2013; Rotarska-Jagiela et al., 2010; Wolf et al., 2011; Woodward et al., 2011). A few studies examined between-RN connectivity in schizophrenia during the resting state (Arbabshirani et al., 2013; Jafri et al., 2008; Khadka et al., 2013; Mamah et al., 2013; Meda et al., 2012; Yu et al., 2012). It should be noted that, apart from Yu et al. (2012) who studied 57 RNs, these studies have focused on less than 20 RNs. All of these reported reduced between-RN connectivity in patients with schizophrenia, except Jafri et al. (2008) and Khadka et al. (2013) who reported both connectivity increases and decreases. In the light of this literature, and by contrast with our previous hypothesis-driven work (Orliac et al., 2013), it seemed relevant to explore RN functional connectivity in schizophrenia as exhaustively as possible. To achieve this, we studied both within-RN and between-RN functional connectivity strength of 34 networks encompassing 98% of the cerebral gray matter, including low-order and high-order networks, and their relationships with schizophrenia symptoms.
in medical history (AH group), or Auditory and Visual Hallucinations in medical history (AVH group). The local ethics committee (CPP de Basse-Normandie, France) approved the study. All participants gave written informed consent. 2.2. Image acquisition Data acquisition was performed on a 3T Philips Achieva MRI scanner. Structural data were acquired using a high-resolution, threedimensional T1-weighted volume (repetition time (TR)=20 ms; echo time (TE)=4.6 ms; flip angle=10°; inversion time=800 ms; turbo field echo factor=65; sense factor=2; field of view=256×256×180 mm; 1×1×1 mm3 isotropic voxel size), and a T2*-weighted, multi-slice acquisition (T2*-weighted fast-field echo; TR=3500 ms; TE=35 ms; flip angle=90°; sense factor=2; 70 axial slices; 2×2×2 mm3 isotropic voxel size). Spontaneous brain activity was monitored using BOLD fMRI while the participants performed a resting-state condition for 8 min (T2*-echo planar imaging; 240 volumes; TR=2 s; TE =35 ms; flip angle=80°; 31 axial slices; 3.75×3.75×3.75 mm3 isotropic voxel size). Immediately before fMRI scanning, participants were instructed to “keep their eyes closed, to relax, to refrain from moving, to stay awake, and to let their thoughts come and go.” 2.3. Image processing Pre-processing of the functional data was based on the methods described in Naveau et al. (2012). Briefly, it included slice-timing correction, motion correction, co-registration to structural scan, spatial normalization to the MNI template and spatial smoothing (6 mm Gaussian kernel). Each individual's structural scan was segmented into gray matter, white matter, and cerebrospinal fluid using the unified segmentation approach (Statistical Parametric Mapping 5; www.fil.ion. ucl.ac.uk/spm5). Time series for white matter, cerebrospinal fluid, and the six motion parameters were regressed out of the data. Finally, fMRI data were temporally filtered using band-pass filtering (0.01 Hz < f < 0.1 Hz). 2.4. Motion control Head motion correction has become a prominent concern in the field of resting state FC, especially during clinical studies. As a matter of facts, several authors have reported that small amounts of movement can produce substantial changes in the timecourses of resting state data, and cause spurious but spatially structured patterns in FC (Power et al., 2012; Satterthwaite et al., 2012). As a consequence, an index of quality control was computed for each participant: the Framewise Displacement (FD) of head position, calculated as the sum of the absolute values of the 6 translational and rotational realignment parameters (Power et al., 2012). Every participant exhibiting FD higher than 0.5 mm in more than 10% of the 240 volumes was excluded from the study. Then, a single motion index was calculated for each participant: the Cumulated Framewise Displacement (C-FD), defined as the sum of FDs over all 240 images. C-FD values were tested for group differences and added as covariate of no interest in our connectivity analyses (see 2.9.) to control for this potentially confounding factor.
2. Methods 2.1. Participants Twenty-nine Patients with Schizophrenia (PS group) attending the Department of Psychiatry of Caen University Hospital and twenty-nine matched healthy controls (HC group) were included in the study. We used a sample expanded from a previous study (Orliac et al., 2013). PS and HC groups were matched for age, sex, handedness, and educational level on a one-to-one basis. All participants were required to be between 18 and 60 years of age. All were screened for magnetic resonance imaging (MRI) contraindications, and participants with a history of a major medical condition, neurological disease, or substance abuse were excluded from the study. PS group participants were diagnosed by an experienced clinician using the Mini International Neuropsychiatric Interview (MINIplus v.4.5). They were required to have been stable on antipsychotic medication for at least four months prior to the study. The Positive and Negative Syndrome Scale (PANSS) (Kay et al., 1987) was used to assess positive (PANSS-P) and negative (PANSS-N) symptoms. Daily antipsychotic medication dosage at the time of inclusion was recorded and converted into chlorpromazine equivalents (mg/d). Our findings pointed towards a need for post-hoc analyses. For this purpose, PS patients were classified into three subgroups according to their hallucinatory status (items A6b and A7b of the MINIplus): No Hallucinations in medical history (NH group), Auditory Hallucinations
2.5. Reference maps of the RNs Reference maps for RNs were estimated from BIL & GIN cohort (Mazoyer et al., 2016) resting-state datasets (n=282), using a group ICA approach based on multi-scale individual component clustering (MICCA, see Naveau et al. (2012)). The MICCA analysis retained 34 non-artifactual RNs. The positive map of each RN was thresholded using a mixture model (p > 0.95) (Beckmann and Smith, 2004). Additionally, voxels exhibiting a probability of being included in the 20
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
Table 1 Population demographics and clinical characteristics. Group
p value for Group comparison
N
PS 28
HC 29
Sex (F/M) Age (years; mean ± SD) Handedness (left/right) Education (years; mean ± SD)
(7/21) 36.1 ± 8 (4/24) 12 ± 1.6
(7/22) 34.5 ± 8.9 (4/25) 12.8 ± 2.4
0.94 0.75 0.96 0.08
Illness duration (years; mean ± SD) Medication (Cpz-eq; mg/d; mean ± SD) PANSS-P (mean ± SD) PANSS-N (mean ± SD) Hallucinatory status (NH/AH/AVH)
12.3 ± 7.9 339.6 ± 255.7
n/a n/a
n/a n/a
12.9 ± 5 12.8 ± 4.3 11 / 8 / 9
n/a n/a n/a
n/a n/a n/a
AH: Auditory Hallucinations in medical history; AVH: Auditory and Visual Hallucinations in medical history; Cpz-eq: average Chlorpromazine equivalent; HC: Healthy Controls; n/a: not applicable; NH: No Hallucinations in medical history; PANSS-N: Negative subscale of the Positive and Negative Syndrome Scale; PANSS-P: Positive subscale of the Positive and Negative Syndrome Scale; PS: Patients with Schizophrenia group.
SPM gray matter canonical map below 0.3 were filtered out. The resultant spatial map of each RN is displayed in Fig. S2 Supplementary data. The numbering of the RNs refers to the hierarchical functional clustering of RNs proposed by Doucet et al. (2011), see Fig. S1.
et al. (2010) Network-Based Statistics (NBS), a new approach designed to identify significant differences between graphs. Briefly, NBS is intended to check for the family-wise error rate using the principles of cluster-based thresholding of statistical parametric maps. We used the NBS toolbox with default settings (T=3.1, p < 0.05 NBS, 5000 permutations), and C-FD was added as covariate of no interest in the model. Second, ANCOVA was conducted to test for a group difference on each BNFC value with the more stringent Bonferroni-Holm correction, still controlling for C-FD (df1=1/df2=53, p < 0.05 BonferroniHolm corrected, n=561).
2.6. Dual regression The dual regression technique (Filippini et al., 2009) was used to build for each individual the level spatial maps corresponding to the 34 RNs, with associated time series. We used the dedicated FSL tool (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/DualRegression). Each voxel value of these maps was a Z-normalized regression beta coefficient, interpreted as a measure of the strength of the association between the voxel time series and the whole network time series, i.e., functional connectivity within this network.
2.9.3. Correlation with clinical variables When a significant group difference was highlighted for a FC variable, a two-tailed non-parametric Spearman test was used to assess the correlation between PANSS subscales and the FC values of this variable within the PS group, controlling for C-FD. Post-hoc analyses were conducted regarding the hallucinatory status: when a correlation was highlighted between an FC variable and PANSS-P score, ANCOVA was conducted to test for an effect of the hallucinatory status on FC, controlling for C-FD (df1:2/df2:22, p < 0.05). When statistically significant, ANCOVA was followed by a two-tailed Tukey's Honest Significant Difference test (HSD; p < 0.05). To check for a possible confounding effect of antipsychotic medication on FC, complementary regression analyses were carried out adding medication dosage as a covariate.
2.7. Within-network functional connectivity For each RN and each individual, the Within-Network Functional Connectivity (WNFC) strength was computed. The WNFC score was defined as the mean Z-score of the voxels located within the corresponding reference RN mask. 2.8. Between-network functional connectivity FC between each possible pair of RNs, annotated Between-Network Functional Connectivity (BNFC), was computed using Pearson's correlation, resulting in a 34×34 connectivity matrix for each participant. This connectivity matrix allowed us to build a weighted connectivity graph for each participant, composed of 34 nodes (representing the RNs) and 561 edges (whose weights represented Pearson's r between two time series, i.e. FC between two RNs). Correlation coefficients were transformed using Fisher's r to z.
3. Results 3.1. Motion control One participant (belonging to the PS group) was excluded from the study due to significant head motion. There was no difference between the two groups of remaining participants regarding C-FD (two-tailed Ttest, T(55)=0.41, p=0.68).
2.9. Statistical analysis 2.9.1. WNFC WNFC values were tested for group difference. First, we tested for a group effect using a Multivariate ANalysis Of VAriance (MANOVA, p < 0.05) including all 34 RNs as dependent variables. Then, ANalysis of COVAriance (ANCOVA) was conducted to test for a group difference on each RN considered individually, controlling for C-FD (df1=1/ df2=53, p < 0.05 Bonferroni-Holm corrected, n=34).
3.2. Clinical data Table 1 provides detailed demographic and clinical data for the remaining participants. None of the matching criteria were statistically different between the groups. With regard more specifically to the PS group, there was no difference between the three “Hallucinatory status” subgroups concerning the PANSS-P score (Analysis Of VAriance, F(2,25)=1.33, p=0.28), the PANSS-N score (ANOVA, F(2,25)=0.64, p=0.54), or medication dosage (ANOVA, F(2,25)=1.14, p=0.33).
2.9.2. BNFC BNFC values were tested for group difference. First, we used Zalesky 21
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
Fig. 1. Between network connectivity is altered in schizophrenia. Connectivity graph is displayed using the hierarchical functional model from Doucet et al. (2011). Gray edges represent the most significant edges in our control group (r > 0.4). Bonf: Bonferroni-Holm correction; HC: Healthy Controls; NBS: Network Based Statistics; RN: Resting-state Network; PS: Patients with Schizophrenia.
3.3. Group comparisons
3.4. Correlation with clinical variables
3.3.1. WNFC MANOVA showed weaker WNFC (Wilks' λ=0.23, χ2(34) =55.7, p=0.01) in the RNs of the PS group. Considered individually, 5 RNs showed significantly reduced WNFC in PS (Fig. S3 Supplementary data): a right parieto-frontal RN (RN10, p < 0.001), a temporo-insular RN (RN26, p < 0.001) and 3 occipital RNs: RN31 (p < 0.001); RN32 (p=0.001); and RN33 (p < 0.001).
3.4.1. WNFC WNFC of RN32 (occipital RN) was negatively correlated with the PANSS-P score in the PS group (rs=−0.43, p=0.02), controlling for CFD. There was no effect of the hallucinatory status on this connectivity variable. This correlation remained significant (rs=−0.41, p=0.03) after adding medication dosage as a covariate. 3.4.2. BNFC Controlling for C-FD, BNFC was negatively correlated with the PANSS-P score in four edges (Fig. 2A): RN24-RN33 (rs=−0.44, p=0.02), RN25-RN33 (rs=−0.43, p=0.02), RN26-RN32 (rs=−0.43, p=0.03), and RN26-RN34 (rs=−0.50, p=0.008). ANCOVA revealed a trend toward an effect of hallucinatory status on the BNFC of RN26RN34, controlling for C-FD (F(2,22)=3.23, p=0.06). In this edge, posthoc Tukey's HSD highlighted that BNFC was significantly lower in AVH patients compared to NH patients. Controlling for C-FD, BNFC was positively correlated (Fig. 2B) with the PANSS-P score in RN6-RN31, an edge linking an occipital RN to the posterior cingulate/precuneus RN (rs=0.42, p=0.03). ANCOVA showed an effect of the hallucinatory status on BNFC (F(2,22)=5.29, p=0.02), and post-hoc Tukey's HSD highlighted that BNFC was significantly higher in AVH patients compared to both NH and AH patients. Correlations between BNFC values and PANSS-P score were no longer significant after adding medication dosage as a covariate. Posthoc analyses were carried out to disentangle the interaction between BNFC, PANSS-P score and medication dosage. First, it appeared that PANSS-P score was strongly correlated with medication dosage
3.3.2. BNFC The NBS analysis uncovered 19 edges with between-group BNFC differences (Fig. 1). Fifteen edges showed reduced BNFC in the PS group, and all but one (RN26-RN28) involved an occipital RN. Most of these (n=13) were positive edges in HC that exhibited lower positive (n=8) or negative (n=5) values in the PS group. The two remaining edges, both involving RN18 (a parieto-frontal RN), were negative edges exhibiting lower negative values in the PS group. This difference remained significant using Bonferroni-Holm correction for four edges. All of these were between an occipital RN (RN31 or RN33) and a temporal RN (RN24, RN25 or RN26). Four edges showed a BNFC increase in the PS group, and all involved RN6, which is an RN encompassing mainly precuneus and posterior cingulate. Three of these are negative edges in HC that become closer to zero in PS. The remaining edge (RN6-RN31) is positive, exhibiting higher values in the PS group. None of these differences remained significant using Bonferroni-Holm correction.
22
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
Fig. 2. Correlation between edge functional connectivity, PANSS subscales and hallucinatory status. Blue edge: connectivity decrease related to greater symptomatology; Red edge: connectivity increase related to greater symptomatology. (A) negative correlations with the positive subscale. (B) positive correlations with the positive subscale. (C) negative correlations with the negative subscale. AH: Auditory Hallucinations; AVH: Auditory and Visual Hallucinations; BNFC: Between Network Functional Connectivity; NH: No Hallucinations; PANSS-N: Negative subscale of the Positive and Negative Syndrome Scale; PANSS-P: Positive subscale of the Positive and Negative Syndrome Scale; RN: Resting-state Network (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).
cant after adding medication dosage as a covariate for both RN26-RN28 (rs=−0.39, p=0.04) and RN26-RN31 (rs=−0.43, p=0.02).
(rs=0.56, p=0.002). Then, among all BNFC values exhibiting group differences, two edges were strongly correlated with medication dosage: RN26-RN32 (rs=−0.60, p < 0.001), and RN26-RN34 (rs=−0.65, p < 0.001). The effect of the hallucinatory status on BNFC remained significant after adding medication dosage as a covariate, for the two edges: RN26RN34 (F(2,22)=3.46, p=0.05) and RN6-RN31 (F(2,22)=5.35, p=0.01). Controlling for C-FD, BNFC was negatively correlated (Fig. 2C) with the PANSS-N score in two edges involving RN26, a temporo-insular RN: RN26-RN28 (rs=−0.39, p=0.04) and RN26-RN31 (rs=−0.46, p=0.01). Correlations between BNFC and PANSS-N score remained signifi-
4. Discussion Our results show that both within-RN and between-RN FC are disrupted in schizophrenia, with a global trend toward weaker FC. Within-RN connectivity (WNFC) was reduced in all the RNs, and mainly in visual networks. Between-RN connectivity (BNFC) analyses revealed that almost all the edges showing weaker connectivity in patients with schizophrenia involved an occipital visual network. Moreover, the most significant differences were found in between two 23
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
Although our results showed a trend toward weaker FC in patients with schizophrenia, between-network connectivity was increased in four edges in this group. All these edges were between RN6 (precuneus and posterior cingulate), on one hand, and an RN belonging to the extrinsic system, on the other. Increased connectivity resulted in diminished anticorrelation between RN6, whose regions are considered to be major components of the intrinsic system (Golland et al., 2008), and three sensory-motor networks (RN20, RN21 and RN22) belonging to the extrinsic system. Diminished anticorrelation between these two systems has already been reported in schizophrenia patients (Abbott et al., 2010; Hasenkamp et al., 2011; Whitfield-Gabrieli et al., 2009). According to Wotruba et al. (2014), diminished anticorrelation between the intrinsic and extrinsic system may lead to confusion between internal information and external stimuli processing. In our study, the FC increase between RN6 and an associative visual network (RN31) was correlated with positive symptoms, and this increase was significantly more pronounced in the AVH group. The precuneus, and more specifically its dorsal regions, seems to be engaged in visual mental imagery processes (Mellet et al., 2000; Zhang and Li, 2012). Thus, we suggest that this aberrant coupling may lead to confusion between the processing of internal visual mental imagery (RN6) and the processing of external visual stimuli (RN31), resulting in delusions and hallucinations. In our study, no patient exhibited visual hallucinations only (i.e. without auditory hallucinations), which prevents a firm conclusion about a specific relationship between visual hallucinations and the FC of RN6-RN31. However, two arguments point in that direction: FC of RN6-RN31 was significantly higher in the AVH group compared to the AH group; and the putative role of RN6 and RN31 itself points towards a dysfunction specifically affecting the visual modality. Interestingly, Jardri et al. (2013) reported increased blood oxygen level-dependent signal in associative visual regions resembling RN31 during visual hallucinations in adolescents experiencing a brief psychotic disorder. In this study temporal and spatial instability of the DMN was correlated with the severity of hallucinations. Moreover the authors reported a negative correlation between associative visual regions and the DMN time course during visual hallucinations. This apparent discrepancy with our results may be surprising, as RN6 is usually regarded as a DMN subcomponent. It should be noted that, in our study, patients in the AVH group have experienced visual hallucination in their medical history, but they were not necessarily hallucinating at the time of scanning. This suggests a different connectivity pattern between DMNrelated regions and visual associative regions, depending on the occurrence of visual hallucinations: anticorrelation during the hallucinatory experience (state marker), and diminished anticorrelation between the hallucination periods (trait marker). This is in line with a recent study exploring the networks dynamics during the different stages of hallucinations in PS (Lefebvre et al., 2016), which reported such a sequential pattern between the DMN and hallucination-related components. This study had several limitations. First, the significance of our results must be discussed with caution. RNs were elicited during rest, and the determination of their putative role relies on their close correspondence with the networks elicited by task-based studies (Smith et al., 2009). Another limitation of this study was the fact that the analyses regarding the hallucinatory status were conducted using a non-specific global severity score (MINIplus), and that the occurrence of hallucination wasn’t assessed during scanning. As a matter of fact, the present study wasn’t initially designed to specifically investigate the pathophysiology of hallucinations in PS, but when discovering our preliminary results we were struck by the fact that sensory networks appeared to be the more affected. This pointed towards a need for posthoc analyses probing the possible relationship between the alteration of these RN and the occurrence of auditory and/or visual hallucinations. The only clinical data available distinguishing between auditory and verbal hallucinations was from MINIplus, which reports the occurrence of hallucinations in medical history, not only at the time of scanning.
medial occipital networks (RN31 and RN33) and three closely linked sensory networks (RN24, RN25 and RN26). RN24 is a low-level auditory network. RN 25 and RN26 encompass several associative regions from different sensory modalities: auditory (Heschl's gyrus for RN25), visual (cuneus for RN25 and lingual gyrus for RN26), auditoryvisual integration (bilateral superior temporal sulcus for RN26) and somato-sensory integration (bilateral insula for RN25 and RN26). On this basis, we assume that FC of these two RNs may be related to how the brain binds signals from multiple sensory modalities together to produce unified percepts of objects and events, i.e. crossmodal binding (Bushara et al., 2003; Senkowski et al., 2008). Altogether, our results suggest that FC disruption in schizophrenia affects visual, auditory and crossmodal binding networks more particularly. These results are in line with two studies (Arbabshirani et al., 2013; Meda et al., 2012) reporting between-RN connectivity disruption involving sensory-motor and visual RNs in schizophrenia. Interestingly, Su et al. (2016), in a recent study using dynamic functional connectivity techniques, reported shared transient dysconnectivity states within visual pathways between schizophrenia patients and their unaffected siblings. Structural connectivity of the visual cortex also seems to be disrupted in schizophrenia. As a matter of facts, Palaniyappan et al. (2013) reported a dysmyelination in visual processing regions, which was most pronounced in patients with the greatest cognitive and functional impairment. These results may be surprising, as schizophrenia is usually seen as a disease mainly involving higher-order cognitive processes. However, a substantial body of literature composed of electrophysiological (Doniger et al., 2002), fMRI (Martínez et al., 2008) and behavioral (Coleman et al., 2009) studies supports the hypothesis of an early visual dysfunction in patients with schizophrenia, affecting more particularly magnocellular processing. Early auditory dysfunction has been reported as well in schizophrenia. The most replicated result is a deficit of the generation of a specific auditory Event-Related Potential component called mismatch negativity (Javitt and Sweet, 2015). This component reflects automatic sensory discrimination, and is associated with cognitive and psychosocial functioning in healthy adults (Light et al., 2007). According to Javitt (2009), several schizophrenia symptoms could be driven by impairments in basic perceptual processes impacting higher-order processes (i.e. "bottom-up" phenomenon). In our study, connectivity disruptions between visual (RN31, RN32, RN33 and RN34) and auditory (RN24) or crossmodal binding (RN25, RN26 and RN28) networks were linked with positive symptoms, negative symptoms and hallucinations. These results suggest that, in schizophrenia, some positive and negative symptoms may be linked with multisensory integration impairment. Interestingly, Williams et al. (2010) yielded similar results using a behavioral paradigm. Concerning hallucinations, connectivity disruption between RN26 and RN34 seemed more pronounced in the AVH group. As stated above, RN26 encompass several associative regions: medial temporal cortex, superior temporal cortex and insula. Several connectivity studies stressed the involvement of these regions in the pathophysiology of hallucinations (Amad et al., 2014; Hare et al., 2016; Lefebvre et al., 2016; Rolland et al., 2015). More specifically, hippocampal cortex seems to be specifically involved in visual hallucinations (Amad et al., 2014). In this seed-based study, the authors reported decreased connectivity between the hippocampal cortex, on the one hand, the left lenticular nucleus, the right thalamus and the superior temporal gyri on the other hand in AVH patients. RN26 encompass (at least a part of) these three regions. Interestingly, in our study WNFC of RN26 was reduced in PS, but there was no effect of PANSS-P score or hallucinatory status on this FC variable. Moreover, Amad et al. (2014) reported increased connectivity between the hippocampus, the medial prefrontal cortex and the caudate nuclei in AVH patients. This discrepancy with our results may be explained by different analysis techniques: the FC of a large scale network like RN26 can’t only reflect the FC pattern of one of its subregions, the hippocampus, like a seed-based analysis can do. 24
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
multiscale hierarchical functional organization. J. Neurophysiol. 105, 2753–2763. http://dx.doi.org/10.1152/jn.00895.2010. Filippini, N., MacIntosh, B.J., Hough, M.G., Goodwin, G.M., Frisoni, G.B., Smith, S.M., Matthews, P.M., Beckmann, C.F., Mackay, C.E., 2009. Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc. Natl. Acad. Sci. USA 106, 7209–7214. http://dx.doi.org/10.1073/pnas.0811879106. Golland, Y., Golland, P., Bentin, S., Malach, R., 2008. Data-driven clustering reveals a fundamental subdivision of the human cortex into two global systems. Neuropsychologia 46, 540–553. http://dx.doi.org/10.1016/j.neuropsychologia. 2007.10.003. Hare, S.M., Ford, J.M., Ahmadi, A., Damaraju, E., Belger, A., Bustillo, J., Lee, H.J., Mathalon, D.H., Mueller, B.A., Preda, A., van Erp, T.G.M., Potkin, S.G., Calhoun, V. D., Turner, J.A., Functional Imaging Biomedical Informatics Research Network, 2016. Modality-Dependent Impact of Hallucinations on Low-Frequency Fluctuations in Schizophrenia. Schizophr. Bull. 〈http://dx.doi.org/10.1093/schbul/sbw093〉. Hasenkamp, W., James, G.A., Boshoven, W., Duncan, E., 2011. Altered engagement of attention and default networks during target detection in schizophrenia. Schizophr. Res. 125, 169–173. http://dx.doi.org/10.1016/j.schres.2010.08.041. Jafri, M.J., Pearlson, G.D., Stevens, M., Calhoun, V.D., 2008. A method for functional network connectivity among spatially independent resting-state components in schizophrenia. NeuroImage 39, 1666–1681. http://dx.doi.org/10.1016/j. neuroimage.2007.11.001. Jardri, R., Thomas, P., Delmaire, C., Delion, P., Pins, D., 2013. The neurodynamic organization of modality-dependent hallucinations. Cereb. Cortex 1991 (23), 1108–1117. http://dx.doi.org/10.1093/cercor/bhs082. Javitt, D.C., 2009. When doors of perception close: bottom-up models of disrupted cognition in schizophrenia. Annu. Rev. Clin. Psychol. 5, 249–275. http://dx.doi.org/ 10.1146/annurev.clinpsy.032408.153502. Javitt, D.C., Sweet, R.A., 2015. Auditory dysfunction in schizophrenia: integrating clinical and basic features. Nat. Rev. Neurosci. 16, 535–550. http://dx.doi.org/10.1038/ nrn4002. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 13, 261–276. Khadka, S., Meda, S.A., Stevens, M.C., Glahn, D.C., Calhoun, V.D., Sweeney, J.A., Tamminga, C.A., Keshavan, M.S., O’Neil, K., Schretlen, D., Pearlson, G.D., 2013. Is aberrant functional connectivity a psychosis endophenotype? A resting state functional magnetic resonance imaging study. Biol. Psychiatry 74, 458–466. http:// dx.doi.org/10.1016/j.biopsych.2013.04.024. Lefebvre, S., Demeulemeester, M., Leroy, A., Delmaire, C., Lopes, R., Pins, D., Thomas, P., Jardri, R., 2016. Network dynamics during the different stages of hallucinations in schizophrenia. Hum. Brain Mapp. 37, 2571–2586. http://dx.doi.org/10.1002/hbm. 23197. Li, H.-J., Xu, Y., Zhang, K.-R., Hoptman, M.J., Zuo, X.-N., 2014. Homotopic connectivity in drug-naïve, first-episode, early-onset schizophrenia. J. Child Psychol. Psychiatry. http://dx.doi.org/10.1111/jcpp.12307. Light, G.A., Swerdlow, N.R., Braff, D.L., 2007. Preattentive sensory processing as indexed by the MMN and P3a brain responses is associated with cognitive and psychosocial functioning in healthy adults. J. Cogn. Neurosci. 19, 1624–1632. http://dx.doi.org/ 10.1162/jocn.2007.19.10.1624. Mamah, D., Barch, D.M., Repovš, G., 2013. Resting state functional connectivity of five neural networks in bipolar disorder and schizophrenia. J. Affect. Disord. 150, 601–609. http://dx.doi.org/10.1016/j.jad.2013.01.051. Martínez, A., Hillyard, S.A., Dias, E.C., Hagler, D.J., Butler, P.D., Guilfoyle, D.N., Jalbrzikowski, M., Silipo, G., Javitt, D.C., 2008. Magnocellular pathway impairment in schizophrenia: evidence from functional magnetic resonance imaging. J. Neurosci. 28, 7492–7500. http://dx.doi.org/10.1523/JNEUROSCI.1852-08.2008. Mazoyer, B., Mellet, E., Perchey, G., Zago, L., Crivello, F., Jobard, G., Delcroix, N., Vigneau, M., Leroux, G., Petit, L., Joliot, M., Tzourio-Mazoyer, N., 2016. BIL & GIN: a neuroimaging, cognitive, behavioral, and genetic database for the study of human brain lateralization. NeuroImage 124, 1225–1231. http://dx.doi.org/10.1016/j. neuroimage.2015.02.071. Meda, S.A., Gill, A., Stevens, M.C., Lorenzoni, R.P., Glahn, D.C., Calhoun, V.D., Sweeney, J.A., Tamminga, C.A., Keshavan, M.S., Thaker, G., Pearlson, G.D., 2012. Differences in resting-state functional magnetic resonance imaging functional network connectivity between schizophrenia and psychotic bipolar probands and their unaffected first-degree relatives. Biol. Psychiatry 71, 881–889. http://dx.doi.org/10. 1016/j.biopsych.2012.01.025. Mellet, E., Tzourio-Mazoyer, N., Bricogne, S., Mazoyer, B., Kosslyn, S.M., Denis, M., 2000. Functional anatomy of high-resolution visual mental imagery. J. Cogn. Neurosci. 12, 98–109. Menon, V., 2011. Large-scale brain networks and psychopathology: a unifying triple network model. Trends Cogn. Sci. 15, 483–506. http://dx.doi.org/10.1016/j.tics. 2011.08.003. Naveau, M., Doucet, G., Delcroix, N., Petit, L., Zago, L., Crivello, F., Jobard, G., Mellet, E., Tzourio-Mazoyer, N., Mazoyer, B., Joliot, M., 2012. A novel group ICA approach based on multi-scale individual component clustering. Application to a large sample of fMRI data. Neuroinformatics 10, 269–285. http://dx.doi.org/10.1007/s12021012-9145-2. Orliac, F., Naveau, M., Joliot, M., Delcroix, N., Razafimandimby, A., Brazo, P., Dollfus, S., Delamillieure, P., 2013. Links among resting-state default-mode network, salience network, and symptomatology in schizophrenia. Schizophr. Res. 148, 74–80. http:// dx.doi.org/10.1016/j.schres.2013.05.007. Palaniyappan, L., Al-Radaideh, A., Mougin, O., Gowland, P., Liddle, P.F., 2013. Combined white matter imaging suggests myelination defects in visual processing regions in schizophrenia. Neuropsychopharmacology 38, 1808–1815. http://dx.doi.org/10. 1038/npp.2013.80.
Finally, we must consider the fact that patients were on antipsychotic medication, which may have an effect on FC (Cole et al., 2013). However, Li et al. (2014) reported FC disruptions in drug-naïve schizophrenia patients, and Sarpal et al. (2015) reported an increase in cortico-subcortical connectivity associated with antipsychotic treatment. According to the latter, FC disruption in schizophrenia may be a state-dependent phenomenon correlated with the severity of the symptoms. In our study, the FC of two edges, both linked to RN26, seemed strongly impacted by medication dosage. In conclusion, this study highlighted within-network and betweennetwork FC disruptions in patients with schizophrenia, with a clear trend toward weaker FC. Our exploratory approach allowed us to point out the fact that sensory networks appeared more functionally altered than high-level cognitive networks. These alterations were correlated with negative symptoms, positive symptoms and hallucinations, indicating abnormalities in visual processing and crossmodal binding in schizophrenia. Moreover, an anomalous synchronization between a visual network and the precuneus was linked with delusions and hallucinations. Altogether, our results support the assumption that some schizophrenia symptoms may be linked with low-order sensory alterations impacting reverberating on higher-order cognitive processes, i.e. the “bottom-up” hypothesis of schizophrenia symptoms. Acknowledgment The authors thank the members of the Cyceron imaging platform for their assistance. This work was supported by the French Health and Research Ministries in a Programme Hospitalier de Recherche Clinique (n° 2003-R06-10). The funding source had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.pscychresns.2017.04. 003. References Abbott, C.C., Kim, D., Sponheim, S.R., Bustillo, J., Calhoun, V.D., 2010. Decreased default mode neural modulation with age in schizophrenia. Am. J. Geriatr. Psychiatry 18, 897–907. http://dx.doi.org/10.1097/JGP.0b013e3181e9b9d9. Amad, A., Cachia, A., Gorwood, P., Pins, D., Delmaire, C., Rolland, B., Mondino, M., Thomas, P., Jardri, R., 2014. The multimodal connectivity of the hippocampal complex in auditory and visual hallucinations. Mol. Psychiatry 19, 184–191. http:// dx.doi.org/10.1038/mp.2012.181. Arbabshirani, M.R., Kiehl, K.A., Pearlson, G.D., Calhoun, V.D., 2013. Classification of schizophrenia patients based on resting-state functional network connectivity. Front. Neurosci. 7, 133. http://dx.doi.org/10.3389/fnins.2013.00133. Beckmann, C.F., Smith, S.M., 2004. Probabilistic independent component analysis for functional magnetic resonance imaging. IEEE Trans. Med. Imaging 23, 137–152. http://dx.doi.org/10.1109/TMI.2003.822821. Bressler, S.L., Menon, V., 2010. Large-scale brain networks in cognition: emerging methods and principles. Trends Cogn. Sci. 14, 277–290. http://dx.doi.org/10.1016/j. tics.2010.04.004. Bushara, K.O., Hanakawa, T., Immisch, I., Toma, K., Kansaku, K., Hallett, M., 2003. Neural correlates of cross-modal binding. Nat. Neurosci. 6, 190–195. http://dx.doi. org/10.1038/nn993. Cole, D.M., Oei, N.Y.L., Soeter, R.P., Both, S., van Gerven, J.M.A., Rombouts, S.A.R.B., Beckmann, C.F., 2013. Dopamine-dependent architecture of cortico-subcortical network connectivity. Cereb. Cortex 1991 (23), 1509–1516. http://dx.doi.org/10. 1093/cercor/bhs136. Coleman, M.J., Cestnick, L., Krastoshevsky, O., Krause, V., Huang, Z., Mendell, N.R., Levy, D.L., 2009. Schizophrenia patients show deficits in shifts of attention to different levels of global-local stimuli: evidence for magnocellular dysfunction. Schizophr. Bull. 35, 1108–1116. http://dx.doi.org/10.1093/schbul/sbp090. Doniger, G.M., Foxe, J.J., Murray, M.M., Higgins, B.A., Javitt, D.C., 2002. Impaired visual object recognition and dorsal/ventral stream interaction in schizophrenia. Arch. Gen. Psychiatry 59, 1011–1020. Doucet, G., Naveau, M., Petit, L., Delcroix, N., Zago, L., Crivello, F., Jobard, G., TzourioMazoyer, N., Mazoyer, B., Mellet, E., Joliot, M., 2011. Brain activity at rest: a
25
Psychiatry Research: Neuroimaging 266 (2017) 19–26
F. Orliac et al.
WNR.0000000000000622. van den Heuvel, M.P., Fornito, A., 2014. Brain networks in schizophrenia. Neuropsychol. Rev. 24, 32–48. http://dx.doi.org/10.1007/s11065-014-9248-7. Whitfield-Gabrieli, S., Thermenos, H.W., Milanovic, S., Tsuang, M.T., Faraone, S.V., McCarley, R.W., Shenton, M.E., Green, A.I., Nieto-Castanon, A., LaViolette, P., Wojcik, J., Gabrieli, J.D.E., Seidman, L.J., 2009. Hyperactivity and hyperconnectivity of the default network in schizophrenia and in first-degree relatives of persons with schizophrenia. Proc. Natl. Acad. Sci. USA 106, 1279–1284. http://dx.doi.org/10. 1073/pnas.0809141106. Williams, L.E., Light, G.A., Braff, D.L., Ramachandran, V.S., 2010. Reduced multisensory integration in patients with schizophrenia on a target detection task. Neuropsychologia 48, 3128–3136. http://dx.doi.org/10.1016/j.neuropsychologia. 2010.06.028. Wolf, N.D., Sambataro, F., Vasic, N., Frasch, K., Schmid, M., Schönfeldt-Lecuona, C., Thomann, P.A., Wolf, R.C., 2011. Dysconnectivity of multiple resting-state networks in patients with schizophrenia who have persistent auditory verbal hallucinations. J. Psychiatry Neurosci. 36, 366–374. http://dx.doi.org/10.1503/jpn.110008. Woodward, N.D., Rogers, B., Heckers, S., 2011. Functional resting-state networks are differentially affected in schizophrenia. Schizophr. Res. 130, 86–93. http://dx.doi. org/10.1016/j.schres.2011.03.010. Wotruba, D., Michels, L., Buechler, R., Metzler, S., Theodoridou, A., Gerstenberg, M., Walitza, S., Kollias, S., Rössler, W., Heekeren, K., 2014. Aberrant coupling within and across the default mode, task-positive, and salience network in subjects at risk for psychosis. Schizophr. Bull. 40, 1095–1104. http://dx.doi.org/10.1093/schbul/ sbt161. Yu, Q., Plis, S.M., Erhardt, E.B., Allen, E.A., Sui, J., Kiehl, K.A., Pearlson, G., Calhoun, V.D., 2012. Modular organization of functional network connectivity in healthy controls and patients with schizophrenia during the resting state. Front. Syst. Neurosci. 5. http://dx.doi.org/10.3389/fnsys.2011.00103. Zalesky, A., Fornito, A., Bullmore, E.T., 2010. Network-based statistic: identifying differences in brain networks. NeuroImage 53, 1197–1207. http://dx.doi.org/10. 1016/j.neuroimage.2010.06.041. Zhang, S., Li, C.R., 2012. Functional connectivity mapping of the human precuneus by resting state fMRI. NeuroImage 59, 3548–3562. http://dx.doi.org/10.1016/j. neuroimage.2011.11.023.
Papo, D., 2013. Why should cognitive neuroscientists study the brain's resting state? Front. Hum. Neurosci. 7, 45. http://dx.doi.org/10.3389/fnhum.2013.00045. Power, J.D., Barnes, K.A., Snyder, A.Z., Schlaggar, B.L., Petersen, S.E., 2012. Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. Neuroimage 59, 2142–2154. http://dx.doi.org/10.1016/j.neuroimage.2011. 10.018. Rogers, B.P., Morgan, V.L., Newton, A.T., Gore, J.C., 2007. Assessing functional connectivity in the human brain by fMRI. Magn. Reson. Imaging 25, 1347–1357. http://dx.doi.org/10.1016/j.mri.2007.03.007. Rolland, B., Amad, A., Poulet, E., Bordet, R., Vignaud, A., Bation, R., Delmaire, C., Thomas, P., Cottencin, O., Jardri, R., 2015. Resting-state functional connectivity of the nucleus accumbens in auditory and visual hallucinations in schizophrenia. Schizophr. Bull. 41, 291–299. http://dx.doi.org/10.1093/schbul/sbu097. Rotarska-Jagiela, A., van de Ven, V., Oertel-Knöchel, V., Uhlhaas, P.J., Vogeley, K., Linden, D.E.J., 2010. Resting-state functional network correlates of psychotic symptoms in schizophrenia. Schizophr. Res. 117, 21–30. http://dx.doi.org/10.1016/ j.schres.2010.01.001. Sarpal, D.K., Robinson, D.G., Lencz, T., Argyelan, M., Ikuta, T., Karlsgodt, K., Gallego, J.A., Kane, J.M., Szeszko, P.R., Malhotra, A.K., 2015. Antipsychotic treatment and functional connectivity of the striatum in first-episode schizophrenia. JAMA Psychiatry 72, 5–13. http://dx.doi.org/10.1001/jamapsychiatry.2014.1734. Satterthwaite, T.D., Wolf, D.H., Loughead, J., Ruparel, K., Elliott, M.A., Hakon, H., Gur, R.C., Gur, R.E., 2012. Impact of in-scanner head motion on multiple measures of functional connectivity: relevance for studies of neurodevelopment in youth. NeuroImage 60, 623–632. http://dx.doi.org/10.1016/j.neuroimage.2011.12.063. Senkowski, D., Schneider, T.R., Foxe, J.J., Engel, A.K., 2008. Crossmodal binding through neural coherence: implications for multisensory processing. Trends Neurosci. 31, 401–409. http://dx.doi.org/10.1016/j.tins.2008.05.002. Smith, S.M., Fox, P.T., Miller, K.L., Glahn, D.C., Fox, P.M., Mackay, C.E., Filippini, N., Watkins, K.E., Toro, R., Laird, A.R., Beckmann, C.F., 2009. Correspondence of the brain's functional architecture during activation and rest. Proc. Natl. Acad. Sci. USA 106, 13040–13045. http://dx.doi.org/10.1073/pnas.0905267106. Su, J., Shen, H., Zeng, L.-L., Qin, J., Liu, Z., Hu, D., 2016. Heredity characteristics of schizophrenia shown by dynamic functional connectivity analysis of resting-state functional MRI scans of unaffected siblings. Neuroreport. http://dx.doi.org/10.1097/
26