- Email: [email protected]
Decreased middle temporal gyrus connectivity in the language network in schizophrenia patients with auditory verbal hallucinations
Neuroscience Letters 653 (2017) 177–182
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Research article
Decreased middle temporal gyrus connectivity in the language network in schizophrenia patients with auditory verbal hallucinations Linchuan Zhang a,1 , Baojuan Li a,1 , Huaning Wang b , Liang Li a , Qimei Liao a , Yang Liu a , Xianghong Bao c , Wenlei Liu a , Hong Yin d , Hongbing Lu a,∗ , Qingrong Tan b,∗∗ a
Department of Biomedical Engineering, Fourth Military Medical University, Xi’an, Shaanxi 710032, PR China Department of Psychiatry, Xijing Hospital, Fourth Military Medical University, Xi’an, Shaanxi 710032, PR China c Department of Health Service, School of Military Preventive Medicine, Fourth Military Medical University, Xi’an, Shaanxi 710032, PR China d Department of Radiology, Xijing Hospital, Fourth Military Medical University, Xi’an, Shaanxi, PR China b
h i g h l i g h t s • Impaired effective connectivity within the language network in schizophrenia patients with AVHs was studied. • The connection from the LIFG to LMTG was significantly decreased in patients with AVHs compared to those without AVHs. • The effective connection from the LIPL to LMTG was significantly decreased compared to the healthy controls.
a r t i c l e
i n f o
Article history: Received 14 October 2016 Received in revised form 9 May 2017 Accepted 21 May 2017 Available online 29 May 2017 Keywords: Schizophrenia Auditory verbal hallucinations Dynamic causal modeling Effective connectivity Language network
a b s t r a c t As the most common symptoms of schizophrenia, the long-term persistence of obstinate auditory verbal hallucinations (AVHs) brings about great mental pain to patients. Neuroimaging studies of schizophrenia have indicated that AVHs were associated with altered functional and structural connectivity within the language network. However, effective connectivity that could reflect directed information flow within this network and is of great importance to understand the neural mechanisms of the disorder remains largely unknown. In this study, we utilized stochastic dynamic causal modeling (DCM) to investigate directed connections within the language network in schizophrenia patients with and without AVHs. Thirty-six patients with schizophrenia (18 with AVHs and 18 without AVHs), and 37 healthy controls participated in the current resting-state functional magnetic resonance imaging (fMRI) study. The results showed that the connection from the left inferior frontal gyrus (LIFG) to left middle temporal gyrus (LMTG) was significantly decreased in patients with AVHs compared to those without AVHs. Meanwhile, the effective connection from the left inferior parietal lobule (LIPL) to LMTG was significantly decreased compared to the healthy controls. Our findings suggest aberrant pattern of causal interactions within the language network in patients with AVHs, indicating that the hypoconnectivity or disrupted connection from frontal to temporal speech areas might be critical for the pathological basis of AVHs. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Schizophrenia (SZ) is a serious mental disorder that associates with abnormalities of brain structure and functions [1–3]. As important indicators to evaluate the severity and clinical efficacy of
∗ Corresponding author at: Department of Biomedical Engineering, Fourth Military Medical University, 169 Changlexi Rd., Xi’an, Shaanxi 710032, China. ∗∗ Corresponding author at: Department of Psychiatry, Xijing Hospital, 145 Changlexi Rd., Xi’an, Shaanxi 710032, China. E-mail addresses: [email protected] (H. Lu), [email protected] (Q. Tan). 1 These authors contributed equally. http://dx.doi.org/10.1016/j.neulet.2017.05.042 0304-3940/© 2017 Elsevier B.V. All rights reserved.
schizophrenia, auditory verbal hallucinations (AVHs) are the most common symptoms of SZ, with an incidence of 60%–80% [4]. SZ is a disorder with unknown etiology and pathophysiology that are further complicated than regional brain abnormalities. Previous neuroimaging studies on AVHs have revealed structural and functional alterations in the language network. A diffusion tensor imaging (DTI) study revealed reduced integrity of the fiber connecting frontal and temporo-parietal language regions [5]. It was further supported by neuroimaging studies revealing that AVHs in SZ was associated with dysconnectivity in the language network [6]. A study of fMRI showed that the emergence of AVHs was the result of mild disruption of function connectivity along the
178
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182
fronto-temporal pathway [7]. Reduced fronto-temporal functional connectivity has also been demonstrated in SZ patients with AVHs during sentence-completion task [8], external speech processing [9], and speaking aloud task [10]. A resting state fMRI study showed that disrupted intrinsic connectivity of the language network could underlie persistent AVHs in patients with SZ [11]. Although the above studies have indicated that the emergence of AVHs is associated with altered structural and functional connectivity within the language network, it still remains unclear how these regions interact with each other to cause the symptoms like AVHs in SZ. As a powerful tool for effective connectivity (describes the causal influences that one region exerts over another [12]) analysis, dynamic causal modeling (DCM) [13] not only enable us to quantify the strength of the connections, but also demonstrate directed information flow from one region to another, therefore provide more information to recover the underlying neural substrates of AVHs and SZ. In fact, there have been some attempts to study abnormal effective connectivity in SZ patients using DCM ´ cic-Blake ´ [9,14,15]. Curˇ et al. found that increased effective connectivity from the posterior cingulate cortex to the ventromedial prefrontal cortex was associated with poorer insight [16]. In addition, abnormal effective connectivity of the posterior cingulate and medial prefrontal cortices have also been related to working memory impairments [17]. Interestingly, measures of effective connectivity have been used to explain differences in the perception of the hollow-mask illusion [18]. However, it still remains unknown how regions in the language network interact with each other abnormally in SZ patients with AVHs. In this study, we aimed to investigate resting-state effective connectivity within the language network in 18 SZ patients with AVHs, 18 patients without AVHs and 37 healthy controls using DCM. Three regions including the left inferior frontal gyrus (LIFG), left middle temporal gyrus (LMTG), and left inferior parietal lobule (LIPL) were selected as regions of interest (ROIs) according to previous findings [19,20]. We hypothesized that the effective connectivity within the language network would be altered in SZ patients with AVHs. 2. Methods 2.1. Subjects This study was approved by the Institutional Board of The Fourth Military Medical University. All participants received a full comprehensive description of the MRI study and gave voluntarily written informed consent before entering the study. A total of 73 subjects were recruited (36 patients with schizophrenia and 37 healthy controls). All patients fulfilled the DSM-IV [21] diagnostic criteria for schizophrenia and was further divided into the AVH(18 subjects) and nAVH group(18 subjects) based on the scores of the Positive And Negative Syndrome Scale (PANSS) [22], as well as detailed symptomatology information acquired through patient interview and examination of patient’s medical records including the severity
of disorganization symptoms, excitement and emotional distress. The AVH and nAVH groups were further characterized in terms of the Auditory Hallucination Rating Scale (AHRS) [23]. For healthy controls, exclusion criteria included current or past history of psychiatric illness and the presence of psychosis in first-degree relatives. Demographic and clinical characteristics of all subjects were given in Table 1. All three groups were matched for age, gender, and education. 2.2. fMRI data acquisition All MRI data were acquired on a 3.0 T MR scanner (MAGNETOM Trio, Siemens AG, Erlangen, Germany) at the Radiology Department of Xijing Hospital. The fMRI acquisition parameters were as follows: repetition time (TR) = 2000 ms; echo time (TE) = 30 ms; flip angle = 90◦ ; field-of-view = 220 mm × 220 mm; 33 slices; slice thickness = 4 mm; interslice gap = 0.6 mm. We collected 240 images for each subject. 2.3. fMRI data preprocessing Preprocessing of the fMRI data was performed using SPM8 software (http://www.fil.ion.ucl.ac.uk/spm). First, a rigid body transformation was used to realign the functional images to the first image of each session to correct for head motion. Then images were normalized to a MNI EPI template. Finally, fMRI images were spatially smoothed using an 8-mm full-width-at-half-maximum (FWHM) Gaussian kernel. 2.4. General linear model After data preprocessing, resting-state fMRI images were modelled using a general linear model with the following regressors: (1) the six rigid body realignment parameters to model head motion that were then used to remove effects of head motion when extracting time courses of the ROIs; (2) one constant regressor to model the baseline; and (3) cosine basis functions with frequencies ranging from 0.0078 (1/128) to model possible signal drift. The cosine basis functions used in the GLM serve as a high-pass filter which filtered out low-frequency signal drift in the current study. 2.5. Regions-of-interest The following three regions were used: the LIFG, LMTG, and LIPL which are the brain regions generally involved in the language network. LMTG is associated with cognitive self-monitoring [24] and memory retrieval [25]. LIFG has been associated with the preparation of overt speech and the generation of inner speech [26,27]. LIPL is considered as the storage of long-term memory mostly in verbal modality [26]. For each of the ROIs, a mask representing this region was created using the Anatomical Automatic Labeling (AAL) atlas [28]
Table 1 Demographic and clinical characteristics.
Age Sex(male/female) Handedness (right/left) Education(years) Duration of illness(months) PANSS Total Score AHRS Score PANSS: Positive and Negative Syndrome Scale. AHRS: Auditory Hallucination Rating Scale.
AVH Patients n = 18
nAVH Patients n = 18
Healthy Controls n = 37
22.56 ± 6.73 10/8 18/0 12.44 ± 2.30 5.94 ± 5.93 106.44 ± 13.55 26.22 ± 8.10
22.67 ± 3.85 9/9 18/0 12.56 ± 2.19 12.44 ± 18.16 88.06 ± 23.90
22.48 ± 5.84 19/18 37/0 12.43 ± 3.13
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182
179
Fig. 1. Significant effective connectivity of the language network (one-sample t-test, p < 0.05, Bonferroni corrected for multiple comparisons). (A) three ROIs: LIFG, LMTG, LIPL within the language network; (B) effective connectivity in AVH group; (C) effective connectivity in the nAVH group; (D) effective connectivity in control group.
provided in the WFU (Wake Forest University) PickAtlas toolbox (http://www.nitrc.org/projects/wfu pickatlas/). These masks were then combined with a thresholded SPM (with the threshold set to 0.05) testing for the baseline generated by first-level (within subject) analyses to constrain the extraction of subject-specific ROI time courses. For each ROI, the fMRI time courses of all the voxels that located within the corresponding ROI mask and showed activation in the thresholded SPM map were extracted. The first principle component of these time courses was finally used to summarize the BOLD response of the ROI. The effects of head motion were also removed from the time courses. 2.6. Dynamic causal modeling 2.6.1. Specification and inversion of the fully-connected model Effective connectivity analysis was conducted using DCM10. Compared to traditional models, DCM integrates the BOLD physiological response model with the functional activation model [29]. It uses differential equations to fit both hemodynamic response and neural dynamics to make the model more accurately [30]. The causal interactions among different brain regions are modelled at neuronal level using neuronal state equations: z˙ ≈ Az +
uj Bj z + Cu
(1)
Here, z denotes the neuronal activities of the ROIs. The N-by-N (N is the number of the ROIs) matrix A represents the effective connectivity from one region to another in the absence of external inputs. Matrices B and C describe changes in the effective connectivity and regional BOLD responses caused by experimental inputs. The neuronal state equations are supplemented by an observation model
which maps neuronal states to observed fMRI signals. Recently, the deterministic DCM described in Eq. (1) has been extended to accommodate state noises [31,32]: z˙ ≈ Az +
uj Bj z + C v + ωx
(2)
Here, x denotes the state noise which models stochastic innovations in the system [32]. This stochastic DCM which is able to model resting-state fMRI data was employed in the current study [33]. After extracting time courses for all three ROIs, we first specified a fully-connected DCM model for each subject. This model hypothesized that any pair of ROIs were reciprocally connected. The fully-connected DCM model for each subject was then inverted using the generalised filtering scheme which provided parameter estimates as well as model evidence of this fully-connected model [34]. The model evidence p(y|m) denotes the probability of obtaining the data y given a model m and is usually used to compare competing hypotheses. 2.6.2. Network discovery The optimal model that can best explain the data may not be the fully-connected model but a reduced one. A reduced model has the same set of ROIs with the fully-connected model while some of the connections presented in the fully-connected model are absent in the reduced one. We have 3 ROIs, the fully-connected model thus includes 9 connections (3 self-connections and 6 inter-regional connections). Any subset of these 9 connections may be absent in the reduced model, in that case, there would be 512 (29 ) possible reduced models. One way to find the optimal model is to specify and invert all the possible models and then use Bayesian model selection. This strategy requires the inversion of hundreds of mod-
180
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182
Fig. 2. Different effective connectivity between groups (post-hoc t-tests, p = 0.05). (A) AVHS < Control; (B) AVHS < Non-AVHS; (C) Non-AVHS < Control (D) Non-AVHS > Control.
els which is quite time-consuming. Thus in the current study, we adopted a network discovery scheme [35]. Instead of specifying and inverting all the possible models, we only needed to invert a single model (the fully-connected model). Provided with the parameter estimates and model evidence of this model, the network discovery scheme searched over all possible reduced models and recover the optimal model that has the highest model evidence. This scheme is quite efficient and has been showed to be more robust and accurate compared with the first strategy [36]. Finally, the group average model was obtained through Bayesian parametric averaging (BPA). For each reduced model, the model evidences over all the subjects were pooled. All possible models were scored according to their model evidences and the one with the highest evidence was recovered as the optimal model. 2.6.3. Statistical analysis Subject specific parameter estimates of the optimal model were taken to the between-subject level analysis using a classical random effects analysis (t-tests). To examine group differences in effective connectivity, a one way ANOVA was first performed and then followed up with post-hoc t-tests. 3. Results Fig. 1 shows the effective connectivity of the language network in the AVH group, nAVH group and controls, respectively (one-sample t-test at p < 0.05, Bonferroni corrected for multiple comparisons). One way ANOVA (p < 0.05) showed that the effective connection from LIFG to LMTG (p = 0.000) and the effective connection from LIPL to LMTG (p = 0.000) were significantly different among the three groups. Fig. 2 illustrates changes in effective connectivity in the patients group (post-hoc t-tests, p < 0.05). The results showed that: (1) com-
pared to healthy controls, the effective connection from LIPL to LMTG was significantly decreased in AVH group (p = 0.000); (2) compared to patients without AVHs, the effective connection from LIFG to LMTG was significantly decreased in AVH group (p = 0.000); and (3) patients without AVHs showed increased effective connection from LIFG to LMTG and decreased effective connection from LIPL to LMTG compared to healthy controls (p = 0.000). 4. Discussion In this study, we observed decreased effective connection from LIPL to LMTG in both patient groups compared to healthy controls. These findings provide a piece of evidence for the “dysconnection hypothesis” of SZ. On the one hand, decreased effective connectivity found in the current study provides further support for previous neuroimaging studies that have reported impaired temporo-parietal functional and structural connectivity. On the other hand, the directionality of the connections that DCM furnishes helps to elucidate the exact mechanisms underlying the pathophysiology of SZ and AVHs. Functional connectivity studies can only tell us that the correlation between the activity of the temporal and parietal regions is impaired in the patients. The results from the current DCM study further revealed that the impairments may be mediated through a lack of causal influences from the LIPL to the LMTG. Considering that patients without AVHs also showed decreased connection form LIPL to LMTG compared to healthy controls, attenuation of this connection may serve as a biomarker of SZ rather than AVHs. We also found that the AVH group showed significantly decreased effective connection from LIFG to LMTG compared to patients without AVHs. Although the presence of AVHs is one of the defining and critical symptoms for patients with schizophrenia, not all patients experience AVHs. Thus the differences in the
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182
effective connectivity between the two groups of patients may shed light on the neural substrates of AVHs. AVHs were considered as the disability of patients to recognize self-generated inner speech [37]. The failure to integrate information regarding inner speech was attributed to verbal generation in the frontal cortex and sensory processing in temporal and parietal regions [38,39]. As for normal subjects, the language pathways have been proposed to diverge into two streams: a ventral stream projects towards MTG which is associated with mapping sound onto meaning, as well as a dorsal stream projects towards posterior Sylvian fissure and then frontal regions which is associated with mapping sound onto articulatory-based representations [40]. In terms of AVHs, impaired self-monitoring has been proposed to be associated with the mechanism underlying this symptom [41]. According to this theory, decreased effective connection in patients with AVHs from LIFG to LMTG found in our study indicated a lack of efficient inhibition of the frontal lobe on the internal speech signal generated by the temporal lobe. Weakening inhibition would then lead to the occurrences of AVHs. Thus the decreased effective connection from LIFG to LMTG may lead to turning inner speeches to verbal hallucinations abnormally. fMRI studies showed that those areas played a central role in the language system of AVHs [42,43]. The anatomic connection between frontal and temporal regions is anterior indirect segment in the left arcuate fasciculus (AF), which is a key anatomical connection between the frontal and temporal speech areas [44–46]. A DTI study on patients with AVHs found disruptions of white matter integrity in the anterior indirect segment of AF [45]. Considering that patients without AVHs showed increased effective connection form LIFG to LMTG compared to healthy controls, our results not only support the important role of the fronto-temporal connection for the presence of AVHs, they further imply that hypoconnectivity or disrupted connection from frontal speech areas to temporal speech areas might be critical for the pathological basis of AVHs. It is important to note that the LIFG, LIPL and LMTG were relatively large (covered 27, 20 and 37 cm3 , respectively) in the current study as the ROIs were defined according to the AAL template. In that case, some regions like the LIPL may contain different subregions that each has different role in language processing which may complicate the interpretation of the results. In order to solve this problem, we may try to obtain more specific ROI locations by also collecting fMRI images while the subjects are performing language processing tasks in the future. In addition, as the fast and optimal spectral DCM scheme [47] and the new Bayesian model reduction (BMR) tool [36] are available now, we may also consider adopting these tools in future studies to further improve the efficiency and robustness of the calculations. Competing interests The authors declare that they have no competing interests. Authors’ contributions HW, HL and QT designed the experiment; HW, QT and HY recruited the subjects and collected the fMRI images; LZ, BL, LL, WL and YL analyzed the data; LZ, BL, QL, XB and HL prepared the manuscript. Acknowledgements This work was supported by the National Natural Science Foundation of China [81301199, 81630032, 81230035 and 81071220]. We thank anonymous reviewers for guidance in clarifying and elaborating this report.
181
References [1] J.W. Buckholtz, A. Meyer-Lindenberg, Psychopathology and the human connectome: toward a transdiagnostic model of risk for mental illness, Neuron 74 (2012) 990–1004. [2] K.J. Friston, Functional and effective connectivity in neuroimaging:a synthesis, Hum. Brain Mapp. 2 (1994). [3] K.E. Stephan, T. Baldeweg, K.J. Friston, Synaptic plasticity and dysconnection in schizophrenia, Biol. Psychiatry 59 (2006) 929–939. [4] N.C. Andreasen, M. Flaum, Schizophrenia: the characteristic symptoms, Schizophr. Bull. 17 (1991) 27–49. [5] A.D. de Weijer, R.C. Mandl, K.M. Diederen, S.F. Neggers, R.S. Kahn, H.E. Hulshoff Pol, I.E. Sommer, Microstructural alterations of the arcuate fasciculus in schizophrenia patients with frequent auditory verbal hallucinations, Schizophr. Res. 130 (2011) 68–77. [6] P. Allen, G. Modinos, D. Hubl, G. Shields, A. Cachia, R. Jardri, P. Thomas, T. Woodward, P. Shotbolt, M. Plaze, R. Hoffman, Neuroimaging auditory hallucinations in schizophrenia: from neuroanatomy to neurochemistry and beyond, Schizophr. Bull. 38 (2012) 695–703. [7] S. Benetti, W. Pettersson-Yeo, P. Allen, M. Catani, S. Williams, A. Barsaglini, L.M. Kambeitz-Ilankovic, P. McGuire, A. Mechelli, Auditory verbal hallucinations and brain dysconnectivity in the perisylvian language network: a multimodal investigation, Schizophr. Bull. 41 (2015) 192–200. [8] S.M. Lawrie, C. Buechel, H.C. Whalley, C.D. Frith, K.J. Friston, E.C. Johnstone, Reduced frontotemporal functional connectivity in schizophrenia associated with auditory hallucinations, Biol. Psychiatry 51 (2002) 1008–1011. [9] A. Mechelli, P. Allen, E. Amaro Jr., C.H. Fu, S.C. Williams, M.J. Brammer, L.C. Johns, P.K. McGuire, Misattribution of speech and impaired connectivity in patients with auditory verbal hallucinations, Hum. Brain Mapp. 28 (2007) 1213–1222. [10] J.M. Ford, D.H. Mathalon, S. Whitfield, W.O. Faustman, W.T. Roth, Reduced communication between frontal and temporal lobes during talking in schizophrenia, Biol. Psychiatry 51 (2002) 485–492. [11] N.D. Wolf, F. Sambataro, N. Vasic, K. Frasch, M. Schmid, C. Schonfeldt-Lecuona, P.A. Thomann, R.C. Wolf, Dysconnectivity of multiple resting-state networks in patients with schizophrenia who have persistent auditory verbal hallucinations, J. Psychiatry Neurosci. 36 (2011) 366–374. [12] K.J. Friston, Functional and effective connectivity in neuroimaging: a synthesis, Hum. Brain Mapp. 2 (1994) 56–78. [13] K.J. Friston, L. Harrison, W. Penny, Dynamic causal modelling, Neuroimage 19 (2003) 1273–1302. [14] S. Benetti, A. Mechelli, M. Picchioni, M. Broome, S. Williams, P. McGuire, Functional integration between the posterior hippocampus and prefrontal cortex is impaired in both first episode schizophrenia and the at risk mental state, Brain J. Neurol. 132 (2009) 2426–2436. [15] D. Dima, J.P. Roiser, D.E. Dietrich, C. Bonnemann, H. Lanfermann, H.M. Emrich, W. Dillo, Understanding why patients with schizophrenia do not perceive the hollow-mask illusion using dynamic causal modelling, Neuroimage 46 (2009) 1180–1186. [16] B. Curcic-Blake, L. van der Meer, G.H.M. Pijnenborg, A.S. David, A. Aleman, Insight and psychosis: functional and anatomical brain connectivity and self-Reflection in schizophrenia, Hum. Brain Mapp. 36 (2015) 4859–4868. [17] G.W. Wu, Y.X. Wang, T.E. Mwansisya, W.D. Pu, H.R. Zhang, C. Liu, Q. Yang, E.Y.H. Chen, Z.M. Xue, Z.N. Liu, B.C. Shan, Effective connectivity of the posterior cingulate and medial prefrontal cortices relates to working memory impairment in schizophrenic and bipolar patients, Schizophr. Res. 158 (2014) 85–90. [18] D. Dima, J.P. Roiser, D.E. Dietrich, C. Bonnemann, H. Lanfermann, H.M. Emrich, W. Dillo, Understanding why patients with schizophrenia do not perceive the hollow-mask illusion using dynamic causal modelling, Neuroimage 46 (2009) 1180–1186. [19] C.J. Price, The anatomy of language: contributions from functional neuroimaging, J. Anat. 197 (Pt. 3) (2000) 335–359. [20] C.J. Price, The anatomy of language: a review of 100 fMRI studies published in 2009, Ann. N. Y. Acad. Sci. 1191 (2010) 62–88. [21] M.B. First, R.L. Spitzer, M. Gibbon, J.B.W. Williams, Structured Clinical Interviewfor DSM-IV Axis I Disorders – (SCID-I/P, Version 2.0), patient edition, Biometrics Research Department, New York State Psychiatric Institute, New York, 1996. [22] S.R. Kay, A. Fiszbein, L.A. Opler, The positive and negative syndrome scale (PANSS) for schizophrenia, Schizophr. Bull. 13 (1987) 261–276. [23] R.E. Hoffman, K.A. Hawkins, R. Gueorguieva, N.N. Boutros, F. Rachid, K. Carroll, J.H. Krystal, Transcranial magnetic stimulation of left temporoparietal cortex and medication-resistant auditory hallucinations, Arch. Gen. Psychiatry 60 (2003) 49–56. [24] J.A. Gray, J. Feldon, J.N.P. Rawlins, A.D. Smith, The neuropsychology of schizophrenia, Behav. Brain Sci. 14 (1991) 1–19. [25] K. Henke, B. Weber, S. Kneifel, H.G. Wieser, A. Buck, Human hippocampus associates information in memory, Proc. Natl. Acad. Sci. U. S. A. 96 (1999) 5884–5889. [26] P.K. McGuire, D.A. Silbersweig, R.M. Murray, A.S. David, R.S. Frackowiak, C.D. Frith, Functional anatomy of inner speech and auditory verbal imagery, Psychol. Med. 26 (1996) 29–38. [27] P.K. McGuire, D.A. Silbersweig, I. Wright, R.M. Murray, R.S. Frackowiak, C.D. Frith, The neural correlates of inner speech and auditory verbal imagery in
182
[28]
[29] [30] [31] [32]
[33] [34] [35] [36]
[37]
[38]
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182 schizophrenia: relationship to auditory verbal hallucinations, Br. J. Psychiatry 169 (1996) 148–159. N. Tzourio-Mazoyer, B. Landeau, D. Papathanassiou, F. Crivello, O. Etard, N. Delcroix, B. Mazoyer, M. Joliot, Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain, Neuroimage 15 (2002) 273–289. K.E. Stephan, L.M. Harrison, W.D. Penny, K.J. Friston, Biophysical models of fMRI responses, Curr. Opin. Neurobiol. 14 (2004) 629–635. W.D. Penny, K.E. Stephan, A. Mechelli, K.J. Friston, Comparing dynamic causal models, Neuroimage 22 (2004) 1157–1172. B. Li, J. Daunizeau, K.E. Stephan, W. Penny, D. Hu, K. Friston, Generalised filtering and stochastic DCM for fMRI, Neuroimage 58 (2011) 442–457. M.L. Seghier, P. Zeidman, N.H. Neufeld, A.P. Leff, C.J. Price, Identifying abnormal connectivity in patients using dynamic causal modeling of FMRI responses, Front. Syst. Neurosci. 4 (2010). W.X. Li B, S. Yao, D. Hu, K. Friston, Task-dependent modulation of effective connectivity within the default mode network, Front. Psychol. 3 (2012). K.S. Karl Friston, Baojuan Li, Jean Daunizeau, Generalised filtering, Math. Prob. Eng. 2010 (2010) 1–35. K.J. Friston, B. Li, J. Daunizeau, K.E. Stephan, Network discovery with DCM, Neuroimage 56 (2011) 1202–1221. K.J. Friston, V. Litvak, A. Oswal, A. Razi, K.E. Stephan, B.C. van Wijk, G. Ziegler, P. Zeidman, Bayesian model reduction and empirical Bayes for group (DCM) studies, Neuroimage 128 (2016) 413–431. C.D. Frith, The cognitive abnormalities underlying the symptomatology and the disability of patients with schizophrenia, Int. Clin. Psychopharmacol. 10 (1995) 87–98. K.J. Friston, C.D. Frith, Schizophrenia: a disconnection syndrome? Clin. Neurosci. 3 (1995) 89–97.
[39] C.D. Frith, The neural basis of hallucinations and delusions, C. R. Biol. 328 (2005) 169–175. [40] G. Hickok, D. Poeppel, Dorsal and ventral streams: a framework for understanding aspects of the functional anatomy of language, Cognition 92 (2004) 67–99. [41] R. Upthegrove, M.R. Broome, K. Caldwell, J. Ives, F. Oyebode, S.J. Wood, Understanding auditory verbal hallucinations: a systematic review of current evidence, Acta Psychiatr. Scand. 133 (2016) 352–367. [42] R. Jardri, A. Pouchet, D. Pins, P. Thomas, Cortical activations during auditory verbal hallucinations in schizophrenia: a coordinate-based meta-analysis, Am. J. Psychiatry 168 (2011) 73–81. [43] R.E. Hoffman, T. Fernandez, B. Pittman, M. Hampson, Elevated functional connectivity along a corticostriatal loop and the mechanism of auditory/verbal hallucinations in patients with schizophrenia, Biol. Psychiatry 69 (2011) 407–414. [44] H. Duffau, The anatomo-functional connectivity of language revisited. New insights provided by electrostimulation and tractography, Neuropsychologia 46 (2008) 927–934. [45] P.A. Geoffroy, J. Houenou, A. Duhamel, A. Amad, A.D. De Weijer, B. Curcic-Blake, D.E. Linden, P. Thomas, R. Jardri, The Arcuate Fasciculus in auditory-verbal hallucinations: a meta-analysis of diffusion-tensor-imaging studies, Schizophr. Res. 159 (2014) 234–237. [46] A.D. de Weijer, S.F. Neggers, K.M. Diederen, R.C. Mandl, R.S. Kahn, H.E. Hulshoff Pol, I.E. Sommer, Aberrations in the arcuate fasciculus are associated with auditory verbal hallucinations in psychotic and in non-psychotic individuals, Hum. Brain Mapp. 34 (2013) 626–634. [47] A. Razi, J. Kahan, G. Rees, K.J. Friston, Construct validation of a DCM for resting state fMRI, Neuroimage 106 (2015) 1–14.
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Research article
Decreased middle temporal gyrus connectivity in the language network in schizophrenia patients with auditory verbal hallucinations Linchuan Zhang a,1 , Baojuan Li a,1 , Huaning Wang b , Liang Li a , Qimei Liao a , Yang Liu a , Xianghong Bao c , Wenlei Liu a , Hong Yin d , Hongbing Lu a,∗ , Qingrong Tan b,∗∗ a
Department of Biomedical Engineering, Fourth Military Medical University, Xi’an, Shaanxi 710032, PR China Department of Psychiatry, Xijing Hospital, Fourth Military Medical University, Xi’an, Shaanxi 710032, PR China c Department of Health Service, School of Military Preventive Medicine, Fourth Military Medical University, Xi’an, Shaanxi 710032, PR China d Department of Radiology, Xijing Hospital, Fourth Military Medical University, Xi’an, Shaanxi, PR China b
h i g h l i g h t s • Impaired effective connectivity within the language network in schizophrenia patients with AVHs was studied. • The connection from the LIFG to LMTG was significantly decreased in patients with AVHs compared to those without AVHs. • The effective connection from the LIPL to LMTG was significantly decreased compared to the healthy controls.
a r t i c l e
i n f o
Article history: Received 14 October 2016 Received in revised form 9 May 2017 Accepted 21 May 2017 Available online 29 May 2017 Keywords: Schizophrenia Auditory verbal hallucinations Dynamic causal modeling Effective connectivity Language network
a b s t r a c t As the most common symptoms of schizophrenia, the long-term persistence of obstinate auditory verbal hallucinations (AVHs) brings about great mental pain to patients. Neuroimaging studies of schizophrenia have indicated that AVHs were associated with altered functional and structural connectivity within the language network. However, effective connectivity that could reflect directed information flow within this network and is of great importance to understand the neural mechanisms of the disorder remains largely unknown. In this study, we utilized stochastic dynamic causal modeling (DCM) to investigate directed connections within the language network in schizophrenia patients with and without AVHs. Thirty-six patients with schizophrenia (18 with AVHs and 18 without AVHs), and 37 healthy controls participated in the current resting-state functional magnetic resonance imaging (fMRI) study. The results showed that the connection from the left inferior frontal gyrus (LIFG) to left middle temporal gyrus (LMTG) was significantly decreased in patients with AVHs compared to those without AVHs. Meanwhile, the effective connection from the left inferior parietal lobule (LIPL) to LMTG was significantly decreased compared to the healthy controls. Our findings suggest aberrant pattern of causal interactions within the language network in patients with AVHs, indicating that the hypoconnectivity or disrupted connection from frontal to temporal speech areas might be critical for the pathological basis of AVHs. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Schizophrenia (SZ) is a serious mental disorder that associates with abnormalities of brain structure and functions [1–3]. As important indicators to evaluate the severity and clinical efficacy of
∗ Corresponding author at: Department of Biomedical Engineering, Fourth Military Medical University, 169 Changlexi Rd., Xi’an, Shaanxi 710032, China. ∗∗ Corresponding author at: Department of Psychiatry, Xijing Hospital, 145 Changlexi Rd., Xi’an, Shaanxi 710032, China. E-mail addresses: [email protected] (H. Lu), [email protected] (Q. Tan). 1 These authors contributed equally. http://dx.doi.org/10.1016/j.neulet.2017.05.042 0304-3940/© 2017 Elsevier B.V. All rights reserved.
schizophrenia, auditory verbal hallucinations (AVHs) are the most common symptoms of SZ, with an incidence of 60%–80% [4]. SZ is a disorder with unknown etiology and pathophysiology that are further complicated than regional brain abnormalities. Previous neuroimaging studies on AVHs have revealed structural and functional alterations in the language network. A diffusion tensor imaging (DTI) study revealed reduced integrity of the fiber connecting frontal and temporo-parietal language regions [5]. It was further supported by neuroimaging studies revealing that AVHs in SZ was associated with dysconnectivity in the language network [6]. A study of fMRI showed that the emergence of AVHs was the result of mild disruption of function connectivity along the
178
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182
fronto-temporal pathway [7]. Reduced fronto-temporal functional connectivity has also been demonstrated in SZ patients with AVHs during sentence-completion task [8], external speech processing [9], and speaking aloud task [10]. A resting state fMRI study showed that disrupted intrinsic connectivity of the language network could underlie persistent AVHs in patients with SZ [11]. Although the above studies have indicated that the emergence of AVHs is associated with altered structural and functional connectivity within the language network, it still remains unclear how these regions interact with each other to cause the symptoms like AVHs in SZ. As a powerful tool for effective connectivity (describes the causal influences that one region exerts over another [12]) analysis, dynamic causal modeling (DCM) [13] not only enable us to quantify the strength of the connections, but also demonstrate directed information flow from one region to another, therefore provide more information to recover the underlying neural substrates of AVHs and SZ. In fact, there have been some attempts to study abnormal effective connectivity in SZ patients using DCM ´ cic-Blake ´ [9,14,15]. Curˇ et al. found that increased effective connectivity from the posterior cingulate cortex to the ventromedial prefrontal cortex was associated with poorer insight [16]. In addition, abnormal effective connectivity of the posterior cingulate and medial prefrontal cortices have also been related to working memory impairments [17]. Interestingly, measures of effective connectivity have been used to explain differences in the perception of the hollow-mask illusion [18]. However, it still remains unknown how regions in the language network interact with each other abnormally in SZ patients with AVHs. In this study, we aimed to investigate resting-state effective connectivity within the language network in 18 SZ patients with AVHs, 18 patients without AVHs and 37 healthy controls using DCM. Three regions including the left inferior frontal gyrus (LIFG), left middle temporal gyrus (LMTG), and left inferior parietal lobule (LIPL) were selected as regions of interest (ROIs) according to previous findings [19,20]. We hypothesized that the effective connectivity within the language network would be altered in SZ patients with AVHs. 2. Methods 2.1. Subjects This study was approved by the Institutional Board of The Fourth Military Medical University. All participants received a full comprehensive description of the MRI study and gave voluntarily written informed consent before entering the study. A total of 73 subjects were recruited (36 patients with schizophrenia and 37 healthy controls). All patients fulfilled the DSM-IV [21] diagnostic criteria for schizophrenia and was further divided into the AVH(18 subjects) and nAVH group(18 subjects) based on the scores of the Positive And Negative Syndrome Scale (PANSS) [22], as well as detailed symptomatology information acquired through patient interview and examination of patient’s medical records including the severity
of disorganization symptoms, excitement and emotional distress. The AVH and nAVH groups were further characterized in terms of the Auditory Hallucination Rating Scale (AHRS) [23]. For healthy controls, exclusion criteria included current or past history of psychiatric illness and the presence of psychosis in first-degree relatives. Demographic and clinical characteristics of all subjects were given in Table 1. All three groups were matched for age, gender, and education. 2.2. fMRI data acquisition All MRI data were acquired on a 3.0 T MR scanner (MAGNETOM Trio, Siemens AG, Erlangen, Germany) at the Radiology Department of Xijing Hospital. The fMRI acquisition parameters were as follows: repetition time (TR) = 2000 ms; echo time (TE) = 30 ms; flip angle = 90◦ ; field-of-view = 220 mm × 220 mm; 33 slices; slice thickness = 4 mm; interslice gap = 0.6 mm. We collected 240 images for each subject. 2.3. fMRI data preprocessing Preprocessing of the fMRI data was performed using SPM8 software (http://www.fil.ion.ucl.ac.uk/spm). First, a rigid body transformation was used to realign the functional images to the first image of each session to correct for head motion. Then images were normalized to a MNI EPI template. Finally, fMRI images were spatially smoothed using an 8-mm full-width-at-half-maximum (FWHM) Gaussian kernel. 2.4. General linear model After data preprocessing, resting-state fMRI images were modelled using a general linear model with the following regressors: (1) the six rigid body realignment parameters to model head motion that were then used to remove effects of head motion when extracting time courses of the ROIs; (2) one constant regressor to model the baseline; and (3) cosine basis functions with frequencies ranging from 0.0078 (1/128) to model possible signal drift. The cosine basis functions used in the GLM serve as a high-pass filter which filtered out low-frequency signal drift in the current study. 2.5. Regions-of-interest The following three regions were used: the LIFG, LMTG, and LIPL which are the brain regions generally involved in the language network. LMTG is associated with cognitive self-monitoring [24] and memory retrieval [25]. LIFG has been associated with the preparation of overt speech and the generation of inner speech [26,27]. LIPL is considered as the storage of long-term memory mostly in verbal modality [26]. For each of the ROIs, a mask representing this region was created using the Anatomical Automatic Labeling (AAL) atlas [28]
Table 1 Demographic and clinical characteristics.
Age Sex(male/female) Handedness (right/left) Education(years) Duration of illness(months) PANSS Total Score AHRS Score PANSS: Positive and Negative Syndrome Scale. AHRS: Auditory Hallucination Rating Scale.
AVH Patients n = 18
nAVH Patients n = 18
Healthy Controls n = 37
22.56 ± 6.73 10/8 18/0 12.44 ± 2.30 5.94 ± 5.93 106.44 ± 13.55 26.22 ± 8.10
22.67 ± 3.85 9/9 18/0 12.56 ± 2.19 12.44 ± 18.16 88.06 ± 23.90
22.48 ± 5.84 19/18 37/0 12.43 ± 3.13
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182
179
Fig. 1. Significant effective connectivity of the language network (one-sample t-test, p < 0.05, Bonferroni corrected for multiple comparisons). (A) three ROIs: LIFG, LMTG, LIPL within the language network; (B) effective connectivity in AVH group; (C) effective connectivity in the nAVH group; (D) effective connectivity in control group.
provided in the WFU (Wake Forest University) PickAtlas toolbox (http://www.nitrc.org/projects/wfu pickatlas/). These masks were then combined with a thresholded SPM (with the threshold set to 0.05) testing for the baseline generated by first-level (within subject) analyses to constrain the extraction of subject-specific ROI time courses. For each ROI, the fMRI time courses of all the voxels that located within the corresponding ROI mask and showed activation in the thresholded SPM map were extracted. The first principle component of these time courses was finally used to summarize the BOLD response of the ROI. The effects of head motion were also removed from the time courses. 2.6. Dynamic causal modeling 2.6.1. Specification and inversion of the fully-connected model Effective connectivity analysis was conducted using DCM10. Compared to traditional models, DCM integrates the BOLD physiological response model with the functional activation model [29]. It uses differential equations to fit both hemodynamic response and neural dynamics to make the model more accurately [30]. The causal interactions among different brain regions are modelled at neuronal level using neuronal state equations: z˙ ≈ Az +
uj Bj z + Cu
(1)
Here, z denotes the neuronal activities of the ROIs. The N-by-N (N is the number of the ROIs) matrix A represents the effective connectivity from one region to another in the absence of external inputs. Matrices B and C describe changes in the effective connectivity and regional BOLD responses caused by experimental inputs. The neuronal state equations are supplemented by an observation model
which maps neuronal states to observed fMRI signals. Recently, the deterministic DCM described in Eq. (1) has been extended to accommodate state noises [31,32]: z˙ ≈ Az +
uj Bj z + C v + ωx
(2)
Here, x denotes the state noise which models stochastic innovations in the system [32]. This stochastic DCM which is able to model resting-state fMRI data was employed in the current study [33]. After extracting time courses for all three ROIs, we first specified a fully-connected DCM model for each subject. This model hypothesized that any pair of ROIs were reciprocally connected. The fully-connected DCM model for each subject was then inverted using the generalised filtering scheme which provided parameter estimates as well as model evidence of this fully-connected model [34]. The model evidence p(y|m) denotes the probability of obtaining the data y given a model m and is usually used to compare competing hypotheses. 2.6.2. Network discovery The optimal model that can best explain the data may not be the fully-connected model but a reduced one. A reduced model has the same set of ROIs with the fully-connected model while some of the connections presented in the fully-connected model are absent in the reduced one. We have 3 ROIs, the fully-connected model thus includes 9 connections (3 self-connections and 6 inter-regional connections). Any subset of these 9 connections may be absent in the reduced model, in that case, there would be 512 (29 ) possible reduced models. One way to find the optimal model is to specify and invert all the possible models and then use Bayesian model selection. This strategy requires the inversion of hundreds of mod-
180
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182
Fig. 2. Different effective connectivity between groups (post-hoc t-tests, p = 0.05). (A) AVHS < Control; (B) AVHS < Non-AVHS; (C) Non-AVHS < Control (D) Non-AVHS > Control.
els which is quite time-consuming. Thus in the current study, we adopted a network discovery scheme [35]. Instead of specifying and inverting all the possible models, we only needed to invert a single model (the fully-connected model). Provided with the parameter estimates and model evidence of this model, the network discovery scheme searched over all possible reduced models and recover the optimal model that has the highest model evidence. This scheme is quite efficient and has been showed to be more robust and accurate compared with the first strategy [36]. Finally, the group average model was obtained through Bayesian parametric averaging (BPA). For each reduced model, the model evidences over all the subjects were pooled. All possible models were scored according to their model evidences and the one with the highest evidence was recovered as the optimal model. 2.6.3. Statistical analysis Subject specific parameter estimates of the optimal model were taken to the between-subject level analysis using a classical random effects analysis (t-tests). To examine group differences in effective connectivity, a one way ANOVA was first performed and then followed up with post-hoc t-tests. 3. Results Fig. 1 shows the effective connectivity of the language network in the AVH group, nAVH group and controls, respectively (one-sample t-test at p < 0.05, Bonferroni corrected for multiple comparisons). One way ANOVA (p < 0.05) showed that the effective connection from LIFG to LMTG (p = 0.000) and the effective connection from LIPL to LMTG (p = 0.000) were significantly different among the three groups. Fig. 2 illustrates changes in effective connectivity in the patients group (post-hoc t-tests, p < 0.05). The results showed that: (1) com-
pared to healthy controls, the effective connection from LIPL to LMTG was significantly decreased in AVH group (p = 0.000); (2) compared to patients without AVHs, the effective connection from LIFG to LMTG was significantly decreased in AVH group (p = 0.000); and (3) patients without AVHs showed increased effective connection from LIFG to LMTG and decreased effective connection from LIPL to LMTG compared to healthy controls (p = 0.000). 4. Discussion In this study, we observed decreased effective connection from LIPL to LMTG in both patient groups compared to healthy controls. These findings provide a piece of evidence for the “dysconnection hypothesis” of SZ. On the one hand, decreased effective connectivity found in the current study provides further support for previous neuroimaging studies that have reported impaired temporo-parietal functional and structural connectivity. On the other hand, the directionality of the connections that DCM furnishes helps to elucidate the exact mechanisms underlying the pathophysiology of SZ and AVHs. Functional connectivity studies can only tell us that the correlation between the activity of the temporal and parietal regions is impaired in the patients. The results from the current DCM study further revealed that the impairments may be mediated through a lack of causal influences from the LIPL to the LMTG. Considering that patients without AVHs also showed decreased connection form LIPL to LMTG compared to healthy controls, attenuation of this connection may serve as a biomarker of SZ rather than AVHs. We also found that the AVH group showed significantly decreased effective connection from LIFG to LMTG compared to patients without AVHs. Although the presence of AVHs is one of the defining and critical symptoms for patients with schizophrenia, not all patients experience AVHs. Thus the differences in the
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182
effective connectivity between the two groups of patients may shed light on the neural substrates of AVHs. AVHs were considered as the disability of patients to recognize self-generated inner speech [37]. The failure to integrate information regarding inner speech was attributed to verbal generation in the frontal cortex and sensory processing in temporal and parietal regions [38,39]. As for normal subjects, the language pathways have been proposed to diverge into two streams: a ventral stream projects towards MTG which is associated with mapping sound onto meaning, as well as a dorsal stream projects towards posterior Sylvian fissure and then frontal regions which is associated with mapping sound onto articulatory-based representations [40]. In terms of AVHs, impaired self-monitoring has been proposed to be associated with the mechanism underlying this symptom [41]. According to this theory, decreased effective connection in patients with AVHs from LIFG to LMTG found in our study indicated a lack of efficient inhibition of the frontal lobe on the internal speech signal generated by the temporal lobe. Weakening inhibition would then lead to the occurrences of AVHs. Thus the decreased effective connection from LIFG to LMTG may lead to turning inner speeches to verbal hallucinations abnormally. fMRI studies showed that those areas played a central role in the language system of AVHs [42,43]. The anatomic connection between frontal and temporal regions is anterior indirect segment in the left arcuate fasciculus (AF), which is a key anatomical connection between the frontal and temporal speech areas [44–46]. A DTI study on patients with AVHs found disruptions of white matter integrity in the anterior indirect segment of AF [45]. Considering that patients without AVHs showed increased effective connection form LIFG to LMTG compared to healthy controls, our results not only support the important role of the fronto-temporal connection for the presence of AVHs, they further imply that hypoconnectivity or disrupted connection from frontal speech areas to temporal speech areas might be critical for the pathological basis of AVHs. It is important to note that the LIFG, LIPL and LMTG were relatively large (covered 27, 20 and 37 cm3 , respectively) in the current study as the ROIs were defined according to the AAL template. In that case, some regions like the LIPL may contain different subregions that each has different role in language processing which may complicate the interpretation of the results. In order to solve this problem, we may try to obtain more specific ROI locations by also collecting fMRI images while the subjects are performing language processing tasks in the future. In addition, as the fast and optimal spectral DCM scheme [47] and the new Bayesian model reduction (BMR) tool [36] are available now, we may also consider adopting these tools in future studies to further improve the efficiency and robustness of the calculations. Competing interests The authors declare that they have no competing interests. Authors’ contributions HW, HL and QT designed the experiment; HW, QT and HY recruited the subjects and collected the fMRI images; LZ, BL, LL, WL and YL analyzed the data; LZ, BL, QL, XB and HL prepared the manuscript. Acknowledgements This work was supported by the National Natural Science Foundation of China [81301199, 81630032, 81230035 and 81071220]. We thank anonymous reviewers for guidance in clarifying and elaborating this report.
181
References [1] J.W. Buckholtz, A. Meyer-Lindenberg, Psychopathology and the human connectome: toward a transdiagnostic model of risk for mental illness, Neuron 74 (2012) 990–1004. [2] K.J. Friston, Functional and effective connectivity in neuroimaging:a synthesis, Hum. Brain Mapp. 2 (1994). [3] K.E. Stephan, T. Baldeweg, K.J. Friston, Synaptic plasticity and dysconnection in schizophrenia, Biol. Psychiatry 59 (2006) 929–939. [4] N.C. Andreasen, M. Flaum, Schizophrenia: the characteristic symptoms, Schizophr. Bull. 17 (1991) 27–49. [5] A.D. de Weijer, R.C. Mandl, K.M. Diederen, S.F. Neggers, R.S. Kahn, H.E. Hulshoff Pol, I.E. Sommer, Microstructural alterations of the arcuate fasciculus in schizophrenia patients with frequent auditory verbal hallucinations, Schizophr. Res. 130 (2011) 68–77. [6] P. Allen, G. Modinos, D. Hubl, G. Shields, A. Cachia, R. Jardri, P. Thomas, T. Woodward, P. Shotbolt, M. Plaze, R. Hoffman, Neuroimaging auditory hallucinations in schizophrenia: from neuroanatomy to neurochemistry and beyond, Schizophr. Bull. 38 (2012) 695–703. [7] S. Benetti, W. Pettersson-Yeo, P. Allen, M. Catani, S. Williams, A. Barsaglini, L.M. Kambeitz-Ilankovic, P. McGuire, A. Mechelli, Auditory verbal hallucinations and brain dysconnectivity in the perisylvian language network: a multimodal investigation, Schizophr. Bull. 41 (2015) 192–200. [8] S.M. Lawrie, C. Buechel, H.C. Whalley, C.D. Frith, K.J. Friston, E.C. Johnstone, Reduced frontotemporal functional connectivity in schizophrenia associated with auditory hallucinations, Biol. Psychiatry 51 (2002) 1008–1011. [9] A. Mechelli, P. Allen, E. Amaro Jr., C.H. Fu, S.C. Williams, M.J. Brammer, L.C. Johns, P.K. McGuire, Misattribution of speech and impaired connectivity in patients with auditory verbal hallucinations, Hum. Brain Mapp. 28 (2007) 1213–1222. [10] J.M. Ford, D.H. Mathalon, S. Whitfield, W.O. Faustman, W.T. Roth, Reduced communication between frontal and temporal lobes during talking in schizophrenia, Biol. Psychiatry 51 (2002) 485–492. [11] N.D. Wolf, F. Sambataro, N. Vasic, K. Frasch, M. Schmid, C. Schonfeldt-Lecuona, P.A. Thomann, R.C. Wolf, Dysconnectivity of multiple resting-state networks in patients with schizophrenia who have persistent auditory verbal hallucinations, J. Psychiatry Neurosci. 36 (2011) 366–374. [12] K.J. Friston, Functional and effective connectivity in neuroimaging: a synthesis, Hum. Brain Mapp. 2 (1994) 56–78. [13] K.J. Friston, L. Harrison, W. Penny, Dynamic causal modelling, Neuroimage 19 (2003) 1273–1302. [14] S. Benetti, A. Mechelli, M. Picchioni, M. Broome, S. Williams, P. McGuire, Functional integration between the posterior hippocampus and prefrontal cortex is impaired in both first episode schizophrenia and the at risk mental state, Brain J. Neurol. 132 (2009) 2426–2436. [15] D. Dima, J.P. Roiser, D.E. Dietrich, C. Bonnemann, H. Lanfermann, H.M. Emrich, W. Dillo, Understanding why patients with schizophrenia do not perceive the hollow-mask illusion using dynamic causal modelling, Neuroimage 46 (2009) 1180–1186. [16] B. Curcic-Blake, L. van der Meer, G.H.M. Pijnenborg, A.S. David, A. Aleman, Insight and psychosis: functional and anatomical brain connectivity and self-Reflection in schizophrenia, Hum. Brain Mapp. 36 (2015) 4859–4868. [17] G.W. Wu, Y.X. Wang, T.E. Mwansisya, W.D. Pu, H.R. Zhang, C. Liu, Q. Yang, E.Y.H. Chen, Z.M. Xue, Z.N. Liu, B.C. Shan, Effective connectivity of the posterior cingulate and medial prefrontal cortices relates to working memory impairment in schizophrenic and bipolar patients, Schizophr. Res. 158 (2014) 85–90. [18] D. Dima, J.P. Roiser, D.E. Dietrich, C. Bonnemann, H. Lanfermann, H.M. Emrich, W. Dillo, Understanding why patients with schizophrenia do not perceive the hollow-mask illusion using dynamic causal modelling, Neuroimage 46 (2009) 1180–1186. [19] C.J. Price, The anatomy of language: contributions from functional neuroimaging, J. Anat. 197 (Pt. 3) (2000) 335–359. [20] C.J. Price, The anatomy of language: a review of 100 fMRI studies published in 2009, Ann. N. Y. Acad. Sci. 1191 (2010) 62–88. [21] M.B. First, R.L. Spitzer, M. Gibbon, J.B.W. Williams, Structured Clinical Interviewfor DSM-IV Axis I Disorders – (SCID-I/P, Version 2.0), patient edition, Biometrics Research Department, New York State Psychiatric Institute, New York, 1996. [22] S.R. Kay, A. Fiszbein, L.A. Opler, The positive and negative syndrome scale (PANSS) for schizophrenia, Schizophr. Bull. 13 (1987) 261–276. [23] R.E. Hoffman, K.A. Hawkins, R. Gueorguieva, N.N. Boutros, F. Rachid, K. Carroll, J.H. Krystal, Transcranial magnetic stimulation of left temporoparietal cortex and medication-resistant auditory hallucinations, Arch. Gen. Psychiatry 60 (2003) 49–56. [24] J.A. Gray, J. Feldon, J.N.P. Rawlins, A.D. Smith, The neuropsychology of schizophrenia, Behav. Brain Sci. 14 (1991) 1–19. [25] K. Henke, B. Weber, S. Kneifel, H.G. Wieser, A. Buck, Human hippocampus associates information in memory, Proc. Natl. Acad. Sci. U. S. A. 96 (1999) 5884–5889. [26] P.K. McGuire, D.A. Silbersweig, R.M. Murray, A.S. David, R.S. Frackowiak, C.D. Frith, Functional anatomy of inner speech and auditory verbal imagery, Psychol. Med. 26 (1996) 29–38. [27] P.K. McGuire, D.A. Silbersweig, I. Wright, R.M. Murray, R.S. Frackowiak, C.D. Frith, The neural correlates of inner speech and auditory verbal imagery in
182
[28]
[29] [30] [31] [32]
[33] [34] [35] [36]
[37]
[38]
L. Zhang et al. / Neuroscience Letters 653 (2017) 177–182 schizophrenia: relationship to auditory verbal hallucinations, Br. J. Psychiatry 169 (1996) 148–159. N. Tzourio-Mazoyer, B. Landeau, D. Papathanassiou, F. Crivello, O. Etard, N. Delcroix, B. Mazoyer, M. Joliot, Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain, Neuroimage 15 (2002) 273–289. K.E. Stephan, L.M. Harrison, W.D. Penny, K.J. Friston, Biophysical models of fMRI responses, Curr. Opin. Neurobiol. 14 (2004) 629–635. W.D. Penny, K.E. Stephan, A. Mechelli, K.J. Friston, Comparing dynamic causal models, Neuroimage 22 (2004) 1157–1172. B. Li, J. Daunizeau, K.E. Stephan, W. Penny, D. Hu, K. Friston, Generalised filtering and stochastic DCM for fMRI, Neuroimage 58 (2011) 442–457. M.L. Seghier, P. Zeidman, N.H. Neufeld, A.P. Leff, C.J. Price, Identifying abnormal connectivity in patients using dynamic causal modeling of FMRI responses, Front. Syst. Neurosci. 4 (2010). W.X. Li B, S. Yao, D. Hu, K. Friston, Task-dependent modulation of effective connectivity within the default mode network, Front. Psychol. 3 (2012). K.S. Karl Friston, Baojuan Li, Jean Daunizeau, Generalised filtering, Math. Prob. Eng. 2010 (2010) 1–35. K.J. Friston, B. Li, J. Daunizeau, K.E. Stephan, Network discovery with DCM, Neuroimage 56 (2011) 1202–1221. K.J. Friston, V. Litvak, A. Oswal, A. Razi, K.E. Stephan, B.C. van Wijk, G. Ziegler, P. Zeidman, Bayesian model reduction and empirical Bayes for group (DCM) studies, Neuroimage 128 (2016) 413–431. C.D. Frith, The cognitive abnormalities underlying the symptomatology and the disability of patients with schizophrenia, Int. Clin. Psychopharmacol. 10 (1995) 87–98. K.J. Friston, C.D. Frith, Schizophrenia: a disconnection syndrome? Clin. Neurosci. 3 (1995) 89–97.
[39] C.D. Frith, The neural basis of hallucinations and delusions, C. R. Biol. 328 (2005) 169–175. [40] G. Hickok, D. Poeppel, Dorsal and ventral streams: a framework for understanding aspects of the functional anatomy of language, Cognition 92 (2004) 67–99. [41] R. Upthegrove, M.R. Broome, K. Caldwell, J. Ives, F. Oyebode, S.J. Wood, Understanding auditory verbal hallucinations: a systematic review of current evidence, Acta Psychiatr. Scand. 133 (2016) 352–367. [42] R. Jardri, A. Pouchet, D. Pins, P. Thomas, Cortical activations during auditory verbal hallucinations in schizophrenia: a coordinate-based meta-analysis, Am. J. Psychiatry 168 (2011) 73–81. [43] R.E. Hoffman, T. Fernandez, B. Pittman, M. Hampson, Elevated functional connectivity along a corticostriatal loop and the mechanism of auditory/verbal hallucinations in patients with schizophrenia, Biol. Psychiatry 69 (2011) 407–414. [44] H. Duffau, The anatomo-functional connectivity of language revisited. New insights provided by electrostimulation and tractography, Neuropsychologia 46 (2008) 927–934. [45] P.A. Geoffroy, J. Houenou, A. Duhamel, A. Amad, A.D. De Weijer, B. Curcic-Blake, D.E. Linden, P. Thomas, R. Jardri, The Arcuate Fasciculus in auditory-verbal hallucinations: a meta-analysis of diffusion-tensor-imaging studies, Schizophr. Res. 159 (2014) 234–237. [46] A.D. de Weijer, S.F. Neggers, K.M. Diederen, R.C. Mandl, R.S. Kahn, H.E. Hulshoff Pol, I.E. Sommer, Aberrations in the arcuate fasciculus are associated with auditory verbal hallucinations in psychotic and in non-psychotic individuals, Hum. Brain Mapp. 34 (2013) 626–634. [47] A. Razi, J. Kahan, G. Rees, K.J. Friston, Construct validation of a DCM for resting state fMRI, Neuroimage 106 (2015) 1–14.