Aberrations in the arcuate fasciculus are associated with auditory verbal hallucinations in psychotic and in non‐psychotic individuals

Abstract

The pathophysiology of auditory verbal hallucinations (AVH) is still unclear. Cognitive as well as electrophysiological studies indicate that a defect in sensory feedback (corollary discharge) may contribute to the experience of AVH. This could result from disruption of the arcuate fasciculus, the major tract connecting frontal and temporo‐parietal language areas. Previous diffusion tensor imaging studies indeed demonstrated abnormalities of this tract in schizophrenia patients with AVH. It is, however, difficult to disentangle specific associations with AVH in this patient group as many other factors, such as other positive and negative symptoms, medication or halted education could likewise have affected tract integrity. We therefore investigated AVH in relative isolation and studied a group of non‐psychotic individuals with AVH as well as patients with AVH and non‐hallucinating matched controls. We compared tract integrity of the arcuate fasiculus and of three other control tracts, between 35 non‐psychotic individuals with AVH, 35 schizophrenia patients with AVH, and 36 controls using diffusion tensor imaging and magnetization transfer imaging. Both groups with AVH showed an increase in magnetization transfer ratio (MTR) in the arcuate fasciculus, but not in the other control tracts. In addition, a general decrease in fractional anisotropy (FA) for almost all bundles was observed in the patient group, but not in the non‐psychotic individuals with AVH. As increased MTR in the arcuate fasciculus was present in both hallucinating groups, a specific association with AVH seems plausible. Decreases in FA, on the other hand, seem to be related to other disease processes of schizophrenia. Hum Brain Mapp, 2013. © 2011 Wiley Periodicals, Inc.

INTRODUCTION

Auditory verbal hallucinations (AVH) are one of the core symptoms of schizophrenia. Its exact pathophysiology remains elusive, although functional MRI studies suggest the involvement of the language system [Diederen et al.,2010; Ford et al.,2007; Heinks‐Maldonado et al.,2007; Jardri et al.,2010; Shergill et al.,2000; Sommer et al.,2008; Whitford et al.,2010]. Activation patterns during the experience of AVH include the classic language areas of Broca and Wernicke and the contralateral homologues, suggesting involvement of abnormalities in language perception as well as production [Diederen et al.,2010,2011; Sommer et al.,2007,2008]. It is conceivable that the activation in inferior frontal areas reflect the production of verbal content that is misattributed to an external source and hence experienced as AVH. It is as yet unclear; why these verbal products are not recognized as self generated but rather attributed to an external source. It is proposed [Feinberg,1978; Ford et al.,2007; Frith,1992] that a malfunction of the corollary discharge system to underlies the failure to recognize self‐produced verbal fragments [Whitford et al.,2010]. Corollary discharge signifies the neuronal circuit that suppresses the sensory consequences of self‐generated actions, in this case covert speech. A malfunction in the corollary discharge system of speech could result from disintegrity of the main white matter tracts connecting inferior frontal and temporo‐parietal language areas. For this reason, these white matter connections have been studied extensively in schizophrenia patients with AVH [Hubl et al.,2004; Seok et al.,2007; Shergill et al.,2007; Whitford et al.,2010] mainly with diffusion tensor imaging (DTI), but recently also with a combination of DTI and magnetization transfer imaging (MTI), [de Weijer et al.,2011]. This latter study observed decreased FA values in the arcuate fasciculus, cortico spinal tract, and uncinate fasciculus, a frequent finding in schizophrenia. In addition, a specific increase in magnetization transfer ratio (MTR) was observed in the arcuate fasciculus only. The combination of high MTR values and low FA indicates a decreased integrity of microstructure (axons or glia cells) in this tract.

However, schizophrenia is a complex syndrome comprising positive, negative, and cognitive symptoms. In addition, patients usually use antipsychotic medication for a long duration which has been shown to affect the brain [Ho et al.,2011] and abort their education or professional career prematurely. It is therefore conceivable that these decreased FA values with increased MTR values in the arcuate fasciculus are not specific for AVH but instead related to another symptom cluster or even to secondary effects, such as medication or drug use or understimulation of the language areas. To test if disrupted integrity of the arcuate fasciculus is specifically associated with AVH, we set out to perform the same assessments in healthy individuals with AVH but without delusions and negative or cognitive symptoms, who never used antipsychotic medication and have normal (school) careers.

AVH are not always indicative of psychiatric disorders, but also occur in a significant minority of otherwise healthy individuals [Johns et al.,2002; Sommer et al.,2010; Tien,1991]. These healthy people with AVH demonstrate good social capacities and function within the normal range [Sommer et al.,2010]. The University Medical Centre Utrecht has acquired a unique group of over 100 subjects [Daalman et al.,2011; Sommer et al.,2010], who experience frequent AVH, i.e., at least once a month, yet do not meet criteria for a psychiatric diagnosis. AVH in these non‐psychotic individuals show similar phenomenological characteristics as patients with schizophrenia, for example in loudness, reality, experienced location, and number of voices [Daalman et al.,2011]. Furthermore, both groups show similar brain activation patterns during AVH when measured with fMRI [Diederen et al.,2011].

To elucidate the specific relation between structural abnormalities of the arcuate fasciculus and AVH we included 35 individuals of this non‐psychotic healthy group with AVH and contrasted them to a group of healthy subjects without AVH and to a schizophrenia patient group with AVH. DTI and MTI scans were acquired from these three groups. If the earlier findings of increased MTR values in the absence of increased FA [de Weijer et al.,2011] along the arcuate fasciculus are found in both hallucinating groups, this suggests a specific relation between this structural deviation and hallucinations.

METHODS

Participants

Thirty‐five patients diagnosed with schizophrenia and 35 non‐psychotic individuals with auditory verbal hallucinations [Sommer et al.,2010] and 36 healthy control subjects without hallucinations participated in the study. All groups were matched for age, gender, and handedness. The data for the patient group included here is a matched selection of the patient data published before [de Weijer et al.,2011]. Included subjects were matched to the new group of non‐psychotic individuals with AVH. As the present analysis needs to directly compare the three groups to shed new light on the nature of the alterations we observed in the AF before, we chose to include parts of the previously published data. The healthy control subjects and non‐psychotic individuals were also matched for educational level. All patients were recruited from the psychiatry department of the University Medical Center Utrecht. Patients were diagnosed using the comprehensive assessment of symptoms and history interview (CASH), [Andreasen et al.,1992] according to DSM‐IV criteria by an independent psychiatrist. The positive and negative syndrome scale (PANSS), [Kay et al.,1987] was used for the assessment of symptoms on the day of the MRI scan. All patients used typical or atypical antipsychotic medication in conventional dosages (Table I). Despite this antipsychotic treatment, they still suffered from severe auditory verbal hallucinations (AVH), occurring at least once an hour. Some of the data from the patient and healthy group were used in our previous publication [de Weijer et al.,2011].

Table I.

Demographic data

	Patient group	Non‐psychotic individuals with AVH	Healthy control group
	N = 35	N = 35	N = 36
Male/female	14/21	13/22	14/22
Age (y)	39.6(13.3)	42.1 (14.6)	41.39 (13.3)
Range (y)	18–61	19–66	21–66
Handedness (r/l)	30/5	30/5	31/5
Age of onset AVH (y)	26.6 (14.1)	14.0 (11.4)
AVH Duration (y)	12.5 (12.5)	27.6 (16.1)
PANSS total scores	58.8 (13.3)
PANSS positive scores	1.9 (3.9)
PANSS negative scores	15.0 (5.4)
PANSS general scores	28.9 (7.0)
A‐typical antipsychotic medication	22
Classic antipsychotic medication	8
No antipsychotic medication	5
GAF		81.9 (8.9)	87.2 (5.9)
SPQ		28.7 (13.9)	9.1 (6.8)
Frequency of AVH	5.5 (0.8)	3.8 (1.1)
Emotional valence of AVH	9 (2.5)	2.1 (3)
Distress due AVH	3.2 (1.9)	0.7 (1.5)
Control on AVH	3.2 (1)	1.8 (1.5)
Education level (CASH)		6.1 (1.3)	6 (2.1)

For both healthy groups absence of psychiatric disorder including substance abuse was checked using the CASH interview. In addition these groups filled out the schizotypal personality questionnaire (SPQ), [Raine,1991], rating their schizotypal tendency.

Demographic details about the patient and the healthy control groups are provided in Table I. From all participants handedness was assessed with the Edinburgh handedness inventory [Oldfield,1971]. The study was approved by the medical ethical committee of the University Medical Center and after explanation to the participants; a written informed consent was obtained. In connection with earlier findings by Daalman et al. [2011], we also included phenomenology characteristics of AVH, such as frequency, emotional valence, distress, and amount of control. In this latter study was reported that on these phenomenology characteristics patients and non‐psychotic individuals with AVH tend to score differently. Namely, patients reported a higher frequency of AVH, a higher emotional valence of the AVH, more distress due to the voices and less control on the AVH compared to non‐psychotic individuals with AVH [Daalman et al.,2011].

Image Acquisition

All MRI scans were acquired on a 3 Tesla Philips Achieva using an eight‐channel SENSE head‐coil. For each participant two sets of DTI scans, a T1‐weighted scan for anatomical reference, and a MTI scan were collected. To increase the signal to noise ratio the DTI set consisted of two transverse DTI scans with the following parameters: single shot EPI‐DTI scan consisting of 30 diffusion‐weighted scans (b = 1,000 s mm⁻²) with non‐colinear gradient directions and an average of five diffusion unweighted scans (b = 0 s mm⁻²), TR/TE = 7,035/68 ms, FOV 240 mm, matrix 128 × 128, 75 slices thickness 2 mm, no gap, SENSE factor 3, EPI factor 35, no cardiac gating. The second set was identical to the first but acquired with reversed k‐space readout (anterior direction) which allowed us to correct for geometric EPI distortions in the image processing step. The DTI scans were used for reconstruction of the fiber tracts. The 3D MTI scan consisted of two volumes, one without and one with a magnetization transfer prepulse. The parameters for the first MTI volume were: TR/TE = 65.8/2.19 ms, FOV 240 × 190 × 180 mm³, matrix 128 × 128, 95 slices, thickness 2.5 mm, flip angle = 18°. The parameters for the second MTI volume were identical to the first MTI volume but than an additional off‐resonance magnetization transfer prepulse (frequency offset 1,100 Hz; 620 degrees; three‐lobe sync‐shaped) was applied. The parameters for the T1‐weighted scan were: TR/TE = 9.87/4.6 ms, flip angle = 8°, FOV 224 × 160 × 168, matrix = 256 × 256, slice thickness 1 mm (no gap). The anatomical scan was used for normalization of all scans to MNI space (see below).

Image Processing

In the following, DTI image preprocessing was performed with software developed in‐house [Mandl et al.,2010]. All subsequent registration steps of images and fiber coordinates as well as fiber selection with ROIs were done in Matlab scripts developed in‐house using SPM5 Matlab functions, among others.

The DTI data set was corrected for susceptibility artefacts by exploiting the fact that DTI was scanned twice and with reversed phase encoding direction [Andersson et al.,2003]. Next the DTI data set was simultaneously realigned and corrected for possible gradient‐induced EPI distortions [Andersson and Skare,2002]. A robust estimation of the diffusion tensors was obtained using M‐estimators [Chang et al.,2005] to limit the influence of possible outliers. From the diffusion tensors an FA image was calculated [Basser and Pierpaoli,1996]. The FA and orientation maps were coregistered to the T1‐weighted image using the diffusion‐unweighted image as a source. Normalized mutual information coregistration as implemented in SPM5 was used (http://www.fil.ion.ucl.ac.uk/spm/software/spm5/). Other diffusion measures, such as axial diffusivity and radial diffusivity were also derived from the diffusion scans. Axial diffusivity (λ_‖‖) represents the largest eigenvalue (λ₁) of the diffusion tensor, while radial diffusivity (λ_⊥) represents the average of the two minor eigenvalues (λ₂ + λ₃)/2 and mean diffusivity (MD) representing the mean of the three eigenvalues of the diffusion tensor (λ₁ + λ₂ + λ₃)/3.

Both magnetization transfer images, with magnetization prepulse (I _M) and without magnetization prepulse (I _O), were also coregistered to the T1‐weighted image with the same SPM5 algorithm. From these two MT images a magnetization transfer ratio (MTR) was computed using: MTR = (I _O − I _M)/I _O. This resulted in a value for each voxel between 0 and 1, 0 for no signal reduction and 1 for maximum signal reduction due to magnetization transfer.

The T1‐weighted image was normalized to MNI space using unified segmentation [Ashburner and Friston,2005] in SPM5.

Fiber Tracking

With DTI scans, four tracts: arcuate fasciculus (AF), cortico spinal tract (CST), cingulum (CGL), and uncinate fasciculus (UF) were reconstructed. These three additional tracts were chosen because of the directional properties they have in common with the AF, rather than their functional role in schizophrenia. Along these different tracts FA and MTR were sampled. Fiber tracking was performed using an implementation of the FACT algorithm [Mori et al.,1999] using in‐house developed software [Mandl et al.,2010]. A brute force approach was used for reconstruction of fiber tracts. The following parameters were used: eight seed points per voxel, minimum FA = 0.2, maximum angle = 53 degrees and maximum average angle with neighbouring voxels = 90 degrees. Then all reconstructed tracts were coregistered to the T1‐weighted scan with the transformation parameters acquired when coregistering the diffusion‐unweighted image to the T1‐weighted scan. Now all data was aligned with the T1‐weighted scan. Next, all reconstructed tracts were tagged with MTR and FA values from the respective images. Finally all fiber tracts that had been reconstructed in native space were transformed to MNI space with normalization parameters acquired as described previously.

Fiber bundles of interest were selected from the complete set of reconstructed tracts in MNI space using a multiple ROI approach [Wakana et al.,2004]. These ROIs were based on an average FA and fiber orientation map from all included subjects (see Fig. 1a). For AF tracts, two ROIs were used per hemisphere with a thickness of 3 mm and drawn in MNI space along two planes at Y‐coordinates −11 and −24. For UF, CG, and CST, two planes were drawn at Z‐coordinates −35 and 49, Y‐coordinates −29 and −2, and Y‐coordinate 6 and 15, respectively. To perform an unbiased selection of fiber tracts for both groups the same ROI was used per tract for all subjects, which was possible as all data is normalized. An exclusive ROI was placed between the hemispheres to remove possible tracts running from one hemisphere to the other.

Figure 1.

(a) Example of anterior and posterior ROIs for selecting the arcuate fasciculus. ROI was based on directional information retrieved from averaged diffusion group data. A ROI with 3 mm thickness was drawn along two planes at Y‐coordinates −11 and −24. For UF, CG, and CST, two planes were drawn at Z‐coordinates −35 and 49, Y‐coordinates −29 and −2, and Y‐coordinate 6 and 15, respectively. (b) Fibertracking results from a subject, displayed in normalized (MNI) space. From top row left arcuate fasciculus and left cortico spinal tract, bottom row left cingulum, and left uncinate fasciculus.

Statistical Analysis

Mean values of the corresponding sampled FA and MTR values were calculated for each fiber in the selected fiber bundle. From these tract means an overall bundle mean was computed (i.e., one value per bundle representing average MTR or FA per bundle). All data was analysed in SPSS 15.0. Mean MTR and diffusivity measures were analyzed with a MANOVA with group as fixed factor. Then group differences for each bundle were tested univariately with a Bonferroni correction for multiple testing. For mean FA we performed a MANCOVA, because it is known that FA decrease has been related to age [Rosenberger et al.,2008], we initially used this factor as covariate and group as fixed factor. A correlation analysis was performed with calculation of a pearson's r coefficient. Outliers were identified based on two standard deviation limits around the group mean.

To investigate differences in asymmetry we calculated a lateralization index (LI) for MTR values and FA values along the arcuate fasciculus according the following formula: . This resulted in two indices, one based on MTR and one based on FA for each group. These lateralization indices were evaluated with an ANOVA.

RESULTS

In line with earlier reports from our group [Daalman et al.,2011], although we used a smaller sample, we also found the same significant differences between patients and non‐psychotic individuals with AVH. Patients tend to report their auditory verbal hallucinations with a higher frequency [t = 7.063, P < 0.001, Cohen's d = 1.731], higher emotional valence [t = 10.056, P < 0.001, Cohen's d = 2.489], accompanied with more distress [t = 10.731, P < 0.001, Cohen's d = 2.697] and patients had less control (reflected as higher scores) over their AVH [t = 4.397, P < 0.001, Cohen's d = 1.078]. Also age of onset was significantly different between groups [t = 3.558, P = 0.001, Cohen's d = 0.933], non‐psychotic subjects with AVH have an earlier age of onset.

In four subjects (two subjects from the patient group and one subject from both healthy groups) not all tracts could be reconstructed. These subjects were excluded from the multivariate analysis. After sampling of FA and MTR values we had to exclude one subject from the non‐psychotic group with AVH and one subject from the patient group because mean MTR values for several tracts exceeded two times standard deviation. No difference in the number of reconstructed tracts for each fiber bundle was found between groups. See Figure 1b for an example of the four reconstructed fiber tracts.

MTR

For each tract the left and right average MTR values were pooled to one average value for each of the different tracts. These four variables were entered into a MANOVA with group as fixed factor. Multivariate testing retrieved no group effect. A univariate/between subjects effect test revealed a significant group effect for arcuate fasciculus values only [F(2, 101) = 4.720 P = 0.011, bonferroni corrected P < 0.0125, η_p ² = 0.085].

A second MANOVA with left and right MTR values of only the arcuate fasciculus as dependent variables resulted in a significant group effect [F(4,200) = 3.095, P = 0.017 Pillai's trace, η_p ² = .058]. A simple contrast (k matrix) was planned for each dependent variable. This standard contrast compared the control group with the patient group and the control group with the group of non‐psychotic individuals with AVH (see Fig. 2a).

Figure 2.

A Mean values and SD of MTR, FA, and RD measures, for all reconstructed tracts. (a) a For left AF; both psychotic and non‐psychotic subjects with AVH had increased MTR values. For the right AF, only in the patient group had increased MTR values. (b) No multivariate group effect had been found for FA values. Only a significant age effect. In general the patient group had decreased FA values. (c) 1c Radial diffusivity measures for the reconstructed tracts. Significant at P < 0.05.

For the left AF both patients and non‐psychotic individuals with AVH had higher MTR values [P = 0.022, Cohen's d = 0.539 and P = 0.015, Cohen's d = 0.614] as compared to the control group. This same contrast revealed a significant increased MTR in the right AF for only the patient group [P = 0.007, Cohen's d = 0.640] as compared to the control group and no difference between the controls and the non‐psychotic group with AVH [P = 0.206, Cohen's d = 0.337]. The contrast of non‐psychotic individuals with AVH and schizophrenia patients is not significant, left AF [P = 0.880, Cohen's d = 0.030] and right AF [P = 0.143, Cohen's d = 0.341].

FA

Mean left and right FA values were pooled to an average value for each tract. These values were used in a MANCOVA with age as a covariate. No group effect for mean FA was found. Univariate tests revealed a significant group effect for the arcuate fasciculus only [F(2,100) = 7.138, P = 0.001, bonferroni corrected P < 0.0125, η_p ² = 0.125] and UF [F(2,100) = 10.774, P < 0.001, bonferroni corrected P < 0.0125, η_p ² = 0.177, CST [F(2,100) = 5.221, P = 0.007, bonferroni corrected, P < 0.0125, η_p ² = 0.095], (see Fig. 2b).

The simple contrast applied for both left and right AF, CST, and UF compared patients and non‐psychotic individuals with AVH to control subjects. Significant differences were found in the patient group only. They had lower FA values in left arcuate fasciculus [P = 0.011, Cohen's d = 0.494], right CST [P = 0.003, Cohen's d = 0.636], and bilateral UF [P = 0.003, Cohen's d = 0.586 and P = 0.002, Cohen's d = 0.554].

Diffusivity Measures

A MANCOVA with age as covariate and group as fixed factor was performed for radial diffusivity (λ_⊥) values for left and right AF. This test revealed a significant age effect [F(2, 98) = 8.253, P < 0.001, η_p ² = 0.144] but no group effect. In general radial diffusivity values were higher in the patient group (see Fig. 2c). RD values for the left AF were significantly increased compared to the healthy control subjects [P = 0.011, Cohen's d = 0.511].

The correlation analysis (Pearson's r) revealed no significant association between AVH features and MRI measurements along left and right arcuate fasciculi for both groups with AVH. The associations of FA and MTR values with age for the three groups are presented as Supporting Information.

Finally, we have found no indication for asymmetry differences for FA and MTR values along the FA. The ANOVA revealed no group effects for LI with MTR values [F(2,100) = 1.068, P = 0.341, η_p ² = 0.021] and LI with FA values [F(2,100) = 2.183, P = 0.118, η_p ² = 0.042].

DISCUSSION

To disentangle the relationship between structural changes in the language network and the AVH phenomenon from other factors related to schizophrenia, this study compared integrity of the arcuate fasciculus between non‐psychotic individuals with auditory verbal hallucinations (AVH), schizophrenia patients with AVH and a non‐hallucinating control group. In both groups with AVH we observed an increase in MTR values of the arcuate fascicular tract, indicating a specific association with the tendency to hallucinate. For patients only, we observed a decreased fractional anisotropy in the arcuate fasciculus and in most other tracts. This suggests that decreased fractional anisotropy is not specifically related to AVH, but to other aspects in the disease or perhaps schizophrenia in general.

The interpretation of the increased MTR in the arcuate fasciculus of both groups with AVH is not straightforward, as the underlying pathophysiology of MTR deviations is not well understood. MTR is sensitive to concentrations of macromolecules such as myelin [Barkovich,2000; Henkelman et al.,2001; van Buchem et al.,1999; Wolff and Balaban,1994] but also to local amounts of free water [Henkelman et al.,2001]. Increased MTR in both groups with AVH could indicate increased myelination [Barkovich,2000] along the arcuate fasciculus. One could hypothesize that frequent AVH would make extensive use of the language networks and hence induce more extensive myelination of these heavily used white matter pathways [Bozzali and Wrabetz,2004]. However, in the arcuate fasciculus of schizophrenia patients we observed the combination of decreased FA and increased MTR, whereas increased myelin concentrations would yield higher FA values due to stronger diffusion anisotropy in the tracks of interest as a result of stronger myelination of axons. Therefore, the current finding can most likely be attributed to microstructural changes in the arcuate fasciculus, leading to more free water in the tracts resulting in increases in MTR and decreases in FA, as was observed for the patients. In the non‐psychotic individuals with AVH a similar MTR increase in the arcuate fasciculus was detected, but without any co‐occurring FA decreases. This may also be attributable to increases in free water, originating from microstructural changes within the white matter tract of the arcuate fasciculus, but to a lesser extent than in the patients, because an extensive increase in free water would also need to be reflected in a decrease in FA in non‐psychotic individuals. Indeed, the non‐psychotic group experienced less frequent and less severe AVH, which could be the result of milder alteration of the tracts microstructure. Therefore it could be that MTR is a more sensitive marker of subtle neurobiological alterations in the fiber tracts compared to fractional anisotropy.

Microstructural changes in the arcuate fasciculus probably indicate a degradation of the axons and/or glia cells in that tract, which may prevent effective corollary discharge from the inferior frontal speech regions to the temporo‐parietal language perception areas. Ineffective corollary discharge may lead to inadequate recognition of self‐generated verbal thoughts, thus provoking the experience of hearing voices. We expect that such microstructural changes occur.

Our findings indicate that similar, although less severe, neuronal deviations underlie AVH in non‐psychotic individuals with AVH as in patients with schizophrenia. A similar pathophysiological mechanism for AVH in both groups was also suggested by a recent fMRI study which revealed equal cerebral activation during AVH in schizophrenia patients and in non‐psychotic hallucinating individuals [Diederen et al.,2011]. The current finding adds to this conclusion, as microstructural disintegrity of the arcuate fasciculus was suggested in diseased as well as in non‐psychotic persons with AVH.

No other studies have assessed structural connectivity in the language network in non‐psychotic individuals with AVH. Yet, several authors studied tract integrity in comparable groups, such as persons at a high risk to develop schizophrenia, family members of schizophrenia patients and patients with schizotypal personality.

Several voxel‐based analyses [Hao et al.,2009; Hoptman et al.,2008; Munoz Maniega et al.,2008] revealed decreases FA values for non‐psychotic siblings of patients with schizophrenia, which suggest a genetic factor in FA aberrations. Voxel‐based and ROI studies in schizotypical individuals [Hazlett et al.,2011; Nakamura et al.,2005] revealed locally decreased FA values in this group as well. The finding of FA aberrations in these groups do not parallel our findings in the non‐psychotic individuals with AVH, suggesting that part of the pathology (perhaps related to negative and cognitive symptoms or the predisposition for these symptoms) is absent in the healthy spectrum of hallucinations.

There are a few limitations to the interpretation of the present findings, as it should be noted that MTR is an indirect reflection of white matter state in the brain. Although disrupted microstructure appears the most likely explanation for decreased MTR values in the arcuate fasciculus of both hallucinating groups, other deviations cannot be completely ruled out. Post‐mortem brain investigations or animal models of different types of white matter pathophysiology are necessary to demonstrate the relation between MTR deviations and histopathology.

In conclusion, non‐psychotic individuals with AVH display a similar deviation in the arcuate fasciculus as hallucinating patients with schizophrenia, namely increased MTR values. This might reflect degradation of the axons and/or glia cells in that tract. A similar pathophysiological mechanism, for example inadequate corollary discharge, underlies AVH in both groups and may therefore be a causal deficit in auditory verbal hallucinations.

Supporting information

Additional Supporting Information may be found in the online version of this article.

Supporting Information Figure 3