Cingulum bundle integrity in schizophrenia with auditory verbal hallucinations: A diffusion tensor imaging tractographic study

Diffusion tensor imaging (DTI)-based neuroimaging studies in schizophrenia have consistently reported disruptions of several white matter tracts in individuals with schizophrenia1,2. However, only a few studies have assessed the white matter disruptions in relation to specific psychotic symptoms such as auditory verbal hallucinations (AVHs)3,4. Understanding the neural mechanisms for AVH is of considerable interest and requires a symptom-based approach5.

The cingulum bundle (CB) is a major white matter tract that connects the neocortex (e.g. cingulate gyrus) with the limbic system, possibly acting as a bridge between emotion and cognition6. Initially thought to be a unitary tract, more recently, it has been observed that fibers from the cingulum bundle project posteriorly to entorhinal cortex and temporal lobe (posterior cingulum) and anteriorly to pre-motor, pre-frontal regions and the striatum (anterior cingulum)7,8. The disconnection seen in these fibers, which connect the limbic system and the cortical regions, could lead to internally self-generated thoughts being misinterpreted as externally generated sensory stimulus. This in turn could be a contributory factor towards development of psychotic symptoms9–13.

Reduced integrity of cingulum, especially anteriorly, has been frequently reported in studies on schizophrenia14–18. Not much literature is, however, available on its relationship with AVH. The PubMed/Medline-based search [using search terms ‘Diffusion Tensor Imaging’ (Mesh) AND auditory hallucinations (All Fields) AND cingulum (All Fields)], supplemented by cross-reference search from relevant reviews9,19, yielded only a few relevant studies20–25. Further, available studies reporting such an association have inferred it from correlations between AVH scale scores and reduced integrity of CB as part of their analysis, rather than pre-defined symptom-based groups with AVH as a defining criterion.

The present study was aimed to compare the integrity of CB, using quantitative tract measurements of fractional anisotropy (FA), among AVH-positive, AVH-negative schizophrenia patients and healthy controls.

Material & Methods

Study participants were recruited from the outpatient clinic and inpatient settings from the department of Psychiatry, All India Institute of Medical Sciences, New Delhi, India, and their DTI assessments were carried out at the department of Neuroimaging and Interventional Neuroradiology at the same institute. The data were collected from July 2015 to July 2016. The study was approved by the Institutional Ethics Committee and a written informed consent was obtained from all the participants who fulfilled the inclusion criteria. A part of the study was conducted on two other fasciculi (arcuate and superior longitudinal), the findings of which have been described elsewhere4.

Study groups and selection criteria: The study sample comprised of three study groups, viz. group I-AVH-positive schizophrenia, group II AVH-negative schizophrenia and group III healthy controls, with pre-defined selection criteria, as described below.

Right-handed individuals of either gender, aged between 18 and 50 yr, with Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5)26 diagnosis of schizophrenia with total duration of illness of at least two or more years were selected. Group I included individuals with schizophrenia who had significant lifetime/current AVH (operationally defined as hallucinations present for at least 50 per cent of days in current month and in previous exacerbation/s of illness). Group II included individuals with schizophrenia with no lifetime AVH [operationally defined as a score of 0 for auditory hallucinations on scale for assessment of positive symptoms (SAPS) along with no history of ‘hearing voices’ as per patient and corroborated from family and treatment records, wherever possible]. Group III consisted of healthy controls with no psychiatric diagnosis as per Mini International Neuropsychiatric Interview I (MINI) version 7.0.027 and no family history of psychiatric illness. Healthy controls were selected from non-biological relatives and acquaintances of patients after an informed consent.

Patients with an ambiguous or transitory history of AVH, current history of any major psychiatric disorder (other than schizophrenia in groups I and II) and lifetime history of any major psychiatric disorder (other than schizophrenia and major depression in groups I and II) and nicotine dependence in all groups were excluded from the study. Those with history of receiving electroconvulsive therapy (ECT) or repetitive transcranial magnetic stimulation (rTMS) or any other neuromodulatory treatment were also excluded from the study. Those with intellectual disability and significant neurological or medical illness were excluded in addition to standard magnetic resonance imaging (MRI) exclusions (e.g. metallic implants).

Screening and clinical assessment: Patients were screened using MINI 7.0.027 and Edinburgh’s handedness inventory28. For clinical assessment, a semi-structured proforma was used for relevant sociodemographic and clinical variables. Clinical assessments included SAPS29, scale for assessment of negative symptoms (SANS)30, clinical global impression-schizophrenia-severity31 and psychotic symptom rating scale (PSYRATS)32. Assessment instruments were applied in a single session. The DTI was done within one week of clinical assessments by prior appointment.

MRI data acquisition: A 3T (Tesla) MRI (Ingenia; Philips Medical Systems, Netherlands) was performed by using a 32-channel head coil. DTI was performed using a single-shot echo-planar sequence with the array sensitivity – encoding technique. Motion-probing gradient orientations were applied along at least 32 directions, and the b factor was set at 1000 sec/mm². Acquisition parameters were repetition time (TR)/echo time (TE), 8000/83.3 msec; bandwidth, 143 kHz; matrix size, 128 × 128; section thickness, 2 mm without intersection gap; field of view, 30 × 30 cm and number of excisions, 2.

MRI data analysis: The DTI datasets were processed using DTI studio version-3.033 (developed and distributed by Prof Mori’s laboratory) after transfer to a computer with windows platform. The details of MRI analysis have been described elsewhere4.

Statistical analysis: Data analysis was carried out using Statistical Package for the Social Sciences (SPSS) version-20.0 (IBM Corp. Armonk, NY, USA). Power analysis for one-way ANOVA was conducted in G-POWER34, to determine a sufficient sample size using an alpha of 0.05, a power of 90 per cent, an effect size of 0.4 (using Cohen’s convention) and two tails. There was an equal allocation of participants into each group. Based on the aforementioned assumptions, the desired sample size came out to be 84. To keep a margin for visual artefacts, or wasting of MRI films, a margin of 10 per cent was taken, which came to a total sample size of 84+8=92. Hence, a sample of at least 30 patients in each group was planned as part of the study. FA values were exported from DTI studio and analyzed using SPSS. The Shapiro-Wilk test was done to test for normality of the data (Supplementary Tables I–V). For the descriptive data, means, standard deviations, frequencies, percentages, medians and ranges were used to describe the data. For finding associations and correlations, appropriate parametric and non-parametric tests were applied.

Results

A total of 92 patients were recruited (30 each in group I and III, and 32 in group II). Mean age in the three groups was 32, 27.5, and 27 yr, respectively. Males predominated in group I and II (63 and 70%, respectively) while group II had equal number of males and females. The differences in the gender distributions were, however, not significant.

Sociodemographic and clinical variables: Table I shows the sociodemographic and clinical profile of the three groups. No significant difference was found across three study groups with respect to age, gender and education. The SAPS score was different across AVH and non-AVH groups (7.1+2.6, 3.5+1.9; U=116.500, P<0.001). No other differences were found in the clinical variables between groups I and II, apart from illness duration (group I: 9.6+7.3 yr, group II: 5.5+4.1 yr; U=331.000, P=0.035).

DTI data: Figures 1A and B depict, the CB as per DTI studio tractography technique. Table II shows the FA values in bilateral CB. FA values showed significant difference across the three groups in the left cingulate gyrus part of CB (CB-CG) (H=10.342; df=2; P=0.006). Post-hoc pair-wise comparisons revealed that FA was significantly lower in group I compared to group II which in turn had no significant difference as compared to group III.

Close

Fig. 1

DTI tracts formed by tractography on DTI studio software. (A) Cingulate gyrus of the right cingulum bundle, (B) Hippocampal part of the right cingulum bundle. DTI, diffusion tensor imaging.

Export to PPT

Figure 2 shows the error plots of FA values in CB-CG and hippocampal part of CB (CB-H) across the three groups. Group I showed significantly lower FA values in the left CB-CG as compared to groups II and III. There was no point of overlap between group I and other two groups (Fig. 2D) depicting significant difference of group I from the two groups (groups II and III). There was no significant difference in FA values between the groups II and III. The relationship between FA values and clinical variables within groups I and II was explored. No relationship was found between FA values and age of individuals, severity of psychosis/hallucinations, total duration of illness or treatment duration.

Close

Fig. 2

Error plots of comparative fractional anisotropy values in cingulate gyrus (CG) and hippocampal (H) part of cingulum bundle between the three groups; group I (schizophrenia with AVH), group II (schizophrenia without AVH), group III (healthy controls). (A) right cingulum bundle (CG), (B) right cingulum bundle (H), (C) left cingulum bundle (H), (D) left cingulum bundle (CG). AVH, auditory verbal hallucination.

Export to PPT

Discussion

The present study is one of the few DTI studies to report reduced white matter integrity of CB in AVH positive schizophrenia compared to schizophrenia without AVH and healthy individuals. The study has attempted to overcome some of the previous methodological issues in similar studies14–18,20–25. The study included individuals with DSM-5 diagnosis of schizophrenia whose illness duration was minimum of two years or more to ensure diagnostic stability and has used rigorous operational criteria to either establish or rule out lifetime AVH in the respective study groups, which included both disease controls and healthy controls in addition to AVH positive schizophrenia. The present sample size of 92 exceeds that in several previous DTI studies on neural correlates of hallucinations (samples varying between 20 and 66)20,21,23,25,35–38. Further, DTI applied motion-probing gradient orientations along 32 directions which acquire more precise estimations of anisotropy and directionality of tensor8.

The present study found a reduced connectivity in the left CB-CG in AVH-positive schizophrenia based on a quantitative tract-based measurement of FA, a marker of reduced microstructural integrity of white matter axonal tracts. The FA value in the left CB-CG of AVH-positive schizophrenia was decreased by 6.19 per cent compared to AVH negative schizophrenia controls and by 6.17 per cent compared to healthy controls. The FA value was not significantly different between AVH-negative schizophrenia and healthy controls.

The main finding from the present study is that of a reduced connectivity (low FA) in the left CB-CG in current sample of AVH-positive schizophrenia. This finding broadly concurs with the findings from prior studies that have observed a significant association between hallucination scores and CB disruptions (i.e. reduced FA)20,23,24, though a few studies reported no association21 or even conflicting results22,25. Similar to the present study, Ćurčić-Blake et al23 also reported a decreased FA in the left cingulate, among few other tracts, in AVH-positive schizophrenia. Further, the white matter integrity correlated negatively with severity of AVH in that study23. Whitford et al24 found that a lower FA in the right dorsal anterior cingulum correlated with patient scores of hallucinations. Zhang et al20 demonstrated decreased FA values in multiple tracts including bilateral cingulum in 21 first episode schizophrenia patients with AVH and 12 chronic schizophrenia patients with AVH, compared to 26 healthy controls. Unlike the current study, they did not compare differences between hallucinators and non-hallucinators. Bopp et al21 found an alteration in the anterior part of the right CB in schizophrenia; however, FA values did not show a significant correlation with AVH scores21. In another study (33 schizophrenia patients and 40 healthy controls), the propensity to experience AVH in schizophrenia patients was significantly associated with an increased FA, rather than decreased FA, in the anterior cingulum25. Another study carried out on 88 schizophrenia patients and 40 healthy controls reported a positive correlation between FA values in anterior CB and ‘positive’ symptom score (delusions, hallucinations, bizarre behaviour), but it did not assess hallucination score separately22. Coming to correlations within group I itself, the CB-FA values were not correlated to severity of hallucination scores (PSYRATS–H). There is no singular explanation which can be put forth in definite terms for this particular finding. The study group I had an overall higher hallucination severity as a group, probably as patients were recruited from a tertiary care centre. It is possible that the relationship between CB-FA values and hallucination severity scores may not be a simple, linear relationship or that relationship may be potentially influenced by other clinical or neuroimaging variables pertaining to other brain areas. Finally, it is also possible that the reduced connectivity (low FA) in the left CB-CG is a marker for AVH-positive schizophrenia, though not necessarily a marker of severity of AVH.

Cingulate gyrus is a principal driver of the cingulum which has been linked to several brain functions. The functions most strongly linked to different parts of the CB are emotion, motivation, executive functions, including aspects of attention, pain and memory7. Allen et al9 showed that the hallucinator group made more external misattributions and showed altered activation in the anterior cingulate, besides superior temporal gyrus, compared with non-hallucinators and controls. A cingulate-based deficiency has been proposed to be related to an inability to distinguish one’s own internal thoughts from external events, which form a possible basis for hallucinations11. Further studies are required to ascertain its specific role in generation of AVH.

CB being a large tract, with various fibres joining and leaving from, may be affected by a number of disorders in addition to schizophrenia, such as mood disorders39–40. In the present study, psychiatric comorbidities (other than lifetime depression) were excluded carefully using MINI 7.0.0. Lifetime depression (as per MINI 7.0.0) was found to be comparable between AVH-positive and AVH-negative schizophrenia with no statistically significant difference and is unlikely to confound or affect the study findings.

The study findings must be interpreted in light of certain limitations as discussed below. The AVH group had a relatively longer duration of illness compared to non-AVH group, though their severity was comparable and FA values did not correlate with illness duration. In DTI studio, fibre tracking was performed based on deterministic tractography without taking crossing-fibre into considerations. Furthermore, it is difficult to comment on whether these findings are part of a global white matter disruption or not, since the study did not use a whole-brain approach. The role of any other possible factors, including nutrition or medication, could not be accounted for. All participants with schizophrenia were on prescribed antipsychotic medications and drug-naive sample was difficult to recruit in view of ethical reasons. The duration of untreated psychosis or lifetime antipsychotic equivalents could not be calculated for either group, as many patients or their caregivers did not recall the names, doses or duration of medications taken by them in the past or did not bring or had lost older treatment records. Consequently, there was a limited access to information as far as past medication history was concerned. However, there are studies showing no effect of antipsychotic medication on white matter integrity21,41. The data derived from DTI may be influenced by magnetic susceptibility effects, and motion artefacts, all of which may be influenced by the acquisition parameters. Having another disease control group (e.g. depression with AVH) in future may help to study AVH across multiple diagnosis.

To conclude, a significantly reduced FA in CB-CG emerged as a possible marker for AVH in schizophrenia. Reduced integrity of the left CB-CG should be investigated further for its clinical relevance in the context of global findings on white matter integrity in AVH-positive schizophrenia. These findings in larger samples and need to be replicated in diverse diagnosis with AVH as a cross-cutting characteristic in the future. Such studies might help determine the potential targets for interventions in patients with persistent AVH and may raise the possibility of subtyping schizophrenia in newer ways.