Background
Schizophrenia patients with auditory verbal hallucinations (AVH) have reduced structural integrity in the left arcuate fasciculus (AFL) compared to healthy controls. However, it is neither known whether these changes are specific to AVH, as opposed to hallucinations or schizophrenia per se, nor how radial and/or axial diffusivity are altered. This study aimed to test the hypothesis that reductions to the structural integrity of the AFL are specifically associated with AVH in schizophrenia.
Method
Diffusion tensor imaging scans and clinical data were obtained from the Australian Schizophrenia Research Bank for 39 schizophrenia patients with lifetime AVH (18 current, 21 remitted), 74 schizophrenia patients with no lifetime AVH (40 with lifetime hallucinations in other modalities, 34 no lifetime hallucinations) and 40 healthy controls.
Results
Fractional anisotropy was significantly reduced in the AFL of patients with lifetime AVH compared to both healthy controls (Cohen's d = 1.24) and patients without lifetime AVH (d = .72), including compared to the specific subsets of patients without AVH who either had hallucinations in other modalities (d = .69) or no history of any hallucinations (d = .73). Radial, but not axial, diffusivity was significantly increased in patients with lifetime AVH compared to both healthy controls (d = .89) and patients without lifetime AVH (d = .39). Evidence was found for a non-linear relation between fractional anisotropy in the AFL and state AVH.
Conclusion
Reduced integrity of the AFL is specifically associated with AVH, as opposed to schizophrenia in general or hallucinations in other modalities. Increased radial diffusivity suggests dysmyelination or demyelination of the AFL may play a role in AVH.
Auditory verbal hallucinations (AVH) are proposed to result from altered connectivity between frontal and temporal left hemisphere language regions (Feinberg, 1978, Frith, 1992, Ford et al., 2007 and Whitford et al., 2010). A key white matter tract linking these regions is the left arcuate fasciculus (AFL), which connects the inferior frontal lobe, including Broca's area, to the posterior superior temporal lobe, including Wernicke's area (Catani et al., 2007). Diffusion tensor imaging (DTI) has examined whether structural alterations to the AFL are associated with AVH. The most commonly employed measure in DTI is Fractional Anisotropy (FA), an index of the structural integrity of white matter fibres, with lower FA indicating reduced integrity.
A recent meta-analysis by Geoffroy et al. (2014) found reduced FA in the AFL (henceforth FA-AFL) of schizophrenia patients with AVH, compared to healthy controls. Because such comparisons cannot demonstrate specificity of changes to AVH, which could instead be due to schizophrenia per se, the authors recommended more studies comparing patients with and without AVH. The only two studies to do this (Catani et al., 2011, 28 first-episode psychosis patients with AVH, 18 without; Benetti et al., 2015, 17 chronic schizophrenia patients with AVH, 11 without) both failed to find significant differences. However, they only had sufficient power to detect large effect sizes, potentially causing Type II errors. For example, Benetti et al.'s data shows an FA-AFL reduction in the AVH group of a medium effect size. The first aim of our study was to test the hypothesis that FA-AFL is lower in patients with schizophrenia with AVH, compared to both patients with schizophrenia without AVH, and healthy controls, in a study powered to detect medium effect sizes.
A second recommendation of Geoffroy et al. (2014) was that DTI studies of AVH should report on radial diffusivity and axial diffusivity, as these are associated with specific biological changes. Increased radial diffusivity is associated with reduced myelination (Song et al., 2005), reduced axial diffusivity with axonal damage (Song et al., 2003), although caution is needed in such interpretations in the absence of post-mortem examination (Wheeler-Kingshott and Cercignani, 2009). Radial but not axial diffusivity has been found to be higher in schizophrenia patients with AVH compared to healthy controls (de Weijer et al., 2011), yet schizophrenia patients with AVH have also been found to have higher radial diffusivity than non-psychiatric patients with AVH, suggesting a lack of specificity to AVH (de Weijer et al., 2013). The second aim of our study was hence to test the hypothesis that schizophrenia patients with AVH have higher radial (but not axial) diffusivity than both patients without AVH, and healthy controls.
We also aimed to undertake more exploratory analyses. First, given the role of the AFL in language (Catani et al., 2005), we hypothesised decreases in FA-AFL in patients with AVH compared to patients with hallucinations restricted to other modalities (e.g., visual hallucinations). Second, as some previous studies have employed patients with both current and remitted AVH (e.g., Benetti et al., 2015), we aimed to test whether FA-AFL changes associated with AVH reflected a state or trait phenomena. Finally, we aimed to undertake the first empirical test based on the proposal that if AVH represent the brain's attempt to incorporate disjointed neural activity then this integration may only be possible up to a certain level of temporal desynchronisation (Whitford et al., 2010). Whilst intermediate asynchronies could result in AVH, severe asynchronies may not be able to be incorporated into a coherent auditory experience, thus not giving rise to AVH, and mild asynchronies may integratable into the flow of normal experience (Whitford et al., 2010). We hence aimed to test the hypothesis that, within schizophrenia patients, changes in FA-AFL relative to healthy controls would have a non-linear (quadratic) relationship to the probability of current AVH
