Neuroanatomical voxel-based profile of schizophrenia and bipolar disorder

Abstract

Although schizophrenia (SCZ) and bipolar disorder (BD) share elements of pathology (Ellison-Wright and Bullmore, 2009), the neural mechanisms underlying these disorders are still under investigation. Up until now, many neuroimaging studies investigated the brain structural differences of SCZ and BD compared with healthy controls (HC), trying to identify the possible neuroanatomical markers for the two disorders. However, just a few studies focused on the brain structural changes between the two diagnoses. The present review summarises the findings of the voxel-based grey matter (GM) comparisons between SCZ and BD, with the objective to highlight the possible consistent anatomical differences between the two disorders. While the comparisons between patients and HC highlighted overlapping areas of GM reduction in insula and anterior cingulate cortex, the SCZ–BD comparisons suggest the presence of more generalised GM deficits in SCZ compared with BD. Indeed, in a number of studies, SCZ patients showed lower GM volumes than BD patients in fronto-temporal cortex, thalamus, hippocampus and amygdala. Conversely, only a couple of studies reported GM deficits in BD compared with SCZ, both at the level of cerebellum. In summary, the two disorders exhibit both common and specific neuroanatomical characteristics, whose knowledge is mandatory to develop innovative diagnostic and treatment strategies.

The two major forms of psychosis, schizophrenia (SCZ) and bipolar disorder (BD), have been historically regarded as separate illnesses. However, the Kreaepelian dichotomous conceptualisation of the two disorders has been recently challenged by evidence of an intimate relationship. Indeed, SCZ and BD exhibit considerable overlaps in terms of genetic risk factors (Lichtenstein et al. 2009), clinical features (Fischer & Carpenter, 2009), neuropsychological impairment (Hill et al. 2013), as well as morphological brain changes compared with healthy controls (HC), including impaired white matter (WM) connectivity (Brambilla et al. 2005, 2009), ventricular enlargement and global brain volume reduction (Arnone et al. 2009). These similarities contribute to raise questions on the degree of distinctiveness of the two disorders.

A comprehensive understanding of the neurobiological characteristics of SCZ and BD may help to shed light on their common and specific pathophysiological bases. A large number of region-based and voxel-based approaches has already been applied to identify the structural abnormalities associated with SCZ and BD (Arnone et al. 2009; Ellison-Wright & Bullmore, 2010; Yu et al. 2010; Bora et al. 2012; Selvaraj et al. 2012). Whole-brain voxel-based morphometry (VBM) studies highlighted overlapping areas of grey matter (GM) reduction in SCZ and BD compared with HC, mainly located in bilateral insula and anterior cingulate cortex. The same studies also provided evidence for the larger extent and magnitude of the GM deficits in SCZ than in BD, suggesting a specific involvement of dorsolateral prefrontal cortex, superior temporal cortex, medial frontal gyrus, posterior cingulate cortex and thalamus in SCZ (Ellison-Wright & Bullmore, 2010).

Having said this, it has to be noticed that most of the current knowledge of the neuroanatomical differences between SCZ and BD relies on meta-analyses of comparisons between each group of patients and HC. There are just a few studies that compared GM structure between SCZ patients and BD ones. In the present review, we focus on the VBM studies directly comparing SCZ patients to BD type I patients by providing an overview of their findings, in order to shed light on possible unique anatomical underpinnings of the two disorders.

Ten studies met the criteria for inclusion (the numerosity and clinical characteristics of the SCZ and BD groups, the type of comparison (single v. multi centre) and the main findings of the studies are listed in Table 1). In these studies, SCZ and BD were compared between each other and to HC, as well as to schizoaffective disorder (SAD) patients in Ivleva et al. (2012, 2013); Amann et al. (2016), and to their first-degree relatives in Ivleva et al. (2013). It is worth noticing that Yüksel et al. (2012) included SAD patients in the SCZ group. Although the clinical characteristics of patients varied from study to study, the large majority considered chronic patients (except from Farrow et al. 2005) and BD patients with lifetime psychosis (except from Molina et al. 2011; Amann et al. 2016). The differences in clinical symptoms and pharmacological therapies between studies represent a confounding factor that should be taken into account when interpreting their findings.

Table 1. Selection of studies comparing SCZ and BD in terms of GM volume using voxel-based approaches

SC, single centre; MC, multi centre; SCZ, schizophrenia; BD, bipolar disorder; GM, grey matter.

Except of Cui et al. (2011), which did not find significant differences between the two disorders, and Farrow et al. (2005), Molina et al. (2011), which detected reciprocal GM deficits in the two groups, the other studies only found regions of GM reduction in SCZ compared with BD. Although the regions interested by these deficits were rather heterogeneous across studies, the overall results provide further proof of the greater GM damage associated with the SCZ pathology.

As mentioned above, volumetric deficits in BD compared with SCZ were detected only in two studies (Farrow et al. 2005; Molina et al. 2011). The singularity of these findings may be partially related to the much smaller number of BD patients compared to SCZ ones (8 of BD v. 25 of SCZ in Farrow et al. (2005), 19 of BD v. 38 of SCZ in Molina et al. (2011)) characterising the two datasets. In the follow-up study by Farrow and colleagues (2005), after 2 years from illness onset, BD patients showed less GM in the left temporal cortex, right occipital cortex and left cerebellum. Cerebellar deficits emerged also in chronic BD patients v. chronic SCZ patients (Molina et al. 2011). The authors additionally found lower GM volume in BD than in SCZ in left anterior cingulate, which is in line with the hypothesis that genetic risk for BD is associated with anterior cingulate deficits (McDonald et al. 2004).

With regard to the GM deficits associated with SCZ, a variety of cortical and subcortical regions emerged from the BD–SCZ comparisons. Three studies found in SCZ patients GM reductions at the level of cerebellum (Molina et al. 2011; Ivleva et al. 2013; Amann et al. 2016), and basal ganglia (McDonald et al. 2005; Brown et al. 2011; Ivleva et al. 2013).

A number of works reported hippocampal (McDonald et al. 2005; Nenadic et al. 2015a ; Brown et al. 2011) amygdalar (McDonald et al. 2005; Brown et al. 2011) and thalamic (McDonald et al. 2005; Molina et al. 2011; Ivleva et al. 2013; Nenadic et al. 2015a ) deficits in SCZ when compared with BD. Given the key function of these structures in learning, memory, attention and information transmission, the GM deficits of these structures in SCZ patients seem to be consistent with the relevant cognitive impairment associated with SCZ (Andreasen et al. 1994; Brambilla et al. 2013).

At the cortical level, a GM reduction in the frontal gyri was found to characterise SCZ compared with BD from the first phases of the illness (Farrow et al. 2005; McDonald et al. 2005; Molina et al. 2011; Ivleva et al. 2013; Nenadic et al. 2015a ). Some studies reported lower GM volume in SCZ than in BD in the temporal lobe, mainly in insula and temporal gyri (McDonald et al. 2005; Ivleva et al. 2013; Nenadic et al. 2015a , b ). A few works also reported occipito-parietal deficits associated with SCZ, mainly in lingual gyrus and precuneus (McDonald et al. 2005; Ivleva et al. 2012), as well as deficits in cingulum (Ivleva et al. 2013) and subgenual cortex (Yuksel et al. 2012). The widespread GM deficits emerged from these studies may be related to the lower WM metabolism in frontal, parietal and temporal areas characterising SCZ in comparison with BD (Altamura et al. 2013).

In summary, the findings of the above SCZ–BD comparisons suggest the presence of GM differences between the two disorders, mainly consisting of volumetric deficits in SCZ compared with BD. While a minority of studies found GM deficits in BD, repeatedly in the cerebellum, most of them detected in SCZ GM reductions in fronto-temporal cortex, thalamus and hippocampal-amygdalar region, supporting the hypothesis that fronto-striato-thalamic and temporal deficits are present in SCZ (McDonald et al. 2004).

The above GM changes may reflect partially different aetiologies, changes in the illness progression over years, different medication effects, or a combination of these factors. A plausible explanation comes from post-mortem examinations (Selemon & Rajkowska, 2003), which found in dorsolateral prefrontal cortex altered packing with increased neuronal density in SCZ, as opposed to decreased neuronal density in BD, suggesting specific anatomical underpinnings for the two disorders. Future research in this direction, using novel morphometric parameters (such as local gyrification (Nenadic et al. 2015 b ) and labelled cortical distance (Ratnanather et al. 2014)) and advanced multimodal processing techniques (such as support vector machine algorithms), opens the door to the development of instruments with higher diagnostic specificity. Significant evidence on SCZ and BD can also come from trans-diagnostic analyses that look at common dimensions of functioning across the two disorders (e.g., Goodkind et al. 2015), in line with the recent Research Domain Criteria.