In this study, three different whole-brain voxel-based imaging techniques identified the medial prefrontal cortex as a prominent site of abnormality in schizophrenia. This region has not previously been a focus of interest in the disorder, although some post-mortem studies have noted microstructural abnormalities in the anterior cingulate cortex,39
and it is an area where apparent hyperactivation,12
and more recently failure of deactivation14, 15
has been found (see below). We also found evidence of convergent abnormality in the dorsolateral prefrontal cortex, which is an area of long-standing interest in schizophrenia. Here, however, the laterality was less consistent across techniques.
Our findings differ from two other studies that have applied whole-brain multimodal imaging to schizophrenia.19, 20
Calhoun et al.19
examined 15 chronic schizophrenic patients and 15 controls using VBM and fMRI while performing an auditory oddball task. They used joint independent component analysis (jICA) of both modalities to determine areas of difference between the groups. Significant differences for VBM and fMRI were located in clearly different cortical regions, and there was no anatomical convergence. Lui et al.20
carried out VBM on 68 patients with first-episode schizophrenia and 68 controls. Similar to our study, they found a cluster of grey matter volume reduction in the anterior cingulate cortex (although on the right only), along with two other clusters in the right superior and right middle temporal gyri. However, when the authors used these three regions as seeds for a functional connectivity analysis using resting fMRI, no differences between the patients and the controls emerged.
Our findings are also different from those of other studies using single modalities of brain imaging. In particular, we found a considerably less extensive pattern of brain abnormality in the VBM and DTI analyses than in previous studies. On the one hand, this undoubtedly reflects our use of a conservative threshold of P
=0.01 corrected. Thus, while two recent meta-analyses of VBM studies in schizophrenia5, 40
found, similar to this study, clusters of reduced volume in the medial prefrontal cortex and the dorsolateral prefrontal cortex, these were accompanied by other clusters, notably in the insula/inferior frontal cortex and medial temporal lobes. Some of these additional areas were also evident in our analysis using a threshold of P
=0.05 (see Supplementary Material).
On the other hand, there are grounds for believing that VBM abnormality in schizophrenia may not be as widespread as originally thought. When the technique was originally introduced in 2000–2001,41, 42
it utilized a measure of grey matter concentration or density, the proportion of grey matter within a given voxel after spatial normalization of the images. Later, a modulated or ‘optimized' method became available, which gives a more intuitive estimate of grey matter volume. Fornito et al.40
meta-analysed studies using measures of grey matter concentration in schizophrenia and found clusters of significant reduction in the medial and lateral frontal cortex, temporal cortex and insula bilaterally. However, meta-analysis of studies using the volume measure resulted in a more restricted pattern of differences: changes in the temporal lobe and insula were no longer seen, and the largest areas of reduced volume were in the medial aspect of the left superior frontal gyrus plus the left orbitofrontal and fusiform regions.
DTI studies of schizophrenia have found evidence of abnormality in many different white matter areas, although these include the corpus callosum,17
especially in recent studies.21, 43
Once again, however, there are grounds for believing that the true pattern may be more restrictive: Ellison-Wright and Bullmore44
meta-analysed 15 studies which used a whole-brain voxel-based approach, and found that only two locations were consistently identified across studies. One of these was in the deep white matter of the left frontal lobe, close to the genu of the corpus callosum. The other area was in the left temporal lobe white matter. These areas are strikingly similar to those we found using a threshold of P
The linchpin for the convergence of evidence in our study is the failure of deactivation in the medial frontal cortex, which was equally prominent at thresholds of P
=0.05 or 0.01. One of the most important functional imaging findings to emerge in recent years has been that prefrontal cortex dysfunction in schizophrenia goes beyond any simple concept of hypofrontality, with a series of studies documenting areas of increased activation (‘hyperfrontality') during performance of working memory tasks.7, 8, 9, 10
The leading interpretation of this finding has been that of cortical inefficiency: even when schizophrenic patients are able to perform a cognitive task normally, they have to ‘work harder to keep up' with its demands, and this leads to a compensatory brain functional response characterized by greater and/or wider activation of the dorsolateral prefrontal cortex than in healthy subjects.10, 13
However, a meta-analysis of fMRI studies using the n
casts doubts on this interpretation, as (a) hyperfrontality was found in the medial prefrontal cortex, not the dorsolateral and other lateral areas where hypofrontality was seen; and (b) this medial frontal area was not activated in the meta-analysis of either patients alone or controls alone. Subsequently, it has been argued that the hyperfrontality seen in the medial frontal cortex in schizophrenia actually represents a failure to deactivate14
(because of ‘reverse subtraction' from a high baseline, this can result in apparent hyperactivation, see Gusnard and Raichle45
What is the function of this medial prefrontal area identified by multimodal imaging? As mentioned above, it includes, but is not limited to, the anterior cingulate cortex, an area which has been implicated in mood, attention, emotional regulation, error detection and decision making.39, 46
Perhaps more significantly, it corresponds closely to one of the two midline nodes of the default mode network, a system of brain regions that are active at rest but deactivate during performance of a wide range of attention demanding cognitive tasks.16
The default mode network has been proposed to be involved in functions such as conceiving the perspectives of others, retrieving autobiographical memories and envisioning the future, processes that can be subsumed under a broad heading of mentation detached from the external world.47
Going further, it has been argued that such functions underlie the experience and maintenance of one's sense of self.48
A further possibility is that the default mode network underlies a state of ‘watchfulness', a passive, low-level monitoring of the external environment for unexpected events in conditions when active attention is relaxed.47
Any and all of these possibilities seem to have scope for explaining the symptoms of schizophrenia.