Our findings confirm an association of MMN reduction recorded at Fz in schizophrenia with MRI-derived measures of cortical gray matter estimates in areas subserving auditory processing (eg, Heschl's gyrus; compare Salisbury et al19
). The most robust correlation was found for MMN measures derived from frequency deviants suggesting that frequency MMN is more linearly associated with pathological changes in schizophrenia than duration or intensity MMN. We have previously raised the question of whether deviance in different sound features could provide complimentary evidence of the pathology underlying MMN reduction in schizophrenia. The observation of Salisbury et al19
suggests that at least some aspect of this pathology is progressive and that the frequency MMN amplitude change is accompanied by a change in the gray matter volume of auditory processing areas. Our present findings are certainly consistent with this observation and with our previous findings showing a progressive decline of MMN in the course of illness predominantly in response to frequency but not to duration and intensity deviants.25
When different forms of deviant are used, there are many factors that complicate a comparison between the resultant MMNs elicited (see Todd et al25
for discussion). However, we hypothesize that it is likely that the multiple tonotopic representations of frequency information in auditory cortex could explain why gray matter volume is more linearly associated with frequency MMN in the present study. These multiple representations might afford frequency MMN some degree of resilience to initial disruption of auditory processing, but as the disruption spreads engulfing other tonotopic representations, frequency MMN declines, rendering it particularly sensitive to a linear progression of the pathology.25
That frequency-derived MMN in particular is closely associated with bilateral gray matter reduction in Heschl's gyrus in schizophrenia suggests a neuropathology that affects the integrity of the tonotopic organization of frequency representation and sound processing in primary auditory cortex with progression of illness. These findings are consistent with observations that perceptual resolution of frequency differences (frequency discrimination) is impaired in schizophrenia42
and becoming more severe with chronicity of illness.43
While our findings suggest that frequency MMN is more closely associated with illness chronicity than MMN to duration and intensity deviants, it is possible that our sample size did not provide sufficient power to detect potentially weaker associations of gray matter deficits, for instance, with intensity-elicited MMN. It is also possible that the reduction in MMN to intensity (and potentially duration) change is not as well suited to linear association with progressive pathology. There is evidence that these features are represented in a distributed fashion across the auditory cortex,44
anterior cingulate, and presupplementary motor area45
and as such, could be less resistant to early signs of the pathology or simply less linearly affected.
As age and duration of illness are highly correlated, and gray matter volume declines with age over the human life span,46
reduction of MMN amplitude and its association with reduced gray matter could be confounded by normal ageing. Our findings, however, clearly indicate that pairwise age- and gender-matched healthy control subjects do not show an association of MMN with gray matter. Furthermore, frequency MMN is correlated with age only in patients and not in healthy participants. Hence, the association of frequency-elicited MMN with gray matter measures appears to be specific to progressive neuropathology in schizophrenia and not driven by genuine age-related variability across the 2 measures when accepting age as a proxy measure of duration of illness.
Medication is another important factor when interpreting our findings. Both human as well as animal research47,48
suggest reduction of gray matter with haloperidol exposure and to exposure to second-generation antipsychotics, though to a lesser extent when compared with first-generation antipsychotics.17,49,50
However, there is no consistent evidence that antipsychotic medication affects MMN generation (eg, Catts et al51
; Schall et al52
), although a relatively intact MMN to frequency deviants predicts a favorable response to clozapine treatment in chronic schizophrenia patients who fail to respond to other antipsychotics.53
Nonetheless, more systematic pharmacological research, including animal models, is required to establish whether MMN generation is unaffected by exposure to antipsychotic medication and whether medication effects on gray matter are linked to illness-related changes of cortical gray matter.
Some recent anatomical evidence from postmortem data is consistent with impaired auditory information processing in schizophrenia.54
The authors stereologically quantified the density of synaptophysin-immunoreactive axon terminals in the gray matter of Brodmann areas 41 and 42. Feedback auditory pathways in layer 1 of area 41 and layer 3 of area 42 were intact in schizophrenia, while reduced terminal densities were found in feed-forward pathways of layer 3 in area 41. This may affect the spread of activation within primary auditory cortex in response to acoustic stimuli and, in turn, may affect the performance of neural circuits involved with sound discrimination and MMN generation.
Importantly, the authors found no evidence that long-term exposure to antipsychotic medication reduces synaptophysin-immunoreactive axon terminals in nonhuman primates.54
If confirmed by more data, this finding would be consistent with a lack of effect of antipsychotics on MMN generation in schizophrenia. However, further animal research is required to better understand the neurocircuitry of auditory mismatch processing and how individual pathways contribute to auditory mismatch generation.
Our data revealed that associations between frequency-elicited MMN and reduced gray matter measures in schizophrenia were not limited to auditory areas; they also extended into other cortical areas involved in motor organization and executive function. The topographic distribution of MMN/gray matter associations in schizophrenia is consistent with current source density studies (eg, Shalgi and Deouell22
) and functional brain imaging data (e.g., Dittmann-Balcar et al20
; Schall et al21
) implicating temporal and frontal cortical areas in the generation of MMN. Such an association was not detected in healthy control subjects or for duration and intensity-elicited MMN in schizophrenia. In this respect, associations of MMN in response to frequency deviants with frontal gray matter reduction indicate a more widespread brain pathology—beyond auditory cortex—that is linked to MMN reduction with longer duration of illness.
We also confirmed earlier reports of a correlation of smaller MMN amplitudes with poor sociooccupational functioning in our schizophrenia sample,24
although in our sample, this correlation was once again most robust for frequency MMN. However, one explanation of this relationship that emerges from our data is that the relationship between decreased MMN and impaired functional skills occurs, not because these deficits are causally linked but because they are both related to frontal lobe pathology.
In summary, the current study shows that MMN is a useful tool to investigate important features of the disorder, such as progression of illness, clinical outcomes, and potentially treatment response, as well as the importance of the choice of the physical characteristics of the deviant stimuli when conducting MMN research in schizophrenia. MMN also holds the potential to provide more profound insights into the underlying pathophysiology by identifying core neurophysiological deficits of the disorder, how they develop, and how they give rise to the clinical features of the condition.