BackgroundMismatch negativity (MMN) is an event-related potential (ERP) that is severely impaired in long-term schizophrenia. However, though largely debated, numerous electroencephalography (EEG) studies suggest that MMN is less reduced in individuals at their first episode of psychosis (FEP). Pitch MMN (pMMN) is not reduced, although duration MMN (dMMN) may be reduced compared to healthy controls (HC), and may show left hemisphere deficits early on. Few MMN studies in FEP have utilized magnetoencephalography (MEG). MEG could potentially be more sensitive to circumscribed differences, given its ability to more effectively localize MMN sources within the auditory cortex. Using MEG, we compared pMMN and dMMN between FEP and HC to identify potential source location- and timing-dependent differences in early disease course.MethodsSimultaneous MEG and EEG recordings of pMMN and dMMN were obtained with an oddball paradigm from 15 FEP and 19 HC. Of those, 12 FEP and 11 HC had usable EEG recordings, with the remainder lost due to a circuit board failure. MMN was calculated by subtraction of the sensor-space average to the standard tones from the average to the deviant tones. For EEG, mean amplitudes across fronto-central electrodes (F1, Fz, F2, FC1, FCz, FC2) were calculated over a 30 msec window at group peak latency. For MEG data, source activity was determined in Brainstorm using minimum norm estimation, multisphere head models, and individual MRI cortical surface reconstructions. Using the HCP Brain Atlas, regions of interest (ROIs) were determined based on areas of peak activity at two time windows for each task: an Early time window of 105–115 ms for both pMMN and dMMN, and a Late time window of 125–135 ms for pMMN and 155–165 ms for dMMN. Mean amplitudes were calculated over six auditory cortex ROIs: A1, Brodmann’s Area 43, Brodmann’s Area 52, Lateral Belt, Medial Belt, and Parabelt. Mixed-model ANOVA was then used to compare groups, regions, hemispheres, time windows, and tasks using SPSS software.ResultsFor EEG peak MMNs, there were no significant differences between groups in pMMN or dMMN (p’s > 0.1). For MEG sources, however, ANOVA indicated a Region x Hemisphere x Group interaction (p = 0.038), but no significant regional differences between Pitch and Duration or for either Early or Late time windows (p’s > 0.1). When ROI data were averaged across tasks and time windows, the right hemisphere showed no difference between groups, whereas the left hemisphere showed a trend for reduced MMN activity in FEP (p = 0.069). Post-hoc comparisons of the six left hemisphere ROIs revealed that all regions but Brodmann’s Area 43 showed significant reduction in FEP (p’s < 0.05).DiscussionThese findings support previous literature that indicate relatively preserved EEG scalp measures of MMN in FEP. However, the MEG source data indicate a deficit in the left hemisphere auditory cortex of FEP. This left hemisphere reduction was observed in 5 out of 6 source regions and at both time windows. This suggests broad left hemisphere auditory cortex pathophysiology early in disease course to which EEG may be relatively less sensitive by comparison with MEG. In conclusion, FEP may be associated with a selective left hemisphere MMN deficit in auditory regions on the superior temporal plane, but not the opercular cortex (i.e. Brodmann’s Area 43).
Read full abstract