Objective: This study aimed to investigate the association between mismatch negativity (MMN) and volumes of several brain regions measured using a semi-automated method in patients with schizophrenia and healthy controls. Methods: MMN in response to duration deviants and magnetic resonance imaging were acquired from 36 schizophrenia patients and 14 healthy controls. FreeSurfer was used for volumetric analysis. MMN amplitudes, brain volumes and their association were compared between schizophrenia and controls. Correlation analysis and multiple linear regression analysis were used to examine the correlated variables of MMN. Results: MMN amplitude was significantly lower in the schizophrenia group. In schizophrenia, MMN was positively correlated with age and negatively correlated with left hippocampal and right pars opercularis volumes. The association between left hippocampal volume and MMN in schizophrenia remained significant after controlling for potential confounders. Conclusions: Smaller hippocampal volume may play a role in the abnormal manifestation of MMN in schizophrenia. Significance: The significant association between MMN and left hippocampal volume may suggest unique neurobiological contribution of hippocampus in auditory processing in schizophrenia.

Introduction

Mismatch negativity (MMN), a type of event-related potential (ERP), is generated when a discernable change occurs in a series of repetitive standard stimuli (Naatanen et al., 1978). MMN represents the pre-attentive process of auditory discrimination and is associated with the function of auditory memory and involuntary attention shifting (Naatanen and Michie, 1979; Javitt et al., 1995; Naatanen et al., 2007). MMN deficit is extensively observed in schizophrenia patients (Shelley et al., 1991; Javitt et al., 1993; Umbricht and Krljes, 2005). By administering different auditory oddball paradigms (such as sounds with distinct duration, frequency, or intensity), research has revealed MMN to be related to the diagnosis of schizophrenia and a few functional indexes in schizophrenic patients, such as social cognition and independent living capability (Light and Braff, 2005; Wynn et al., 2010; Haigh et al., 2017). Conversely, duration of illness has been found unrelated to MMN deficit in a recent meta-analysis (Erickson et al., 2016). MMN deficit in response to duration deviants is stable in chronic schizophrenia patients; collectively, with the similar finding in first-degree relatives, it is considered as a candidate for endophenotype of schizophrenia (Michie et al., 2002; Turetsky et al., 2007; Rissling et al., 2010; Nagai et al., 2013b, 2013a; Umbricht et al., 2006; Solis-Vivanco et al., 2014). Structural abnormalities in several brain regions have been found to be associated with schizophrenia (Shenton et al., 2001; Brent et al., 2013). A previous review reported structural differences in the left medial temporal lobe and left superior temporal gyrus in schizophrenia (Honea et al., 2005) and in areas that contribute to MMN generation (Alho, 1995). Furthermore, it was found that decreased MMN amplitude to frequency deviants is correlated with reduced gray matter volume of the left Heschl’s gyrus in schizophrenia (Salisbury et al., 2007). As a candidate endophenotype of schizophrenia, the relationship between MMN and brain volumes is an interesting topic, for it may provide more understanding of the interrelatedness of structural and functional changes pertaining to the disease. Previous studies mainly focused on superior temporal and inferior frontal lobe involvement in MMN generation and auditory deviant processing (Wible et al., 2001; Schall et al., 2003; Rasser et al., 2011). Although it has been reported that mismatch activation could be detected in hippocampal areas at 350 msec with intracranially-placed electrodes (Rosburg et al., 2007), whether hippocampal activity could affect scalp-recorded MMN is less studied. Previous studies exploring structural brain abnormalities in schizophrenia have adopted manual tracing procedures to delineate regional brain volumes. This approach has a major dependency on raters and a potential risk of rater bias. In recent years, semi-automated volumetric parcellation methods have been developed and utilized in schizophrenia research (Fischl, 2012; van Erp et al., 2014; Arnold et al., 2015). Automated measurements are more efficient, reproducible and useful for larger samples. Volume estimations from semi-automated approaches were found to have strong correlation with manual tracing, suggesting a satisfying validity (Morey et al., 2009; Arnold et al., 2015). The current study aimed to: (1) compare MMN and brain volumes of the healthy controls and schizophrenia patients; and (2) examine the correlation between MMN and preselected brain regions in the schizophrenia group. We hypothesized that MMN was associated with volumes of brain regions such as the hippocampus, the superior temporal gyrus or the inferior frontal gyrus in patients with schizophrenia.