Evaluation of Mismatch Negativity As Biomarker for Language Impairment in Autism Spectrum Disorder

Background: Currently, autism spectrum disorder (ASD) is diagnosed using the Diagnostic Statistical Manual of Mental Disorders-Fifth Edition (DSM-V) and children often go undiagnosed until around the age of three. Moreover, current language assessments are designed to behaviorally measure language skills, therefore requiring that a child have language or be “of language age” in order to participate. As a result of these diagnostic limitations, speech and language interventions for children with ASD plus language impairment (ASD+LI) are often not initiated until a child is of preschool age. Finding an early and objective way to identify language impairment (LI) in ASD has the potential to lead to earlier speech and language intervention for individuals “at risk” for the disorder. Magnetoencephalography (MEG) studies use the Mismatch Field component (MMF) to investigate how the brain processes speech sounds. Previous MEG studies by Roberts et al. (2011) utilizing the MMF component have shown that increased MMF latency (i.e., longer processing time) is a predictor of LI in children with ASD (sensitivity 82.4%; specificity 71.2%). 


Objectives: Since MEG is expensive and not widely used with infants or young children, we attempted to replicate these results using the mismatch negativity (MMN), the electroencephalography (EEG) equivalent of MMF. EEG is inexpensive and can be used with children of all ages making it an appropriate method to identify LI in children on the autism spectrum. We explored increased MMN latency as a potential biomarker for LI in autism. 


Methods: EEG was recorded in children ages 5-10 with ASD+LI, ASD-LI and typically developing controls (TD) during a passive auditory oddball experiment presenting speech sounds. During the recording children were instructed to watch a movie and ignore the sounds.

Results: Contrary to previous MMF findings, individuals with ASD+LI demonstrated decreased MMN latency in the left hemisphere in response to vowel sounds compared to those with ASD-LI and TD controls. A positive correlation between left hemisphere MMN latency and language scores on the Clinical Evaluation of Language Fundamentals-Fifth edition was found when combining both ASD groups. No correlation between MMN latency and language score was found in the control group. Parent report revealed that all individuals with ASD who participated in this study were hyper-sensitive to sounds.

Conclusions: Our results show that children with ASD+LI and hyper-sensitivity to auditory stimuli detect change in the auditory stream faster than children with ASD-LI and TD controls. Our results support the theory that children with ASD+LI have increased connectivity in primary sensory cortices at the expense of computational connectivity between association areas of the brain (Belmonte et al., 2004). This may account for faster auditory processing time despite low language scores in these children. Further research needs to be done in order to determine if grouping children by hyper- versus hypo-sensitivity to auditory stimuli could explain conflicting results between studies and elucidate a neurophysiological biomarker of language impairment in subgroups of children with ASD.