Differences in Auditory Evoked Potentials Between Children with Autism Spectrum Disorder with Versus without Language Impairment: A Methodological Comparison

Background:  Atypical auditory processing is commonly reported in individuals with autism spectrum disorder (ASD), particularly in those with co-occurring language impairment (LI). One method commonly used to investigate these processes is recording of auditory evoked potential (AEPs)using electroencephalography. There is considerable interest in understanding whether atypical auditory cortical processing underlies LI in individuals with ASD, however, findings are inconsistent. Bomba and Pang (2004) identified two factors that explain some discrepancies in the literature: the heterogeneity of individuals diagnosed with ASD and the lack of an appropriate control group. A third possibility receiving little consideration in the literature to date is whether the choice of data analysis method contributes to discrepancies across studies.

Objectives:  To compare differences in AEPs between children with ASD with normal language development (ALN) and those with language impairment (ALI) using three analysis methods: peak amplitude/latency, time-frequency, and global waveform resemblance.

Methods:  Seventeen children aged 7-11 years (ALN: n=10, ALI: n=7) participated in a passive AEP paradigm where 225 trials of a 50ms, 490Hz tone were presented binaurally. A 128-channel EGI system recorded AEPs during stimulus presentation. Children with ALN had standard scores ≥ 85 on the Clinical Evaluation of Language Fundamentals – 4, while those with ALI scored < 85 (< 1 SDbelow the mean). The AEPs of children with ALN and ALI recorded at channel Fz were compared across three analysis methods. 1) Peak amplitude/latency analysis: P1 and N2 AEP components were identified from the grand averaged waveform of all participants. The mean amplitude and peak latency within 60ms (±30ms) around the maximum of P1 and N2 were calculated and compared across groups. 2) Time-frequency analysis: The AEPs of all participants underwent wavelet transformation from 5-100Hz in steps of 0.98 Hz using Morlet wavelets. The power and circular variance of AEPs at the gamma oscillation range (30-50Hz) were calculated and compared for both 200ms pre- and 200ms post-stimulus intervals. 3) Global waveform resemblance analysis: An AEP segment from 0-400ms post-stimulus presentation was compared between children with ASD to our previously established normative AEPs of children with typical development aged 7-10 years. Using the intraclass correlation coefficient as an indicator of overall resemblance, the AEPs of children with ASD were assigned an age-equivalent based on the comparison that yielded the highest resemblance score.

Results:  In amplitude/latency analysis, only the N2 mean amplitude was significantly different between ALI and ALN (p<0.005). Neither power nor circular variance of gamma-oscillation were significantly different between the two groups. Despite having the same mean chronological age, children with ALI had a younger AEP-age-equivalent than children with ALN (7.4 and 8.4 years respectively, p<0.01) on the global resemblance analysis. 

Conclusions:  Our results demonstrate that different AEP analyses can lead to different conclusions about putative auditory processing differences between children with ALI and ALN. Interpretation of the relation between AEPs and language functioning in ASD should take into consideration the choice of data analysis method and the neural processes thought to underlie different cortical responses.