On Electrophysiologic and Multimodal Signatures of Language Impairment in ASD

Background:  

Brain level traits, or endophenotypes, of ASD may serve as diagnostic or prognostic biomarkers. Perhaps more importantly, such brain measures may identify neural systems that lend themselves (1) to patient stratification and thus improve group homogeneity (stratification biomarkers), (2) to measuring therapeutic response, and (3) in translation to the preclinical environment where behavioral analogy is an important, but limited tool.

Objectives:  

To draw together results from multiple imaging modalities in a large cohort of children with ASD and both typically developing and clinical controls.

Methods:  

Approximately 200 children (6-15yrs) with ASD and age/IQ-matched typically developing (TD) subjects were administered tests of auditory processing while whole-head MEG data were collected. Diagnosis was confirmed via ADOS and ADI-R and/or SCQ. Language impairment was quantified using the Core Language Index (CLI) of the CELF-4. An additional cohort of children with specific language impairment (SLI; language impairment in the absence of ASD) were also recruited. Whole-head biomagnetometer data (Omega, 275-channel, VSMMedTech Inc.) were obtained during presentation of isolated sinusoidal tones (500 and 1000 Hz) as well as oddball mismatch tone and vowel paradigms. Dependent variables were left and right auditory cortex latencies for M50, M100 and the mismatch field (MMF). MRI (Siemens 3T Verio™) was performed for anatomic registration and for estimation of regional white matter integrity using diffusion tensor imaging  (DTI; 30 directions, b=1000s/mm²). In a subset of subjects, levels of GABA in auditory cortex were obtained using the MEGAPRESS MRS sequence (TE=68ms).

Results:  

Examining pure tones, M50 and M100 latency were delayed in ASD versus TD. The latency prolongation in ASD persisted even after measures of language impairment (CLI of the CELF-4) and IQ (FSIQ or PRI) were added as covariates. No latency prolongation was observed in the cohort of children with SLI.

MMF latency was prolonged in ASD, especially in the subset of children with ASD with language impairment (CELF CLI < 85). MMF latency was similarly prolonged in children with SLI.

DTI of the acoustic radiations was abnormal in ASD, specifically fractional anisotropy (FA) was reduced and exhibited a flatter developmental trajectory than observed in typical development.

DTI of the left superior longitudinal fasciculus (SLF) was abnormal in ASD, specifically exhibiting elevated mean diffusivity (MD), especially in the subset of children with ASD with language impairment. Although mean diffusivity was also elevated in children with SLI, the most pronounced elevation was observed in the cohort with ASD and language impairment. A 2x2 ANOVA (with factors of ASD/noASD and LI/noLI) with PRI and age as covariates revealed significant main effects (p<0.05) of both ASD and LI.

GABA levels in superior temporal gyrus (STG; including auditory cortex) were significantly lower (p<0.05) in ASD than TD.

Conclusions:  

There is considerable emerging physiologic evidence for brain abnormalities in ASD. Some abnormalities appear to be proportional to clinical severity (in the domain of interest: language).

Converging evidence from multiple modalities supports neurobiological interpretation of developmental anomalies that are both general to ASD and also specific to the language impairment aspect of the phenotype.