Examining the Interaction of Social Function and Language Skills in ASD

Background: The autism spectrum is associated with significant heterogeneity in terms of etiology, biology, and phenotype and characterizing this heterogeneity is key to development of targeted treatments. In this context, characterization of variability in the language domain has received much attention. Language impairments are a key feature of ASD and predictors of outcome in this population. Moreover, variability in this domain is suggested to be associated with underlying biological mechanisms. Although DSM-5 has deemphasized language ability and removed diagnostic distinctions between Asperger's syndrome and autistic disorder, language impairment remains a "specifier" ASD. One outstanding issue in the area is how this specifier interacts with core ASD symptom domains. Specifically challenging this examination is the confounding effect of communication function on the interaction between language social function in ASD. 

Objectives:  The objective of this study was to investigate the interaction between structural language ability and social functioning in ASD.

Methods:  A sample of children with ASD (n=141, age: 10.2+3.0, 111 male) was recruited. To tease out the confounding effect of communication on the language-social reciprocity interaction, the Reciprocal Social Interaction subtotal from the Autism Diagnostic Observation Schedule (ADOS)- Module 3 was used to characterize reciprocal social function. Language ability was quantified using the Oral and Written Language Scales, respectively. Receptive and expressive language were considered separately and as a composite as recommended by DSM-5. Using a cut-off of 85 in each language domain (one standard deviation below the mean), children were stratified into high and low language groups. This cut-off value also corresponded to the result of clustering the sample based on language ability using the k-means algorithm.

Results: Reciprocal social interaction score decreased significantly as language comprehension, oral expression, and the oral language composite increased (p<0.0001). In all three cases, there was a significant group x language effect on the social interaction scores (p<0.03).  In particular, the reciprocal social interaction score decreased significantly with scores on language domains in the "high language" group but not in the "low language" group.

Conclusions:  Our results suggest that structural language ability may define two subgroups within the autism spectrum. The interaction between the social and language domains is different between the two subgroups and may be reflective of distinct underlying biology.