Modeling the Phenotypic Architecture of Autism Symptoms From Time of Diagnosis to Age 6

Background: Autism Spectrum Disorder (ASD) is a heterogeneous disorder. Recent studies showed that ASD heterogeneity can be captured by classifying newly diagnosed children into three classes (or subgroups) that differ in symptom severity and configuration (Georgiades et al., 2013). Despite the fact that notable progress has been achieved in the classification of children with ASD at the time of diagnosis, longitudinal research is needed to better understand how symptom heterogeneity unfolds as children with ASD develop.   

Objectives: The main objective of the current study was to model the underlying latent class structure (phenotypic architecture) of core autism symptoms - social communication deficits (SCD) and Fixated Interests and Repetitive Behaviours (FIRB) - from the time of diagnosis to age 6.   

Methods: The sample comprised 280 children (86% males) participating in a longitudinal study of ASD. Factor mixture modeling (FMM) was performed using data on 26 items from the Autism Diagnostic Interview – Revised algorithm indexing the SCD and FIRB autism symptom domains. A set of goodness-of-fit criteria were used to select the best fitting model of the underlying latent class structure of autism symptoms. The FMM analysis was repeated twice – at time of diagnosis (Time 1) and when the children turned age 6 (Time 2).  

Results: At Time 1, a “2-factor/3-class” structural model provided the best fit to the data (Class 1=35%; Class 2=11%; Class 3=54% of the sample). This model replicates the one reported in the Georgiades et al. (2013) study. However, the same model failed to converge at Time 2 at which point a more parsimonious “2-factor/2-class” model provided the best fit to the data (Class A=32%, Class B=68% of the sample). According to this factor mixture model 6-year old children with ASD can be classified in two distinct classes characterized by significantly different levels of severity on the SCD and FIRB symptom dimensions. Compared to children from Class B, children in Class A have significantly higher adaptive and language skills and present with lower emotional/behavioural problems. Furthermore, there is a difference in the way boys and girls are distributed across the two ASD classes; girls tend to be assigned to the less severe, higher functioning class (Class A: 61.5%; Class B: 38.5%) while the reverse is true for boys (Class A: 27.4%; Class B: 72.6%). Finally, children across the two classes did not differ in terms of the age at which they were diagnosed.  

Conclusions: Study findings suggest that there is a change in the underlying latent class structure of autism symptoms during the first few years after diagnosis. Specifically, it appears that there is a reduction in symptom heterogeneity in children with ASD from the time of diagnosis to age 6. These findings demonstrate the dynamic nature of the ASD phenotype and speak to the importance of repeated classification assessments of symptoms, functional skills, and behaviours as children develop.