The Etiological Relationship Between Dimensional Traits and Categorical Diagnostic Constructs of ASD

Background:  Evidence from twin and family studies to date has shown that liability to autism impairments in social communication and interest patterns extends beyond the traditional categorical diagnostic boundaries, thus leading to broadening of diagnostic criteria to include other developmental subtypes. The Autism Spectrum Disorder diagnosis was therefore re-conceptualised as the extreme end of dimensionally distributed set of autistic traits in the population, rather than a separate behavioural entity. Most recent twin studies further confirmed that the genetic factors  responsible for the variation of autism trait scores in the population are likely to be the same for the extreme trait scores at the end tail of the same distribution. 

Objectives:  To define the relative genetic (A), shared (C) and unique (E) environmental influences on the hypothetical overlap between the continuous measure of autism traits (Children Autism Spectrum Test, CAST) in a large population of UK twins and a sub sample of the same cohort in which at least one twin has received a clinical diagnosis of Broad Spectrum or ASD, assessed by DAWBA/ADOS/ADI-R and an overall Consensus Diagnosis (CD). Additionally, we wanted to validate findings of a recent study reporting the importance of shared (C) environmental influences (Hallmayer et al, 2011) in clinical autism/ASD diagnosis. 

Methods:  The genetic and environmental parameters were estimated in a bivariate continuous-ordinal liability threshold model. To account for the selected nature of the sample the thresholds were fixed to 'known' population z- values: the 1st threshold discriminating between categories 0 and 1 was fixed at 5% (Broad Spectrum) and the 2nd threshold discriminating between 1 and 2 was set to 1% (ASD). 

Results:  Cross-twin within-trait correlations for CAST, ADOS and CD indicated no role of shared environment (C), except for ADI-R; whereas dominant rather than additive genetic effects were indicated for the DAWBA. The cross-twin cross-trait correlations between CAST and ADI-R, ADOS and CD all indicated mainly genetic and unique environmental influences on the covariance. The phenotypic overlap (r_ph) between CAST and all clinical measures were moderate to high (.52 to .65) and mostly influenced by genes (r_ph-A for DAWBA=.40, ADI-R=.58, ADOS=.56 & CD=.60) and the remainder explained by unique environments (r_ph-E: DAWBA=.12, ADI-R=.03, CD=.05) but not significantly so for ADOS (r_ph-E=-.02). No shared environmental factors acted on the covariance between CAST and clinical measures.

Conclusions:  In the first study of this kind, we revealed a phenotypic overlap between continuous and ordinal measures of autistic traits and ASD and that this is largely due to genetic factors.