Social communication difficulties represent an autistic trait, which is heritable and persistent during the course of development. Little is known however about variations in heritability and genetic association signals spanning childhood and adolescence.
Our study aimed to undertake a genome-wide investigation of social communication symptoms during childhood and adolescence using both single-time point and longitudinal approaches, and an estimation of the heritability at different developmental stages.
We performed a genome-wide association study on mother-reported social communication problems as captured by the Social and Communication Disorders Checklist (SCDC), which were assessed in a large UK population-based birth cohort at 8, 11, 14 and 17 years of age (Avon Longitudinal Study of Parents and Children, N ≤ 5628; 2883 males, 2745 females). We selected a 2-stage analysis approach in order to facilitate a computationally efficient genome-wide screen. During the first stage, we derived single-time point estimates using a Quasi-Poisson regression framework. During the second stage, we modelled the continuous change in social communication symptoms using the best-fitting multi-level Poisson model for all single-time point SNP signals below our selection threshold (P≤1E-05). Using ranktransformed sex- and principal component adjusted social communication scores with a normal data distribution, we also conducted ‘Genome-wide Complex Trait Analysis (GCTA)’ in order to estimate the proportion of additive phenotypic variance explained by all SNPs (narrow-sense heritability).
Quasi-Poisson regression identified 34 independent signals (P≤1E-05) across the four investigated time-points. The strongest single-point SNP signal (rs4453791) with genome-wide evidence for association was observed near a sodium channel encoding gene (voltage-gated, type XI, alpha subunit; SCN11A) on chromosome 3p22.2 at the age of 17 years (beta[changes in SCDC log-scores per risk allele]=0.23; P=9.31E-09). There was no evidence however that this signal contributed to association with SCDC scores at an age of 8 or 11 years (P<0.26), although there was moderate evidence for association with social-communication difficulties at the age of 14 years (beta=0.13; P=0.0014). Combining the effect estimates in a multilevel Poisson framework did not increase the evidence for association (beta=0.085, P=0.014) although there was evidence for a SNPxAge interaction effect (P=0.0013). In parallel, GCTA revealed that the narrow-sense heritability was highest at the beginning (age 7 years, h2=0.23, P=8.0E-05) and at the end of puberty (age 17 years, h2=0.45, P = 3.2E-09). Measures within mid puberty however demonstrated considerably lower heritability (age 11 years: h2=0.16, P=0.0059; age 14 years: h2=0.078, P=0.12).
Our analysis revealed that neither genetic associations nor heritability estimates for parent-reported social communication difficulties may remain stable during the course of childhood and adolescence, despite the observed phenotypic stability of the most severely affected individuals observed in previous research. Our findings emphasize the presence of age-specific genetic associations and highlight the model complexities of longitudinal frameworks aiming to capture these signals. We are now undertaking analyses investigating factors contributing to the observed variation in heritability. For replication purposes we will also investigate the observed SNP signals in autism samples, such as families from the Autism Genetic Resource Exchange.
See more of: Genetic Factors in ASD
See more of: Biological Mechanisms