Psychiatric comorbidities in Asperger syndrome are related with polygenic overlap and differ from other Autism subtypes

Abstract

There is great phenotypic heterogeneity within autism spectrum disorders (ASD), which has led to question their classification into a single diagnostic category. The study of the common genetic variation in ASD has suggested a greater contribution of other psychiatric conditions in Asperger syndrome (AS) than in the rest of the DSM-IV ASD subtypes (Non_AS). Here, using available genetic data from previously performed genome-wide association studies (GWAS), we aimed to study the genetic overlap between five of the most related disorders (schizophrenia (SCZ), major depression disorder (MDD), attention deficit hyperactivity disorder (ADHD), obsessive-compulsive disorders (OCD) and anxiety (ANX)), and AS, comparing it with the overlap in Non_AS subtypes. A Spanish cohort of autism trios (N = 371) was exome sequenced as part of the Autism Sequencing Consortium (ASC) and 241 trios were extensively characterized to be diagnosed with AS following DSM-IV and Gillberg's criteria (N = 39) or not (N = 202). Following exome imputation, polygenic risk scores (PRS) were calculated for ASD, SCZ, ADHD, MDD, ANX, and OCD (from available summary data from Psychiatric Genomic Consortium (PGC) repository) in the Spanish trios' cohort. By using polygenic transmission disequilibrium test (pTDT), we reported that risk for SCZ (P_scz = 0.008, corrected-P_SCZ = 0.0409), ADHD (P_ADHD = 0.021, corrected-P_ADHD = 0.0301), and MDD (P_MDD = 0.039, corrected-P_MDD = 0.0501) is over-transmitted to children with AS but not to Non_AS. Indeed, agnostic clustering procedure with deviation values from pTDT tests suggested two differentiated clusters of subjects, one of which is significantly enriched in AS (P = 0.025). Subsequent analysis with S-Predixcan, a recently developed software to predict gene expression from genotype data, revealed a clear pattern of correlation between cortical gene expression in ADHD and AS (P < 0.001) and a similar strong correlation pattern between MDD and AS, but also extendable to another non-brain tissue such as lung (P < 0.001). Altogether, these results support the idea of AS being qualitatively distinct from Non_AS autism and consistently evidence the genetic overlap between AS and ADHD, MDD, or SCZ.

Fig. 1. Polygenic transmission of ASD, comorbid disorders and BMI AS trios (N = 39), and Non_AS trios (N = 202).

Transmission disequilibrium is represented as standard deviations of the mid-parent distribution. Colored geometric lines represent 95% confidence intervals. P-values over geom error bars measure the probability that the mean of the pTDT deviation distribution is higher than 0 (two-sided, one-sample t-test). For each disorder, pTDT values were calculated based on the P threshold in which highest transmission were found for the whole ASD trios’ population (N = 379). Significant over-transmissions were confirmed with random permutation of AS subgroup. “*” Permutation P-value < 0.05; “+” Permutation P-value < 0.1.

Fig. 2. Spearman correlation between gene expression diferences in AS against Non_AS and gene expression differences in comorbid disorder (ADHD, MDD, and SCZ).

Correlation under various P cutoffs from S-PrediXcan results (imputed gene expression differences in cases vs. controls from ADHD, MDD, SCZ, and BMI GWAS), were assessed. Predicted expression relationships were studied in brain frontal cortex, cerebellum, and lung. BMI was used as a negative control disorder. *P < 0.05; **P < 0.01; ***P < 0.001. Gene numbers and correlation results are described in Supplementary Table 6.