The Role of Chromosome X in ASD Sex Bias

Background: ASD is observed at a higher rate in males than females (4:1); a simple explanation for this bias is X-linked risk variants in the male. While Fragile X (FMRP) is the most common cause of ASD there are few ASD families that display clear X-linked inheritance leading to gene discovery (with the exception of NLGN4X). Alternative mechanisms of sex bias include female protection by virtue of the diploid chromosome X (chrX) or an excess of chrX-based common inherited risk variants. Analysis of the distribution of Social Responsiveness Scale (SRS) scores in 515 female individuals from multiplex families of the AGRE collection showed a bimodal distribution not found in males. This raises the possibility that a common allele leads to higher sociability and therefore ASD protection but only in the diploid homozygous state (i.e. not in males or subset of females). To account for a sex bias of 4:1 the variant would require a minor allele frequency of 14% and to be present on a region of chrX that escapes X-inactivation and has no homologue on chrY.

Objectives: To identify if a common variant on chrX leading to ASD protection is present in an identified subset of females with ASD.

Methods: 261 unaffected and 254 affected females from the AGRE sample set were used for initial discovery. Both sample groups were split into high and low SRS score subsets (divided by an SRS score of 45 and 87 in the unaffected and affected groups respectively). Principle component analysis was used to select only individuals of European ancestry. Standard genotype data cleaning procedures were used including genotyping call rate >99%, Hardy Weinberg Equilibrium (p-value <0.05), and SNP call rate >90%. Only SNPs within regions shown to escape X-inactivation, as defined by Carrel and Willard 2005, and outside of the pseudo-autosomal regions and X-transposed region were considered for analysis. Analysis was restricted to SNPs with a minor allele frequency of 14% ±10% in line with the hypothesis. A replication set comprising 710 unaffected females, similarly classified into higher and lower SRS scoring subsets, and 244 affected females was used.

Results: We found a single SNP (rs5936079) at Xp22.2 that approached significance with a Bonferroni corrected p-value fractionally greater than 0.05 in the discovery set for the comparison of high SRS scoring unaffected females versus low SRS scoring affected females. The minor allele was C with a frequency of 20%; odds ratio was 4.76 (95% CI: 2.22-10.0). In the SSC replication set this same SNP had a p-value of 0.42 in high SRS scoring unaffected females versus all affected females (regardless of SRS).

Conclusions: No SNP was identified matching our hypothesis after correction for multiple comparisons in either the discovery or replication set. However the top scoring SNP was observed at the frequency predicted by the hypothesis and merits follow up in a larger sample set. Our method of pursuing specific association hypotheses based on predicted allelic characteristics represents a novel approach to investigating biological phenomena in genotyping data.