Objectives: To characterize how the distribution of de novo and rare variants in ASD probands differs between the sexes and to assess whether this data supports specific models of ASD sex bias.
Methods: Samples from the Simons Simplex Collection were analyzed by genotyping array (N=2,326 families; 309 female probands, 2,017 male probands) and whole-exome sequencing (N=964 families; 147 female probands; 761 male probands) to identify de novo and rare variants predicted to alter protein composition.
Results: Female probands consistently show a higher proportion of de novo variants than their male counterparts for both CNVs and loss of function (LoF, i.e. nonsense, canonical splice site and frameshift variants) variants. For de novo CNVs that include exons the difference is modest (9.1% of females vs. 5.4% of males, OR 1.76 (95% CI: 1.11-2.78), p=0.01), however considering only de novo CNVs implicated in clinical syndromes, such as 16p11.2, leads to a difference of a similar order to the sex bias observed in ASD (4.9% of females vs. 1.4% of males, OR 3.26 (95% CI: 1.82-7.14), p<0.001). From whole-exome sequencing a similar pattern emerges with a higher rate of de novo LoF in females (22.4% of females vs. 10.6% of males; OR 2.4 (95% CI 1.5-3.9), p<0.001); a similar non-significant trend is observed when the analysis is restricted to genes with ≥2 de novo LoF (a threshold at which they are considered ASD-associated) though only 21 individuals have such variants (4.1% of females vs. 2.0% of males; OR 2.12 (95% CI: 0.72-5.94), p=0.13). Finally while de novo CNV and LoF variants are observed on chrX, they are not seen at a greater rate than on other chromosomes.
Conclusions: In individuals diagnosed with ASD, risk-associated de novo variants occur more often in females than in males. Such an observation is consistent with a model of relative protection to ASD liability in females, but with the caveat that de novo events contribute sufficient risk that the protective mechanism is overwhelmed. In such a model the increase in de novo events is the consequence of a smaller proportion of females in whom ASD is caused by low risk inherited variation to which the females have protection. This model would explain the comparative similarity in recurrence rate of siblings to male and female probands (the ‘Carter Effect’). In summary the observation of increased de novo events in females with ASD supports a model of protection to inherited genetic risk in females.