Strategies for Detecting Increased ASD Risk in Male and Female Probands

Background: An excess of male cases has consistently been observed in autism spectrum disorder (ASD), however no clear mechanism for this sex bias has been demonstrated. The possibility of a female protective effect (FPE) has received considerable attention, especially given the excess of de novo CNVs in females with ASD in several cohorts. Under this FPE model we would expect an excess of ASD risk in females (as is observed for de novomutations) and an increased ASD recurrence rate in the siblings of females with ASD, compared to the siblings of males with ASD (often called the Carter Effect). However, despite several studies with large sample sizes the Carter Effect has not been observed for ASD diagnosis. The evidence for the FPE mediating ASD sex bias is therefore contradictory.

Objectives: To assess the likelihood of observing increased ASD risk in females under an FPE model directly (de novomutation) and indirectly (Carter Effect).

Methods: A family-based simulation was used to estimate the power to detect the FPE directly (proband de novo mutation rate) and indirectly (sibling ASD recurrence rate). To perform this calculation estimates of the ASD risk architecture were required. Two models were used: 1) ASD is caused entirely by inherited genetic risk (the ideal scenario for observing the Carter Effect); and 2) 50% of ASD risk is from common variation, 3% from de novo and 47% from unique environmental effects (estimates based on Gaugler et al, Nature Genetics, 2014). 

Results: Under the first model (100% inherited risk), approximately 3,000 probands were required to detect the Carter Effect with 80% power. To date, only one analysis of ASD recurrence rate has exceeded this threshold (Grønberg, JAMA Pediatrics, 2013). However, under the second model (50% inherited risk, 3% de novo) this power was reduced considerably. Even with a samples size of 10,000 the probability of detecting the Carter Effect was less than 20%. Under this second model the probability of detecting a significant difference in the burden of de novo mutation between affected females and males was 80% with only 500 probands.

Conclusions: The sample size required to reliably detect the Carter Effect is at least an order of magnitude greater than the largest study of ASD recurrence rate performed to date. In contrast, the direct observation of a higher burden of known ASD risk factors (such as de novo mutation) can be observed with the sample sizes currently available for analysis. Therefore, a model of ASD sex bias being predominantly a consequence of a female protective effect fits the available data well.