Uncovering the Role of Biological Sex in the Divergent Genetic Profiles of Early and Late-Diagnosed

Autism is a neurodevelopmental disorder with a significant male prevalence bias. While recent evidence suggests that genetic heterogeneity is indexed by age at diagnosis, males are also typically diagnosed earlier, such that the extent to which these age-specific findings are confounded by shared genetic signal with sex-specific genetic architecture remains unclear. To test this, we leveraged sex- and age of diagnosis-stratified GWAS summary statistics for autism and applied a multiple regression framework within Genomic Structural Equation Modeling (Genomic SEM). This approach allowed us to disentangle the overlapping genetic variance between sex, diagnostic timing, and 68 clinically relevant phenotypes reflecting a host of different psychiatric, cognitive, health, and social outcomes. Consistent with prior findings, we identified significantly divergent genetic associations with external traits between early- and late-diagnosed autism subtypes. We also identified few findings for sex-specific associations, along with high genetic correlations across these two traits, which questions the biological basis for disproportionate prevalence rates in males. Multiple regression models confirmed that early-specific associations were not confounded by biological sex. Conversely, the correlations between late-diagnosed autism and the sex-stratified autism GWAS were near 1, indicating that late-specific associations may be confounded. Follow-up analyses using an internalizing factor provided support for sex differences in clinical presentation and age at diagnosis, with female-diagnosed and late-diagnosed autism estimated to have stronger genetic correlations with internalizing traits compared to male and early-diagnosed counterparts. This study underscores the importance of subtyping in genetic analyses and provides a framework to disentangle confounded genetic pathways.