Background: Elevated prenatal sex steroids and maternal conditions that are related to sex steroids (e.g., polycystic ovary syndrome) have been positively associated with autism likelihood. It is unclear if this is detectable in the maternal circulation, if it relates to maternal autistic traits, and whether it is also predictive of autistic traits in infants.
Methods: Maternal serum samples were collected as part of routine prenatal screening from pregnant women taking part in the longitudinal Cambridge Ultrasound Siblings and Parents (CUSP) study (n=219) (gestational age: mean=12.7 [SD=0.8] weeks). Concentrations of the following were measured via immunoassays: testosterone (T), estradiol (E2), dehydroepiandrosterone sulphate (DHEAS), progesterone (P); sex hormone-binding globulin (SHBG). Human choriogonadotropin (hCG) and pregnancy-associated plasma protein A (PAPP-A) were collected from clinical records corresponding to the same serum samples. Participants completed the adult Autism Spectrum Quotient (AQ). Infants were followed-up with the Quantitative Checklist for Autism in Toddlers (Q-CHAT) between 18-20 months old (mean=570 days, SD=21.3 days).
Results: Maternal AQ scores significantly correlated with circulating levels of total E2 (Pearson’s r=0.20, p=0.036) and the bioactive fraction of E2 (Pearson’s r=0.26, p=0.008) in univariate and multiple regression models. Total E2, DHEAS and a steroidogenic factor (derived from total E2, DHEAS and T) were all associated with Q-CHAT scores in multiple regression models that controlled for covariates and for an interaction with infant sex. This interaction also had a significant effect, leading to a positive correlation between hormone levels and Q-CHAT scores in males but not in females (interaction term: semipartial correlation r=0.23, p=0.018). The opposite was found for standardised hCG values and Q-CHAT scores, with a positive association in females but not in males (interaction term: semipartial correlation r=-0.22, p=0.009). Limitations: This longitudinal clinical study was relatively small and statistical power was further reduced by the need to account for different effects according to sex. The findings will need to be confirmed in a larger cohort and with clinically diagnosed cases of autism.
Conclusion: In line with previous findings, this study’s results suggest that increased steroid synthesis prenatally is related to autistic traits and that this is detectable in the maternal circulation.