Estimating masculinity in neurodevelopment disorders

Neurodevelopmental disorders (NDDs) display a striking sex bias, with three males to every one female diagnosed with autism spectrum disorder (ASD) and a two-to-one ratio observed for intellectual disability (ID). This disparity has fueled a body of research examining aberrant sex hormone exposure as a potential contributor to NDDs. These studies primarily utilized digit ratio, a potential indirect measure of fetal androgen exposure, to predict individual risk for NDDs. However, results to date have been inconclusive, and increasing evidence suggests 2D:4D ratio is an inadequate predictor of NDD risk. Emerging research indicates a strong association between facial morphology and NDDs, with evidence suggesting facial masculinity may better capture latent risk factors related to mental disorders. The goal of this study is to compare the capacity of digit ratio and automated facial landmarking to predict NDD diagnosis and genetic risk in a sex-independent manner. A total of 577 individuals spanning a spectrum of social, behavioral, communication, and cognitive capabilities were enrolled in devGenes, a genetic registry of families with NDDs. Of these, 370 individuals (61% female) reported no NDD; the remaining participants reported a diagnosis of ASD (n=129, 22% female), ID (n=97, 30% female), and/or another NDD (n=131, 27% female). Reports were corroborated by examination of medical records. Digit ratios were calculated from second- and fourth-digit measures obtained from subjects’ hand scans. Automated facial landmarking was conducted on subjects’ facial photographs, and facial features were extracted and combined into a single predictor of sex, using a machine learning model. In addition, participants were categorized into typically developing, high functioning (autism with language impairment or epilepsy) and low functioning (autism with intellectual disorder). Genetic support of predicted masculinity was investigated using genotype array data from a subset of 174 individuals. Polygenic risk scores were calculated for 174 individuals from14 traits including mental disorders, cognitive performance, sex hormone, social behavior and physical trait. Surprisingly, increasing facial masculinity was significantly associated with increasing severity symptoms of NDD (p=0.03 for female and p=0.01 for male). Digit ratio was not associated with NDD severity. On the other hand, genetic analysis showed that increasing masculinity, based on digit ratio and facial landmarks, was associated with increasing polygenic estimates for educational attainment and social dissatisfaction. In addition, the facial masculinity measure specifically associated with polygenic risk for mental disorders while the masculinity measure from digit ratio associated with polygenic estimates for sex hormone traits.