Context

Autism is a highly heterogeneous neurodevelopmental disorder (1:00). Supekar discusses the skewed gender ratio of women to men (ratio of 1 to 4) in autism diagnoses and asserts that this lack of female representation is a key source of heterogeneity in autism symptom presentation, treatments, and research (2:20). He stresses the historical consistency of the sex/gender bias (7:15) and discusses why we know so little about behavioral symptoms and brain signatures of females with autism (4:00). Supekar presents the unanswered question of why there are fewer females with autism as a fundamental building block for the future of autism research and treatments (3:45). 

The speaker describes the large-scale science approach, which collects publicly available data from multiple studies, sites, languages, populations, and investigators (9:40). Such datasets provide large sample sizes, robust and replicable findings, and population variance (10:15). Supekar uses the large-scale science approach to address three research questions (12:04): 

1. What are the behavioral/symptom presentations of autism in females, and how do they differ from those in males (12:40)?

Supekar and his colleagues compared female and male clinical data (13:00) and found that girls with autism exhibit less severe restricted/repetitive behavior (RRB) than autistic boys (14:00). These findings have been replicated in multiple large-scale studies (15:20). The presenter outlines subsequent research on co-occurring conditions (17:25), which revealed a lower prevalence of ADHD and a higher prevalence of epilepsy in autistic females compared to males (18:03). He also describes evidence suggesting an influence of age and gendered socio-ecological contexts on symptom presentation in autistic females (19:10). Supekar states that these findings show significantly different phenotypes between autistic males and females where the female presentation is more nuanced (19:53). 

2. What is the structural organization of autistic brains in females, and how do they differ from those in males (20:27)? 

Researchers applied multivariate pattern analysis (MVPA) (22:18) to clinical and structural MRI data to assess gray matter volume maps. Their findings revealed that structural brain organization in females with autism is significantly different from that of autistic males, especially in brain regions belonging to the motor, language, and visual-motor systems (23:33). Supekar’s team subsequently correlated those brain regions with clinical scores and found a relationship between RRB domains and gray matter volume in girls with autism (25:10). The speaker suggests these results reveal gender discrepancies in autistic structural brain organization that are associated with significantly different sex/gender phenotypes (27:04). 

3. What is the functional organization of autistic brains in females, and how do they differ from those in males (29:30)?

Supekar and his colleagues used deep learning neural networks (33:16) to assess differences in clinical and functional MRIs in multiple brain regions (32:00). Despite the heterogeneity of the dataset, the model was consistently (86%) accurate in differentiating males and females with autism (36:28). Further, this study revealed sex/gender differences in brain systems associated with visual-spatial attention in autistic individuals (37:25). Such findings had never been recorded in autism literature before and have since been replicated in a fully-independent cohort (38:15). 

Brain regions showing differences were subsequently correlated with clinical scores, which revealed that regions belonging to motor systems predicted the severity of RRB in autistic females (40:30) but not in autistic males. Supekar asserts that these findings suggest the presence of domain-specific effects associated with RRB in females and some level of female protective effect (5:54) in those brain regions (41:35).

Supekar summarizes the research findings (44:40), noting that gender differences in autistic structural and functional brain organization significantly differed from normative sex differences (24:16; 39:20). He lists the implications of their results, including the need for sex-specific diagnostic instruments and treatments. The presenter outlines priorities for future research (46:53) before the question and answer session (50:12).