Understanding the polygenic architecture of autism with dimensional measures of behavior and gender

Binary measures of traits, like cases versus controls, have been the traditional measures used by genetic researchers. However, these measures are often used simply because they are the most convenient, not necessarily because they are the best traits for accurately reflecting the underlying biology. This may be particularly true for psychiatric research. For example, autism is not as definitive as those who have received an autism diagnosis versus those who have not been diagnosed. Some autistic people severely struggle with communication and may be unable to speak, whereas other autistic people may only struggle with nonverbal communication. This highlights the need for the central conceptual frameworks that unify the main approach to this thesis work: First, to consider psychiatric and behavioral traits as dimensional, not binary. And second, to consider that people with the same psychiatric diagnosis do not necessarily have the same psychiatric traits, so investigation of subtle behavioral traits is warranted. This phenotypic-forward approach was applied to genetic research that leveraged the additive effects of common genetic variants. The cumulative sum of the many common genetic variants that have been associated with a particular trait is a polygenic score (PGS). Because many autistic people are also gender diverse (meaning transgender or gender nonbinary), the first part of this thesis sought to investigate how dimensional measures of gender diversity and mental health are related to PGS. However, considering the lack of prior research in the genetic factors of gender diversity, the first step was assessing community attitudes towards genetic research in sexual orientation, gender identity, and mental health. We surveyed N = 768 adults for their opinions on this topic. We found that the most important factors that determined attitudes towards genetic research in sexual orientation, gender identity, and mental health were simply broader towards mental health research and genetic research more generally. In the second study, we collected dimensional gender diversity data using the Gender Self Report (GSR) and also dimensional mental health data in N = 696 adults in the SPARK autism cohort. We found that the two GSR subscales, Binary and Nonbinary, were phenotypically positively associated with more mental health problems. However, genetic investigation using PGS found that Binary and Nonbinary were not associated with psychiatric PGS, but mental health problems were positively associated with psychiatric PGS. Instead, Binary and Nonbinary were positively associated with the cognitive performance and non-heterosexual sexual behavior PGS, and mental health problems were not associated with the cognitive performance and non-heterosexual sexual behavior PGS.

The second part of this thesis shifted focus to investigation of dimensional behavioral traits that are directly related to the core traits of autism. A diagnosis of autism requires the person to have social and communication difficulties, and a strong desire for sameness and predictability that manifests itself as focused interests, repetitive movements, and an insistence for routines. In N = 6, 064 autistic children in SPARK, we used 12 dimensional autistic traits from three different assessments of social difficulties, repetitive behaviors, and coordination problems. Genetic investigations including associations with 14 PGS for psychiatric diagnoses and personality traits, GWAS, and SNP-based heritabilities revealed that these autistic traits varied in their SNP-based heritabilities and PGS associations, highlighting how autism is heterogeneous both phenotypically and genetically. The repetitive behavior traits had the greatest genetic signal and this genetic signal was most generalizable to a general population sample (the ABCD cohort). Lastly, we investigated the main effects of PGS on broader behavioral problems in N = 7, 091 children from SPARK and ABCD, and we also examined if these PGS effects changed depending on environmental adversity (i.e., gene-by-environment interactions). The environmental adversity variable itself had many PGS associations, including positive associations with psychiatric PGS and negative associations with educational attainment and cognitive performance PGS, which indicates that it was not a "pure" environmental variable but was actually entangled with many polygenic factors. Additionally, environmental adversity and many PGS had independent, additive effects on behavioral problems, and the PGS effects were most penetrant in adverse environments.