94. IMPACT OF CHANGES IN DIAGNOSTIC CRITERIA ON PERFORMANCE OF POLYGENIC RISK SCORES IN ADHD

Background The predictive power of polygenic risk scores (PRS) for ADHD has increased dramatically with the recent increases in GWAS sample size. However, in of the many genotyped cohorts contributing to the PGC ADHD working group participants were diagnosed prior to the release of the DSM-5 which did not allow for a co-diagnosis of ADHD and Autism/ASD. The removal of this restriction in DSM-5, and the changes in criteria for Autism between DSM-IV and DSM-5 have resulted in marked changes in the prevalence of ADHD and Autism. At the same time, an increase in awareness of neurodiversity has resulted in a growing number of people self-identifying as Autistic, ADHD or AuDHD. These changes in diagnostic criteria and increasing self-identification raise the question of measurement invariance. That is, does the ADHD PRS perform differently among those diagnosed under DSM-IV vs DSM-5, and how do those who have been formally diagnosed compare to those who self-identify? Methods We examined these questions using data from the Australian ASD and ADHD study (total N=6048; genotyped N=2109). Within this study we recruited adults who self-identify as having ADHD and/or Autism who were 1) diagnosed under DSM-III, 2) diagnosed under DSM-IV, 3) diagnosed under DSM-5, 4) were in the process of being diagnosed, or 5) were self-diagnosed. To examine these questions, we calculated ADHD and Autism PRS using SbayesR from the most recent PGC GWAS. Results Among those who reported a diagnosis of ADHD there was no significant difference in the distributions of those diagnosed under DSM-III vs DSM-IV. However, the mean ADHD PRS was significantly lower among those diagnosed under DSM-5 vs DSM-III or DSM-IV. Mean PRS of those currently engaged in the diagnostic process did not differ significantly from those with a diagnosis. Similarly, there was no significant difference between those who self-identified as compared to those with a diagnosis. Discussion The implication of these results both for research and clinical applications will be discussed.