Looking beyond labels: defining dimensional psychiatric symptom clusters with insulin-related somatic measures based on large population-based genetic datasets

It is known that diagnostic labels of complex psychiatric disorders, like attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD) and obsessive-compulsive disorder (OCD), may not be distinctive at the level of their etiologies. Understanding the biology underlying psychiatric disorders is challenging because of phenotypic and genetic heterogeneity and by frequent comorbidities that also extend to somatic conditions, e.g. diabetes mellitus type 2 and obesity that share genetic risk. Looking beyond the labels of psychiatric disorders with disorder-like traits has shown to be fruitful for genetic analyses. Therefore, we aim to explore the continuum of psychiatric symptoms in combination with insulin-related somatic measures to explore the existence of shared genetic architectures. We hypothesize that this ‘genetics-first’ approach can identify potential biological profiles of subgroups within psychiatry. For our approach we utilized publicly available summary statistics of the largest genome-wide association studies (GWASs) of ADHD, ASD and OCD symptoms and insulin-related measures available at the time of performing our analyses. For exploratory analyses, we prioritized insulin-related somatic traits that are genetically correlated to ADHD, ASD or OCD. Thereafter we applied multivariate genomic structural equation modeling (genomicSEM) to investigate latent genetic factors underlying GWASs of ADHD symptoms (n=17,666), OCD symptom and cases (n= 17,992) and social behavior linked to ASD (n=342,461) with five insulin-related traits (body mass index (BMI), fasting glucose, fasting insulin, glycated hemoglobin, and homeostatic model assessment for insulin resistance (HOMA-IR), n=98,210-697,734). Using exploratory and confirmatory factor analyses we explored the best fit to our data. A three-factor model fitted our data the best (AIC=57, CFI=0.99, SRMR=0.068). The first factor was loaded by solely insulin-related traits (BMI, fasting insulin and HOMA-IR). The second latent factor included ADHD symptom scores together with BMI, fasting glucose and glycated hemoglobin. The third factor included OCD symptom and case scores loaded together with glycated hemoglobin. At the moment we are running a multivariate GWAS to explore genetic variants underlying these factors. Moreover, we will apply gene-wide and gene-set analyses to identify relevant genes for these clusters and interpret GWAS outcomes of our results. Our findings show that latent genetic factors exist reflecting shared liabilities underlying psychiatric traits with insulin-related measures. The observed factors driven by both psychiatric traits and insulin-related traits suggest potential novel subgroups in psychiatric disorders such as ADHD and OCD. The study provides groundwork to further conduct in depth analyses of the role of insulin-related traits in psychiatry. Furthermore, this project suggests that biologically informed subgroups might capture the heterogeneity of psychiatric disorders and allow more personalized treatment approaches.