Polygenic risk prediction of psychiatric and medical comorbidity burden in anorexia nervosa using a data-driven approach

Anorexia nervosa (AN) is a serious eating disorder with high morbidity and mortality. In line with clinical and epidemiological studies that have established comorbidity to be the norm in psychiatric disorders, psychiatric as well as medical comorbidities have been shown to be frequent among individuals with AN. Notably, a recent Danish epidemiological study has identified comorbidity burden to be associated with increased risk of readmission and mortality in AN, highlighting the need to further examine the etiology and burden of comorbidity. The objective of the present study is to examine the genetic underpinnings of psychiatric and medical comorbidity burden in patient with AN using Danish population registers. The study population consists of individuals born in Denmark between May 1, 1981 and December 31, 2009 who received an AN diagnosis (ICD-10: F50.0 and F50.1) by December 31, 2016, followed up until the end of 2018. Information on comorbidities was obtained from the Danish National Patient Register and the Danish Psychiatric Central Research Register. In our ongoing analysis, the Danish subcohort of Anorexia Nervosa Genetics Initiative and Eating Disorders Genetics Initiative serves as our target cohort, comprising ∼7,000 AN cases with genotype data. Using Cox regression models adjusted for birth year and genomic principal components, polygenic risk scores (PRS) calculated using LDpred2-auto and meta-PRS for 422 phenotypes across 44 categories are used as predictors of psychiatric and medical comorbidity burden in AN cases. We will also present our findings on whether certain PRS categories predict specific combinations of comorbidities within AN patients. Additional sensitivity analyses will focus on ancestral diversity and other clinically relevant variables such as sex assigned at birth. To our knowledge, this is the largest and most detailed genomic examination of psychiatric and medical comorbidities in AN to date. Our findings have the potential to serve as a crucial step toward identifying distinct clinically relevant phenotypes within AN as assessed by polygenic burden to a large variety of complex traits and diseases in a data-driven approach, elucidating the biological mechanisms associated with AN outcome based on comorbidities, as well as developing risk prediction models toward personalized precision medicine.