Genome-wide analysis of binge-eating disorder identifies the first three risk loci and implicates iron metabolism

Abstract

Binge-eating disorder (BED) is the most common eating disorder yet its genetic architecture remains largely unknown. Studying BED is challenging because it is often comorbid with obesity, a common and highly polygenic trait, and it is underdiagnosed in biobank datasets. To address this limitation, we apply a supervised machine learning approach to estimate the probability of each individual having BED based on electronic medical records from the Million Veteran Program. We perform a genome-wide association study on individuals of African (n = 77,574) and European (n = 285,138) ancestry while controlling for body mass index to identify three independent loci near the HFE, MCHR2 and LRP11 genes, which are reproducible across three independent cohorts. We identify genetic association between BED and several neuropsychiatric traits and implicate iron metabolism in the pathophysiology of BED. Overall, our findings provide insights into the genetics underlying BED and suggest directions for future translational research.

Valid inference for machine learning-assisted GWAS

“…We use the same notations 𝛽 " + , ,% )*&'( , 𝛽 " +,% &'( and 𝛽 " + , ,% &'( to denote three GWAS summary statistics as described in the main texts. The estimator in existing methods can be written as the non-negative weighted sum of these three GWAS estimators: 𝑤 6 𝛽 " for MTAG (𝜌 is the genetic correlation, ℎ + , 4 and ℎ + 4 are the heritability for imputed and observed phenotypes), and 𝑐 = 1 for other methods.…”

“…Genome-wide association study (GWAS) is a powerful tool for identifying genetic variants associated with complex human traits 1 . However, even in the era of biobank cohorts with tens of thousands of individuals, high-quality phenotype data is often lacking due to the costly technology for phenotypic measurement, invasive procedure for sample collection, or a lack of commitment to study participation [2][3][4][5][6][7] . These challenges severely reduce the statistical power of GWAS on many valuable phenotypes, compromising genetic discoveries and efforts to uncover new therapeutic targets 3 .…”

The Eating Disorders Genetics Initiative (EDGI) United Kingdom

“…No GWASs of clinically diagnosed bulimia nervosa, binge-eating disorder, or ARFID have yet been conducted, but GWASs using proxy phenotyping approaches have been performed. In a GWAS of machine learning predicted binge-eating disorder based on details in medical records, three loci were genome-wide significant, after adjusting for body mass index (BMI; (Burstein et al, 2022). The first GWAS of ARFID used proxy phenotyping within a subsample of the autism study SPARK, including 3,142 autistic children and 2,205 of their parents (Koomar et al, 2021).…”

Anorexia nervosa polygenic risk, beyond diagnoses: relationship with adolescent disordered eating and behaviours in an Australian female twin population

“…To our knowledge, AN is currently the only clinically diagnosed eating disorder to have been the subject of large GWASs, although recruitment for GWASs on BN and BED are underway (Bulik et al, 2022; Steiger & Booij, 2020). A GWAS on BED has been conducted recently, however cases were predicted using machine learning as clinical BED diagnoses were not collected (Burstein et al, 2022). Machine learning was applied to distinguish between clinically diagnosed BED and obesity, but this does mean there may be some misassignment.…”