Binge eating disorder is the most common eating disorder in the United States, thought to affect as many as 3 percent of people during their lifetimes. Yet it remains poorly understood.
Now, researchers from the Icahn School of Medicine at Mount Sinai have made important progress with the first genome-wide analysis of binge eating disorder (BED). The study, published in Nature Genetics in August, identified genes that appear to be associated with BED risk. The study also found evidence that iron metabolism may play a role in the disease.
“By applying machine learning to the study of binge eating disorder, we’ve gained important insights into this poorly understood condition, and a new tool for exploring other underdiagnosed diseases,” says Panos Roussos, MD, PhD, Professor of Psychiatry, and Genetics and Genomic Sciences at Icahn Mount Sinai and Director of the Center for Disease Neurogenomics, who is a co-author of the study. “By combining Neuroscience with genomics and big data analysis, we can discover more about how the brain works and ultimately prevent psychiatric disease.”
A Fresh Look at Binge Eating Disorder
Binge eating disorder has significant impacts on a person’s health and well-being. “It can cause substantial distress and impairment in quality of life,” says Trevor Griffen, MD, PhD, a psychiatrist and neuroscientist who collaborated on the recent study while he was a fellow in child and adolescent psychiatry at Mount Sinai. “BED often co-occurs with other psychiatric disorders, such as depression, ADHD, and substance use, and seems to be a nexus of metabolic dysfunction, with associations to conditions like diabetes and high blood pressure.”
Trevor Griffen, MD, PhD
Yet it took a long time for the scientific community to recognize BED as a distinct disorder. It was first included as a new diagnosis when the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) was published in 2014. As a result, the diagnosis is all but absent in the electronic health records and large biobanks that researchers tap into for large-scale genetic analyses. Luckily, the Mount Sinai team developed a workaround.
“A big part of this study was using machine learning to figure out the people most likely to have BED,” says lead author David Burstein, PhD, a data scientist at Mount Sinai who works in the labs of Dr. Roussos and study co-author Georgios Voloudakis, MD, PhD, Assistant Professor of Psychiatry, and Genetics and Genomic Sciences.
Using electronic health record data from more than 767,000 people through the Million Veterans Project, Dr. Burstein and his colleagues applied machine learning approaches to sift through medical diagnoses, prescription medicines, body mass index (BMI) data, and other factors, looking for patterns that would predict if a person had BED. Applying their model to smaller cohorts of people with diagnosed BED, they showed the approach could meaningfully predict the disorder, even in the absence of a formal diagnosis.
Genes Point to New Binge Eating Disorder Treatments
Applying the machine learning model to some 362,000 people for whom genetic information was available, the researchers zeroed in on several genetic loci that appear to be associated with BED risk. One of the genes implicated in the new study is MCHR2, which is associated with the regulation of appetite in the brain. Two others, LRP11 and APOE, have previously been shown to play a role in cholesterol metabolism.
David Burstein, PhD
Another gene identified in the study, HFE, is involved in iron metabolism. The identification of HFE aligns with recent research suggesting iron metabolism may have an important role in regulating overall metabolism, Dr. Griffen says. In particular, iron overload seems to be associated with binge eating, the team found. Interestingly, iron deficiency has been implicated in pica, a disorder that drives people to eat non-food items such as soil or hair.
“There have been hints that iron is a player in the eating disorder world,” Dr. Burstein says. “This new study is more evidence that the mineral deserves a closer look.”
The findings also point toward new directions for treating BED. So far, treatment has mostly focused on repurposing therapies used for other disorders, such as depression or ADHD.
“This study identifies genes and systems that could serve as potential targets for treatments that finally address the underlying biology of BED,” Dr. Griffen says. “It also continues to build evidence that there are biological and genetic drivers of binge eating behaviors. The more we get that message out there, the more we can decrease stigma associated with binge eating.”
A New Tool for Eating Disorder Research
Dr. Griffen is continuing to collaborate with Dr. Roussos and Dr. Voloudakis to expand on their findings, with plans to develop mouse models and dig deeper into the mechanisms. Ultimately, their goal is to develop new treatments that target the underlying biology of BED.
Meanwhile, the researchers are eager to apply their new computational approach to other diseases such as bulimia nervosa—another common eating disorder for which no genome-wide analysis has ever been done.
“Being able to infer a diagnosis from medical records is really significant, not only for BED but for other eating disorders, which are often extremely underdiagnosed” and therefore challenging to study using electronic health records, Dr. Burstein says.
The approach can also extend the science into populations that have been overlooked in past research. Most research on eating disorders has focused on white females. Using machine learning, researchers can more thoroughly study eating disorders in males and populations with other racial or ethnic backgrounds.
“This is exciting work, with so many potential future directions,” Dr. Burstein says.