Rethinking Anorexia Nervosa

Recent P&S research may change attitudes about anorexia nervosa and suggest new ways to treat it
By Susan Canova | Photographs by Jörg Meyer

Over the past 40 years, Timothy Walsh, MD, has seen treatments for anorexia nervosa come and go. When he started treating patients who have eating disorders, most psychiatrists blamed a perfectionist mother and kept parents away from patients during treatment.

Later, SSRI antidepressants became a popular pharmacological treatment for anorexia nervosa. Unfortunately, a randomized clinical trial conducted at Columbia, led by Dr. Walsh’s colleague Evelyn Attia, MD, showed that the SSRI fluoxetine does not help anorexia nervosa patients gain weight or improve otherwise.

More recently, anxiety was hypothesized to play a critical role in anorexia nervosa. But an anti-anxiety medication did not improve caloric intake in a clinical trial led by P&S psychiatrist Joanna Steinglass, MD.

“Over the years physicians have thrown everything they have at anorexia nervosa,” says Dr. Walsh, “but it is still stubbornly hard to treat, particularly as patients get older. And that’s because little is known about the way anorexia nervosa gets started and why it persists.”

Two recent studies at Columbia may start to offer an answer while also suggesting new ways to treat the disorder. A new animal model developed by researchers in the Naomi Berrie Diabetes Center has uncovered the mix of vulnerability factors that trigger anorexia nervosa and a drug that may be able to interrupt the process.

And a new theory put forth by Dr. Walsh, Dr. Steinglass, and other Columbia psychiatrists and psychologists suggests that treating anorexia nervosa as a habit may help people with anorexia nervosa change their destructive eating patterns.


Who’s Who

Evelyn Attia, MD, professor of psychiatry at CUMC and director of the Columbia Center for Eating Disorders

Karin Foerde, PhD, assistant professor of clinical psychology (in psychiatry)

Daphna Shohamy, PhD, associate professor of psychology

Joanna Steinglass, MD, a Florence Irving Associate Professor of Clinical Psychiatry

B. Timothy Walsh, MD, the William and Joy Ruane Professor of Pediatric Psychopharmacology, director of the division of clinical therapeutics at the New York State Psychiatric Institute, and founder and former director of the Columbia Center for Eating Disorders

Lori Zeltser, PhD, associate professor of pathology & cell biology


“In anorexia nervosa, the defining behavior is the restriction of food intake,” says Dr. Walsh. “And patients restrict food in a specific way: They restrict the number of foods they eat and avoid food with fat. All anorexia nervosa patients do this. Often patients will come into our inpatient treatment program and gain substantial weight, but after they leave, they go back to eating the same sorts of food they ate before, setting the stage for relapse. That’s really telling us something about the nature of the illness and why it’s tough to change.”

Columbia researchers began to explore the idea that this behavior may not simply be an example of supreme self-control, as many people view anorexia nervosa, but one that becomes ingrained as an automatic habit. Researchers have begun to apply burgeoning knowledge of how the brain controls behavior to understand anorexia nervosa.

In recent years, cognitive neuroscience has used psychological tasks and brain imaging techniques to identify a behavior’s underlying neural circuits. Drs. Walsh and Steinglass turned to Daphna Shohamy, PhD, a cognitive neuroscientist on the Morningside campus, and her colleague Karin Foerde, PhD, now a faculty member in psychiatry, to develop a food choice task that measures an anorexia nervosa patient’s preference for low-fat foods. Their goal: to see if patients with anorexia nervosa were using different brain circuits when making choices of what to eat.

In the food-choice task, participants were asked to rate the healthiness of various foods and each food’s tastiness. Patients and women without an eating disorder alike rated high-fat foods as less healthy, but the participants with anorexia nervosa rated these foods as much less tasty. And when asked to choose a food to eat, people with anorexia nervosa overwhelmingly selected low-fat foods.

Imaging using fMRI shows that people with anorexia nervosa engage a different part of the brain when making these choices: the dorsal striatum, which plays a role in the acquisition and continued expression of habits.

Dr. Walsh says the imaging results, which were published in the November 2015 issue of Nature Neuroscience, do not prove the habit theory but are consistent with it. “One implication of this line of thought is you have to address the behavior directly,” he says. “Just talking about the stresses of life and psychological factors—though that’s clearly important—won’t be sufficient. The way to change a bad habit is to replace it with a new, better habit.”

That can be difficult, as people raised in the United States learn when they travel to London or Australia. “In the U.S., we’re taught at an early age to look left and then right before crossing the street,” Dr. Walsh says. “It takes real mental effort to reverse the sequence, but you’ll eventually replace the habit after lots of practice. We may need to help the patient make such efforts to adopt new habits of eating and help them practice over and over again until healthier food choices become ingrained.”

The researchers’ previous studies suggest that replacing the habit of eating low-fat foods makes a big impact on recovery. “We’ve seen that the people who make the biggest change in what they eat, not just the amount they eat, do better long term,” says Dr. Walsh.

Dr. Steinglass and others are now testing habit-changing treatments in small clinical trials at Columbia. And Dr. Walsh and his colleagues are continuing to test their theory with people in the very early stages of anorexia nervosa, who may not have developed strong habits.

Dr. Attia, who directs the Columbia Center for Eating Disorders and has treated people with anorexia nervosa for more than 20 years, says the theory resonates with many of the center’s patients. “When we describe our research to patients, they say, ‘Wow, that’s me,’” says Dr. Attia. “Patients will use that word: habit. They say they just can’t break their eating habits. The theory fits the clinical picture of what happens to people, not leading up to the development of the problem but once the behaviors have become entrenched.”



If anorexia nervosa is a habit, how does the habit get started?

“People think that if they’re thinner, they’ll feel better, and that drives the initial dieting,” says Dr. Walsh, “but anorexia nervosa remains uncommon. Why do only a relatively few people develop it?”

Lori Zeltser, PhD, is an obesity researcher in the Department of Pathology & Cell Biology and the Naomi Berrie Diabetes Center, where she investigates developmental influences on neural circuits that control eating behavior. She had not thought much about the science of anorexia nervosa, but a few years ago she saw a call for research proposals from the Klarman Family Foundation, which was eager to tap the expertise of scientists outside the field of eating disorders.

Dr. Zeltser views the disorder through her experience as an obesity researcher—and as someone who has first-hand knowledge of the disorder’s impact. In high school and college she witnessed friends’ struggles with anorexia nervosa.

“Because anorexia nervosa primarily affects adolescent girls, the media and even many doctors have a perception that the disorder is just an extreme form of irrational teenage girl behavior,” she says, “and not a disease with a physiological basis, like addiction or obesity. What I learned as an obesity researcher is that the body engages many parallel systems to fight against weight loss. Even people who are very highly motivated usually can’t keep the weight off. But people with anorexia nervosa can maintain dangerously low body weight. And it even feels good. I suspected/hypothesized that there must be a biological basis for the ability to suppress the powerful drive to regain body weight.”

But even Dr. Zeltser was surprised to learn about the strong genetic underpinnings of anorexia nervosa as she continued to read about the disorder. One study found that first-degree female relatives of patients with anorexia nervosa are 10 times more likely to develop anorexia nervosa themselves than people in the general population. And twin studies indicate that the heritability of the disorder is between 48 percent and 88 percent.

“Patients will use that word: habit. They say they just can’t break their eating habits.”
— Evelyn Attia

“That’s huge; it’s even higher than obesity or diabetes,” says Dr. Zeltser. “There’s no way to have heritability estimates that high without involving a genetic component.”

To better understand anorexia nervosa, Dr. Zeltser decided to first create a better animal model of the disorder. “The development of an animal model for anorexia nervosa would be a landmark advance, because it’s one of the things holding us back,” says Dr. Walsh. “Current models haven’t changed our understanding of the illness in terms of illuminating why people get the illness and what we can do to help.”

Dr. Zeltser’s mouse model may fit the bill, Dr. Walsh says. “Her model is an attempt to identify these vulnerability factors and tease apart their contributions.” The paper describing the model was published in early 2016 in Translational Psychiatry.

Because of the strong genetic component of anorexia nervosa, Dr. Zeltser decided to create a model carrying a gene variant (BDNF-Val66Met) that has been linked to anorexia nervosa. (BDNF, short for brain-derived neurotrophic factor, is known to play a role in the development of brain circuits that control anxiety and eating behaviors.)

But because no single gene is sufficient to cause anorexia nervosa on its own, “this means that other environmental factors are required to elicit the disease state,” Dr. Zeltser says.

The obvious environmental factor is dieting, so Dr. Zeltser reduced the amount of food the animals ate. Peer pressure—expressed in the desire to be thin—is another driver of anorexia nervosa in most patients and one that, at first glance, cannot be modeled in a mouse.

“We took a step back and reasoned that what we see in human anorexia nervosa can be classified more generally as social stress. Social stress, social anxiety: That’s what teenage girls are feeling,” Dr. Zeltser says. “We produced social stress in mice by housing the normally social animals in isolation, a manipulation that impacts the same brain regions as social stress in humans.”

Dr. Zeltser expected to see subtle differences in eating or body weight when all three factors—the genetic variant, food restrictions, and social stress—were present. But among mice with all three risk factors, 40 percent completely stopped eating just a few days into the experiment, sometimes resulting in death. “I never thought we would induce such a dramatic effect,” she says.

“The combination of all three risk factors was critical to elicit a robust effect on feeding,” says Dr. Zeltser. “If we eliminated any one variable, the incidence of anorexic behavior was markedly reduced.”

The age of the mice also turned out to have a dramatic effect. Anorexia nervosa in people primarily affects adolescents and young women; in Dr. Zeltser’s experiments, adolescent mice were particularly vulnerable to developing anorexia nervosa. When food restrictions and social stress were applied in mature mice, the genetically susceptible mice did not become anorexic.

The findings suggest dieting is a key trigger in anorexia nervosa. That sounds obvious, but it provides important insight into the “chicken vs. egg” debate about the role of dieting: Does the disorder lead to dieting, or does dieting lead to the disorder?

For Dr. Zeltser, the finding suggests dieting in adolescence is potentially risky. “We’re careful to say teenagers shouldn’t be drinking alcohol because of the greater risk of dependency. It may be the same with dieting.”



After first feeling frustrated that not all of the mice with all three risk factors exhibited anorexic behavior, Dr. Zeltser decided to use the opportunity to look for differences in the brains of the mice that stopped eating versus mice with the same genetic and environmental risk factors that ate normally.

And she found a difference. The vulnerable mice, she found, have increased levels of a particular receptor in the brain. Most excitingly, when the mice that stopped eating were given a compound that blocked the receptor, they started to eat again.

Analogous compounds are already being tested in early clinical trials for other psychiatric conditions. However, Dr. Zeltser needs to do more work before they can be tested in people with anorexia nervosa.

“We see in mice that the drug has no effect on eating behavior when levels of the receptor are not elevated in the brain,” says Dr. Zeltser, “so I don’t think this drug will work for every person with anorexia nervosa. We’ll need a personalized medicine approach.”

A new mouse model carries a gene variant that has been linked to anorexia nervosa.

The next critical steps are to identify the relevant brain targets in the mouse model and then to develop methods to identify people with elevated levels of the receptor.

Together, the Zeltser and Walsh studies suggest a two-hit system. “We think that there is an initial override in the system that allows the individual to suppress eating behavior in the face of extreme weight loss, and then the circuits regulating habits kick in to maintain the behavior,” says Dr. Zeltser. But until better treatments are found, Dr. Walsh says, early aggressive treatment is the best option.

“When I talk to doctors about anorexia nervosa, I tell them to be aggressive and to focus on changing the behavior,” he says. “Once it gets started, it seems to take on a life of its own.”