Interview with Stephanie Dulawa, Ph.D.

I spoke to Stephanie Dulawa, professor at University of California San Diego, about two articles she co-authored on the subject of Activity Based Anorexia (ABA). The first came out in 2012, and was co-written with Stephanie Klenotich. The second came out in 2021, and was co-written by with Jie Zhang.

Here are some of the topics we discuss:

  • How to induce something like anorexia in lab animals

  • Whether animal models are relevant for humans

  • Is anorexia purely dysfunctional, or does it serve some purpose?

  • What are the hopes for a drug treatment for anorexia?

Notes: The interview transcript below has been edited and organized for concision and clarity. Any mistakes are probably mine. Neither interviewer nor interviewee necessarily endorse each other's opinions. None of this is medical advice.

Q. Could you give the elevator summary of your 2012 article with Stephanie Klenotich? I call this “the Stephanies paper.”

A. This article was an outline for other researchers who are studying aspects of anorexia nervosa (AN). We can’t model an entire psychiatric disorder in an animal, but we can model particular aspects. By doing this, we hope to learn about how the brain behaves and how we might intervene.

For anorexia, what we can model is the paradoxical response to negative energy balance. In AN, you’re losing weight and beginning to starve. But instead of eating more, you continue to do things that decrease your body weight. We can put laboratory mice into an environment that causes them to do the same thing.

What we do has two components. One is restricting the mice’s food availability. We give them plenty of food, but only during certain hours of the day–say, noon to three o’clock. If we just do that, the animals adapt. At first they might lose some weight, but they learn to eat everything they need in those three hours.

The other is housing them alone and giving them a running wheel. Mice really like to run and exercise. They might lose a little bit of weight when they run, but they’re not going to run themselves to death.

If we do both of these things at once, the brain (or body) breaks in a very strange way: the animals will run more on the wheel and not eat when food is available. In fact, they start to eat less and less. They can get to a point where they’ve lost 20% of their body weight, but they won’t approach their food. They just keep running. Food is freely available to them but they just don’t make the choice to eat it anymore. This mimics a core feature of AN in humans.

Q. The literature refers to this as “Activity Based Anorexia,” or ABA. What fraction of mice can develop this ABA? Is it more like 20% or 80%? Does it happen quickly?

It varies by mouse strain. The general public doesn’t realize it because they don’t see all the research, but anorexia is quite genetic. 80% is a good guess, and it does start quickly. But the speed at which different mice get into trouble varies. For some it can be three days; others last for two weeks.

By the way, we don’t let the mice starve to death. We feel like we have our answer when they lose 25% of their body weight. Then we let them recover.

Q. There are other articles about ABA in rats, pigs, and hamsters. Given that, should we assume most mammals are vulnerable to it, including humans? Is studying this in humans off-limits for ethical reasons?

A. As far as I know, the mammals that have been tested have shown it. When we look for ABA we tend to find it.

Studies where people are systematically starved aren’t ever done now. However, some studies look at people who already have anorexia. Some of these show similarities to the animal studies. For example, animals tend to run more right before their food comes. Anorexics also seem to increase their restlessness right before meals.

We also see examples that back up the idea of ABA in humans in athletics. Take Olympic figure skaters. They exercise at a high level for hours and hours every day. Sometimes they’re following regimens where they only eat after 6 PM. That approximates the ABA protocol, and there’s a lot of anorexia in that field.

Q. The ABA protocol has been known since the 1950s. But the general public, and even people who treat anorexics, don’t seem to be aware of it. Is there some scientific reason for this? Some result that makes us think that it isn’t relevant for humans?

A. I think science isn’t always disseminated well to the public. People focus on things like fashion magazines and pressure to be thin. Those things could play a role, but you can go back and find examples of anorexia existing long before we had modern western culture.

Thinking of social pressures as an absolute cause has perhaps been a distraction. Genetics and biology play a very strong role. But just because anorexia is mostly genetic or biological, it doesn’t mean it’s fixed. We can still treat it!

Q. Your 2021 article emphasizes that ABA in animals doesn’t model all of AN in humans. That is, we don’t see body image disturbance or fear of gaining weight. But are we sure about that? Could mice have their own versions of these things?

A. We just have to say that we don’t know. They might have something, a sensation of not needing to eat. But we can’t get at that directly.

Q. I read about a study about people with anorexia misjudging how well they could fit through narrow openings. Could we do something similar with mice to determine whether they have a distorted body image?

A. That’s very interesting! I don’t think that experiment has been done with mice. You could imagine setting up something where they have to quickly squeeze through a hole.

Q. Is it plausible that the paradoxical response to negative energy balance is an attractor in biological space? More than one thing could lead to it? Maybe limited food availability is one cause, exposure to pictures of skinny models is another?

A. One theoryand I should stress that it’s a theoryis that mammals have a built-in famine response program.

Say you’re an animal and you’ve been walking around for a few days. You don’t see very much food to eat in the area; there’s a lot less than usual. What do you do? You could stay in the area and hunker down to save energy. But that might be riskywhat if you don’t find anything to eat?

Now say you take off, you run like crazy, and you try to get to another area that has more food. Some people theorize that ABA is that mechanism saying “there’s not enough food here, this could be a serious problem, I need to migrate out of this area.”

Take birds as an example. They’re living somewhere with a good food source. Then winter comes and that food source is starting to go away. They get up and fly unbelievable distances and land in some nice, sunny place where they have lots of food again.

ABA may be pushing this phenomenon into a place that’s dysfunctional. In the controlled experiments we don’t let the mice get anywhere. I’m sure that if we let them run to somewhere with food, they would eat it. But we keep them in the hyperactive phase.

Q. This is Shan Guisinger’s Adapted to Flee Famine idea, right? She suggests that it might be useful for a population to have some fraction of individuals with a strong version of this migratory response.

A. That rings true. Different traits can either be an advantage or disadvantage depending on the situation you’re in. If you’re trapped in the north, you’re freezing, and your tribe is about to starve…

Most traits in a population will have something like a bell curve distribution. You might have some individuals who can get into migration mode very easily. For those individuals at the extreme, any cue might kick them into that mode. Let’s say they start a diet. They just want to lose a couple pounds. But when they restrict their food intake for a few days their body reacts strongly. They get restless, they start to care less about eating, they constantly want to be active.

We often see diets trigger anorexia. But it can also be something like an illness that causes you to eat less. The stomach flu, say, or a surgery.

Q. So we think some people are just sensitive to the loss of body fat, and the reason doesn’t matter?

A. This is something we really want to understand. What exactly are the signals that are triggering this?

Body fat is really an organ. It’s alive, and it secretes chemicals. People who have been obese at any time in their lives usually don’t become anorexic. There’s a real metabolic vulnerability for people who have usually been underweight.

Also, the ABA model breaks down if we give mice high-fat food. We normally use a chow with a healthy balance of carbohydrates, fats, and proteins. But with high-fat food, the animal doesn’t go into ABA; it just eats.

Q. Your 2021 article reviews some recent findings from genomics that conclude AN should be thought of as a “metabo-psychiatric” disorder. Could you give an elevator summary of that research? Is that term applied to anything else?

A. The researchers had access to the genes of tens of thousands of individuals, some with anorexia and some without. Essentially what they found was that set of genetic loci that predispose someone to AN also relate to metabolic function.

I don’t think anyone is describing anything else as metabo-psychiatric yet. But I wouldn’t be surprised if metabolic traits are a risk factor for other psychiatric disorders. It’s interesting: One of the biggest genetic correlations they found was between AN and OCD. And people with OCD tend to be lean.

I think it’s valuable for patients and their families to know that being anorexic isn’t a decision. It’s not that the patient just wants to be thin. That view is stigmatizing, and maybe keeps people from advocating for themselves or their family members.

But if we can recognize that AN is an illness like any other, that it has to do with biological predisposition, that metabolism plays a role, that helps. Those are not things that are just a person’s choice.

Q. We usually think of AN as something that affects mostly females. The female:male ratio is usually cited as 10:1. Is that the case for ABA?

A. I suspect there are important sex differences, but male mice will actually develop ABA. Sometimes it takes them a little bit longer. There are definitely studies of just male mice.

Also, it’s amazing how disorders turn out not to be as rare as we think. There are more people that have the symptoms, but they’re not quite as severe. 1% incidence becomes maybe 2% or 3% if we account for heterogeneity. This applies to OCD and autism spectrum disorders as well.

Q. What do you think an ABA-informed treatment plan for AN would look like?

A. Unfortunately I can’t comment quite as much on the clinical side, since I don’t have the experience. But it’s an interesting question for clinicians to consider.

Researchers tend to think about potential drug treatments, because there are none right now. We try to discover neural mechanisms and find targets where we can have a drug effect.

Q. I noticed that in the ABA studies, animals tend to recover all of their lost weight and then some. Should that be a goal for treating humans with AN?

A. Animals do overshoot their original weight once we restore their access to food. You could say that it’s a failure of ABA as a model for AN, because it’s so difficult for anorexics to get back to normal.

But it still might capture something relevant. Sometimes restricting anorexics start to binge - the body just says “I can’t take this anymore!” Clinicians have told me that it’s actually a good sign if an anorexic is binging. Because the most dangerous state of the illness has them not eating at all.

So the mice going beyond their original weight might provide a model for early AN recovery. It might help us understand what’s happening in the brain when that’s going on.

The mice don’t revert to eating less again once we’ve removed them from the paradigm. I think part of the reason is we’re using relatively normal mice, not ones that have a specific genetic vulnerability to anorexia. Interestingly, ABA was reported in pigs that were bred to produce lean meat. They developed the condition after weaning. That’s a period in which they’re not getting as much food. It caused them to become very active and restless in their pens. There was no need for the running wheel.

Q. To close us out, could you talk about what you think the future holds for AN research and treatment?

A. I really hope that the work to develop drug treatments continues. There are a good number of researchers who are interested in this right now. These would be very helpful to have in addition to psycho-social treatment. But they would hopefully also improve people’s understanding of AN. They might change perceptions to be more aligned to the truth, which is that disorders like AN aren’t that different from diabetes or cancer.