A tale of two anorexia genetics studies

Over the past month, there have been two major, news-grabbing studies on the genetics of anorexia. Both of these studies carried headlines saying something like “Anorexia genes found!”

There are several problems with this:

  1. There are no such thing as “anorexia genes”
  2. These two studies were VERY different and found VERY different things.

Given that the genetic underpinnings of EDs are a Big Deal in the research and lay communities right now, and that I haven’t covered specific genetics studies thus far, I thought that these studies are actually very good ways to explain how we discover genetic links to illnesses like EDs, and what they do (and don’t) tell us.

The missing heritability

Researchers call it the ‘missing heritability’ issue. It’s not specific to EDs or psychiatric illnesses, although this is where you will hear it discussed most frequently. The issue is that twin studies on anorexia have found that roughly 55% of the reason some twins developed AN and others didn’t was due to genetic factors- factor in confidence ratios, and that percentage can go upwards of 80% (Bulik et al., 2006). The problem was that when researchers went spelunking through the billions of As, Ts, Gs, and Cs in the human genome, they couldn’t consistently identify any gene that was conclusively linked to an increase AN risk. The twin studies were consistent across different populations, indicating that genetics was a major factor in AN (and all EDs), but any actual genes remained mysterious.

Scientists were never expecting to find ‘the’ anorexia gene. They know there would never be one. Genes don’t cause disease–they make mRNA which makes proteins. A gene’s sequence can alter the amino acids that build a protein. The sequence of amino acids in a protein determine the protein’s shape, and the protein’s shape determines how well it works. How well a protein does its job can alter your risk of developing an illness, such as an eating disorder. Given that an eating disorder involves problems not just with food and appetite, but also emotion regulation and reward, there will likely be a whole host of genetic mutations that slightly alter your risk of developing an ED. This can make these genes harder to find because their impact across a population will be small, and researchers won’t be able to find the proverbial ‘smoking gun.’

When researchers begin to look for these types of mutation that contribute risk for developing an illness, they frequently use a technique known as genome-wide association studies (GWAS, pronounced “gee-wass”). In GWAS, researchers search for variations in single nucleotides (known as Single Nucleotide Polymorphisms, or SNPs- pronounced “snips”). They look at the gene sequences in people with an illness and those without and see if any SNPs are more or less common in the individuals with the illness. If they do identify a SNP, then the researchers say that this gene (or area of the genome) is associated with an illness.

It’s what researchers did in a recent paper in Molecular Psychiatry, in which the researchers found an association with variants of the EPHX2 gene and AN (Scott-Van Zeeland et al., 2013). The researchers started with genetic information from 334 individuals, of which 261 had early onset, severe AN and 73 were controls, and began sequencing 152 genes known to be associated with feeding behavior or other aspects of AN (such as GABA, dopamine, and other genes), including the area immediately around the gene. Of all of these 152 genes, the researchers only identified several variants in two different genes.

When they looked closer at a larger group of AN cases and controls, the researchers found that only one gene (EPHX2) was significantly associated with AN in a variety of populations. The gene, whose full name is epoxide hydrolase 2, is an enzyme that regulates cholesterol production. Many sufferers of AN suffer from high cholesterol in the acute stage of illness, which has generally been attributed to the effects of starvation. These new results hint that it might not just be starvation.

From a press release:

It isn’t yet clear how EPHX2 variants that cause an abnormal metabolism of cholesterol would help trigger or maintain anorexia. But Schork noted that people with anorexia often have remarkably high cholesterol levels in their blood, despite being severely malnourished. Moreover, there have been suggestions from other studies that weight loss, for example in people with depression, can lead to increases in cholesterol levels. At the same time, there is evidence that cholesterol, a basic building block of cells, particularly in the brain, has a positive association with mood. Conceivably, some anorexics for genetic reasons may feel an improved mood, via higher cholesterol, by not eating.

“The hypothesis would be that in some anorexics the normal metabolism of cholesterol is disrupted, which could influence their mood as well as their ability to survive despite severe caloric restriction,” said Schork.

The problem with GWAS studies like this is that they are prone to false positives. Sort through enough genetic data and you’re bound to find a link between some SNP and your illness, just like if you start flipping coins you’ll get a streak of tails. You might think you have a trick coin, but it’s really just the vagaries of chance. The researchers tried to guard against it by searching through a variety of populations and getting a large sample size, which is good that variants in the gene were significantly associated with AN in all of these. However, it’s still just one study and all of the populations were European (of course, gene variants can be population-specific, since many populations didn’t mix together that frequently until very recently). My basic response is: it’s promising and interesting and definitely needs more research, but I have seen many strong results fail to be replicated. If someone else finds these results, then I’ll say we’re on to something.

Rare mutations, strong effects

GWAS tends to search for common-ish variants with small effects. Other types of genetic studies look for rare mutations with much stronger effects. On the diagram below, GWAS is at the bottom right. The next study I’m going to talk about is at the top left.


Image credit: Wikimedia Commons

In a separate study published in the Journal of Clinical Investigation, researchers looked at two families in which multiple members had AN or BN across three generations (Cui et al., 2013). In one family, 10/20 members had full-syndrome AN or BN (they excluded EDNOS just to be on the safe side, in case the abnormal eating behavior wasn’t a true eating disorder and instead only looked like one on the surface, due to growing up in an environment where eating disorders were the norm)–tremendously higher than the prevalence you would expect. The researchers first identified regions of interest on various chromosomes and then fully sequenced the genomes of 1-2 people from each family who had AN to look for specific mutations.

AN pedigrees

In the first family, they found a single nucleotide mutation in the estrogen-related receptor α (ESRRA) gene. All ten individuals with an ED had this mutation, and nine of ten without an ED did not. The protein coded for by ESRRA is a transcription factor and helps to regulate the activity of other genes. Interestingly, so is the other gene mutation found in the second family, HDAC4 (histone deacetylase 4). Even more interesting is that these two genes bind to each other in the cell. The effects of both of these mutations is to decrease ESSRA activity, which somehow increases the risk of developing anorexia.

Unlike the first study, these mutations were identified by direct genome sequencing, and it’s basically pretty much certain that these mutations are responsible for the astronomical rates of AN and BN. The caveat is this: most people with EDs probably don’t have these mutations. These are rare mutations with VERY strong effects, unlike most gene variants thought to contribute to AN, which are slightly more common and have smaller effects. Thus a person with an ED likely carries several or even many of these variants that help to tip them over into developing an ED under certain circumstances. Each person’s risk of developing AN will be a combination of environmental and genetic effects. Some mutations have a much stronger genetic effect, such that environmental factors play a much smaller role in the developing of an ED. Other people may have environmental factors that dominate their own risk. Most people will be in between.

A genetic link to AN doesn’t mean that your chances of developing anorexia are somehow foretold on a DNA magic 8 ball hidden deep in the nucleus of your cells. It doesn’t mean that there’s nothing you can do about your disorder or that you’re totally screwed by your double helix. It just means that some people are much more likely to develop an ED than others.

Yes, both of these studies found genes linked to AN, but they are very different genes with very different effects. The genetics of EDs is extremely complicated!


Bulik, C. M., Sullivan, P. F., Tozzi, F., Furberg, H., Lichtenstein, P., & Pedersen, N. L. (2006). Prevalence, heritability, and prospective risk factors for anorexia nervosa. Archives of general psychiatry63(3), 305. doi:10.1001/archpsyc.63.3.305

Cui, H., Moore, J., Ashimi, S. S., Mason, B. L., Drawbridge, J. N., Han, S., … & Lutter, M. (2013). Eating disorder predisposition is associated with ESRRA and HDAC4 mutations. The Journal of Clinical Investigation123(11), 0-0. doi:10.1172/JCI71400

Scott-Van Zeeland, A. A., Bloss, C. S., Tewhey, R., Bansal, V., Torkamani, A., Libiger, O., … & Schork, N. J. (2013). Evidence for the role of EPHX2 gene variants in anorexia nervosa. Molecular psychiatrydoi: 10.1038/mp.2013.91

Twitter Digg Delicious Stumbleupon Technorati Facebook Email

18 Responses to “A tale of two anorexia genetics studies”

  1. Thank you for explaining this research for your readers. By coincidence I was talking about this study this morning with my husband. Last week I had read a short article listed on your blog, under “Eating Disorder News.” The article did mention the findings need to be duplicated to determine their validity.

    I read way more into this article than what was there. I guess some days I wish there could be a little pill that would take away this “hell” in my life. Your willingness to tackle these finding and writing them in laymen terms is very much appreciated.

  2. “A genetic link to AN doesn’t mean that your chances of developing anorexia are somehow foretold on a DNA magic 8 ball hidden deep in the nucleus of your cells. It doesn’t mean that there’s nothing you can do about your disorder or that you’re totally screwed by your double helix. It just means that some people are much more likely to develop an ED than others.”

    Your conclusion is perhaps the most comforting explanation of all and thank you for saying this……

  3. Great and informative post! I’m not too familiar with genetics/genetics studies so this was really helpful. I do know that there’s the popular misconception that a singular “AN gene” will be identified but unfortunately — like you said — psychological disorders are complex, i.e. polygenic in nature. And yes, we cannot discount the power of environmental influences!

    But anyway, after reading this post, I’m totally motivated to read up on more gene association studies. Thanks, Carrie!

  4. It bugs me SO much that they cut out anyone with EDNOS “in case the abnormal eating behavior was the result of growing up in a family with seriously messed up attitudes about food and not a true eating disorder”.
    There are so many things wrong with this statement and this “logic”, I don’t even know where to begin.
    You’d think people who research EDs for a living would understand how messed up that “logic” is.

    • That sentence didn’t quite come out like I meant it to, so I’ve edited it slightly. Thanks for pointing out the problem- I didn’t like it when I first posted but I couldn’t put my finger on why.

      You have to keep in mind that the people you include as part of a genetics study is going to be MUCH MUCH stricter in terms of diagnostic criteria than you would for other clinical studies. When you don’t know what genes you’re looking for, you need to be ABSOLUTELY SURE that the people you classify as “cases” have the disease in question.

      With EDNOS, the researchers couldn’t always be absolutely sure that the people actually had an eating disorder or they simply looked a lot like someone who did. With AN and BN, the diagnostic criteria are very clear-cut. You meet them or you don’t. With EDNOS, they’re not so clear cut, and it’s hard to distinguish between disordered eating and an eating disorder. That’s where the confusion comes in and that’s why they were excluded.

      It’s impossible to say how much genes and environment contribute to a specific individual’s eating disorder- we can only do so for large populations. Most of the EDNOS people did not carry the mutations in question, which means their ED (assuming they had one) was caused by other genetic mutations and/or environmental factors.

      It DOESN’T mean that EDNOS isn’t an eating disorder and it DOESN’T mean that EDNOS has no genetic basis and it DOESN’T mean that people with EDNOS don’t deserve treatment. It just means that determining for sure whether someone has EDNOS can be tricky for these types of studies. I do believe that they probably should have been considered cases, but I don’t do this type of genetic research, so I’m not as familiar with what they include and what they don’t.

      It’s very much a result of this specific study protocol and says nothing about EDs in general. I agree that EDNOS gets the short end of the stick on most ED research, but it’s so heterogeneous that it’s harder to determine what’s what.

  5. I’m stunned about the bit where anorexics have high cholesterol. In all my research and reading up on eds, how did I never happen to run across this? Or was it subconscience denial and avoidance of data? My cholesterol is routinely high. I have continually justified my need to continue cutting fats b/c of this. Damn, eds can twist things up and make a smart person think stupid.

  6. Most physicians don’t know about this, so don’t beat yourself up. It’s the “standard” medical knowledge.

    Hell, when I see my physician, despite having told her many times that high cholesterol runs in my family and that mine has stayed high regardless of diet and exercise habits, she still says “diet and exercise!” Of course, this is completely ignoring my history and (what really irks me) is that she never asks me about what I’m eating and how much I’m exercising.

    Rant aside, high cholesterol in AN is actually a sign to eat more fats rather than less. If it remains after weight restoration, then it’s something to try and manage but probably without dietary restrictions given risk of relapse.

    • Thanks for letting me know I’m not a dope for not noticing this. :)

      Doctors usually look at my cholesterol and are baffled, then I take that as affirmation that I must be smarter than them, and know that I need to limit more fats.

      Eat more fats… can you give some specific examples of good fats to eat? Yes, of course, I could look this up. And, of course, my dietitian has made suggestions. But I like you more. I tend to hear things better from someone I like.

      Would you mind sharing some suggestions? :)

      • Olive oil, avocado, nuts (any kind- peanuts, almonds, cashews, macadamias, etc), coconut oil (if you like the taste. I like coconut but in small quantities, so I’ve generally passed up this option). Stir fries in canola and sesame oil are good.

        There’s also nothing wrong with dairy fats and butter in moderation, too.

  7. Once again, Carrie, you do a phenomenal job of breaking down the scientific literature into something that lay people can understand. Re – high cholesterol levels in AN – I’ve heard of this phenomenon (though too my knowledge have never experienced it myself), and always assumed it had something to do with the metabolism of body fat during acute (or subacute) starvation. The genetic “twist” cited in the first study is…interesting.

  8. Jennifer Mahnke, MD Reply October 24, 2013 at 11:06 am

    I was fascinated with the cholesterol findings. My son and I have both had anorexia nervosa. My cholesterol has always been high, and I oddly found it improved as I had fewer ED behaviors. The worse (more fats in particular) my diet was (and the heavier I was BMI close to 25), the better my cholesterol!I will be interested to hear more on this and am hoping there will be some studies on mitochondrial DNA.

    • More fats in the diet isn’t necessarily bad, so there’s that. And some of us naturally have BMIs that are around 24-25 and have an AN history and this is where our bodies just seem to want to be at. So it could be that your cholesterol improved because your body still thought it was starving at a lower weight. Mine does. I’ve never correlated with cholesterol, but all the other signs are there.

      I don’t know that I have seen any studies on mtDNA and EDs, although that could be very interesting.

  9. Hi Carrie,
    Thanks for the post. I had read about both these studies, and the snazzy headlines. I treat people with eating disorders and I think it would be awesome if there was a gene that could be identified and targeted. My job might disappear but quite a few people would live much better lives.

    Unfortunately, life is never so simple and genes only give us a glimpse into the problem. As you said, gene expression, environment, and a whole host of other factors play a role.

    I suspect that even if a gene therapy was developed, my clients would still need therapy and medical monitoring and meal plans and everything else they need right now. It would just be another tool in our bag.

    Thanks again!

    • Except for those rare families with the really strong mutations, I don’t really think that gene therapy will ever have much therapeutic use for EDs- too many genes with small effects.

      Identifying risk genes is important in understanding etiology and (verrrrry long-term) for developing new treatments, but right now, it doesn’t affect as much how EDs are treated. We know we need to stop behavior usage, normalize nutrition, deal with co-morbids and prevent relapse. What needs to happen isn’t going to change with knowing the genes (other than perhaps developing newer, better medications), but how it happens might.

  10. Carrie, I tracked you here after reading an article you had written for Lancet on the “dangers of new synthetic drugs.” I’m an 81 year old MD in California with an interest in the subject of medical nosology and how it has been impacted by the drug war.

  11. Super fascinating! I was really intrigued to hear about a possibility of a genetic link because my birthmom was anorexic. I am adopted and my adoptive family is relatively ED free. But i have struggled with ED since age 12


  1. News You Can Use: October 27-November 3 2013 « Eating Disorder Pro - October 29, 2013

    […] A Tale of Two Anorexia Genetics Studies - Over the past month, there have been two major, news-grabbing studies on the genetics of anorexia. Both of these studies carried headlines saying something like “Anorexia genes found!”. There are several problems with this. Learn More. […]

Leave a Reply