Propping Open the Doors to the Ivory Tower: Follow-up to #ICED2015


Published: May 1, 2015

After delivering the keynote address at #ICED2015 in which I discussed, “Towards an Engaged Science of Eating Disorders: Opening the Doors to the Ivory Tower,” I received some excellent questions from attendees. To answer them, I engaged the help of three of my collaborators whose work I showcased in the talk. Thanks to Brendan Bulik-Sullivan, Ian Carroll, and David Pisetsky for joining me in this continuation of engaged science!

Interview with Brendan Bulik-Sullivan on Genetic Correlations:

Dr. Bulik asks: Brendan, you reported negative genetic correlations between anorexia nervosa and some immunological diseases, yet clinically we see lots of immunological illnesses in individuals with eating disorders and their family members, perhaps suggesting a positive phenotypic correlation. Could you help us understand this?

Brendan replies: I wouldn’t call the three genetic correlations between AN and immunological diseases that we report in our paper ( statistically significantly negative. The confidence intervals contain zero. We can, however, rule out large positive genetic correlations.

The increased prevalence of immunological disease in people with AN and their family members could be caused by something other than genetic correlation between AN and immunological disease. For example, if anorexia-related behavior (e.g., prolonged starvation) increases risk for immunological disease, then we would observe increased prevalence of immunological disease among AN patients compared to the general population. However, we might not see a strong positive genetic correlation: AN is rare, so the casual path from AN to immunological disease would account for only a very small fraction of immunological disease cases.

Screen Shot 2015-05-01 at 1.28.33 PM[5]

Brendan Bulik-Sullivan Bioinformatics Analyst, Broad Institute

In addition, we only looked at a three immunological diseases: Crohn’s disease, ulcerative colitis, and rheumatoid arthritis—the diseases for which we were able to obtain public GWAS data. There are many other immunological diseases besides these three. Unlike psychiatric disorders, for which the cross-disorder genetic correlations are typically quite high, immunological diseases typically don’t have strong positive genetic correlations with each other. Immunological diseases do tend to share a lot of genes, but the directions of effect often don’t align (e.g., some variants that increase risk for rheumatoid arthritis are protective for Crohn’s disease; some variants increase risk for both; see e.g., this paper). Consequently, the genetic correlation results that we obtain for the three traits in our paper have little bearing on the genetic correlations between AN and the other immunological diseases that we were not able to investigate.

I think that using data from genetic studies to better understand comorbidity is a promising strategy for future inquiry for all heritable diseases and AN in particular. Nevertheless, the results from our study do not yet provide definitive answers about how and whether anorexia and immunological diseases are related genetically.

Interview with Dr. Ian Carroll on Intestinal Microbiota:

Dr. Bulik asks: Dr. Carroll, we are very excited that you have turned your expertise to understanding the role of the intestinal microbiota in anorexia nervosa. Lots of interesting questions arose after my keynote at #ICED2015, and this was the most common. We see from the literature that individuals with obesity appear to have a adaptations of the intestinal microbiota that make them more efficient at extracting energy from their diet. But your data from individuals who were acutely underweight and then re-nourished suggest that in the low weight state, individuals with AN might also be extremely efficient at extracting energy from their diet which allows them to survive on so few calories. How do we reconcile these two observations, especially if we know that the bacteria present in obese individuals and individuals with AN differ? Ian Carroll, PhD Assistant Professor of Medicine, UNC Chapel Hill

Ian Carroll, PhD
Assistant Professor of Medicine, UNC Chapel Hill

Dr. Ian Carroll replies: The key word here is adapted. We need to consider the entire microbial ecosystem in the gut. In an obese individual, the intestinal bugs are naturally responding to the environment. They are exposed to lots of sugar, protein, and fat and what that does to the ecosystem is select for bugs that can thrive in that situation. The current theory is that because these nutrient loving bugs have bloomed, they’re able to process nutrients in a more efficient way. That allows an obese individual to get more energy from the diet than someone who does not have this type of adapted ecosystem.

Originally, I had thought that this would happen in a stepwise fashion so that we might see the opposite in anorexia nervosa. I assumed—wrongly—that people with anorexia nervosa would actually be less efficient at extracting energy from the diet. But I was wrong because I wasn’t thinking like a microbial ecologist. This is not a stepwise process. Both obesity and anorexia nervosa are extreme states. Obesity is a state of extreme nutrient availability and anorexia is a case of extreme nutrient poverty. Just like in obesity, in anorexia you’re selecting for bugs that can survive in an extreme environment. In the anorexic environment, the nutrient loving bugs die off because they don’t have the sugar, fats, and proteins necessary to fuel their growth. So you end up selecting for bugs that can live in a nutrient impoverished environment that are able to efficiently extract energy for survival. The theory is and preliminary data suggest that these efficient bugs have outcompeted the other bugs for real estate in the anorexic microbiota. So the composition of the obese microbiota and anorexic microbiota are very different (i.e., different bugs) because of different environmental stressors and different nutrient supplies, but the functional impact is the same. Both are efficient at energy extraction, but it’s not the same players.

Interview with Professor David Pisetsky on Immunology

Dr. Bulik asks: Dr. Pisetsky, thank you for the opportunity to present some of our work at #ICED2015. Lots of good questions arose from the audience and I was hoping you could shed some light on the answers. We are all very intrigued by your preliminary observations of “uncoupling” of immune responses in anorexia nervosa. It makes sense that cytokines would be low given the absence of adipose tissue in acute anorexia nervosa…

  • But what might this “uncoupling” signal?
  • Have you seen it in other conditions?

Could your findings have any implications for treatment in the future?

Dr. Pisetsky replies: I am glad there was so much interest in our work, I will address the questions one at a time. The study of alarmins like HMGB1 in a psychiatric disease is a new undertaking and there are few precedents for “uncoupling” in any condition.  In general, the production of cytokines and alarmins are linked given the stimuli (for example, infection) that drives their expression.  Thus, bacterial endotoxin induces both HMGB1 and pro-inflammatory cytokine upregulation.  Typically, we would expect alarmins and cytokines to track together, but some events like psychological stress can increase cytokines, but we don’t know how they influence alarmins.  Again, most of the time, cytokines would induce the expression of alarmins and coupling would occur.

The question then arises about the other direction.  Can an elevation of HMGB1 occur without a comparable increase in cytokines?  We think that the answer can be yes if a major source of cytokine production (i.e., adipose tissue) is lost as in anorexia nervosa.  Thus, even though a stimulus is present (for example, HMGB1 from a damaged tissue), cytokine production may not occur because the responding cells are not there in sufficient number.

David 2[3]

David Pisetsky, MD, PhD Professor of Medicine and Professor of Rheumatology, Duke University

Uncoupling may not be unique to anorexia nervosa, but this topic has not been well studied.  We think that it may happen in rheumatoid arthritis patients who have been effectively treated since the therapies such as steroids or anti-cytokine agents may take away the responding population of immune cells. Those are preliminary findings and relate to observed discrepancies between levels of a marker like C-reactive protein (CRP) and clinical findings and in this case, the uncoupling occurs in response to the effects of a drug treatment.

If our observations of uncoupling hold in a larger sample, there are definitely implications for the treatment of anorexia. First, it would provide a new target for treatment (i.e., HMGB1) as well as biomarker of illness (i.e., HMGB1 levels or relative HMGB1 and cytokine levels).  A second point is whether cytokine levels increase as individuals with anorexia are renourished and gain adipose tissue.  Increasing adipose tissue will increase their capacity to produce cytokines, and increased cytokines can contribute to a worsening of both their physical state (i.e., they will feel more ill) and their psychological state because of the effects of cytokines on mood and affect.  In that case, treatment with anti-cytokines may be considered to make the recovery process less uncomfortable.  Both anti-cytokines and anti-HMGB1 agents do have neuropsychological effects.  The biomarkers (i.e., relative levels of cytokines and HMGB1) would hopefully guide use of these strategies in different stages of illness.  Given the array of immunological agents that are available, there are potentially many approaches that can be considered. Uncoupling may provide a clue to the best approach.