Journalists like to report on novel and exciting findings regardless of how likely they are to be replicated or how well they fit with everything else known about the topic. It is an all too common occurrence that a small pilot study which has favourable results, creates a buzz and gets into our heads, only to produce negative results once the sample size is increased. But the latter, negative finding, rarely makes it to the printing press. So, we are left thinking the results of the pilot study are correct, when they might be an outlier or a false positive.
Given this, I wanted to summarize an article that provides a critical overview of the current neuroimaging studies in anorexia nervosa. We can use this as an introduction to neuroimaging in AN: to get a sense of the scientific consensus, the trends that emerge and what can be learned from the data.
When reading studies on eating disorders, please keep in mind:
The ill state of the EDs, particularly of AN, can be accompanied by severe metabolic, electrolyte, and endocrine disturbances. As such, research in ill EDs is always potentially confounded by these effects. Even in recovered participants, however, it has to be taken into consideration that observed differences may represent either premorbid traits contributing to the development of the illness, or a consequence of having been previously ill. In addition, there is a restricting type (AN-R) and a binge-purging type (AN-B/P) of AN and ﬁndings in one sub-type may not be representative for all AN individuals – Frank & Kaye (2012)
Furthermore, the neuroimaging field itself suffers from a lot of problems (I highly recommend this fantastic discourse on the matter here: “The dilemma of weak neuroimaging papers“. It is so good, I’ve re-read it). An accomplished and well-known MIT scientist, Nancy Kanwisher, wrote in a comment: “I have occasionally asked respected colleagues what percent of published neuroimaging findings they think would replicate, and the answer is generally very depressing. My own guess is *way* less than 50%.”
So, just keep those points in mind as you continue reading this post.
I chose a review by Pietrini and colleagues, from 2011, in part because I think the way they chose to summarize the data is really nice. Pietrini et al reviewed neuroimaging studies published from 1950-2009, focusing predominantly on those published within the last 15 years and restricting themselves to studies using SPECT, PET and fMRI techniques.
How to compare heterogeneous studies?
Pietrini et al compared sub-types of AN when possible, and when the data was unavailable, listed the findings in a “AN-NS” (not specified) category (I’m not completely sure why this decision was made and emphasized, given the high rates of diagnostic crossover). To compare many different studies, for each study they assigned all the areas that were investigated (called “regions/volumes of interest”) to one of eight neuroanatomical areas (frontal cortex, temporal cortex, cingulate cortex, parietal cortex, occipital cortex, insular cortex, hypothalamus/thalamus and striatum).
If you get confused by words like “dorsal”, “ventral”, “caudal”, “lateral”, “medial”, etc.. check out this page on directional terms.
For a great quick (and not very detailed) functional neuroanatomy reference click here, or here for more on the insular cortex, cingulate cortex, hypothalamus/thalamus. It may be useful to refer to it as you go through this post. I tried to find good links, but there are a few links to Wikipedia because I couldn’t (quickly) find better ones.
Summary of Main Findings:
Anorexia Nervosa – Restricting and Binge/Purge Subtypes
- both subtypes show similar profiles are resting conditions:
- reduced activation in frontal cortex; normalizes after recovery
- reduced activity in the cingulate cortex
- increased activity in parietal cortex to images of bodies
- three main regions that are consistently involved: frontal cortex, cingulate cortex and parietal cortex
Anorexia Nervosa – With Diagnostic Subtype Not Specified and Binge/Purge Subtypes (AN-NS)
Peculiarly, data that didn’t specify the diagnostic subtypes has a slightly different profile from either the AN-R and AN-BP neural profiles. The authors write: “It is of note that part of the common pattern of altered cerebral function in AN-R and AN-BP disappears when data derived from those studies in which the AN subtype was not assessed are compared.”
- atresting conditions:
- reduced bilateral activity in temporal cortex
- altered metabolism in parietal cortex (mostly left-side)
- several studies show increased activity in striatum (but not all)
- increased activity in left temporal cortex during exposure to food or images of food
- reduced activity in right parietal cortex to various stimuli (images of bodies, Stroop test)
- regions that seem to be consistently involved: temporal cortex (particularly the fusiform gyrus and amygdala), parietal cortex, and striatum (mostly caudate nucleus).
For All Groups (AN-R, AN-BP, AN-NS):
- normal activity in all regions for all sub-types after recovery with one study showing a reduced in the occipital cortex for AN-R
- occipital cortex, insular cortex, hypothalamus/thalamus don’t seem to differ (at least not consistently) from controls.
If one were to look through the figures summarizing all the findings from various studies, one would see that while there are some definite trends, there are also many contradictory findings.
This is to be expected: (1) there are many different neuroimaging techniques, (2) the authors grouped findings into 8 large brain regions (there may be significant and important differences within one of those areas), (3) most of the studies have small sample sizes – increasing the chances of false positives or false negatives [detecting something that isn’t there or not detecting a real effect, if it is small, for example], (4) different patient groups in terms of BMI, gender, age, phase of disorder, (5) different statistical methods, study designs, parameters, and more…
I’m pointing this out primarily because I think one should always be critical of media reports about a particular study showing “X” region in the brain of eating disorder patients is “hyper/hypo-active compared to a control group” and thus it means “insert complex behaviour or through process here”.
It is nearly impossible for most people without sufficient knowledge in the field or access to scholarly literature to evaluate how the results fit in with previous findings, how likely they are to replicated, how big is the effect is, and are there alternate interpretations to those proposed by the authors (usually YES!) and so on…
The authors note that while they decided to subdivide the data into AN-R and AN-BP subtypes, they admit there is still debate as to how independent and different these subtypes really are (given the high rates of crossover).
In summary, it seems that the frontal, parietal and cingulate cortices are most consistently involved in AN; but all observed alterations do seem to normalize after recovery. So, it is still an open question whether they are the result of starvation and malnutrition or causative traits (or a bit of both).
Within those large regions, the specific areas that show altered neuronal activity are: the dorsal lateral prefrontal cortex (DLPFC), the inferior parietal lobule (IPL) and the anterior cingulate cortex (ACC).
What do we know about these brain regions (brief descriptions from Pietrini et al.)?
Dorsal Lateral Prefrontal Cortex:
- motor planning, planning the sequence of events
- extensively connected to other brain regions
- emotional impulse regulation (also limbic structures are important for this)
Inferior Parietal Lobule:
- visuo-spatial processing
Anterior Cingulate Cortex:
- dorsal ACC: cognitive processing of stimuli, connected to DLPFC and IPL
- ventral ACC: connected to insula, amygdala and striatum; “involved in identifying emotional significance of stimuli and in assessing their motivational information to produce an adequate affective response”
Given that the core symptoms in AN include: inadequate self-perception of body weight, shape and dimensions of one’s own body, and the DLPFC and IPL are important in the “representation of one’s own ‘‘body schema’’ (the mental map each one has of his own body)”, the findings of altered activity in these regions to images of bodies suggests that these brain regions are working differently in AN patients, compared to controls.
What about the reduced activity in the ACC? This finding is consistent, as Pietrini et al write, “with the hypothesis of a reduced activity.. of the emotion/anxiety processing network that links sensory and hedonic experiences to a proper motivational value of reward.”
In their conclusion, Pietrini et al provide a nice summary of what these brain regions are involved in:
the body schema network, the frontal visual system, the attention network, the arousal and emotional processing systems and the reward processing network.
Pietrini, F., Castellini, G., Ricca, V., Polito, C., Pupi, A., & Faravelli, C. (2011). Functional neuroimaging in anorexia nervosa: A clinical approach European Psychiatry, 26 (3), 176-182 DOI: 10.1016/j.eurpsy.2010.07.011