An interesting new paper from a British/Danish collaboration uses a clever trick based on genetics to untangle the messy correlation between obesity and mental health.They had a huge (53,221) sample of people from Copenhagen, Denmark. It measured people's height and weight to calculate their BMI, and asked them some simple questions about their mood, such as "Do you often feel nervous or stressed?"
Many previous studies have found that being overweight is correlated with poor mental health, or at least with unhappiness ("psychological distress"). And this was exactly what the authors found in this study, as well.
Being very underweight was also correlated with distress; perhaps these were people with eating disorders or serious medical illnesses. But if you set those small number of people aside, there was a nice linear correlation between BMI and unhappiness. When they controlled for various other variables like income, age, and smoking, the effect of BMI became smaller but it was still significant.
But that's just a correlation, and as we all know, "correlation doesn't imply causation". Actually, it does; something must be causing the correlation, it didn't just magically appear out of nowhere. The point is that shouldn't make simplistic assumptions about what the causal direction is.
It would be easy to make these assumptions. Maybe being miserable makes you fat, due to comfort eating. Or maybe being fat makes you miserable, because overweight is considered bad in our society. Or both. Or neither. We don't know.
Finding this kind of correlation and then speculating about it is where a lot of papers finish, but for these authors, it was just the start. They genotyped everyone for two different genetic variants known, from lots of earlier work, to consistently affect body weight (FTO rs9939609 and MC4R rs17782313).
They confirmed that they were indeed associated with BMI; no surprise there. But here's the surprising bit: the "fat" variants of each gene were associated with less psychological distress. The effects were very modest, but then again, their effects on weight are small too (see the graph above; the effects are in terms of z scores and anything below 0.3 is considered "small".)
The picture was very similar for the other gene.This allows us to narrow down the possibilities about causation. Being depressed clearly can't change your genotype. Nothing short of falling into a nuclear reactor can change your genotype. It also seems unlikely that genotype was correlated with something else which protects against depression. That's not impossible; it's the problem of population stratification, and it's a serious issue with multi-ethnic samples, but this paper only included white Danish people.
So the author's conclusion is that being slightly heavier causes you to be slightly happier, even though overall, weight is strongly correlated with being less happy. This seems paradoxical, but that's what the data show.
That conclusion would fall apart, though, if these genes directly effect mood, and also, separately, make you fatter. The authors argue that this is unlikely, but I wonder. Both FTO and MC4R are active in the brain: they influence weight by making you eat more. If they can affect appetite, they might also affect mood. A quick PubMed search only turns up a couple of rather speculative papers about MC4R and its possible links to mood, so there's no direct evidence for this, but we can't rule it out.
But this paper is still an innovative and interesting attempt to use genetics to help get beneath the surface of complex correlations. It doesn't explain the observed correlation between BMI and unhappiness - it actually makes it more mysterious. But that's a whole lot better than just speculating about it.
Lawlor DA, Harbord RM, Tybjaerg-Hansen A, Palmer TM, Zacho J, Benn M, Timpson NJ, Smith GD, & Nordestgaard BG (2011). Using genetic loci to understand the relationship between adiposity and psychological distress: a Mendelian Randomization study in the Copenhagen General Population Study of 53,221 adults. Journal of internal medicine PMID: 21210875
12 comments:
This is hilarious! So glad I was browsing Research Blogging tonight.
Also..could there be a causation between stress and over eating?
I don't think it's merely genetics that would explain "conditions" like this, by the way. there's also epigenetics. They're starting find strong links between environmental triggers and the natural silencing of certain genes.
Check out the tubby mice in this article:
Obesity, Epigenetics and Gene Regulation
Should be noted that BMI is a rather inaccurate way (i'm eufemistic here) to express Fatness.
A heavyweight trained to the max boxer can have the same BMI as a couch potato.
Even heart specialists start question it's merits:
The relationship between body mass index, treatment, and mortality in patients with established coronary artery disease
http://eurheartj.oxfordjournals.org/content/30/21/2584.full
I don't understand what this paper is supposed to show (I don't seem to have access to the full text).
So BMI was associated with 'psychological distress' but two genes which were associated with BMI were also associated with less distress. I'm not sure how this is supposed to show that BMI was inversely correlated with distress.
If the genetic variants had been associated with BMI and associated with distress then I could see how by stratifying by genotype you could then show a within genotype negative relationship between BMI and stress but that's not what we're saying.
As I understand it 'instrumental variable analysis' is used in stuff like economics and is based on the assumption that the instrumental variable affects the predictor but not the outcome variable i.e. it depends entirely on the assumption that these genes affect distress only via BMI, that is they can have no effect on distress other than BMI and therefore they allow us to better interrogate the relationship between the predictor and the outcome, hopefully missing out on some confounding effects.
As you state, we've no reason to think that these genes only have an effect on BMI (they'd be the first genes to do so) and this undermines the whole rationale of instrumental variable analysis.
To accept this study at face value you believe the whole BMI-distress relationship is a massive artefact (mechanism unknown) and that there is a secret hidden small reverse relationship and simultaneously you believe that there are two genes that have a single effect on BMI through some mechanism that affects other aspects of physiology sufficiently minimally to have no effect on distress by any other mechanism.
As I understand it you've understood the paper; I guess it depends how plausible you find the idea that having a higher BMI, all other things being equal, makes you a bit happier.
I don't find that especially implausible, although I think I find the idea that these genes have a more direct rather more plausible - if they affect appetite, they might well affect dopamine and reward, and we know that that's a mood-y pathway.
If they showed the same effect with entirely non-brain-related fat genes - something to do with peripheral metabolism or absorption of fats - I'd be more convinced.
But I think it's an interesting methodology.
I'm ultra-unconvinced. The usual idea behind the instrumental variable method is to rule out reverse-causation. In this case the correct conclusion is that you cannot rule it out (assuming you believe the genes can't affect distress other than via BMI). So it may be distress causes high BMI rather than vice versa. To conclude that actually BMI causes less distress, despite the population correlation, is simply perverse.
All i see a paper using a incorrect yardstick (BMI) and draw conclusions from using it.
Since BMI doesn't have much relation to actual fat/bodymass , the conclusions are irrelevant and only have amusement value.
BMI Not Accurate Indicator Of Body Fat
http://www.medicalnewstoday.com/articles/64577.php
"A research team from Michigan State University and Saginaw Valley State University measured the BMI of more than 400 college students - some of whom were athletes and some not - and found that in most cases the student's BMI did not accurately reflect his or her percentage of body fat."
100's of studies like that.
Petrossa: They also used Waist-to-Hip ratio and found the same thing, I didn't mention that in my post, to save space.
BMI is certainly not perfect but in the population as a whole it's not a bad measure. Most people aren't athletes. The ones who are, add noise to the measure, but they don't add enough noise to make it useless.
If someone has a BMI of 35 it's very unlikely that they're not obese. they might be the world's strongest 5 foot weightlifter but, probably not.
"Nothing short of falling into a nuclear reactor can change your genotype."
Genotypes are robust, but not that robust. Infection with a retrovirus, perhaps transmitted via a mosquito bite, is quite adequate to do the job, and of course, epigenetic differences can change genotype expression . . . a pertinent point given that inherited epigenetic factors impact obesity.
Kind of a no brainer asking fat people if they feel stressed. But I suspect the causal link (if there is one) points the other way. Eating calms you down for a brief period, so anxious people often eat a lot, and eat carbs because they work better, but which also cause weight gain. So anxious people are more likely than average to be fat. So fat people are more likely to be anxious. Duh.
Of course if you did this with really thin people it would work as well - though because I'm in the first camp I have no insight into why anxious people don't eat. I have to actually be vomiting to not eat.
@Petrossa BMI is a good indicator around the averages - your extreme examples show it's limitations, but all measures have limitations. I also combine it with other measures waist/hip ratio, and a rough body fat measure.
J
I've studied the matter intensively. BMI is not a reliable indicator of fat/body mass relation by any means. Not to spam, but here are my thoughts on the matter: (many scientific sources quoted)
http://petrossa.wordpress.com/2010/04/16/the-disease-fat-does-not-exist/
Remember the first Subway commercials starring Jared? I always said the fat Jared looked much happier.
Have not read the paper, but am curious: are the data consistent with a model where people who have higher BMI "because of" the "fat gene variant" do not show the BMI-happiness negative correlation ? However, if BMI is higher for other reasons, then it does correlate negatively with happiness.. Maybe some of the commenters were saying this too ?
Post a Comment