Thursday, December 10, 2015

New Evidence Strengthens the Link Between Hypothalamic Injury, Obesity, and Insulin Resistance

Obesity involves changes in the function of brain regions that regulate body fatness and blood glucose, particularly a region called the hypothalamus.  My colleagues and I previously showed that obesity is associated with inflammation and injury of the hypothalamus in rodent models, and we also presented preliminary evidence that the same might be true in humans.  In our latest paper, we confirm this association, and show that hypothalamic injury is also associated with a marker of insulin resistance, independently of BMI.


A key reason why it's so hard to lose fat is that the brain defends against fat loss by ramping up hunger and the seductiveness of food, and shutting down calorie expenditure (1).  Essentially, the brain of a person with obesity "wants" him to to be obese, and if he tries to lose fat, it mounts a starvation response designed to undermine the effort.  This is primarily coordinated in a brain region called the hypothalamus.

Why does this happen?  If we can answer this question, we might be able to understand how to prevent and treat obesity more effectively.  Naturally, a number of researchers are working on the problem.

Our research team (led by Josh Thaler and Mike Schwartz) previously showed that diet-induced obesity in rodents causes changes in the hypothalamus that suggest inflammation and injury (2).  We saw increases in the expression of inflammation-related genes, and changes in the size, shape, and number of specific brain cells called microglia and astrocytes.  These cells protect the brain against threats by activating themselves through a process called gliosis, and the changes we observed in fat rodents suggested that they were doing exactly that.

Yet ultimately we're interested in human obesity, not rat obesity.  The problem is that it's a lot harder to study human brains than rat brains, because humans don't usually want to give them up.  So we developed a method to look for gliosis in the brains of living people.  This relies on a technique called magnetic resonance imaging (MRI), which is sort of like a fancy X-ray that's better at examining soft tissues like the brain.

Using MRI, my colleagues Ellen Schur and Ken Maravilla looked for signs of the same cellular changes in the hypothalamus that we had observed in rodents (increased T2 relaxation time, to be precise).  And we found them.  The higher a person's body mass index (BMI), the more we tended to see MRI evidence of gliosis in their hypothalamus.  This suggested that people with obesity might also have hypothalamic inflammation and injury that makes it harder for them to lose weight-- and may have also contributed to their obesity in the first place.

Yet this experiment was preliminary, because the MRI data weren't specifically collected to look for hypothalamic injury.  Also, we were operating under the assumption that the MRI signal we were detecting was related to actual gliosis-- which we hadn't directly tested yet.  In our new study in the journal Obesity, we sought to address these concerns, and look for a possible role of gliosis in insulin resistance as well.

The study

In our latest study led by Ellen Schur and Ken Maravilla, we recruited 70 male and female volunteers, 18-50 years old, of all body weights (3).  We then put them in the MRI scanner and looked for evidence of gliosis in the hypothalamus.  We divided the group into thirds based on the degree of suspected gliosis, and compared the top third to the bottom third to see if they differed in other respects, like body weight and insulin levels.

We also took blood samples to measure blood glucose and insulin, and calculate an estimate of insulin resistance called HOMA-IR.

In a second experiment, we analyzed post-mortem human brain tissue both by MRI and by tissue staining, and compared the MRI gliosis signal to a direct microscopic measure of gliosis.  This was to make sure that our MRI signal was actually a good measure of gliosis.

The results

In the overall cohort, we were able to confirm that people with higher BMI also tend to show MRI evidence of gliosis in the hypthalamus.  This replicates our preliminary finding from the previous paper.

When comparing the top third of our cohort to the bottom third, we saw striking differences in two areas.  First, people in the top third of gliosis were much more likely to be obese (64 vs. 39 percent).  Second, their insulin levels were nearly twice as high as people in the bottom third of gliosis.  Similarly, HOMA-IR, an estimate of insulin resistance, was almost twice as high.

Surprisingly, higher insulin and HOMA-IR levels in the third with more gliosis was partially independent of their body weight.  In other words, the association between gliosis and insulin resistance couldn't be fully explained by the fact that they carried more fat.

In the second experiment, we found that there was a strong correlation between MRI evidence of gliosis and direct evidence of gliosis, as seen in stained human brain sections.


Our new paper strengthens the evidence that MRI can be used to detect hypothalamic gliosis in humans, and that it is associated with obesity.  It also confirms that our MRI signal actually measures gliosis, and not something else.  And lastly, it suggests that hypothalamic gliosis is also associated with insulin resistance, regardless of whether or not a person is obese.

These findings have several interesting implications.  First of all, they give researchers a powerful new tool for studying the role of the brain in obesity.  We believe that hypothalamic inflammation and injury play a role in obesity, and now we can measure it in living humans.  This means we might get a real-time glimpse at a process that may be a fundamental driver of fat gain-- and see what factors, dietary or otherwise, promote or suppress it.  Previously, I co-led a study showing that hypothalamic gliosis and obesity are reversible in mice when we put them back on a strict whole-food, low-fat diet (4).  Could the same be true in humans?  Or could we achieve the same outcome with a different diet?

That said, it's too early to know exactly how useful this tool will be for research, and it's much too early to know whether it will be useful to individuals.  There is a lot of variability in the data, and the associations aren't tight enough that we can reliably categorize someone as lean or obese based on gliosis alone (r = 0.31).

Another interesting implication is that hypothalamic gliosis could be related to blood glucose regulation in addition to body fat regulation.  The hypothalamus plays a key role in regulating blood sugar, and the neurons that do so are intertwined with those that regulate body fatness.  So gliosis could be directly relevant to diabetes as well-- even among people who aren't obese.


raphi said...

MRI => (hypothalamic) gliosis => obesity = r of 0.31
"hypothalamic gliosis is also associated with insulin resistance, regardless of whether or not a person is obese".

It is my understanding that you would like to see reduced hypothalamic inflammation (gliosis) become a targeted obesity treatment, despite hypothalamic gliosis being less predictive of specific conditions (IR &/or obesity) than a generalized inflammatory state. How should the weak correlation & the targeted treatment be reconciled?

I think we can already partially answer your question "hypothalamic gliosis and obesity are reversible in mice when we put them back on a strict whole-food, low-fat diet (4). Could the same be true in humans?" Dr.Kevin-Lee illustrates how the answer is likely 'no' (

For inflammation to interfere with 'body fat set-point' should we not already be able to point to the what this set-point consists of? It seems to me that it has been accepted as a Platonic concept rather than pinned down as well-defined biological structure.

Nick Mackowski said...

Would you be able to clarify and elaborate on hypothalamus inflammation and injury? I really enjoy your work, thank you!

RLL said...

Stephan - This looks like a big piece of the many pieces needed before diabetes can really be understood.

My definition of 'understood': the newly diagnosed diabetic goes to an endo, who then does a series of tests, and then Rx's diet, meds, exercise that results in truly normal glucose metabolism. Diabetics now can have unexpected highs and lows, no doctor or endo is able or willing to explain why. It is still a matter of Saints Tinker and Tweak.

Stephan Guyenet said...

Hi Raphi,

You wrote: "It is my understanding that you would like to see reduced hypothalamic inflammation (gliosis) become a targeted obesity treatment". Not quite. What I would like is for the hypothesis to be tested so we can see if it has predictive or therapeutic value. If we can reduce hypo inflammation effectively, and it's demonstrated to have therapeutic value for obesity, then yes I would like to see it used therapeutically. But we're not there yet.

I acknowledge that there is a lot of variability in the association between hypo inflammation/injury and obesity. A lot of that is probably because the MRI method of measuring it is crude. MRI is a very blunt tool for this task, but it's the best we can do in living people. Tissue staining is much more accurate but it requires a formaldehyde-preserved brain. It's also possible that hypo inflammation/injury doesn't play a causal role in obesity, rather it's simply an epiphenomenon. There is pretty good evidence that hypo inflammation plays a causal role in mouse diet-induced obesity, but we'll need more work to evaluate that hypothesis in humans.

The links you provided were to weight loss studies comparing a moderate LC to moderate LF. I'm not sure how well they address the hypothesis. The mouse diet we used was a 14% fat whole food diet with 100% adherence. I'm certainly willing to accept the possibility that humans might not respond the same way as mice to that diet, but that hypothesis has not been tested yet. It is worth noting that the high-starch vegans (e.g. McDougall, Barnard) tend to be extremely lean even in their 60s and beyond. That degree of leanness is something you rarely see among aging low-carb figures. The high-starch vegans eat diets that are much lower in fat and higher in fiber than what is used in LC vs. LF RCTs. Nothing to hang your hat on for sure, but it is intriguing given the animal evidence.

Regarding the adiposity setpoint, it is a concept, but we also know a great deal about the biological systems that underlie it. We know from 60+ years of experimentation in animals and humans that adiposity responds to short-term perturbations "as if" it were regulated by a negative feedback control system (like your thermostat). We have also identified a negative feedback control system between adipose tissue and the brain, involving leptin. We don't know all the details yet, but it's clear that adipose tissue, leptin, and the hypothalamus (and other brain regions, to a lesser degree) form a negative feedback adiposity control system that plays a key role in determining the "setpoint" or "settling point" or whatever you want to call it.

Hi Nick,

You'll have to be more specific.


Yes, I think understanding the brain mechanisms will be a major "missing link" for diabetes. My recent mentor Mike Schwartz is working hard on this, and making good headway. It's looking like the brain plays an even greater role in blood glucose control than anyone realized, although much work needs to happen before we know how the findings apply to humans, and whether or not they can be targeted therapeutically.

If you're waiting for doctors to be able to quickly diagnose and fix T2 diabetes using personalized strategies though, I think you'll be waiting a long time. The human body is a complicated place and diabetes is usually the result of many years of accumulated damage. There's still a lot of guesswork and tinkering in most areas of medicine, and I don't see that going away anytime soon. Also, people usually don't comply with lifestyle recommendations like regular exercise, leading many docs to become jaded and stop trying those approaches.

Stylooke said...

Hi Stephan, great article. Would you agree with me that rodents in general respond worse to dietary fat (in large amounts) than do humans? I wonder whether the moderate-high fat populations such as the kitavians who are also very lean suffer from gliosis. My best guess would be that hypercaloric diets tend to increase injury and hypocalorics (which are most easily adhered to by consuming a real whole food diet) will reduce injury.

Best, Nils

Stephan Guyenet said...

Hi Nils,

I used to think that rodents respond differently to dietary fat than humans, but looking deeper into the rodent and human lit, and thinking further about my own lab experience, has made me question that. Here's why. The "high-fat diets" that we use to promote obesity in rodents and other animals aren't just high in fat, they also have a high calorie density and other "junk food" characteristics. They're made entirely from refined ingredients and they're highly palatable to rats. They have a texture and flavor similar to a moderately sweet raw cookie dough.

I'm not aware of any studies that put rats on a whole-food-based, lower calorie density diet in which the calories came primarily from fat. My strong suspicion, based on studies that have manipulated the calorie density of high-fat rodent diets, is that they wouldn't become obese.

Conversely, when you put humans on a "junk food" diet that's high in fat, calorie-dense, highly palatable, and is composed of refined ingredients, they spontaneously overeat and rapidly gain fat just like rats on a "high-fat diet". So at this point I see no reason to believe our responses to macronutrients are fundamentally different.

Basically, in both humans and rodents, it seems to be less about carbs vs. fat, and more about calorie density, palatability, and how refined the food is (that may be less true at the macronutrient extremes, where other mechanisms may join the fray). Added fat is the easiest and most effective way to increase the calorie density and palatability of food-- but that doesn't necessarily mean that a high-fat diet has to be high in calorie density or palatability.

For me, this way of seeing things resolves the "paradox" that people who eat high-fat junk diets tend to gain fat, while people who go on low-carb high-fat diets tend to lose fat.

Teech said...

Aren't the rats usually fed a seed oil? Isn't that another factor that could come into play? There is some preliminary evidence that shows that soybean oil is obesogenic. I wonder if that's doing the damage.

raphi said...

Hi Stephan,

Thanks for your reply.

Barnard & McDougall are lean, yes, but I wouldn't go much further in holding them up as exemplars of health.

It's interesting that you do not think humans and mice respond very differently to macronutrient ratios. Or at least, only as a far second to the presence or absence of 'crapinabag' (as Peter Dobromylskyj calls them) diet features. By your definition it seems, these are: calorie-density (stemming from refinement) and hyper-palatability. Maybe also an imbalanced omega6-to-omega3 ratio too?

You seem to be rubbing elbows with Spreadbury's 'acellular carbs' hypothesis which postulates that the small intestine is only suited (long-term) to starchy/sugary foods with no more than 23% maximal density of non-fibrous carbs by mass.

I don't think that palatability part of the 'crapinabag' features. Whole-foods can be equally palatable (or hedonistically enjoyable) as junk foods without instigating long-term appetite dysregulation (leading to obesity). Palatability doesn't seem to 'add' any explanatory power to overeating. I think your hyper/hypothalamic gliosis theory is leaner & more plausible without it.


Stylooke said...

Hi Stephan,

Thank you for your elaborate response. The argument you set out to make seems solid. Whole foods diets are healthful compared to junk-food diets, regardless of macronutrients (though some optimum may exist on a personal level). It is no coincidence obesity in pets has risen dramatically as well, their foods are made as addictive as possible.

The main advantage of whole food carbs over whole food fats would then seem to be fermentable fiber, which is naturally much higher in carb-rich foods (in general anyway).

I have personally stopped adding copious amounts fat to my cooking (except for some coconut oil), it just didn't feel 'natural' anymore.

Best, Nils

Moldy Salami said...

Hi Stephen,
Bit off-topic but:
Have you changed your eating habits somewhat since 2011, where you mentionned on a Healthy Skeptic podcast a typical daily menu? Wondering if new evidence you came across in the 4 years after may have changed your habits.

Stephan Guyenet said...

Hi Bebidoo,

I honestly don't remember what I said at that time, but I doubt things have changed substantially. I probably eat less added fat now. I don't eat a lot of added fat anymore, with most coming from extra-virgin olive oil. Probably somewhat less fat overall too, as a result of adding less fat. I eat a lot of potatoes and a lot of beans/lentils. Especially since we harvested 800 lbs of taters from our garden this year. I probably eat less red meat too.

Clark Endrizzi said...

Hi Stephen,
I just stumbled upon your blog and love the perspective (academic) and insight you bring to the conversation.

I've been somewhat binge reading your articles for the last several days.

I do have some thoughts for this blog, and would love to see if they have merit to you. I think Bebidoo comments highlight something that I've found as well. When people are new to your stuff there is a lot of information to digest. As you are a blog driven by science I'm sure certain things have changed over in time in light of research and so forth.

That said, it would be really useful to have some permanent pages on your site that are effectively "round ups" that each cover the big topics that people like myself would really enjoy understanding your position on (and being updated as things change).

For excellent pages might be:
1. Your diet and exercise (lifestyle). No doubt people are going to be very interested in how you eat and excersize, which undoubtedly is based on your own research.
2. Position on popular diets out there (when I see diets I more mean way of life diets, not fad diets). I've seen a lot of references to paleo and others but a clear position would be insightful. For example I've read "Grain Brain" from Dr Pearlman, which largely promotes a paleo diet, and was a little uncomfortable with how strongly he seemed to suggest the evidence for his diet clearly being the fix for everything (headaches, weight, and every other malady). Very dr Oz'ish in it's hype. I love to read your stuff because it's so clearly driven by sound science and would provide context beyond the hype.
3. Something clear on the latest position of science on Obesity in general. Is there good sugars vs bad sugars. How much is obesity pre-programmed?

Anyway, I know it's a big ask. Maybe it's why you are writing a book. But you have so many great articles on here, it's just not always clear what your latest conclusions are (understanding that they change).

karl said...

Actually rats are herbivores and have evolved to eat a very low fat diet and don't normally eat any sugar so these so-called (fraudulent?) test diets are not going to be anything near normal - even the 'control diets' are not normal diets for rats. (they also have mitocondria that leak ROS a lot more than humans). Pigs make much better test animals for diets than rodents.

The ability to image such things in the brain with MRI is questionable at best - there are other better imaging systems to see diffuse damage. Rat brains are also quite small to begin with...

Normal people - during sleep - become insulin resistant - thus fat cells can release some fatty acids - chronically high insulin causes insulin resistance and what you really have is dis-regulation if insulin is always elevated. What we want is for the insulin levels to come back down to a level so that insulin resistance is back doing its regulation job and that lets us lose some weight.

Again - it is insulin resistance allows people to lose weight. The mantra that it is the insulin resistance that is causative is simply wrong - changes in insulin sensitivity are part of the normal control mechanism. Insulin is known to activate LPL in adipocytes - By contrast, insulin has been shown to decrease expression of muscle LPL.

Moving to the other side - We also know that HSL is inhibited by insulin.

Now it is known that dietary carbohydrates increase insulin - so normalizing insulin means one must eat less so there is no excess. The other end is that excess PUFA fats from plant oils and corn fed livestock - not something we have evolved to eat - messes with insulin sensitivity and causes inappropriate insulin sensitivity in adipocytes inducing weight retention/and gain.

There is some evidence that the viruses we get exposed to can induce inappropriate immune response that sometimes damages the brain - we can't ignore the effect of the autonomic immune systems on adipose tissue. There are people that are obese below the waist and way too skinny above the waist - most likely a break in some nerve connections.

Unknown said...

Hi! It seems to me that obesity is driven by a rather complicated feedback-loop: Behavior (manifested through diet/lifestyle) => inflammation (+ other stuff) => behavior => inflammation (+ other stuff) => … . If this is so, then obesity might be treated by disrupting this feedback-loop at one of several places; “restricting calories only” might not be the best place.

It would be interesting to know, however, whether exogenously induced hypothalamic inflammation causes behavioral changes in diet or lifestyle.


Jane said...

Hi Stephan
I tried to post this a few days ago and I think I hit the wrong button.

It seems to me that there is a simple explanation for the apparent palatability of the refined high fat diet. High fat diets can cause oxidative stress, and the satiety system is sensitive to oxidative stress because it uses ROS for signalling. POMC neurons signal satiety by raising ROS, and AgRP/NPY neurons signal hunger by lowering ROS.

"Previous studies have proposed roles for hypothalamic reactive oxygen species (ROS) in the modulation of circuit activity of the melanocortin system. Here we show that suppression of ROS diminishes pro-opiomelanocortin (POMC) cell activation and promotes the activity of neuropeptide Y (NPY)- and agouti-related peptide (AgRP)-co-producing (NPY/AgRP) neurons and feeding, whereas ROS-activates POMC neurons and reduces feeding. .."

The authors showed that high fat feeding upregulated antioxidant systems which increased food intake by lowering ROS too low to signal satiety, and low enough to signal hunger.

So what looks like palatability could actually be oxidative-stress-mediated toxicity. Of course we know that high fat diets damage the hypothalamus, that's exactly what your post is about. Perhaps a whole food high fat diet would not do it, as you suggest.

Richard B said...

Does human diet supplementation with sodium butyrate reduce insulin resistance and inflammation?

Ellen said...

"It is worth noting that the high-starch vegans (e.g. McDougall, Barnard) tend to be extremely lean even in their 60s and beyond. That degree of leanness is something you rarely see among aging low-carb figures. "

Hmm..I'm not so sure about this. Anecdotally - Taubes seems rather lean and sometimes a bit plump. Just like Mcdougall - who in some vids looks plump and in others very lean.

The "obese" low carb guys were usually morbidly obese or obese. I don't think anyone who has that level of obesity in their life can really reverse it and get to real levels of leanness.

I think the bit of yo-yo dieting that the obese struggle with applies across the board. Joe Cross (who is juicing/vegan) seems to bubble up and down with his weight too.

julianne said...

Aaron Blaidsell at UCLA has put rats on junk food diets and whole food diets - ad libertum. The refined food diet was low fat.

Food quality and motivation: a refined low-fat diet induces obesity and impairs performance on a progressive ratio schedule of instrumental lever pressing in rats.