Does Hyperinsulinemia Drive Diet-Induced Obesity?

Unless you've been living under the internet equivalent of a rock in this community, no doubt you've heard there was a paper published in Cell and Metabolism entitled Hyperinsulinemia Drives Diet-Induced Obesity Independently of Brain Insulin Production.  It's a TWICHOOB's dream title ... until you actually read the paper, or even the abstract and freely available charts and diagrams.  Too bad most did not.  My link is to the full text I'm sharing with my readers through Google docs.   But you actually don't need to read this whole paper get the basic gist of what the study entailed and the results.  However, before we even do that, consider this.  How do you induce obesity in a C57Bl6/J mouse?  You feed it a high fat diet, HFD.    And as the graphic clearly shows, the researchers believe the results of their study warrant a change in the chain of
causality/progression in the development of diet-induce obesity, DIO.  As studies in support of TWICHOO go, this study is about the worst you can come by!  Because on their more natural rather low fat high carb diet, these mice do not fatten.  That is, a low fat, high carb, insulin spiking chow does not induce the hyperinsulinemia that is implicated in the "new model".  It is a high fat diet that does.  At least in C57Bl6/J mice.  

This didn't stop the TWICHOOB brigade from jumping on this study as proof of their theories.  Odd how one sought to mock the researchers, I guess because it's not enough for them to finally "see the light", we must denigrate them anyway.  But this post is not about the response of the low carb community.  Suffice it to say, however, if you are not interested in what this study says outside of the prism of LC-dogma, there is no reason to read on.  This study ain't the scrolls or tablets you've been waiting for!

So what did this study show?  I weighed in ever so briefly on this here:  Remember LIRKO? Grey & Kipnis? which linked to Stephan Guyenet's first response on his blog:  Is it Time to Re-write the Textbooks on Insulin and Obesity?  I must say that I'm so very happy to see lead researcher James D. Johnson venture into the blogosphere to correct some of the misinterpretations and/or criticisms of his interpretations of the study, and I'm honored he chose my blog as one place where he's shared his thoughts (in the comments on my "Remember" post linked above.)  Jim has also taken to Twitter where we've exchanged some thoughts there as well.  I favorited a bunch of tweets hoping I could share a link to those, but no dice (then again I'm a social media spaz so perhaps there's a way).  In the meantime, anyone interested in reading his tweets with me and others, here's his Twitter account ... go while they're still "fresh"!  

Before addressing the study, I want to discuss another culprit in interpretations of the results:  the media.  But "the media" begins with press releases such as this one, from UBC stating:

A serendipitous discovery by a researcher at the University of British Columbia could overturn widely accepted notions about healthy eating habits.

How?  In tweets and comments, Johnson has been adamant to remind folks -- especially those blaming dietary carbs --that they did not test different diets here.  They used a standard HFD known to induce obesity.  So I think to the extent that there is some overzealous interpretation of this study, it begins with this press release!  There's absolutely nothing about this study that tells you, me, or anyone else anything about healthy eating habits.

James Johnson, an associate professor of cellular and physiological sciences, found that in animal models, too much insulin can be harmful.

I challenge that such a conclusion can be drawn from his study, and this is certainly why I harp on the LIRKO mouse -- a mouse that is sick, for sure, but not sick-fat with sick fat, while being profoundly hyperinsulinemic with no ill effects directly attributable to insulin and its action.  I'll cite more from the bulk of the release later, but it is interesting to note that:

Johnson, a member of UBC’s Life Sciences Institute, was not intending to study obesity. He was initially exploring whether beta cells in the pancreas, which produce insulin, were stimulated to multiply by their own insulin secretion. His discovery that low-insulin mice couldn’t gain weight was unexpected – as was the finding that most of those mice, despite dramatically lower insulin levels, still didn’t develop diabetes.

While I don't doubt that Jim has familiarized himself with a lot of the relevant research in this area, I tend to think that were he knee deep in obesity research per se, his findings would not have been all that surprising after all.  In my opinion, the conclusions drawn in the study are over-reaching and not supported, as worded,  by the published data.

OK ... on to the study itself.

In looking into this study, I learned something new about mice vs. humans.  We have one insulin gene, INS, while mice have two, Ins1 and Ins2.  The Ins2 is the homologue to human INS and is expressed in all cells, while the Ins1 gene is pancreas specific.  Here is one paper on the two-gene system in mice (and rats) for anyone interested.  I think this is important to keep in mind, as all genetically altered mice are, in terms of whether or not such a mutation is possible or observed in humans.  For example, the ob/ob mouse has been useful in elucidating many things regarding the role of leptin in obesity.  However, we all know the "end" to that story vis a vis humans ... this sort of leptin deficiency is far too rare in humans to explain even the "baseline" levels of obesity in the couple-three percent range.  

click on picture to enlarge further

So what they did here was create an insulin deficient -- yet sufficient -- mouse that was incapable of insulin hypersecretion.  Indeed in their graphic, they show a half-pancreas for their Ins1 knockout.  At right they show the Ins1 gene expression for each diet (12wks) and the islet insulin content (8wks).  To me it is interesting to note that Ins1 +/- on the normal diet (blue bars) had only slightly lower insulin content than controls (dark vs. light), and on the HFD (dark red) had almost identical insulin content to normal controls.  Thus, the finding that these mice were not diabetic is hardly surprising!  

I'm going to refer to these mice, left to right as CD+, CD-, HFD+ and HFD-.  It's much easier than writing this out over and over.

The major outcome here was that the HFD- mice did not produce excessive insulin and the beta cell mass (measured as cell area) did not increase.   They also did not become obese like their HFD+ cousins.  The fasting insulin is shown next to that data in my cropped images from Figure 2 above.  Curiously, both groups of HFD mice seem to be mildly hyperinsulinemic vs. controls at the 5 week (presumably once weaned onto chow) baseline.  Statistical significance is not noted, but the ** indicates difference between CD- and CD+ where levels for CD- are similar to HFD- and lower than HFD+ at baseline.  It appears that there is a "spike" in the HFD- mice but this falls off, presumably because they don't expand beta cell mass and cannot produce additional insulin.  Meanwhile the other finding in the study is that these mice remain lean and healthy and normoglycemic.  This is attributed to upregulation of UCP1 activity in white adipose tissue, WAT.  UCP1 is the thermogenic uncoupling protein expressed in brown fat that is more abundant and relevant to maintaining whole body temperature for small animals like mice.  It is apparently present in WAT as well.

Rather than storing the excess lipid intake, these mice burned off the fatty acids -- in their fat tissue!  This sounds somewhat similar to what occurs in the FIRKO -- lacking insulin receptors in fat tissue -- on a standard diet:  

At the whole body level, FIRKO mice demonstrated an increase in basal metabolic rate and respiratory exchange ratio. Analysis of gene expression in white adipose tissue (WAT) of FIRKO mice from 6 to 36 months of age revealed persistently high expression of the nuclear-encoded mitochondrial genes involved in glycolysis, tricarboxylic acid cycle, beta-oxidation and oxidative phosphorylation as compared to expression of the same genes in WAT from controls that showed a tendency to decline in expression with age .... these data suggest that maintenance of mitochondrial activity and metabolic rates in adipose tissue may be important contributors to the increased lifespan of the FIRKO mouse.

All of which points to elevated basal insulin levels and intact adipose insulin signaling as requirements for the development and maintenance of obesity.  Before I go into mouse models that refute this, let's harken back to the differences between mice and man, and even before that, let's remember something -- this is a genetically engineered mouse that is not found in any wild-type variant.  As extends to humans, the only examples of reduced insulin production I can think of are some MODY and LADA diabetics.  I doubt their insulin production is mimicked by this model, but this is the closest we get in humans to a relevant physiological state.  We do not see upregulation of metabolism in WAT to process metabolic excesses, and produce normoglycemic, healthy, lean humans.  One could argue that "compensators" -- humans for whom insulin levels are elevated and "seen" by sensitive cells -- are far healthier than these hypoinsulinemics.    These so-called metabolically normal obese people are significant in number.  This is the reason why, however imperfect a perfect measure, but the HOMA index is used to assess IR -- basically [fasting insulin] x [fasting glucose].  Compensators may have elevated fasting insulin, but are not considered IR if their glycemic control remains intact.

The definition of IR was what was behind my last twitter exchange with Jim, as they believe their mouse is unique in being able to isolate the role of hyperinsulinemia without IR complicating the picture.  The HFD- mice are mildly hyperglycemic early on (11 weeks, corresponding with when the slight hyperinsulinemia starts to subside), but goes away.  But here's where the unique observation of the upregulated uncoupling in WAT comes into play, and the role that insulin plays in this.  Do we see that in humans?  I'm struggling to find the relevance here if what would normally happen IS hyperinsulinemia and obesity, but it doesn't happen because of a deliberate genetic defect.  It is interesting that the mice compensate rather than experience elevated NEFA and all the ensuing metabolic havoc, but where's the relevance?  It's not at all unlike comparing what injected insulin does to endogenous insulin production.  If eating carbs doesn't cause a bolus of insulin in your thigh fat, it's not really relevant.  

So the problem as I see it with the wording and notion of insulin "driving" obesity.  If this simplistic model is a good one, you could not have obesity without hyperinsulinemia, and you would always see obesity with hyperinsulinemia.   We do not.  In this study, it was basically demonstrated that hyperinsulinemia was a necessary state for obesity to develop on a high fat diet.  What might be the mechanism?  As we all know insulin "fixes and traps" fatty acids in fat cells.  The results of this study would predict that insulin also suppresses UCP1 activity in WAT.  If this is the case, then the LIRKO mouse should be fat.  I would be interested in a non-inventive (as in making up "facts" easily shown to be false about this mouse) explanation for why LIRKO is not obese.  In this mouse, the lack of insulin receptors in the liver is what induces hyperinsulinemia and increased beta-cell mass instead of the diet.  Yet this mouse doesn't fatten.

I believe that the results here are still consistent (don't prove or support to the exclusion of other hypotheses) with the existing model.  A recent paper showed that just a few days of high fat feeding in mice caused IR of the brain and inappropriate lipolysis of WAT.  That's right -- too much fat let go.  The fatty acids stimulate basal insulin so as to tuck them back in but also overload peripheral mitochondria leading to IR further increasing the need for insulin.  I don't think there is even necessarily a progression occurring.  It's possible that normal mice become obese, IR and HI pretty much concurrently, because their physiology is ill-equipped to trap a high dietary fat load.  The mutant mouse that can't mount the appropriate insulin response to keep up has an "out" -- it can "waste" fatty acids through uncoupling.  Does this happen in humans?   Are the obesity resistant incapable of raising basal insulin?   When insulin production wanes wit the loss of beta cell mass, do obese T2's  suddenly become non-diabetic as their adipose  UCP1's rev up?

Which brings us back to misinterpretations of this study, which brings me back to the press release.  It is clear, both from reading the paper and discussions with the author, that the insulin levels being discussed are basal insulin, not postprandial.

Johnson concluded that extra insulin produced in the normal mice by the high-fat diet caused their obesity, which strongly suggests that mice – and, by extension, humans – may make more insulin than they need. The findings may mean that the key to maintaining a healthy weight is to continually return insulin levels to a healthy baseline by extending the gaps between meals and ignoring the widespread recommendations to consume small amounts throughout the day. In other words, cut out the snacks – and make sure not to overcompensate at mealtime.

Huh?  First ... are obese people known for eating small amounts throughout the day?  

This paragraph seems to have come from other research not included in this paper.  In comments the following study was brought up:  Time-Restricted Feeding without Reducing Caloric Intake Prevents Metabolic Diseases in Mice Fed a High-Fat Diet, from which the diagram at right comes  (click to view full size).    I'll save analysis of that one for when I get the full-text of the study (carbsane at gmail dot com) but here's what it boils down to.  Normal mice fed CD or HFD, each diet either ad libitum or time-restricted feeding.  Guess which group got fat?  HFDal, but not HFDtrf ... but also neither regular chow diet.  The HFDal has high fasting insulin and an exaggerated insulin response on glucose challenge.  The "protection" is attributed to increased beta-oxidation in the time-fed group.  But what's that nagging thought in my head?  It's the confusion/conflation of diet-induced postprandial insulin and basal insulin ... and the macronutrient that's causing the change.  

“As crucial as insulin is for storing blood sugar, it can also be too much of a good thing,” Johnson says. “If we can maintain insulin levels at a happy medium, we could reverse the epidemic of obesity that is a risk factor for so many ailments – diabetes, heart disease, and cancer.”

When one thinks diet and insulin, they think carbs -- even the non-low carbers.  So I think statements such as the two above in the press release are contributing to the confusion.   I guess what the timed-feeding study says is that if you are a mouse and you want to consume an obesogenic diet, you should practice intermittent fasting.  But if you want to eat a normal diet ... one that doesn't potentially lead to obesity and metabolic disruption, then meal timing really doesn't matter.  And if you're a human, you don't eat 24 hrs a day anyway.    However the study we're talking about didn't even look at this!  

So does this study demonstrate that insulin is bad?  Not really.  Because you cannot separate the fact that the HFD+ mice got obese and all the changes in adipokines, macrophage infiltration, and all the ensuing inflammation.  So was it the insulin or the obesity?  The HFD- mice didn't get fat, or "skinny fat" for that matter, so there was no NEFA to wreak havoc.  And we're left wondering what causes the hyperinsulinemia anyway.  How does the HFD cause hyperinsulinemia?

On Twitter Jim asked me what I think would stimulate insulin secretion (basal) and my response was NEFA.  He responded that fatty acids are weak stimulators.  Twitter is not the best for such discussions so let me say that I believe FA's are strong enough stimulators of insulin secretion to elevate basal secretion.  They are not on par with glucose, but that's not what we're talking about anyway.  However, when I countered with the question as to what about the HFD stimulated the insulin, Jim said that wasn't known.   My vote is again on NEFA.  It has recently been shown in mice that a few days of HFD leads to insulin resistance in the brain, inappropriate lipolysis of WAT and fatty acid release.  

What we do know, however, is that dietary carbohydrate is not the cause.  And this is actually a pretty damning study for TWICHOO whatever your interpretation of the data.  If carbs somehow caused hyperinsulinemia, then CD+ mice would be fat, and CD- mice would not be.  That's not what we see.    Just how many more studies such as this do we need before the TWICHOOBs finally get that?

No magic obesity pill, yet: While existing insulin-blocking drugs could prevent weight gain, they carry serious side effects that outweigh their benefit. Further research might lead to drugs that block excess insulin production or blunt its effect on certain targeted tissues.

Or we can remove the crap foods from our diet.

This post is not as polished as I would have liked.  But I decided to publish it up anyway so the discussion moves forward.  Jim is going to be on Superhuman Radio tonight (Tues 12/18) to discuss this.  I forget the exact time hopefully someone will stop by and drop a link.  Tune in!