Is it Time to Re-write the Textbooks on Insulin and Obesity?

A recent study in Cell Metabolism by Dr. Arya Mehran and colleagues found a result that, according to a press release, "could overturn widely accepted notions about healthy eating habits" (1), and has set the Internet abuzz.

In this study, researchers generated mice that lack one copy of the pancreatic insulin gene, and compared them to mice carrying both copies (2).  Then, they exposed both groups to a fattening diet, and found that mice lacking one copy of the insulin gene secreted less insulin than the comparison group (i.e., they did not develop the same degree of hyperinsulinemia).  These mice were also completely resistant to fat gain, while the comparison group became obese.  The authors came to some rather large conclusions based on these results, suggesting that the "accepted model" that hyperinsulinemia is the result of obesity is "incompatible with our results that put the insulin hypersecretion genetically upstream of obesity".  Ergo, diet causes hyperinsulinemia, which causes fat gain.  It's a familiar argument to those who frequent Internet diet-health circles, except in this case the hyperinsulinemia is caused by a high-fat diet.

The problem is that the "accepted model" they want to replace overnight didn't come out of thin air-- it emerged from a large body of research, which was almost completely ignored by the authors.  When carefully considered, this evidence suggests an alternative explanation for the results of Dr. Mehran and colleagues.

In a landmark paper published in the journal Nature in 1997, Dr. Teoman Uysal and colleagues studied a mouse line lacking the inflammatory gene TNF-alpha (3).  Due to a suppression of inflammatory signaling, these animals do not develop insulin resistance when placed on a fattening diet.  As a result, they do not develop hyperinsulinemia at all-- insulin levels remain the same as lean controls fed a normal diet.  This suggests that the hyperinsulinemia of obesity is indeed a compensatory response to insulin resistance.  Get rid of insulin resistance, and you get rid of hyperinsulinemia.  Meanwhile, in the comparison group fed the same fattening diet, fasting insulin increased by five-fold.  Yet despite these huge differences in fasting insulin, both groups of mice developed "marked obesity".  Here are the relevant figures:

The same thing was reported for the iNOS knockout mouse, which also does not develop insulin resistance for a similar reason (4).  In that case, the mice with normal insulin actually became fatter than the hyperinsulinemic comparison group, but both groups gained fat.  The finding that obesity does not depend on hyperinsulinemia has been replicated multiple times in other animal models that do not develop insulin resistance on fattening diets (5, 6).  It is unfortunate that Dr. Mehran and colleagues did not cite these papers or attempt to reconcile them with their own findings.

Roughly 20% of obese humans are insulin sensitive and have normal circulating insulin levels, an issue I discussed in my 2012 AHS talk (7).  Consistent with the evidence in animal models I just discussed, this demonstrates that hyperinsulinemia is not required for obesity in humans.  If the interpretation of Dr. Mehran and colleagues were correct, this should be impossible or at least very rare.  Observational studies in humans overall have found that elevated insulin levels do not predict future fat gain (8), offering further evidence against the idea that hyperinsulinemia is required for obesity.  Since there have been a lot of studies, they can be cherry picked, but if you consider them as a whole, the majority of studies that found an association actually reported that higher insulin predicts less fat gain over time (8).

Well, where do we go from here?  We can either stand on opposite sides of the line and shout at one another, or we can try to put the pieces together into a cohesive framework that explains ALL the evidence.  Luckily, such a framework exists and it's pretty simple.

The elevated insulin levels that accompany obesity are a compensatory response to insulin resistance-- this is clear from the papers I cited above, among many others.  Therefore, when insulin resistance develops, insulin secretion goes up in parallel, maintaining approximately the same relationship between insulin secretion and insulin sensitivity, so that relatively normal metabolic control is maintained (if this didn't happen, life-threatening metabolic havoc would rapidly ensue, e.g. diabetes or hypoglycemia).  This explains why suppressing insulin resistance and hyperinsulinemia in parallel has little or no impact on fat gain.  In that case, total insulin action on fat tissue (i.e., the normal relationship between insulin secretion and insulin sensitivity) is preserved.

However, when you uncouple insulin secretion from insulin sensitivity, you change the normal balance between the two, and you can affect fat mass.  That's why the fat-specific insulin receptor knockout mouse is resistant to obesity (9).  It's also one of the reasons why people lose fat when they develop diabetes, which is a (relative or absolute) deficiency of insulin action.  Correcting the insulin deficiency of diabetes frequently causes fat gain for the same reason.  In the paper by Dr. Mehran and colleagues, they suppressed insulin secretion without increasing insulin sensitivity, creating an insulin deficiency state similar to mild diabetes, and the result was the same: elevated blood glucose and resistance to fat gain.  I fail to see why this is earth shattering.

Although the result is interesting from an academic perspective, it isn't relevant to common obesity where insulin resistance and insulin secretion parallel one another, and it certainly does not erase all previous evidence suggesting that hyperinsulinemia isn't required for obesity in mice or humans.  I think we can hold off on re-writing those textbooks for the time being.