The Insulin Paradox

I think this post has needed a bump for a while.     I also think the data on growing animals eating a keto diet should be cause for at least caution in encouraging such diets for kids, teens or even young adults.


One of the more difficult things to sort out from the enormous amount of information out there is the actual effect, if any, of dietary composition on resulting body composition.  This is complicated from the get-go by many rodent studies that are done on reproductively mature animals that are still growing.  How to compare this to humans who cease growing vertically once they reach adulthood?  Next we have to sort through the studies that are done, and the implications of them, in the context of weight loss/underfeeding vs. those in the context of weight gain/overfeeding.  Clearly there are compensatory mechanisms at play in both situations that are drastically different.  Lastly, and here's where we probably have the least amount of information, and practically nothing in humans, what role does one's dietary composition have on their body composition over the long haul for a "normal" weight stable adult?    Verifiable science basically fails us on this last point, and we're left mostly to the "wolves" of the diet/fitness industry -- e.g. body builders,  trainers and nutritionists for elite athletes, etc. -- for their anecdotal reports.  I don't mean to disparage these folks with the term wolves (hence the quotes), but we can never separate the reports from the reality when there's money on the line in a world where ads are required to include "results not typical" disclaimers.  While I'm not saying all of these folks, or even a majority, are of the 2am infomercial variety, the claims remain largely unverifiable.  It would sure be cool if there was a privacy-sparing way to collect verifiable body composition, dietary and activity data for a large population of free-living humans, eh?

There can be no doubt that insulin plays a major role in regulating metabolic traffic in the fat cells.  Insulin stimulates fatty acid uptake into adipocytes and suppresses lipolysis from the fat cells so that high insulin would seem to favor fat storage.  This has been exploited by one Gary Taubes and others to mislead a lot of people over the past several years into thinking that insulin is thus the primary regulator of fat tissue mass through this mechanism.  The so-called "lipophilia" hypothesis -- that high insulin levels trap fats in adipose tissue making less available for other cells that "starve" -- implies that insulin plays a pivotal role in fuel partitioning,  favoring partitioning to fat stores when elevated.  The unspoken implication is, then, that when insulin levels are low, fuel partitioning to lean tissue is preferred.   I think the reason this is generally unspoken is because it would not make sense if presented as such directly.

And therein lies what I'll call 

When insulin levels are elevated, the uptake/esterification of fatty acids in fat tissue is enhanced while lipolysis/release of fatty acids stored as triglycerides is suppressed,  thus favoring fatty acid accumulation in adipose tissue.
The low insulin state favors partitioning of fuel (mass) to fat tissue stores vs. lean tissue.

Huh?  Clearly I've lost my mind now.  Please hold off the men in white coats coming to drag me off to a real insane asylum long enough for me to explain.  Because this post has been quite a long time in coming.  I have read literally hundreds of papers on this subject over the past few years.  My regular readers and/or folks who have interacted with me outside this blog may have noticed that I possess a number of, shall we say, outlier qualities.  Among them is an amazing memory, and a quirky one at that.  What happens with me is that often I pick up on a seemingly insignificant passage in something I read, or seemingly insignificant feature in something I see.  These things tend to stick in the back of my mind, and the next time I see it repeated it jumps out at me.   Then out of the blue I read a statement that ties all of these past remembered things together. I don't know if I'm explaining this very well, or should even bother to try, but this is how posts like this one evolve.  And the evidence I'll provide in this post is but a fraction of what I've come across these past years.  I kept picking up on the fact that, time and again, although it wasn't the focus of the studies at hand,  the dietary milieu that creates a low (chronic) insulin state favors fuel partitioning to fat mass over lean mass.  Low insulin is associated with low nutrient availability, thus a conservation mode.  And a conservation mode is associated with building less lean tissue and hanging on to fat stores.

Before presenting some evidence, I'd like to point out that a lot of this information comes from the longevity realm.  The most common dietary intervention studied in this regard is calorie restriction -- usually somewhere in the two-thirds to three-quarters of normal intake range.  Calorie restriction (CR) is known to (a) slow the metabolic rate and (b) reduce insulin levels.  In many studies, CR results in reduction of all macronutrients as the subjects are merely fed a reduced amount of a diet of the same composition.  This has sparked the interest of low carb advocates who are interested in showing that carbohydrate restriction is sufficient to produce the longevity benefits from calorie restriction.  Many in the longevity realm also look to protein restriction alone -- and certain amino acids more than others -- to see if the same benefits can be realized, or perhaps if specific macro restriction may even be more effective.  Overall, the target "bad guy" hormone seems to be insulin, a hormone elicited postprandially by both carbohydrate and protein.    But as I'm about to show, if one is successful in lowering insulin levels, the trade-off is less lean mass and/or more fat mass compared to the normo-insulin state.

A Smattering of the Evidence

C. elegans (aka worms):  
Speaking of my memory, it was the quote I'm about to provide that really touched off this post.  I had remembered it almost verbatim, but try as I might I couldn't find the paper it came from when I went looking back.  Just today, I found it on my HD with a non-descript Pub# title of all things.  Phew!  I'm not nuts, I really did read that! From DAF-16 Target Genes That Control C. elegans Life-Span and Metabolism, here's the quote:  (note:  if you would like the full text, email me carbsane at gmail dot com)
The C. elegans daf-2 pathway controls longevity, metabolism, and development and is orthologous to the mammalian insulin and insulin-like growth factor 1 signaling cascade. Decreased daf-2 signaling causes up to threefold life-span extension, increased fat storage, and constitutive arrest at the dauer diapause stage.
That's what jumped off the page at me and all of these other recollections flooded in.  What's this reduced insulin (or insulin-like) action and increased fat storage stuff?  Huh?  I've quoted   this study previously:
In C. elegans, mutations in several genes, such as daf-2 and tph-1, cause an extension in life span and also result in an increased fat content in the intestine [the main fat storage organ in these worms]  speculated that these mutants shift from fat-metabolizing to fat-storing adults. Additionally, mutations in these genes result in an increase in TG content. We asked whether tub-1 mutants that have a higher TG content also show a change in life span ... in addition to increased fat storage, mutations in tub-1 lead to a 20% extension of life span.  
The worm data is clear.  The long-living mutants all seem to have some alteration in their insulin-like signaling that is analogous to reduced insulin levels.  And these long-living mutants also have higher than wild-type fat accumulation.

Normal "Adult" C57BL/6 Mice:    
In this study, 8 week old mice were broken into four groups.  One group continued ad libitum standard chow, a calorie restricted group received 66% of calories from the ad libitum standard chow, the remaining two groups received ad libitum either a ketogenic diet or a high fat chow (usually used for diet-induced obesity, similar to SAD).  Mice normally grow slightly on their standard diet -- roughly 27.5g mice add about 2g mass over two months.  I've consolidated the results in the graphic below.
The body compositions and insulin levels were assessed after 6 weeks (day 42) on the timeline of the top graphs.  Focusing on just the black and white bars on the left -- contrasting the standard chow with the KD chow -- we see that although the KD mice actually lost weight, their fat mass is virtually identical to the standard chow mice, but in terms of lean mass, the KD mice lost lean mass.  And while the standard controls were hardly hyperinsulinemic, the KD mice are virtually type 1's with extremely low insulin.  
Summary:  During normal adulthood (slight growth) over six weeks, mice on a diet that dramatically reduced basal insulin levels retained/maintained fat mass similar to standard controls while losing lean body mass.

Young (30d) Normal Wistar Rats
In this study, normal rats were weaned onto one of two diets, control = standard chow, and KD = ketogenic.  The body weight and composition were assessed weekly for six weeks.  The results are shown at right.  Although the KD rats gained less total body mass, they gained more fat mass, and thus their fat/total mass ratio was unfavorable.  It doesn't get much more clear than this.  
These growing rats, on a low insulin diet, squirreled away more fat tissue while directing fewer nutrient resources towards the building of lean mass.

Overfed Adult Humans:
Yes, that Bray overfeeding study.  The insulin levels of the subjects were not measured to my understanding, and none of the overfeeding diets could be described as "low insulin" per se.  Still, it is common knowledge that protein and carb increase postprandial insulin and since the low protein group actually reduced both modestly from baseline, getting all of their surplus calories from fat, while the high protein group increased protein consumption considerably thus getting roughly half of their excesses from fat and protein, comparing these two groups does compare relative insulin.  At baseline, these participants consumed roughly 2400 cal/day and for 8 weeks of overfeeding, they consumed roughly 3350 cal/day.  Their carb consumption was constant throughout ALL phases in the study across all groups at about 350g/day.  Thus participants overate fat (only in the LP group) and protein and gained weight.  As this table shows, when gaining weight, the excess fuel is clearly shunted to fat mass over lean in the lower insulin state.
It would have been interesting to see longer term results here.  Another way of looking at this is that without the extra insulinogenic contribution of protein, ALL of the surplus energy is partitioned to fat storage, but when one adds considerable insulin (and likely IGF-1) action in the form of half of the surplus calories from protein, considerable fuel is partitioned to building lean mass and the ample carbohydrate in the diet is NOT, counter to claims, sucked into and trapped in the fat tissue.

Human Children on Ketogenic Diets:
I can't find the exact reference at the moment, and unfortunately I cannot find one where body composition was assessed before and after.  However, I recall a study that followed kids on such diets for several years.  The ketogenic diet is associated with stunted growth (height) which would presumably be due to the low protein content in the diet.  Since vertical growth requires lean mass (e.g. bone) growth, it is reasonable to presume that, as in animal models, lean mass building is spared in favor of building and/or conserving fat reserves.  Here's an abstract to a similar study to the one I'm thinking of.  It appears the degree of ketosis was correlated with the degree of height-for-age decreases.  Those in higher ketosis saw decreases in height-for-age while those in moderate ketosis did not.  

OK, as I said, a smattering.  I welcome links to studies that confirm or contradict what I've written here.   This is but a fraction of the studies and I just don't have time to collect them all now.  Still the theme repeats.  Low insulin dietary milieu favors increased fat mass vs. lean mass.  I can hear the hair pulling now about the amazing results of low carb in weight loss scenarios.  I think this seeming contradiction is easily explained in a manner consistent with this data.   When the typical SAD-eating obese person goes low carb (not listening to current gurus), they end up pretty much cutting carbs from their diet but not altering their fat and protein intake much.  Thus, whether by percent or not, the LC diet is reduced in calories while remaining high in protein.  The combination of the energy deficit and protein insulin (and likely IGF-1) stimulation likely acts to preserve lean mass while burning off stored fat compared to the CRD.  Conventional CRD's, especially for women, generally impose a smaller caloric deficit than many spontaneously achieve low carbing (at least initially), and if based on protein as a percent are effectively low protein diets leading predictably to LBM losses.  

This unmistakeable reality about insulin may go a long way towards explaining the disconnect between dramatic low carb weight loss successes and the long term shortcomings of the very same way of eating that produced them.  


Sanjeev said…
> favors fuel partitioning to fat mass over lean mass
Tying into the more recent understanding of insulin's anti-catabolic role, does this formulation also make sense to you -

low insulin favours lean mass catabolism over fat mass catabolism
CarbSane said…
I have to think on that a bit b/c clearly enough LC weight loss studies show a significant breakdown of fat mass over lean. I think our bodies break down lean only when absolutely needed when fat stores are constantly in flux and in a way, they are there for just the purpose of providing energy when dietary sources are sparse.
Jim said…
Evelyn, I think there is another issue besides overfeeding and underfeeding. I think the main physiological difference between Mark Sisson and Jimmy Moore is that Sisson has so much less fat mass. Long-term low-carbing when you have a lot of fat mass is detrimental as you suggest, but I am not sure that it is detrimental if your body fat percentage is low. Indeed some people may actually experience better glucose tolerance when they have more fat mass. Perhaps Sisson, Feinman, Taubes, Groves, and other lean long-term low-carbers would experience insidious weight gain if they raised the carbs, but these luminaries don't necessarily understand that people with more body fat than they have might actually have better glucose tolerance than they do and need a lower-fat regimen to lose body fat.
Mr. Wrong said…
I am curious Evelyn, did my reference to DAF 16 on your 'Protein, Insulin-like Growth Factor-1 (IGF-1) and Glucose Homeostasis' post possibly put this gene back to the front of your mind?

Also, I am unable to find the original study but I remember reading also that glucose feeding upregulated DAF 2(longevity reducing) and turned off DAF16(longevity promoting). While removing glucose from the diet did just the opposite. After seeing these results the researcher switched to low carb (although possibly prematurely, these are worms after all).

Have you read of this? Would this imply that insulin resistance might be protective? It would appear to me that insulin resistance is not necessarily the problem but possibly the body's natural defense against flooding your body with glucose.
Kindke said…
Sanjeev & CarbSane,

Do you both favour high fasting insulin over low fasting insulin then?
CarbSane said…
Hi Jim, I don't know Groves, but pics I've seen of Feinman, he could easily fit the bill of fattIer. Taubes says his wife calls him meat man, and his regular diet is indeed low carb but probably sufficiently high in protein that it's not really ketogenic. He's a former athlete which I would never have guessed from pics but I'm told he looks the part in person. Mark clearly had all his muscles and such before he went low carb and he, as well, despite his new schtick, is hardly ketogenic. If I recall, his sample menu was under 60% fat and over 100g carb (albeit he counted every little carb in every veggie). So I think once you've built the lean, it seems all that's needed to preserve it is sufficient protein which most low carbers should get unless they're eating a true ketogenic diet. The only reason I believe the new Boy Wonder is achieving good results at this point is that he's working out like 3 hours a day and consuming 4-5000 cal/day. So even at 10% protein that's sufficient for his 5'10" 175 lb body (I think that's what he quotes he weighs). I think this man is an idiot for doing this self experiment, but that's just me ....

I always ask low carbers who insist it's about the insulin and not the calories why Dr. Atkins' "fat fast" (that lives on for lengthier periods of time with Dr. Salerno's Silver Cloud diet) was strictly calorie controlled at 1000-1200 cal/day. It is no wonder this diet breaks stalls because it is likely (a) imposes a caloric deficit and (b) has around 40g protein per day so triggers LBM breakdown.

I also can't get over the calories these guys actually eat. Mark's sample menu came to like 2500 cal/day and he was clear he almost never eats all meals. In his recent interview with Sean Croxton (the free bonus from the Paleo Summit) Mark seemed to almost brag how he hadn't had breakfast just some cream in his coffee. Then they ate taco-less fish tacos for lunch -- hardly a high fat or calorie fest, a little oil to fry halibut? -- followed by a workout. Even if he's having a ribeye and veggies for dinner, that's HARDLY a lot of anything.

people with more body fat than they have might actually have better glucose tolerance than they do and need a lower-fat regimen to lose body fat.

CarbSane said…
Welcome Mr. Wrong! I just looked back in my reader. I apologize I missed your comment, so while it's a fun coincidence that wasn't it ;) (I was looking for another article by Kimura on my HD and it was a reference in this PDF so up it popped -- I love computers!)

I think you're talking about Cynthia Kenyon. My take is that it is more biologically expensive to maintain lean mass so in a low insulin state favoring reduced metabolism and conservation the worms last longer -- in the lab with other worms of similar bent.
CarbSane said…
I don't. Low fasting insulin is indicative of metabolic balance. But the ketos in the cited studies have low 24h insulin AUC and it would appear that this favors conservation. A person eating a truly keto diet (low protein, very high fat) may well be recomposing to a fattIer person. We hear all the time how there are all these "skinny fat" people on the SAD. It would appear that VLC low postprandial insulin favors such a body type. Do you think the evidence says otherwise?
Sanjeev said…
In situations where healthy (or optimally functioning, stressor - free) bodies don't produce insulin why would I prefer high insulin (if I'm healthy)
Nigel Kinbrum said…
"It would appear to me that insulin resistance is not necessarily the problem but possibly the body's natural defense against flooding your body with glucose."
At a cellular level, insulin resistance is a cell's defence against over-filling.

At a body level, insulin resistance results in hyperglycaemia which is damaging. IR is therefore a poor defence against flooding your body with glucose.

On ketogenic diets, cellular insulin resistance is desirable as it spares glucose for red blood cells & the parts of the brain that burn only glucose (due to the cells having no mitochondria).
Jim said…
Maybe I need to get out more, but I don't ever see "skinny fat" amongst true low-carbers. They are either fat as pigs or else lean and fit. To see "skinny fat" or frail you need to look at the Walter Willett types with their whole grains, legumes, vegetable oil, and avoidance of animal foods.
CarbSane said…
I don't want to get on the bad side of folks by naming names, but there are a number of ex-obese low carbers who would qualify as skinny fat. At one point I would say even Jimmy Moore would have fallen under that category. This is not to say that there aren't examples on the other side of the fence, but ... Also, I've never seen Rosedale, Feinman, Eades or even Taubes for that matter, unposed without a shirt.
Harry said…
The 'skinny-fat' body composition is more a function of exercise than diet. Simply put, if you don't have enough muscle (mostly generated by resistance training) and you have normal to above normal levels of body fat, you're going to look skinny fat.

So, the two conditions (low muscle mass, normal/high fat mass) are pretty achievable on all sorts of diet, so long as one is committed to a sedentary life.

I've seen all sorts of skinny fat clients with all sorts of dietary backgrounds...the only common feature is a lack of resistance training...and the remedy reflects that.

Mr. Wrong said…
I have to agree with you Harry. I tried vegan and after losing 55 lbs in 6 months I was down to 150lbs. I was a 6'1" rail with a pooch. I started eating 'Primal' and got back to 165-170 where I stayed for many months but still was skinny fat. It was not until I started resistance training that I lost the belly fat and gained an extra 5 lbs of muscle.
ProudDaddy said…
Even though you're 100% correct, Harry, you've uttered the two words that a lot of people (I almost typed "women", whew!) do not want to hear: resistance training! I've been trying to get Carbsane to include the subject more often, so thanks for the help.
bentleyj74 said…
Dittoing Harry and posting a helpful example of what body fat percentages look like on different individuals with regard to their overall composition. That also touches on scale weight variability. You have to scroll down a bit to get to the example pics.

There is a difference between having a high fat percentage in general and having a high fat percentage relative to lean mass [underdeveloped]. We are very reluctant to call a spade a spade despite protestations to the contrary.

I'm going to be blogging about that actually in the nearish future.
Kindke said…
Adding to Harry's comment I just want point out that the Myostatin/Follistatin axis is far more important for lean mass aslong as your insulin isnt zero like type 1 diabetics.

I can gain muscle mass even on a strict ketogenic diet aslong as I exercise. Meanwhile, if I leave my leg in a cast and dont move it, the muscle will atrophy regardless of my how much my diet spikes insulin.
Jason Sandeman said…
I can give a small insight onto this - Prior to my diagnosis as a diabetic, (type 1,) I lost a lot of weight. 60 lbs to be exact.
I lost the fat first, and was in danger of losing the lean mass.
This is what happens with an uncontrolled diabetic with heavy insulin resistance. The body does not react to the insulin, and therefore thinks the insulin is missing, and starts to catabolize in order to gain energy. In essence, going into ketosis.
Fat is the first to go, then lean mass. The reasoning, (and I have nothing real to back this up, it's just a hypothesis) is that fat is easier to convert to energy than lean proteins. Plus, when protein is converted to glucose, IIRC the result is an acidic state - which is dangerous to an undiagnosed T1D in the form of Ketoacidosis.

I hope that sheds some light on things. I know that I am diabetic, and some will say that skews things, but the logic makes sense to me. While I was producing insulin, my body was resistant. I burned my pancreas out, so now I produce next to none. If I was to stop taking my insulin, I would lose weight quickly - but perhaps too quickly.
There is an actual eating disorder called diabulemia which is the ceasing of following the insulin regiment in order to lose weight. Perhaps this sheds light on the subject as well.
Jason Sandeman said…
Sorry, the rambling. LOL
I meant to say the formula is reversed.

low insulin favours fat mass catabolism over lean mass catabolism
Tsimblist said…
Thank you, that is a helpful example. I purchased a body fat scale back in January. My intent is to track changes in my body fat. But when I configure my profile for the scale, I must select which model to use to interpret the raw data. I had narrowed my choices down to "2011 Version" and "Athlete Version 2011" but couldn't decide which model applied to my case. The example pics tell me that the athlete model definitely does not apply to me.
Blogger said…
With bistroMD you know that not only will you get tasty entrees, but that every meal and every day in bistroMD's weight loss programs is balanced to bistroMD's specific nutritional platform to help promote healthy weight loss.

STEP 1 - Select one of the weight loss plans for 5 or 7 days of entrees.
STEP 2 - Review your menu before ordering and pick the entrees you want for each day and week.
STEP 3 - Order your diet program online.
STEP 4 - Your entrees are delivered to your door.

ORDER NOW - delivered to your doorstep.