Response to insulin series part 3

Part 3 of this five-part series deals with the paradox of certain insulinemic foods that do not cause weight gain. In particular, the focus is on dairy since it is one of the most insulinemic foods (meaning, it causes a strong insulin response).

At the very get go, we can see an inherent flaw in Krieger’s argument logic. Here is the particular quote that starts off his first content paragraph:

If the carbohydrate/insulin hypothesis were true, then we would predict that foods that are extremely insulinemic would be uniquely fat promoting.

This is another example of confusing the tenants of the point he is trying to disprove. Krieger is challenging (according to his first article and the series premise) the carbohydrate/insulin hypothesis that carbs raise insulin and cause fat gain. With the above statement, he ignores the role of carbohydrates and goes straight towards attempting to disprove insulin as a fat promoter in and of itself. What that statement should correctly read is “If the carbohydrate/insulin hyp. were true, we would predict that foods that are high in carbohydrate are also insulinemic and uniquely fat promoting.”

There is a big difference in these two statements. Insulin cannot be said to be directly the cause of fat gain. Instead it is associated with fat gain. By eliminating the carbohydrate variable from the statement, he has wrongfully changed the correlation into a causation. Therefore, arguing causation has set him up for seemingly startling null conclusions with regard to insulin and fat accumulation. In fact, however, we cannot dissociate insulin from carbohydrate without creating a new hypothesis. So Krieger does NOTHING to disprove the carbohydrate/insulin hypothesis. Instead, he lectures us about dairy.

Despite his flaws in the initial assertion, Krieger does highlight some interesting points about dairy that I would like to cover. He states that dairy products are insulinemic, and yet they don’t promote weight gain. Now I think the latter part of that statement is debatable since milk is in fact used in some cultures for intentional weight gain, as well as the fact that dairy is an agent specifically manufactured to induce rapid growth in baby mammals. Regardless, Krieger is correct in stating that dairy is strangely insulinemic in that it causes a large insulin response and yet contains low amounts of carbohydrate, which comes from the only sugar, lactose. This is due to the branch chain amino acid (BCAA), leucine, which was discussed at length in part 1 of the Krieger series and in my critique. Milk (whey) contains ample quantities of BCAA’s, particularly leucine, and essential BCAA’s are potent stimulants of insulin release. Essential AA’s are called such because the body cannot synthesize them and hence they must be acquired from the diet. The accompanying insulin spike is important because leucine bypasses the liver to be directly used by cells, especially those of muscle tissue. Therefore, insulin is released to shuttle leucine into the cells.

It is important to know how leucine is handled upon ingestion to understand the paradox of high insulin response to a low carbohydrate containing food such as dairy. At this point, Krieger reproduces three figures, which I’ve copied below, adapted from Nilsson et al. displaying the AUC’s (area under the curve) for insulin, glucose and the insulinogenic index in response to seven test foods (such as white bread and milk) and food compound mixtures (such as gluten/lactose and whey/lactose).

It is unfortunate that not all variables are isolated in the referenced study, and so results are based off of suboptimal observations. Still, the figures make a convincing point that complete dairy products and some independent components of dairy are uniquely insulinemic and yet do not necessarily cause a high blood glucose response. (One note about the glucose AUC figure is that it really gives us no predictive ability about what food or mixtures will cause high glucose and which will not because, as we are already observing the unexpected insulinogenic effects with whey and lactose, the variables are not isolated and I believe this is a poor example. We basically have no notion of the potential synergistic effects of some of these food mixtures on blood sugar and insulin release upon consumption. We have no baseline). As said before and in the Krieger article, this is a result of the amino acid, leucine. However, what leucine does in the body compared to what carbohydrates do is radically different. The leucine gets shipped to our muscle tissue where it in conjunction with insulin signals for protein synthesis (the creation of new protein). Leucine and amino acids in general also stimulate the release of glucagon to counteract the hypoglycemia caused by the release of insulin. The glucagon tells the liver to free store glycogen to convert to glucose and also regulates glucose production through lypolysis (breakdown of fat tissue). Carbohydrates, on the other hand, do not stimulate glucagon and instead stimulate high insulin with no signal to counteract hypoglycemia. Over time, this is what blunts cellular insulin sensitivity and leads to overproduction of fat.

So while all this dairy induces an insulin response much higher than we would expect based on its carbohydrate content, it is the positive, quickly attenuating, transient spike that we would expect (and want) due to the type and concentration of the branched chain amino acid, leucine. Also, I guess a lot of science is really about who you ask (or what study you link), as Holt et al. suggests that the insulin index of white bread is still considerably higher (2x as high or more) than protein rich foods including cheese, eggs, fish, and beef (milk was not included in this study). Holt et al. fed isoenergetic portions of individual foods to a group of healthy subjects and then took finger prick blood samples at different postprandial time intervals to measure insulin and glucose response. The Nilsson study Krieger linked used “reconstituted milk, cheese, whey, cod and gluten” and if you read their methods, all but the cod, cheese and bread were served as a liquid (usually powder dissolved in tap water). The milk was actually 96g “roller-dried skimmed milk” powder combined with 937g water (you can learn about the roller drying process here in which high heat is used to vaporize the water). Essentially, this study tells us nothing about how real, whole food affects glucose and insulin levels upon ingestion. To summarize, yes protein rich foods elicit insulin responses disproportionately higher than expected based on their glycemic values alone, especially those containing essential amino acids and the branched chain amino acid leucine. However, according to Holt et al., protein rich foods still only cause about half of the insulin response as that induced by white bread (see figure below), and since dairy has such a low glycemic load then we of course expect only a modest rise in blood glucose values.

From Holt. et al. 1997, AM J Clin Nutr

The result of dairy consumption, based on its protein and carbohydrate content, is about the same as consuming any other source of high quality protein. Amino acids stimulate insulin release to push the BCAA’s into tissue where it can stimulate protein synthesis and glucagon release. As for the increased insulin response to a high GI meal with the addition of milk, again it is a combination of variables, and again the results can only tell us about the response to the test meal specifically rather than the individual foods.

Now, all of this lecture, both on my part and on Krieger’s has done nothing to refute or invalidate the carbohydrate/insulin hypothesis that maintains that carbohydrate consumption over time can lead to excess insulin production, insulin insensitivity, and overproduction of fat. I find this comment in particular to be most illuminating as it shows the limitations in his thinking:

The above data illustrates one of the problems with the carbohydrate/insulin hypothesis…it assumes that carbohydrate is the primary stimulus of insulin secretion.  However, it is clear that amino acids and incretins play significant roles in insulin secretion as well.  And as I pointed out in part 1 of this series, the blood sugar response of a food only explains 23% of the variation in the insulin response.  Thus, a lot more goes into insulin secretion than the blood sugar response from eating carbohydrate.

Nothing about the carbohydrate/insulin hypothesis itself has assumed anything about primary stimulants of insulin. All the hypothesis proposes is that carbohydrates are uniquely fat producing because of the dis-regulation of insulin secretion over time. The assumptions he refers to are the assumptions mainly of lay people when metabolic biochemistry is reduced to simplistic explanations to convey more complex processes. To me this statement is an example of the fact that Krieger does not actually know what he is arguing against. His stance is that the carbohydrate/insulin hypothesis is wrong and yet he continues to argue about only one component of the hypothesis, that it requires elevated insulin levels and so he provides evidence for other insulin inducing foods… that aren’t carbohydrate. And so he fails to address the hypothesis in its entirety and instead shows useless data on uncontentious observations. Yes we know protein induces insulin secretion. Yes we know that protein (this includes dairy) can be part of a diet and in fact help people lose weight rather than gain. This has nothing to do with insulin dis-regulation. Instead, it has everything to do with the fact that these foods, within a certain population of people (since lactose intolerant people are excluded from benefiting from healthful effects of dairy consumption), actually maintain or help reinstate proper metabolism. A proper metabolism includes insulin secretion. This is nothing new or surprising.

To conclude, I will quote and respond succinctly:

Thus, if the carbohydrate/insulin hypothesis were true, then we would predict that a diet high in dairy products should promote weight and fat gain.

No, because we know that dairy is not a high carbohydrate food and is both low glycemic and induces a low glycemic load. Thus it does not necessarily apply to the carbohydrate/insulin hypothesis. Even still, for some people milk actually is not tolerated on a fat-loss, low carbohydrate program.

Of course, these are animal studies.  What about humans?  In one study, low-fat dairy products did not promote weight gain, while high-fat dairy products did.  Hmmm, could it be that the weight gain in this study was simply caused by excess calories and not insulin?

If you want to make a point about something by citing studies, make sure it’s good science. The article linked is a model of how NOT to conduct a dietary experiment. Grab volunteers, don’t control their diets, provide no macronutrient baseline for their diets, don’t track what they eat during experimentation, and then randomize their experimental feeding by having the lunch ladies at the residence halls slip them extra full-fat or low-fat milk and yogurt. And if you’re trying to decide whether it’s fat or carbohydrate that induces weight gain, at least try to isolate the calorie content of each experimental meal. The above experiment provided nearly twice as many calories in the high fat dairy than the low fat dairy (2177kcal versus 1357kcal, and I believe that was in the dairy supplement alone). Also, the resulting change in weight? The full fat high calorie dairy group gained +1 lb on average and the low fat low calorie dairy group lost 0.2 lbs. Dang. Big changes there. James, please go ahead and calculate the additional calories consumed by the high calorie dairy group per day over 8 weeks and let me know how many pounds you would expect them to gain if we invoked thermodynamics in our reasoning.

In a 9-month study, increased intake of dairy products did not affect weight maintenance, but the high dairy group exhibited evidence of greater fat oxidation.

The experimental results of this study say that the high dairy group was able to consume greater energy without greater weight gain and they mobilized more fat. So, Krieger now favors a study that says an increase in calorie consumption will have an automatic compensatory effect on energy expenditure (thereby nullifying his proposition that we gain and lose weight by increasing or decreasing calorie consumption). Also, just glancing through the volley of studies he links in the end of this article shows you that the results are all over the place. One says high fat dairy makes people fat (while Krieger argues that dairy is not fat promoting), others say dairy made some people fat and others lose weight (depending on if you lived in Kansas or if you lived in Tennessee). Another says that dairy consumption has no impact on fat mass, but then later it says dairy consumption predicted fat loss. Nothing is very consistent. The studies are wrought with procedural problems, and still Krieger has done nothing to invalidate the carbohydrate/insulin hypothesis.

This pretty much ends the third discussion in the insulin article series. To drive the point home we learn that in his n=1 study, Krieger consumes plentiful dairy (Greek yogurt, cheese, cottage cheese, milk, whey; all very high protein and low carb), assumes he must have very high insulin, but maintains that he is not fat. While I can’t comment much here, I’d like to see a little bit better anecdotal evidence. At least show readers some test results of your fasting and post-prandial insulin levels. I bet they are actually rather low. I too consume plenty of Greek yogurt, cheese, beef, fish and other foods high in protein, am not overweight (BMI 21, fat ~15-19%) and my fasting insulin is <2 (which is very low).



  1. […] The third part of my critique to the insulin article series written by James Krieger is now live. Click on the Insulin article series tab and go to “part 3″ or just click here. […]

  2. Great work, keep it up.

  3. When are you going to post part 4? I’m wading through what you have explained in the 1st 3 parts, and although quite technical, I think that I am getting the main ideas.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s