Sunday, January 15, 2012

What Causes Insulin Resistance? Part V

Previously in this series, we've discussed the role of cellular energy excess, inflammation, brain insulin resistance, and micronutrient status in insulin resistance.  In this post, I'll explore the role of macronutrients and sugar in insulin sensitivity.

Carbohydrate and Fat

There are a number of studies on the effect of carbohydrate:fat ratios on insulin sensitivity, but many of them are confounded by fat loss (e.g., low-carbohydrate and low-fat weight loss studies), which almost invariably improves insulin sensitivity.  What interests me the most is to understand what effect different carbohydrate:fat ratios have on insulin sensitivity in healthy, weight stable people.  This will get at what causes insulin resistance in someone who does not already have it.

The first study to address this question was published in 1935 by Dr. H.P. Himsworth (Himsworth HP. Clin Sci 2:67. 1935).  He found that insulin sensitivity was increased by feeding a high-carbohydrate diet and decreased by feeding a low-carbohydrate diet, but these effects were only observed at very high (70-80%) and very low (less than 10%) carbohydrate intakes, respectively.

Studies since then have been somewhat equivocal, with some showing higher insulin sensitivity on high-carbohydrate diets (1, 2), and others unable to replicate the finding (3).  One potential explanation is that the difference in carbohydrate:fat ratios was not sufficient to see a difference in the studies that did not confirm the effect.  As per Dr. Himsworth's finding, reduced insulin sensitivity seems to require a very low carbohydrate intake.  Qualitative differences between the fat and carbohydrate sources may also have played a role-- unrefined carbohydrates improve insulin sensitivity more than refined carbohydrates.

Although very low-carbohydrate diets can reduce insulin sensitivity (and this squares with the poor glucose tolerance of many people who have followed very low-carbohydrate diets long-term), I think it is fair to ask whether this is harmful or simply adaptive (pathological vs. physiological).  When glucose is scarce in the diet, the body initiates a metabolic program to spare glucose for the brain, which is more dependent on glucose than other organs.  Since insulin signals tissues to increase glucose uptake, it is logical to think that insulin resistance might be a way of reducing glucose use by peripheral tissues so that the brain gets a larger share.  It's not clear whether or not this type of adaptive mechanism can fully account for the insulin resistance on such diets.  In rodents, insulin resistance on purified high-fat diets rapidly becomes pathological (i.e. it is clearly not an adaptive response because it leads to health problems), but I think we have to be cautious about extrapolating that finding to humans eating a mixed diet.  I would call this question unresolved. 

Much has been made of the glycemic index (speed of absorption) of carbohydrate as a determinant of insulin sensitivity, but a comprehensive review of controlled trials reveals that the glycemic index of carbohydrate generally has no measurable impact on insulin sensitivity (4). 

I think a simple principle that can explain these findings is that a healthy body increases insulin sensitivity in response to increased demand for glucose disposal.  In other words, the more carbohydrate in the diet, the more the body increases insulin sensitivity to appropriately metabolize it. This is particularly apparent at the extremes of carbohydrate intake. 

Sugar and Fructose

For the purposes of this discussion, I'm defining sugar as sucrose (table sugar, cane sugar), its close cousin high-fructose corn syrup, and natural sugars derived from fruit and honey.  When referring to fructose and glucose, I'll refer to them specifically.  Starch is made up of long chains of glucose that are released upon digestion.  Sucrose is composed of 50% glucose, and 50% fructose, linked together.  The form of high-fructose corn syrup that is typically used in soft drinks is roughly 55% fructose and 42% glucose, and the form that is used in most other foods is roughly 42% fructose and 53% glucose.

Sugar consumption per se has not been consistently linked to insulin resistance or diabetes risk in observational studies (update 1/20/12: fructose intake has just been associated with insulin resistance in adolescents.  The association seems to be mediated by increased visceral fat).  Most of the research on insulin resistance has focused on the role of fructose, because there is not much evidence that glucose or starch themselves are involved.  A number of studies have investigated the role of fructose consumption in insulin sensitivity.  Fructose or sucrose feeding consistently causes insulin resistance in rodents, but the effect requires unnaturally large amounts if given in the context of a low-fat diet (5, 6). 

In humans, high-dose refined fructose feeding can cause insulin resistance in as little as a week (7), however this requires an amount of fructose that far exceeds what can be obtained in a normal diet.  Somewhat lower, but still very high (~100 g/d) fructose intakes do not lead to insulin resistance in healthy, lean adults when fed for four weeks (8).  I think it is notable that in Dr. Lynda Frassetto's Paleo diet study, the diet was fairly high in sugar from fruit, fruit juice and carrot juice, yet participants experienced impressive improvements in insulin sensitivity (9).  This shows that naturally sourced sugars from fruit and vegetables are compatible with high insulin sensitivity in the context of a healthy diet.  As another example, the inhabitants of the island of Kitava eat a diet rich in fruit and thus unrefined sugar, yet in the context of a traditional diet and lifestyle, they do not develop insulin resistance (10). 

However, in overweight and obese people, a high intake of refined fructose can exacerbate insulin resistance.  Dr. Peter Havel showed this in a thorough study in 2009 (11).  After 10 weeks of drinking fructose- or glucose-sweetened beverages for 25% of calories, both groups gained the same amount of body fat (~3%), yet the fructose group gained it disproportionately in the abdominal (belly) region, and also showed a decrease in insulin sensitivity of 17%.  Another interesting study showed that a very high fructose intake (200 g/d) caused insulin resistance and increased blood pressure in overweight and obese volunteers, and that these effects were partially blocked by allopurinol, a drug that lowers circulating uric acid (12).  This supports the hypothesis that fructose exerts some of its effects by increasing uric acid, but it seems to take a lot of fructose to significantly increase plasma uric acid levels.

In another interesting study, Dr. Judith Hallfrisch and colleagues fed insulin resistant and normal men diets containing 0%, 7.5% and 15% fructose for 5 weeks (13).  None of the diets influenced fasting insulin, but the 15% diet increased the insulin response to a sucrose drink.  There was no glucose control group, so we don't know if the same response would have been observed with glucose or other sugars.  The level of sweetness of the different diets was not matched.

In a more realistic study, Dr. Maria Maersk and colleagues showed that adding one liter (equivalent to 2.75 12 oz cans; 106 g/d of sugar or 53 g/d of fructose) of cola to the habitual diet for 6 months resulted in an increase in liver and visceral fat, but no change in fasting insulin or estimated insulin sensitivity in overweight volunteers, relative to the same volume of water, milk or diet cola (14).  There was no significant effect on body fatness, although there was a trend toward an increase in the soda group.  I suspect that if these trends continued, insulin resistance would have developed eventually, but that may simply have been secondary to increased fat mass. In any case, regardless of its effects on insulin sensitivity, this study supports the idea that sweetened soda is not healthy! 

Overall, these studies show that very high fructose intakes can worsen insulin resistance in people who carry excess fat, but they do not offer strong support to the idea that normal intakes of fructose cause insulin resistance in those who are lean (particularly when the sugar comes from natural sources).  It is possible that insulin resistance would develop over a longer period of refined sugar consumption, particularly if visceral and liver fat continued to accumulate, but this has not been demonstrated to my knowledge.  Although some uncertainty remains, the evidence I've seen does not convince me that sugar intake within the typical range causes insulin resistance through an inherent metabolic effect of fructose, and I have seen nothing that suggests we should avoid eating whole fresh fruit.  However, since added sugar increases the energy density and palatability/reward value of food, it can contribute to insulin resistance indirectly via increased food intake and body fatness in susceptible people.  To be clear, I still think refined sugar is unhealthy and should be largely avoided.  In a future post, I'll explore the effect of sugar on body fatness in more detail.


KCJerry said...

I have been on the no wheat diet for the past year, lost twenty pounds, my BMI is near perfect now. However my glucose has shot up too 100 from the low 90's during the same period. Is this what you are describing? I have had less sugar this year than any year of my life too. Taste is gone after getting rid of the wheat.

Tam said...

Thanks as always for your thoughtful comments on these issues.

Jenny said...


One problem with studies of insulin resistance in humans is that they rely on the HOMA formula to estimate it (a forumla that uses fasting insulin and blood sugar to compute a measure of IR.) This turns out to be highly misleading if there is any beta cell dysfunction, which, it turns out is the case in most people with the human genes that lead to diabetes (as opposed to the oft studied mouse genes which cause diabetic lab mice which are completely different genes.)

The study that documented this is discussed on my blog HERE.

The observed IR in people on very low carb diets may turn out to have to do with downregulation of the enzymes the liver uses to process dietary carbohydrate and resolves within 2 days. It is entirely different from the kind of IR that stems from nonresponsive cellular insulin receptors.

There are so many different causes of IR that it probably isn't one condition and deducing it by measuring glucose and circulating insulin miss this. IR can be caused by inefficient livers full of liver fat, failing insulin production, mitochondria that don't burn glucose properly, abnormal glucagon secretion (which turns out to be surprisingly common in people who go on to develop Type 2.) etc.

vladex said...

this is just another reductionist diet hypothesis which will as always prove wanting. What I am reading about is that insulin resistance is caused by the brain under stress, psychological,physical and nutritional. One recent study says that the hormone oxytocin can reverse insulin resistance regardless of the diet. Oxytocin is secreted during positive and stimulating social interactions and attachments. The lack of such leads one to be vulnerable to stress and disease.

JBG said...
This comment has been removed by the author.
JBG said...

Minor editing nitpick, Stephan, for future reference.

"I still think refined sugar is unhealthy and should be eaten in moderation."

Most people will get the right meaning, but the (common!) phrasing is not precise. What you mean is that refined sugar should not be eaten *beyond* moderation. You are not intending to suggest that people should eat refined sugar *to* moderation or indeed to eat refined sugar at all.

Stephan Guyenet said...

Hi Jenny,

I agree. We do clamp studies in our lab, so I'm acutely aware of the issue of measuring vs. estimating insulin sensitivity. The study by HP Himsworth, as far as I can tell, assessed insulin sensitivity directly. I'm not sure what method he used, because I can't get my hands on the full text, but that's what I can gather from the papers that referenced it.

The rest of the studies I referenced used a mix of techniques, some direct measures and some estimates such as HOMA. I fully agree that HOMA is just an estimate and it has its pitfalls. If I were to restrict myself only to studies that did clamps, there wouldn't be much to cite, so I compromised and included studies that used less direct methods. You can take these with a grain of salt if you wish.


Thanks, I fixed it.

EL 66K said...

Stephan, can reduced absorption of the fructose at high quantities play a role in all this?

Galina L. said...

I remember reading in the "Good calories, Bad calories" by G. Taubes description of different group of people who went from their native diets to the Western one high in refined carbohydrates. It took years(from 10 up to 20) for people in the group to develop deceases typical for Westerners. It is understandable that it is not easy to conduct 20 years long studies, but if epidemiological date sited by GT is correct, it would be expected to see no rapid changes in health among healthy volunteers on the regiment. Probably, the studies mentioned were too old to be considered and weren't put into consideration for that reason?

LeonRover said...

Hi Stephan,

This is a nice series, thank you.


In Diabetes Care 2 3 :1 7 1–175, 2000, HELEN YENI-KOMSHIAN, and Reaven conclude that

"The total integrated insulin response to oral glucose is the best surrogate measure of insulin resistance, accounting for approximately two-thirds of the variability in insulin-mediated glucose disposal. Fasting insulin concentration accounted for approximately one-third of the variability in insulin-mediated glucose disposal, and the use of fasting plasma glucose and insulin concentrations to calculate more sophisticated estimates of insulin
resistance appears to offer little advantage over the fasting plasma insulin concentration."

Further, in Diabetes Care. 2008 July; 31(7): 1433–1438, Kim & Reaven make the case, which the name of the encapsulates "Insulin Resistance and Hyperinsulinemia - You can't have one without the other".

Under these circumstances, higher than normal fasting insulin levels are the measure of something gone awry. Multiplying FI by BG - HOMA-IR - does nothing to further our understanding of the situation beyond the observed individual values.

The term IR, which to me means the down-regulation of insulin receptors in some tissue, has become devalued, very often as the specific tissue - liver, muscle or fat - is not stated by the writer, and one is left to assume a total body resistance.

I cannot see how those MDs and Endos advising Diabetics about their hyperglycemia and pancreatic insulin functioning are being faithful to the Hippocratic Oath when an OGTT is performed without providing the corresponding insulins. One could easily estimate AUCs for both BG and insulin.

I speak as one who in his past built stimulus-response models and is quite used to the standards of pharmaco-dynamic and pharmaco-kinetic papers.


tfarny said...

Very interesting - but since we know we are talking about pathologies which develop over a span of years or decades in humans, why trust conclusions drawn entirely from short-term studies? Given a smallish sample, the effect sizes you'd observe in the short-term are likely to be far too small to reach p .05 significance if the time frame in the study is a fraction of the time frame required to produce pathology in the real world.

Yves said...

Do you have any plans for talking about the role of intestinal permeability causing low grade endtoxemia, and what may be the causes and solution? I know you touched upon it in another post. You mentioned you did not think high-fat diets were a cause in humans, despite being a predictable cause of increased permeability and unfavorable flora changes in rodents.

Also, Thank you vidicating fructose/fruit! I was getting tired of the fructose-phobia on paleo blogs!

Nyx said...

Hi, I was curious about this and I think I found the Himsworth study you are looking for, or at least it seems to be a separate publication by the same author discussing the same study. See if this doesn't have what you were looking for:

Indrid Cold said...

"However, since added sugar increases the energy density and palatability/reward value of food, it can contribute to insulin resistance indirectly via increased food intake and body fatness in susceptible people"

What makes the difference in susceptible people?

I used to eat nothing but junk food and I got fat. I remained fat even with healthy eating. I got fed up and I hired a consultant .
I lost all of the weight eating junk and don't avoid it at all now.

how could I have been susceptible before but now not?

Stephan Guyenet said...

Hi LeonRover,

Thanks for your perspective. Commonly, "insulin resistance" without any qualifiers is used to refer to an overall impairment of the ability of the body to suppress circulating blood glucose in response to insulin. Personally I think that's an acceptable use of the phrase.

Hi Yves,

I think it's probably important, but it's hard to know what factors influence it. The best I can say at this point is it's probably a good idea to feed the intestinal flora with an appropriate amount of various fermentable fibers.

Hi Nyx,

Awesome, thanks for passing that along. It's not the original paper but he does describe his methods. There's some other good stuff in that PDF as well.

It looks like Himsworth used an insulin tolerance test to estimate insulin sensitivity. That involves injecting a standard dose of insulin and measuring how much it depresses circulating glucose. It's not quite as good as a clamp, but it is a direct measure of insulin sensitivity rather than an estimate, so it's a pretty good measure IMO.

Nyx said...

I'm really glad that I was able to help in some small way. I really appreciate all the information you share on your blog, although I am still struggling to understand it all. It is very generous of you to spend so much time and energy to share these things with the world.

kyle said...


What are your thoughts on the role of omega-6 pufa consumption and insulin resistance?

In the post on inflammation, you mention aspirin and salicylic acid's ability to improve insulin sensitivity. Wouldn't a diet low in omega-6 result in improved insulin sensitivity as well? As there would be less substrate for prostaglandin formation. I realize inflammation is a general term covering many pathways, but I suspect that omega-6 consumption plays a major role in insulin resistance.

Cliff said...


"These data indicate that prolonged exposure to saturated fat enhances GSIS(glucose stimulated insulin secretion) (but this does not entirely compensate for insulin resistance), whereas unsaturated fat, given in the diet or by infusion, impairs GSIS."

"In summary, dietary fish oil supplementation adversely affected glycemic control in NIDDM subjects without producing significant beneficial effects on plasma lipids. The effect of safflower oil supplementation was not significantly different from fish oil"

"After omega-3 fatty acid withdrawal, fasting glucose returned to baseline. Omega-3 fatty acid treatment in type II diabetes leads to rapid but reversible metabolic deterioration, with elevated basal hepatic glucose output and impaired insulin secretion but unchanged glucose disposal rates. Caution should be used when recommending omega-3 fatty acids in type II diabetic persons."

"Fasting blood glucose and insulin levels were significantly higher on the linoleic acid diet compared with the oleic acid diet (P < 0.01 and P < 0.002, respectively). Plasma cholesterol and LDL cholesterol levels were also significantly higher on the linoleic acid diet (P < 0.001). Likewise, fasting chylomicron apo B48 and apo B100 (P < 0.05) and postprandial chylomicron and VLDL apo B48 and B100 (P < 0.05) were also higher on the linoleic acid diet. "

"The FA composition of serum phospholipids (S-PL) measured by gas liquid chromatography and insulin action during a 2-step hyperinsulinemic isoglycemic clamp (1 and 10 mU/kg. min) were determined in 21 newly diagnosed DM2(diabetic) subjects (DMN), in groups of long-term DM2 patients treated with hypoglycemic agents (DMH; n = 21) or diet alone (DMD; n = 11), and in 24 healthy subjects (HS)....Increased contents of highly unsaturated n-6 family FA (P <.01), arachidonic acid in particular, were found in all groups of diabetics compared with HS."

"The blood glucose concentration tended to increase during MaxEPA treatment, and to decrease during the placebo period, the changes under the two regimes being significantly different (P less than 0.01). In addition, the rate constant for glucose disappearance (k value) for the intravenous insulin-tolerance test, which reflected the peripheral insulin sensitivity, tended to decrease during MaxEPA treatment and increase during administration of the placebo, there being a significant difference (P less than 0.03) between the changes during the two treatments."

Nyx said...

Here is something I'm confused about, I don't know if you can clear this up or not ... my sons have a lot of digestive problems, which has led me to read a lot of stuff on small bowel bacterial overgrowth. Based on the stuff I've read, I was under the distinct impression that even normal people have a very limited ability to digest fructose unless glucose is present. that in order to absorb a fructose molecule, it has to be carried across the intestinal wall (or something like that) by a glucose molecule. Am I wrong about that? How could they test the impact of fructose based purely on the amount that a person ingested, without controlling for simultaneous glucose ingestion, etc.?

Robert Andrew Brown said...

@ Cliff

Thanks for the thought proving references which I am mulling

A riposte (-:

"Design: This was a prospective population-based cohort study in 51,963 men and 64,193 women free of T2D, cardiovascular disease, and cancer at baseline with valid dietary information. Dietary intake, physical activity, and anthropometric measurements were collected. A Cox regression model was used to evaluate the association of fish, shellfish, and long-chain n−3 fatty acid (in g/d) with risk of T2D.

Results: Fish, shellfish, and long-chain n−3 fatty acid intakes were inversely associated with T2D in women. The relative risks [RRs (95% CI)] for quintiles of fish intake were 1.00, 0.96 (0.86, 1.06), 0.84 (0.75, 0.94), 0.80 (0.71, 0.90), and 0.89 (0.78, 1.01) (P for trend = 0.003) and for shellfish were 1.00, 0.91 (0.82, 1.01), 0.79 (0.71, 0.89), 0.80 (0.71, 0.91), and 0.86 (0.76, 0.99) (P for trend = 0.006). In men, only the association between shellfish intake and T2D was significant. The RRs (95% CI) for quintiles of fish intake were 1.00, 0.92 (0.75, 1.13), 0.80 (0.65, 1.00), 0.89 (0.72, 1.11), and 0.94 (0.74, 1.17) (P for trend = 0.50) and for shellfish intake were 1.00, 0.93 (0.76, 1.12), 0.70 (0.56, 086), 0.66 (0.53, 0.82), and 0.82 (0.65, 1.02) (P for trend = 0.003). "

Robert Andrew Brown said...

@ Cliff

"We prospectively analyzed data in 3088 older men and women (mean age: 75 y) from the Cardiovascular Health Study (1992–2007). Plasma phospholipid n−3 FAs were measured by using gas chromatography, and incident diabetes was ascertained by using information on hypoglycemic agents and serum glucose.

With the use of objective biomarkers, long-chain n−3 FAs and ALA were not associated with a higher incidence of diabetes. Individuals with the highest concentrations of both types of FAs had lower risk of diabetes.

Nyx said...

It looks like the studies that failed to replicate the Himsworth findings did not actually raise the carbs to the 70 to 80% level. I really wonder whether anyone has faithfully adhered to this "superhigh" carb diet and not found any impact? I found another article by Hugh Trowell that reported huge improvements for diabetics who were given a superhigh carb / superlow diet that was also high in fiber. Meanwhile, I found some other reference to a more recent study that claimed that it had found that a high carb low fat diet didn't help, but again when you looked at it, the ratio was nowhere near 70/80 for carbs and 10 or under 10 for fat. I wonder what is up with that?

Don S said...

Any comment on the role that insoluble fiber has on increasing insulin sensitivity? (As in ) and others.

Or on the recent 1/6 Science article by Taubes reviewing the putative link of higher insulin levels to cancers?

Stephan Guyenet said...

Hi Don,

Yes, fiber improves insulin sensitivity probably via its effects on the gut flora (which may in turn act on inflammatory pathways). Fiber consumption will be part of my recommendations for avoiding insulin resistance when I post them tonight.

Regarding Taubes's article, it was fine besides his exaggeration of the effect of obesity on overall cancer risk. He wrote "Obese and diabetic individuals have a far higher risk than lean healthy people of getting cancer", but this is quite a stretch.

Obesity accounts for ~4-7% of cancers overall, which is meaningful but not exactly overwhelming. If you look at specific cancers, the increase in risk can be up to ~40%, but it's not clear that insulin is to blame (it is definitely plausible though, given the recent findings he reviewed).

The article was fine, he was just highlighting some recent findings in the literature. Fortunately he didn't try to make any tall claims about carbs or sugar-- I'm sure the editors would have smacked it down if he did.