Thirty Years of Glycemic Index Dogma, and The American GI Man

Sixteen post bumps for 2016 ... No. 3

I think this post from around a year ago, in light of the new book by David Ludwig is a necessary bump.

It's not every diet guru that is a principal investigator on a "groundbreaking" study that directly debunks the thesis of their book.  In this regard, Ludwig is, if not one of a kind, the leader of the pack.

---

---

Originally published on January 19, 2015

The former British colonies/territories may be known for a lot of things, but in the nutrition world, I'm going to go out on a limb here and state that the Glycemic Index - GI - is the trifecta.

In Canada, David Jenkins pioneered the concept, for applications in diabetes treatment.  This cause was picked up and concurrently championed in Australia by Janette (Jenny) Brand-Miller.  A little late to the party, in one of the original Thirteen, David Ludwig took up the cause in the USofA.  Are these the men and woman who made us fat?

Adapted from Adele Hite's Eathropology
image link

Yes, that's pretty silly ... but no more so than Adele Hite and others like her, who in the same breath decry the error in equating correlation with causation as they point to plots like the one above to indict the low fat diet.

USEFUL??

Let's be honest here, can we?  Even worse than the Big Fat Lie that Americans have actually cut fat from their diets, is the idea that in the process of cutting fat, we've replaced it with refined carbohydrates at the behest of government and health organizations.   Whether or not the term "glycemic index" was used, the notion of "complex carbohydrates" has long been in the nutrition and weight loss lexicon.   The Snackwell phenomenon was not the fault of USDA guidelines, and folks who gained weight eating them by the sleeve did so because they ate them by the sleeve.  True, the glycemic index never reached the same level of universal "official" acceptance, but if I still had any of the women's rags from the 1980s and 90s lying around, you would find far more references to fiber/complex carbs than to saturated fats in them.   And there was no fat deficiency for most, as low fat still means plenty of the stuff.  Indeed most may recall the labeling changes that had to be instituted to where the term "low fat" had to actually mean below a certain amount in a serving, water weight was no longer allowed to count when advertising a % fat free, etc.

image link

The food pyramid is filled with the spirit of the low glycemic index with its whole grains, fruits and vegetables.   Ludwig and other hangers-on to the GI contend that low fat has meant a turn to high glycemic refined carbs.  Even more dishonest than this, is any ridiculous belief that french fries and Snickers bars would be metabolically healthier or less obesogenic food choices than a baked potato or shredded wheat.

Thirty Years Later ...

Last month, the following appeared in the New York Times:  Questioning the Idea of Good Carbs, Bad Carbs.  It was prompted by a study published in JAMA:   Effects of High vs Low Glycemic Index of Dietary Carbohydrate on Cardiovascular Disease Risk Factors and Insulin Sensitivity, and here's the JAMA Press Release: Low-Glycemic Index Carbohydrate Diet Does Not Improve Cardiovascular Risk Factors, Insulin Resistance.  The lead author is Frank Sacks ... nemesis to low carb advocates for the 2009 diet comparison study showing no difference between macro content of diets in terms of weight loss etc.  

The current study compared 4 diets in a randomized cross-over fashion of 5 week exposure to each diet.  Not all subjects completed all diets, but each subject completed at least 2, and each pair of diets compared had from 135-150 subjects consuming each of the two diets.   The outcomes were insulin sensitivity (by AUC of glucose and insulin on an OGTT), LDL, HDL, triglycerides and systolic blood pressure.  All diets were DASH type foods and were either high carb (58%)  with high or low GI, or lower carb (40%) with high or low GI.

Results  At high dietary carbohydrate content, the low– compared with high–glycemic index level decreased insulin sensitivity from 8.9 to 7.1 units (−20%, P = .002); increased LDL cholesterol from 139 to 147 mg/dL (6%, P ≤ .001); and did not affect levels of HDL cholesterol, triglycerides, or blood pressure. At low carbohydrate content, the low– compared with high–glycemic index level did not affect the outcomes except for decreasing triglycerides from 91 to 86 mg/dL (−5%, P = .02). In the primary diet contrast, the low–glycemic index, low-carbohydrate diet, compared with the high–glycemic index, high-carbohydrate diet, did not affect insulin sensitivity, systolic blood pressure, LDL cholesterol, or HDL cholesterol but did lower triglycerides from 111 to 86 mg/dL (−23%, P ≤ .001).

Conclusions and Relevance  In this 5-week controlled feeding study, diets with low glycemic index of dietary carbohydrate, compared with high glycemic index of dietary carbohydrate, did not result in improvements in insulin sensitivity, lipid levels, or systolic blood pressure. In the context of an overall DASH-type diet, using glycemic index to select specific foods may not improve cardiovascular risk factors or insulin resistance.

The purpose of this post is not to discuss this study in detail, but rather the reporting, etc.  To me, the third decade of glycemic index research and advocacy is instructive for another reason.  It appears to be a textbook case of scientists -- presumably well meaning -- getting too invested in their hypothesis to see the writing on the wall.   On the one hand, 

“The takeaway is a good message for people.  They can pick foods that are part of a healthy dietary pattern without wondering if they’re high or low glycemic. They don’t have to learn that system.”

Dr. Frank M. Sacks 

"The dogma out there is that a high glycemic index is bad.  I hope that ultimately the glycemic index will be left on the shelf.”

Dr. Robert H Eckel

On the other hand,

{paraphrase} ... short term clinical trials of increased fiber and whole grain intake often fail to show any benefit.  But public health recommendations are based on longterm trials and high quality observational studies. And unlike very low carbohydrate diets, which show quick results, low glycemic diets may require more time ...  “A low G.I. diet may be more like the tortoise than the hare.  It takes longer, but gets you there in the end.”

Dr. David S. Ludwig

From the get-go, the GI is fundamentally flawed.  Even its close relative, the glycemic load, GL cannot rescue it.  

• Single foods are almost never consumed in isolation
• Standardizing to mass or carbohydrate content each brings issues into play as to the relevance of the comparison
• Some "single foods" are actually combinations,  e.g. french fries = potatoes + oil, and those foods can differ in GI from the combination

Further complicating matters, many plain foods can vary in GI by reasonable variation in prep (sliced, mashed), cooking time, cooking method (baked vs. boiled) and even shape (pasta!).   Sacks' study is yet another in which the GI failed to pan out experimentally.  Ludwig sounds like an advocate, not a scientist here.  It's not like he doesn't have his own ample body of research that should lead him to much the same conclusion.  

Twenty Years of The Glycemic Index - A Dozen Years Ago, 2002

The mechanism or hypothesis proposed, seems plausible.  The graphic below from a website promising "The Body of Your Dreams" pretty much sums it up:

original source

And so for 20 years, scientists went about testing the hypothesis ... or did they?  In 2002, a 20 Year "summit" regarding the utility of the GI was held and a special journal issue put out:   The glycemic index at 20 y.  The authors of that summary?  Eckel and Ludwig.  

During the past 20 y, > 100 scientific studies have examined the application of the GI or GL to diabetes mellitus, obesity, cardiovascular disease, behavioral disorders, and physical performance (8–,10). Several popular nutrition books, with combined sales of several million copies, advocate the consumption of low-GI diets (11–,13). Moreover, use of the GI has been endorsed by the FAO/WHO (14) and numerous other international health-related organizations.

However, there is by no means a consensus regarding the utility of the GI to human health and nutrition. Many clinicians and researchers, especially in the United States, have questioned the relevance and practicality of the GI (15,16). The topic has sparked controversy in a variety of national scientific settings, including meetings held by the Department of Agriculture (17) and the National Academy of Sciences (18). At present, neither the American Diabetes Association (19), the American Heart Association (20), nor the American Dietetic Association (21) recognize a role for GI in disease prevention or treatment.

I'll get to a few papers from 2002 in just a bit here, but Eckel and Ludwig's concluding paragraph is interesting:

The postprandial effects of carbohydrate-dense, high-GI foods described above may help to explain why low-fat diets have not lived up to their potential to inhibit weight gain when consumed ad libitum.

Note the "may"

Postprandial hyperglycemia and hyperinsulinemia are consequences of typical Western diets ...

Sounds a bit hyperbolic, but OK

... that are not seen in rural societies where high-carbohydrate, low-GI diets are typically consumed.

Except for potatoes and rice (yes, even the white sort).  I note, however, that the implication here is that there is not an obesity problem in such rural societies which was generally true in most cases.

This is especially true of persons who are sedentary, overweight, and genetically prone, in whom insulin resistance is common. Faster digestion and absorption and higher insulin responses after high-GI meals [might] dictate differences in satiety and energy partitioning that, over the long term, favor expansion of the fat stores.

This sounds awfully definitive given the preceding paragraphs and the whole of the evidence available in 2002, and really should have had the [might] in there.  But, here's the clincher!

Carefully designed multicenter studies to assess the efficacy of high-carbohydrate, low-GI diets in the treatment and prevention of obesity are clearly justifiable on scientific grounds.

Translation:  The failure of the evidence to date to clearly support the hypothesis supports the need for more studies in hopes of supporting the hypothesis.  Reopeat:

{paraphrase} ... short term clinical trials of increased fiber and whole grain intake often fail to show any benefit.  But public health recommendations are based on longterm trials and high quality observational studies. And unlike very low carbohydrate diets, which show quick results, low glycemic diets may require more time ...  “A low G.I. diet may be more like the tortoise than the hare.  It takes longer, but gets you there in the end.”

Dr. David S. Ludwig, 2014

What happened to this/these trial(s)?   The one(s) specifically to answer the question of whether ad libitum intake of a high carb high GI diet paired vs. a high carb low GI diet would result in differences in weight (loss or gain) over at least a one year, preferably two year (or more, heck a girl can dream) period.  The study or studies to demonstrate that the proposed blood sugar roller coaster impacted satiety and intake such as to result in weight gain or loss irrespective of protein or fat intake. Apparently, Ludwig's excuse for the "inconclusive" nature of the 2014 Sacks study is that the tortoise still hasn't been put up to the test  --despite his 15 year plus track record of continuous NIH funding no less!!!

What We Knew in Two Thousand Two ...

Let's look at a few 2002 round ups, shall we?  I'll begin with a pair of competingly-titled reviews/opinion pieces:

1.  Should obese patients be counselled to follow a low-glycaemic index diet? Yes.   Pawlak, Ebbeling & Ludwig

A reduction in dietary fat has been widely advocated for the prevention and treatment of obesity and related complications. However, the efficacy of low-fat diets has been questioned in recent years. One potential adverse effect of reduced dietary fat is a compensatory increase in the consumption of high glycaemic index (GI) carbohydrate, principally refined starchy foods and concentrated sugar. Such foods can be rapidly digested or transformed into glucose, causing a large increase in post-prandial blood glucose and insulin. Short-term feeding studies have generally found an inverse association between GI and satiety. Medium-term clinical trials have found less weight loss on high GI or high glycaemic load diets compared to low GI or low glycaemic load diets. Epidemiological analyses link GI to multiple cardiovascular disease risk factors and to the development of cardiovascular disease and type 2 diabetes. Physiologically orientated studies in humans and animal models provide support for a role of GI in disease prevention and treatment. This review examines the mechanisms underlying the potential benefits of a low GI diet, and whether such diets should be recommended in the clinical setting.

I have to comment on that bolded part, because it is an unscientific mantra that has been repeated as if fact for far too long.  When a person goes on a low fat diet to lose weight, they remove some fat.  They do not replace it with sugar and three slices of Wonder bread for breakfast.  They are not counseled to increase their consumption of Twinkies.  I have a copy of this review but had devoted considerable time to looking into the references of a JAMA article by Ludwig:  The glycemic index: physiological mechanisms relating to obesity, diabetes, and cardiovascular disease (also 2002).   I have another blog post in the works on the mechanisms posited by Ludwig as fact.  [EDIT to add links:   The Mechanisms of the Glycemic Index: A Fatty Acid Roller Coaster?, More on the Mechanisms of the Glycemic Index: A Fatty Acid Roller Coaster?]   However a brief scan of the references in the Yes article, I do not see those put forth in the JAMA review as regards metabolic health, although similar claims are put forth regarding lipids, etc.  In JAMA, Ludwig cites a 1994 meta-analysis of sorts by Brand-Miller Importance of glycemic index in diabetes .  I've included the summary table for the 11 studies Brand-Miller refers to as "medium to long-term" in duration.

(Number of studies included in averages is shown below the average line)

I think this is self explanatory as regards the use of "medium to long".  To emphasize the missing values and insignificant differences, I've used gray squares to denote measured but not significant, and deleted all ND = not determined entirely.   Lots of holes even in these mostly short term studies.  Because the first Jenkins study was also cited in Ludwig JAMA as a separate citation, I took a look at that one in more detail.  You'll notice the -15% for cholesterol which implies a reasonably favorable difference.  Here is the augmented table for that.

As you can see, the high GI group had a statistically significant reduction in LDL and a non-statistically significant rise (improvement) in HDL that offset to flatline the TC response.  For the low GI group, a non-statistically significant reduction in  both LDL and HDL combined to produce a statistically significant reduction in TC.  These details are obscured by the summary table in Brand-Miller and falsely imply that the low-GI response was favorable to that of the high-GI which is simply not true.  Both produced comparable reductions in LDL while the increase of HDL from 39 to 49 is a clinically significant improvement seen with the high GI diet.   I would add that it is also inappropriate to extrapolate the utility of GI in treatment of diabetes (see title of Brand-Miller study!) to the subject of weight control in the general population.

But this didn't seem to dissuade the American GI Man!  No doubt, he was privy to a counter argument put forth ...

2.  Should obese patients be counselled to follow a low-glycaemic index diet? No.  Anne Raben

ABSTRACT:  In diabetes research the glycaemic index (GI) of carbohydrates has long been recognized and a low GI is recommended. The same is now often the case in lipid research. Recently, a new debate has arisen around whether a low-GI diet should also be advocated for appetite- and long-term body weight control. A systematic review was performed of published human intervention studies comparing the effects of high- and low-GI foods or diets on appetite, food intake, energy expenditure and body weight. In a total of 31 short-term studies (< 1 d), low-GI foods were associated with greater satiety or reduced hunger in 15 studies, whereas reduced satiety or no differences were seen in 16 other studies. Low-GI foods reduced ad libitum food intake in seven studies, but not in eight other studies. In 20 longer-term studies (< 6 months), a weight loss on a low-GI diet was seen in four and on a high-GI diet in two, with no difference recorded in 14. The average weight loss was 1.5 kg on a low-GI diet and 1.6 kg on a high-GI diet. To conclude, there is no evidence at present that low-GI foods are superior to high-GI foods in regard to long-term body weight control. However, the ideal long-term study where ad libitum intake and fluctuations in body weight are permitted, and the diets are similar in all aspects except GI, has not yet been performed.

And that, the bolded statement, right there, is the crux of the matter.  Why wasn't it done already BY 2002?  After all, many of the "supporting documents" were around a decade old or more when the GI world did their 20 year "gut check".

The Lost Glycemic Decade(s)?

My discussion here is regarding the low GI diet and advocacy of this paradigm.  I am well aware that analogies can be drawn with some of the USDA/government recommendations, etc.  I cannot rehash all of that, and in the end, any "they did it too" (valid or not) arguments do not absolve scientists working on the GI from their missteps.

The GI began with a plausible mechanism of the "blood sugar rollercoaster".  It was codified by evaluating certain foods for this property in the isolated food, single "meal", context over the course of several hours.  There were already some problems here, as mentioned previously.  But as this was a novel concept, and it was developed by these scientists as a somewhat subjective measure, it is up to the scientists to validate their concept and demonstrate how this evaluation of a food quality extrapolates to a real-life mixed-meal context over the short term.  If successful there, it is further incumbent upon the researchers to test their hypothesis over the long term.  Long term meaning at least six months, preferably a year or two.  The mechanism ...

High GI Meal

↓

Exaggerated Glucose & Insulin Spike

↓

Reactive Hypoglycemia

↓

Hunger

↓

Increased Intake

... however plausible, has only been partially substantiated in some short term (hours) studies on single foods or isolated meals.  Efforts to translate this to the longer term and document the mechanism have either failed, or, worse yet, never even been attempted.   There's also that troubling inconsistency of the satiety index wildly favoring potatoes, which have a high glycemic index.  Not to mention that I know of no epidemiologic study linking potatoes with obesity or diabetes.

In 1990, Brand was one of the first to look into the trials of the Pima Indians.  What was it about their food, perhaps, that had caused the exploding rates of obesity and diabetes in modern Pima populations?  

Plasma glucose and insulin responses to traditional Pima Indian meals

... Their diets today are considerably different from those before 1930, which were dominated by wild and cultivated desert legumes, with cacti, fish, and small seeds as supplements. Mesquite, corn, and legumes together provided -40-50% of food intake by weight. ...

...At one time the Pima and Papago Indians may have consumed more legumes per capita per day than any other ethnic population in the world. ...  

Ya don't say ;-)  Please tell me again about the psychological and metabolic damage that legumes cause oh mighty Hartwig/Wolf/Sisson/Cordain etc.!!  So the study assessed starch availability for several traditional Pima foods (in vitro) and the glucose and insulin responses to meals standardized to 25 grams of available carbohydrate.  The variation was quite great, from 125 grams hominy (with ~ 5 grams fiber and protein and under 3 grams fat) to 454 grams acorns (with ~ 30 grams protein and ~ 80 grams fat and fiber) -- e.g. hardly a fair comparison -- and there was a variation of GI from 40 for corn down to 16 for acorns, and Insulin Index of 53 for corn to 19 for acorns.   The study is a rather intensive look at specific foods and the hormonal response to a meal.  From this, and the observational evidence of the change in weight and health of the Pima on a diet of these foods vs. the SAD.  So the discussion turned to the slow-release nature of the carbohydrate in traditional Pima foods, and the mucilaginous fiber content.  This same fiber that protects desert plants from drought was posited to protect the Pima from their genetic disposition towards diabetes.  Fair enough and plausible hypothesis.

To test these hypotheses, the low GI foods, which formerly dominated the diets of indigenous desert peoples, must be shown to improve β-cell function and/or insulin sensitivity over the long term in individuals predisposed to diabetes, as well as to show digestion and absorption. Short-term consumption of legumes and other low GI foods by Caucasian diabetics has resulted in improved glucose tolerance and glycemic control. Similar studies comparing traditional desert foods and modern processed foods need to be carried out in healthy and prediabetic Pima Indian subjects.

Janette C. Brand, 1990 

Leaving aside the Pima for a moment ... where have these studies been all these years?   If the "movers and shakers" in the low GI camp aren't designing and carrying out these experiments, who will?  More importantly, at what point do these researchers get called to account and challenged on their repeated assertions?

It is not enough to keep calling for further study.  And after 20 years, Ludwig wrote (repeat from above) 

Carefully designed multicenter studies to assess the efficacy of high-carbohydrate, low-GI diets in the treatment and prevention of obesity are clearly justifiable on scientific grounds.

Ludwig, and countless others have done research on the GI in the intervening dozen years.  Few would count towards the realization of this call to action.  Some by Ludwig himself.   Was it a matter of funding?  I tend to doubt that.  Readers here are all familiar with a number of diet comparison studies -- from Shai to Gardner to Dansinger and on down the line.  There were certainly opportunities along the way to include a low GI diet in the mix (and arguably it has without the specific name, e.g. Zone can be considered low GI).  I don't have the time to do any sort of comprehensive search for other studies ... again, the onus is on those trying to substantiate and support their hypothesis.  Not me.  

Still, Google Scholar didn't turn up much of anything.  There was an 18 month study in Brazil by Walter Willet's group.  Strange one -- "sticky rice versus parboiled rice was one of the major determinants of the difference in GI between the 2 diets, beyond the amount of beans" -- where the low GI dieters managed less fiber than the high GI ones (among other anomalies).  But ultimately:

Long-term weight changes were not significantly different between the HGI and LGI diet groups; therefore, this study does not support a benefit of an LGI diet for weight control. Favorable changes in lipids confirmed previous results.

And there was a study where over 800 subjects lost at least 8% of body weight (avg. 11 kg = 24 lbs) on an ultra low 800 kcal/day diet.  They were then randomized to one of five ad libitum maintenance diets for 6 months: the 4 variations of high/low GI and high/low protein, and a control.

 I'll let the weight changes from this table (complete) speak for themselves after stressing we are talking 6 months ad libitum after weight loss on a semi-starvation diet.  I don't think any further analysis is really needed.  Their conclusion:

In this large European study, a modest increase in protein content and a modest reduction in the glycemic index led to an improvement in study completion and maintenance of weight loss.

Fewer of the higher protein and lower GI participants dropped out (~26 vs 37%), the high GI.  Other than that, the results here hardly support some change in energy partitioning based on different glucose and insulin responses.  Unless I'm reading this wrong, all continued to lose some fat mass, while the high glycemic groups put on the most fat-free mass.

There are likely many more studies.  Feel free to include links (and preferably a synopsis of design and outcome) etc. in comments.

Ludwig & The Glycemic Index:  2006, 2012 and Beyond

Yes ... you've probably guessed that, as usual, I'm going somewhere with this post.  You see, a couple of years ago, Ludwig, with colleague Cara Ebbeling, published a widely circulated and hyped study:  Effects of Dietary Composition on Energy Expenditure During Weight-Loss Maintenance.  The problem for the GI crowd, is that even if one is to accept the results of LC vs. LF as in any way definitive of some metabolic advantage, etc., the "middle of the road" low glycemic index arm was unimpressive -- see Table 3.  Sometimes there was no difference, sometimes LGI was like LC, sometimes like LF.  Unremarkable.  And then there was another publication, which focused on the supposed mechanism of low GI,  from this same clinical trial, that went all but ignored.   It too, was a bust for the proposed mechanism for GI.

This clinical trial, was first registered in April of 2006.  This means that many of the diet comparison trials that may have soured investigators on the utility of low GI had not yet been conducted.  WHY were the primary outcomes -- listed presumably in order of importance, and indicated by the title of the primary publication -- energy expenditure followed by IR and thyroid?  Why not insulin secretion, glucose exposure, satiety, etc.??  All the stuff of the Low-GI Dogma?  Nobody can know for sure what they were thinking, but the fact that America's GI Man wasn't championing that definitive study that all had been calling for, while apparently still hanging onto the dream with public statements in 2014, seriously calls his scientific chops into question.

The trial ended in 2010.  And while all the statistical machinations took two years to complete, please do not tell me that the general outcome was unknown to the researchers upon cursory review of the raw data.  According to their revelations, Taubes and Attia began to discuss things with Ludwig following the results of this "groundbreaking" study in 2012.  Whatever else one might think of that relationship, why is the Low-GI arm still part of the upcoming NuSI funded study?   You cannot compare LGI to LF there -- it's not fixed carbohydrate with LGI vs. HGI type.  No, it's 20% protein with LGI of 40% each carb and fat, while LF is 20% fat and 60% carb.  

According to their website, the NuSI funded "repeat" of this study got underway last September.  The cost?  Oh ... around 14 million.  If the timeline is anything like the 2012 publication, we could be looking at 2020 before we know anything.  NuSI's mission is ostensibly to answer the questions left unanswered by crappy nutritional science research to date.  And yet, this study will do nothing to answer the burning questions of over a dozen years ago, or go anywhere to substantiate the Low GI (refined carb!) hypothesis proposed over 30 years ago. 

It seems like Ludwig has changed nutritional teams in his advocacy.  He's also apparently ditched the blood sugar roller coaster for the metabolic advantage schtick.  Time will tell how that works out for him, NuSI and nutritional science. 

Stay tuned ... I hope to get the follow up out there soon!!