The Glycemic Index, Carbs & Protein, and Weight Loss
Hey all ... just a quick note. It's been one of those weeks and all where I've started like a half a dozen blog posts and gotten distracted or needed to do some "cleanup" on them so I've held them in the draft bin. Lots going on in real life in addition to the online side. But as often happens when I'm writing about a topic, a related paper pops up and something catches my eye. Such was the case when I looked into some of Dr. David Ludwig's work on the glycemic index for background on a post, and another study was mentioned along with his work.
Studies such as this one from Ludwig -- High Glycemic Index Foods, Overeating, and Obesity -- and, frankly, common sense to a degree -- paint high glycemic index foods as "bad" because:
The rapid absorption of glucose after consumption of high-GI meals induces a sequence of hormonal and metabolic changes that promote excessive food intake in obese subjects.
And this makes sense, and it actually does happen in a subset of the population. Eat high-GI carbs, blood sugar spikes, insulin spikes (probably slightly delayed as the beginning of "IR"), blood sugar plummets below baseline and hunger ensues.
Which is why I found this next study so intriguing:
Comparison of 4 Diets of Varying Glycemic Load on Weight Loss and Cardiovascular Risk Reduction in Overweight and Obese Young Adults
- 129 subjects, 98 women, 31 men
- Overweight or obese BMI ≥ 25 but weight less than 150 kg
- Age 18 to 40
- Willing to eat red meat and maintain current activity level
- Exclusions: chronic illness, regular medication other than birth control pills, eating disorders, special diets, pregnancy, food allergy, and insufficient command of the English language.
Subjects were randomized (stratified for weight and gender) to diet assignments which "resulted in 4 well-matched groups". There were four study diets, all of which were reduced fat with moderate fiber content. They call them Diets 1 through 4, I'll add an acronym designation to keep track: The usual H/L for high/low, C= carb, G= glycemic:
- Diet 1 = HCHG 55%C / 30%F / 15%P
- Diet 2 = HCLG 55%C / 30%F / 15%P
- Diet 3 = HPHG 45%C / 30%F / 25%P
- Diet 4 = HPLG 45%C / 30%F / 25%P
These were reduced calorie plans (1400 women/1900 men) and "to maximize compliance, all key CHO and protein foods and some preprepared meals were provided." The intervention lasted 12 weeks.
As with all such studies, actual consumption missed some targets, but not too badly.
Here are the results:
It is interesting that roughly two-thirds of HPHG lost over 5% of body weight with the HCLG group not far behind, while only around one-third of the HCHG or HPLG group lost this amount of weight. Another way to look at this is that when you have normal (15%) protein, glycemic index seems to matter and favor the low end, but when you have high protein (25%), the higher glycemic index carbs are favored. *** Speculation: The folks that lost the most weight were the ones that adhered to the diet. Both lower GI and higher protein have been shown to be satiating. Those would be the diets in the middle there with roughly equivalent fat mass loss. Perhaps, however, when you have one you cannot have the other? In other words if you have protein for satiety, it is better to have the glucose spike and vice versa. As I said ... speculation. What is troubling is that the HPHG group thus lost more lean mass en route to losing the most weight.
Sex influenced fat mass changes, with a significant interaction with diet (P=.008). In a subanalysis of women (n=98), there were highly significant mean ± SE differences in fat mass change (−2.5±0.5 kg, −4.5±0.5 kg, −4.6±0.5 kg, and −2.9±0.5 kg for diets 1, 2, 3, and 4, respectively; P=.007)
In just the women, the fat masses were again greater for the middle two diets. Again, these were the two diets where you had almost double the number of "successes" (over 5% initial body weight lost) vs. the other two. Adherence is likely the factor here.
Oh but insulin!!
For some reason, they did a 10 hour profile in just 11 of the women who completed this study (average age ~25, BMI ~30). As per the caption, they consumed three meals and a snack per the caption of the diet. Each woman consumed each diet in randomly assigned order, on separate days, with 3-7 days between tests. It is important to point out that these are the same 11 women, consuming each diet in a supervised setting over a 10 hour period.
The greatest differences in both the insulin and glucose responses to the diets were seen with breakfast, with the least difference at dinner. I would also note that there appear to be no significant differences in the insulin levels between the meals. By that I mean if you look 2 hours post breakfast, they are roughly the same despite insulin spiking to over 7X baseline with the HCHG meal. We see the same at 5 hours and from 6 hours on through dinner at 9 hours (I note the time scale is not linear and is stretched after each meal). Insulin levels returned to baseline levels between meals with all diets.
Despite the significantly greater post-meal glucose and insulin spikes with the HCHG diet, there was no effect of either GI or protein on baseline insulin levels, thus no evidence of chronic hyperinsulinemia compared with the other diets.
The 10-hour insulin exposure (the AUC) was not correlated with the weight and/or fat loss response to the diets in the main study. Comparing the HCHG to the HPLG (Jonathan Bailor's diet) there is roughly double the insulin response to the meals, but comparable weight and fat loss, as well as percentages of subjects who succeeded in losing over 5% initial body weight.
Furthermore (scan down to my table below), fasting insulin levels decreased with ALL diets by 10% to 17%, as did FBG. Thus there were improvements in all HOMA scores but none rose to a level of significance between diets.
A high carbohydrate, high glycemic index, calorie restricted diet (free-living) resulted in the least weight and fat loss, but this effect cannot be attributed to "starving cells" and "lipophilia" brought on by the diet.
Here is where things get a bit interesting with this plot. If both HCLG and HPHG produce similar weight and fat loss along with comparable insulin exposures and glycemia, are they equivalent in all respects? The plot at right would appear not. See Ev? The quality of those calories matters so get with the program! [/snark: I've never said it didn't matter].
But first, I'd like to preface this discussion with a bit of a surprise on my part. The LoBAG diets, the Ebbeling study (used a 25% protein diet) and basically every paleo clinical trial are just a few that come to mind that have not shown this effect though we don't always have GI info, many of those end up essentially swapping carb for protein. And yet for total cholesterol and "bad" LDL-C it sure looks pretty bad. I constructed the table (pardon the shoddy formatting, etc.) from the data given to put it in context to baseline. I also want to make clear that I left out standard deviations, etc., and to the best of my knowledge any apparent differences at baseline did not rise to the level of statistical significance.
The lipids are also reported in millimolar and I didn't have time to convert them all, but for the "best case" (HCLG) vs. "worst case" (HPHG) it is noteworthy that the latter began the study being slightly worse off lipid-wise. Converting to mg/dL we have:
HPHG: TC from 199 to 208 , LDL from 129 to 139 , Trigs from 125 to 109
HCLG: TC from 182 to 175 , LDL from 112 to 106 , Trigs from 115 to 111
I think this serves as more of a cautionary tale than anything else, but looking at the bottom rows of my table, the percentage changes were not that considerable. The combination of best weight loss and worst lipids is intriguing and yet it can't be blamed on the protein or animal protein per se as the trends were reversed pairing high protein with low GI.
A Final Note on Glycemic Index:
I consider this concept to be a generally disproven one on a population basis. Moreso than any other dietary manipulation, merely changing the GI does not appear to have much effect in RCT's (e.g. recommending lower GI -- I consider differences in this study to be due to shorter time frame and the fact that foods were specifically provided -- in most comparison studies lower GI doesn't do much if anything). However, I do believe this is because GI is one of those things that is highly individual. Some are insulin hypersecreters, some have a degree of beta-cell dysfunction and/or hepatic IR. The list could continue. The GI from a table is hardly a good predictor of your own personal GI and whether or not you may encounter a modicum of reactive hypoglycemia that causes hunger.
I am intrigued by the relative "failure" of the high protein low GI diet. Perhaps its a fluke, or perhaps if you are going to eat a high protein diet, it may be best to eat that protein with white rice and potatoes. At least in this study the known satiating effects of lower GI/higher fiber foods and higher protein intake do not appear to combine synergistically. Things to ponder ...