"Oh Heavens!" all you readers of both sexes will cry out, "oh Heavens above! But what a wretch the Professor is! Here in a single word he forbids us everything we most love, those little white rolls...and those cookies...and a hundred other things made with flour and butter, with flour and sugar, with flour and sugar and eggs! He doesn't even leave us potatoes, or macaroni!"
- Jean Anthelme Brillat-Savain, 1825 (quoted from Why We Get Fat by G Taubes)Refined starches and sugary foods have historically been considered uniquely fattening. The simplest explanation for this property is that these foods raise blood glucose and evoke more insulin secretion than other foods. The glycemic index (GI), or better yet the glycemic load (GL), are good candidates for determining the most fattening foods. Given this, people who tend towards weight gain and obesity could eliminate or severely restrict some foods, eat some foods in moderation, and even be able to eat other foods ad libitum, in order to prevent weight gain. But how confident can we be with these measures?
The GI offers novel physiological information, but it is limited by the fact that it only measures the response to a set amount of carbohydrates. Since different foods are eaten in different ways and are eaten in different amounts, the GL is likely better in both theory and practice. A very recent paper by Jennie Brand-Miller's group contains a comprehensive analysis of several dietary predictors, including total carbohydrate content, glycemic index, glycemic load, fiber, and protein, for the glucose and insulin responses to a variety of single foods and meals. They found that the GL was the best predictor of glucose and insulin responses to single foods and meals. Absolute carbohydrate content only predicted the glucose and insulin responses for single foods, but had no predictive power for meals; fiber and protein content were also relatively poor predictors. Given the shear number of foods tested - 116 in total - and that they also tested complete meals, we can be confident that the GL is a strong predictor of the acute insulin response to food, and therefore a reasonable estimate of the obesogenicity of a food. But how does it hold up in practice?
Lower GL diets have shown consistent benefits in terms of weight loss and improvement in blood lipids, but these benefits are modest. A systematic analysis by the Cochrane Collaboration found that low GL diets demonstrated more weight loss than did calorie restricted diets. The GL diet only provided an additional two to three pounds of weight loss, on average, but importantly, they did so without requiring intentional caloric restriction - the subjects did not have to go hungry. And this may be the critical point. Or as Dr. David Ludwig puts it, "this conclusion [that GL diets produce only modestly more weight loss] does not consider the fundamental relationship between psychology and physiology. A person's ability to maintain adherence over time may be influenced by the way in which a diet affects hunger and metabolism (Ludwig and Ebbeling NEJM 2010)." Lowering insulin can lower appetite, which will allow a diet to become a lifestyle.
Implementing the GL may reduce morbidity from obesity irrespective of weight loss. After reviewing the clinical literature in 2010, the authors concluded that "clinicians can do better than recommending a conventional low-fat, high-carbohydrate diet for preventing type 2 diabetes." They identified low GL diets as the best candidate for reducing the risk of diabetes. Furthermore, since the inception of the GI, it has been known that improving the quality of dietary carbohydrates can improve blood lipids in hyperlipidemic patients (Jenkins 1985). However, the results from a GL diet can still be improved.
The major limitation of focusing on the post-meal glucose and insulin responses is that it ignores any food-specific effect on chronic glucose and insulin levels. Sucrose, or table sugar, is one part glucose and one part fructose. Starches, such as potatoes, are only composed of glucose. Eating a starchy food will raise blood glucose rapidly because it is primarily glucose, but the same amount of carbohydrate as sucrose will not raise blood glucose, and therefore insulin, to the same magnitude simply because there is less glucose per carbohydrate content. Fructose must first be metabolized by the liver into either glucose or fat. Thus, a can of soda has a lower glycemic index than potato, rice, or bread (53 vs. mid-70's), and it also has a lower glycemic load. However this does not mean that a soda is less fattening than a potato.
The fundamental problem with sugar is that it contains fructose, and when enough sugar is chronically consumed, it wrecks havoc on our metabolism. This, at least, is according to Dr. Robert Lustig, the physician-researcher central to the article Is Sugar Toxic? by Gary Taubes. Fructose causes problems for the brain and the liver. In the brain, excess fructose can lead to something known as "leptin resistance," which causes the brain to underestimate the body's fat stores, and therefore the brain favors sedentary behavior and overeating to make up the difference. Excess fructose also acts on the liver to induce insulin resistance, which perpetuates weight gain and induces metabolic syndrome. Unfortunately, the GI and GL miss this phenomenon, which could explain why simply low GL diets only produce modest results.
But as long as we understand that sugar-ladened foods should be restricted, evaluating a diet by the glycemic load should work to reverse, or at least prevent, weight gain. But can this readily be translated from the clinic into our kitchens? One major objection is that the foods the GL identifies as detrimental are the same foods that everyone loves the most - the breads, pastries, pastas, and desserts - and don't want to give-up. Many clients clung to the Weight Watchers point system when the organization got rid of it because it "allowed" all foods as long clients stayed in a caloric deficit. But Weight Watchers acknowledged that a calorie is not just a calorie, and some foods will hinder weight loss. Furthermore, replacing refined grains and fruit juices with non-starchy vegetables, legumes, nuts, and some more fat is hardly deprivation, and is more similar to what we ate prior to the obesity epidemic.
David Jenkins's impression of the glycemic index was that the foods with the lower GI are commonly found in non-westernized countries eating traditional diets, whereas the highest GI foods are American staples (The Hartford Courant, 2 January 1983); and that the "diseases of western civilization" accompany the elevated glycemic index. This is probably not a coincidence. Because of this association, the GL is not merely another example of "nutritionism." In his article Unhappy Meals, Michael Pollan describes nutritionism as "applied reductionism" where we are no longer worried about eating food, but are instead only concerned with nutrients such as carbohydrates, saturated fat, and caratenoids. The GI and GL are not just more nutrition jargon, rather they are reductionist analyses for understanding our initial observation, namely that some foods make us fat.