Effect of macronutrients on the glycemic index

Dear Editor:

Meng et al. (1) measured the effect of macronutrients and fiber on postprandial glucose responses and meal glycemic index (GI) and glycemic load determinations and concluded that uncertainty in the determination of meal GI and glycemic load values is introduced when carbohydrate-containing foods are consumed concurrently with protein but not with carbohydrate-, fat-, or fiber-containing foods. I will not dwell on the fact that the authors did not measure the GI correctly and that the variability in their GI results (SD ∼47) is >96% (27 of 28) of international laboratories around the world (2) and thus not representative of the typical variation in GI values. Nevertheless, their results show that GI almost perfectly predicted the relative effect on glycemic responses of mixing different carbohydrate foods together in their studies 1 and 4 (Figure 1A), and I am pleased that the authors acknowledge this fact. However, I criticize the authors for focusing their conclusions on GI rather than on postprandial glycemic responses because, in doing so, the authors use the term “GI” incorrectly and therefore misunderstand and abuse the purpose and utility of GI.

FIGURE 1.

Data shown from Figure 1 in Meng et al. (1). Values are means ± 95% CIs. (A) The x axis shows the predicted RGR for the test meals fed in studies 1 and 4 calculated as follows: RGR = 1.49 × GI × (1 − e^{−0.0222 g}), where “GI” is the calculated meal GI and “g” is grams of available carbohydrate in the meal (3); the y axis shows the observed AUC estimated from Meng et al. (1) Figures 1B1 and 1B4. Lines are regression lines that are not forced through the origin (study 1, r = 0.991; study 4, r = 0.997). (B) Comparison of the effect of adding tuna protein to bread on glycemic response expressed as a percentage of bread alone (filled circles) reported by Meng et al. (1) with that of adding whey protein to glucose in 2 studies [open circles (4) and open triangles (5)]. (C) Comparison of the effect of adding butter to bread on glycemic response expressed as a percentage of bread alone (filled circles) reported by Meng et al. (1) with that of adding polyunsaturated margarine to bread (6). avCHO, available carbohydrate; GI, glycemic index; RGR, relative glycemic response.

The GI is a measure of the ability of the available carbohydrate in carbohydrate-rich foods to increase blood glucose on a gram-for-gram basis. GI is one of many independent determinants of the glycemic response elicited by mixed meals; others include the amount of carbohydrate consumed and the amounts of types of fat, protein, and fiber consumed. Thus, to avoid confounding factors, one measures the GI of a food fed alone. The meal GI calculation yields a value that indicates the ability of the available carbohydrate in the meal to increase blood glucose on a gram-for-gram basis. However, the other components in the meal (amount of carbohydrate, protein, and fat) also influence the glycemic response independently of each other and of GI and this must to be taken into account. Thus, the GI of a mixed meal is not measured, it is calculated. Adding protein or fat to bread does not reduce the GI of the bread, it reduces the glycemic response when compared with that elicited by bread alone. It is naive, incorrect, and dissimulating to blame GI for failing to predict a glycemic response when one has added a confounding factor to the study design and then failed to take it into account. Meng et al. (1) are not the first to do so (7, 8).

I also criticize the authors for drawing generalized conclusions about the effects (or lack thereof) of protein, fat, and fiber on glycemic responses from their results when they studied only one type of protein, one type of fat, and one type of fiber. It is well known that different types of fiber affect glycemic responses differently (9), and the same is true for fat and protein. On the other hand, there are few studies, to my knowledge, on the effect of adding graded doses of protein and fat to carbohydrate. Thus, dose-responses of protein and fat reported by Meng et al. (1) are a useful addition to the literature. Tuna protein was found to reduce the glycemic response (AUC) by ∼0.57%/g protein added to 50 g available carbohydrate, an effect less than half of that elicited by whey protein [∼1.4%/g per 50 g carbohydrate (4, 5); Figure 1B]. Different fats also have different effects: for example, a test meal of glucose, skim-milk powder, and butter elicited a significantly higher glycemic response than the same test meal containing the same amount of fat as safflower oil (10). Meng et al. (1) found that adding butter to bread had a small but nonsignificant effect on glucose AUC, with the percentage reduction tending to be less than that of adding polyunsaturated margarine to bread (6) (Figure 1C).

The GI of mixed meals is calculated, not measured. GI is measured for carbohydrate-rich foods fed alone, and these values are used to calculate the GI of mixtures of foods. The resulting meal GI value is one of many independent determinants of the glycemic response elicited by mixed meals; other determinants include the amount of carbohydrate consumed and the amounts and types of protein, fat, and fiber. The latter have to be taken into account when considering the utility of GI in the context of mixed meals.