To determine the glycemic index (GI) dependence on the training state of healthy adult males.
Subjects and design:
Young, adult males of normal body mass index and normal glucose tolerance were tested twice with a 50 g reference glucose solution and twice with a breakfast cereal containing 50 g of available carbohydrates in a randomized order. Ten subjects were sedentary (SE), 12 were moderately trained (MT) and 12 were endurance trained (ET). Blood glucose, insulin and glucagon were measured.
The GI differed significantly between SE and ET subjects (P=0.02, mean difference: 23 GI units, 95% CI=3–42 GI units). The GI of the MT subjects was intermediary, but did not differ significantly from the SE or ET subjects. The insulin index did not differ significantly between the groups (P=0.65).
The GI of the commercially available breakfast cereal depended on the training state of the healthy males. The training state is the first reported factor influencing the GI that is subject specific rather than food specific.
The concept of the glycemic index (GI) was introduced by Jenkins et al. 25 years ago (Jenkins et al., 1981). Since then the GI concept has become widely used and even the Food and Agriculture Organization/World Health Organisation recommends the use of the GI (FAO/WHO, 1998). A food's GI is a percentage representing the area under the blood glucose response curve (over 2 h and measured at defined intervals) relative to the blood glucose response curve of a reference food (glucose or white bread) (FAO/WHO, 1998). One of the GI's characteristics is that it describes a specific food and does not seem to be related to subject-specific factors such as age, gender, body mass index (BMI), ethnicity or absolute glycemic response (Wolever et al., 2003). An important subject specific factor, a subject's training state or fitness, however, has not been investigated since the GI's introduction 25 years ago. Thus, we recently have explored whether the GI might differ according to the training state of healthy people, and were surprised to find a substantial (25 GI units) variation in a commercially available breakfast cereal's GI when comparing endurance trained (ET) to sedentary (SE) subjects (Mettler et al., 2006). To our knowledge, this was the first report of a subject-specific factor impacting the GI. As this result would challenge the actual prescription of the GI determination, and to some extent the existing GI values, it was mandatory to validate the results obtained in our recent study. Therefore, this study's main aim was to repeat the GI determination of a breakfast cereal with SE and ET subjects. Secondary aims were to identify possible reasons for the GI's dependence on the training state and to include a group of moderately trained(MT) subjects in order to detect a potential dose–response relation between training state and GI.
Materials and methods
The study was carried out in a random crossover order and consisted of a duplicate assessment of the reference food (glucose) and of a commercially available breakfast cereal. At least 7 days before the first of these four trials, which were separated by 3–7 days, each subject performed a VO2max test. The study's design was approved by the ethical committee of the ETH Zurich. All participants of the study gave their written, informed consent.
All subjects were young (inclusion criteria: age 18–40 years), non-smoking, normoglycemic (glycosylated hemoglobin (HbA1c): 4.8–5.9%; fasting blood glucose <5.6 mmol l−1) males of normal BMI (19–25 kg m−2). None of the subjects’ nearest relatives had diabetes. Forty-one subjects who fulfilled all inclusion criteria were divided into three groups according to their reported training volume. Seven subjects dropped out before completing all trials because of illness, unexpected change in their workplace or other personal reasons, such as not feeling comfortable with the arm vein catheter or inaccurate adherence to the standardization procedure. The remaining subjects consisted of 10 SE males who were neither involved in any exercise-like activity in their leisure time nor engaged in any physically intense occupational activity, 12 MT males who exercised two to three times per week and 12 ET males who trained at least four times per week, and some of them being competitive athletes (Table 1). The subjects’ physical activity level had to be constant for at least 6 months before the start of and throughout the study.
Body mass was measured in the fasted state with light clothing (underwear and t-shirt) on a scale (Multi Range, Sauter GmbH, Albstadt 1-Ebingen, Germany). Height was measured in the morning and body fat was measured with a skinfold caliper (Harpenden Skinfold Caliper, British Indicators, West Sussex, UK) at three skinfolds (chest, abdominal, femoral). Body density and body fat were calculated according to Jackson et al. (1980) and Siri (1993).
The maximal oxygen uptake (VO2max) was determined on a cycle ergometer (Ergoline 800S, Ergoline GmbH, Bitz, Germany). The subjects were asked not to have any large meal within 4 h of the test and not to have any small snack within 2 h of the test. The physically active subjects were asked not to compete the week before the test and not to train vigorously in the 2 days before the test. The protocol consisted of a 5 min warm-up at 50 W, a start power of 80 and a 15 W increment every 30 s (Buchfuhrer et al., 1983; Zhang et al., 1991). The pedal frequency was not standardized during the warm-up, but it was set at or above 70 (±3) r.p.m. throughout the test. Oxygen uptake was measured breath by breath with the Quark b2 System (COSMED, Rome, Italy). The highest oxygen uptake reached, averaged over 20 s, was designated as the VO2max. All subjects reached a VCO2/VO2 above 1.1 and a subjective rating of at least 18 on the BORG scale.
The reference food for the GI determination was glucose (50 g dissolved in 200 ml water) and the commercially available breakfast cereal was Kellogg's Special K (Kellogg AG, Zug, Switzerland). The subjects were served 57.6 g of the cereal with 150 ml of partially skim milk (2.8% fat), so that the test meal provided a total of 50 g available carbohydrates, 14 g protein, 4.6 g fat and 1.7 g fiber. An additional 200 ml tap water was served with both meals. The amount of available carbohydrates was calculated according to nutritional information provided by the manufacturer and additionally verified by analysis of the digestible (available) starch and carbohydrate as well as indigestible fibers with the AOAC/AACC method (McCleary et al., 1997, 2002) using enzymatic kits of Megazyme International (Wicklow, Ireland).
In addition to maintaining their habitual physical activity level throughout the entire experimental period, each subject's physical activity level was standardized for 2 days before each trial according to the subject's habitual lifestyle and activity pattern. Thus, the SE subjects refrained from physical activity above their normal level during the 2 days preceding each trial. In contrast, the MT subjects were not allowed to exercise the day before each trial, and the ET subjects were not allowed to perform any vigorous exercise during the 24 h before each trial.
All subjects semiquantitatively recorded their habitual diet the day before their first trial by estimating their serving sizes. To replicate their pretrial diet, the subjects received a copy of their dietary protocol and were asked to follow and report it again. A dinner, calculated to provide about one-third of the subject's daily energy requirements, was provided to all subjects (pasta with tomato sauce, apple puree and cheese). No food was permitted after dinner until the beginning of the trial. Water was allowed ad libitum until bedtime and restricted to one glass in the morning until 1 h before the trial. Alcoholic drinks were restricted to not more than two glasses and drinks containing caffeine to not more than two cups of coffee or 1 l of soft drinks containing caffeine the day before the trial.
The subjects arrived at the laboratory between 0745 and 0830 hours after an overnight fast of at least 10 h, and just after arrival a catheter was placed into an antecubital vein. Thereafter, adherence to the pretrial standardization was controlled by questionnaire, and subjective well-being and satiety scores were evaluated with a visual analog scale. At the next step and about 10–15 min after catheter placement, the fasting blood samples were drawn. Venous blood (6 ml) was taken from the catheter with an ethylenediaminetetraacetic acid containing S-Monovette (Sarstedt AG, Sevelen, Switzerland) for glucose, insulin and glucagon analysis. Capillary blood for glucose analysis was taken from the opposite arm simultaneously to the venous blood sampling. (This procedure was repeated after 15, 30, 45, 60, 90 and 120 min according to the standard GI measurement procedure (FAO/WHO, 1998).) Just after the fasting blood samples were taken, the reference glucose solution or the breakfast cereal was served. Both the reference solution and the breakfast cereal had to be eaten within 8 min at most.
Analysis of blood parameters
Capillary whole-blood glucose was measured with the Glucotrend 2 system (Roche Diagnostics GmbH, Mannheim, Germany). The 6 ml venous blood sample was stabilized with 300 μl Trasylol (Bayer, Zurich, Switzerland), put on wet ice and centrifuged for 15 min at 1800 g in a cooled centrifuge (Omnifuge 2.0 RS, Heraeus Sepatech, Osterode, Germany) within 15 min. Then, the plasma was dispersed into polypropylene tubes and frozen at −80°C. Insulin and glucagon were prepared with a multiplex immunoassay (Linco Research, St Charles, MO, USA) according to the manufacturer's standard protocol and were read on a Luminex 100 (Bio-Rad Laboratories AG, Reinach, Switzerland). Venous glucose was measured by an enzymatic colorimetric method, using a centrifugal analyzer (Cobas-Mira, Roche, Basel, Switzerland).
Calculations and statistical methods
The incremental area under the curve (IAUC) values for glucose and insulin were calculated geometrically, ignoring areas below the fasting value (Brouns et al., 2005). The IAUC was determined for each test. The values of the two reference glucose solutions and the two cereal trials were averaged, and the GI and insulin indices were calculated by dividing the average cereal IAUC by the average reference glucose solution IAUC.
Statistical analysis was performed with SAS for Windows (version 8.2, SAS institute Inc., Cary NC, USA) using analysis of variance with Bonferroni adjustment for multiple comparisons. Data are presented as mean±s.e., unless otherwise stated. A P-value of <0.05 was considered significant.
The fasting values of all parameters (capillary and venous glucose, insulin, glucagon, subjective well-being and satiety rating) were neither statistically different between the subject groups nor between the trials. All subjects were normoglycemic and there was no period effect on any parameter.
The GI derived from capillary blood, which is the recommended method for GI analysis (Wolever et al., 2003; Brouns et al., 2005), was significantly lower for the ET subjects than for the SE subjects (P=0.02, mean difference: 23 GI U, 95% CI=3–42 GI U; Figure 1). The MT subjects’ capillary GI value was between the two other groups, but did not differ significantly from either of them (Figure 1). The GI calculated from the venous glucose (P=0.73) and the insulin index (P=0.65) did not differ between the groups (Figure 1). Capillary and venous glucose peak time, as well as insulin peak time, did not differ between the groups or the trials. The postprandial glucagon response differed neither between the groups nor the trials (data not shown). The average coefficients of variation (CV) for the duplicate IAUC determinations of the glucose trials were 0.21 (capillary glucose), 0.25 (venous glucose) and 0.23 (insulin). For the cereal trials, the average CV were 0.26, 0.26 and 0.22.
A breakfast cereal's GI was determined in young male subjects who differed in their training states. Similar to our previous study, which found a GI difference of 25 U with a breakfast cereal between SE and ET subjects (Mettler et al., 2006), the present study found that the cereal's GI was significantly higher (P=0.02) with the SE subjects than with the ET subjects (mean difference 23 GI units, 95% CI=3–42GI units). The results available today suggest that the GI of at least some breakfast cereals may be higher with SE subjects than with ET subjects. The GI of the MT, which was between the GI of the SE and the ET, supports the notion that GI depends on the subject's training state.
The most obvious reason for a difference in the GI of SE and ET subjects could be a different insulin metabolism according to the training state. It is often believed that endurance training leads to improved insulin sensitivity (Rogers et al., 1990; Nassis et al., 2005), thereby influencing the glycemic response. However, improved insulin sensitivity should improve the glycemic response to both a glucose solution and a food containing carbohydrates, thus not impacting the relation between the two glycemic responses (i.e., the GI). Accordingly, in the present study, the absolute insulin response, which was calculated as the insulin area under the curve according to the GI formula, is significantly higher in the SE subjects than in the MT and ET subjects (Figure 1), whereas the insulin index, as the relative insulin response between the glucose and the cereal, was not significantly different (P=0.65) between the SE, MT and ET subjects (Figure 1).
Another possible reason that GI could depend on the training state could be the insulin response curve's pattern, rather than the area under the curve. This notion is supported by a recent study, in which the peak time of the insulin response, but not the area under the curve, was related to a GI difference between two different breakfast cereals (Schenk et al., 2003). A divergent insulin secretion could theoretically be related not only to a food but also to a subject. A stronger insulin boost in the first postprandial minutes with an earlier insulin peak in one of the two trials and in only one of the subject groups could have lead to a relatively different glucose behavior, and thus a training state-dependent GI in the present study. However, the insulin data do not support this hypothesis, either, and cannot explain the different GI values between the SE and ET subjects.
Design and other issues
This study was designed to represent the habitual lifestyle of subjects with different physical activity levels. As a consequence, a different standardization between the subject groups was necessary. It would not have made sense to prescribe an exercise session the day before the trials to the usually SE subjects, nor would it have been sound to prohibit all types of exercise to the more or less daily active ET subjects. However, it is certainly important to standardize within each subject and within each group.
This study cannot determine whether the training state's observed effect is mainly of chronic origin or is owing to the last exercise bout only (acute origin). However, physical activity of different intensity than usual performed the day before measuring postprandial glycemia does not seem to influence postprandial glycemia (Young et al., 1989; Campbell et al., 2003). Neither is the within-subject variation in the postprandial glycemic response reduced when controlling or not controlling the subject's diet and physical activity on the day before measuring the postprandial glycemia (Campbell et al., 2003). This suggests that the absolute glycemia is, at least in young and healthy subjects, probably not primarily affected by the acute effects of physical activity. However, because the GI is a relative glycemic response within a specific subject, an imaginable acute effect of physical activity on the absolute glycemic response between different subjects is secondary anyway.
As several studies indicate that the GI is independent of a subject's glucose tolerance or diabetes (Brouns et al., 2005), it might appear that physical activity, which may influence the glucose tolerance (Rogers et al., 1990), might also be unrelated to the GI. This, however, does not seem to be the case as the results of the present and our previous study suggest (Mettler et al., 2006). As neither this study's data nor any found in the literature allow us making an educated guess about the underlying cause of the different GIs, we can only speculate about the reasons that GI depends on the subject's training state. Several food-specific factors such as the protein (Spiller et al., 1987), fat (Owen and Wolever, 2003) and fibers (Jenkins and Jenkins, 1985) content, and the food preparation methods (Foster-Powell et al., 2002) or the second meal effects (Granfeldt et al., 2006), may influence the absolute glycemic response. However, to impact the GI, one or several of these or other unknown factors must affect the glycemia of SE and ET subjects differently.
Interestingly, the GI derived from the venous blood glucose did not show even a slight tendency of difference between the ET and SE subjects. As venous blood glucose is not recommended to be used for GI measurements because of other reasons (Brouns et al., 2005), the absence of a difference in the venous GI is of secondary importance for the GI concept, because the available GI data, which are used for comparison, have been obtained from capillary blood. From a physiological point of view, however, it remains interesting all the same, as it would not be expected that the capillary and venous GI would differ. The present data suggest that a different rate of glucose disappearance between the capillary and venous blood may lead to the observed GI difference.
Conclusion and future prospects
This study's aims were to validate our previous results, which showed that GI may depend on the subjects’ training state, and to identify potential reasons for this. The present results seem to confirm the previous results and contribute to the actual understanding of the factors affecting the glycemic response and the GI. From the methodological viewpoint it seems mandatory to further investigate possible impacts of subject-specific factors on the GI.
So far, the training state of young male subjects has been shown to affect the GI of breakfast cereals. It is now important to expand this knowledge not only to different foods but also to other population groups such as female or elderly subjects, who might have insulin sensitivity or glycemic responses different from male or young subjects (Rogers et al., 1990; Unwin et al., 2002; Borissova et al., 2005).
Borissova AM, Tankova T, Kirilov G, Koev D (2005). Gender-dependent effect of ageing on peripheral insulin action. Int J Clin Pract 59, 422–426.
Brouns F, Bjorck I, Frayn KN, Gibbs AL, Lang V, Slama G et al. (2005). Glycaemic index methodology. Nutr Res Rev 18, 145–171.
Buchfuhrer MJ, Hansen JE, Robinson TE, Sue DY, Wasserman K, Whipp BJ (1983). Optimizing the exercise protocol for cardiopulmonary assessment. J Appl Physiol 55, 1558–1564.
Campbell J, Glowczewski T, Wolever TM (2003). Controlling subjects’ prior diet and activities does not reduce within-subject variation of postprandial glycemic responses to foods. Nutr Res 23, 621–629.
FAO/WHO (1998). Carbohydrates in Human Nutrition: Report of a Joint FAO/WHO Expert Consultation. FAO/WHO: Rome.
Foster-Powell K, Holt SH, Brand-Miller JC (2002). International Table of Glycemic Index and Glycemic Load Values: 2002. Am J Clin Nutr 76, 5–56.
Granfeldt Y, Wu X, Bjorck I (2006). Determination of glycaemic index; some methodological aspects related to the analysis of carbohydrate load and characteristics of the previous evening meal. Eur J Clin Nutr 60, 104–112.
Jackson AS, Pollock ML, Ward A (1980). Generalized equations for predicting body density of women. Med Sci Sports Exerc 12, 175–181.
Jenkins DJ, Jenkins AL (1985). Dietary fiber and the glycemic response. Proc Soc Exp Biol Med 180, 422–431.
Jenkins DJ, Wolever TM, Taylor RH, Barker H, Fielden H, Baldwin JM et al. (1981). Glycemic index of foods: a physiological basis for carbohydrate exchange. Am J Clin Nutr 34, 362–366.
McCleary BV, Gibson TS, Mugford DC (1997). Measurement of total starch in cereal products by amyloglucosidase–alpha–amylase method: Collaborative Study. J AOAC Int 80, 571–579.
McCleary BV, McNally M, Rossiter P (2002). Measurement of resistant starch by enzymatic digestion in starch and selected plant materials: collaborative study. J AOAC Int 85, 1003–1111.
Mettler S, Wenk C, Colombani PC (2006). Influence of the training status on the glycemic index. Int J Vitam Nutr Res 76, 39–44.
Nassis GP, Papantakou K, Skenderi K, Triandafillopoulou M, Kavouras SA, Yannakoulia M et al. (2005). Aerobic exercise training improves insulin sensitivity without changes in body weight, body fat, adiponectin, and inflammatory markers in overweight and obese girls. Metabolism 54, 1472–1479.
Owen B, Wolever TM (2003). Effect of fat on glycaemic responses in normal subjects: a dose–response study. Nutr Res 23, 1341–1347.
Rogers MA, King DS, Hagberg JM, Ehsani AA, Holloszy JO (1990). Effect of 10 days of physical inactivity on glucose tolerance in master athletes. J Appl Physiol 68, 1833–1837.
Schenk S, Davidson CJ, Zderic TW, Byerley LO, Coyle EF (2003). Different glycemic indexes of breakfast cereals are not due to glucose entry into blood but to glucose removal by tissue. Am J Clin Nutr 78, 742–748.
Siri WE (1993). Body composition from fluid spaces and density: analysis of methods. 1961. Nutrition 9, 480–491.
Spiller GA, Jensen CD, Pattison TS, Chuck CS, Whittam JH, Scala J (1987). Effect of protein dose on serum glucose and insulin response to sugars. Am J Clin Nutr 46, 474–480.
Unwin N, Shaw J, Zimmet P, Alberti KG (2002). Impaired glucose tolerance and impaired fasting glycaemia: the current status on definition and intervention. Diabet Med 19, 708–723.
Wolever TM, Vorster HH, Bjorck I, Brand-Miller J, Brighenti F, Mann JI et al. (2003). Determination of the glycaemic index of foods: Interlaboratory Study. Eur J Clin Nutr 57, 475–482.
Young JC, Enslin J, Kuca B (1989). Exercise intensity and glucose tolerance in trained and nontrained subjects. J Appl Physiol 67, 39–43.
Zhang YY, Johnson MC, Chow N, Wasserman K (1991). Effect of exercise testing protocol on parameters of aerobic function. Med Sci Sports Exerc 23, 625–630.
We thank Astrid Peterhans, Fabienne Schwitz and Patricia Kressig for their medical assistance, and the Swiss Foundation for Nutrition Research and the Federal Council of Sports for partially funding this study.
Study data have not been previously published and are not currently submitted for publication.
Guarantor: PC Colombani.
Contributors: SM was responsible for and coordinated the execution of the study and laboratory analysis, performed the statistical analysis and drafted the manuscript. SM, CW and PCC obtained the funding, planned and designed the study. SM, FLR and NSK executed the study. PCC edited the manuscript and FLR, NSK and CW commented on the manuscript.
About this article
Cite this article
Mettler, S., Lamprecht-Rusca, F., Stoffel-Kurt, N. et al. The influence of the subjects’ training state on the glycemic index. Eur J Clin Nutr 61, 19–24 (2007). https://doi.org/10.1038/sj.ejcn.1602480
- glycemic index
- insulinemic index
Effect of pre-exercise carbohydrate diets with high vs low glycemic index on exercise performance: a meta-analysis
Nutrition Reviews (2017)
European Journal of Nutrition (2014)
Dietary fiber and the glycemic index: a background paper for the Nordic Nutrition Recommendations 2012
Food & Nutrition Research (2013)
Progress in Retinal and Eye Research (2011)
Nutrition Reviews (2011)