| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research Communication |
1 From the Human Performance Laboratory, Department of Kinesiology and Health Education, The University of Texas at Austin (SS, CJD, TWZ, and EFC), and the Stable Isotope Laboratory, Pennington Biomedical Research Center, Baton Rouge, LA (LOB).
2 Supported in part by the Quaker Oats Company. Correspondence: 3 Address reprint requests to EF Coyle, Department of Kinesiology and Health Education, The University of Texas at Austin, Bellmont Hall 222, Austin, TX 78712. E-mail: coyle{at}mail.utexas.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objective: We compared the plasma glucose kinetics of low- and high-GI breakfast cereals.
Design: On 2 occasions, plasma insulin concentrations and plasma glucose kinetics (by constant-rate infusion of [6,6-2H2]glucose) were measured in 6 healthy males for 180 min after they fasted overnight and then consumed an amount of corn flakes (CF) or bran cereal (BC) containing 50 g available carbohydrate.
Results: The GI of CF was more than twice that of BC (131.5 ± 33.0 compared with 54.5 ± 7.2; P < 0.05), despite no significant differences in the rate of appearance of glucose into the plasma during the 180-min period. Postprandial hyperinsulinemia occurred earlier with BC than with CF, resulting in a 76% higher plasma insulin concentration at 20 min (20.4 ± 4.5 compared with 11.6 ± 2.1 µU/mL; P < 0.05). This was associated with a 31% higher rate of disappearance of glucose with BC than with CF during the 30–60-min period (28.7 ± 3.1 compared with 21.9 ± 3.1 µmol · kg-1 · min-1; P < 0.05).
Conclusion: The lower GI of BC than of CF was not due to a lower rate of appearance of glucose but instead to an earlier postprandial hyperinsulinemia and an earlier increase in the rate of disappearance of glucose, which attenuated the increase in the plasma glucose concentration.
Key Words: Carbohydrate • glucose kinetics • protein • nutrition • hyperinsulinemia
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A fundamental assumption regarding low-GI foods is that they produce a low glycemic response as a result of a slower rate of digestion of carbohydrate in the intestinal lumen, and that this subsequently slows the absorption of glucose into the circulation (1, 2, 4–7). The plasma glucose concentration, however, is a function of both the rate of appearance of glucose (Raglucose) into the systemic circulation and the rate of disappearance of glucose (Rdglucose) from the systemic circulation. Some properties of foods that may influence Raglucose (ie, glucose absorption from the intestines into the portal and peripheral circulation) include the intrinsic properties of the carbohydrate (eg, gelatinization and amylose-to-amylopectin ratio) and the fiber and fat contents of the food (8–10). The Rdglucose (ie, glucose uptake from the blood by tissue) is primarily influenced by insulin secretion and action on tissue (11). Although glucose is the primary macronutrient that stimulates insulin release, protein may augment insulin release when ingested with carbohydrate (12–14).
The rate at which the carbohydrate in a food is absorbed into the blood as glucose can only be inferred by directly measuring the plasma glucose kinetics (ie, Raglucose) and not from the GI. For example, a food can have a low GI as the result of having either a relatively low Raglucose or a relatively high Rdglucose. To our knowledge, no studies comparing the GI of foods have actually measured plasma glucose kinetics. Thus, direct data regarding the underlying factors responsible for differences in the GI of foods seem absent from the literature. Therefore, the purpose of the present study was to describe the underlying glucose kinetics responsible for the different glycemic responses of popular low- and high-GI breakfast cereals (2).
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Diet and exercise control
The day before the initial trial, the subjects were instructed to record their diet and then to repeat this diet the day before subsequent trials. Notably, the final meal of the day was a standardized snack of high glycemic carbohydrate (0.5 g/kg). This was provided to ensure that the initial plasma substrate concentrations and any measured metabolic differences between the trials were not a consequence of the final meal (15). Physical activity was standardized the day before all experimental trials, with subjects undertaking cycling exercise for 1 h at 
60% of their age-predicted maximum heart rate.
Experimental meals
The experimental cereals were a bran cereal (BC; All-Bran Original; Kellogg’s, Battle Creek, MI) and a corn flakes cereal (CF; Kellogg’s Corn Flakes). These cereals were chosen because they are popular breakfast cereals and were previously shown to have a low (BC) and high (CF) GI (16). A placebo drink (flavored water) was used as the fasting control. In another trial, a glucose-water solution (GL) was given as the reference food so that the GI could be calculated for the cereals. Four hundred milliliters water was consumed with the meal during each trial. Each experimental meal (except the placebo drink) provided 50 g available carbohydrate. This amount was chosen according to the standardized protocol for determination of the GI as outlined by Wolever et al (2). The macronutrient composition of each meal is provided in Table 1
.
|
10 h); on arrival, catheters were inserted into a forearm vein of each arm for blood sampling and isotope infusion, respectively. A primed, constant-rate infusion of [6,6-2H2]glucose (prime = 35 µmol/kg, constant infusion = 0.40 µmol · kg-1 · min-1; Isotec Inc, Miamisburg, OH) administered by using calibrated syringe pumps (Harvard Apparatus, South Natick, MA) was then initiated and maintained for 1.5 h before and for 3 h after ingestion of the experimental meal. There was no isotope infusion or measurement of plasma glucose kinetics during the GL meal, because this trial served only as the reference for calculation of the GI of the cereals. During the 180 min after meal ingestion, the subjects remained seated in a standardized position and performed significant bodily movement only at 31, 91, and 151 min postprandially. At these times, they were allowed to urinate if necessary and were then asked to cycle a stationary ergometer for 3 s, performed 4 times with 2 min of seated recovery. This was part of another study and it was directly determined that this brief exercise did not alter plasma glucose kinetics (the CF trial was performed both with and without the exercise bouts; data not shown).
Blood sampling and analysis
Blood was sampled (6 mL) before the isotopic infusion began, immediately before ingestion of the experimental meal (0 min), and 20, 30, 60, 90, 120, 150, and 180 min postprandially for determination of plasma glucose and insulin concentrations and plasma glucose kinetics. Blood samples (4 mL) were also taken 54, 84, 114, 144, and 174 min postprandially for analysis of the plasma glucose concentration and plasma glucose kinetics only. During the GL meal, blood was sampled (6 mL) only at 0, 20, 30, 60, 90, 120, 150, and 180 min for analysis of plasma glucose and insulin concentrations. Four milliliters from each sample was placed into prechilled tubes containing 0.2 mL EDTA solution (25 mg/mL) for determination of plasma glucose concentrations and plasma glucose kinetics. Blood (2 mL) for plasma insulin analysis was placed into prechilled tubes containing 0.2 mL EDTA (25 mg/mL) and aprotinin (0.5 U/mL). Plasma was separated by centrifugation (3 x g for 20 min at 4 °C), transferred to 12 x 75 mm plastic tubes, and immediately frozen at -80 °C for subsequent analysis. Plasma glucose concentrations were measured by use of a Colorimetric assay (Trinder; Sigma Inc, St Louis), and plasma insulin concentrations were measured by radioimmunoassay (Linco Research, Inc, St Charles, MO).
Isotope enrichment sample preparation
Plasma samples (300 µL) were deproteinized with 450 µL 0.3 N Ba(OH)2 and 450 µL 0.3 N ZnSO4, mixed by vortexing, and incubated in an ice bath for 20 min. After centrifugation at 3 x g for 15 min at 4 °C, the supernatant fluid was passed down an ion-exchange column and was then rinsed 5 times with 400 µL distilled water. The samples were captured in 13 x 75 mm glass screw-cap tubes and dried overnight by using compressed air. Acetic anhydride (75 µL) and pyridine (75 µL) were added to the dried samples, which were then incubated at 100 °C for 1 h and subsequently dried with nitrogen gas. The samples were then resuspended in ethyl acetate and injected into a gas chromatograph–mass spectrometer with an autosampler (5890 Series II 5988a mass spectrometer; Hewlett-Packard, Palo Alto, CA), and the 200 and 202 masses of the pentaacetate derivative of glucose were monitored with selective ion monitoring.
Calculations
The area under the curve (AUC) for glucose and insulin was obtained by calculating the AUC over the 180-min postprandial period by using the trapezoidal method. Only values above fasting concentrations were included. The GI was calculated by dividing the respective glucose AUC for BC and CF by the glucose AUC for GL and multiplying by 100. Glucose kinetics (ie, Raglucose and Rdglucose) were calculated by using the one-pool, non–steady state equations of Steele (17), modified for use with stable isotopes:
 | (1) |
 | (2) |
 | (30 |
where F is the isotope infusion rate, Vd is the effective volume of distribution, C is the plasma concentration of the tracee, (E2 - E1)/(t2 - t1) is the change in enrichment (ie, E = tracer-to-tracee ratio) between 2 consecutive samples (t2 - t1= x min), and GCR is the glucose clearance rate. Vd was estimated to be 150 mL/kg (18).
Glucose kinetics were calculated for the following time periods: 0–20 min, 20–30 min, 30–54 min, 54–60 min, 60–84 min, 84–90 min, 90–114 min, 114–120 min, 120–144 min, 144–150 min, 150–174 min, and 174–180 min. Total Raglucose and Rdglucose in µmol were subsequently calculated for each time period, and then the average glucose kinetics over a 30-min time period were calculated by dividing the total Raglucose or Rdglucose by 30 min (ie, 0–30 min, 30–60 min, 60–90 min, 90–120 min, 120–150 min, and 150–180 min). The GCR represents an index of the glucose uptake relative to the prevailing plasma glucose concentration and thus gives an index of insulin-mediated plasma glucose uptake.
Experimental design and statistics
A two-way repeated-measures analysis of variance (ANOVA; cereal x time) was used to determine significant main effects and interactions. When there was a significant interaction or main effect of time, mean contrasts according to modified Bonferroni inequalities were used to analyze dependent variables (plasma glucose and insulin concentrations, Raglucose, Rdglucose, and GCR) at specific time points. A one-way repeated-measures ANOVA was used to analyze the GI and overall glucose AUC (ie, 180-min postprandial period). Data for the placebo mean are not presented because, as expected, plasma glucose and insulin values did not differ significantly from 0 min, ie, before meal ingestion. Statistical significance was defined as P < 0.05. All results are presented as means ± SEMs.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). As a result, the glucose AUC for BC was significantly lower than that for CF from 30 to 120 min (all P < 0.05; Figure 1B). Moreover, the glucose AUC for the 180-min postprandial period (192.5 ± 38.4 compared with 85.7 ± 12.1 mmol · min/L; P < 0.05) and the GI (131.5 ± 33.0 compared with 54.5 ± 7.2; P < 0.05) for CF were more than twice those for BC.
|
). There was a main effect of time for both BC and CF, whereby Raglucose significantly increased above the fasted state during the initial 120 min after meal ingestion (P < 0.05).
|
). In light of this, the insulin AUC was more than twice as great for BC during the 0–30-min postprandial period (288.7 ± 71.0 compared with 128.4 ± 42.9 µU · min/mL; P < 0.05; Figure 3B). The augmented postprandial hyperinsulinemia for BC was only transient, however, because there was no significant difference between the cereals at 30 min (25.5 ± 3.5 compared with 22.1 ± 4.6 µU/mL for BC compared with CF; NS) or at any time point thereafter. Similarly, there were no further differences in the insulin AUC during the 30–180-min postprandial period nor any significant difference in the overall insulin AUC (1523.3 ± 155.7 compared with 1602.4 ± 138.9 µU · min/mL for BC compared with CF; NS).
|
). Rdglucose was not significantly different during the initial 0–30-min postprandial period (17.5 ± 1.5 compared with 14.1 ± 0.9 µmol · kg-1 · min-1 for BC compared with CF) or during the 60–180-min period (all NS). Similar to Rdglucose, the GCR was only significantly different between the cereals during the 30–60-min period, with the GCR for BC being 54% higher than that for CF (5.3 ± 0.8 compared with 3.5 ± 0.6 mL · kg-1 · min-1; P < 0.05, Figure 4B). Furthermore, the GCR increased above the fasted state during the 30–60-min postprandial period for BC only and was delayed until 60–90 min for CF (P < 0.05). The increase in the GCR during the 30–60-min period was significantly correlated with the initial plasma insulin response as measured by the 0–30-min insulin AUC (r = 0.71, P < 0.009).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The fundamental assumption of the GI concept is that low-GI foods produce a lesser increase in the plasma glucose concentration as a result of slower rates of gastric emptying and digestion of carbohydrate in the intestinal lumen and, subsequently, a slower rate of absorption of glucose (ie, Raglucose) into the portal and systemic circulation (1, 2, 4–7). In the present study, we used stable-isotope methods to directly determine that although the GI of CF was more than 2-fold higher than that of BC, the postprandial change in Raglucose was not significantly different between CF and BC. It is perhaps surprising that despite the high fiber content of BC compared with CF (38.5 compared with 2.2 g), the Raglucose of BC was not significantly different from that of CF. However, BC is composed primarily of insoluble wheat bran fiber (92%), which has been shown to have little effect on reducing Raglucose when co-ingested with carbohydrate (19, 20). Also, BC contains a higher proportion of simple sugars than does CF (23 compared with 4 g), and this could be another reason that the Raglucose of BC was not significantly different from that of CF.
It is important, however, to bear in mind that although CF contains a small quantity of simple sugars, it still has a comparable Raglucose to BC. This observation is supported by the fact that although a food may be high in complex carbohydrate (eg, corn starch), this does not necessarily mean that the glycemic response will be low (21, 22). Differences in the gastric emptying rate could influence Raglucose, although given the fact that Raglucose was not significantly different between CF and BC, it is reasonable to hypothesize that the rate of available carbohydrate movement from the stomach into the intestines was similar. Independent of those factors that may influence Raglucose, it is noteworthy that we directly determined that the postprandial change in Raglucose was not significantly different between CF and BC. Accordingly, the present study highlights that the GI is not solely a function of Raglucose.
An increase in the plasma glucose concentration above postabsorptive concentrations is the primary stimulus for increased insulin secretion. Given that Raglucose was not significantly different between trials, the change in plasma insulin concentration would be expected to be comparable between cereals. However, the postprandial increase in the plasma insulin concentration occurred earlier for BC than for CF and resulted in a 76% higher plasma insulin concentration at 20 min and a 125% higher insulin AUC during the 0–30-min postprandial period. In the present study, although the amount of available carbohydrate in the 2 cereals was identical, BC contained 3.5 times more protein than did CF (15.4 compared with 4.3 g). When protein is added to carbohydrate, insulin secretion and the calculated insulin AUC are augmented (12–14, 23–26); several studies have found that adding protein to carbohydrate attenuates the glycemic response compared with that for carbohydrate alone, in association with a greater insulin response (12, 13, 23, 25, 26). Notably, the addition of 30 g wheat protein, the type of protein present in BC, to 59 g carbohydrate resulted in augmented hyperinsulinemia compared with carbohydrate alone (14). Therefore, we believe that in our study the protein-carbohydrate interaction of BC was likely responsible for the more rapid plasma insulin response during the initial 20-min postprandial period.
During the postprandial period, Rdglucose is primarily stimulated by an increase in insulin concentration (11). Rdglucose did not increase above postabsorptive levels for either BC or CF (ie, 30–60 min) until the plasma insulin concentration significantly increased above overnight-fasted concentrations. The markedly earlier insulin response for BC was associated with an earlier increase in Rdglucose, resulting in a 31% higher Rdglucose for BC than for CF during the 30–60-min postprandial period. Importantly, although this was the only time period during which Rdglucose was significantly different between the 2 cereals, this earlier increase in Rdglucose is what attenuated hyperglycemia and the GI of BC. Individuals with impaired glucose tolerance show a delayed rise in postprandial insulin secretion compared with insulin-sensitive control subjects (27, 28). When the delayed rise in plasma insulin concentration is corrected through early insulin supplementation, abnormal hyperglycemia is corrected, in parallel with an earlier increase in Rdglucose (27, 28). Although we saw no significant difference in the insulin AUC over the 180-min postprandial period, it is the markedly earlier rise in plasma insulin concentration seen for BC that appears functionally important for increasing Rdglucose early in the postprandial period, thereby reducing the postprandial hyperglycemia.
The important role of insulin in increasing glucose extraction from the blood is further supported by the GCR, which provides an indication of insulin-mediated glucose uptake by tissue (11). In the present study, the GCR increased above fasting levels during the 30–60-min period with BC only, being 54% higher than that for CF. In parallel with the delayed insulin response with CF, an increase in the GCR was delayed with CF until the 60–90-min period. Interestingly, this temporal delay of 30 min between insulin appearance in plasma and insulin metabolic action (as indicated by the increased GCR) has been extensively described (29). Furthermore, Carey et al (30) found that the rate of postprandial muscle glycogen storage, an indirect measure of plasma glucose uptake, was significantly correlated with the initial plasma insulin response (r = 0.87). Similarly, we found a strong correlation (r = 0.71, P < 0.009) between the 30–60-min GCR and the initial plasma insulin response (ie, 0–30-min insulin AUC), further supporting the importance of the early insulin response in increasing plasma glucose removal from the circulation and consequently in attenuating postprandial hyperglycemia.
The concept of GI is based primarily on the interpretation of the glycemic curve and in some instances the relation between the in vivo postprandial glycemic curve and in vitro measurements of glucose availability (31, 32). Although we appreciate that not all foods have a low GI as a result of a greater Rdglucose, we have highlighted herein that the plasma glucose concentration and the GI are indeed a function of both Raglucose and Rdglucose. It is therefore perhaps tenuous to draw conclusions regarding the "behavior," and hence the therapeutic applicability, of a food solely from plasma glucose and insulin concentrations without appropriate support from measures of glucose kinetics. Indeed, if we did not directly determine the glucose kinetics of BC and CF, then we would be uncertain whether the lower glycemic response of BC than of CF was due to a reduced Raglucose secondary to its high fiber content or to enhanced Rdglucose secondary to an earlier increase in plasma insulin concentration. Further research with the use of isotope methods to quantify the plasma glucose kinetics of a broader range of foods, in particular foods containing different types of carbohydrate and different amounts of fat and protein, is necessary to further expand the functional usefulness of the GI. It is also important to emphasize the need for frequent blood sampling during the initial 30 min of the postprandial period (eg, every 10 min) for assessment of plasma insulin concentrations because it is the first-phase insulin response, and not necessarily the total AUC, that seems important in attenuating postprandial hyperglycemia and hence the GI.
In conclusion, to our knowledge, this is the first study to systematically investigate the underlying glucose kinetics responsible for the different glycemic response of 2 popular, commercially available foods of low and high GI. The high fiber content of BC did not appear to affect glucose absorption into the systemic circulation, because Raglucose was not significantly different between the 2 cereals. Despite a comparable Raglucose, the glycemic response of CF was more than twice that of BC. Instead, the lower postprandial glycemia in BC was due to an augmented Rdglucose during the 30–60-min period in association with an earlier and significantly greater insulin response during the initial 20-min period. Overall, the present findings further the concept of the GI by emphasizing that differences in Rdglucose, and not just Raglucose, can account for differences in GI between foods.
|
ACKNOWLEDGMENTS |
|---|
SS, CJD, and ECF were involved in the design of the experiment. SS and CJD undertook the data collection. SS performed all data analysis with assistance from CJD (plasma glucose and insulin) and TWZ (plasma glucose, glucose tracer, and calculations). LOB supervised and provided technical expertise in mass-spectrometry analysis of the glucose tracer and calculations. SS was the primary writer of the manuscript. ECF served as laboratory director and supervising professor and also oversaw the writing of the manuscript.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. O. Weickert and A. F. H. Pfeiffer Metabolic Effects of Dietary Fiber Consumption and Prevention of Diabetes J. Nutr., March 1, 2008; 138(3): 439 - 442. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Englert, K. Wells, W. Long, M. S. Hickey, and C. L. Melby Effect of Acute Prior Exercise on Glycemic and Insulinemic Indices J. Am. Coll. Nutr., June 1, 2006; 25(3): 195 - 202. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Gruendel, A. L. Garcia, B. Otto, C. Mueller, J. Steiniger, M. O. Weickert, M. Speth, N. Katz, and C. Koebnick Carob Pulp Preparation Rich in Insoluble Dietary Fiber and Polyphenols Enhances Lipid Oxidation and Lowers Postprandial Acylated Ghrelin in Humans J. Nutr., June 1, 2006; 136(6): 1533 - 1538. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. F. Sheard, N. G. Clark, J. C. Brand-Miller, M. J. Franz, F. X. Pi-Sunyer, E. Mayer-Davis, K. Kulkarni, and P. Geil Dietary Carbohydrate (Amount and Type) in the Prevention and Management of Diabetes: A statement by the American Diabetes Association Diabetes Care, September 1, 2004; 27(9): 2266 - 2271. [Full Text] [PDF]
|
|
|
|
|
|
T. M.S. Wolever, J. E. Campbell, D. Geleva, and G. H. Anderson High-Fiber Cereal Reduces Postprandial Insulin Responses in Hyperinsulinemic but not Normoinsulinemic Subjects Diabetes Care, June 1, 2004; 27(6): 1281 - 1285. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
] or high [corn flakes (CF); 
and 
] glycemic index. Each cereal provided 50 g available carbohydrate. Two-way ANOVA showed a main effect of time and a significant cereal-by-time interaction (both P < 0.05). Mean contrasts according to modified Bonferroni inequalities were used to analyze significance at specific time points (P < 0.05). *BC significantly different from time 0. 
CF significantly different from time 0. 
BC significantly different from CF.
