No difference in the 24-hour interstitial fluid glucose profile with modulations to the glycemic index of the diet

Methods The study was a randomized, balanced, two-way crossover intervention of 2 × 1-wk periods of lower and higher GI diets. Participants were 12 overweight healthy adult women (mean body mass index ± standard deviation 27.5 ± 2.3 kg/m 2 ). Changes in GI were achieved through substitution of key staple carbohydrate-rich foods. After a 4-d run-in on each dietary regimen, participants wore the continuous glucose-monitoring system over 2 d of identical controlled feeding in the laboratory, separated by 1 d of ad libitum consumption at home. Results On controlled days, diets differed in GI by 15 U and provided equal energy, macronutrients, and fiber. On ad libitum days, diet diaries revealed a difference in GI of 14 ± 1 U (mean ± standard error), with no detectable difference in energy, macronutrient, or fiber intake. No differences were observed in glucose profiles between higher and lower GI interventions in the controlled or ad libitum setting. There was significant agreement in area under the glucose curve on repeated controlled feeding days (intraclass correlation 0.75). Conclusion This study indicates that a difference in dietary GI of 14–15 U is insufficient to alter daylong glycemia as measured in interstitial fluid by the continuous glucose-monitoring system. Keywords Glycemic index Postprandial glycemia Continuous glucose monitoring Introduction Reducing the glycemic index (GI) of carbohydrate-rich foods in the diet may decrease metabolic risk [1] . The reduction and stabilization of blood glucose concentration is central to the postulated mechanism for this, and a fundamental tenet of the GI hypothesis is that incorporating low versus high GI foods into mixed meals in a whole-diet context will produce metabolically favorable glucose profiles. Reducing hyperglycemia reduces insulin demand and may have beneficial effects on insulin sensitivity, lipid profiles, and β-cell function. Reductions in hyperglycemia and hyperinsulinemia may also lower oxidative stress, which exacerbates a number of features of the metabolic syndrome, including insulin resistance, inflammation, and hypertension [2] . Because prolonged exposure to hyperglycemia can promote macro- and microvascular diseases [3] , it is important to investigate the potential of dietary strategies such as reduced GI to modulate daylong glucose levels. Several previous intervention studies have used continuous glucose-monitoring systems (CGMSs) to monitor the impact of low or high GI diets on daylong glucose profiles, with mixed findings. Brynes et al. observed reductions in 24-h area under the curve for glucose (AUC G ) after 7–10 d of consuming a low GI diet in healthy, normal-weight adults [4] and patients with type 2 diabetes [5] compared with baseline habitual diets. In the first of these studies, however, there was also a significant increase in fiber intake. No information was given for the second study in diabetics. Henry et al. [6,7] favorably altered 24-h glucose profiles in two separate studies using simple food substitutions at breakfast, lunch, and dinner: first, by substituting low for high GI breads (although the low GI bread contained more fiber/100 g than the high GI bread) [6] , and second, with beverages formulated with a low GI plant-based sweetener instead of sucrose-sweetened beverages (beverages contained equal quantities of carbohydrate) [7] . Hui et al. [8] compared 12-h glycemic profiles on single days of consuming high, medium, or low glycemic load meals (GI, carbohydrate, and fiber intakes not stated). There was no obvious dose–response effect of glycemic load on postprandial glucose peak height. Glucose profiles were relatively stable on the low glycemic load days, but there was no discernable difference between the medium and high glycemic load days, leading the researchers to suggest a threshold effect. A previous dietary intervention study carried out by our research group employed a CGMS to investigate the effects of incorporating lower or higher GI versions of carbohydrate-rich foods into the diets of overweight hyperinsulinemic women at main meals in ad libitum quantities. Despite a mean difference of 8.4 GI U in dietary intake, no differences were found in 24-h glucose profiles [9] . The present study was specifically designed to investigate the effects of incorporating lower or higher GI foods into mixed meals on 24-h CGMS glucose profiles in controlled laboratory and home settings in participants following the dietary instruction but eating ad libitum. Materials and methods Twelve overweight but otherwise healthy women were recruited from the local community to participate in a dietary intervention study at MRC Human Nutrition Research (Cambridge, United Kingdom). Potential participants were screened by telephone interview. Eligibility criteria were having a body mass index of 25–30 kg/m 2 and age of 18–65 y. Participants were excluded if they had diabetes, cardiovascular disease, hypertension, inflammatory conditions, allergy or intolerance to intervention foods, or if they were pregnant, breast-feeding, or taking medication that might interfere with glucose metabolism. They were also excluded if they had irregular eating patterns, did not habitually consume three meals per day, or if they were following a weight-reducing or low carbohydrate diet. The study was approved by the Peterborough and Fenland local research ethics committee and procedures followed were in accordance with their ethical standards. All participants gave written informed consent. Data were collected from May to December 2006. The study was a randomized, balanced, two-way crossover design of 1-wk intervention periods separated by at least 1-wk washout, and underwent an identical protocol on both intervention weeks. Participants were requested to keep activity and dietary patterns constant between intervention weeks and to refrain from excessive alcohol consumption. On the first day (D1), they attended the laboratory in the morning for measurement of height and weight. They were supplied with appropriate intervention foods for their diet period and instructed to substitute these into their habitual diets on at least three occasions per day, suggested to be once at each main meal, in the quantities they would normally consume. Quantities provided were unlimited and no further advice was given regarding other dietary components. The intervention foods were lower or higher GI versions of key staple carbohydrate-rich foods, according to diet period. These foods comprised breads, breakfast cereals, and rice, plus pasta (lower GI period only) or potatoes (higher GI period). GI values of the specific foods had previously been determined in our laboratory [10] , and GI values of equivalent foods differed by 28.5 U on average (range 20–45 U). Participants returned to the laboratory on the morning of the fifth day (D5) after an overnight fast for a controlled diet day in which all meals were provided to them. Energy, macronutrient, and fiber content of the meals were standardized across all participants and matched between lower and higher GI interventions. Meals were designed to provide a total of 8.6 MJ to meet the daily energy requirements of a woman 30–60 y old of average height (1.63 m) and with a body mass index of 28 kg/m 2 (World Health Organization, 1985). This was provided as 25% total energy at breakfast, 30% at lunch, 35% at dinner, and 10% as an evening snack. Participants were required to consume all of the food provided to them. Details of meals and nutritional composition are given in Tables 1 and 2 . Immediately upon arrival at the laboratory on D5, participants were weighed and fitted with a CGMS (CGMS System Gold; Medtronic MiniMed, Northridge, CA, USA). This was fitted and set up according to the manufacturer's instructions. The CGMS was worn for 72 h, giving 288 measurements of interstitial fluid glucose level per 24-h period. Participants were instructed on how to standardize the monitor by entering four glucose readings per 24-h period, as measured in capillary whole blood obtained by finger prick (Accu-Chek Advantage System, Roche Diagnostics Ltd., Lewes, United Kingdom). Participants were instructed to perform these readings first thing in the morning, before lunch, before dinner, and before bed. Because it was not critical that these were evenly spaced, and the period immediately after a meal was to be avoided, these times were selected as being memorable for participants. This method of glucose measurement was chosen because it was important that the participants were able to perform calibrations away from the laboratory. The glucometer used has been shown to have a correlation coefficient of 0.99 with a whole blood reference method (hexokinase) [11] and to be among the most precise of five tested, with a 6.5% deviation from the reference method [12] . Once the 1-h initialization period for the CGMS was complete, participants were provided with a fixed preweighed breakfast incorporating lower/higher GI breakfast cereal and bread as appropriate for the intervention period and were instructed to eat everything provided. Four hours later, participants were provided with a fixed lunch incorporating lower/higher GI bread, and after another 4.5 h, a fixed dinner including lower/higher GI rice. Tea or coffee was provided with meals, but otherwise only water was allowed. Participants remained sedentary during this time. After dinner, they were allowed to return home, with a lower/higher GI cereal bar to consume that evening as a snack. On the following day (D6), participants continued to wear the CGMS and followed the dietary advice regarding consumption of intervention foods at home, eating ad libitum. They recorded everything consumed during this period, using household measurements. On the seventh day (D7) of each intervention week, participants returned to the laboratory after an overnight fast and remained all day to undergo a repeat of the controlled diet protocol from D5, with identical fixed meals, fed at identical times. They returned in a fasted state on the following day (D8) for removal of the CGMS and measurement of body weight. Data analysis The CGMS data were downloaded into MiniMed Solutions 3.0B and then exported to Microsoft Excel 2000 (Microsoft Corporation, Redmond, WA, USA). Incremental AUC G was calculated using the trapezoid rule for 24 h and shorter defined periods. Food intake data were analyzed using the in-house DIDO database, which incorporates GI values obtained from measurements made at MRC Human Nutrition Research (HNR) [10] , and also from published values [13] . All statistical analyses were carried out using STATA 9.1 (STATA Corp. LP, College Station, TX, USA). The primary endpoint of total 24-h AUC G on D5 was analyzed using a mixed-effects model with a random subject effect and a fixed lower versus higher GI treatment effect. Period effects were investigated using an additional covariate. The same model was also used to analyze the secondary endpoints of 24-h AUC G on D6, plus fasting glucose concentration, 24-h mean glucose concentration, and standard deviation (SD) of mean glucose (used as an indicator of the variability of the glucose concentrations over the day) on D5 and D6. Additional secondary endpoints, analyzed in the same way, were AUC G for shorter determined periods (after meals, overnight) and postmeal maximum and minimum glucose concentrations on D5. Models assumed that the endpoint was normally distributed and data were transformed to normality where required. The repeatability of measurements between D5 and D7 of the same intervention week was assessed using Bland-Altman plots, which assess the level of agreement between two measurements, and the intraclass correlation coefficient, which measures the homogeneity within groups of replicate measurements, relative to the total variation between groups, and can be used to quantify the reproducibility of a variable. Results The mean age ± SD of participants was 46.9 ± 15.3 y and mean body mass index ± SD was 27.5 ± 2.3 kg/m 2 . Dietary data on the free-living day (D6) are presented in Table 3 . Diet diaries revealed a difference in dietary GI of 14 U, with no differences in energy or macronutrient intakes between higher and lower GI diets. There was significant agreement between 24-h AUC G on the repeated controlled feeding days (D5 and D7), with an intraclass correlation coefficient of 0.75 ( P = 0.003), and good agreement shown by the Bland-Altman plot, with just 1 of 24 points (4.17%) falling outside the limits of agreement. However, data from the CGMS were very similar on both dietary treatments. There were no significant differences in 24-h AUC G , mean 24-h glucose, or the SD of mean 24-h glucose between lower and higher GI interventions under controlled (D5) or ad libitum (D6) conditions ( Figs. 1 and 2 , Table 4 ). Neither were there any significant differences between diets in AUC G or maximum or minimum glucose concentrations during any of the intermeal intervals or overnight on D5. Discussion This study found no effect of incorporating lower or higher GI versions of key staple carbohydrate-rich foods with similar energy, macronutrient, and fiber contents into mixed diets on daylong interstitial fluid glucose profiles measured by the CGMS. There were no differences during controlled diet manipulation days in the laboratory or when participants were following dietary advice and eating ad libitum in the home setting. Comparison of the repeated days in the controlled setting revealed significant agreement of 24-h AUC G when participants were following identical diet and activity protocols, making it unlikely that background variability in the measurement was masking a difference in AUC G due to the lower or higher GI foods. This finding is consistent with that of a previous dietary intervention study in which participants wore the CGMS in the home setting during the final weeks of two 12-wk periods of lower/higher GI ad libitum dietary interventions [9] . The present study has demonstrated that incorporating lower or higher GI foods into the diet as part of mixed meals in a controlled setting does not result in differing glucose profiles, even when a larger difference in dietary GI is achieved. Participants in the present study were overweight but otherwise healthy women. Those in our previous study were overweight women selected on the basis of being hyperinsulinemic but not diabetic. It must be noted therefore that these findings are not generalizable to other populations and do not preclude the possibility of a different response to dietary GI modulation in persons with type 2 diabetes or at differing levels of metabolic disease risk. One explanation may be that lower or higher GI foods do not produce differential glucose responses when consumed as part of mixed meals. Although a number of studies have demonstrated that the glucose response to a mixed meal can be predicted reasonably accurately from the GI values of the constituent foods [14–16] , this finding is not universal [17] . The utility of the GI of single foods or even single mixed meals in quantifying the glycemic effect of whole diets has been questioned previously, with no relation observed between dietary GI and hemoglobin A 1c , a marker of average glycemia over the preceding few months [18] . Therefore, although GI values of component foods may be able to predict acute responses to single high-carbohydrate mixed meals consumed in the fasted state, and the ranking of foods with respect to GI may be unaltered when consumed as a part of similar mixed meals, the application of this to whole diets over extended periods is still unclear. A 14- to 15-point difference in GI is substantial in the context of dietary guidelines based on widely available and frequently consumed foods. However, it is possible that a larger difference in GI than was achieved in the present study is required to produce detectably different 24-h glucose profiles. A number of previous studies have made greater changes to dietary GI, but these are frequently achieved by making broader changes to the diet such as increasing intakes of fruit, vegetables, and pulses. These changes are therefore often accompanied by changes in intakes of dietary fiber, meaning that effects cannot be attributed to GI per se. Moreover, they may represent an unrealistic dietary change at a population level. The lack of a difference in 24-h glucose patterns in this study is in contrast to a previous study that used the CGMS to investigate effects of a low GI diet. Brynes et al. [4] observed reductions in fasting glucose, mean glucose, 24-h AUC, and 8-h overnight AUC on a reduced GI diet compared with the baseline habitual diet, and there was a significant correlation between change in GI calculated from food diaries and change in AUC. Although the difference in dietary GI was smaller than that achieved in the present study, fiber intakes were increased by 60% from 13.5 to 22.3 g/d. This is likely to have largely been due to increased soluble fiber intake, because participants were provided with a low GI bread high in β-glucans. Increases in fiber, particularly soluble fiber, have been associated with improvements in glucose and insulin parameters, including reductions in fasting glucose concentration [19] . This could therefore explain the discrepancy in the findings of the two studies. An alternative possibility is that the CGMS does not reflect postprandial blood glucose concentrations sufficiently accurately to detect what may be relatively small and brief differences induced by the dietary changes. The CGMS measures glucose in interstitial fluid, where glucose concentrations lag behind blood concentrations by around 10 min [20] . This represents the passage of glucose from the circulation into the interstitial space, which is under the control of transport and regulatory mechanisms. In theory, these mechanisms may buffer extreme changes and thereby blunt the acute differences in meal responses that are detectable in capillary blood. However, no significant differences were found between GI values calculated using the CGMS and the conventional method of capillary whole blood, with GI differences between foods being detectable by the CGMS [21] . Moreover, we have observed significant reductions in 24-h AUC glucose when changing the total carbohydrate intake of the diet from 60% or 55% to 45% (with concomitant increases in protein intake from 10% or 15% to 25%; unpublished data). Conclusion In summary, this study has found no differential effect of incorporating higher or lower GI versions of staple carbohydrate-rich foods into the diet on daylong interstitial fluid glucose patterns measured by CGMS. Acknowledgments The authors thank Caroline Stokes, Celia Prynne, and Anna Gent for analysis of diet diaries; Darren Cole for work on the GI database; Sue Bryant and staff of the Volunteer Suite at HNR for assistance with clinical procedures; staff at the Nutritional Biochemistry Laboratory at HNR; and Fiona Tulloch at Addenbrooke's Hospital Department of Biochemistry for sample analysis. References [1] L.M. Aston Glycaemic index and metabolic disease risk Proc Nutr Soc 65 2006 125 134 [2] L.S. Augustin S. Franceschi D.J. Jenkins C.W. Kendall C. La Vecchia Glycemic index in chronic disease: a review Eur J Clin Nutr 56 2002 1049 1071 [3] G. Orasanu J. Plutzky The pathologic continuum of diabetic vascular disease J Am Coll Cardiol 53 suppl 2009 S35 S42 [4] A.E. Brynes J. Adamson A. Dornhorst G.S. Frost The beneficial effect of a diet with low glycaemic index on 24 h glucose profiles in healthy young people as assessed by continuous glucose monitoring Br J Nutr 93 2005 179 182 [5] A.E. Brynes J.L. Lee R.E. Brighton A.R. Leeds A. Dornhorst G.S. Frost A low glycemic diet significantly improves the 24-h blood glucose profile in people with type 2 diabetes, as assessed using the continuous glucose MiniMed monitor Diabetes Care 26 2003 548 549 [6] C.J. Henry H.J. Lightowler E.A. Tydeman R. Skeath Use of low-glycaemic index bread to reduce 24-h blood glucose: implications for dietary advice to non-diabetic and diabetic subjects Int J Food Sci Nutr 57 2006 273 278 [7] C.J. Henry K.J. Newens H.J. Lightowler Low-glycaemic index sweetener-based beverages reduce 24-h glucose profiles in healthy adults J Hum Nutr Diet 22 2009 77 80 [8] L.L. Hui E.A. Nelson K.C. Choi G.W. Wong R. Sung Twelve-hour glycemic profiles with meals of high, medium, or low glycemic load Diabetes Care 28 2005 2981 2983 [9] L.M. Aston R. Laccetti A.P. Mander R. Hall C.S. Moore S.A. Jebb No difference in the 24-hour gulcose profile in interstitial fluid with manipulations to the glycaemic index of the diet Obesity Facts 2 suppl 2 2009 150 [10] L.M. Aston J.M. Gambell D.M. Lee S.P. Bryant S.A. Jebb Determination of the glycaemic index of various staple carbohydrate-rich foods in the UK diet Eur J Clin Nutr 62 2008 279 285 [11] J. Podczasy G. Flodstrom L. Doucette Clinical evaluation of the Accu-Chek Advantage blood glucose monitoring system Biosens Bioelectron 12 11 1997 iii–iii(1) [12] M. Cohen E. Boyle C. Delaney J. Shaw A comparison of blood glucose meters in Australia Diabetes Res Clin Pract 71 2006 113 118 [13] K. Foster-Powell S.H. Holt J.C. Brand-Miller International table of glycemic index and glycemic load values: 2002 Am J Clin Nutr 76 2002 5 56 [14] T.M. Wolever C. Bolognesi Prediction of glucose and insulin responses of normal subjects after consuming mixed meals varying in energy, protein, fat, carbohydrate and glycemic index J Nutr 126 1996 2807 2812 [15] T.M. Wolever D.J. Jenkins The use of the glycemic index in predicting the blood glucose response to mixed meals Am J Clin Nutr 43 1986 167 172 [16] I. Chew J.C. Brand A.W. Thorburn A.S. Truswell Application of glycemic index to mixed meals Am J Clin Nutr 47 1988 53 56 [17] A. Flint B.K. Moller A. Raben D. Pedersen I. Tetens J.J. Holst The use of glycaemic index tables to predict glycaemic index of composite breakfast meals Br J Nutr 91 2004 979 989 [18] E.J. Mayer-Davis A. Dhawan A.D. Liese K. Teff M. Schulz Towards understanding of glycaemic index and glycaemic load in habitual diet: associations with measures of glycaemia in the Insulin Resistance Atherosclerosis Study Br J Nutr 95 2006 397 405 [19] B.M. Davy C.L. Melby The effect of fiber-rich carbohydrates on features of Syndrome X J Am Diet Assoc 103 2003 86 96 [20] M.S. Boyne D.M. Silver J. Kaplan C.D. Saudek Timing of changes in interstitial and venous blood glucose measured with a continuous subcutaneous glucose sensor Diabetes 52 2003 2790 2794 [21] R. Chlup D. Jelenova P. Kudlova K. Chlupova J. Bartek J. Zapletalova Continuous glucose monitoring—a novel approach to the determination of the glycaemic index of foods (DEGIF 1)—determination of the glycaemic index of foods by means of the CGMS Exp Clin Endocrinol Diabetes 114 2006 68 74