Clinical Care/Education/Nutrition

O R I G I N A L A R T I C L E

Improved Glycemic Control and Lipid Prole and Normalized Fibrinolytic Activity on a LowGlycemic Index Diet in Type 2 Diabetic Patients
ANETTE E. JRVI, BSC BRITA E. KARLSTRM, DR MED SC YVONNE E. GRANFELDT, PHD INGER E. BJRCK, PHD NILS-GEORG L. ASP, MD, PHD BENGT O.H. VESSBY, MD, PHD

OBJECTIVE To evaluate the effects of varying the glycemic index (GI) of carbohydraterich foods on metabolic control in type 2 diabetic patients. RESEARCH DESIGN AND METHODS In a randomized crossover study, 20 patients, 5 women and 15 men, were given preweighed diets with different GIs during two consecutive 24-day periods. Both diets were composed in accordance with dietary recommendations for people with diabetes. The macronutrient composition and type and amount of dietary ber were identical. Differences in GI were achieved mainly by altering the structure of the starchy foods. RESULTS Peripheral insulin sensitivity increased signicantly and fasting plasma glucose decreased during both treatment periods. There was a signicant difference in the changes of serum fructosamine concentrations between the diets (P 0.05). The incremental area under the curve for both blood glucose and plasma insulin was 30% lower after the low- than after the high-GI diet. LDL cholesterol was signicantly lowered on both diets, with a signicantly more pronounced reduction on the low-GI diet. Plasminogen activator inhibitor-1 activity was normalized on the low-GI diet, ( 54%, P 0.001), but remained unchanged on the high-GI diet. CONCLUSIONS A diet characterized by low-GI starchy foods lowers the glucose and insulin responses throughout the day and improves the lipid prole and capacity for brinolysis, suggesting a therapeutic potential in diabetes. Diabetes Care 22:1018, 1999

ifferences in glycemic responses to various carbohydrate-rich foods are related to differences in the rate at which the carbohydrate is digested and absorbed. A low glycemic response has been reported to facilitate blood glucose regulation and to improve lipid metabo-

lism in diabetes (1,2). The glycemic index (GI) was introduced in the early 1980s as a way of ranking foods according to their glycemic effect (3). The current European dietary recommendations for patients with type 2 diabetes (4) suggest that most of the dietary energy intake should come from a

From the Clinical Nutrition Research Unit (A.E.J., B.E.K., B.O.H.V.), Department of Public Health and Caring Sciences/Geriatrics, Uppsala University, Uppsala; and the Department of Applied Nutrition and Food Chemistry (Y.E.G., I.E.B., N.-G.L.A.), Chemical Center, Lund University, Lund, Sweden. Address correspondence and reprint requests to Anette Jrvi, Department of Public Health and Caring Sciences/Geriatrics, Clinical Nutrition Research Unit, PO Box 609, S-751 25 Uppsala, Sweden. E-mail: [email protected]. Received for publication 15 April 1998 and accepted in revised form 29 September 1998. Abbreviations: apo, apolipoprotein; GI, glycemic index; I, mean insulin concentration; Lp(a), lipoprotein(a); M, mean glucose uptake; NEFA, nonesteried fatty acids; PAI-1, plasminogen activator inhibitor-1; tPA, tissue-type plasminogen activator. A table elsewhere in this issue shows conventional and Systme International (SI) units and conversion factors for many substances.

combination of carbohydrates and monounsaturated fatty acids with a special emphasis on carbohydrate-containing foods rich in soluble ber and foods with a low GI. In the healthy eating guidelines for people with diabetes in Australia and New Zealand, the GI concept is evident (5). Recent recommendations by the joint Food and Agriculture Organization (FAO)/World Health Organization Expert Consultation in a report entitled Carbohydrates in Human Nutrition (6) also support the choice of low-GI foods in healthy individuals. On the other hand, in the American Diabetes Associations Nutrition Recommendations and Principles for People with Diabetes Mellitus (7), it is stressed that the rst priority should be given to the total amount of carbohydrate consumed, rather than the source of the carbohydrate. Thus, there is still no international consensus as to the clinical usefulness of the GI concept in the dietary management of diabetes. GI has been criticized for not being applicable to mixed meals, and hence the long-term effects of low-GI foods have been questioned (810). However, there is evidence that dietary changes that involve replacement of foods with a higher GI by those with lower indices result in improved glycemic control and reduced fasting serum lipids (2,1116). Various properties of foods, e.g., particle size, botanical structure, and properties of the starch, have been found to inuence the metabolic responses to starchy foods. Processing steps in manufacturing foods may strongly inuence such food properties (17,18). The aim of this study was to evaluate the effects of two diets with pronounced differences in GI on metabolic control in type 2 diabetic patients, to establish whether the GI approach is useful in a realistic diabetic diet for a longer period of time. The carbohydrate-rich foods in both diets were basically made from the same ingredients. Differences in GI were achieved mainly by altering the botanical food structure or the chemical starch structure of the starchy foods, thus making it

10

DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999

Jrvi and Associates

Table 1Nutrient composition Nutrient Carbohydrate (g/% of energy) Starch (g) Dietary ber, corrected for resistant starch (g) Glucose (g) Fructose (g) Sucrose (g) Protein (g/% of energy) Fat (g/% of energy) Fatty acids Saturated (g/% of energy) Monounsaturated (g/% of energy) Polyunsaturated (g/% of energy)) Energy kcal kJ Low-GI diet 255/55 215 38 11 13 9 84/18 57/27 19/9 25/12 14/7 1,880 7,870 High-GI diet 239/54 205 34 10 12 8 78/18 60/29 18/9 26/12 15/7 1,820 760

diet in combination with oral antidiabetic drugs. One took metformin only and the other 15 took sulfonylurea, nine in combination with metformin. No insulin-treated subjects were included. All medication remained unchanged throughout the study. Diets The two test diets were planned with the aim of achieving large differences in the GI of the starchy foods, in particular, but with no differences in the amounts of energy, protein, fat, carbohydrate, starch, and dietary ber. The diets were composed in accordance with the previous dietary recommendations for patients with diabetes (1921). The nutrient composition of the test diets had been calculated and analyzed chemically (Table 1). According to the database from the Swedish National Food Administration, the energy derived from protein, fat, and carbohydrate was 16, 28, and 55%, respectively. Based on the sources of carbohydrate in the diets, a mean GI was calculated. On the average, GI was calculated on the basis of data for 94 2.5% (9296) of all the carbohydrates in the diets. The average GIs of the low- and high-GI diets, as expressed in rela-

Data represent the mean for all 7 days of three analyses for each diet. Chemical analyses were performed on a menu calculated to contain 2,000 kcal (8,370 kJ). Calculations of % of energy were based on the energy levels calculated from the analysis of carbohydrates, protein, and fat. Fatty acids were calculated from the relative fatty acid composition and the energy levels calculated from the analysis of carbohydrates, protein, and fat. Energy was calculated from the analysis of carbohydrates, protein, and fat.

possible to eliminate potential effects of variations in dietary ber or nutrients. RESEARCH DESIGN AND METHODS Study design In a randomized crossover study, patients with type 2 diabetes were given two diets with different food structure with either a low or a high GI during two consecutive 24-day periods. Throughout the study, the participants were free-living, and they continued their customary life and activities without any change in degree of physical activity. Admission and change of diets took place on the same day of the week. The participants started in randomized order with either of the two different diets. All patients had given their informed consent before entering the study, and the study design was approved by the Ethical Committee of the Medical Faculty of Uppsala University, Uppsala, Sweden. Subjects The study comprised 20 patients, 5 women and 15 men, with mean ages of 65 (range 5676) and 67 (5077) years, respectively. The mean body weight (mean SD) of the 20 patients on admission was 76.3 7.5 kg, with a BMI of 25.3 2.7 kg/m2. Individuals with a BMI 27 were not included. The average HbA1c on admission was 7.2 1.4%. The known duration of the diabetic

disease varied between 0.5 and 17 years. Four of the patients were treated with diet only and the other 16 were treated with

Table 2GIs of starchy foods included in the two test diets Contribution% Low-GI (carbohydrate basis) diet (ref. no.) 23 23 21 21 21 21 3 3 2 2 2 2 2 2 2 2 1 1 0.5 0.5 58 (22) 100 (22) 67 (23) 100 (23) 65* 86* 35 (24) 98 (25) 36 (26) 70 40 (26) 74 36 (30) 70 (30) 40 74 68* 96* 50* 81* High-GI diet (ref. no.)

Starchy food Whole grain barley bread Whole meal barley bread White durum pasta White durum bread Parboiled rice Sticky rice Whole grain barley porridge Whole meal barley porridge Whole red lentils Ground red lentils Whole white beans Ground white beans Whole red kidney beans Ground red kidney beans Whole brown beans Ground brown beans Ordinary maize starch muffins High-amylose maize starch muffins Ordinary maize our High-amylose maize our

The structure of the starchy foods was changed with the purpose of varying the GI (with white bread as reference) of the diets while maintaining an identical gross nutrient composition. All other dietary constituents were kept identical in both test diets. *GI predicted from in vitro assay according to Granfeldt et al. (30). Estimated assuming an increase in GI due to milling, as in red kidney beans. Properties assumed to be similar to those in the corresponding white bean product (27). Ordinary maize products are 25% amylose; high-amylose products are 70% amylose.

DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999

11

Lowglycemic index diet and type 2 diabetes

tion to that of white wheat bread, were 56.8 3.6 (52.960.9) and 82.7 3.1 (76.985.1) U, respectively. The value for the diet with the low GI was on the average 31% (26.2 3.8 U) lower than that for the high-GI diet. GI values were calculated on the foodstuffs, and their GI values are given in Table 2. The diets were based on common foods, and both diets were basically prepared from the same ingredients. The difference in GI was mainly achieved by manipulating the structure of the starchy foods (Table 2). For most food items, the GIs had previously been measured in healthy subjects (2227) or derived by using data from the literature (28,29). With regard to foods for which GI data were missing, the GIs were predicted by using an enzymatic in vitro procedure (30). This method is based on enzyme incubation of chewed samples and allows evaluation of products as eaten. A hydrolysis index is calculated from the area under the hydrolysis graph with the test product and expressed as a percentage of the corresponding area with white bread as a reference. Hydrolysis indices thus calculated have been shown to predict GIs for a large number of cereal- and legume-based foods with good accuracy (r = 0.877, P 0.005) (30). Except for the foodstuffs with different GIs, all other ingredients in the two diets were kept identical. Both diets were based on a 1-week menu that was repeated during each test period. Four different energy levels were used: 1,600 kcal (6.7 MJ), 2,000 kcal (8.4 MJ), 2,400 kcal (10.1 MJ), and 2,800 kcal (11.8 MJ). The relative contents of nutrients were identical at different levels of energy intake. Dietary instructions were given by a dietitian before entry into the study. The participants were provided with all the food throughout the study period. The food was individually prepared in a metabolic kitchen, and the estimated energy intake of the participants was calculated on the basis of their initial body weight in an attempt to keep the body weight constant throughout both diet periods. The female patients received 30 kcal (125 kJ) per kilogram of body weight, and the male patients, 35 kcal (145 kJ). If a patient gained or lost weight, the energy level was adjusted. An example of a menu for 1 day on the lowand the high-GI diets, respectively, is given in Table 3. The food was collected by the subjects three times a week. In a written 1week menu, they were told which dishes and food items should be eaten for breakfast, lunch, dinner, and in-between meals. All prepared dishes were ready to be heated.
12

Table 3Menu for 1 day on the low- and high-GI diets at the 2,000-kcal level Meal Breakfast Pasta porridge: 73 g durum pasta* 18 g margarine 30 g strawberries 3 g sugar 150 ml low-fat milk Lunch 54 g durum pasta* Salmon sauce: 5 g our 10 g margarine 100 ml low-fat milk 50 g salmon 10 g leek 45 g green beans 45 g bread made of whole barley seeds* 3 g margarine 105 g orange Dinner Chili con carne: 40 g white beans, whole* 50 g minced meat 30 g onion 3 g garlic 60 g tomato, crushed 10 g tomato, puree 15 g margarine 47 g parboiled rice* Tomato salad: 65 g tomato 5 g canola oil 2 g vinegar Snack 75 g bread made of whole barley seeds* 6 g margarine 150 ml low-fat milk Apple rice: 32 g parboiled rice* 5 g honey 40 g apple 5 g lemon juice
*Foods with different GI values.

Low-GI day

High-GI day 127 g bread made from durum wheat* 18 g margarine 30 g strawberries 3 g sugar 150 ml low-fat milk 95 g bread made from durum wheat* Salmon sauce: 5 g our 10 g margarine 100 ml low-fat milk 50 g salmon 10 g leek 45 g green beans 45 g bread made of ground barley seeds* 3 g margarine 105 g orange Chili con carne: 40 g white beans, ground* 50 g minced meat 30 g onion 3 g garlic 60 g tomato, crushed 10 g tomato, puree 15 g margarine 47 g sticky rice* Tomato salad: 65 g tomato 5 g canola oil 2 g vinegar 75 g bread made of ground barley seeds* 6 g margarine 150 ml low-fat milk Apple rice: 34 g sticky rice* 5 g honey 40 g apple 5 g lemon juice

Instructions were given on how to store and prepare the food. The participants were requested to eat all the food that they received. No other food items except water, mineral water, coffee, and tea were allowed. If, for some reason, a participant was not able to eat all of the food, he or she was instructed to return the food to the meta-

bolic ward to be weighed and subtracted from the calculated amount in the diet. Chemical analysis of the diets For analysis of the nutrient composition, duplicate portions for each day were prepared separately on three occasions. Each portion was homogenized with water, and

DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999

Jrvi and Associates

samples were removed and stored at 18C. For measurements of carbohydrates, dietary ber, and protein, the samples were freeze-dried. The content of insoluble and soluble dietary ber was analyzed by a gravimetric method according to Asp et al. (31,32). Starch was determined as described by Holm et al. (33), and the other carbohydrates were measured by enzymatic methods (Roche Diagnostics, Mannheim, Germany). The total fat content and fatty acid composition were also determined. A quantity of 10 g of the homogenate (see above) was extracted with 50 ml methanol, 100 ml chloroform, and 150 ml of 0.2 mol/l sodium dihydrogen phosphate; 10 ml of the chloroform phase was evaporated to dryness; and the total fat content was determined by weighing. Another aliquot of the chloroform phase was used for determination of the fatty acid composition. The fatty acids were transesteried to methyl esters and analyzed by gas-liquid chromatography as described above (34). Laboratory methods The participants were admitted to a metabolic ward for 2 days at the beginning and end of each dietary period for laboratory measurements. On the 1st day, blood samples were drawn after an overnight fast for determination of plasma glucose, serum fructosamine, plasma insulin, C-peptide, serum lipoproteins, serum apolipoproteins, nonesteried fatty acids (NEFAs), fatty acid composition of the serum cholesterol esters, serum tocopherols, and plasminogen activator inhibitor-1 (PAI-1) activity. In addition, blood samples were drawn at xed time points throughout the day from 0730 to 1800 (prole day) for assay of plasma glucose and plasma insulin at 60, 120, 180, 240, 300, 360, 480, and 540 min after breakfast and of C-peptide, serum cholesterol, serum triglycerides, and NEFA concentrations at 120 and 300 min after breakfast. The subjects followed the study diets on the prole day, with breakfast at 0800, lunch at 1215, dinner at 1615, and snacks at 1000 and 1400. On the 2nd day, the insulin sensitivity was evaluated by a euglycemic-hyperinsulinemic clamp technique. The subjects were weighed three times a week in indoor clothing without shoes. Plasma glucose concentrations were determined by the glucose oxidase method (35). The determination of serum fructosamine was performed using a Roche reagent kit (07-366694) (Roche, Basel), based on nitroblue tetrazolium reduction in

alkaline medium, in a Cobas Bio centrifugal analyzer. Plasma insulin assays were performed by the enzyme-immunological test for quantitative determination of human insulin in vitro (enzyme-linked immunosorbent assay/1-step sandwich assay; Boehringer Mannheim Immunodiagnostics for the ES300). Plasma C-peptide was measured according to the method of Heding (36). NEFAs were determined with a Wako NEFA C-kit (994-75409) (Wako, Neuss, Germany), modied for use in a Monarch apparatus (Instrumentation Laboratories, Lexington, MA). Lipoprotein concentrations in serum were determined after an overnight fast. VLDL, LDL, and HDL were isolated by a combination of preparative ultracentrifugation (37) and precipitation with a sodium phosphotungstate and magnesium chloride solution (38). Triglyceride and cholesterol concentrations were measured in serum and in the isolated lipoprotein fractions by enzymatic methods using the IL Test cholesterol method 181618-10 and the IL Test triglyceride enzymatic-colorimetric method 181610-60 in a Monarch apparatus (Instrumentation Laboratories). The concentrations of serum apolipoprotein (apo) A-1 and B were determined by immunochemical assay (Orion Diagnostica, Espoo, Finland) in a Monarch apparatus (Instrumentation Laboratories). Lipoprotein(a) [Lp(a)] was measured by a Pharmacia apo(a) radioimmunoassay method (Pharmacia, Uppsala, Sweden). This is based on the direct sandwich technique, in which two monoclonal antibodies are directed against separate antigenic determinants on the apo(a) in the sample. The concentration is expressed in U/l. According to the manufacturer, 1 U of apo(a) is approximately equal to 0.7 mg Lp(a). The fatty acid compositions of the serum cholesterol esters were determined by gas-liquid chromatography as described earlier (34). The concentrations of - and -tocopherol in serum were assayed by high performance liquid chromatography using a uorescence detector, as described by hrvall et al. (39). PAI-1 activity in plasma was measured with Spectrolyse/pL kits from Biopool AB (Ume, Sweden), using polylysine as a stimulator (40). The euglycemichyperinsulinemic clamp technique according to DeFronzo et al. (41) was used as described in detail by Pollare et al. (42). Insulin sensitivity is expressed as the insulin sensitivity index (M/I60120), which is a measure of the tissue sensitivity to insulin expressed per unit of insulin obtained by

dividing the mean glucose uptake (M) by the mean insulin concentration (I) during the last 60 min of the clamp study. Statistical analyses The analyses take into account the design of the experiment, the scales, and the distributions of variables (43). For continuous variables with normal distributions (for some variables after logarithmic transformation), an analysis of variance model with factors for treatment, participant, and time was used. Comparisons were made of the two dietary periods, and each dietary period was also compared with the results on admission. The mean value of period one and period two minus the baseline value was used as the basis of a test of different carryover effects between the two treatments. The test was performed as a t test between the two sequences. For continuous variables with normal distributions, a t test for different carryover effects at the 10% level was used. If this test was not signicant, a t test for different treatment effects at the 5% level was used. If the carryover tests were signicant, only data from the rst dietary period were used in comparisons of treatment effects. If the usual assumptions for t tests did not hold, or if the data were on an ordinal scale, the t test was replaced by the Mann-Whitney U test. Since no carryover effects were identied, data from both periods with a high-GI diet were added, irrespective of the order of treatment, as were both periods with a low-GI diet. An analysis of covariance with change in body weight as a covariate was carried out to separate the effects of treatment from those of changes in body weight. Some of the data were unbalanced, since a few values were missing. In consequence, the results are presented as least-square means, which form the basis of the statistical tests and estimates in the analyses and take the imbalance into account. RESULTS Body weight The intention was to keep the subjects weight stable during the entire study. Adjustments in the energy levels were therefore made if there were any changes in body weight. Owing to the large volume of the diets, it was not always possible to completely adjust the energy intake to avoid a reduction of the body weight. Consequently, there was a slight and similar reduction in body weight during both diet
13

DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999

Lowglycemic index diet and type 2 diabetes

Table 4Effects on glucose and insulin metabolism after 3 weeks on low- and high-GI diets Baseline Fasting plasma glucose (mmol/l) HbA1c (%) Fructosamine (mol/l) Fasting plasma insulin (mU/l) Insulin sensitivity (M/I) 10.3 3.2 7.2 1.4 353 78 13.6 8.4 4.3 2.8 Low GI 8.8 3.1 6.7 1.3 347 72 12.5 4.5 5.6 2.8 % Change 14.6* 5.9 1.8 8.3 30 High GI 9.0 3.1 6.9 1.3 356 75 12.7 5.0 5.2 2.5 % Change 13.3* 3.6 0.8 6.3 21 P 0.471 0.107 0.050 0.930 0.305

Data are means SD. P includes values for differences between the low- and high-GI diets. Signicant changes during the dietary periods when compared with baseline: *P 0.001; P 0.01; P 0.05.

periods, 1.5 7.21 kg on the high-GI diet and 1.4 7.08 kg on the low-GI diet. All results presented have therefore been adjusted for changes in body weight. Glucose and insulin metabolism The fasting plasma glucose concentration fell signicantly by 14% (P 0.001) during both the low- and the high-GI period (Table 4). The incremental area under the plasma glucose response curve during the 9-h prole day was 31% lower (P 0.05) after the period with the low-GI diet than after the high-GI period (Fig. 1). The changes in the serum fructosamine concentrations differed signicantly (P = 0.05) between the two diet periods (Table 4). The incremental area under the curve for plasma insulin during the day was 27% lower (P 0.01) after the low-GI diet (Fig. 2), while the peripheral insulin sensitivity measured by the clamp technique (M/I) increased signicantly during both periods (Table 4). The C-peptide levels were signicantly higher after the high-GI diet, compared with the low-GI diet, at 120 (P = 0.001) and 300 (P 0.01) min after breakfast. Serum lipoprotein lipids The serum cholesterol concentration was markedly reduced after both dietary periods when compared with the value on admission (Table 5).The reduction in subjects on the low-GI diet was signicantly more pronounced than that in those on the high-GI diet ( 5%, P 0.01). Serum triglycerides were reduced to the same extent on both diets. HDL cholesterol levels were also decreased after both diet periods. LDL cholesterol was reduced by 29% (P 0.001) after the low-GI diet and by 22% (P 0.001) after the high-GI diet when compared with that on admission. When comparing the two periods, LDL cholesterol was 8% (P 0.01) lower in the low-GI diet than in the high-GI diet (Table 5).
14

Serum apolipoprotein concentration The concentration of apoB was markedly decreased after both diet periods, signicantly more after the low- ( 24%) than after the high- ( 19%) GI period. The serum concentration of apoA-1 was also signicantly decreased after both periods, slightly more after the period with the lowGI diet. The concentrations of Lp(a) decreased to a similar extent during both diet periods (Table 5). NEFA concentration There were no changes in the fasting values of NEFA, but there were signicant differences between the dietary periods during the day, the NEFA levels being 40% higher at 120 and 180 min and 31% lower at 300 min on the low-GI period compared with the high-GI period (Fig. 3). Fatty acid composition of the serum cholesterol esters The fatty acid composition of the serum cholesterol esters was similar after the two dietary periods. Compared with admission,

the proportions of -linoleic acid and docosahexanoic acid had increased, and those of palmitoleic and stearic acid had decreased, indicating good compliance to both diets (data not shown). Serum tocopherols The serum tocopherol concentration changed signicantly and to about the same extent during both treatment periods. The concentration of -tocopherol was increased after both dietary periods: by 9% (P 0.05) after the low-GI period, and by 14% (P 0.05) after the high-GI period, while that of -tocopherol was increased by 43% (P 0.001) after the low- and 41% (P 0.001) after the high-GI period. PAI-1 activity PAI-1 activity decreased substantially on the low-GI diet by 58% (P 0.01), but remained unchanged on the high-GI diet. When comparing the two periods, the PAI1 activity was 53% (P 0.001) lower after the low-GI than after the high-GI diet. (Fig. 4).

Figure 1Increments of plasma (P) glucose in patients with type 2 diabetes after the period of lowGI diet ( ) compared with that after the high-GI diet ( ). Areas under the curve (area units) for the low- and the high-GI diet were 1,495 671.8 and 2,124.8 1,468.96, respectively. P ** 0.01, *** P 0.001.
DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999

Jrvi and Associates

suggested that the GI of the diet in free-living subjects with type 2 diabetes is distributed with a mean of 85 5 U (7097.8) (46). This is comparable with the GI of the high-GI diet in the present study. The present study conrms and extends the results of some of the previous long-term studies evaluating the effects of a similar GI reduction (2,13,15,44). The difference in the area under the day-long curve for plasma glucose and plasma insulin corresponded very well with the predicted difference in mean GI, which was 26 U. The higher NEFA levels on the low, as compared with the high, GI diet 2 and 3 h after breakfast are probably secondary to a lower plasma insulin concentration, which, however, cannot explain the Figure 2Increments of plasma (P) insulin in patients with type 2 diabetes after the period of lowGI diet ( ) compared with that after the high-GI diet ( ). Areas under the curve (area units) for the lower NEFA value on the low-GI diet after 5 low- and the high-GI diet were 12,927 6,729.2 and 17,724 10,318.0, respectively. P 0.05, * ** P h. Compared with the value on admission, 0.01, ***P 0.001. serum lipoprotein lipid and apolipoprotein concentrations were substantially reduced CONCLUSIONS Our study shows ber, which apparently differed from the on both diets. This study shows that a that a strictly controlled diet for people habitual diet of the participants. The reduction of GI from a level comparable to with diabetes, with a low GI for the major improvement in insulin sensitivity was that of a diabetic diet, according to the curstarchy foods, results in a considerably more accentuated, however, on the low-GI rent recommendations of the American Diaimproved metabolic prole when com- diet ( 30%) than on the high-GI diet betes Association (21), is associated with a pared with a corresponding high-GI diet. ( 21%). In this study, the average GIs of signicantly more pronounced reduction of The glucose and insulin responses through- the low- and high-GI diets were 57 and 83 LDL cholesterol and apoB. The extent of the out the day were lower, the LDL cholesterol U, respectively. Compared with many con- total reduction of serum and LDL cholesterol level was decreased, and the PAI-1 activity trolled dietary studies using the GI concept, compared with that on admission is compalevels were normalized after the period the GI of our low-GI diet was even further rable to that obtained by treatment with with the low-GI diet. There was a dramatic reduced (2,12,13,44). hydroxy-methyl-glutaryl (HMG)-CoA reducreduction in plasma glucose, serum choIn a meta-analysis of 11 medium- to tase inhibitors, such as simvastatin (47) and lesterol, and LDL cholesterol, as well as an long-term studies using the GI approach, pravastatin (48), which lowered plasma choincrease in insulin sensitivity, even after the Miller (45) concluded that the GI of the dia- lesterol by 25 and 20% and LDL cholesterol diet period with the high GI. This is prob- betic diet was, on average, lowered from 91 by 35 and 26%, respectively, in recent trials. ably due to the fact that both diets were to 75 (white bread = 100). In one of the This is the rst time that lowering the composed in accordance with the dietary more successful studies, the GI value was dietary GI has been shown to improve the recommendations (1921), i.e., low in fat, reduced to 52, which is comparable with the capacity for brinolysis. Earlier studies modied in fat quality, and high in dietary value for our low-GI diet (15). It has been have demonstrated that dietary changes
Table 5Serum lipoprotein and serum apo concentrations at baseline and after 3 weeks on diets with low- and high-GI diets Baseline Serum cholesterol (mmol/l) Serum triglycerides (mmol/l) HDL cholesterol (mmol/l) HDL triglycerides (mmol/l) VLDL cholesterol (mmol/l) VLDL triglycerides (mmol/l) LDL cholesterol (mmol/l) LDL triglycerides (mmol/l) LDL/HDL cholesterol ApoA-1 (mg/dl) ApoB (mg/dl) Apo(a) (U/l) 5.79 0.78 1.80 1.00 1.06 0.26 0.10 0.05 0.56 0.48 1.28 0.98 4.03 0.78 0.42 0.10 3.96 1.15 125.8 16.24 104.3 16.25 39.8 296.2 Low GI 4.23 0.73 1.25 0.58 0.88 0.28 0.09 0.06 0.37 0.21 0.94 0.47 2.87 0.70 0.33 0.09 3.66 1.57 99.3 17.95 78.9 15.61 307.0 329.61 % Change 27* 30* 17 10 34 27 29* 20* 8 21* 24* 28* High GI 4.46 0.87 1.22 0.57 0.87 0.27 0.07 0.04 0.41 0.27 0.99 0.57 3.13 0.90 0.34 0.09 3.84 1.24 102.5 15.56 84.3 14.67 304.2 335.36 % Change 23* 32* 19 35 27 23 22* 18 3 19* 19* 27* P 0.002 0.877 0.700 0.086 0.494 0.117 0.003 0.573 0.121 0.036 0.006 0.222

Data are means SD. P includes values for differences between the low- and high-GI diets. Signicant changes during the dietary periods when compared with baseline: *P 0.001; P 0.01.

DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999

15

Lowglycemic index diet and type 2 diabetes

of the diets and their effect on satiety. The participants could not eat any more, and they stated that they had never eaten so much before without gaining weight. This demonstrates that the diet usually eaten by many diabetic patients probably is relatively energy dense, with more dietary fat, and low in dietary ber. When it comes to the practical application of the GI concept in the management of diabetes, it is important to look at the diet as a whole. The GI is useful only when comparing different foodstuffs within the same food group and should not be used as an isolated concept. As judged from the present study, advice concerning fat content and fat quality should be compleFigure 3 Responses of serum (s) NEFAs after two diets with different food structures and GIsmented in by advice regarding the carbohypatients with type 2 diabetes. This gure represents admission ( ) compared with the low-GI diet ( ) drate quality, in order to prevent and the high-GI diet ( ). *P 0.05, **P 0.01, ***P 0.001. cardiovascular disease. To make the choice easier for the patients, this advice should such as a reduction of the fat intake and an high GIs of the carbohydrate-rich foodstuffs focus on food items and their effects. Peoincrease in ber intake may improve the were mainly achieved by manipulating the ple with diabetes should thus be encourbrinolytic activity. Consequently, a longer structure of the starchy foods. The results aged to eat traditional foods with low GIs, period on a low-fat high-ber diet emphasize the importance of food form such as pasta, parboiled rice, whole grain markedly increased the plasma tissue-type (17,18). However, although the test diets breads, and different kinds of beans. Today plasminogen activator (tPA) (49,50). In an were based on identical ingredients, the the choice of low-GI foods is limited, espeepidemiological study (51), a high intake of low-GI foods can be expected to contain cially when it comes to breakfast products fruit and vegetables was associated with somewhat larger amounts of resistant starch such as bread and cereal products. There is reduced PAI-1 levels. When the normal (57). Like dietary ber, resistant starch will therefore a need for development of new diet was enriched with dietary ber, such as be fermented by the colonic microora. The foods with a low GI. The food industry has guar gum and oat husk, there was a metabolic effects of the short-chain fatty the ability as well as the responsibility to decrease in the PAI-1 activity (52,53). acids formed remain to be established. modify the GI of different products and to According to Boman et al. (54), a low However, in a study by Thorburn et al. create low-GI foods. dietary ber intake appears to be associated (58), increased colonic fermentation was The present data indicate that the GI is with high PAI-1 activity. It should be associated with improved glucose tolerance. a useful concept in the management of diastressed that there were no differences in In our study, there was a certain reduc- betes. In addition to improving glucose the amount and source of dietary ber tion in body weight after both dietary peri- homeostasis, a low-GI diet seems to between the low and high-GI diets in the ods. It was almost impossible to adjust the improve lipid metabolism, and is therefore present study. In one study, a diet high in energy levels to ensure total weight stabil- also useful in the management of hypercomplex carbohydrate and dietary ber ity, presumably because of the large volume lipidemia. In addition, this study has also and low in fat showed highly signicant reductions in the levels of plasminogen, tPA, and PAI-1. However there was no information about the source of the carbohydrates (55). A positive association between the VLDL triglyceride concentration, plasma insulin level and PAI-1 activity has been suggested (56). In this study, there were no differences in the triglyceride concentrations at the end of the two dietary periods, while the incremental area under the curve for plasma insulin during a characteristic day on the low-GI diet was signicantly lower. The results thus suggest that the reductions in PAI-1 activity may have been due to the reduced insulin levels rather than to decreased triglyceride levels. Figure 4Comparison of the effects on the activity of PAI-1 of two diets with different food structures The low and high-GI diets were basi- and GIs in patients with type 2 diabetes. PAI-1 activity values on admission and after the low- and highcally made from the same ingredients. The GI diets were 22.2 14.7, 9.4 11.0, and 20.2 16.1, respectively. P **0.01, ***P 0.001.
16 DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999

Jrvi and Associates

shown that a diet with a low GI improves the capacity for brinolysis. A low-GI diet thus has several advantages in the treatment of insulin resistance syndrome.
Acknowledgments This study was supported by grants from the Swedish Council for Forestry and Agricultural Research, the Swedish Diabetes Association, the Swedish Nutrition Foundation, the Cerealia Foundation for Research and Development, the Ernfors Foundation, and the School of Household Management Fund. References 1. Jenkins DJA, Ghafari H, Wolever TMS, Taylor RH, Jenkins AL, Barker HM, Fielden H, Bowling AC: Relationship between the rate of digestion of foods and post-prandial glycaemia. Diabetologia 22:450455, 1982 2. Jenkins DJA, Wolever TMS, Buckley G, Lam KY, Guidici S, Kalmusky J, Jenkins AL, Patten RL, Bird J, Wong GS, Josse RG: Low glycemic-index starchy foods in the diabetic diet. Am J Clin Nutr 48:248254, 1988 3. Jenkins DJA, Wolever TMS, Taylor RH, Barker H, Fielden H, Baldwin JM, Bowling AC, Newman HC, Jenkins AL, Goff DV: Glycemic index of foods: a physiological basis for carbohydrate exchange. Am J Clin Nutr 34:362366, 1981 4. Diabetes and Nutrition Study Group of the European Association for the Study of Diabetes: Recommendations for the nutritional management of patients with diabetes mellitus. Diabetes Nutr Metab 8:14, 1995 5. International Diabetes Institute: Diabetes: Eating for Health . Melbourne, International Diabetes Institute, 1994 6. Carbohydrates in Human Nutrition: report of a Joint FAO/World Health Org. Expert Consultation. FAO Food Nutr Pap66:1 140, 1998 7. American Diabetes Association: Nutrition recommendations and principles for people with diabetes mellitus. Diabetes Care 19 (Suppl. 1):1619, 1996 8. Jenkins DJA, Jenkins AL: Nutrition principles and diabetes: a role for lente carbohydrate? Diabetes Care18:14911498, 1995 9. Hollenbeck CB, Coulston AM: The clinical utility of the glycemic index and its application to mixed meals. Can J Physiol Phar macol69:100107, 1991 10. Coulston AM, Hollenbeck CB, Liu GC, Williams RA, Starich GH, Mazzaferri EL, Reaven GM: Effect of source of dietary carbohydrate on plasma glucose, insulin and gastric inhibitory polypeptide responses to test meals in subjects with non-insulindependent diabetes mellitus. Am J Clin Nutr 40:965970, 1984 11. Brand JC, Colagiuri S, Crossman S, Allen A,

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Roberts DCK, Truswell AS: Low-glycemic index foods improve long-term glycemic control in NIDDM. Diabetes Care 14:95101, 1991 Wolever TMS, Jenkins DJA, Vuksan V, Jenkins AL, Buckley GC, Wong GS, Josse RG: Benecial effect of a low-glycaemic index diet in type 2 diabetes. Diabet Med 9:451458, 1992 Wolever TMS, Jenkins DJA, Vuksan V, Jenkins AL, Wong GS, Josse RG: Benecial effect of low-glycemic index diet in overweight NIDDM subjects. Diabetes Care 15:562564, 1992 Fontvieille AM, Rizkalla SW, Penfornis A, Acosta M, Bornet FRJ, Slama G: The use of low glycemic index foods improves metabolic control of diabetic patients in a 10 week study. Diabet Med9:444450, 1992 Jenkins DJA, Wolever TMS, Kalmusky J, Giudici S, Giordano C, Wong GS, Bird JN, Patten R, Hall M, Buckley G, Little JA: Low glycemic index carbohydrate foods in the management of hyperlipidemia. Am J Clin Nutr 42:604617, 1985 Jenkins DJA, Wolever TMS, Kalmusky J, Giudici S, Giordano C, Patten R, Wong GS, Bird JN, Hall M, Buckley G, Csima A, Little JA: Low glycemic index diet in hyperlipidemia: use of traditional starchy foods. Am J ClinNutr 46:6671, 1987 Brand JC, Nicholsson PL, Thorburn AW, Truswell AS: Food processing and the glycemic index. Am J Clin Nutr 42:11921196, 1985 Behall KM, Scholeld DJ, Canary J: Effect of starch structure on glucose and insulin responses in adults. Am J Clin Nutr 47:428432, 1988 Diabetes and Nutrition Study Group of the European Association for the Study of Diabetes: Nutritional recommendations for individuals with diabetes mellitus. Diabetes Nutr Metab1:145149, 1988 Nutrition subcommittee of the British Diabetic Associations Professional Advisory Committee: Dietary recommendations for people with diabetes: an update for the 1990s. Diabet Med9:189202, 1992 American Diabetes Association: Nutritional recommendations and principles for individuals with diabetes mellitus (Position Statement). Diabetes Care 14 (Suppl. 2): 2027, 1991 Liljeberg H, Granfeldt Y, Bjrck I: Metabolic responses to starch in bread containing intact kernels versus milled our. Eur J Clin Nutr 46:561575, 1992 Granfeldt Y, Bjrck I, Hagander B: On the importance of processing conditions, product thickness and egg addition for the glycemic and hormonal responses to pasta: a comparison with bread made from pasta ingredients. Eur J Clin Nutr45:489499, 1991 Granfeldt Y, Liljheberg H, Drews A, New-

25.

26.

27.

28. 29. 30.

31. 32.

33. 34.

35.

36. 37.

38.

39.

man R, Bjrck I: Glucose and insulin responses to barley products: inuence of food structure and amylose-amylopectin ratio. Am J Clin Nutr59:10751082, 1994 Liljeberg H, Granfeldt Y, Bjrck I: Products based on a high ber barley genotype, but not common barley or oats, lower postprandial glucose and insulin responses in healthy humans. J Nutr 126:458466, 1996 Tovar J, Granfeldt Y, Bjrck I: Effect of processing on blood glucose and insulin responses to starch in legumes. J Agric Food Chem40:18461851, 1992 Tovar J, Bjrck I, Asp N-G: Starch content and a-amylosis rate in precooked legume ours. J Agric Food Chem 38:18181823, 1990 Jenkins DJA, Wolever TMS, Jenkins AL: Starchy foods and glycemic index. Diabetes Care 11:149159, 1988 Foster-Powell K, Miller JB: International tables of glycemic index. Am J Clin Nutr 62: 871S890S, 1995 Granfeldt Y, Bjrck I, Drews A, Tovar J: An in vitro procedure based on chewing to predict metabolic response to starch in cereal and legume products. Eur J Clin Nutr 46:649660, 1992 Asp N-G: Denition and analysis of dietary bre. Scand J Gastr oenterol 22 (Suppl. 129):1620, 1987 Asp N-G, Johansson CG, Hallmer H, Siljestrm M: Rapid enzymatic assay of soluble and insoluble dietary bre. J Agric Food Chem31:476482, 1983 Holm J, Bjrck I, Drews A, Asp N-G: A rapid method for the analysis of starch. Strke 38:224226, 1986 Boberg M, Croon L-B, Gustafsson I-B, Vessby B: Platelet fatty acid composition in relation to fatty acid composition in plasma and to serum lipoprotein lipids in healthy subjects with special reference to the linoleic pathway. Clin Sci 68:581587, 1985 Hjelm M, de Verdier CH: A methodological study of the enzymatic determination of glucose in blood. Scand J Clin Lab Invest 15:415428, 1963 Heding LG: Radioimmunological determination of human C-peptide in serum. Dia betologia 11:541548, 1975 Havel RJ, Eder HA, Bragdon JH: The determination and chemical composition of ultracentrifugally separated lipoprotein in human serum. J Clin Invest34:13451353, 1955 Siegler L, Wu WT: Separation of serum high-density lipoprotein for cholesterol determination: ultra-centrifugation vs precipitation with sodium phospho-tungstate and magnesium chloride. Clin Chem 27:834841, 1981 hrvall M, Tengblad S, Vessby B: Lower tocopherol serum levels in subjects with
17

DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999

Lowglycemic index diet and type 2 diabetes

abdominal adiposity. J Int Med234:5360, 1993 Eriksson E, Rnby M, Gyzander E, Risberg B: Determination of plasminogen activator inhibitor in plasma using t-PA and a chromogenic single pointpoly-D-lysine stimu- 47. lated assay. ThrombosisRes 50:90101, 1988 DeFronzo RA, Tobin JD, Andres R: Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Phys iol 237:E214E223, 1979 48. Pollare T, Lithell H, Selinus I, Berne C: Application of prazosin is associated with an increase of insulin sensitivity in obese patients with hypertension. Diabetologia 31:415420, 1988 Jones B, Kenward MG: Monographs on Sta tistics and Applied Probability 34: Design and49. Analysis of Cross-OverrT ials. London, Chapman and Hall, 1989 Jenkins DJA, Wolever TMS, Collier GR, Ocana A, Rao AV, Buckley G, Lam Y, Mayer A, Thompson LU: Metabolic effects of a low 50. glycemic index diet. Am J Clin Nutr 46:968975, 1987 Miller JCB: Importance of glycemic index in diabetes. Am J Clin Nutr 59 (Suppl. 3):747S752S, 1994 Wolever TMS, Nguyen P-M, Chiasson J-L, Hunt JA, Josse RG, Palmason C, Rodger 51. NW, Ross SA, Ryan EA, Tan MH: Determinants of diet glycemic index calculated retrospectively from diet records of 342 individuals with non-insulin-dependent diabetes mellitus. Am J Clin Nutr 59:12651269, 1994 The Scandinavian Simvastatin Survival Study Group: Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet 8934:13831389, 1994 Shepherd J, Cobbe SM, Ford I, Isles CG, Lorimer AR, Macfarlane PW, McKillop JH, Packard CJ, the West of Scotland Coronary Prevention Study Group: Prevention of coronary heart disease with pravastatin in men with hypercholesterolemia. N Engl J Med 333:13011307, 1995 Marckmann P, Sandstrm B, Jespersen J: Favorable long-term effect of a lowfat/high-ber diet on human blood coagulation and brinolysis. Arterioscler Thr omb 13:505511, 1993 Marckmann P, Sandstrm B, Jespersen J: Low-fat, high-ber diet favorably affects several independent risk markers of ischemic heart disease: observations on blood lipids, coagulation and brinolysis from a trial of middle-aged Danes. Am J Clin Nutr 59:935939, 1994 Nilsson TK, Sundell IB, Hellsten G, Hallmans G. Reduced plasminogen activator inhibitor activity in high consumers of fruits, vegetables and root vegetables. J Int Med227:267271, 1990 Landin K, Holm G, Tengborn L, Smith U: Guar gum improves insulin sensitivity, blood lipids, blood pressure, and brinolysis in healthy men. Am J Clin Nutr 56:10611065, 1992 Sundell-Rnby B: Food intake, brinolysis and risk factors for cardiovascular disease . PhD thesis. Ume, Sweden, University of Ume, 1993 Boman K, Hellsten G, Bruce , Hallmans G, Nilsson TK: Endurance physical activity, diet and brinolysis. Atherosclerosis 106:6574, 1994 Mehrabian M, Peter JB, Barnard RJ, Lusis AJ: Dietary regulation of brinolytic factors. Atherosclerosi84:2532, s 1990 Hamsten A, Eriksson P: Fibrinolysis and atherosclerosis. Baillir es Clin Hematol 8:34563, 1995 Jenkins DJA, Jenkins AL, Wolever TMS, Collier GR, Rao AV, Thompson LV: Starchy foods and ber: reduced rate of digestion and improved carbohydrate metabolism. Scand J Gastr oent22 (Suppl. 129):132144, 1987 Thorburn A, Muir J, Proietto J: Carbohydrate fermentation decreases hepatic glucose output in healthy subjects. Metabolism 42:780785, 1993

40.

52.

41.

53.

42.

54.

55. 56. 57.

43.

44.

45. 46.

58.

18

DIABETES CARE, VOLUME 22, NUMBER 1, JANUARY 1999