点击显示 收起
1 From the Department of Nutritional Sciences, Faculty of Medicine, and the Department of Statistics, University of Toronto; the Kellogg Company, Battle Creek, MI; and the Clinical Nutrition and Risk Factor Modification Centre and the Division of Endocrinology and Metabolism, St Michael's Hospital, Toronto.
2 Supported by the Kellogg Company and the Medical Research Council of Canada (grant no. UI-13990). 3 Address reprint requests to TMS Wolever, Department of Nutritional Sciences, University of Toronto, Toronto, Ontario, Canada M5S 3E2. E-mail: thomas.wolever{at}utoronto.ca.
ABSTRACT
Background: Results of 6-wk studies suggest that high-carbohydrate diets are deleterious for people with type 2 diabetes.
Objective: Our objective was to see whether long-term replacement of dietary monounsaturated fatty acids (MUFAs) with carbohydrate from breakfast cereals with either a high or a low glycemic index (GI) affected blood glucose and lipids in subjects with type 2 diabetes.
Design: Subjects with type 2 diabetes (n = 91) were randomly assigned to receive 10% of energy from a low-GI breakfast cereal, a high-GI cereal, or oil or margarine containing MUFA for 6 mo. Eating breakfast cereal was prohibited for subjects in the MUFA group.
Results: Seventy-two subjects completed the trial. The subjects who received cereals consumed 10% more energy from carbohydrate than did the subjects in the MUFA group. Changes in glycated hemoglobin, body weight, and fasting cholesterol and triacylglycerol did not differ significantly among groups. HDL cholesterol increased by 10% in the MUFA group compared with subjects who consumed either high- or low-GI cereals (P = 0.002). The ratio of total to HDL cholesterol was higher in the subjects who consumed the high-GI cereal than in the MUFA group at 3 mo but not at 6 mo (diet x time interaction, P = 0.041). During 8-h metabolic profiles, mean plasma insulin was higher and mean free fatty acids were lower in the 2 cereal groups than in the MUFA group (P < 0.05).
Conclusions: A 10% increase in carbohydrate intake associated with breakfast cereal consumption had no deleterious effects on glycemic control or blood lipids over 6 mo in subjects with type 2 diabetes. The increase in plasma insulin and the reduction in free fatty acids associated with higher carbohydrate intake may reduce the rate of progression of diabetes.
Key Words: Type 2 diabetes • monounsaturated fat • carbohydrate • cereal • glycemic index • cholesterol
INTRODUCTION
To reduce the risk of cardiovascular disease, people with type 2 diabetes (1) are advised to reduce their intakes of saturated fat and cholesterol. However, there is debate as to whether the saturated fat should be replaced by carbohydrate or monounsaturated fatty acids (MUFAs). In the 1980s, in line with advice for the general population (2), people with diabetes were advised to eat high-carbohydrate diets (3). However, the results of recent studies suggest that high-carbohydrate diets cause persistent deterioration in glycemic control, exaggerate hyperinsulinemia (4), raise serum triacylglycerol, and lower HDL cholesterol (4, 5) compared with diets that are rich in MUFAs. Thus, it has been argued that high-carbohydrate diets increase the risk of cardiovascular disease and should not be used in the treatment of type 2 diabetes (6). One practical implication is that ready-to-eat breakfast cereals should be avoided by people with diabetes because their consumption is associated with high carbohydrate intakes (7, 8).
Deleterious effects of high-carbohydrate diets have been suggested on the basis of studies lasting 2–6 wk, which is too short to determine sustained effects of diet on body weight or glycated hemoglobin (Hb A1c). Ad libitum, high-fat diets are associated with weight gain (9–11), which is generally considered undesirable for people with diabetes. In addition, the type of carbohydrate has not been considered. The consumption of foods with a low glycemic index (GI) improves blood glucose and lipids in people with type 2 diabetes (12–15). The GI of commercially available breakfast cereals varies over a 2-fold range, from 120–130 for low-fiber corn and rice cereals to 65 for psyllium-enriched, high-fiber cereals (16, 17). There is evidence that low-GI cereals reduce blood glucose throughout the day (18) and that psyllium-enriched cereals reduce serum cholesterol (19). Thus, our purpose was to determine the effect of exchanging 10% of dietary energy from MUFAs with carbohydrate from breakfast cereals on blood glucose and lipids in subjects with type 2 diabetes over a 6-mo period and to see whether low-GI cereals had any advantage over high-GI cereals.
SUBJECTS AND METHODS
Subjects were recruited through 3-d newspaper advertisements on 4 occasions over 16 mo. Prospective volunteers were invited to attend a lecture explaining the purpose and procedures of the study. A total of 196 people were screened; 109 were eligible to participate in the study and 91 of these were recruited (20–25 from each round of advertisements). To be eligible to participate, subjects had to be nonpregnant, be aged 40–80 y, and have had diabetes for 6 mo with no biochemical evidence of impaired renal or hepatic function. The subjects had to have a body mass index (BMI; in kg/m2) <36, an Hb A1c concentration > 0.065 (6.5%) and <0.101 (10.1%), and a serum triacylglycerol concentration < 10.0 mmol/L (885 mg/dL). The subjects could be treated by diet alone or with a stable dose (for 6 wk) of metformin, sulfonylurea, or both. Subjects who were being treated with insulin or acarbose were excluded. Nine healthy control subjects underwent the 8-h metabolic profiles. The study protocol was approved by the St Michael's Hospital Research Ethics Board. The subjects gave written, informed consent and obtained their physician's written consent to participate. The physicians were requested not to change doses of diabetes, lipid, or antihypertensive medications.
The study had a randomized, parallel design with 3 diet treatment arms, each lasting 6 mo. Subjects from each round of recruitment were stratified on the basis of age, sex, Hb A1c, BMI, and type of diabetes, lipid-lowering, and antihypertensive medication used. The 3 treatments were carbohydrate from a high-GI breakfast cereal (CHO-HGI), carbohydrate from a low-GI breakfast cereal (CHO-LGI), or high MUFA intake without breakfast cereals. Treatment foods were provided to the subjects and prescribed to meet 10–15% of daily energy requirements. Daily energy requirements were calculated from data from the Lipid Research Clinics Prevalence Study (20), with 1254 kJ (300 kcal) added for weight maintenance or 836 kJ (200 kcal) subtracted for weight loss (BMI > 27). The subjects were free-living and attended the clinic approximately once per month. At each clinic visit, body weight was measured and compliance was assessed as the difference between the amount of treatment food provided at the previous clinic visit and the amount returned at the present visit. All subjects were encouraged to exercise and were specifically advised that they could lose as much weight as they desired as long as they consumed the prescribed amount of treatment food.
Fasting blood samples were drawn for analysis of Hb A1c and serum glucose, total and HDL cholesterol, and triacylglycerol at screening, baseline, 3 mo, and 6 mo. Additional blood samples were drawn for analysis of Hb A1c 1 wk before the start and at the end of the study; the average of the 2 Hb A1c results measured at baseline and at the end of the study were used in the statistical analysis.
At baseline and at the end of the study, the subjects came to the nutrition center after 10–14 h overnight fasts for an 8-h metabolic profile. After a cannula was inserted into a forearm vein for blood sampling and a fasting blood sample was obtained, any medications were taken and a breakfast meal was consumed. Additional blood samples were taken at hourly intervals after the subjects started to eat. Immediately after the 4-h blood sample was taken, lunch was consumed and 4 additional blood samples were taken at hourly intervals.
For the baseline metabolic profile, the same standard breakfast and lunch meals, representing a high-carbohydrate diet, were provided to all diabetic subjects and healthy control subjects. The energy content of the meals was individually adjusted to provide 40–45% of daily energy requirements. At the end of the study, the breakfast meals included the amount of treatment food used at home during the study. Lunch meals were identical at the beginning and at the end of the study, containing 2078 ± 25 kJ (497 ± 6 kcal) energy, 22.0 ± 0.3 g protein, 50.8 ± 0.6 g carbohydrate, 22.8 ± 0.3 g total fat, 8.1 ± 0.1 g saturated fat, 7.5 ± 0.1 g MUFAs, 5.9 ± 0.1 g polyunsaturated fat, 8.0 ± 0.1 g fiber, and a GI of 90.5 ± 0.1. The subjects were offered water, coffee, or tea (with an optional 30 g milk and artificial sweetener) at breakfast, midmorning (2 h), and at lunch, and the beverages chosen remained standard for each subject on both profile days.
The high-GI cereals were corn flakes (GI, 121; Nature's Path; Delta, Canada), puffed rice (Arrowhead Mills, Hereford, TX), and crispy rice (GI, 132; Our Compliments, The Oshawa Group Ltd, Toronto). The low-GI cereals were Bran Buds with Psyllium (GI, 67; Kellogg Co, Battle Creek, MI) and a prototype oat-loop cereal enriched with psyllium (GI, 61) (21). The subjects were allowed to consume only the breakfast cereals provided. The subjects in the MUFA group were not allowed to eat breakfast cereals and were given nonhydrogenated salted or unsalted soft tub margarine (Fleischman's 20% Corn Oil Margarine; Lipton, Toronto) or olive oil (President's Choice Extra Virgin Olive Oil; Sunfresh Ltd, Toronto), or both. The treatment foods were incorporated into individualized diabetes meal plans by the study dietitian. Either a portion or all of the prescribed food was consumed at the breakfast meal and the remaining amounts were consumed later in the day according to individual preference. The subjects were advised to increase their consumption of treatment foods slowly during the first month to avoid abdominal discomfort.
The subjects were instructed on how to complete 3-d food records, 2 of which were collected during the baseline period and 4 during the study to assess nutrient intakes. The food records were analyzed by using NUTRIPUT (version 2.02; University of Toronto, Toronto), incorporating the US Department of Agriculture Handbook no. 8 database with missing values for dietary fiber added (22) and the values for carbohydrate modified to reflect glycemic carbohydrate (ie, total carbohydrate minus dietary fiber). The GIs of the diets were calculated as described previously (23).
Hb A1c was measured in the Department of Clinical Biochemistry, St Michael's Hospital, by using Diamat HPLC [Bio-Rad Laboratories (Canada) Ltd, Mississauga, Canada]. Fasting serum glucose, total cholesterol, and triacylglycerol were measured enzymatically by using a Vitros Analyser 950 (Johnson & Johnson Clinical Diagnostics, Rochester, NY); HDL cholesterol was measured after precipitation of other lipoproteins with dextran sulfate and magnesium chloride. LDL cholesterol was calculated as total cholesterol - (HDL cholesterol + triacylglycerol/2.2) (only for triacylglycerol <4.51 mmol/L). Plasma samples from the 8-h metabolic profiles were used for measurement of glucose (2300 STAT glucose analyzer; Yellow Springs Instruments, Yellow Springs, OH), insulin (Insulin RIA, Pharmacia, Dorval, Canada), free fatty acids (ACS-ACOD Method; WAKO Chemicals USA, Richmond, VA), and triacylglycerol [Triacylglycerol (GPO-Trinder); Sigma Diagnostics, St Louis).
Subjects with changes in the type or dose of oral hypoglycemic agent in the first 3 mo of the study were dropped from the study and their results excluded. If the change occurred after 3 mo, the subjects were dropped from the study but the baseline and 3-mo results were included in the analysis. Blood lipid results of subjects with changes in hypolipidemic agents were excluded from the analysis.
The primary endpoint was Hb A1c and the study was designed to have an 80% chance of detecting a difference in Hb A1c of 0.5% with P < 0.05. Changes from baseline and 3 and 6 mo were subjected to repeated-measures analysis of variance (ANOVA), with examination for differences in diet (CHO-HGI, CHO-LGI, or MUFA), time (3 and 6 mo), and diet x time interaction by using the general linear model procedure (SAS release 6.11; SAS Institute Inc, Cary, NC). Adjustment was made for uneven sample sizes and seasonal effects due to recruitment rounds occurring at different times of the year. All pairwise comparisons of individual means for effects found to be significant in the ANOVA were carried out by using Tukey's procedure to control for multiple comparisons. Dietary intake and metabolic profile data were subjected to repeated- measures ANOVA. When significant interactions between diet and time were found, individual means were compared by using Tukey's procedure. For metabolic profile data, after demonstration of a significant diet x period interaction, changes in outcome measures from baseline were compared by using one-way ANOVA, and the Neuman-Keuls procedure was used to protect for multiple comparisons. Mean 8-h plasma glucose, insulin, fatty acid, and triacylglycerol concentrations were calculated by determining the total areas under the curves with use of the trapezoidal rule and dividing by 8 h. Results are expressed as means ± SEMs.
RESULTS
Age, BMI, and Hb A1c ranged from 42 to 79, 18.4 to 35.7, and 0.047 to 0.1, respectively, in the control and diabetic subjects (Table 1). There were no significant differences in sex, age, or BMI between the 3 study groups and the control group at baseline. There were no significant differences in Hb A1c, type of diabetes management, and antihypertensive and lipid-lowering agents used among the 3 diet groups. Hb A1c was lower in the control group than in subjects with diabetes. Of 91 randomly assigned subjects, 19 (21%) dropped out, 12 before and 5 after 3 mo, because of a change in the dose of the oral agent (6 in the CHO-HGI group, 1 in the CHO-LGI group, and 3 in the MUFA group), dislike or intolerance of treatment (1 in the CHO-HGI group, 3 in the CHO-LGI group, and 2 in the MUFA group), a motor vehicle accident (1 in the MUFA group), a nonfatal heart attack (1 in the MUFA group), and a fatal stroke (1 in the MUFA group). One subject from each diet group was excluded from lipid analyses because of changes in lipid-lowering medications.
View this table:
TABLE 1.. Characteristics of subjects with type 2 diabetes by diet group and of healthy control subjects at baseline1
Except for small differences in energy, the baseline diets did not differ significantly among the 3 diet groups (Table 2). Recorded energy intake increased by 400 kJ from baseline with each of the 3 diets. Carbohydrate intake increased by 5% and protein and total fat intakes decreased in the subjects consuming the CHO-HGI diet. Carbohydrate intake decreased by 6% in the MUFA group, mainly because of the high MUFA intake. Carbohydrate intake did not change significantly with the CHO-LGI diet but protein intake tended to increase and fat decreased significantly. Thus, the CHO-HGI group consumed 3.5% more carbohydrate than did the CHO-LGI group, whose intake, in turn, was 7% greater than that of the MUFA group. Both cereal groups consumed 9–10% less fat than did the MUFA group. Fiber intake doubled with the CHO-LGI diet; the cereal contributed 30.6 ± 0.9 g fiber/d. The GI of the diet increased in the CHO-HGI group and decreased in the CHO-LGI group; there was no significant change in the MUFA group.
View this table:
TABLE 2.. Daily nutrient intakes in subjects with type 2 diabetes by diet group at baseline and during the study1
Disappearance of treatment foods, as a percentage of the prescribed amount from months 2–6, was 111 ± 4% in the CHO-HGI group, 87 ± 2% in the CHO-LGI group, and 104 ± 5% in the MUFA group; the difference between the CHO-LGI group and the other 2 groups was significant (P < 0.001). Intake of treatment foods accounted for 13 ± 1% of energy in the CHO-HGI and CHO-LGI groups and for 18 ± 1% in the MUFA group (P < 0.001).
At baseline, weight, fasting glucose, Hb A1c, HDL cholesterol, the ratio of total to HDL cholesterol, and triacylglycerol were not significantly different between groups; however, total and LDL cholesterol were significantly higher in the MUFA than in the CHO-LGI group (Table 3). Changes in body weight, Hb A1c, fasting serum glucose, total and LDL cholesterol, and triacylglycerol did not differ significantly between diets (Figures 1 and 2). However, there was a trend for a diet x time interaction for change in serum total cholesterol (P = 0.060); the value between 3 and 6 mo decreased in the CHO-HGI group and increased in the MUFA group (Figure 2). HDL cholesterol was persistently higher in the MUFA group (by 12%) than in the CHO-HGI and CHO-LGI groups (P = 0.002; Figure 2). There was no significant main effect of diet on the change in the ratio of total to HDL cholesterol, ie, the mean of the 3- and 6-mo changes in total:HDL cholesterol did not differ significantly between diets (P = 0.091). However, there was a significant diet x time interaction (P = 0.041). At 3-mo, total:HDL cholesterol was 12% higher in the CHO-HGI group than in the MUFA group, but by 6 mo the difference was no longer significant (Figure 3).
View this table:
TABLE 3.. Baseline values of outcome measures in each diet group1
FIGURE 1. . Mean (±SEM) changes from baseline in body weight, glycated hemoglobin (Hb A1c), and fasting serum glucose after 3 and 6 mo of treatment with carbohydrate from breakfast cereal with a high glycemic index (•), carbohydrate from breakfast cereal with a low glycemic index (), or monounsaturated fatty acids (). SEs not shown if they overlap other symbols or SEs.
FIGURE 2. Mean (±SEM) changes from baseline in fasting serum total, LDL, and HDL cholesterol and triacylglycerol after 3 and 6 mo of treatment with carbohydrate from breakfast cereal with a high glycemic index (•), carbohydrate from breakfast cereal with a low glycemic index (), or monounsaturated fatty acids (). SEs not shown if they overlap other symbols or SEs. Means with different superscript letters are significantly different, P < 0.05.
FIGURE 3. Mean (±SEM) changes in the ratio of fasting serum total to HDL cholesterol after 3 and 6 mo of treatment with carbohydrate from breakfast cereal with a high glycemic index (•), carbohydrate from breakfast cereal with a low glycemic index (), or monounsaturated fatty acids (). Means with different superscript letters are significantly different, P < 0.05.
Of the 72 subjects who completed the study, 5 had incomplete metabolic profiles because of religious holidays (n = 2), dislike of multiple blood sampling, lack of venous access, and illness. There were no significant differences in nutrient intakes on profile days between groups at baseline (Table 4). At the end of the study, the breakfast meal for the MUFA group contained 20 g less carbohydrate and 9 g more fat than at baseline, whereas that for the CHO-LGI group contained 20 g more fiber than at baseline. The GI of the breakfast meal at baseline was 92, increasing to 97 with the CHO-HGI diet and decreasing to 72 and 82 with the CHO-LGI and MUFA diets, respectively.
View this table:
TABLE 4.. Compositions of the breakfast meals during the metabolic profile tests in subjects with type 2 diabetes and healthy control subjects1
During the 8-h baseline metabolic profiles, plasma glucose was significantly higher in subjects with diabetes than in control subjects (F[1,8] = 55.5; Figure 4) and there was a significant main effect of time of day (F[8,8] = 17.3) and time x group interaction (F[8,591] = 17.2). Mean 8-h plasma insulin in subjects with diabetes (155 ± 10 pmol/L) was not significantly different from that in control subjects (139 ± 18 pmol/L). However, there was a significant main effect of time and time x group interaction such that plasma insulin in subjects with diabetes was significantly higher than in control subjects 3, 4, 6, 7, and 8 h after breakfast and significantly lower than in control subjects 1 h after breakfast. Mean plasma free fatty acids in subjects with diabetes (0.22 ± 0.01 mmol/L) was 32% greater than in control subjects (0.15 ± 0.01 mmol/L) (P = 0.013). There was a significant main effect of time (P < 0.001) and time x group interaction (P = 0.009); subjects with diabetes had significantly higher fatty acids than did control subjects at fasting, 1 h, and 6 h. For triacylglycerol there were also significant main effects of diet (P = 0.002) and time (P < 0.001) and significant time x group interaction (P = 0.001). Subjects with diabetes had significantly higher triacylglycerol concentrations than did control subjects at 6, 7, and 8 h.
FIGURE 4. Mean (±SEM) plasma glucose, insulin, free fatty acids, and triacylglycerol concentrations over 8 h in 67 subjects with diabetes () at baseline and in 9 healthy control subjects (•). Breakfast was consumed at 0 h (0800) and lunch at 4 h. SEs not shown if they are smaller than the symbol. *,**Significantly different from subjects with diabetes: *P < 0.05, **P < 0.01.
A comparison of the metabolic profiles of the 3 diet groups (diet effect) at the beginning and end of the study (period effect) showed significant main effects of diet and period for fasting and 4-h plasma glucose and 0–8-h mean plasma glucose (Table 5), the values being lower in the MUFA group than in the other 2 diet groups and increasing from the beginning to the end of the study in all 3 diet groups. There was a significant diet x period interaction at the 1- and 2-h time points. In the CHO-HGI group, plasma glucose tended to increase at 1 and 2 h, changes that differed significantly from the CHO-LGI group at 1 h and from the MUFA group at 1 and 2 h (Figure 5). In the CHO-LGI group, plasma glucose was significantly lower than at baseline 1 h after breakfast. Mean 8-h glucose increased more in the CHO-HGI group than in the CHO-LGI or MUFA group, but the differences were not significant (Figure 6). The incremental area under the blood glucose curve after breakfast (BAUC) increased in the CHO-HGI group (from 16.1 ± 1.2 to 18.8 ± 1.3 mmol•h/L, P = 0.062) and decreased in the CHO-LGI group (from 16.8 ± 1.2 to 15.3 ± 1.3 mmol•h/L; P = 0.058) and the MUFA group (from 15.9 ± 1.4 to 11.9 ± 0.8 mmol•h/L; P < 0.001). The change in BAUC with the CHO-HGI diet differed significantly from the changes with the MUFA and CHO-LGI (P < 0.001).
View this table:
TABLE 5.. Significance of main effects and interactions for metabolic profile data before (0 h) and 1–8 h after consumption of the breakfast meals
FIGURE 5. Mean (±SEM) plasma glucose, insulin, free fatty acids, and triacylglycerol concentrations over 8 h before (•) and 6 mo after () treatment with carbohydrate from breakfast cereal with a high glycemic index (left column), carbohydrate from breakfast cereal with a low glycemic index (middle column), or monounsaturated fatty acids (right column). Breakfast was consumed at 0 h (0800) and lunch at 4 h. SEs not shown if they are smaller than the symbol. *Significantly different from baseline, P < 0.05. Pairs of means with different subscript letters indicate that changes from baseline on the different diets differed significantly, P < 0.05.
FIGURE 6. Mean (±SEM) changes in mean 8-h plasma glucose, insulin, free fatty acids, and triacylglycerol concentrations after treatment with carbohydrate from breakfast cereal with a high-glycemic index (), carbohydrate from breakfast cereal with a low glycemic index (), or monounsaturated fatty acids (). The total length of the vertical axis is 70% of the mean concentration at baseline. Means with different superscript letters are significantly different, P < 0.05.
There was no significant main effect of diet or period on plasma insulin, but the diet x period interaction was significant at 2, 3, and 4 h, and the 0–8-h mean was significant (Table 5). Plasma insulin in the MUFA group was significantly lower 2 and 4 h after breakfast than at baseline. The changes in plasma insulin in the MUFA group differed significantly from those in the CHO-HGI group at 2, 3, and 4 h and from those in the CHO-LGI group at 2 and 4 h (Figure 5). Mean 8-h plasma insulin in the MUFA group was 12% lower than at baseline (NS), 15% lower than that in the CHO-LGI group (P < 0.05), and 19% lower than that in the CHO-HGI group (Figure 6).
There was no significant main effect of diet or period on plasma free fatty acids, but the diet x period interaction was significant at 4, 5, and 6 h and for the 0–8-h mean (Table 5). Plasma fatty acids were significantly lower than at baseline at 6 h in the CHO-LGI group and significantly higher than at baseline at 4 and 5 h in the MUFA group. The change in the MUFA group was significantly different from those in both cereal groups at 4, 5, and 6 h (Figure 5). Mean 0–8-h fatty acids tended to fall from baseline in both cereal groups, but the difference was not significant. However, 0–8-h mean FFAs were 28% higher in the MUFA group and 23% than in the CHO-HGI and CHO-LGI groups, respectively (Figure 6).
There was no significant effect of period on plasma triacylglycerol and only a marginally significant effect of diet at 1 and 6 h, but the diet x period interaction was significant at 2, 3, and 4 h (Table 5). Plasma triacylglycerol tended to increase in the CHO-LGI group, decrease in the CHO-HGI group, and not change from baseline in the MUFA group. The change after the CHO-LGI diet was significantly different from that after both the CHO-HGI and the MUFA diets 2, 3, and 4 h after breakfast (Figure 5). Mean 8-h plasma triacylglycerol in the CHO-LGI group was higher than in the CHO-HGI group (Figure 6).
DISCUSSION
The results of this study show that exchanging 10% of energy from MUFA with carbohydrate from breakfast cereals has no long-term deleterious effects on glycemic control or blood lipids in subjects with type 2 diabetes. In addition, they suggest that the results of studies lasting 6–12 wk may not reflect the long-term effects of high-carbohydrate diets on cardiovascular risk in type 2 diabetes.
High-carbohydrate diets have low energy density, which is associated with reduced energy intake (24, 25). This is consistent with data from subjects with type 2 diabetes that suggest that advice to reduce dietary fat and energy results in greater long-term weight loss than does advice merely about energy restriction (26). Thus, we expected increased MUFA intake to result in greater weight gain relative to the consumption of breakfast cereals. However, we saw no significant differences between the diets in recorded energy intake or change in body weight. This may have been because the sources of MUFAs were not consumed alone but were used in cooking or added to other foods. This may have resulted in the subjects paying more attention to what they were eating than merely eating a specific breakfast cereal. Whatever the reason, these results show that, with careful dietary counseling, sources of MUFA can be incorporated into the diet of patients with type 2 diabetes and produce no weight gain over a 6-mo period.
It has been suggested that high-carbohydrate diets cause persistent deterioration of glycemic control in subjects with type 2 diabetes (4). However, the validity of this conclusion is questionable because it was based on higher mean 24-h plasma glucose concentrations measured under controlled conditions in a metabolic ward. There was no change in Hb A1c or fasting plasma glucose over the 6-wk study period (4). High-carbohydrate diets increase postprandial blood glucose and insulin in healthy subjects (27) and subjects with diabetes (28) but may lower fasting plasma glucose and improve glucose tolerance (29), effects that would tend to counteract the increased postprandial glucose (30). In the present study there was a nonsignificant trend for Hb A1c to be higher in the CHO-HGI than in the MUFA group at 3 mo, which presumably reflected increased postprandial blood glucose. However, this trend had reversed by 6 mo, with Hb A1c and fasting glucose becoming nonsignificantly higher in the MUFA group despite lower postprandial glucose. Although it is tempting to speculate that the latter trend may have been significant in a longer study, the present data do not show any significant difference in glycemic control between diets.
The CHO-LGI diet persistently reduced the glycemic effect of the breakfast meal. Although this was associated with smaller increases in mean Hb A1c and fasting glucose than was the CHO-HGI diet, the differences were not significant. The -glucosidase inhibitor acarbose improves Hb A1c for 1 y (31), suggesting that reducing postprandial glucose responses with no change in carbohydrate intake should have long-term benefits. Thus, we believe that the change in the overall GI of the diet achieved by varying the GI of breakfast cereals only was not large enough to have had a significant effect on Hb A1c. Indeed, the change in the GI of the diet in our study (10) is smaller than differences of 12–28 achieved in most other studies in which glycemic control was improved (32). Thus, more than one type of low-GI food may need to be incorporated into the diet to achieve measurable long-term improvements in glycemic control.
A recent meta-analysis concluded that high-carbohydrate compared with high-MUFA diets raise serum triacylglycerol and reduce HDL cholesterol in subjects with type 2 diabetes (5), changes associated with increased cardiovascular risk. However, these effects were not seen in studies with modest (<15% of energy) increases in carbohydrate intake (33), and, none of the 10 studies included in the analysis lasted 6 wk (5). Thus, the fact that we found no effect of a 10% increase in carbohydrate intake on serum triacylglycerol is consistent with the results of previous studies. Despite a significant fall in HDL cholesterol, which persisted for 6 mo in the CHO-HGI group, total:HDL cholesterol increased only temporarily. Total:HDL cholesterol may be a better predictor of cardiovascular risk than are the concentrations of total, LDL, or HDL cholesterol or triacylglycerol alone (34). Although they need confirmation, our data suggest that the results of studies lasting only 6–12 wk may not represent the long-term effects of high-carbohydrate diets on cardiovascular risk in subjects with type 2 diabetes.
The lack of effect of the CHO-LGI diet on serum cholesterol was unexpected because 2–3 servings of psyllium-enriched cereal daily reduced total and LDL cholesterol in hyperlipidemic subjects (35, 36). The number of subjects studied was based on power analysis for Hb A1c; serum cholesterol was a secondary endpoint. A larger subject sample would be needed to detect the expected 5% change in cholesterol, especially in normocholesterolemic subjects (36). The significant increase in mean 8-h triacylglycerol in the CHO-LGI group than in the CHO-HGI group was also somewhat surprising but is consistent with a nonsignificant trend toward a similar effect seen in hyperlipidemic subjects (36).
The reduction in plasma insulin associated with MUFAs is consistent with the results of previous studies and has been interpreted as representing a beneficial reduction in hyperinsulinemia (4). However, although people with diabetes may have fasting hyperinsulinemia, they have a reduced ability to secrete insulin in response to a rise in blood glucose. Progressive ß-cell failure is characteristic of type 2 diabetes and is likely the cause of the relentless rise in blood glucose seen with time (37). There is no evidence that insulin treatment has any adverse effects in type 2 diabetes; if anything, it is associated with fewer long-term cardiovascular complications (38). Thus, we argue that a reduction in plasma insulin is deleterious in type 2 diabetes. In this context, it is of interest that postprandial plasma insulin was not lower in the CHO-LGI group despite a significant reduction in the glycemic response. This may represent enhanced ß-cell function because, acutely, psyllium-enriched test meals elicit low glucose and insulin responses in subjects with type 2 diabetes (39).
High plasma free fatty acids have been implicated in the pathogenesis of diabetes for >30 y (40, 41) and are often discussed in the context of insulin resistance (42, 43) and its associated abnormalities (44). However, there are few data on how diet influences free fatty acids in diabetes. High plasma free fatty acids increase hepatic glucose output, reduce peripheral glucose utilization (45), and reduce insulin secretion (46), the pathogenic mechanisms responsible for diabetes (47). The present data are consistent with this in that MUFAs raised free fatty acids, reduced insulin, and tended to increase fasting and postlunch glucose to the greatest extent. High-fat meals acutely increase postprandial free fatty acids in healthy (21) and diabetic (28) subjects, and the current data suggest that these effects persist in subjects with diabetes. A high-fat diet may increase postprandial free fatty acids because, with lower postprandial insulin, there is less inhibition of hormone-sensitive lipase and more release of free fatty acids from adipose tissue. Alternatively, the rise in free fatty acids may be derived directly from dietary fat. Increased free fat intake reaches the circulation as chylomicrons, which are acted on by lipoprotein lipase to release free fatty acids. However, not all of these fatty acids are taken up by adipose tissue, especially in subjects with diabetes, because impaired trapping of chylomicron-derived free fatty acids is associated with insulin resistance (48). The reduction in fatty acids and the increase in triacylglycerol in the CHO-LGI group may have been due to increased colonic fermentation and increased acetate production (49, 50). It is of interest that the only time free fatty acids were significantly lower than at baseline was at 6 h, a time when serum acetate is at its highest during the day, at least in nondiabetic subjects (51).
Regular consumption of breakfast cereals is associated with a high carbohydrate intake. Thus, if high-carbohydrate diets are deleterious, breakfast cereals should be avoided by people with diabetes. However, people aged >65 y, who are most likely to have diabetes, also are more likely to use ready-to-eat breakfast cereals than is any other segment of the population except children (8). Breakfast cereals contribute to the achievement of adequate intakes of vitamins and minerals (7, 8), which is important for the elderly, who are most at risk of nutrient deficiencies. Our results suggest that there is no reason for people with diabetes to avoid breakfast cereals.
We conclude that a 10% increase in carbohydrate intake associated with breakfast cereal consumption has no deleterious effects on glycemic control or blood lipids over 6 mo in subjects with type 2 diabetes. A longer study is needed to confirm whether a high-carbohydrate diet has only a temporary effect on total:HDL cholesterol and to determine whether the reduction in plasma insulin and increase in free fatty acids associated with a high MUFA intake have deleterious effects on the progression of diabetes.
REFERENCES