点击显示 收起
1 From The Division of Endocrinology, Diabetes and Clinical Nutrition (GTG, AA, PBD, and WEC) and the General Clinical Research Center (KM and MPM), Oregon Health & Science University, Portland, OR
2 Supported by General Clinical Research Center grant PHS 5MOI-RR00334 and by the Oregon Health & Science University Foundation.
3 Reprints not available. Address correspondence to WE Connor, Department of Medicine, L465 Oregon Health & Science University, 3181 SW Sam Jackson Park Road, Portland, OR 97239-3098. E-mail: connorw{at}ohsu.edu.
See corresponding editorial on page 537.
ABSTRACT
Background: An important therapeutic goal for patients with type 2 diabetes is weight loss, which improves metabolic abnormalities. Ad libitum low-fat diets cause weight loss in nondiabetic populations. Compared with diets higher in monounsaturated fat, however, eucaloric low-fat diets may increase plasma triacylglycerol concentrations and worsen glycemic control in persons with type 2 diabetes.
Objective: We investigated whether, in type 2 diabetes patients, an ad libitum low-fat diet would cause greater weight loss than would a high-monounsaturated fat diet and would do this without increasing plasma triacylglycerol concentrations or worsening glycemic control.
Design: Eleven patients with type 2 diabetes were randomly assigned to receive an ad libitum low-fat, high-carbohydrate diet or a high-monounsaturated fat diet, each for 6 wk. The diets offered contained 125% of the estimated energy requirement to allow self-selection of food quantity. The response variables were body weight; fasting plasma lipid, lipoprotein, glucose, glycated hemoglobin A1c, and fructosamine concentrations; insulin sensitivity; and glucose disposal.
Results: Body weight decreased significantly (1.53 kg; P < 0.001) only with the low-fat diet. Plasma total, LDL-, and HDL-cholesterol concentrations tended to decrease during both diets. There were no interaction effects between diet and the lipid profile response over time. Plasma triacylglycerol concentrations, glycemic control, and insulin sensitivity did not differ significantly between the 2 diets.
Conclusion: Contrary to expectations, the ad libitum, low-fat, high-fiber diet promoted weight loss in patients with type 2 diabetes without causing unfavorable alterations in plasma lipids or glycemic control.
Key Words: Lipoproteins • glycemia • energy density • fiber • type 2 diabetes • body weight
INTRODUCTION
The optimal diet for persons with diabetes has long been a subject of controversy. Dietary therapy was the only treatment available in the era before insulin therapy (1). As elucidated by Joslin (2), dietary carbohydrate had to be restricted in patients with type 1 diabetes because of impaired carbohydrate metabolism. Such diets were ketogenic and were composed largely of fat and protein—the "good foods of life," eg, meat, cream, butter, cheese, and eggs. Joslin also commented that this diet was atherogenic and made the point that, if patients did not die of diabetic ketoacidosis, they would probably die of coronary artery disease.
With the discovery of insulin in 1921 (3), it became possible to introduce carbohydrate-containing foods into the diabetic diet, but this approach was poorly accepted. Research in the 1960s, however, pointed out that a higher-carbohydrate, low-fat diet could be used in diabetic patients to lower cholesterol without increasing the plasma triacylglycerol concentration (4). More recently, a diet low in saturated fat has become accepted (5), but controversy has focused on whether the diabetic diet should be higher in monounsaturated fat or higher in carbohydrates. It was thought that a high-monounsaturated fat (high-mono) diet would avoid the possible plasma triacylglycerol- and glucose-elevating effects of a high-carbohydrate diet and still contain less saturated fat and cholesterol than the earlier diabetic diet (6).
This study was therefore carried out to define more precisely the optimal energy distribution of monounsaturated fat and carbohydrates in the diabetic diet. The need for such a definition comes from the recommendation of the American Diabetes Association (5) that 60-70% of total energy be derived from a combination of monounsaturated fat and carbohydrates. It was suggested that more monounsaturated fat and less carbohydrate be prescribed, especially in diabetic patients with lipemia. On the other hand, it was recommended that dietary fiber be increased, and fiber is present in carbohydrate-containing plant foods.
Although high-mono diets have been associated with improvement in glycemic control and dyslipidemia in controlled metabolic studies (6), such results are not necessarily applicable to free-living subjects. In those metabolic studies, subjects were not allowed to regulate their own energy intake. We hypothesized that, if patients were allowed to adjust their energy intake on the basis of satiety, a high-carbohydrate, high-fiber, low-fat diet might be superior to a high-mono diet. Accordingly, the purpose of the present study was to compare 2 ad libitum diets in diabetic patients—one high in monounsaturated fat and the other low in fat and high in fiber and complex carbohydrates—to ascertain which diet would lead to greater weight loss and greater improvements in dyslipidemia and glycemic control.
SUBJECTS AND METHODS
Subjects
Eleven subjects (8 women and 3 men) with type 2 diabetes mellitus treated with oral glucose-lowering medication, diet, or both were recruited for the study. Baseline characteristics of the subjects are given in Table 1. Nine of the 11 patients were being treated with glyburide or glyburide and metformin, 1 was being treated with troglitazone alone, and 1 was being treated with diet alone. The subjects had fairly good glycemic control at baseline (Table 1). Exclusion criteria included insulin therapy within the previous 2 mo, glycated hemoglobin A1c (Hb A1c) > 12%, medical conditions affecting plasma lipoprotein metabolism, use of lipid-lowering medication within the previous 6 wk, proliferative retinopathy or nephropathy, coronary events within the previous 6 mo, treatment with oral glucocorticoids, or fasting plasma cholesterol concentrations > 300 mg/dL or triacylglycerol concentrations > 700 mg/dL. The mean initial plasma lipid and lipoprotein concentrations of the subjects are given in Table 2. Stable doses of thiazide, ß-blocker, or other antihypertensive medication were permitted. No changes were made in oral glucose-lowering therapy during the study. Written informed consent was obtained from all subjects. The study was approved by the Institutional Review Board at the Oregon Health & Science University.
View this table:
TABLE 1. Baseline characteristics of 11 patients with type 2 diabetes mellitus1
View this table:
TABLE 2. Changes in body weight and fasting plasma lipid and lipoprotein concentrations during low-fat and high-monounsaturated fat (high-mono) diets in 11 patients with type 2 diabetes mellitus
Design
Subjects were fed low-fat or high-mono metabolic diets in random order for 6 wk, and the 2 diets were separated by a 6-12-wk washout period. Both diets were offered at 25% above estimated energy requirement (7) to allow self-selection for quantity of food. Diets were fed by using a 4-d menu cycle. Menus for the 2 diets were similar, with the fat and carbohydrate composition changed by differences in recipes and serving sizes; subjects were thus blinded to dietary treatment, insofar as this was possible. Typical menus for the 2 diets are shown in Appendix A, which illustrates the manipulations of the foods that were performed to meet the low-fat and high-mono dietary prescriptions. In general, high-fat items and oils on the high-mono diet were partially replaced on the low-fat diet with fat-free oils and foods higher in complex carbohydrates.
View this table:
TABLE 3. Composition of the experimental diets1
All meals were prepared by the Metabolic Kitchen of the General Clinical Research Center at Oregon Health & Science University. Subjects consumed one meal per day at the Clinical Research Center. The other meals including the weekend meals were packaged for home consumption. Subjects were instructed to eat to satisfaction and return uneaten foods, which were weighed to allow calculation of the total energy intake and nutrient consumption by using a nutrient analysis database (FOOD PROCESSOR NUTRIENT ANALYSIS SOFTWARE, version 6.1; ESHA Research, Salem, OR; 8). Subjects were encouraged to consume their meals on a regular schedule and were instructed to maintain their usual exercise level during the study.
Experimental diets
The low-fat diet provided 20% of energy as fat, and the high-mono diet provided 40% of energy as fat (26% of energy was monounsaturated fat; Table 3). The low-fat diet provided 65% of energy as carbohydrates compared with 45% as carbohydrates for the high-mono diet; refined sugar made up 10% of energy intake in both diets. The low-fat diet was higher in fiber and water content, weighed more, and had a lower energy density (kcal/g diet) than did the high-mono diet. Although both diets were low in saturated fat, the low-fat diet was lower than the high-mono diet in saturated fat and cholesterol. The difference in saturated fat and cholesterol between the 2 diets was intentional, in that we wished to study the effects of a diet in its entirety (low-fat compared with high-mono), rather than the effects of individual dietary components. A low-fat diet will generally be lower in saturated fat and cholesterol and higher in dietary fiber than will a high-mono diet, so that the composition of the 2 diets in our study likely mirrored the composition of these diets as they would be eaten in the "real world."
Laboratory analyses
Laboratory analyses were performed on blood samples collected after a 12-h fast before and after each dietary period. Fasting plasma lipid and lipoprotein concentrations were measured in our Lipid Laboratory with the use of standard procedures established by the Lipid Research Clinics Program (9). Fasting plasma glucose concentrations were measured by using the glucose oxidase method. Hb A1c concentrations were measured as an index of glycemic control over the previous 2-3 mo by using HPLC on a Diamet analyzer (Bio-Rad, Hercules, CA). Plasma fructosamine concentrations were measured by using an automated colorimetric assay to obtain an index of glycemic control over the preceding 2-4 wk (10). Hyperinsulinemic, euglycemic clamp studies were performed by using a modification of the DeFronzo method (11, 12); the glucose infusion rate [mg glucose · (m2/h)] required to maintain a steady blood glucose concentration in the face of a constant insulin infusion gives a measure of glucose disposal and insulin sensitivity; lower glucose infusion rates indicate less insulin sensitivity.
Statistical analysis
Changes in body weight, plasma lipids and lipoproteins, and glycemic variables were analyzed with the use of a two-factor analysis of variance model with repeated measures on time and diet (13). In designing the study, preliminary data were used to obtain power calculations (80% power and an of 5%) for detecting a difference of 1.0 in Hb A1c concentrations, a difference of 20% in plasma fructosamine concentrations, and a difference of 15% in fasting plasma glucose concentrations. P values were calculated for the main effects of time, diet, and the time x diet interaction. Of most interest was the time x diet interaction effect, which tested for differential responses to the 2 different diets over time. If the time x diet interaction was significant, multiple comparisons were performed by using Tukey’s test (13). There were no carryover effects. Changes in body weight were computed as the difference between the mean weight on the last 3 d and that on the first 3 d of the dietary phase. The insulin clamp studies and dietary intake data were compared with the use of a paired t test (13). Statistical analyses were performed by using statistical analysis software (SIGMA STAT, version 2.03; Jandel Scientific, San Rafael, CA).
RESULTS
Weight change and energy balance
The most striking finding in this study was that the ad libitum low-fat, high-fiber diet induced a significant weight loss, whereas the high-mono diet did not. The statistically significant time x diet interaction for the body-weight changes indicated a differential response of body weight to the 2 diets (Table 2). Further analysis using Tukey’s multiple comparisons procedure indicated that weight loss was statistically significant only during the low-fat diet (–1.53 ± 1.21 kg; P < 0.001). Body weight decreased 1.0 kg in 8 of the 11 participants during the low-fat diet and in only 3 of the 11 subjects during the high-mono diet. Subjects were offered a mean of 3555 kcal/d during both diets (Table 3), which is 25% above the eucaloric energy requirement of 2848 ± 281 kcal/d. The subjects consumed 212 fewer kcal per day during the low-fat diet than during the high-mono diet (P < 0.02). The predicted difference in the weight loss between the 2 diets, based on the 212 kcal/d difference in energy intake, was –1.15 kg [assuming a total relative caloric deficit on the low fat diet of 212 kcal x 42 d = 8904 kcal, and a deficit of 3500 kcal = 1 lb (2.2 kg)]. The predicted differential weight loss was comparable to the observed difference of –1.06 kg between the 2 diets.
Nutrient intake
The composition of the food offered and the food consumed is shown in Table 3. The composition of the food offered and that of the food returned was determined by weighing all foodstuffs and then consulting the nutrient analysis database. The composition of the food consumed was determined by subtracting the nutrient value of the foods returned from that of the foods as offered. The foods returned for both diets reflected the composition of the diets as they were offered. The mean weight of food consumed, the intakes of total and soluble fiber, and the water content were significantly greater, and energy density was lower, during the low-fat diet than during the high-mono diet.
Fasting plasma lipid and lipoprotein concentrations
The mean fasting plasma lipid and lipoprotein concentrations before and after the low-fat and high-mono diets are given in Table 2. There were no significant time x diet interaction effects for any of the plasma lipid or lipoprotein measurements, which indicated that the response of the plasma lipids and lipoproteins over time did not differ significantly between the low-fat and the high-mono diets. Plasma total, LDL-, and HDL-cholesterol concentrations decreased from both diets. Both diets lowered the mean plasma LDL-cholesterol concentration to <100 mg/dL, which is the treatment goal for diabetic patients (14). Mean plasma triacylglycerol and VLDL-cholesterol concentrations did not change significantly during either diet. The triacylglycerol responses of some subjects were of interest, however—particularly those of the subject with the highest baseline plasma triacylglycerol concentration (639 mg/dL). In that subject, the fasting plasma triacylglycerol concentration decreased substantially and similarly during both diets (40.0% during the low-fat diet and 38.6% during the high-mono diet). There were no significant differences in the ratios of plasma LDL to HDL cholesterol or of total to HDL cholesterol.
Glycemic control
Plasma glucose, fructosamine, and Hb A1c concentrations and glucose infusion rates during the hyperinsulinemic, euglycemic clamp studies did not differ between the 2 diets (Table 4); the changes in plasma glucose, fructosamine, and Hb A1c concentrations were, in fact, very small and unlikely to be of clinical importance. Whereas the difference in glucose infusion rates during the clamp studies was of potential clinical significance (9% higher during the low-fat diet), the variability in the infusion rates was high, so that this difference was not statistically significant.
View this table:
TABLE 4. Fasting plasma glucose, fructosamine, and glycated hemoglobin A1C concentrations before and after low-fat and high-monounsaturated fat (high-mono) diets in 11 patients with type 2 diabetes mellitus1
We recognize that the Hb A1c concentrations obtained at the end of each dietary phase indicated diabetic control over the preceding 2-3 mo, not just the 6 wk of the particular feeding period. However, major deviations in glycemic control over each 6-wk dietary phase would likely result in some change in the Hb A1c concentration obtained at the end of that particular dietary phase, and, therefore, Hb A1c was included in the analysis.
DISCUSSION
The unique feature of this study was that an ad libitum low-fat, high-complex carbohydrate diet caused weight loss in type 2 diabetes patients, whereas an ad libitum high-mono diet did not lead to weight loss. The low-fat diet did not cause the plasma triacylglycerol concentrations to increase and did not worsen glycemic control, contrary to past studies (15–18) in which "eucaloric" low-fat diets were prescribed to maintain body weight. We documented that weight loss occurred because of a reduction in energy intake. Because energy density is an important determinant of satiety and energy intake (19), we hypothesized that the decreased energy intake during the low-fat diet was related to the fact that its energy density is lower than that of the high-mono diet (20). Foods of high energy density (which usually are fats and oils) produce less satiety and promote overconsumption of energy compared with foods with a low energy density, eg, high-fiber foods (19, 20). The lower energy density of the low-fat diet in our study was the result of its reduced fat content and greater content of fiber and water, both of which add weight and bulk without adding calories. Although we did not quantify satiety, many of our subjects expressed the sensation of "fullness" during the low-fat diet.
Eighty to ninety percent of patients with type 2 diabetes mellitus are overweight (21). Because of the central role of visceral fat accumulation and insulin resistance in type 2 diabetes (22), an important therapeutic goal is reduction of body weight. Experimental animals fed low-fat diets lose weight (23), and human populations consuming low-fat diets have low rates of obesity and type 2 diabetes (24). Controlled studies (25, 26) in nondiabetic persons suggested that ad libitum low-fat diets induce weight loss more than do high-fat diets. Furthermore, the results of 3 meta-analyses (27–29) show that the amount of weight lost with low-fat diets is proportional to the percentage of reduction in dietary fat intake. In the Bray meta-analysis (29), which reviewed 28 clinical trials, a reduction of 10% in the daily energy from fat (example, from 30% to 20% of energy) was associated with a weight loss of 16 g/d. For our study, in which the percentage of dietary energy from fat was reduced from 40% to 20% for the low-fat diet, the predicted differential weight loss according to the Bray meta-analysis would be 42 x 16 g/d x 2 = 1344 g, or 1.34 kg. This result is very close to the 1.06-kg observed difference in weight loss between the 2 diets in our study. Our results thus closely conform to those in the scientific literature on this subject. An important issue is whether weight loss that results from following a low-fat diet can be maintained in the long term. Future longer-term studies are needed to answer this question.
There have been few studies of the effects of ad libitum low-fat diets on weight loss in subjects with impaired glucose tolerance or overt type 2 diabetes. In a study by Poppitt et al (30), overweight persons with the metabolic syndrome who received an ad libitum low-fat, high-complex carbohydrate diet for 6 mo lost 4.25 kg and plasma triacylglycerol concentrations fell; these results are similar to those in the present study. In 2 large prospective studies, the Finnish Diabetes Prevention Study (31) and the US Diabetes Prevention Program (32), nondiabetic persons with impaired glucose tolerance or elevated fasting plasma glucose concentrations who received counseling to decrease dietary fat and increase fiber intake and physical activity lost more weight than did control subjects, and 58% less of those subjects than of the control subjects progressed to diabetes. The results of these studies and of our own study support the use of low-fat, high-complex carbohydrate diets for the treatment of obesity in patients with the metabolic syndrome as well as in those with impaired glucose tolerance or overt type 2 diabetes.
A particularly important result of our study was that the low-fat diet did not worsen the fasting plasma lipid and lipoprotein profile. It is important that plasma triacylglycerol concentrations did not increase during the low-fat diet. Abbott et al (33) similarly found that plasma triacylglycerol concentrations did not increase in Pima Indians with type 2 diabetes who were fed a low-fat, high-complex carbohydrate, high-fiber diet for 5-7 wk under controlled research ward conditions. In other dietary studies in patients with type 2 diabetes, however, plasma triacylglycerol concentrations increased during low-fat, high-carbohydrate diets (15, 16). In these other studies, the patients were fed weight-maintaining diets (15, 16), whereas the diabetic patients in our study were allowed to adjust their energy intakes in response to appetite. The patients in our study consumed less energy during the bulky low-fat diet, which resulted in weight loss that may have prevented a rise in plasma triacylglycerol concentrations. Schaefer et al (25) showed in healthy persons that ad libitum low-fat diets caused weight loss without increasing plasma triacylglycerol concentrations, and these results were similar to those in our study. It is important that both the low-fat and the high-mono diets in our study reduced the mean plasma LDL-cholesterol concentration to <100 mg/dL. Plasma HDL-cholesterol concentrations decreased from baseline during both diets, possibly as a result of a reduced intake of saturated fat (34).
Another finding of our study is that glycemic control did not deteriorate during the low-fat diet. This result stands in contrast to previous reports that low-fat, high-carbohydrate diets may cause a deterioration in glycemic control in type 2 diabetes (16–18). In these studies, subjects who were fed eucaloric, low-fat, high- carbohydrate diets had higher fasting and postprandial glucose and insulin concentrations and greater 24-h urinary glucose excretion than did patients fed eucaloric high-mono diets. The worsening of glycemic control in those other studies, however, may have been a consequence of forced eucaloric feeding. In some of these studies (17, 18), the diets were matched for fiber content. This matching represents an artificial situation because low-fat, high-complex carbohydrate diets tend to be naturally higher in fiber than are higher-fat diets. If the low-fat diets in the previous studies had been allowed to contain more fiber than the high-mono diets, it is possible that glycemic control might not have deteriorated, given the effect of dietary fiber in improving glycemic control (35).
There are several possible reasons that blood glucose control did not deteriorate during the low-fat diet in our study. One likely reason is that subjects were in negative energy balance and thus lost weight. Another possible reason is that the low-fat diet in our study contained more fiber than did the high-mono diet. In particular, the low-fat diet was higher in soluble fiber, which may improve glycemic control in patients with type 2 diabetes (35).
In summary, an ad libitum, low-fat, high-fiber, high-complex carbohydrate diet resulted in greater weight loss than did a high-mono diet, and the former did not increase plasma triacylglycerol concentrations from baseline or worsen glycemic control in patients with type 2 diabetes mellitus. Long-term consumption of a low-fat diet might be associated with even greater weight loss and additional improvement of plasma lipid concentrations. We conclude that ad libitum low-fat, high-fiber diets may be very useful in the dietary management of type 2 diabetes mellitus.
View this table:
APPENDIX A. Sample daily menu for the low-fat and high–monounsaturated fat (high-mono) diets
ACKNOWLEDGMENTS
We thank Gary Sexton, General Clinical Research Center biostatistician, for his help with the statistical analyses.
GTG and WEC contributed to the design of the study, were responsible for the day-to-day conduct of the study, and were primarily responsible for writing the manuscript. GTG was responsible for analysis of the data. AA contributed to the design of the study, carried out the insulin clamp studies, and helped manage the patients. KM wrote the Methods section of the manuscript as it pertained to diet. MPM and staff at the General Clinical Research Center provided the food, performed the dietary calculations, and weighed the patients. PBD contributed to the design of the study and manuscript review. None of the authors had any financial or personal interest in any company or organization sponsoring the research.
REFERENCES
Related articles in AJCN: