Diabetes Mellitus: A Fundamental and Clinical Text
3rd Edition

There are a number of ways in which the absorption of carbohydrate foods can be prolonged and the nutrient load can be spread over time (Table 74.2). The effects attributed to slowing absorption (Table 74.3) are many and relate to a range of metabolic abnormalities seen in type 2 DM associated with macrovascular and microvascular complications. Not all benefits have been associated with each method of slowing absorption. For this reason, the food frequency paradigm is used to illustrate the general principle because it is associated with most metabolic effects and is perhaps the cleanest model. This model
eliminates the potentially confounding differences in macronutrients and micronutrients between test and control treatments.

In its simplest form, the metabolic effects of absorption rate can be assessed by either taking glucose as a bolus or sipping it at an even rate over an extended period. When 50 g of glucose was taken in solution by healthy volunteers either as a bolus over 5 minutes or sipped at an even rate over 3 hours, major differences were seen in the insulin response (24) (Fig. 74.1). After sipping, the peak blood glucose level was reduced, although the incremental blood glucose area was not significantly different between treatments. However, the insulin area was reduced by 54%. A corresponding reduction was seen in plasma C-peptide levels on sipping compared with bolus ingestion, whereas free fatty acid (FFA) and branched-chain amino acid levels decreased similarly on both treatments despite the lower insulin levels on sipping (24) (Fig. 74.1). Furthermore, sipping appeared to avert the need for a counterregulatory response seen at 180 minutes after the glucose bolus because blood glucose levels showed no undershoot (24) (Fig. 74.1). Associated with the blood glucose undershoot after the glucose bolus was an increase in serum growth hormone and urinary catecholamine excretion. At the same time, there was a rebound increase in FFA levels and higher levels of branched-chain amino acids than seen after sipping. An intravenous glucose tolerance test given at 4 hours [and therefore performed against either suppressed FFA levels (postsipping) or high FFA levels (postbolus)] confirmed the more rapid glucose disappearance (K_g) after sipping, suggesting less resistance to glucose uptake (24) (Fig. 74.2).

The implications of this study are that provision of carbohydrate at a reduced rate over a longer time reduces the need for insulin. This effect may relate to altered FFA metabolism. Adipose tissue hormone-sensitive lipase is sensitive to insulin at levels that are below those required for cellular uptake of glucose (25). The small increase in insulin secretion above baseline after nibbling is nevertheless sufficient to suppress FFA release. Assuming that elevated serum FFA levels inhibit glucose uptake and utilization (26), the low levels of insulin acting over a prolonged period may be all that is required to prevent the increase in glucose in peripheral blood, provided that the glucose is absorbed sufficiently slowly from the gut. Further evidence for increased sensitivity to insulin after sipping is provided by the lower levels and more prolonged suppression of the insulin-sensitive branched-chain amino acids leucine, isoleucine, and valine, which, like glucose, are cleared by muscle under the action of insulin. The inhibitory effect of FFAs on glucose utilization is supported by the lower mean (K_g) value seen after intravenous glucose at 4 hours after the glucose bolus. At this time, FFA levels were elevated compared with the postsipping FFA level.

Studies of Ceriello and others suggest that elevated levels of glucose and FFAs and their subsequent metabolism reduce serum and tissue antioxidants and in turn increase insulin resistance (27,28,29,30). In this context, vitamin E supplementation was shown to improve carbohydrate tolerance in type 2 diabetes (31) which raises the question of whether slow-release carbohydrate results in less oxidative stress.

The principle of prolonging absorption also appears to have an application in type 2 DM. Figure 74.3 shows data derived from a physician volunteer with type 2 DM who took 240 g of glucose over a 12-hour period of observation, either in bolus form at four hourly intervals (with 80 g of glucose taken on each occasion) or sipped at an even rate over the entire 12-hour period (32).
As indicated (Fig. 74.3), on sipping glucose, levels declined progressively over the day. Mean insulin levels and total urinary glucose output were both reduced on sipping compared with bolus administration. With sipping, the respiratory quotient increased progressively through the day, as opposed to the sawtooth pattern seen on bolus administration, with each increase in respiratory quotient corresponding to a bolus (32). The reduced requirement for insulin secretion and the diminished urinary glucose losses are the potential advantages seen with spreading the nutrient load. The mechanisms involved in the improved carbohydrate tolerance are likely to be the same as for nondiabetic subjects.

Another illustration of the meal frequency phenomenon relates to the second meal effect or the effect that one meal may have on the next. Early on, it was recognized that one carbohydrate load may facilitate the disposal of the next (the Staub-Traugott effect) (33,34). This does not appear to be the case if the first meal is a low-carbohydrate meal by virtue of being low in total calories or high in fat (35,36). However, those meals that result in prolonged suppression of FFA levels appear to result in an improved glucose tolerance to the second meal. Studies in which viscous fiber (guar gum) was used to prolong absorption also demonstrated the improved second meal glucose tolerance and prolonged suppression of FFA levels accompanied by the prolonged suppression of 3-hydroxybutyrate (3-OHB) (37) (Fig. 74.4).
Similar effects on glucose tolerance have been seen when conventional foods have been fed as a meal or nibbled over a 4-hour period before a (second meal) test meal challenge (35).

It has been claimed that the second meal effect and the Staub-Traugott phenomenon referred to previously are not related phenomena (38). It is also true that the fat and protein may be major determinants of the glycemic response (39,40,41). However, there still appears to be sufficient evidence to implicate FFA metabolism in addition to alterations in counterregulatory hormones (42), which in the postprandial period may also promote glycogenolysis and glucose mobilization. The relationship of FFA concentration to subsequent postprandial glycemia has been illustrated by studies in healthy subjects demonstrating that postprandial glycemia was directly related to the antecedent FFA concentration (43) (Fig. 74.5).

Of even greater interest is the concept that blood glucose levels do not necessarily relate to the flux of glucose through the system (44). Tracer studies have demonstrated that although glucose absorption after an 80-g glucose tolerance test continues at a high rate for an approximately 3-hour period, the glycemia lasts only for 90 minutes or less in healthy subjects (45). Thus, during the latter part of the glucose tolerance test, the rate of peripheral tissue uptake equals or exceeds the rate of glucose absorption. By inference, the glucose increase after a second meal (taken while glucose is still being absorbed from the first meal and FFA levels therefore are suppressed) should meet less “resistance,” that is, should cause less rise in the peripheral blood glucose level—hence the “improved” second
meal glucose tolerance and the advantage of increased meal frequency (nibbling vs. gorging).

Interest in the health implications of altered food frequency dates back at least as far as Sanctorius in the late sixteenth and early seventeenth centuries. His observations were encapsulated in one of his axioms: “He who eats but once a day destroys himself whether he eats little or a lot.” However, it was not until the early 1930s that Ellis (46) applied the principle to the management of DM. He showed that hospitalized patients with type 1 and possibly some type 2 DM could be managed more effectively with better glycemic control and less insulin if they were provided with small but frequent amounts of glucose and insulin throughout the day (46). However, it remained for Fabry (47) in the late 1950s and 1960s to make the claim, based on a prospective study of 1,359 men over 60 years of age in Prague, that increased meal frequency was associated with reduced CHD, DM, and obesity. At that time, his observations sparked considerable interest in the effects of meal frequency in improving glucose tolerance (48), lowering serum cholesterol levels (49,50), and favorably influencing adipose tissue enzyme levels that promote fatty acid mobilization rather than storage (51).

More recently, studies were carried out in normal volunteers who were placed in an extreme model of increased food frequency at 17 meals daily (one meal per hour for 16 hours) compared with a three-meal-per-day schedule for 2 weeks in a randomized crossover design (52). The day profiles at the end of each period (52) (Fig. 74.6) indicated a flatter blood glucose profile on nibbling, although the incremental glucose area (above time zero) was not different between the two groups. However, both mean insulin and C-peptide levels over the day were lower on nibbling, confirming the previous observation in which “sipping” was used to “spread the nutrient load.” Nevertheless, the difference in FFAs and 3-OHB levels over the day did not reach significance between treatments (Fig. 74.6), nor were the branched-chain amino acid levels (not shown) significantly different between treatments.

Single-day feeding studies have also been carried out in type 2 DM, showing an advantage with increased meal frequency (53,54). Studies of hourly snacks compared with 4 meals during the 10 hours of observation (Fig. 74.7) showed both lower mean glycemia and insulinemia and a flatter gastric inhibitory polypeptide profile (53). In addition, 24-hour urinary C-peptide output was reduced. However, again as with the nondiabetic volunteers, no significant difference was seen in the FFA, 3-OHB, or branched-chain amino acid responses. Other studies have produced similar results (54). Isocaloric feeding of six meals compared with two large meals reduced glucose excursions, insulin, and FFA levels during the day (54).

Studies conducted in the mid-1960s noted reductions in total cholesterol levels with increased meal frequency (49,50). More recently, studies have demonstrated that the reduction was in low-density lipoprotein cholesterol when 3 meals were compared with 9 or 17 meals daily for 8 and 2 weeks, respectively (52,55,56), using both metabolic and ad libitum diets. After 17 meals daily, lower levels of apolipoprotein B were also demonstrated (52) (Fig. 74.8). Population studies of middle-aged and older volunteers have confirmed that total cholesterol levels were lower in those who ate more frequent meals daily (57). In addition, mean plasma lipid levels over the course of the day
were lower with greater meal frequency (58). Stable isotope studies indicate that cholesterol synthesis was reduced at greater meal frequency (59), and studies using urinary mevalonic acid excretion as a water-soluble marker of cholesterol synthesis indicated that the change in cholesterol levels also related to the change in urinary mevalonic acid output (58). The reduction in cholesterol synthesis has been attributed to the lower insulin levels observed because insulin is known to stimulate 3-hydroxy-3-methyl-glutaryl coenzyme A reductase activity, a rate-limiting enzyme in cholesterol synthesis (60). A further possible reason for the reduction in serum cholesterol level on nibbling is the increased loss of bile acids that would result from the increased frequency of cycling of bile acids through the gut with each snack. The resulting increased losses of the cholesterol molecule as bile acids would again promote the cholesterol-lowering effect of increased meal frequency. There is no solid body of studies assessing the effects of food frequency on DM control and hyperlipidemia. The original studies looked hopeful, but two ad libitum studies using nine meals per day compared with three showed no advantage (61,62). More studies are required to indicate why normolipidemic subjects respond and hypercholesterolemic subjects may not.

Serum uric acid levels were reduced and urinary uric acid excretion was increased with increased food frequency (58). As with the reduction in serum cholesterol levels, the effect of lower insulin levels was invoked as an explanation (58). It was suggested that insulin promoted renal reabsorption of uric acid, and this idea has been discussed in the context of hyperinsulinemic states (63).

Additional physiologic effects of food frequency have been explored that are relevant to DM. In humans, some acute studies of the effects of meal frequency have demonstrated a reduced thermogenic response to increased meal frequency, whereas others have demonstrated greater thermic responses to larger, less frequent meals (64,65,66,67,68). However, 24- to 48-hour studies in metabolic chambers failed to show differences in energy expenditure related to meal frequency, and it therefore seems unlikely that alterations in the thermic effect of foods will provide a major weight loss advantage for increased meal frequency (69). However, assessment of satiety in acute studies suggests that satiety fluctuates less over the day and thus does not dip down into the “hunger” range (53). The all-important long-term studies have not been undertaken to determine whether in the long term fewer calories are consumed. Until then, there remains the concern that snacking may increase the body weight of those who most need to lose weight. Nevertheless, regardless of whether increased meal frequency as such is broadly applicable in practice, the demonstration that it can improve certain aspects of lipid and carbohydrate metabolism makes it a valuable model for other methods of spreading the nutrient load.

Viscous gums, gels, and mucilages, depending on their viscosity, have also been shown to reduce the rate of absorption of
sugars and to flatten the postprandial glycemia and the serum insulin response (70). These fibers have also been shown to reduce low-density lipoprotein cholesterol and apolipoprotein B levels (71,72,73,74,75,76). The so-called particulate insoluble fibers such as wheat bran increase fecal bulk but have little effect on postprandial glycemia (70) or other metabolic events.

The lack of enthusiasm in using viscous fibers in the treatment of DM stems from the relative difficulty in making palatable formulations for long-term use. This has limited the number and size of the studies undertaken. Initial test meal studies in diabetic patients appeared promising (77). However, in the 1980s, the results of studies did not indicate a clear improvement in glycemic control (78). As a result, guidelines have considered soluble fibers to play no role in glycemic control (5,6). More recently, a number of studies have appeared suggesting a possible role for soluble viscous fibers in the treatment and prevention of DM complications. These include studies of guar gum in healthy men that demonstrated reduced low-density lipoprotein cholesterol and blood pressure and increased urinary sodium loss (79). In parallel test and control studies in type 2 DM, hemoglobin A_1C (HbA_1C) levels were reduced in those taking the test fiber (guar) but not in the control subjects, and a similar picture of reduction in the test situation was seen in relation to low-density lipoprotein cholesterol and the low-density lipoprotein:high-density lipoprotein (HDL) cholesterol ratio (80). Furthermore, long-term studies of guar lasting 48 weeks in volunteers with type 2 DM showed a mean reduction in HbA_1C on guar, which increased significantly on cessation of guar supplementation (81). These studies do not demonstrate the gold standard of improved glycemic control in which HbA_1C is shown to decrease in long-term, randomized, cross-over design studies (78). However, this goal is often difficult to achieve even for approaches advocated for the treatment of DM (e.g., monounsaturated fats) and for which there is considerable support (6). Use of viscous fibers in the treatment of DM therefore continues to be of interest but requires further work to maximize the palatability and effectiveness of preparations for trial purposes.

Particulate fibers in the diet, typically in the form of wheat bran, has been associated in prospective cohort studies with freedom from cardiovascular disease and diabetes (82,83,84,85). Over the years a number of intervention studies have indicated that wheat bran reduces blood glucose in subjects with impaired glucose tolerance (86), improves glucose tolerance in healthy individuals, and reduces serum cholesterol concentrations (87,88). However, there is no clear physiologic reason why wheat bran should have metabolic advantages, although it has been suggested that the magnesium content of wheat bran may be responsible for the improved carbohydrate tolerance (84,89). It is also possible that some aspect of the whole grain is responsible, and improved carbohydrate tolerance has been reported when insulin-resistant subjects were fed wheat bran (90). Current evidence does not suggest any benefit of wheat bran in the treatment of type 2 diabetes either in terms of glycemic content or improvement in the bood lipid profile when type 2 diabetes patients received 20 g additional wheat fiber in bread and breakfast cereals over a 3-month period in a randomized controlled cross-over study (91).

The original impetus to classify foods using a glycemic index was to provide a ranking of the rates at which different starchy foods were digested (92,93). It had been well documented that different starchy foods produced different glycemic effects (94). It was hoped that selection of foods with lower glycemic indices would contribute to prolonging the absorption of nutrients, thereby improving the glycemic profile (95,96) and reducing insulin requirement and fasting lipids (97).

Nevertheless, certain acute (up to 1 day) mixed-meal studies during the middle to late 1980s suggested that a glycemic index classification of foods had no clinical utility (98,99,100). However, high-fiber diets have been shown to improve insulin sensitivity in healthy subjects (101) and in CHD patients (102). Metabolic benefits also have been shown in subjects with diabetes. Since the late 1980s, when the effects of high– and low–glycemic index diets on glycemic control were compared within populations of type 1 and 2 DM patients, 10 of 14 studies, of 2 weeks’ to 12 months’ duration, reported reductions in serum fructosamine or HbA_1C levels (103,104,105,106,107,108,109,110,111,112,113,114,115,116) (Table 74.4). In a recent metanalysis, the mean difference in glycated proteins between the low– and high–glycemic index diets was –7.4%, thereby favoring low–glycemic index diets (117). In some cases, changes in blood measurements have been noted despite relatively small differences in glycemic index between test and control diets. Furthermore, some studies also noted reductions in serum lipids (103,108,109,110,111). The changes are therefore similar to those seen with viscous fiber, despite the absence of these fibers in some of the low–glycemic index diets (118). Cohort studies have provided support for the importance of the glycemic index of the diet in preventing the development of DM in men (119) and women (120) and in reducing the risk for cardiovascular disease (121). The glycemic index of the antecedent diet expressed as the glycemic load (mean glycemic index times total carbohydrate intake) related to the development of DM 6 years later (Fig. 74.9).

Prospective studies suggested that low–glycemic index diets may decrease the risk of CHD independently (121) and as part of a healthy lifestyle (122). In relation to blood lipids, a prospective study showed that high glycemic index was associated with high triglyceride levels (123), whereas two cross-sectional studies found the dietary glycemic index to be inversely related to HDL cholesterol (124,125).

Another potential benefit seen for the glycemic index is in weight control and satiety (126,127). The effect of high–, medium–, or low–glycemic index breakfast meals on subsequent ad libitum food intake was studied in obese teenage boys (126). Reductions in energy intake of 53% and 81% in the medium– and low–glycemic index groups, respectively, compared with the high–glycemic index group, was observed 5
hours after breakfast. These results suggest that in isoenergetic meals, slowly digested carbohydrate-rich foods may allow a sense of satiety to last longer than rapidly digested foods. A characteristic effect of high–glycemic index foods is the hypoglycemic undershoot as a possible consequence of high insulin and low glucagon levels, which would induce glucose storage, inhibition of lipolysis, and reduction of glucose availability for metabolic oxidation (hypoglycemic undershoot). This metabolic state is similar to a fasting state and would trigger glucagon release and hunger signals. Low glycemic index foods, however, tend to maintain glucose and insulin at a moderate level, avoiding the hypoglycemic state. Also, low–glycemic index foods that are rich in dietary fiber may produce distension of the gastrointestinal tract, which may further explain the enhanced satiety level. Cholecystokinin (CCK), a gut peptide that induces satiety, is thought to be directly affected by gastric volume. Meal glycemic index has been found inversely proportional to CCK response and satiety (128), suggesting a possible role of gastric volume and of bulky foods in the maintenance of appetite suppression.

There have been concerns that high carbohydrate intakes at the expense of fat, particularly monounsaturated fat (129,130,131), could result in elevated postprandial insulin and glucose levels, an increase in fasting triglycerides and very-low-density lipoproteins, and suppression of HDL levels, which could translate into a greater risk of heart disease (132,133,134). However, not all carbohydrate-rich diets produce the same effects on blood lipids, because low–glycemic index diets may be associated with a more favorable lipid profile than high–glycemic index diets. Lowering the dietary glycemic index by at least 12
points reduced triglycerides by approximately 9% in 10 of 11 studies (135). This effect appeared most marked when triglyceride levels were increased (97). Recent data also showed that a high-carbohydrate diet made of low–glycemic index foods significantly increased HDL levels by 5.4% compared with an isocaloric high-carbohydrate/high–glycemic index diet (114). In addition, cross-sectional data (124,125) showed that dietary glycemic index was inversely related to HDL cholesterol levels, which in turn were inversely related to triglycerides, and that glycemic index was a stronger predictor of serum HDL levels than dietary fat (124). Other investigators have shown that the unwanted HDL reductions seen with some high-carbohydrate diets may be transient (136).

α-Glucosidase inhibitors such as acarbose, which reduce the rate of absorption of starch, sucrose, and, to a lesser extent, maltose (137), have been shown in large, multicenter trials to result in a significant reduction in HbA_1C in type 2 DM (138,139). Furthermore, use of acarbose in the context of the UKPDS trial in diabetic patients improved glucose control, expressed as reduced HbA_1C, to a degree similar to that achieved by current hypoglycemic therapy (e.g., metformin and sulphonylurea) (22). Finally, in the STOP-NIDDM trial, subjects with impaired glucose tolerance who received 100 mg of acarbose three times daily showed a significantly reduced rate of conversion to diabetes versus the control group (140). These issues are addressed in detail in Chapter 79 of this text. Changes in serum lipids in general are not observed. Nevertheless, findings of this nature provide additional encouragement that the principle of spreading the nutrient load by dietary means, in addition to altering the amount and nature of the macronutrients, may one day have a role in modifying glycemia in the management of diabetes.

The hypothesized metabolic effects relate to the rate at which glucose is absorbed from the small intestine. A reduced rate of glucose absorption after consuming low–glycemic index carbohydrate foods will reduce the postprandial increase in gut hormones (e.g., incretins) and insulin. The prolonged absorption of carbohydrate seen over time will maintain suppression of FFAs and the counterregulatory responses while at the same time achieving a lower blood glucose level. Over time, with the reduction in FFA levels and the increase in respiratory quotient (RQ) with tissue insulinization, glucose is withdrawn from the circulation at a faster rate. Consequently blood glucose levels return toward baseline despite continued glucose absorption from the small intestine. The peak postprandial blood glucose increase is therefore reduced together with the incremental blood glucose area above baseline. This improvement may in part be the result of sustained tissue insulinization, suppression of FFA release (50,51), and the absence of a counterregulatory endocrine response (47,51). Also, an improved second meal glycemia, reminiscent of the Staub-Traugott effect (where the first meal improves glucose tolerance of the second meal) (33,34,35) with consequent lower FFAs is seen with low–glycemic index meals (35).

Another mechanism indirectly related to disease is the possible role of the glycemic index in modulating body weight through satiety (141). Short-term studies on altering glycemic index and load have indicated that the lower the glycemic index and load of the first meal, the less food is consumed in the subsequent meal (126).

There is considerable interest in the links between insulin resistance, the generation of reactive oxygen species, tissue damage, and the liberation of proinflammatory cytokines and acute phase proteins, the latter appearing to be powerful markers of chronic diseases, notably CHD (142). It is possible that the dietary glycemic index may play a role in this sequence of events. In one study, C-reactive protein has been found to relate to high–glycemic load diets (143).

Finally, studies have demonstrated that the postprandial increase in glucose is associated with depression of serum antioxidants, including lycopene and vitamin E (144,145). Presumably the higher the glycemia, the greater the postprandial depression of serum antioxidants (145). The concept is evolving that increased insulin resistance results from oxidative stress. Supplementing subjects with the antioxidant vitamin E has been shown to improve glycemic control (146). Studies such as these suggest a possible beneficial role for low–glycemic index diets through reducing oxidative damage.

Longer-term studies, however, are required to define the relevance of these interesting findings.

A relatively new general approach to the dietary therapy of type 2 DM is a conscious attempt to try to spread the nutrient load or lengthen the absorption time. This approach covers the effects of altered meal frequency, viscous dietary fibers, low–glycemic index foods, and inhibitors of carbohydrate absorption. In its simplest form, it is illustrated by studies of altered meal frequency (nibbling vs. gorging). Reducing the size and increasing the frequency of carbohydrate feedings, either by sipping glucose or nibbling meals, has been shown acutely to result in lower mean blood glucose and insulin levels over the day in normal volunteers and patients with type 2 DM and reduced 24-hour urinary C-peptide losses. In the longer term in nondiabetic subjects, fasting and postprandial total and low-density lipoprotein cholesterol levels are reduced, together with fasting apolipoprotein B and serum uric acid levels, as risk factors for CHD. These and other physiologic effects make slowing carbohydrate absorption (lente carbohydrate) a potentially useful therapeutic modality. Clinically slowing the rate of carbohydrate absorption by the use of α-glucosidase inhibitors is now a recognized therapeutic modality in the treatment of type 2 DM. In terms of diet, more slowly absorbed or low–glycemic index carbohydrate foods appear to reduce the risk both of DM and cardiovascular disease.

The mechanism for the glycemic advantage of prolonging absorption may relate to the more efficient uptake and metabolism of glucose over time, possibly in part as a result of prolonged
suppression of FFA levels. The lower insulin levels observed may confer additional advantages in reducing the hepatic drive to cholesterol synthesis, reducing uric acid reabsorption by the kidney, increasing HDL levels and possibly decreasing C-reactive protein. Risk factors for CHD may therefore be reduced. Furthermore, increased bile acid losses through the colon may contribute to lower cholesterol levels. This effect may be the result of increased enterohepatic cycling of bile with increased meal frequency or of reduced ileal bile acid retrieval due to entrapment by viscous fibers. The metabolic effects of these nutritional maneuvers will further increase the value of spreading the nutrient load to reduce the risk of complications in DM.