Breakfast Glycaemic Response in Patients with Type 2 Diabetes: Effects of Bedtime Dietary Carbohydrates

Breakfast glycaemic response in patients with type 2 diabetes: Effects of bedtime dietary carbohydrates

M Axelsen1*, R Arvidsson Lenner 2,PLoÈnnroth1 and U Smith 1

1The Lundberg Laboratory for Diabetes Research, Department of Internal Medicine, Sahlgrenska University Hospital, S-413 45 GoÈteborg, Sweden; and 2Department of Clinical Nutrition, Department of Internal Medicine, Sahlgrenska University Hospital, GoÈteborg University, Sweden

Objectives: Bedtime carbohydrate (CHO) intake in patients with type-2 diabetes may improve glucose tolerance at breakfast the next morning. We examined the `overnight second-meal effect' of bedtime supplements containing `rapid' or `slow' CHOs. Design: Randomized cross-over study with three test-periods, each consisting of two days on a standardized diet, followed by a breakfast tolerance test on the third morning. Setting: The Lundberg Laboratory for Diabetes Research, Sahlgrenska University Hospital, GoÈteborg, Sweden. Subjects: Sixteen patients with type 2 diabetes on oral agents and=or diet. Interventions: Two different bedtime (22.00 h) CHO supplements (0.46 g available CHO=kg body weight) were compared to a starch-free placebo (`normal' food regimen). The CHOs were provided as uncooked cornstarch (slow-release CHOs) or white bread (rapid CHOs). Results: On the mornings after different bedtime meals we found similar fasting glucose, insulin, free fatty acid and lactate levels. However, the glycaemic response after breakfast was 21% less after uncooked cornstarch compared to placebo ingestion at bedtime (406Æ 46 vs 511Æ 61 mmol min 171, P < 0.01). In contrast, it did not differ when the evening meal consisted of white bread (451Æ 57 mmol min 171) compared to placebo. According to an in vitro analysis, uncooked cornstarch contained 4 times more slowly digestible starch as compared to white bread. Conclusions: A bedtime meal providing uncooked cornstarch improved breakfast tolerance the next morning while, in contrast, this was not found following a bedtime meal of white bread. The results are consistent, therefore, with the concept that an increased intake of slowly digestible carbohydrates exert an overnight second- meal effect in patients with type 2 diabetes. Sponsorship: This study was supported by grants from the Swedish Medical Research Council (project 3506) and the Ingabritt and Arne Lundberg Foundation. Descriptors: blood glucose; insulin; dietary carbohydrates; diabetes; type 2; dietary starch

Introduction glucose concentrations (Bolli, 1988; Ferrannini et al, 1988; Firth et al; McMahon et al, 1989; Mitrakou et al, 1990). Type 2 diabetic patients exhibit important changes in Finding a regimen which alleviates the insulin resistance in diurnal insulin sensitivity, with a more pronounced insulin the morning should, thus, be considered critically important resistance in the morning as compared to in the afternoon in the management of type 2 diabetes. The potential role of (Atiea et al, 1987; Dimitriadis et al, 1988; Perriello et al, the diet, such as timing and release pro®le of various 1998; Shapiro et al, 1991). This perturbation has a negative carbohydrates, remains to be de®ned. effect on both the fasting and morning postprandial blood The concept of a second-meal effect has attracted inter- est in recent years (Abraira et al, 1987; Jenkins et al, 1982; Shaheen & Fleming, 1987; Wolever et al, 1995, 1988). *Correspondence: Dr M Axelsen, The Lundberg Laboratory for Diabetes Brie¯y, it may be de®ned as the ®rst of two consecutive Research, Department of Internal Medicine, Sahlgrenska University carbohydrate loads leading to improved glucose tolerance Hospital, S-413 45 GoÈteborg, Sweden. to the second. Wolever et al, (1988) showed in healthy Contributors: MA: conception and design; provision of study materials and patients; administrative, technical, and logistic support; collection and subjects that breakfast carbohydrate tolerance can be assembly of data; analysis and interpretation of data; statistics; drafting of improved when low-glycaemic-index foods are eaten the the article; critical revision of the article for important intellectual content. previous evening. Recently, we have shown that a large RAL: Conception and design; administrative, technical, and logistic amount of bedtime carbohydrates, provided as uncooked support; critical revision of the article for important intellectual content. PL: conception and design; critical revision of the article for important cornstarch, also leads to an improved glucose tolerance the intellectual content. US: Conception and design; administrative, technical, next morning in type 2 diabetic patients (Axelsen et al, and logistic support; analysis and interpretation of data; statistics; drafting 1997). However, the mechanism(s) for the overnight of the article; critical revision of the article for important intellectual second-meal effect by cornstarch ingestion were not fully content; obtaining of funding. elucidated. Since uncooked cornstarch possesses a slow Guarantor: M Axelsen and U Smith. Received 28 January 1999; revised 26 February 26 1999; accepted 15 release pro®le in vivo (Axelsen et al, 1999; Chen et al, March 1999. 1984; Collings et al, 1981; Hayde & Widhalm, 1990; Bedtime carbohydrates in type 2 diabetes M Axelsen et al 707 Wolfsdorf & Crigler Jr, 1997), the lente characteristic of Table 2 Standardized meals prior to the breakfast study the food may be critical in eliciting the overnight second- meal effect in type 2 diabetic patients. Weight g The aims of this present study, therefore, were to Breakfast at 07.00 h examine the overnight second-meal effect of a bedtime Wholemeal bread 70 supplement of `rapid' and `slow' starch and to compare that Butter 8 Bologna 15 to the effect of a starch-free placebo (a `normal' meal- Beef 10 regimen) in type 2 diabetic patients. In addition, an in vitro Orange 105 analysis was undertaken to examine the content of rapidly Snack at 10.00 h and slowly digestible starch in uncooked cornstarch and in Wholemeal bread 50 two reference foods. Butter 5 Ham 18 Lunch at 12.00 h Day 1: Chili con carne 300 Material and methods Day 2: Goulash 310 Sixteen patients with type 2 diabetes were recruited by White bread 50 Butter 8 means of newspaper advertisement. They had a glycated Berries 100 hemoglobin A1c below 10% (ref. 3.3 ± 5.3%) and fasting Coffee cream 30 blood glucose below 12 mmol=l. None of the patients were Sugar 10 treated with insulin. Treatments and clinical characteristics Snack at 15.30 h Banana 120 of the patients are shown in Table 1. All participants gave Dinner at 18.00 h their informed consent and the study was approved by the Shrimps 200 Ethical Committee of GoÈteborg University. Wholemeal bread 50 Butter 8 Metabolic study Mayonnaise 30 Pear 110 The patients were studied in three consecutive periods in Evening snack at 20.00 h random order, with 5 ± 10 d separating each period. During Wheat-bun 50 the test periods the patients adhered to a standardized diet at home for 2 d. In the evenings at 22.00 h a bedtime meal was taken. A meal tolerance test was performed in the Table 3 Ingredients of the bedtime meals laboratory on the third morning. The standardized diet (Table 2), which was prepared in advance, was stored at Placebo Cornstarch White bread 7 18C and thawed on the day of consumption. The energy g=kg body wt g=kg body wt g=kg body wt content was 8.4 MJ (2000 kcal), with 19% of energy (E%) Water 2.00 1.67 1.67 from protein, 36 E% from fat and 45 E% from carbo- Low-sugar fruit juice, 0.12 0.12 0.12 hydrates, and the ®ber content was 46 g=d. Beverages dry matter were standardized individually. The patients were informed Pectin 0.033 Ð Ð to abstain from exercise and, if applicable, to carefully Uncooked cornstarch Ð 0.55 Ð White bread Ð Ð 0.96 follow the prescribed times for medical treatments. Available carbohydrates Ð 0.46 0.46 Three different bedtime meals at 22.00 h were tested from starch (Table 3). The uncooked cornstarch (Maizena1, CPC Proteina < 0.01 0.08 Foods AB, Kristianstad, Sweden) and bread supplements Fata Ð < 0.01 0.03 provided an identical amount of available carbohydrates, aValues according to food tables. 0.46 g=kg body weight. The placebo contained only a thickening-agent (pectin) and, hence, no available starch. The cornstarch and placebo ingredients were pre-mixed and was placed in the left arm, which was kept in heating pads dispensed in jars. In the bread period, the juice was taken to arterialize the blood. Samples were collected at 7 10 along with the bread. The meals provided the same amount and 0 min for analysis of glucose, insulin, free fatty acids of water, and no other liquids were allowed for 1 h before and lactate. The standardized breakfast consisted of whole- or after the meals. grain bread with butter, cheese (28% butterfat) and ham The patients came to the laboratory at 07.30 h on day and was served at 08.00 h, with coffee or tea in a standar- three, fasting from 22.10 h the previous evening. Hypo- dized amount. The calculated energy content was 2.0 MJ glycaemic agents, if any, were taken at the laboratory 20 min (484 kcal) and 1.8 MJ (420 kcal) for males and females, before being served a standard breakfast. A venous catheter respectively, of which 14 E% were derived from protein, 40 E% from fat and 46 E% from carbohydrates. Postpran- dial blood samples were collected every half hour for 4 h. Table 1 Clinical characteristics of the patients Blood glucose was analyzed by an automatic glucose Sex (M=F) 14=2 analyzer (YSI, Yellow Springs Instruments, Texas). The Age (y) 61Æ 9 (40 ± 73) remaining samples were centrifuged, frozen and stored at Fasting blood glucose (mmol=l) 8.4Æ 1.7 (6.6 ± 11.7) 7 20C until analysis. Plasma free fatty acids were deter- HbA1c(%) 6.5Æ 0.9 (5.0 ± 7.8) BMI (kg=m2) 29.2Æ 5.0 (22.5 ± 44.2) mined with an enzymatic colorimetric method (Wako waist=hip-ratio 0.95Æ 0.07 (0.85 ± 1.08) Chemicals GmbH, Neuss, Germany). Plasma insulin was Oral agents: determined with radioimmunochemical analyses (Pharma- Metformin (n) 4 cia Insulin RIA, Pharmacia AB, Uppsala, Sweden) and Sulphonylurea (n) 8 plasma lactate by an automatic analyzer (YSI, Yellow MeansÆ s.d. (range), BMI; body mass index. Springs Instruments, Texas). Bedtime carbohydrates in type 2 diabetes M Axelsen et al 708 Starch analyses A starch analysis was undertaken for: (1) white bread; (2) uncooked cornstarch; and (3) pumpernickel-bread with 100% kernels (Schwarzbot, Delba, Delbrck, Germany). The white bread and cornstarch samples were from the same batch of foods used in the metabolic study. The pumpernickel bread was included as a reference product for a `slowly digestible' starchy food. Because the cornstarch ingredients were pre-mixed before the actual ingestion by the patients, an additional digestibility analysis was performed on uncooked cornstarch which had been left in water for 48 h. The amounts of rapidly digestible starch, slowly digestible starch, resistant starch and total starch were analyzed by in vitro enzymatic hydrolysis and the released glucose measured by glucose oxidase (GOD-PAP, Boehringer), as previously described (Englyst et al, 1992). Brie¯y, chewing was mimicked by passing the bread through a standardized mincer (0.9 cm diameter holes). Samples were put in tubes containing pepsin, pectin, hydrochloric acid and glass-beads and were shaken in a water bath (37C) for 30 min. Rapidly digestible starch (RDS) was de®ned as the glucose released after 20 min incubation with pancreatin (Pancrex V, Paines & Byrne, Greenford, Middx, UK) and amyloglycosidase (EC 3.2.1.3, Novo Norisk, Copenhagen). After a further 100 min, a second sample was removed, de®ned as slowly digestible starch (SDS). Total starch (TS) was de®ned as the glucose released by complete enzymatic hydrolysis of starch after gelatinization in boiling water and treatment with potas- Figure 1 (A) Mean glucose (top) and (B) insulin (bottom) levels during a sium hydroxide to disperse retrograded amylose. Resistant standardized breakfast in 16 type 2 diabetic patients after ingestion at starch (RS) was de®ned as the starch not hydrolyzed after 22.00 h the previous evening of uncooked cornstarch (*) and white bread (s) or carbohydrate free placebo (u). Vertical bars represent standard 120 min incubation (TS 7 (RDS SDS), and available error (SE). IAUC; area under incremental curve. starch (Table 3) de®ned as TS-RS.

Statistical analyses The data are presented as meansÆ s.e.m., unless otherwise consumption. The post-breakfast IAUC or peak levels for stated. Data are expressed as fasting values, peak responses insulin (Figure 1B), and AUC or nadir levels for FFA (data and the incremental area under the curve above baseline not shown), were similar after the different bedtime snacks. (IAUC) (Matthews et al, 1990). Statistical signi®cances However, the lactate IAUC after breakfast was higher between cornstarch and placebo, and between white bread after cornstarch consumption (104Æ 14 vs 83Æ 14 mmol and placebo, were evaluated by Wilcoxon's paired rank- min 171 for cornstarch and placebo, respectively, P < sum test. All signi®cances are two-tailed and a P-level of 0.05), but not after intake of white bread (95Æ 24 mmol < 0.05 was considered statistically signi®cant. min 171).

Results Starch analyses The results from the in vitro digestibility analyses are Metabolic study shown in Table 4. The total starch content per 100 g On the morning after uncooked cornstarch, white bread or ranged from 32 ± 89 g. Expressed as % of total starch, placebo at 22.00 h, similar fasting levels (08.00) of blood rapidly digestible starch represented 88% of the starch in glucose (8.2Æ 0.5, 8.2Æ 0.5 and 8.1Æ 0.4 mmol=l), insulin white bread (Table 4). In contrast, uncooked cornstarch and (11.9Æ 1.4, 11.6Æ 1.3 and 12.5Æ 1.4 mU=l), FFA pumpernickel bread contained similarly low proportions (0.67Æ 0.6, 0.70Æ 0.06 and 0.69Æ 0.06 mmol=l) and lactate (38% and 48%) of rapidly digestible starch, and high (1.1Æ 0.1, 1.2Æ 0.1 and 1.2Æ 0.1 mmol=l) were obtained. proportions (55% and 51%) of slowly digestible starch. However, the IAUC for glucose after breakfast (Figure 1A) The contribution of resistant starch from all foods were was 21% less following cornstarch compared to placebo minor (data not shown), the highest being 6 g=100 g dry- consumption the previous evening (406Æ 46 vs weight in uncooked cornstarch. We found no effect on 511Æ 61 mmol min 171, P < 0.01). In contrast, the post- starch digestibility by soaking the cornstarch in room- breakfast glucose IAUC after white bread was tempered water for 48 h (Table 4). 451Æ 57 mmol min 171, which did not differ signi®cantly from placebo consumption in the previous evening. The blood glucose peak after breakfast subsequent to cornstarch Discussion (12.0Æ 0.5 mmol=l) was lower than the peak following placebo consumption the previous evening (12.7Æ An increase in carbohydrates in the evening meal does not 0.4 mmol=l) (P < 0.05) but, again, did not differ after affect glycaemic control the next morning in type 2 diabetic white bread (12.4Æ 0.5 mmol=l) compared to after placebo patients (Axelsen et al, 1997; Beebe et al, 1990). However, Bedtime carbohydrates in type 2 diabetes M Axelsen et al 709 Table 4 Starch digestibility in vitro The clinical implications of this present study remain to Types of starcha be shown. A bedtime carbohydrate supplement will raise the nocturnal glucose levels, clearly an unwanted change in Rapidly digestible Slowly digestible Total type 2 diabetic patients. However, a low dose ( 25 g) also starch starch starch (TS) elicits bene®cial effects on morning glucose metabolism, at g=100 g (% of TS) g=100 g (% of TS) g=100 g least in patients with well-preserved endogenous insulin secretion (Axelsen et al, unpublished observations). Hence Uncooked cornstarch, 34.3 (38%) 49.2 (55%) 89.1 a temporal change in carbohydrate intake, whereby the dry bene®cial effects of a diminished breakfast carbohydrate Uncooked cornstarch 31.3 (35%) 51.5 (57%) 89.7 wet for 48 h content (Stilling et al, 1998) are combined with a bedtime White bread 42.1 (88%) 5.8 (12%) 47.9 slow-release carbohydrate supplement, may be worthy of Pumpernickel bread 15.6 (48%) 16.5 (51%) 32.2 study in type 2 diabetic patients. In addition, a lipotoxic effect of free fatty acids on the pancreatic insulin release aResistant starch not shown. (Shimabukuro et al, 1998) and an elevated endogenous glucose production (Perriello et al, 1988) are present also in we speculated that this is due to the carbohydrates being obese patients with impaired glucose tolerance. Whether supplied by `rapid' foods. White bread is, by de®nition bedtime intake or slow-release carbohydrates may alleviate (Wolever et al, 1991), a high glycaemic food. The novel these perturbations and, therefore, prevent or delay the ®nding of this present study is that a bedtime meal provid- development of manifest diabetes in these individuals, is ing carbohydrates from uncooked cornstarch positively an intriguing possibility that warrants further studies. affected breakfast tolerance the next morning while, in contrast, this was not found following a bedtime meal consisting of white bread. According to the in vitro analy- Conclusions sis, uncooked cornstarch had a similar release pro®le as pumpernickel bread, a well-recognized `slow' food (Jen- A bedtime meal providing carbohydrates from uncooked kins et al, 1986). The data, thus, suggest the importance of cornstarch positively affects the glucose tolerance the next a slow release pro®le of the carbohydrates in eliciting an morning in type 2 diabetic patients, while this effect is not overnight second-meal effect in type 2 diabetic patients. achieved with a bedtime meal consisting of white bread. The purpose of this present study was to de®ne the Uncooked cornstarch contain half the amount of rapidly preferable kind of carbohydrates a dietary supplement digestible carbohydrates but 4 times more slowly diges- should contain in order to exert an overnight second-meal tible starch as compared to white bread. The results are effect in type 2 diabetics. The results were consistent with consistent, therefore, with the concept that an increased the concept that slowly digestible carbohydrates exert a intake of slowly digestible carbohydrates exerts an over- second-meal effect. In addition to types of carbohydrates, night second-meal effect in patients with type 2 diabetes. the amounts used are probably also of importance. In both the white bread and cornstarch periods, the amounts of References carbohydrates were increased compared to the `normal' period. In contrast, when comparing the glycaemic Abraira C, Buchanan B & Hodges L (1987): Modi®cation of glycogen deposition by priming glucose loads: the second-meal phenomenon. response after breakfast in the white bread and cornstarch Am. J. Clin. Nutr. 45, 952 ± 957. periods, it tended to be less after cornstarch than after bread Atiea J, Ryder R, Vora J, Owens D, Luzio S, Williams S & Hayes T but this effect was less pronounced than compared to the (1987): Dawn phenomenon: Its frequency in non-insulin-dependent normal meal regimen (Figure 1). However, it may be diabetic patients on conventional therapy. Diabetes Care 10, 461 ± 465. Axelsen M, LoÈnnroth P, Arvidsson Lenner R & Smith U (1997): argued that the high fat content of the breakfast meal Suppression of the nocturnal free fatty acid levels by bedtime cornstarch (40% of the energy) may diminish the `glucotoxic' effect in NIDDM. Eur. J. Clin. Invest. 27, 157 ± 163. of high glucose peaks, due to inhibition of the gastric Axelsen M, Wesslau C, Arvidsson Lenner R, LoÈnnroth P &; Smith U emptying (Cunningham & Read, 1989). If so, the bene®cial (1999): Bedtime uncooked cornstarch supplement prevents nocturnal effect of cornstarch compared to white bread may be larger hypoglycaemia in intensively treated IDDM subject. J. Intern. Med. 245, 229 ± 236. than that demonstrated here. Beebe C, Van Cauter E, Shapiro E, Tillil H Lyons R, Rubenstein A & The factor(s) responsible for the overnight second-meal Polonsky K (1990): Effect of temporal distribution of calories on effect have not yet been established. Free fatty acids have a diurnal patterns of glucose levels and insulin secretion in NIDDM. number of effects, such as decreased peripheral glucose Diabetes Care 13, 748 ± 755. BjoÈrklund A, Yaney G, McGarry J & Weir G (1997): Fatty acids and beta- disposal (Boden & Chen, 1995; Groop et al, 1989), a cell function. Diabetologia 40, B21 ± B26. `lipotoxic' effect on pancreatic insulin release (BjoÈrklund Boden G & Chen X (1995): Effects of fat on glucose uptake and utilization et al, 1997; Paolisso et al, 1998) and an increased endo- in patients with non-insulin-dependent diabetes. J. Clin. Invest 96, genous glucose production (Golay et al, 1987; Saloranta et 1261 ± 1268. al, 1991), all of which are present in type 2 diabetic Bolli G (1988): The dawn phenomenon: Its origin and contribution to early morning hyperglycemia in diabetes mellitus. Diabete & Metabolism 14, patients. Wolever et al (1995) showed in healthy subjects 675 ± 686. that the second-meal effect is associated with reduced free Chen Y, Cornblath M & Sidbury J (1984): Cornstarch therapy in type 1 fatty acid levels before and after the second meal. In glycogen storage disease. N. Engl. J. Med. 310, 171 ± 175. addition, we found in type 2 diabetic patients, that the Collings P, Williams C & Macdonald I (1981): Effects of cooking on serum glucose and insulin responses to starch. Br. Med. J. 282, 1032. free fatty acid levels late at night (2 ± 7 am) were reduced Cunningham K & Read N (1989): The effect of incorporating fat into after increased intake at bedtime of `slow' carbohydrates different components of a meal on gastric emptying and postprandial ( 50 g) at 22.00 h, while this was not found after increased blood glucose and insulin responses. Br. J. Nutr. 61, 285 ± 190. intake of similar amounts of `rapid' carbohydrates (Axelsen Dimitriadis G, Vlachonikolis I, Hatziagellaki E, Linos A, Kordonouri O, et al, unpublished observations). Alexopoulos E & Raptis S (1988): The `dawn phenomenon' in patients with type II diabetes mellitus. Diabetes Nutr. Metab. 1, 37 ± 41. Bedtime carbohydrates in type 2 diabetes M Axelsen et al 710 Englyst H, Kingman S & Cummings J (1992): Classi®cation and measure- Perriello G, DeFeo P, Torlone E, Fanelli C, Santeusanio F, Brunetti P & ment of nutritionally important starch fractions. Eur. J. Clin. Nutr. 46, Bolli G (1988): Early night exaggerated hepatic glucose production, Suppl 2, S33 ± S50. later reduction of glucose clearance, and non-responsiveness of b-cell Ferrannini E, Simonson D, Katz L, Reichard G, Bevilacqua S, Barret E, all contribute to abnormal postabsorptive glucose homeostasis in sub- Olsson M & DeFronzo R (1988): The disposal of an oral glucose load in jects with impaired glucose tolerance. Second EASD Satellite Sympo- patients with non-insulin-dependent diabetes. Metabolism 47, 79 ± 85. sium on control of Metabolism, Lyon 2 ± 3 Sept., 1988. Firth R, Bell P, March H, Hansen I & Rizza R (1986): Postprandial Perriello G, Pimenta W, Pampanelli S, Porcellati F, Lepore M, Lucidi P, hyperglycemia in patients with noninsulin-dependent diabetes mellitus: Cordoni M, Brunetti P & Bolli G (1988): Evidence of day-night changes roles of hepatic and extra-hepatic tissues. J. Clin. Invest. 77, 1525 ± 1532. in insulin sensitivity in type 2 diabetes mellitus (abstract). Diabetologia Golay A, Swislocki A, Chen Y-D & Reaven G (1987): Relationships 41, Suppl 1, A194. between plasma-free fatty acid concentration, endogenous glucose Saloranta C, Franssila-Kallunki A, Ekstrand A, Taskinen M-R & Groop L production, and fasting hyperglycemia in normal and non-insulin- (1991): Modulation of hepatic glucose production by non-esteri®ed dependent diabetic individuals. Metabolism 36, 692 ± 696. fatty acids in type II (non-insulin-dependent) diabetes mellitus. Diabe- Groop L, Bonadonna R, Del Prato S, Ratheiser K, Zyck K, Ferrannini E & tologia 34, 409 ± 415. DeFronzo R (1989): Glucose and free fatty acid metabolism in non- Shaheen S & Fleming S (1987): High-®ber foods at breakfast: in¯uence on insulin dependent diabetes mellitus. Evidence for multiple sites of plasma glucose and insulin responses to lunch. Am. J. Clin. Nutr. 46, insulin resistance. J. Clin. Invest. 84, 205 ± 213. 804 ± 811. Hayde M & Widhalm K (1990): Effects of cornstarch treatment in very Shapiro E, Polonsky K, Copinschi G, Bosson D, Tillil H, Blackman J, young children with type I glycogen storage disease. Eur. J. Pediatr. Lewis G & VanCauter E (1991): Nocturnal elevation of glucose levels 149, 630 ± 633. during fasting in noninsulin-dependent diabetes. J. Clin. Endocr. Metab. Jenkins D, Wolever T, Jenkins A, Giordano C, Giudici S, Thompson L, 72, 444 ± 454. Kalmusky J, Josse R & Wong G (1986): Low glycemic response to Shimabukuro M, Zhou Y-T, Levi M & Unger R (1998): Fatty acid-induced traditionally processed wheat and rye products: bulgur and pumper- b cell apoptosis: A link between obesity and diabetes. Proc. Natl. Acad. nickel bread. Am. J. Clin. Nutr. 43, 516 ± 520. Sci. 95, 2498 ± 2502. Jenkins D, Wolever T, Taylor R, Grif®ths C, Krzeminska K, Lawrie J, Stilling B, Mehlsen J, Hamberg O, Larsen J, Gram N & Madsbad S (1998): Bennett C, Goff D, Sarson D & Bloom S (1982): Slow release carbohy- Effect of starch-free bread on metabolic control in type 2 diabetes drate improves second meal tolerance. Am. J. Clin. Nutr. 35, 1339 ± 1346. [letter]. Lancet 352, 369 ± 370. Matthews J, Altman D, Campbell M & Royston P (1990): Analysis of Wolever T, Bentum-Williams A & Jenkins D (1995): Physiological serial measurements in medical research. Br. Med. J. 300, 230 ± 235. modulation of plasma free fatty acid concentration by diet: implications McMahon M, Marsh H & Rizza R (1989): Effects of basal insulin in non-diabetic subjects. Diabetes Care 18, 962 ± 970. supplementation on the disposition of a mixed meal in obese patients Wolever T, Jenkins D, Jenkins A & Josse R (1991): The glycemic with non-insulin-dependent diabetes mellitus. Diabetes 38, 291 ± 303. index: methodology and clinical implications. Am. J. Clin. Nutr. 54, Mitrakou A, Kelley D, Veneman T, Jenssen T, Pangburn T, Reilly J & 846 ± 854. Gerich J (1990): Contribution of abnormal muscle and liver glucose Wolever T, Jenkins D, Ocana A, Rao V & Collier G (1988): Second- metabolism to postprandial hyperglycemia in NIDDM. Diabetes 39, meal effect: low-glycemic-index foods eaten at dinner improve 1381 ± 1390. subsequent breakfast glycemic response. Am. J. Clin. Nutr. 48, Paolisso G, Tagliamonte M, Rizzo M, Gualdiero P. Saccomanno F, 1041 ± 1047. Gambardella A, Giugliano D, Onofrio F & Howard B (1988): Lowering Wolfsdorf J & Crigler Jr J (1997): Cornstarch regimens for nocturnal fatty acids potentiates acute insulin response in ®rst degree relatives of treatment of young adults with type I glycogen storage disease. Am. J. people with type II diabetes. Diabetologia 41, 1127 ± 1132. Clin. Nutr. 65, 1507 ± 1511.