Dm

ORIGINAL ARTICLES

Effects of a-Glucosidase Inhibition and Viscous Fibre on Diabetic Control and Postprandial Gut Hormone Responses F. Requejo”, L. 0. Uttenthalb, S. R. BloomC “Servicio de Endocrinologia, Hospital Universitario San Carlos, and bDepartamento de Bioquimica, Facultad de Medicina, Universidad Complutense, Madrid, Spain, and ‘Department of Medicine, Royal Postgraduate Medical School, Hammersmith Hospital, London, UK

Twelve sulphonylurea-treated Type 2 diabetic patients underwent treatment for 2-week periods with the absorbable a-glucosidase inhibitor BAY m1099 (50 mg thrice daily) and with guar granules (5 g thrice daily) separately and together in a sequence-randomized double-blind placebo-controlled study. BAY mlO99 and guar reduced the mean fasting plasma glucose from 10.0 f 0.7 mmol I-’ to 8.7 f 0.5 ( p < 0.05) and 8.3 ? 0.7 mmol I-’ (p < 0.01), respectively. Both agents also lowered home-monitored postprandial blood glucose, with BAY m1099 exerting the greater effect. Guar, but not BAY m1099, lowered serum cholesterol from 5.43 -C 0.52 to 5.29 f 0.31 mmol I-’ ( p < 0.05). BAY m1099 reduced the test breakfast plasma responses of glucose (p < 0.001) and gastric inhibitory polypeptide (GIP, p < 0.01) and increased those of peptide tyrosine-tyrosine ( p < 0.05) and motilin (p < 0.01). Cuar also reduced plasma glucose concentrations after a test breakfast (p < 0.05) and increased the response of neurotensin ( p < 0.05). Combining treatments gave no further reduction of postprandial blood glucose concentration and was associated with an increased incidence and severity of gastrointestinal side-effects. KEY WORDS

a-Glucosidase inhibition Serum lipids Diet

Viscous fibre

Type 2 diabetes Gut hormones

Introduction

Patients and Methods

In recent years two methods of improving diabetic control by slowing intestinal carbohydrate absorption have been extensively studied: the use of added or natural viscous fibre’,2 and the use of inhibitors of intestinal carbohydrate d i g e ~ t i o n .However, ~ viscous fibre, whether added or native, is inconvenient to prepare or take in adequate amounts, while inhibitors of carbohydrate digestion are apt to cause flatulence and intestinal discomfort. Theoretically there might be advantages of greater benefit or reduced adverse effects when used as combined therapy. Such synergy has been demonstrated in an acute study on normal subject^.^ The purpose of the present study was to test moderate doses of added viscous fibre (guar granules) and an a-glucosidase inhibitor separately and together in Type 2 diabetic patients on sulphonylurea therapy, assessing effects on home-monitored blood glucose levels, fasting serum lipids, and test-meal responses of plasma glucose and gut hormones. As part . of the aim was to assess clinical applicability of the treatments, the study was conducted from the routine diabetic clinic on a heterogeneous sample of Type 2 diabetic patients.

Patients Type 2 diabetic patients in the age range 18-75 years were recruited, being currently on treatment with energyrestricted diet and sulphonylurea drugs and with fasting blood glucose concentrations between 7.0 and 15.0 mmol I-’ in the preceding 3 months. Women of childbearing age and patients with malignancy, severe cardiovascular, renal, hepatic or intestinal disease were excluded. Withdrawal criteria included development of fasting blood glucose concentrations exceeding 15.0 mmol I-’ on two consecutive follow-up visits, and the development of unacceptable side-effects. Twelve patients entered the study, seven men and five women, ten Caucasoid and two Negroid, age 59 (range 41-72) years and body mass index 30 ( 2 4 4 0 ) kg m-*. Two patients were being treated with oxyprenolol, one also taking bendrofluazide and the other hydralazine. A further two patients were treated with digoxin and hydroxycobalamin, respectively. Prior permission for the study had been granted by the Research Ethics Committee of the Royal Postgraduate Medical School and all patients gave their informed, written consent.

Protocol Correspondence to: Professor S. R. Bloom, Department of Medicine, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road, London, W12 ONN, UK.

0742-3071 /90/060515-06$05.00 & Sons, Ltd.

0 1990 by John Wiley

A randomized, double-bl ind, four- period, fou r-treatment, crossover design was used. The twelve patients entered Accepted 9 March 1990 DIABETIC MEDICINE, 1990; 7: 51 5-520

515

Dm

ORIGINAL ARTICLES sequentially a predetermined treatment sequence randomization scheme according to a 4 x 4 Latin square with three replications. Each period consisted of a 2week placebo tablet (Bayer UK, Newbury, UK) and placebo granules (round cellulose particles which did not change their form in the presence of food) run-in/ wash-out period, followed by a 2-week period on one of four treatments: 1. placebo tablets and granules; 2. tablets of BAY mlO99 (Miglitol: N-hydroxyethyl-ldeoxynojirimycin (Bayer UK)) 50 mg thrice daily plus placebo granules; 3. guar granules (Guarem, Rybar Laboratories, Amersham, UK) 5 g thrice daily plus placebo tablets; 4. tablets of BAY m1099 50 mg thrice daily plus guar granules 5 g thrice daily. The length of the treatment period was determined by the 4-week limit for the administration of BAY m1099 to human subjects imposed by the Committee for Safety of Medicines. Patients were instructed to take the tablets with the first bite of each main meal and to sprinkle the granules on the food as this method of administration has been shown to be far more effective than the conventional method of separate ingestion of g ~ a r The .~ usual sulphonylurea treatment and other medications were continued throughout. At the end of each treatment period, patients’ weight, fasting plasma glucose, serum fructosamine, serum total cholesterol and total triglycerides, routine haematology and serum biochemistry were determined. Symptoms were assessed by asking patients to score a standard randomized list as absent (score 0); mild (score 1); moderate (score 2); or severe (score 3). Patients were then given a test meal with the medication corresponding to the preceding 2-week period. The test meal was taken at 0900 h after an overnight fast and consisted of cornflakes, milk, toast, butter, marmalade, sugar, and tea with a total energy content of 2.2 MJ (519 kcal) distributed between protein 8 % (12 8); fat 27 % (16 g); sucrose 25 % (33 g) and other carbohydrates (40 % (54 g). It was consumed within 15 min and 10-ml blood samples were taken at times -10, 0, 20, 60, 120, and 240 min in relation to the start of the meal. On the penultimate day of each week of the study, patients recorded at home a seven-point blood glucose profile before and 1.5 h after each main meal and at bedtime, using glucose oxidase reagent strips (BM test Glycemie 1-44, BCL, Lewes, UK) and portable reflectance photometers (Reflolux 11, BCL).

Analytical Methods Plasma glucose concentrations were determined by a glucose oxidase method in a glucose analyser (Beckman I1 Analyser, Beckman instruments, Fullerton, CA, USA). Serum fructosamine was measured by the alkali nitroblue tetrazolium manual method.6 Total serum cholesterol 51 6

and serum triglycerides were measured by enzymatic methods (Technicon, Basingstoke, UK). For measurement of peptide hormones, blood samples of 10 ml were collected in heparinized tubes with 4000 kallikrein inhibitory units of aprotinin and immediately centrifuged. Plasma concentrations of insulin,’ gastric inhibitory polypeptide (GIP),8 C-terminal immunoreactive g l u ~ a g o n , ~ N-terminal ,l~ glucagon-like immunoreactivity,y,’o peptide tyrosine-tyrosine (PYY),’ neurotensin,12 and motilin’j were measured by radioimmunoassay. An index of gut-derived glucagon-like immunoreactivity (enterogl ucagons) was obtained by subtracting C-termi nal immunoreactive glucagon from N-terminal glucagon-like i mmunoreactivity.y,’ O

Data Analysis Values were expressed as mean t SE. Integrated plasma responses were calculated as incremental area under the 0-240 min curve, using the trapezoidal rule. Statistical analysis was performed by the Applied Statistical Research Unit, University of Kent, Canterbury, UK. The results were submitted to analysis of variance, allowing for the imbalance resulting from a single patient withdrawal, and allowing for patient, period, and treatment effects, using the statistical package GENSTAT (NAG, Oxford, UK). As no period or sequence effects were found the data were re-analysed by two-way analysis of variance. Additional paired comparisons of non-normally distributed data were performed by Wilcoxon’s test for pair differences.

Results Metabolic Control There was no significant change in body weight. Treatment with either BAY m1099 or guar alone resulted in a significant fall in fasting plasma glucose from 10.0 t 0.7 to 8.7 t 0.5 and 8.3 0.7 mmol I-’, respectively (Table 1). Combining BAY mlO99 and guar treatments did not lead to further improvements. Serum fructosamine results (normal range 2.1-2.7 mmol I-’) suggested possible but not significant reductions except for the combined treatments where there was clearly no improvement (Table 1). Guar either alone or in combination with Bay m1099 reduced fasting serum cholesterol concentrations from 5.43 0.52 to 0.31 and 5.26 0.29 mmol I-’, respectively. 5.29 Guar also had an effect on reducing serum triglycerides but this was not confirmed in combination with BAY mlO99.

*

*

*

Home-monitored Blood Glucose Concentrations Pre-breakfast glucose values were similar to those obtained in the clinic (Tables 1, 2), and all treatments F. REQUEJO, L. 0 . UTTENTHAL, S. R. BLOOM

DTT7

ORIGINAL ARTICLES

Table 1. Effect of 2 weeks treatment with BAY mlO99 (50 mg thrice daily) and guar (5 g thrice daily) separately and together on metabolic control in Type 2 diabetic patients ( n = 11) ~

__

~

Body weight (kg) Fasting plasma glucose (mmol I-') Serum fructosamine (mmol IF') Serum cholesterol (mmol I-') Serum triglycerides (mol I-')

__

Placebo

BAY mlO99

Cuar

85.6 ? 5.8 10.0 2 0.7 3.23 0.13 5.43 0.52 1.51 0.33

84.9 t 5.7 8.7 0.5" 2.99 0.26 5.68 t 0.31 1.34 2 0.23

85.1 t 5.7 8.3 i 0.7b 2.83 ? 0.23 5.29 0.31' 1.30 t 0.23'

* *

* * *

*

__

__

BAY m1099

+ Guar

*

85.9 5.9 8.9 2 0.5" 3.26 f 0.14 0.29' 5.26 1.47 2 0.36

*

Mean 2 SE. " p < 0.05; and bp < 0.01 compared with placebo by analysis of variance. ' p < 0.05 compared with placebo by Wilcoxon's test. Table 2. Effect of BAY m1099 (50mg thrice daily) and guar (5 g twice daily) separately and together on home-monitored blood glucose concentrations (mmol IF') in Type 2 diabetic patients ( n = 11) Placebo

BAY mlO99

Guar

BAY m1099

+ Guar

~

Before breakfast 1.5 h after breakfast Before lunch 1.5 h after lunch Before supper 1.5 h after supper Bedtime

9.0 2 0.4 11.6 0.8 8.6 t 0.5 9.8 0.7 8.2 0.8 10.3 t 0.8 8.9 0.6

* * * *

8.2 t 0.6" 8.9 0.6' 7.7 0.4" 8.6 0.8' 6.9 0.4" 9.1 ? 0.6" 9.1 t 0.8

* * * *

8.0 5 0.6b 9.3 0.7' 7.4 t O.bb 9.0 0.8 7.2 0.4" 9.7 t 0.8 9.0 1.0

* * * *

*

8.4 0.4 9.6 5 0.7' 7.4 ? 0.6b 8.7 0.7" 7.2 0.6" 9.4 0.8 8.6 2 0.6

* * *

Values are derived from the average of results obtained on days 6 and 13 of each treatment. Mean 2 SE. " p < 0.05; hp < 0.01; ' p < 0.001 compared with placebo by analysis of variance.

resulted in significantly lower plasma glucose concentrations preprandially compared with placebo (Table 2). Treatment with BAY mlO99 significantly lowered postprandial values 1.5 h after breakfast, lunch, and supper compared with placebo. Guar treatment showed a similar effect after breakfast but the reduction was not significant after lunch or supper (Table 2). There was no evidence of added benefit by combining treatments.

lest Meal Responses BAY mlO99 reduced the incremental glucose response to 68 2 9 % (p < 0.001) of placebo control value. Guar had a lesser effect (incremental response 84 ? 11 % of control; p < 0.05) and produced no further improvement when combined with BAY m1099 (Figure 1). No effects on plasma insulin, enteroglucagon, or glucagon were found for any treatment (data not shown). BAY m1099 flattened the GIP response to 64 2 11 % ( p < 0.01) of control, whereas guar had no effect. BAY m1099 alone or in combination with guar increased the plasma response of PYY to 221 ? 48 % and 200 2 35 % of control, respectively ( p < 0.05) whereas guar alone had no effect. Only guar increased the late release of neurotensin to 166 t 41 % of control ( p < 0.05). BAY mlO99 given alone both raised basal motilin concentrations and prolonged the positive phase 'of postprandial GLUCOSIDASE INHIBITION AND FIBRE IN DIABETIC PATIENTS

motilin response to 148 2 28 % of control ( p < 0.01). The addition of guar moderated this effect.

Side-effects and Tolerance No significant effects of treatment were noted on routine haematological and biochemical screening values. Subjective symptom questionnaires showed a total excess score over placebo for BAY mlO99, guar, and combined treatment of 46, 8, and 51, respectively, with excess score due to gastrointestinal symptoms of 25, 5, and 44, respectively (Table 3). The patient on BAY m1099 who developed severe diarrhoea did so after starting treatment with amoxycillin for an intercurrent infection and was withdrawn from the study before completing the BAY m1099 and combined treaments. N o other patient withdrew from the trial.

Discussion At the dosages used in this study, BAY m1099 had a greater effect o n postprandial blood glucose concentrations than guar, as recorded both after test meals and by patients' home monitoring, whereas guar had slightly more effect on fasting blood glucose concentrations. The small, but significant effect of a-glucosidase inhibition on fasting blood glucose concentrations might not be 517

DTT7

ORIGINAL ARTICLES

‘“1

plasma g l u c o s e

(mmol I”)

01

I

I

1

I

I

1

501

plasma P Y Y

01

(

I

I

I

I

Oi

I

I

I

l

-100

I

20

I

60

I

120 time ( m i n l

240

01

I

I

1

I

1

-100

T

T

(pmoi I-’)

I

20

I

I

I

60

I

120

1

I

240

time ( m i n l

Figure 1. Test meal responses (mean k SE, n = 11) of plasma glucose; gastric inhibitory polypeptide (GIP); peptide tyrosine-tyrosine (PYY); neurotensin and motilin, showing effects of placebo (0); BAY m1099 50 mg (A); guar granules 5 g (V);and BAY m1099 50 mg plus guar granules 5 g (+) in Type 2 diabetic patients after 2 weeks of treatment. See text for significance of differences between integrated responses

expected from the primary mechanism of action, were it not for the fact that such an effect has been described for dietary manipulations designed to attenuate the postprandial glucose rise and in previous studies on aglucosidase inhibition.l4~l5 The effect of guar on fasting serum lipids confirms previous findings that guar administration reliably reduces serum cholesterol level^.'^^" BAY mlO99 showed no significant effects on fasting serum lipids over the 2-week period of treatment. However, longer-term a-glucosidase inhibition with acarbose produced a transient fall in total serum ‘cholesterol

51 8

and a rise in the high density lipoprotein cholesterol ratio apparent at 8 to 20 weeks of treatment.18 Thus it cannot be precluded that longer-term use of BAY m1099 might produce similar, slight effects. The decreased release of postprandial GIP from the upper small intestine produced by BAY m1099 is comparable with that produced by acarbose18 and is presumed to be directly related to the decreased availability of transported and metabolizable monosaccharide to the GIP secreting mucosal endocrine cells.19 In contrast with acarbose,18 BAY mlO99 treatment F. REQUEJO, L. 0. UTTENTHAL, S.

R. BLOOM

DTT7

ORIGINAL ARTICLES

Table 3. Subjective excess score of symptoms during treatments with BAY m1099 and guar, separately and together, compared with placebo for Type 2 diabetic patients BAY m 1 099

n Total excess Gastrointestinal symptoms abdominal pain distension meteorism flatulence diarrhoea other

Guar

12 46 25

BAY m1099

+

Guar

12 8

5

4 (n=4)

1 1 1 1 1

6 (n=4) 8 (n=5)" 7

(n=l) (n=l)

(n=l) (n=l) (n=l)

-

"One patient with severe symptoms. "Two patients with severe symptoms

did not produce a significant increase in postprandial enteroglucagon release. O n the other hand, like acarbose, it significantly stimulated the postprandial release of PYY. This is difficult to explain because PYY is co-localized with N-terminal glucagon-like immunoreactivity in the L-cells of the intestinal mucosa, even being localized in the same secretory granules.*O The dissociation of PYY and enteroglucagon responses in the present group of patients may indicate the existence of a separate population of specific PYY cells, so placed or constituted as to be more susceptible to the action of BAY m1099. This possibility could be elucidated by further immunocytochemical studies on human biopsy material. The effect of BAY m1099 to raise basal plasma motilin levels and to prolong the positive phase of the postprandial motilin response i s similar to that of acarbose,18 but seems to be rather more potent in that basal levels are affected within 2 weeks. The fact that oral and intravenous glucose administration inhibits motilin release2' suggests that the effect may in part be due to impaired glucose generation and absorption in the upper small intestine. However, the rise in basal motilin levels with acarbose treatment shows a considerable time-lag in relation to the improvement in basal blood glucose, so that various chronic mechanisms, including a direct effect of the agent, must be considered. The conditions of this study, with the aim of measuring mean effects of standard drug doses, precluded titration of a-glucosidase inhibitor dose in accordance with the desired and unwanted effects in individual patients. As individuals vary greatly with respect to a-glucosidase activities in the intestinal brush border,22 such a titration would be necessary for optimal treatment, and a considerable incidence of side-effects may reasonably be expected if this is omitted. One advantage that might be expected from an aglucosidase inhibitor that is rapidly absorbed, as compared with one that is not, is a lower incidence of intestinal side-effects. However, comparison of the sideeffects of BAY mlO99 with those of acar'bose studied on CLUCOSIDASE INHIBITION A N D FIBRE IN DIABETIC PATIENTS

a similar group of patients18 fails to show any obvious advantage, although the doses of BAY m1099 and acarbose used were approximately equipotent with respect to their effect on postprandial blood glucose. A tendency for side-effects to decline with duration of treatment has been noted with acarbose. Combined treatment with BAY m1099 and guar increased the incidence and severity of gastrointestinal side-effects above those produced by BAY mlO99 alone. The cholesterol-lowering effect of guar was retained, but the combination did not otherwise offer any advantage over treatment with BAY m1099 alone. We conclude that intestinal a-glucosidase inhibition i s an effective means of attenuating postprandial blood glucose rises, at the expense of some gastrointestinal side-effects. It seems to offer a useful addition to the available repertory of diabetic therapy. Guar i s less effective in reducing postprandial blood glucose and its administration is less convenient, but this is balanced by its cholesterol-lowering effect and a lower incidence of side-effects. Because of the exacerbation of gastrointestinal side-effects there seems to be no case for combined use of these agents, at least not with the fixed doseschedules of this study.

Acknowledgements This study was generously supported by Bayer UK, who also provided BAY m1099 and placebo tablets. We thank P. Leigh for his organizational help. We are very grateful to BCL, Lewes, UK for their gift of Reflolux meters. We acknowledge the kind help of M. A. Ghatei and colleagues in the measurement of plasma hormone concentrations.

References 1.

Jenkins DJA, Goff DV, Leeds AR, et a/. Unabsorbable carbohydrate and diabetes: decreased postprandial hyper-

519

DTT7

ORIGINAL ARTICLES 2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

520

glycaemia. Lancet 1976; ii: 172-1 74. Simpson HCR, Lousley S, Geekie M, Carter RD, Hockaday TDR, Mann JI.A high carbohydrate leguminous fibre diet improves all aspects of diabetic control. Lancet 1981; i : 1-5. Puls W, Keup V, Krause HP, Thomas G, Hoffmeister F. Glucosidase inhibition. A new approach to the treatment of diabetes, obesity and hyperlipoproteinaemia. Naturwissenschaften 1977; 64: 536-537. Jenkins DJA, Taylor RH, Nineham R, et a/. Combined use of guar and acarbose in reduction of postprandial glycaemia. Lancet 1979; ii: 924-927. Fuessl HS, Adrian TE, Bacarese-Hamilton AJ, Bloom SR. Guar in NIDD: effect of different modes of administration on plasma glucose and insulin responses to a starch meal. Pract Diabetes 1986; 3: 258-260. Johnson RN, Metcalf PA, Baker JR. Fructosamine: a new approach to the estimation of serum glycosylprotein. An index of diabetic control. Clin Chim Acta 1982; 127: 87-95. Albano JDM, Ekins RP, Martitz G, Turner RC. A sensitive, precise radioimmunoassay of serum insulin relying on charcoal separation of bound and free hormone moieties. Acta Endocrinol 1972; 70: 487-509. Sarson DL, Bryant MG, Bloom SR. A radioimmunoassay for gastric inhibitory polypeptide in human plasma. ) Endocrinol 1980; 85: 487-495. Ghatei MA, Uttenthal LO, Bryant MG, Christofides ND, Moody AJ, Bloom SR. Molecular forms of glucagon-like immunoreactivity in porcine intestine and pancreas. Endocrinology 1983; 112: 917-923. Ghatei MA, Uttenthal LO, Christofides ND, Bryant MG, Bloom SR. Molecular forms of human enteroglucagon in tissue and plasma: plasma response to nutrient stimuli in health and in disorders of the upper gastrointestinal tract. / Clin Endocrinol Metab 1983; 57: 488-495. Adrian TE, Ferri G-L, Bacarese-Hamilton AJ, Fuessl HS, Polak JM, Bloom SR. Human distribution and release of a putative new gut hormone peptide YY. Gastroenterology 1985; 89: 1070-1 077. Lee YC, Allen JM, Uttenthal LO, et a/. The metabolism of intravenously infused neurotensin in man and its chromatographic characterisation in human plasma. Clin

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

Endocrinol Metab 1984; 59: 45-50. Christofides ND, Bryant MG, Ghatei MA, eta/. Molecular forms of motilin in the mammalian and human gut and human plasma. Gastroenterology 1981 ; 80: 292-300. Sailer D, Roder G. Treatment of non-insulin dependent diabetic adults with a new glycoside hydrolase inhibitor (BAY g 5421). Drug Res 1980; 30: 2182-2185. Willms B, Sachse G, Unger H . Short-term effects of acarbose in sulphonylurea-treated diabetics. In: Creutrfeldt W, ed. Proceedings, First International Symposium on Acarbose. International Congress Series 594. Amsterdam: Excerpta Medica, 1982: 261-265. Jenkins DJA, Leeds AR, Slavin B, Mann J, Jepson EM. Dietary fibres and blood lipids: reduction of serum cholesterol in Type I I hyperlipidemia by guar gum. Am ) Clin Nutr 1979; 32: 16-18. Jenkins DJA, Wolever TMS, Taylor RH, Reynolds D, Nineham R, Hockaday TDR. Diabetic glucose control, lipids, and trace elements on long-term guar. Br Med / 1980; 280: 1353-1 354. Uttenthal LO, Ukponmwan 00, Wood SM, et a/. Longterm effects of intestinal alpha-glucosidase inhibition on postprandial glucose, pancreatic and gut hormone responses and fasting serum lipids in diabetics on sulphonylureas. Diabetic M e d 1986; 3: 155-1 60. Sykes S, Morgan LM, English J, Marks V. Evidence for preferential stimulation of gastric inhibitory polypeptide secretion in the rat by actively transported carbohydrates and their analogues. / Endocrinol 1980; 85: 201-207. Ali-Rachedi A, Varndell IM, Adrian TE, et a/. Peptide YY (PYY) immunoreactivity i s co-stored with glucagon-related immunoreactants in endocrine cells of the gut and pancreas. Histochemistry 1984; 80: 487-491. Christofides ND, Bloom SR, Besterman HS, Adrian TE, Ghatei MA. Release of motilin by oral and intravenous nutrients in man. Gut 1979; 20: 102-106. Puls W, Keup U, Krause HP, Thomas G, Hoffmeister F. The concept of glucosidase inhibition and its pharmacological realisation. In: Creutzfeldt W, ed. Proceedings, First lnternational Symposium on Acarbose. International Congress Series 594. Amsterdam: Excerpta Medica, 1982: 16-26.

F. REQUEJO, L. 0. UTTENTHAL, S. R. BLOOM