R

ORIGINAL ARTICLE | OA

Effect of acarbose on additional insulin therapy in type 2

diabetic patients with late failure of sulphonylurea therapy

E. Standl, H. J. Baumgartl, M. FuÈchtenbusch and J. Stemplinger

Institute of Diabetes Research and Department of Endocrinology, Academic Hospital Schwabing, Munich, Germany

Aim: The present study investigated the effect of acarbose on insulin requirements and glycaemic
control in patients with type 2 diabetes receiving exogenous insulin due to secondary failure of
maximum dose sulphonylurea therapy.
Methods: A single-centre, double-blind, randomized, placebo-controlled study was performed in
48 type 2 diabetic patients with late-term failure following at least 3 years of sulphonylurea
therapy requiring additional insulin therapy to determine the impact of acarbose on glycaemic
control and insulin requirements. The primary end points were glycaemic response rate
(responders being prede®ned as patients who achieve a decrease in HbA1c to less than 8% or a
reduction by at least 15% as compared to the baseline values) and the daily insulin dose at
6 months. Secondary parameters assessed included postprandial changes in blood glucose, serum
insulin and C-peptide during the treatment period.
Results: There were signi®cantly more responders in the acarbose-treated group compared
with the placebo group (20/24 patients vs. 10/19 patients; p < 0.05). The mean daily insulin dose
after 24 weeks of treatment was 16.4 6 10.1 IU in the acarbose group and 22.4 6 12.2 IU in the
placebo group (mean 6 s.d.; p < 0.07). Postprandial increases in blood glucose, insulin and
C-peptide were consistently lower in the acarbose-treated group than in the placebo group. For
example, the mean increase in 2-h postprandial serum insulin remained almost unchanged in the
acarbose group at the end of 24 weeks of treatment compared to an increase to 43 6 29 mU/ml (mean
6 s.d.) at the end of the study period for the placebo group.
Conclusions: The ®ndings of this study suggest that the addition of acarbose to sulphonylurea/
insulin combination therapy can improve glycaemic control in type 2 diabetic patients. Acarbose
may also reduce insulin resistance and hyperinsulinaemia.
Keywords: acarbose, type 2 diabetes, insulin, sulphonylurea, therapy
Received 13 October 1998; returned for revision 9 November 1998; revised version accepted 18 March 1999

Acarbose is a pseudotetrasaccharide of microbial

Introduction
The use of combined insulin and sulphonylurea therapy
is a therapeutic option in patients with type 2 diabetes
who experience late-term failure of sulphonylurea
monotherapy. The aim of this combination is to use a
sulphonylurea to maximize utilization of the remaining
insulin secretion, while using exogenous insulin to
compensate for the rather small insulin de®ciency
which is still present despite sulphonylurea therapy [1].
origin with a similiar structure to carbohydrates. It
binds competitively to the a-glucosidases at the surface
of the brush border of the small intestine. Acarbose thus
delays carbohydrate digestion, since a-glucosidases are
required for the hydrolysis of carbohydrates [2].
The metabolic effects of acarbose include attenuation of
the postprandial increase in blood glucose and insulin
levels, decreased hyperinsulinaemia and increased
insulin sensitivity. Many clinical studies involving

Correspondence:
E. Standl, Institute of Diabetes Research, Academic Hospital Schwabing, Koelner Platz 1, Munich, Germany.

ã 1999 Blackwell Science Ltd Diabetes, Obesity and Metabolism, 1, 1999, 215±220 | 215
L
OA | Effect of acarbose on insulin therapy E. Standl et al.

patients with type 2 diabetes have shown the bene®cial Table 1. Baseline data of the study population (means 6 s.d.)
effects of acarbose, used alone and in combination with
sulphonylurea therapy, on glycaemic control as indicated Acarbose Placebo
by decreased HbA1c [3±8]. Age (years) 59.3 6 8.5 62.9 6 9.4
The present study investigated the effect of acarbose on Body weight (kg) 70.5 6 11.7 66.7 6 10.8
insulin requirements and glycaemic control in patients Height (cm) 167.0 6 10.7 165.9 6 9.1
with type 2 diabetes receiving exogenous insulin due to BMI 25.2 6 2.2 24.1 6 2.0
Sex (ratio m:f) 14:10 10:14
secondary failure of sulphonylurea therapy.
Duration of diabetes (months) 137.7 6 81.7 146.7 6 68.0
median 110 120
Sulphonylurea for (months) 103.3 6 66.8 112.2 6 64.7
Patients and Methods
median 72 104
This was a 24-week, phase IV, single-centre, double-blind, Diabetic complications (n)
Neuropathy 5 5
randomized, placebo-controlled, group-comparison
Retinopathy 8 5
study involving outpatients with type 2 diabetes receiving Nephropathy 1 0
combined sulphonylurea/insulin therapy following HbA1c (%) 10.9 6 1.0 11.0 6 1.2
secondary failure of sulphonylurea therapy (of at least
3 years duration).
Before inclusion in the study, patients were hospita-
lized for 5 days, maintained on their previous maximum ization. The acarbose dose was gradually increased in a
dose of the sulphonylurea glibenclamide (2±0±1 3.5 mg stepwise fashion (from 50 mg b.i.d. on days 1±3, up to
tablets, representing a special high effectiveness 100 mg b.i.d. on days 4±6, up to 100 mg t.i.d. during
formulation in Germany), and assessed for true second- weeks 2±4, and up to a maximum of 200 mg t.i.d.
ary failure of sulphonylurea vs. compliance failure. For starting from week 5 to the end of the study). The same
this purpose, a 1000 kcal/day antidiabetic diet was procedure was used for placebo in a double-blind
strictly reinforced during this period. Inclusion criteria manner. If intolerable adverse events related to acarbose
for patients entering the study were: a fasting capillary (or placebo) occurred, the dose could be reduced to the
blood glucose level > 8.9 mmol/l and a 1-h postprandial next lowest dose within the dosage regimen.
capillary blood glucose level > 13.9 mmol/l on day 5 in The prede®ned primary end points of the study were
the context of a HbA1c level > 8.5%. Additional inclu- the daily insulin dose after 6 months of treatment and
sion criteria were: manifestation of type 2 diabetes after glycaemic response rate. Patients were classi®ed as
the age of 40 years, duration of diabetes since diagnosis responders if their HbA1c value had fallen to below 8%
of at least 5 years and body weight Broca index < 115%. or was at least 15% below their baseline value at the 6-
Patients also had to have received sulphonylurea month follow-up visit, according to local guidelines [9].
therapy for at least 3 years. All patients had normal Patients were asked to attend the outpatient depart-
plasma creatinine levels and liver function test results. ment of the clinic in a fasting state for individual
Patients receiving therapy with b-blockers or thiazide examinations 4, 8, 12, 16, 20 and 24 weeks after
diuretics were excluded. Table 1 shows the baseline initiation of treatment. Venous blood samples were
characteristics of the study groups at screening. taken in a fasting state and 1 and 2 h after the intake of a
All patients were then started on insulin treatment in standard breakfast (consisting of one roll, a portion of
addition to the maximum dose sulphonylurea therapy. jam containing 6 g of carbohydrate, 50 g of cottage cheese
Furthermore, patients were randomized to receive acar- and a caffeine-free beverage). Fasting HbA1c, glucose, C-
bose or placebo treatment together with the start of insulin peptide, insulin, cholesterol, HDL cholesterol and
therapy which was initiated with 6 IU of a standard triglycerides were determined in addition to 1-and 2-h
mixture of 25±30% of regular and NPH insulin and postprandial glucose, C-peptide and insulin. Measure-
adjusted if necessary. The titration of the insulin dose ments of blood pressure and body weight were
was based on 1-h postprandial blood glucose levels. If 1-h performed before starting the study and during all
postprandial blood glucose values were > 12.2 mmol/l, the examinations.
insulin dose was increased by 2±4 IU at 4-week intervals. If Capillary whole blood and venous plasma glucose
1-h postprandial blood glucose values were < 7.8 mmol/l, levels were measured using the hexokinase method [10].
the insulin dose was reduced by 2±4 IU. HbA1c levels were measured using high pressure liquid
From day 1 of the insulin therapy, patients also chromatography (Bio-Rad Laboratories, Munich, Ger-
received acarbose or placebo according to their random- many) and serum insulin and C-peptide levels were

216 | Diabetes, Obesity and Metabolism, 1, 1999, 215±220 ã 1999 Blackwell Science Ltd
 R
E. Standl et al. Effect of acarbose on insulin therapy | OA

evaluated by radioimmunoassay [11]. Samples were

analysed for lipid status using kits and reagents
supplied by Immuno Corp. (Heidelberg, Germany) [12].
Patient compliance was monitored carefully through-
out the treatment period. Patients were asked to return
unused tablets to the investigator at each visit and the
number of tablets was counted to assess compliance. At
least 80±100% of each patient's prescribed dose had to
have been taken. All adverse events or unusual events
were documented, with a detailed description of signs,
symptoms, time of onset and duration. In addition, the
documentation had to state the action taken (including Fig. 1 Responders (HbA1c fallen below 8% or reduction by at
least 15% as compared to baseline values) and non-responders
discontinuation), the outcome and possible causes.
after 24 weeks of treatment in the acarbose group compared to
The study was implemented in accordance with the the placebo group.
rules and regulations of the German Drug Law, the
German Guidelines for Good Clinical Practice, and the
European Guidelines for Good Clinical Practice. The
study was also conducted in accordance with the
Principles of the Helsinki Declaration. All patients
participating in the study gave written informed consent
and approval was given by the local ethical committee.

Statistical Analysis

Results are given as mean 6 s.d. (s.e.m.) or median

values, as appropriate. Comparisons were made using
the Wilcoxon rank test or c2 test.

Results

A total of 48 patients was included in the study (24 in

each group). Five patients (all in the placebo group)
were regarded as invalid for inclusion in the ef®cacy
analysis. The reasons for exclusion from the analysis
included poor compliance (one patient dropped out
after 6 days); clear evidence of absolute insulin depen-
dence with increasing blood glucose values
> 22.2 mmol/l despite insulin combination therapy
(two patients with probably late onset type 1 diabetes
and need for insulin monotherapy with four injections/
day after days 4 and 8, respectively); unforeseen need
for surgery (one patient at day 12); and loss to follow up
(one patient at week 4).
There were signi®cantly more responders in the
acarbose-treated group compared with the placebo-
treated group (20/24 patients vs. 10/19 patients,
p < 0.05; Fig. 1). The mean drop in HbA1c level, however,
was comparable in both groups at the end of the study
(Fig. 2) and amounted to a total of » 2.4% HbA1c in
relation to the time of ®rst screening. On the other hand,
at 24 weeks, the mean daily insulin dose was 16.4 6 10.1
IU (mean 6 s.d.) in the group treated with acarbose and
Fig. 2 HbA1c pro®les (means 6 s.d.; valid patients n = 43).
22.4 6 12.2 IU in the placebo group. This compares with
baseline values of 6.4 6 1.9 IU for the acarbose group
and 5.9 6 2.0 in the placebo group and corresponds to a
mean difference between the 24-week value and
pretreatment value of 9.9 IU (acarbose) or 16.5 IU
(placebo). The difference between the treatment groups
fell slightly short of the level of statistical signi®cance
(p = 0.07; Fig. 3).
The postprandial increases in blood glucose, insulin
and C-peptide were consistently lower in the acarbose-
treated group. When analysing the increase in blood
glucose levels after intake of the standard breakfast, the
differences in the 1-h postprandial values between the
treatment groups are distinct after just 4 weeks (Fig. 4).
When comparing the postprandial increase in serum
insulin, greater differences between the two treatment
groups were evident in the 2-h postprandial values than
in the 1-h postprandial values. The mean increase in

ã 1999 Blackwell Science Ltd Diabetes, Obesity and Metabolism, 1, 1999, 215±220 | 217
L
OA | Effect of acarbose on insulin therapy
Fig. 3 Total daily dose of therapeutic insulin (IU) in the
acarbose vs. placebo group at start vs. end of the study period
(means 6 s.e.m.).
Fig. 4 Increase in blood glucose (1-h postprandial±fasting)
(means 6 SEM). Two-h postprandial increases shown by circles
(without s.e.m.).
serum insulin 1-h postprandial at baseline was
21 6 18 mU/ml (mean 6 s.d.) in the acarbose group
compared to 24 6 17 mU/ml in the placebo group
(Fig. 5). There was a further slight increase to
27 6 16 mU/ml in the acarbose group and a greater
increase to 36 6 20 mU/ml in the placebo-treated group
after 24 weeks of treatment. The mean increase in 2-h
postprandial serum insulin levels remained almost
unchanged in the acarbose group after 24 weeks, but
increased to 43 6 29 mU/ml at the end of the treatment
period in the placebo group.
Values for C-peptide decreased during the course of
the study in the acarbose group from 0.40 6 0.4 pmol/ml
to 0.27 6 0.2 pmol/ml but rose slightly in the placebo
group from 0.42 6 0.3 pmol/ml to 0.46 6 0.3 pmol/ml.
There was thus a marked difference between the two
treatment groups (0.19 pmol/ml, Fig. 6). In general, the
218 | Diabetes, Obesity and Metabolism, 1, 1999, 215±220
E. Standl et al.
Fig. 5 Increase in serum insulin levels (1-h postprandial±
fasting) in the acarbose vs. the placebo group during 24 weeks
of treatment (means 6 s.e.m.).
Fig. 6 Increase in C-peptide levels (1-h postprandial±fasting) in
the acarbose vs. the placebo group during 24 weeks of treatment
(means 6 s.e.m.).
mean 2-h postprandial C-peptide increases did not
reveal any additional effects.
There was a slight decrease in triglyceride levels in
the acarbose group between day 0 and week 24 (median
decrease 0.18 mmol/l) and a slight increase in the
placebo group (median increase 0.59 mmol/l), but the
differences were not statistically signi®cant (the median
was used instead of the mean because no normal
distribution is given).
Regarding the results obtained for total cholesterol
and HDL cholesterol, there was an increase in total
cholesterol by 0.28 mmol/l in the acarbose group and by
ã 1999 Blackwell Science Ltd

E. Standl et al.
0.05 mmol/l in the placebo group. HDL cholesterol
increased by 0.11 mmol/l (acarbose) and by 0.08 mmol/
l (placebo). There was a trend towards an increase in
body weight (b.w.) during the course of the study in both
treatment groups: acarbose, + 1.7 kg; placebo, + 1.5 kg.
A total of 22 patients suffered from adverse events.
Fifteen of these patients were in the acarbose group and
seven in the placebo group. Most of the adverse events
were gastrointestinal in nature. No patients withdrew
from the study due to adverse events.
Discussion
In this study, the addition of acarbose to a newly
implemented combined insulin/sulphonylurea therapy
in patients failing on maximum dose sulphonylurea
monotherapy was found to improve glycaemic response,
as re¯ected in the percentage of patients who achieved
HbA1c values below 8% or a reduction of at least 15% as
compared to the baseline during the course of 24 weeks
of treatment with acarbose. Only 52% of patients
receiving insulin/sulphonylurea therapy achieved this
treatment target, compared to 83% of patients receiving
acarbose in addition to insulin/sulphonylurea therapy.
This clinical observation is in accordance with a
number of previous studies and underlines the conclu-
sion that acarbose can improve glycaemic control in
patients with type 2 diabetes receiving long-term
treatment with combined insulin/sulphonylurea ther-
apy [13±15]. In addition, acarbose was found to reduce
the requirement for insulin. In a recent study by
Chiasson et al. [16] the insulin dose was decreased by
15% or more in 36% of patients receiving insulin plus
acarbose but in only 6% of patients receiving insulin
plus placebo [4]. In our investigation, the mean daily
insulin dose after 24 weeks of treatment was 16 6 10.1
IU (mean 6 s.d.) in the acarbose group and 22 6 12.2 IU
in the placebo group.
The descriptive analysis of the secondary end points
of this study reveals a number of interesting ®ndings
which could encourage further clinical research, parti-
cularly since triple combination therapy with insulin,
glibenclamide and acarbose in patients with type 2
diabetes has not been investigated thoroughly to date.
Postprandial blood glucose, insulin and C-peptide
values were, in general, lower in the acarbose-treated
group during the study period. The observation that
postprandial plasma glucose concentrations are signi®-
cantly affected by acarbose has been seen in various
studies in which acarbose has been added to sulphony-
lurea and in patients treated with insulin [4,5].
Although the reduction in plasma glucose levels
ã 1999 Blackwell Science Ltd
R
Effect of acarbose on insulin therapy | OA
induced by acarbose in healthy persons has always
been associated with decreased plasma insulin levels,
con¯icting results have been reported in patients with
type 2 diabetes since sulphonylurea stimulates insulin
secretion and insulin therapy suppresses it [6,17]. We
hypothesize that the changes in postprandial blood
glucose and serum insulin pro®les observed following
additional therapy with insulin can only be explained
conclusively by progressive insulin resistance possibly
in connection with hyperinsulinaemia. The additional
use of acarbose seems to stop or delay this process. In
these patients, a reduction in insulin resistance, as
achieved by acarbose-induced inhibition of a-glucosi-
dase, might be an appropriate aim of therapy
[13,14,16,18]. In addition, in this study, acarbose did
not affect the serum lipid pro®le, although this has not
been observed in other studies [19,20].
In summary, the present study underlines a large
number of positive effects of a-glucosidase inhibition in
a patient group treated with triple therapy following late
sulphonylurea failure. This is of considerable practical
interest since acarbose is very well tolerated, even in
patients with multiple morbidities, and is not associated
with severe side-effects such as liver toxicity or lactic
acidosis, as seen with troglitazone or metformin [21±23].
References
1 Bachmann W., Sieger C., Haslbeck M., Lotz N. Combina-
tion of insulin and glibenclamide in the treatment of
adult-onset diabetes (type 2). Diabetologia 1981; 21: 245±
250.
2 Hanefeld M. Acarbose monotherapy in non-insulin-
dependent diabetes mellitus: the German experience.
In: Lefebvre P.J., Standl E., eds. New Aspects in
Diabetes. Berlin, New York: de Gruyter, 1992: 147± 161.
3 Clissold S.P., Edwards C. Acarbose: a preliminary
review of its pharmacodynamic and pharmacokinetic
properties and therapeutic potential. Drugs 1988; 35:
214±243.
4 Chiasson J.-L., Josse R.G., Hunt J.A., Palmason C., Rodger
N.W., Ross S.A. The ef®cacy of acarbose in the treatment
of patients with non-insulin-dependent diabetes melli-
tus. Ann Intern Med 1994; 928±935.
5 Sachse G., Willms B. Effect of the alpha-glucosidase-
inhibitor BAY-g-5421 on blood glucose control in
sulphonylurea-treated diabetics and insulin-treated dia-
betics. Diabetologia 1979; 17: 287±290.
6 Reaven G.M., Lardinois C.K., Green®eld M.S., Schwartz
H.C., Vreman H.J. Effect of acarbose on carbohydrate and
lipid metabolism in NIDDM patients poorly controlled
by sulphonylureas. Diabetes Care 1990; 13 (Suppl.):
32±37.
7 May C. Wirksamkeit und VertraÈglichkeit von einschlei-
Diabetes, Obesity and Metabolism, 1, 1999, 215±220 | 219
L
OA | Effect of acarbose on insulin therapy
chend dosierter Acarbose bei Patienten mit nichtinsu-
linp¯ichtigem Diabetes mellitus (Typ-II-Diabetes) unter
Sulfonylharnstofftherapie. Diabetes Stoffwechsel 1995;
4: 3±8.
8 Coniff R.F., Shapiro J.A., Seaton T.B., Bray G.A. Multi-
center, placebo-controlled trial comparing acarbose
(BAY b5421) with placebo, tolbutamide and tolbuta-
mide-plus-acarbose in non-insulin-dependent diabetes
mellitus. Am J Med 1995; 98: 443±451.
9 Standl E., Usadel K.H., Mehnert H. Grundlagen des
Diabetesmanagements. In: Mehnert H., Standl E., Usadel
K.H., eds. Diabetologie in Klinik und Praxis. Stuttgart:
Thieme, 1999: 103± 119.
10 Bergmeyer H.U., Bernt E., Schmidt F., Stock H. D-
glucose: determination with hexokinase and glucose-6-
phosphatase dehydrogenase. In: Bergmeyer H.U.,
Gawehn K., eds. Methods of Enzymatic Analysis. New
York: Academic Press, 1974: 1196± 1201.
11 Kuzuya T., Saito T., Yoshida S., Matsuda A. Human C-
peptide immunoreactivity (CPR) in blood and urine:
evaluation of a radioimmunoassay method and its
clinical application. Diabetologia 1976; 12: 511±518.
12 Wieland H., Seidel D. Improved techniques for assess-
ment of plasma lipoprotein patterns. Clin Chem 1973;
19: 1139±1141.
13 Goto Y., Toyota T., Oikawa S., Maruhama Y., Tamura
M., Komatsu K. Long-term treatment of BAY g 5421
(acarbose) in non-insulin-dependent diabetes mellitus.
In: Lefebvre P.J.Standl E., eds. New Aspects in Diabetes.
Berlin, New York: de Gruyter, 1992: 251± 257.
14 Foster D.W. Insulin resistance, a secret killer? New Engl
J Med 1989; 320: 733±734.
15 Domke A., Willms B. Wirksamkeit und VertraÈglichkeit
einer laÈngerfristigen Acarbosetherapie bei Diabetikern
220 | Diabetes, Obesity and Metabolism, 1, 1999, 215±220
E. Standl et al.
mit drohendem SekundaÈrversagen einer Sulphonylharn-
stoffbehandlung. Med Klinik 1994; 89: 187±192.
16 Chiasson J.-L., Josse R.G., Leiter L.A., Mihic M., Nathan
D.M., Palmonson C. The effect of acarbose on insulin
sensitivity in subjects with impaired glucose tolerance.
Diabetes Care 1996; 19: 1190±1193.
17 Gerard J., Jandrain B., Pirnay F., Pallikarakis N.,
Krzentowski G., Lacroix M. Utilization of oral sucrose
load during exercise in humans: effect of the alpha-
glucosidase inhibitor acarbose. Diabetes 1986; 35: 1294±
1301.
18 Laube H., Linn Th., Heyen P. The effect of acarbose on
insulin sensitivity and proinsulin in overweight subjects
with impaired glucose tolerance. Exp Clin Endocrinol
Diabetes 1998; 106: 231±233.
19 Akazawa Y., Koide M., Oishi M., Azuma T., Tashiro S.
Clinical usefulness of acarbose and ®ber in the treatment
of diabetes mellitus. Therapeutics 1982; 36: 848±849.
20 Noyon R., Pagano Mirani-Oostdijk C., v. Gent C.M.,
Froblicle M., Hillebrand J., Terpstra J. Long-term effect of
acarbose on diurnal serum triglyceride, glucose, insulin
and adipose tissue lipoprotein lipase levels in patients
with primary endogenous hypertriglyceridaemia, with
or without type II diabetes. Neth J Med 1986; 29: 157±
164.
21 Watkins P.B., Whitcomb R.W. Hepatic dysfunction
associated with troglitazone (letter). New Engl J Med
1998; 338: 916±917.
22 Kuramoto K., Shimizu N., Toda G. Liver dysfunction
associated with troglitazone. J Clin Ther Med 1998; 14:
461±466.
23 Bailey C.J., Turner R.C. Metformin. New Engl J Med
1996; 334: 574±579.
ã 1999 Blackwell Science Ltd