Articles Acarbose compared with metformin as initial therapy in
patients with newly diagnosed type 2 diabetes:
an open-label, non-inferiority randomised trial
Wenying Yang, Jie Liu, Zhongyan Shan, Haoming Tian, Zhiguang Zhou, Qiuhe Ji, Jianping Weng, Weiping Jia, Juming Lu, Jing Liu, Yuan Xu,
Zhaojun Yang, Wei Chen
Summary
Background Metformin is the only ?rst-line oral hypoglycaemic drug for type 2 diabetes recommended by
international guidelines with proven e?cacy, safety, and cost-e?ectiveness. However, little information exists
about its use in Asian populations. We aimed to ascertain the e? ectiveness of the α-glucosidase inhibitor acarbose,
extensively adopted in China, compared with metformin as the alternative initial therapy for newly diagnosed
type 2 diabetes.
Methods In this 48-week, randomised, open-label, non-inferiority trial, patients who were newly diagnosed with
type 2 diabetes, with a mean HbA
1c
of 7·5%, were enrolled from 11 sites in China. After a 4-week lifestyle modi? cation
run-in, patients were assigned to 24 weeks of monotherapy with metformin or acarbose as the initial treatment,
followed by a 24-week therapy phase during which add-on therapy was used if prespeci? ed glucose targets were not
achieved. Primary endpoints were to establish whether acarbose was non-inferior to metformin in HbA
1c
reduction at
week 24 and week 48 timepoints.The non-inferiority margin was 0·3%, with an expected null di? erence in the
change from baseline to week 48 in HbA
1c
. Analysis was done on a modi? ed intention-to-treat population. This study
was registered with Chinese Clinical Trial Registry, number ChiCTR-TRC-08000231.
Findings Of the 788 patients randomly assigned to treatment groups, 784 patients started the intended study drug.
HbA
1c
reduction at week 24 was –1·17% in the acarbose group and –1·19% in the metformin group. At week 48, the
HbA
1c
reduction was ?1·11% (acarbose) and ?1·12% (metformin) with di? erence 0·01% (95% CI ?0·12 to 0·14,
p=0·8999). Six (2%) patients in the acarbose group and seven (2%) patients in the metformin group had serious
adverse events, and two (1%) and four (1%) had hypoglycaemic episodes.
Interpretation This study provides evidence that acarbose is similar to metformin in e? cacy, and is therefore a viable
choice for initial therapy in Chinese patients newly diagnosed with type 2 diabetes.
Funding Bayer Healthcare (China) and Double Crane Phama.
Introduction
Despite advances in treatment for type 2 diabetes, an optimum strategy for glycaemic control remains elusive. Metformin is the only ? rst-line oral hypoglycaemic drug for type 2 diabetes with proven e? cacy, safety, and cost-e?ectiveness that is recommended by international guidelines.1 Robust evidence for metformin has been generated mostly from white populations2,3 with extrapolations for other populations;4 few studies have assessed metformin in other populations, especially in eastern Asian patients with lower BMI5 and exaggerated postprandial glucose excursion.6,7
The α-glucosidase inhibitors acarbose and voglibose are commonly used as monotherapy for mild diabetes, and in combination with other oral drugs or insulin for more advanced diabetes, in China and other eastern Asian countries. The reason for di?erences in use of α-glucosidase inhibitors between white and Asian population remains ill de? ned.
A previous head-to-head study to compare α-glucosidase inhibitors with metformin as the initial therapy for type 2 diabetes has not been reported. One previous study in
patients with impaired glucose tolerance8s howed similar
e?cacy of metformin and acarbose in reducing the
incidence of new-onset diabetes after 3 years of follow-
up. Another study9 showed that the two drugs led to a
similar reduction in HbA
1c
in more advanced diabetes.
The mechanisms of action for both metformin and
α-glucosidase inhibitors are related to the gastrointestinal
tract. However, information about Chinese dietary
patterns and the e?cacy of acarbose is lacking. In
addition, the relationship between glucose, insulin,
glucagon, and glucagon-like peptide-1 (GLP-1), needs to
be clari? ed especially for intervention with acarbose and
metformin. 10–18
We therefore did a non-inferiority trial to compare
acarbose with metformin as the initial therapy in
Chinese patients newly diagnosed with type 2 diabetes.
In addition to glycaemic control, we investigated e? ects
on levels of insulin, glucagon, and GLP-1; β-cell insulin-
secretory capacity and insulin sensitivity; and the
in? uence of dietary carbohydrate on glycaemic control.
Published Online
October 18, 2013
https://www.wendangku.net/doc/ba7893841.html,/10.1016/
S2213-8587(13)70021-4
See Online/Comment
https://www.wendangku.net/doc/ba7893841.html,/10.1016/
S2213-8587(13)70107-4
China–Japan Friendship
Hospital, Beijing, China
(Prof W Yang MD,
Prof Z Yang MD); Shanxi Province
People’s Hospital, Taiyuan,
China (Prof J Liu PhD); The First
Hospital of China Medical
University, Shenyang, China
(Prof Z Shan PhD); West China
Hospital, Sichuan University,
Chengdu, China
(Prof H Tian MD); Xiangya
Second Hospital of Central
South University, Changsha,
China (Prof Z Zhou PhD); Xijing
Hospital, Fourth Military
Medical University, Xi’an, China
(Prof Q Ji PhD); The Third
A? liated Hospital of Sun Yat-
sen University, Guangzhou,
China (Prof J Weng PhD);
Shanghai Jiaotong University
A? liated Sixth People’s
Hospital, Shanghai, China
(Prof W Jia PhD); Chinese People’s
Liberation Army General
Hospital, Beijing, China
(Prof J Lu MD); Gansu Provincial
Hospital, Lanzhou, China
(Prof J Liu PhD); Beijing Chao
Yang Hospital, Beijing, China
(Y Xu MD); and Peking Union
Medical College Hospital,
Beijing, China (Prof W Chen PhD)
Correspondence to:
Prof Wenying Yang, Department
of Endocrinology, China–Japan
Friendship Hospital, Beijing
100029, China
ywying_1010@https://www.wendangku.net/doc/ba7893841.html,
Articles
Methods
Participants
For this non-inferiority, multicentre, randomised controlled trial we recruited 788 patients newly diagnosed with type 2 diabetes, aged between 30 and 70 years, from 11 clinical sites in China, after completion of the Chinese national diabetes and metabolic disorders study.6 All patients were diagnosed within the past 12 months with type 2 diabetes according to 1999 WHO criteria, had either not received oral anti-diabetic drugs or had been on short-term (1 month) treatment that had been discontinued 3 months before enrolment, had suboptimum glucose control (HbA
1c between 7% and 10% and fasting plasma glucose [F PG]
≤11·1 mmol/L), and had a BMI of 19–30 kg/m2.
Exclusion criteria were a history of renal disease with a
plasma creatinine concentration of 133 μmol/L (1·5 mg/dL)
or more; severe gastrointestinal diseases; cardiac diseases
(a history of unstable angina or myocardial infarction
within the previous 6 months or New York Heart
Association class III or IV congestive heart failure); hepatic
diseases, or an aspartate amino t ransferase or alanine
aminotransferase concentration at least twice as high as the
upper limit of the normal range; chronic hypoxic diseases
(emphysema or cor pulmonale); haematological diseases;
endocrine diseases (hypo t hyroidism, hyperthyroidism,
Cushing’s syndrome); uncontrolled hypertension (systolic
pressure ≥160 mm Hg or diastolic pressure ≥95 mm Hg);
acute illness; a history of intestinal surgery; women with
the potential to become pregnant, who were preparing for
conception, or who were pregnant or breastfeeding;
participation in any drug clinical trials during the past
3 months before enrolment; mental disorders; drug or
other substance misuse; requirement for insulin therapy;
diabetic ketoacidosis; or hyperosmolar non-ketotic coma.
Key withdrawal criteria included an allergic reaction or
intolerance to study drugs; inability to continue according
to protocol requirements; unwillingness to follow the
study; or FPG greater than 11·1 mmol/L with hypoglycaemic
drugs titrated to the maximum dose. All patients provided
written informed consent and con? rmed their willingness
to participate. The protocol was approved by an ethics
committee from each clinical site and was implemented in
accordance with provisions of the Declaration of Helsinki
and Good Clinical Practice guidelines.
Randomisation and masking
Randomisation codes were generated with a computer
programme (SAS version 9.10) for eligible patients with
PG between 7·0 and 11·1 mmol/L. Patients were
randomly assigned (1:1) to each of the two treatment groups
(block size 8) at 11 centres. Neither patients nor investigators
involved in the study were masked to treatment allocation. Procedures
A 4-week screening and run-in phase was used for
therapeutic lifestyle modi?cation with group patient
education, according to Chinese diabetes management
guidelines. The rule of “start low, go slow” was followed for
the drug intervention to avoid and attenuate possible
gastrointestinal e?ects. After the 4-week run-in phase,
patients were assigned to receive sustained-released
metformin hydrochloride up to 1500 mg, once daily
(500 mg per tablet, Beijing Double Crane Pharma, Beijing,
China), or up to 100 mg of acarbose, three times daily
(50 mg per tablet, Bayer Healthcare, Beijing, China),with
24-week monotherapy and 24-week add-on therapy with
insulin secretagogues if needed Acarbose was started from
50 mg once a day at dinner during the ?rst week and
Figure 1: Trial pro? le
Total dropout rate was 18·8%. ITT=intention to treat.
Articles titrated up to 50 mg twice a day at lunch and dinner in the
second week, 50 mg three times a day at three meals in the
third week, and 100 mg three times a day from the fourth
week onwards. Metformin was started at 500 mg once a day
after dinner for the ? rst 2 weeks and titrated up to 1000 mg
once a day after dinner in the third week and to 1500 mg
once a day after dinner from the fourth week onwards.
According to 2007 Chinese management guidelines for
type 2 diabetes, add-on therapy with insulin secretagogues
was started at week 24 in patients whose HbA
1c
was higher
than 7%, or in those who had FPG higher than 7 mmol/L
or postprandial glucose of more than 10 mmol/L for
3 consecutive days by self-monitored blood glucose.
The primary endpoints were reduction in HbA
1c
at
24 weeks and 48 weeks. Key secondary endpoints included
the proportion of patients with HbA
1c
of 6·5% or less;
change in F PG, postprandial 2-h glycaemic pro? le,
bodyweight, insulin, glucagon, GLP-1, and insulin
sensitivity or β-cell function by HOMA index, all measured
at baseline, 24 weeks, and 48 weeks. We measured plasma
glucose, insulin, glucagon, and GLP-1 before and after a
standardised breakfast (70 g instant noodle equivalent to
an energy intake of 500 kilocalories) at baseline, 24 weeks,
and 48 weeks. W e also investigated the relation between
glucose pro? le and dietary pattern at baseline, 24 weeks,
and 48 weeks. Safety endpoints included hypoglycaemic
episodes, adverse events, vital signs, 12-lead electro-
cardiogram, biochemistry, lipid pro? le, haematology
measures, and routine urine tests.
BMI was calculated at baseline. Visits with patients were
scheduled every 2 weeks, or every 4 weeks after week 4.
Blood pressure, waist circumference, and bodyweight were
measured at all visits. HbA
1c
levels were measured and
seven-point glucose pro?les were requested (on the day
before each visit for investigators’ clinical appraisal for
safety and e? cacy; data not reported since not a primary
nor secondary outcome; OneTouch Ultra glucose meter,
Johnson & Johnson, Shenzen, China). Plasma creatinine
con c en t rations were measured at baseline and at 4, 12, 24,
and 48 weeks. Plasma creatinine, lipids (including HDL
cholesterol, LDL cholesterol, triglycerides), and alanine
aminotransferase were measured at local sites’ laboratories
at baseline, 24 weeks, and 48 weeks. Plasma samples were shipped in dry ice to a central laboratory at the China–Japan Friendship Hospital (Beijing, China). HbA
1c
was measured by high-performance liquid chromatography at the China–Japan Friendship Hospital (Biorad Variant-II, Biorad, CA, USA)(normal range, 4·5–6·2%). Serum insulin (Beckman insulin kit, Prague, Czech Republic), plasma glucagon (Linco GL-32K kit, CA, USA), and active GLP-1 (Linco GLP1A-35HK kit, CA, USA) were measured by radioactive immunoassay in an authorised laboratory by certi? ed technicians from China National Nuclear Corporation (with XH6020 4-detector RIA automatic gamma counter, Xi’An nuclear instrument factory/China state-owned 262 factory). All values had to meet quality-control standards, including a coe? cient of variation of 25% or less.
F or analysis of dietary macronutrients, patients com-pleted a questionnaire about dietary information the day before visits. Energy intake from carbohydrates, proteins, and fats was calculated by independent and dedicated clinical nutrition experts at baseline, 24 weeks, and 48 weeks. For calculation of HOMA index, we used the following formulae (INS=fasting insulin): HOMA-IR=INS × FPG/22·5; HOMA-B=20 ×FINS/(FPG–3·5); early insu l in secretion index=ΔI30/ΔG30; whole body insulin sensitivity index=10 000/(F P
G [mg/dL] × F INS) × (mean glucose [mg/dL] × mean insulin).1/2
Statistical analysis
We concluded non-inferiority if the upper limit of the 95% CI for the treatment di? erence was less than 0·3%
Articles
in change in HbA
1c
from baseline to week 48. With the assumption of a SD of 1·3%, 295 patients per treatment group (a total of 590 patients) were required to achieve 80% power for the per protocol analysis. To allow for a 20% dropout rate, we aimed to randomise a total of 738 patients (369 patients per treatment group).
We did e? cacy analyses using prespeci? ed modi? ed intention-to-treat and per-protocol populations. Patients included in the modi? ed intention-to-treat analysis had received at least one dose of study drug, had e? cacy data at baseline, and had at least one post-baseline measurement of the respective variable. We included all patients randomly assigned to treatment groups with documented safety data in the safety analysis. We used SAS (version 9.1.0) for data analysis, and used the
last observation carried forward (LOCF) approach for
missing data for the primary endpoints only. For HbA
1c data, we used ANCOVA with treatment and centre as
factors, and baseline as a covariate, and included
treatment by centre interaction. We calculated least-
squares means and 95% CIs for mean di? erences
between the two treatment groups. We used repeated
measures ANCOVA to assess e? cacy parameters over
the course of the study. Category variables were
analysed with χ2 tests, CMH-χ2 tests, or Fisher’s exact
test where appropriate. These tests were also used to
assess di? erences in the incidence of adverse events
between treatments.
45Articles
Role of the funding source
The Chinese Diabetes Society was involved in study design and implementation, appointment of contract research organisations, data analysis and interpretation, and medical writing. Independent sta? masked to randomisation analysed data and evaluated safety. Bayer Healthcare (China) provided ? nancial support and acarbose, and Beijing Double Crane Pharma provided metformin. Neither Bayer nor Double Crane had a role in study design, implementation, data analysis, interpretation, or writing of the report. All the authors had full access to all the data in
the study and the corresponding author had ?
nal responsibility for the decision to submit for publication.Results
F igure 1 shows the study pro? le. F rom Nov 8, 2008, to June 27, 2011, we screened 1099 patients and randomly allocated 788 to the two treatments. Four withdrew consent before drug intervention. 784 patients commenced study drug (393 metformin and 391 acarbose). 16% of patients in
Figure 2: Primary outcome according to treatment group
Δ acarbose–Δ metformin represents the di? erence in the primary outcome of HbA 1c reduction between acarbose group and metformin group at 24 weeks and 48 weeks.
Articles
the acarbose group (40 at the ? rst 24-week monotherapy phase and 25 at the add-on therapy phase) and 20% of patients in the metformin group (46 at the ? rst 24-week monotherapy phase and 33 at the add-on therapy phase) discontinued study drugs before the end of follow-up. Only ? ve patients in the acarbose group and three patients in the metformin group received insulin secretagogues as add-on therapy after 24-week monotherapy and until study completion. No patients stopped the study on the basis of our withdrawal criteria (F PG of more than 11·1 mmol/L with maximum dose of hypoglycaemic drugs based on clinical decisions). The two treatment groups were balanced with respect to baseline characteristics (table 1).
At week 24 and at week 48, HbA 1c was reduced compared with baseline in both acarbose and metformin groups (table 2, ? gure 2). At week 48, the least-squares mean treatment di? erence between acarbose and metformin for reduction of HbA 1c was 0·01% (two-sided 95% CI –0·12 to 0·14; p=0·8999), demonstrating non-inferiority (? gure 2).The reduction in FPG at week 48 was greater in patients taking metformin than in those taking acarbose, whereas reduction in 2-h postprandial glucose was greater in patients taking acarbose than those taking metformin (table 2). We observed a progressive decrease in bodyweight in both treatment groups, although patients taking acarbose had lost more weight than had those taking metformin at week 24 (?0·67 kg, 95% CI –1·14 to –0·20; p=0·0054) and at week 48 (–0·63 kg, –1·15 to –0·10; p=0·0194; table 2). Results from the modi? ed intention-to-treat analysis were in agreement with ? ndings from the per-protocol analysis (appendix).HbA 1c reduction di? ered signi? cantly within the acarbose group as well as the metformin group when patients were strati? ed by baseline HbA 1c (ie, <7%, 7–8%, >8%) by Nemenyi test (all p<0·0001 within subgroups; ? gure 3). There was no signi? cant di? erence in change from baseline HbA 1c between acarbose and metformin when strati? ed by baseline HbA 1c . There was no di? erence in the proportion of patients with HbA 1c of 6·5% or less between metformin (67%) and acarbose (69%) at 24 weeks (p=0·6231) and between metformin (62%) and acarbose (64%) at 48 weeks (p=0·5422; table 2).The contribution of carbohydrates for energy intake in the study overall was higher than the China dietary
Figure 3: Least-squares mean change in glycated haemoglobin from baseline strati? ed by (A) baseline glycated haemoglobin and (B) baseline percentage of energy intake from carbohydrate
The results by carbohydrate intake from the intention-to-treat and per-protocol populations were consistent.
See Online for appendix
35
40
Articles
recommendations (up to 65%) and international guidelines (45–65%), with mean 67% (SD 9) at baseline, 66% (SD 11) at 24 weeks, and 68% (SD 9) at 48 weeks. There was no signi? cant association between reduction in HbA
1c
and proportion of dietary carbohydrate intake from baseline to week 24 or week 48 for either metformin or acarbose (? gure 3). Neither was there an association between baseline BMI and HbA
1c
reduction (appendix). On the whole, no di?erence in indices for insulin sensitivity and β-cell function were found between acarbose and metformin groups at week 24 or 48 (table 2); however, there was a signi? cant di? erence between groups in whole body insulin sensitivity index at 48 weeks (p=0·0424). We consider this a chance ? nding. After 24-week and 48-week treatment, there was no di? erence between the metformin and acarbose group in the fasting state for serum insulin, glucagon, and GLP-1 concentration. With the standard meal test after 24-week intervention, (?gure 4) acarbose monotherapy was associated with a signi? cantly greater insulin-sparing e? ect compared with metformin (change in serum insulin concentration at 30 min, p=0·0022; at 120 min, p<0·0001; at 180 min, p=0·0657). Similar results were shown after 48-week treatment, although di? erences were not signi?cant at all timepoints (change in serum insulin concentration at 30 min, p=0·2661; at 120 min, p=0·0002; at 180 min, p=0·0521; ? gure 4B). A signi? cant di?erence between metformin and acarbose groups was shown for change in area under the curve for serum insulin after the 180 min standard meal test (at 24 weeks, p=0·0001, and at 48 weeks, p=0·0047; table 2).
Figure 4: Mean glucose (A), insulin (B), glucagon (C), and GLP-1 (active; D) concentrations during standard meal test, by intention to treat
Values show means with SE (at week 0) and least-squares means with SE (at week 24 and 48). The values at weeks 24 and 48 are adjusted for baseline values and centre where appropriate. Note that y axes in C and D do not start at 0. GLP-1=plasma glucagon-like peptide-1 (active). *p<0·05, ?p<0·01, for comparisons between acarbose and metformin groups at week 24 or 48. p values for treatment di? erence (acarbose–metformin) were: (A) week 24: p<0·0001 (0 min), p=0·0892 (30 min), p=0·0421 (120 min), p=0·5149 (180 min); week 48: p=0·0385 (0 min), p=0·0050 (30 min), p=0·0003 (120 min), p=0·0604 (180 min); (B) week 24: p=0·9829 (0 min), p=0·0022 (30 min),
p<0·0001 (120 min), p=0·0657 (180 min); week 48: p=0·1857 (0 min), p=0·2661 (30 min), p=0·0002 (120 min), p=0·0521 (180 min); (C) week 24: p=0·0490 (0 min),
p=0·1707 (30 min), p=0·0176 (120 min), p=0·4075 (180 min); week 48: p=0·8073 (0 min), p=0·3556 (30 min), p=0·9147 (120 min), p=0·2912 (180 min); (D) week 24:
p=0·7073 (0 min), p=0·2652 (30 min), p=0·5722 (120 min), p=0·1467 (180 min); week 48: p=0·3280 (0 min), p=0·5956 (30 min), p=0·0826 (120 min), p=0·3183 (180 min).
Articles
In the metformin group, there was no di? erence between baseline and week 24 in glucagon concentrations during the standard meal test, but there was a signi? cant reduction at week 48 compared with baseline (p<0·0001 at all time points; ? gure 4C). However, in the acarbose group, there was a signi? cant decrease compared with baseline at week 24 at all but the 180 min timepoint (p=0·0009 for fasting, p=0·0476 for 30 min and p=0·0190 for 120 min) and further reduction in concentrations at week 48 (p<0·0001 at all timepoints). The dynamics in the change di? ered in each group: glucagon concentrations remained level after 120 min in the acarbose group, but decreased further from 120 min to 180 min in the metformin group at both week 24 and week 48 (? gure 4C). Overall, however, the two drugs did not di? er signi? cantly from each other in their e? ects on glucagon during the standard meal test. Both acarbose and metformin treatment were associated with a similar increase in GLP-1 concentrations from baseline during a standard meal test at 24 weeks and 48 weeks (p<0·0001 compared with baseline at all timepoints, except p=0·0003 for acarbose at 30 min at week 24, and p=0·0002 for metformin at 30 min, week 48; ? gure 3D). However, the two drugs showed di? erences in the time of peak GLP-1 concentrations: the peak occurred at 120 min in the metformin group and was delayed until after 120 min in the acarbose group (? gure 4D).
Table 3 shows serious adverse events and adverse events
that occurred in more than 5% of patients in either group.
Six serious adverse events in the acarbose group and seven
in the metformin group were reported. Consistent with
the known safety pro? le of both drugs, the most common
adverse events were mild to moderate gastrointestinal
symptoms (100 [27%] patients in the acarbose group vs 107
[29%] in the metformin group). There were no deaths or
severe hypoglycaemic events in either group. Two patients
in the acarbose group and four in the metformin group
reported hypoglycaemic episodes.
Discussion
In this study we have shown that acarbose treatment was
non-inferior to metformin treatment in view of HbA
1c reduction after 48 weeks of treatment. Metformin’s
e?cacy in reducing HbA
1c
is independent of baseline
BMI (panel).19 We also show that 100 mg acarbose three
times a day decreases HbA
1c
to a greater extent in Chinese
patients (1%) than previously reported in white
populations (0·5–1%).20 Higher baseline HbA
1c
(8%) was
associated with greater treatment-emergent glycaemic
reductions (2%). These ? ndings are in agreement with
results from the China vildagliptin phase 3 study,21,22 in
which this drug was compared head-to-head with
acarbose in drug-naive Chinese patients newly diagnosed
with diabetes. That study used the same acarbose dose
(up to 300 mg daily), and reported a greater reduction in
HbA
1c
(1·36%), higher baseline HbA
1c
(8·6%), and a
longer duration of diabetes (1·3 years), but without using
an initial washout phase and lifestyle advice. The changes
in fasting and postprandial glucose with metformin and
acarbose treatment are also in agreement with previous
studies.8,18 However, concomitant with glucose-lowering
e? ect, we did not observe changes in β-cell function or
insulin resistance indices between the two groups.
We did not ? nd a signi? cant correlation between glucose
control and weight reduction in our study.α-glucosidase
inhibitors are thought to have a neutral e? ect on
bodyweight, whereas metformin is associated with mild
weight reduction in overweight and obese patients with
type 2 diabetes.1 Acarbose’s e?ects on bodyweight
reduction in this study are in line with those of previous
studies in Chinese22 and other populations.23 The mean
BMI in the present study and in the China national cross-
sectional epidemiological study,6 under t aken during the
same period, are consistent and are characteristic of the
Chinese population with newly diagnosed diabetes. Thus,
our ?ndings are generalisable to patients with newly
diagnosed type 2 diabetes in China. However, the
underlying mechanisms of weight reduction by acarbose
and its clinical relevance need to be examined further.
China is experiencing rapid nutritional transition from
a traditional high-carbohydrate dietary pattern to a
combination of traditional and western dietary patterns.
Eastern Asian populations, especially in China, have
cultivated cereal (rice) and pulses (soy bean) as major
Articles sources of energy for thousands of years. α-glucosidase
inhibitors delay the degradation of complex carbohydrates
into glucose, so that carbohydrates remain in the
intestine. In view of the perceived di? erences in e? cacy
of acarbose in reduction in HbA
1c
between white and
Chinese populations, one hypothesis of this study was to
test whether the e? cacy of acarbose vs metformin was
correlated with a relative proportion of dietary
carbohydrates. We show here that the relative proportion
of dietary carbohydrates consumed by most newly
diagnosed Chinese patients with type 2 diabetes is higher
than that recommended by Chinese and international
dietary guidelines for macronutrients.24–27We did not ? nd
that e? cacy was signi?cantly associated with baseline
dietary carbohydrate intake, but we detected a trend to
suggest that acarbose (but not metformin) might lower
HbA
1c
to a greater extent in patients with dietary
carbohydrate intake greater than 65·5% than in those
with less than 65·5%. However, this needs investigation
in further studies with larger sample sizes.
Both acarbose and metformin monotherapy were
associated with signi? cant insulin-sparing e? ects in both
the fasting state and after standard meal test. We found
that metformin and acarbose attenuate glucagon response
with late and slow modes of action (no change at week 24 and signi? cant reduction at week 48 for the metformin group; stepwise reduction at 24 weeks and 48 weeks for the acarbose group); this hints at an indirect e? ect of glucagon in these treatments. Both drugs also have signi? cant e?ects on GLP-1, boosting and prolonging concentrations before and after standard meals compared with baseline, although the GLP-1 response after the standard meal test in both groups at baseline seemed blunted. These ? ndings also need to be veri? ed further. Both metformin and acarbose were associated with changes from baseline in insulin, glucagon, and GLP-1 concentrations after the standard meal test. The acarbose group showed a greater insulin-sparing e?ect than the metformin group, and a di?erent pattern of glucagon response with a peak at 120 min; GLP-1 responses also might be prolonged with acarbose compared with metformin. Thus, in addition to a? ecting HBA
1c
, the capacity of acarbose and metformin to a? ect the concentrations of glucose, insulin, GLP-1, and glucagon—and the complex gastrointestinal, endocrine, and metabolic interplay between these molecules—should be considered. This interplay could include glucose-stimulated insulin secretion, potentiation of insulin secretion by GLP-1, and the balance between insulin and glucagon. Relationships between glucose, insulin, GLP-1, and glucagon and the short-term and long-term e? ects of glucose reduction need to be investigated. The two drugs might act via di? erent underlying mechanisms: metformin acting possibly through peroxisome proliferator-activated receptor-α, as shown in in animal studies,11,12 and inhibition of DPP-4 activity,10 as shown in patients with type 2 diabetes; and acarbose acting by increased stimulation of L-cells through delayed absorption and altered transit of dietary carbo-
hydrates. The trend we observed for the delayed but
prolonged GLP-1 e? ect after the standard meal test in the
acarbose group compared with the metformin group is in
line with current evidence.13–18 The delayed but prolonged
GLP-1 response with acarbose might partly explain earlier
inconsistent ? ndings.28
We speculate that the action of α-glucosidase inhibitors
partly mimics the e? ect of gastric bypass surgery,29 with
increased and prolonged stimulation of the enteroinsulinar
axis and a decreased hyperinsulinemic response. Another
possible mechanism of the e?ect of acarbose on bodyweight
might be related to improved microbiota after delayed
carbohydrate passage into the small intestine.30–33 Unlike
the action of acarbose, sparing energy use and cellular
energy depletion with AMPK-dependent and AMPK-
independent pathway,34 metformin exerts bene? cial e? ects
on insulin sensitivity and metabolic pro? le, partly through
modulation of multiple components of the incretin axis,
including gut hormones and gastric emptying.11,12 Both
acarbose and metformin are known from use over many
years in clinical practice to cause gastrointestinal adverse
e?ects including ?atulence, diarrhoea, and abdominal
discomfort; slow titration can attenuate these symptoms.
The main limitation of our study is that there was no
placebo-controlled group, for ethical reasons, although
there was a 4-week run-in phase compared with a previous
study.22 Another limitation is that there was no hyperglycaemic clamp or duodenal nutrient perfusion to
eliminate di?erences in blood glucose and gastric
emptying. The study did not assess the role of other gut
hormones that could be a? ected by these drugs, especially
gastric inhibitory peptide.
Panel: Research in context
Systematic review
We searched PubMed up to March 22, 2013, with the search terms “α-glucosidase inhibitors”, “metformin”, and “randomised trial”. We reviewed randomised clinical trials and meta-analyses published in English and were unable to ? nd a comparative e? ectiveness trial that compared the two strategies as an initial treatment in drug-naive patients with type 2 diabetes.
Interpretation
This study was the ? rst large randomised controlled trial to assess acarbose as an initial treatment in newly diagnosed Chinese patients by comparison with metformin, the only ? rst-line therapy recommended by American Diabetes Association and European Association for the Study of Diabetes guidelines. Our ? ndings show that acarbose was
non-inferior to metformin in reduction of HbA
1c
for these patients. The safety pro? le also showed that acarbose was well tolerated in Chinese patients with less frequent safety events than was seen in white populations. Furthermore, a trend suggested that e? cacy of
acarbose in reduction in HbA
1c
and postprandial glucose was prospectively related to the proportion of dietary carbohydrates in Chinese patients. There was greater weight loss with acarbose than with metformin. Both metformin and acarbose diminish insulin and glucagon concentrations while increasing GLP-1 concentration. The evidence provided by this comparative e? ectiveness trial might help physicians in considering α-glucosisdase inhibitors as an alternative initial therapy for type 2 diabetes.
Articles
This study is the ?rst head-to-head comparison of metformin and acarbose as initial therapy for type 2 diabetes after failure of therapeutic lifestyle modi? cation. We report that both acarbose and metformin are well tolerated and have similar e?cacy as initial therapy for HBA
1c
reduction in Chinese patients with type 2 diabetes. Our ? ndings are consistent with those of previous studies of acarbose with mild weight reduction, which has been neglected previously in clinical practice. Both metformin and acarbose have insulin-sparing and glucagon-sparing e?ects and increase GLP-1 concentrations. The possible interplay between e?ects on glucose, insulin, glucagon, and GLP-1 with these drugs needs to be assessed further. Metformin should remain as ?rst-line treatment for patients with newly diagnosed type 2 diabetes, while patients with exaggerated postprandial excursion can be treated with an α-glucosidase inhibitor as an alternative therapy before cardiovascular bene?ts of acarbose are validated and con? rmed in ongoing studies.35
Contributors
WY designed, submitted, and wrote the report. All authors were involved in study protocol discussion and data collection, and writing the Discussion. Con? icts of interest
We declare that we have no con? icts of interest.
Acknowledgments
We thank Wenbin Yang for his constructive discussion of the manuscript preparation.
References
1 Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of
hyperglycaemia in type 2 diabetes: a patient-centered approach.
Position statement of the American Diabetes Association (ADA) and
the European Association for the Study of Diabetes (EASD).
Diabetologia 2012; 55: 1577–96.
2 UK Prospective Diabetes Study (UKPDS) Group. E? ect of intensive
blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). Lancet 1998; 352: 854–65.
3 Holman RR, Paul SK, Bethel MA, et al. 10-year follow-up of intensive
glucose control in type 2 diabetes. N Engl J Med 2008; 359: 1577–89.
4 Chan JC, Deerochanawong C, Shera AS, et al. Role of metformin in
the initiation of pharmacotherapy for type 2 diabetes: an Asian-
Paci? c perspective. Diabetes Res Clin Pract 2007; 75: 255–66.
5 WHO Expert Consultation. Appropriate body-mass index for Asian
populations and its implications for policy and intervention
strategies. Lancet 2004; 363: 157–63.
6 Yang W, Lu J, Weng J, et al. Prevalence of diabetes among men and
women in China. N Engl J Med 2010; 362: 1090–101.
7 Wang JS, Tu ST, Lee IT, et al. Contribution of postprandial glucose to
excess hyperglycaemia in Asian type 2 diabetic patients using
continuous glucose monitoring. Diabetes Metab Res Rev 2011;
27: 79–84.
8 Y ang W, Lin L, Qi J. The preventive e? ect of acarbose and metformin on
the IGT population from becoming diabetes mellitus: a 3-year
multicentral prospective study. Chin J Endocrinol Metab 2001; 17: 131–34.
9 Ho? mann J, Spengler M. E? cacy of 24-week monotherapy with
acarbose, metformin, or placebo in dietary-treated NIDDM patients: the Essen-II Study. Am J Med 1997; 103: 483–90.
10 Cuthbertson J, Patterson S, O’Harte FP, et al. Investigation of the
e? ect of oral metformin on dipeptidylpeptidase-4 (DPP-4) activity in type 2 diabetes. Diabet Med 2009; 26: 649–54.
11 Cho YM, Kie? er TJ. New aspects of an old drug: metformin as a
glucagon-like peptide 1 (GLP-1) enhancer and sensitiser. Diabetologia 2011; 54: 219–22.
12 Maida A, Lamont BJ, Cao X, et al. Metformin regulates the incretin
receptor axis via a pathway dependent on peroxisome
proliferator-activated receptor-α in mice. Diabetologia 2011;
54: 339–49.13 Seifarth C, Bergmann J, Holst JJ, et al. Prolonged and enhanced
secretion of glucagon-like peptide 1 (7-36 amide) after oral sucrose due to alpha-glucosidase inhibition (acarbose) in type 2 diabetic
patients. Diabet Med 1998; 15: 485–91.
14 En? FY, Imeryüz N, Akin L, et al. Inhibition of gastric emptying by
acarbose is correlated with GLP-1 response and accompanied by CCK release. Am J Physiol Gastrointest Liver Physiol 2001; 281: G752–63.
15 Ranganath L, Norris F, Morgan L, et al. Delayed gastric emptying
occurs following acarbose administration and is a further mechanism for its anti-hyperglycaemic e? ect. Diabet Med 1998; 15: 120–24.
16 G?ke B, Herrmann C, G?ke R, et al. Intestinal e? ects of alpha-
glucosidase inhibitors: absorption of nutrients and enterohormonal changes. Eur J Clin Invest 1994; 24 (suppl 3): 25–30.
17 G?ke B, Fuder H, Wieckhorst G, et al. Voglibose (AO-128) is an
e? cient alpha-glucosidase inhibitor and mobilizes the endogenous GLP-1 reserve. Digestion 1995; 56: 493–501.
18 Lee A, Patrick P, Wishart J, et al. The e? ects of miglitol on
glucagon-like peptide-1 secretion and appetite sensations in obese type 2 diabetics. Diabetes Obes Metab 2002; 4: 329–35.
19 Esposito K, Chiodini P, Bellastella G, et al. Proportion of patients at
HbA1c target 7% with eight classes of antidiabetic drugs in type 2
diabetes: systematic review of 218 randomized controlled trials with 78,945 patients. Diabetes Obes Metab 2012; 14: 228–33.
20 Van de Laar FA, Lucassen PL, Akkermans RP, et al. Alpha-glucosidase
inhibitors for type 2 diabetes mellitus. Cochrane Database Syst Rev
2005; 2: CD003639.
21 Pan C, Yang W, Barona JP, et al. Comparison of vildagliptin and
acarbose monotherapy in patients with Type 2 diabetes: a 24-week, double-blind, randomized trial. Diabet Med 2008; 25: 435–41.
22 Pan C, Yang W, Ji Q, et al. Comparison of vildagliptin and acarbose
monotherapy in patients with type 2 diabetes: a 24-week, multi-
center, double-blind, double dummy controlled, randomized trial.
Chin J Endocrinol Metab 2009; 25: 386–90.
23 Holman RR, Cull CA, Turner RC. A randomized double-blind trial
of acarbose in type 2 diabetes shows improved glycemic control over
3 years (U.K. Prospective Diabetes Study 44). Diabetes Care 1999;
22: 960–64.
24 Ministry of Health of The People’s Republic of China. Dietary
Guidelines for Chinese (2007).
25 Su HY, Tsang MW, Huang SY, et al, for the Task Force for
Development of Transcultural Algorithms in Nutrition and
Diabetes. Transculturalization of a diabetes-speci? c nutrition
algorithm: Asian application. Curr Diab Rep 2012; 12: 213–19.
26 Dyson PA, Kelly T, Deakin T, et al, for the Diabetes UK Nutrition
Working Group. Diabetes UK evidence-based nutrition guidelines
for the prevention and management of diabetes. Diabet Med 2011;
28: 1282–88.
27 Nishida C, Uauy R, Kumanyika S, et al. The joint WHO/FAO expert
consultation on diet, nutrition and the prevention of chronic
diseases: process, product and policy implications.
Public Health Nutr 2004; 7: 245–50.
28 Hücking K, Kostic Z, Pox C, et al. alpha-Glucosidase inhibition
(acarbose) fails to enhance secretion of glucagon-like peptide 1
(7-36 amide) and to delay gastric emptying in type 2 diabetic
patients. Diabet Med 2005; 22: 470–76.
29 Karra E, Yousseif A, Batterham RL. Mechanisms facilitating weight
loss and resolution of type 2 diabetes following bariatric surgery.
Trends Endocrinol Metab 2010; 21: 337–44.
30 Holt PR, Atillasoy E, Lindenbaum J, et al. E? ects of acarbose on
fecal nutrients, colonic pH, and short-chain fatty acids and rectal
proliferative indices. Metabolism 1996; 45: 1179–87.
31 Wolever TM, Chiasson JL. Acarbose raises serum butyrate in human
subjects with impaired glucose tolerance. Br J Nutr 2000; 84: 57–61. 32 Dehghan-Kooshkghazi M, Mathers JC. Starch digestion, large-bowel
fermentation and intestinal mucosal cell proliferation in rats treated with the alpha-glucosidase inhibitor acarbose. Br J Nutr 2004; 91: 357–65.
33 Koytchev R, Richter W, Erkent U, et al. In? uence of acarbose on
blood glucose and breath hydrogen after carbohydrate load with
sucrose or starch. Arzneimittelforschung 2009; 59: 557–63.
34 Rena G, Pearson ER, Sakamoto K. Molecular mechanism of action
of metformin: old or new insights? Diabetologia 2013; 56: 1898–906.
35 Standl E, Schnell O. Alpha-glucosidase inhibitors 2012—cardiovascular
considerations and trial evaluation. DiabVasc Dis Res 2012; 9: 163–69.