Received: 23 August 2016
Revised: 31 October 2016
Accepted: 8 November 2016
DOI 10.1111/dom.12826
ORIGINAL ARTICLE
Sustained influence of metformin therapy on circulating glucagon-like peptide-1 levels in individuals with and without type 2 diabetes David Preiss PhD1† | Adem Dawed MPH2 | Paul Welsh PhD3 | Alison Heggie MRCP4 | Angus G. Jones PhD5 | Jacqueline Dekker PhD6 | Robert Koivula MSc7 | Tue H. Hansen MD8 | Caitlin Stewart MSc3 | Rury R. Holman FRCP9 | Paul W. Franks PhD7 | Mark Walker PhD4 | Ewan R. Pearson PhD2† | Naveed Sattar FMedSci3† and the DIRECT consortium group‡ 1
MRC Population Health Research Unit, Clinical Trial Service Unit and Epidemiological Studies Unit, Nuffield Department of Health, University of Oxford, Oxford, UK
2
Division of Cardiovascular and Diabetes Medicine, Medical Research Institute, University of Dundee, Dundee, UK
3
BHF Glasgow Cardiovascular Research Centre, Institute of Cardiovascular and Medical Sciences, University of Glasgow, Glasgow, UK
4
Institute of Cellular Medicine, Newcastle University, Newcastle upon Tyne, UK
5
NIHR Exeter Clinical Research Facility, University of Exeter Medical School, Exeter, UK
6
Department of Epidemiology and Biostatistics, VU Medical Center, Amsterdam, The Netherlands
7
Department of Clinical Sciences, Genetic and Molecular Epidemiology, Lund University, Malmö, Sweden
8
Section of Metabolic Genetics, Faculty of Health Sciences, Novo Nordisk Foundation Centre for Basic Metabolic Research, University of Copenhagen, Copenhagen, Denmark 9
Diabetes Trials Unit, Radcliffe Department of Medicine, University of Oxford, Oxford, UK
Correspondence David Preiss PhD, CTSU, Richard Doll Building, Old Road Campus, University of Oxford, Oxford, OX3 7LF, UK Email:
[email protected] E. R. Pearson, Division of Molecular and Diabetes Medicine, School of Medicine, University of Dundee, DD1 9SY, UK. Email:
[email protected] Funding information CAMERA was funded by the Chief Scientist Office, Scotland (CZB/4/613). The work leading to this publication has received support from the Innovative Medicines Initiative Joint Undertaking under grant agreement n 115317 (DIRECT), resources of which are composed of financial contribution from the European Union’s Seventh Framework Programme (FP7/2007-2013) and EFPIA companies’ kind contribution. E. R. P. holds a Wellcome Trust New Investigator Award (ref 102820/Z/13/Z).
Aims: To investigate, in the Carotid Atherosclerosis: Metformin for Insulin Resistance (CAMERA) trial (NCT00723307), whether the influence of metformin on the glucagon-like peptide (GLP)-1 axis in individuals with and without type 2 diabetes (T2DM) is sustained and related to changes in glycaemia or weight, and to investigate basal and post-meal GLP-1 levels in patients with T2DM in the cross-sectional Diabetes Research on Patient Stratification (DIRECT) study. Materials and methods: CAMERA was a double-blind randomized placebo-controlled trial of metformin in 173 participants without diabetes. Using 6-monthly fasted total GLP-1 levels over 18 months, we evaluated metformin’s effect on total GLP-1 with repeated-measures analysis and analysis of covariance. In the DIRECT study, we examined active and total fasting and 60minute post-meal GLP-1 levels in 775 people recently diagnosed with T2DM treated with metformin or diet, using Student’s t-tests and linear regression. Results: In CAMERA, metformin increased total GLP-1 at 6 (+20.7%, 95% confidence interval [CI] 4.7-39.0), 12 (+26.7%, 95% CI 10.3-45.6) and 18 months (+18.7%, 95% CI 3.8-35.7), an overall increase of 23.4% (95% CI 11.2-36.9; P < .0001) vs placebo. Adjustment for changes in glycaemia and adiposity, individually or combined, did not attenuate this effect. In the DIRECT study, metformin was associated with higher fasting active (39.1%, 95% CI 21.3-56.4) and total GLP-1 (14.1%, 95% CI 1.2-25.9) but not post-meal incremental GLP-1. These changes were independent of potential confounders including age, sex, adiposity and glycated haemoglobin.
†
Joint senior authors.
‡
Consortium members listed in Supplementary data.
356
© 2016 John Wiley & Sons Ltd
wileyonlinelibrary.com/journal/dom
Diabetes Obes Metab 2017; 19(3):356–363
PREISS
357
ET AL.
Conclusions: In people without diabetes, metformin increases total GLP-1 in a sustained manner and independently of changes in weight or glycaemia. Metformin-treated patients with T2DM also have higher fasted GLP-1 levels, independently of weight and glycaemia. KEYWORDS
antidiabetic drug, GLP-1, metformin
1 | I N T RO D UC T I O N Metformin is recommended as first-line therapy for the majority of individuals with type 2 diabetes mellitus (T2DM).1 This is based on evidence of cardiovascular benefit and also its capacity to maintain or reduce weight. In the UK Prospective Diabetes Study, metformin
influence, such as weight and glycaemia. To address these questions, we performed complementary studies, namely, an ancillary study using data from a randomized placebo-controlled repeated-measures study with 18 months’ follow-up, the Carotid Atherosclerosis: Metformin for Insulin Resistance (CAMERA) study,21 and a cross-sectional study by the Diabetes Research on Patient Stratification consortium,22 the DIRECT study.
monotherapy led to a 39% reduction in the risk of myocardial infarction compared with conventional dietary therapy over 10 years, a finding not explained by the drug’s effect on glycaemia.2 Metformin
2 | MATERIALS AND METHODS
has also been shown to reduce the risk of developing T2DM. In the Diabetes Prevention Program, metformin therapy reduced new-onset
The CAMERA study was a randomized double-blind placebo-
T2DM by 31% and also led to 2.1-kg weight loss compared with pla-
controlled trial designed to investigate the effect of metformin on
3,4
cebo over 2.8 years.
surrogate markers of cardiovascular disease in patients without dia-
The glucagon-like peptide-1 (GLP-1) axis remains at the forefront
betes, aged 35 to 75 years, with established coronary heart disease
of T2DM and cardiovascular research. Major outcomes trials of
and a large waist circumference (≥94 cm in men, ≥80 cm in women;
dipeptidyl peptidase-4 (DPP-4) inhibitors and the first completed out-
NCT00723307). This single-centre trial enrolled 173 adults who were
come trial of a GLP-1 receptor agonist in patients with T2DM indi-
followed up for 18 months each. Patients attended the research cen-
cated cardiovascular safety, although not benefit5–8; however, it was
tre every 6 months in a fasted state. A detailed description of the
recently reported that the potent GLP-1 receptor agonist, liraglutide,
trial and its results has been published previously.21 Participants were
has cardiovascular benefit.9 Furthermore, it has been reported that
randomized 1:1 to 850 mg metformin or matched placebo twice daily
another GLP-1 receptor agonist, semaglutide, has also provided cardi-
with meals, although they could reduce the dose to once daily based
This is supported by recently pub-
on side effects for the duration of the trial. Weight was measured in
lished results from a Mendelian randomization study of a GLP-1
light clothing using a bio-impedance scale. While bio-impedance body
10
ovascular benefit in a major trial.
genetic variant (Ala316Thr; rs10305492) strongly associated with lower fasting glucose levels, which demonstrated a lower risk of car11
diovascular disease,
supporting the concept that GLP-1 may indeed
be protective against cardiovascular disease. In addition, GLP-1 receptor agonists can yield modest weight loss12 and blood pressure reduction, important goals in the management of T2DM. It is unclear whether some of metformin’s benefits may be mediated via GLP-1. To explore this, it is important to establish robustly the effect of metformin on GLP-1, and whether any effect is mediated by changes in related variables such as weight or glycaemia. Various small studies of short duration have investigated the effect of metformin therapy on circulating GLP-1 levels in individuals with and without T2DM.13–20 While results have been inconsistent, some
fat results were available from the trial, we opted to measure circulating leptin levels as a better marker of body fat. The DIRECT study (www.direct-diabetes.org) is part of a European Union Innovative Medicines Initiative project, with the overarching aim of discovering and validating biomarkers of rapid diabetes development, progression and drug response.22 It involves four industrial partners and 21 academic institutes within Europe. As part of Work Package 2, which aimed to identify predictive biomarkers of glycaemic deterioration, deep phenotyping and biochemical assays were performed in 836 people recently diagnosed with T2DM who were receiving either metformin or lifestyle therapy alone at baseline. The 18-month follow-up data are being collected. For this study, complete cross-sectional data were analysed from the baseline visit in 775 participants from all six clinical centres.
have shown increases in active GLP-1 and total GLP-1 in both the fasting and postprandial states. To date, however, no suitable studies have been conducted to investigate robustly whether metformin
2.1 | Sample assays
therapy influences circulating GLP-1 levels in individuals with and
In CAMERA, participants attended 6-monthly visits after overnight
without T2DM, whether any observed effect is sustained in the
fasts and before taking their morning dose of metformin. Blood sam-
longer term (ie, beyond a few weeks), and whether any effect is
ples collected during the trial were centrifuged at 4 C soon after
related to changes in other variables that metformin is known to
sampling, separated and stored at −80 C at the Western Infirmary’s
358
PREISS ET AL.
Clinical Research Facility, Glasgow, for subsequent analyses. As previ21
placebo on total GLP-1 levels at 6, 12 and 18 months, respectively.
6-monthly fasting plasma glucose, fasting insulin
Additional on-treatment analyses were performed to assess whether
and HbA1c levels were analysed We calculated updated homeostasis
any change in total GLP-1 attributable to metformin was related to
model assessment for insulin resistance (HOMA2-IR) index values
simultaneous changes in weight, HOMA2-IR, HbA1c, leptin, or all
using the HOMA Calculator (v2.2.3, https://www.dtu.ox.ac.uk/
four variables combined, by adding these as cofactors.
ously described,
homacalculator/). Using available stored EDTA plasma samples, 6-
In the DIRECT study, fasting active and total GLP-1, and 60-
monthly total GLP-1 levels (Meso Scale Diagnostics, Rockville, Mary-
minute post-meal total GLP-1 levels were compared between metfor-
land) were measured with commercially available electrochemilumi-
min and lifestyle groups using Student’s t-tests. Anthropometric mea-
nescence assay (Meso Scale Diagnostics). Leptin levels were
sures (age, sex, waist–hip ratio, body mass index [BMI]), lifestyle
measured with a commercially available enzyme-linked immunosorb-
factors (smoking and alcohol use), HbA1c, fasting glucose and centre
ent assay (R&D Systems, Abingdon, UK). For total GLP-1, the mean
were investigated regarding any influence of metformin on GLP-1
inter- and intra-assay coefficients of variation (CVs) were 2.6% and
levels using linear regression models.
17.3%, respectively. For leptin, the mean inter- and intra-assay CVs
Because of the natural log transformation for GLP-1 measures,
were 10.1% and 6.3%. All time points for an individual participant
results are presented as the percentage differences in geometric
were run on the same plate, blinded to treatment arm.
means of GLP-1 measures on metformin vs placebo or metformin vs
For the DIRECT study, blood samples were collected in the
lifestyle to aid interpretation. The same approach was taken to pres-
morning after a 10-hour overnight fast. Metformin was stopped for
ent leptin results. Statistical analyses were carried out using the sta-
the 24 hours preceding the study visit and restarted immediately
tistical packages SPSS (version 22, SPSS Inc., Chicago, Ill) and R
thereafter. For a mixed meal test, participants drank 250 mL Fortisip
(version 3.0.1). A two-sided P value of 0.05 was used as the threshold
liquid drink (18.4 g carbohydrate/100 mL) over 2 to 5 minutes. Blood
for statistical significance.
samples were taken immediately prior to the drink (time 0) and then every 30 minutes up to 120 minutes. Samples for GLP-1 measurement were collected using P800 (for active GLP-1) and EDTA tubes
3 | RE SU LT S
(for total GLP-1; Becton Dickinson, Oxford, UK) at 0 and 60 minutes. The same commercial kits were used to measure GLP-1 levels as in
Baseline characteristics of the participants in the CAMERA and
CAMERA. In DIRECT, the mean intra- and inter-assay CVs for active
DIRECT studies are summarized in Table S1 and Table 1, respectively.
GLP-1 were 9% and 10%, respectively. For total GLP-1, these CVs
It was previously reported that metformin led to falls in HbA1c level
were 6% and 9%, respectively.
(0.13% or 1.4 mmol/mol), fasting insulin level (21%), HOMA-IR value (26%) and weight (3.2 kg) compared with placebo over 1.5 years in
2.2 | Ethics and consent
the CAMERA study. In the DIRECT study there was no significant difference in age,
All participants provided written informed consent for participation in
sex, BMI, duration of diabetes or HbA1c between the metformin and
both studies. For the CAMERA study, this included permission for
non-metformin treated groups. Metformin-treated individuals had a
biochemical assays that were not planned at the time of the trial. The
higher fasting glucose level (P < .001) and a slightly higher waist–hip
CAMERA trial was approved by the Medicines and Healthcare Pro-
ratio than those on no treatment (P = .045) in the DIRECT study.
ducts Regulatory Agency and West Glasgow Research Ethics Committee. In DIRECT, each partner clinical centre obtained approval from their respective research ethics review boards.
2.3 | Statistics
3.1 | CAMERA results: Metformin increases fasting total GLP-1 over 18 months The geometric mean for total GLP-1 was 3.52 pmol/L (11.6 pg/mL) in metformin recipients and 3.76 pmol/L (12.4 pg/mL) in placebo recipi-
Normality was assessed for all variables, and non-normally distributed
ents at baseline. Metformin therapy led to significant increases in
data were transformed using the natural log value where relevant
fasting total GLP-1 levels compared with placebo at each of the
(specifically for active GLP-1, total GLP-1, leptin and HOMA2-IR).
6, 12 and 18-month study visits (Table 2 and Figure 1A). The
In the CAMERA study, analyses were performed for the modified
increases in total GLP-1 levels at these visits were 21% (P = .010),
intention-to-treat population (ie, participants with a baseline total
27% (P = .001) and 19% (P = .012), respectively. In repeated-
GLP-1 and at least one subsequent total GLP-1 result). The effect of
measures analysis, metformin increased total GLP-1 level by 23.4%
metformin on total GLP-1 was investigated using two different
(P < .0001) across the entire duration of the 18-month follow-up,
approaches. First, repeated-measures analysis was carried out, allow-
with no evidence of heterogeneity among the study visits (P = .74).
ing a comparison of metformin- and placebo-treated participants over
Leptin levels fell with metformin treatment, in keeping with a
the entire trial (assuming a general covariance structure). Repeated-
reduction in body fat (Table 2). Overall, metformin therapy reduced
measures analyses were only performed after demonstrating that
leptin by 25% (P < .0001) compared with placebo, with similar
there was no significant treatment-by-visit interaction (ie, that any
changes observed at each visit.
observed effect was stable over the trial). Secondly, analyses of
Adjustment for the observed changes in weight, HOMA2-IR,
covariance were carried out to determine the effect of metformin vs
HbA1c and leptin at each visit, whether individually or combined, did
PREISS
359
ET AL.
TABLE 1
Characteristics of participants in the DIRECT study
Characteristics
Metformin (N = 270)
Men, n (%)
151 (55.9)
Median (range) age, y
P
Lifestyle (N = 505) 295 (58.4)
63 (35-75)
.70
64 (35-75)
.064
Duration of diabetes, y
1.18 (0.82)
1.25 (0.75)
.10
Weight, kg
88.9 (16.7)
89.7 (17.0)
.53
BMI, kg/m2
30.1 (4.7)
30.7 (5.1)
.11
Waist–hip ratio
0.97 (0.08)
0.96 (0.08)
.045
HbA1c, %
6.40 (0.60)
6.34 (0.59)
.20
HbA1c, mmol/mol
46.44 (6.57)
45.81 (6.39)
.20
Fasting plasma glucose, mmol/L
7.49 (1.47)
6.94 (1.37)