Background
Type 2 diabetes significantly increases the risk of cardiovascular morbidity and mortality [
1]. Clinical practice guidelines recommend the use of metformin as first-line therapy for the management of type 2 diabetes [
2,
3]. In addition to lowering hemoglobin A1C, long-term data indicate that metformin also may be associated with risk reductions in cardiovascular outcomes [
4]. As the majority of patients with type 2 diabetes often fail to achieve or maintain glycemic control with metformin alone, additional second-line therapies are often required [
5]. Guideline-recommended add-on therapies to metformin include glucagon-like peptide-1 receptor agonists (GLP-1RAs), sulfonylureas, thiazolidinediones, dipeptidyl peptidase-4 inhibitors, sodium-glucose cotransporter-2 inhibitors, and insulin [
2,
3]. Because of a lack of comparative long-term outcomes studies of dual-agent combinations that assess cardiovascular measures, the cardiovascular risk benefits of add-on therapies to metformin are uncertain [
6]. The measurement of various markers of cardiovascular risk (e.g., weight, waist circumference, blood pressure [BP], glycemic markers, lipid levels) in long-term studies of safety and efficacy may help to provide some guidance where long-term outcomes data are lacking.
The EURopean EXenAtide (EUREXA) study was the longest randomized, active-controlled, multinational study of a GLP-1RA with comparative treatment for up to 4.5 years with cardiovascular risk marker data collected throughout the trial as secondary endpoints [
7]. EUREXA compared the efficacy of exenatide twice daily versus glimepiride as second-line add-on therapy in patients with type 2 diabetes who had inadequate glycemic control with metformin alone. Although patients in randomized controlled trials might differ in several aspects from patients in real-life clinical practice, patients enrolled in EUREXA were similar to typical patients with type 2 diabetes with regard to the prevalence of overweight/obesity, lipid abnormalities, and cardiovascular disease at baseline. Primary data on time to treatment failure for patients in the EUREXA study have been published previously [
7]. The objective of the current report was to examine changes in the cardiovascular risk markers, which included waist circumference, BP, serum lipids, and high-sensitivity C-reactive protein (hsCRP, as a surrogate marker of inflammation), over time with exenatide twice daily versus glimepiride.
Research design and methods
Study design and patients
EUREXA was a randomized, multicenter, open-label, two-arm parallel, active comparator-controlled trial of adult patients with type 2 diabetes with inadequate glycemic control using metformin [
7]. Patients were enrolled in 14 countries (Austria, Czech Republic, Finland, France, Germany, Hungary, Ireland, Israel, Italy, Mexico, Poland, Spain, Switzerland, and the United Kingdom). The study was performed with approval from institutional review boards in line with country-specific regulations, and in compliance with Good Clinical Practice and the principles of the Declaration of Helsinki. All patients provided signed informed consent.
Full inclusion and exclusion criteria were published previously [
7]. Briefly, enrolled patients were aged 18–85 years with type 2 diabetes, had a body mass index ≥25 to <40 kg/m
2 with stable body weight for at least 3 months, were taking stable and maximum tolerated doses of metformin, and had inadequate glycemic control [A1C ≥6.5 % (48 mmol/mol) and <9.0 % (75 mmol/mol)]. Patients with active or untreated malignancy, renal or liver disease or dysfunction, chronic anemia or hemoglobinopathy, proliferative retinopathy or macular edema, or severe gastrointestinal disease were excluded.
Eligible patients were randomized in a 1:1 ratio, with stratification by A1C, to receive exenatide twice daily or glimepiride as add-on to current metformin treatment. Exenatide twice daily was initiated at 5 µg twice daily, increasing to 10 µg twice daily after 4 weeks, and was subcutaneously administered by an injection pen in the abdomen, thigh, or upper arm within 60 min before meals that were at least 6 h apart. The recommended starting dose for glimepiride was 1 mg/day orally immediately before the morning meal, with titration every 4 weeks to the maximum tolerated dose, in accordance with the usual practice of the attending physician. As far as possible, patients continued to use the same formulation and dose of metformin throughout the study. There was no restriction on concomitant therapies, except for other anti-diabetes medications, weight-loss agents, and glucocorticoids.
Patients who experienced inadequate A1C control, defined as A1C concentration >9.0 % (75 mmol/mol) at any time after the first 6 months of randomized treatment or >7.0 % (53 mmol/mol) at two consecutive study visits 3 months apart after the first 6 months of randomized treatment, were discontinued and offered alternative treatment.
Study assessments
Patients visited the clinic at baseline, Week 4, and at 3-month intervals after baseline until reaching the primary endpoint or study conclusion. A1C was measured every 3 months after baseline. Markers of cardiovascular risk included body weight (measured at baseline, Week 4, and every 3 months after baseline); waist circumference, BP, and heart rate (measured at baseline, Week 4, and every 6 months after baseline); and serum lipids (total cholesterol, low-density lipoprotein (LDL)-cholesterol, high-density lipoprotein (HDL)-cholesterol, and triglycerides) and hsCRP (measured at baseline and 1-year intervals after baseline). Safety assessments included data on hypoglycemic episodes (classified according to the American Diabetes Association recommendations [
8]) and concomitant medications collected at each study visit as part of routine collection in the trial.
BP was determined three times, seated and after a 5-min rest, and the mean of the second and third measurements was recorded. All serum lipid and hsCRP assays were carried out at a central laboratory (Interlab GmbH, Munich, Germany). A1C was determined by automated high-performance liquid chromatography (Tosoh Bioscience Inc., San Francisco, CA, USA). Serum concentrations of total cholesterol, HDL-cholesterol, and triglycerides were determined using automated enzymatic colorimetric methods (Roche Diagnostics GmbH, Mannheim, Germany), with LDL-cholesterol calculated using the Friedewald formula, and hsCRP measured by a high-sensitivity particle-enhanced turbidimetric assay (Roche Diagnostics GmbH).
Statistical analysis
The intent-to-treat (ITT) efficacy population included randomized patients who received at least one dose of study treatment and had baseline and at least one post-baseline measurement of A1C. Analyses of cardiovascular risk markers and safety measurements used an ITT safety population comprising all patients who received at least one dose of study drug. Continuous variables were analyzed by mixed model repeated measures (MMRM), with terms for treatment, study visit, and interactions, and including the baseline value as a covariate. Study visits with at least 25 % of randomized patients remaining were included in MMRM analyses, and least-squares (LS) means and 95 % confidence intervals were derived from the model for each time point. Differences in rates of hypoglycemia, as episodes per year, were analyzed between groups using a negative binomial model. Between-group differences in use of concomitant medications were analyzed using Fisher’s exact test.
Ad hoc analyses for differences between treatment groups in proportions of patients meeting therapeutic goals at their final visit in the study for waist circumference, systolic and diastolic BP (SBP and DBP, respectively), and lipid concentrations were examined using a logistic regression model; therapeutic goals were those recommended by the American Diabetes Association [
9].
Discussion
When first-line metformin is insufficient to maintain glycemic control in patients with type 2 diabetes, practitioners are faced with an array of options for second-line add-on therapy. Either a GLP-1RA, such as exenatide, or a sulfonylurea, such as glimepiride, are viable second-line therapy options, and primary data from the long-term EUREXA study indicate that add-on exenatide twice daily provides significantly greater durability of glycemic control than add-on glimepiride [
7]. Patients with type 2 diabetes are at an increased risk of cardiovascular morbidity and mortality [
1]. We report that add-on exenatide twice daily was also associated with greater improvement in several cardiovascular risk markers through 36 months than add-on glimepiride. Significantly greater improvements in body weight, waist circumference, SBP, DBP, HDL-cholesterol, triglycerides, and hsCRP were observed in exenatide- compared with glimepiride-treated patients, but changes in heart rate, total cholesterol, and LDL-cholesterol were similar between treatment groups. The improvements in cardiovascular risk markers observed with exenatide twice daily mirror improvements in body weight, BP, and lipids reported from long-term treatment with exenatide once weekly for up to 5 years [
10,
11].
Very few studies have compared the long-term efficacy and safety of GLP-1RAs with sulfonylureas for the improvement of glycemic control and cardiovascular markers. The Liraglutide Effect for Action in Diabetes (LEAD)-3 trial examined the effects of liraglutide 1.2 and 1.8 mg once daily compared with glimepiride 8 mg once daily for a 52-week controlled phase followed by a 52-week open-label extension [
12,
13]. Similar to long-term treatment with exenatide twice daily, liraglutide 1.2 and 1.8 mg treatment for 2 years was associated with significantly improved cardiovascular risk markers including body weight and waist circumference (
P ≤ 0.0002) [
13]. Decreases in SBP and DBP and increases in heart rate were not significant in patients treated with liraglutide, while exenatide was associated with significant decreases in SBP in the present study. Changes in lipids in patients treated with liraglutide were not reported in the original LEAD-3 study or in the 52-week extension [
12,
13]. However, a recent 12-month retrospective cohort study of 115 liraglutide-treated patients indicated a positive benefit on lipids, with significant reductions from baseline in total cholesterol, LDL-cholesterol, and triglycerides (
P < 0.05), and a significant increase in HDL-cholesterol (
P < 0.001) [
14].
Differences in body weight between the two treatment groups were consistent with previous studies. The finding of weight loss with add-on exenatide twice daily in the long-term EUREXA study is supported by long-term data from two open-label extensions of clinical trials where sustained and progressive weight loss (mean change from baseline, −4.7 to −5.3 kg) up to 2–3 years’ duration were observed [
15,
16]. The finding of weight gain with glimepiride treatment is also not unexpected. Data from two long-term studies observed body weight gain (mean change from baseline, 1.3 kg) with glimepiride when compared with two dipeptidyl peptidase-4 inhibitors [
17,
18].
Multiple studies have examined changes in waist circumference in patients treated with add-on exenatide or glimepiride but few inform on long-term changes. Over 1 year, exenatide therapy added on to baseline metformin was associated with a significant mean reduction from baseline in waist circumference (−5 %) [
19]. In a retrospective study of 179 patients treated with liraglutide or glimepiride in combination with metformin, liraglutide-treated patients had a significantly greater mean reduction in waist circumference from baseline to 18 months compared with glimepiride (−3.0 vs 0.0 cm;
P < 0.001) [
20]. One small randomized study comparing glimepiride and thiazolidinedione treatment added to metformin observed an increase in waist circumference from baseline of 1.86 cm with glimepiride [
21]. In the present study we have found a difference of −4.5 cm of waist circumference at 36 months between those patients treated with exenatide versus glimepiride. This finding may be largely explained by the observed changes in weight loss; however, there is also some evidence to suggest that the positive benefits of GLP-1RA therapy on waist circumference may be related to increased serum concentrations of atrial natriuretic peptide and B-type natriuretic peptide [
22].
Increased levels of hsCRP have been shown to be related to features of the metabolic syndrome [
23] and incident type 2 diabetes [
24], as well as with increased total atheroma volume [
25] and elevated coronary heart disease mortality in patients with type 2 diabetes [
26]. In the present study, both medications were associated with decreases in hsCRP through 36 months, with a greater decrease observed in the exenatide group. These findings confirm previous preliminary results from retrospective analyses [
27], consistent with an association between exenatide and reduced hsCRP. It is unclear to date, however, whether this reduction in hsCRP levels is the result of the observed reversal in features of the metabolic syndrome or of a drug-specific, anti-inflammatory effect of exenatide.
Reducing BP is important in patients with type 2 diabetes to reduce risk of macrovascular and microvascular complications [
28,
29]. Both SBP and DBP were reduced from baseline and maintained during 36 months of add-on exenatide therapy. The mechanism of BP reduction in patients treated with exenatide has not been clearly delineated. However, it appears that the short-term dynamics (arterial vasodilation) of BP is possibly related to the concomitant effects of reduced A1C and body weight with exenatide treatment [
30,
31].
Recent studies are conflicting about the association of weight changes with improvements in BP. In the Action for Health in Diabetes (Look AHEAD) study, 1-year weight loss was found to be associated with significant improvements in SBP through 4 years [
32]. When looking at individual categories of weight loss in the Practice-based Opportunities for Weight Reduction (POWER-UP) trial, patients with weight loss of 10 % or more from baseline had smaller mean improvements in SBP and DBP compared with patients with no change in weight or weight gain through 24 months [
33]. At least one analysis of eight clinical trials evaluating exenatide once-weekly treatment in patients with type 2 diabetes showed greater decreases in SBP in patient quartiles with the greatest body weight loss from baseline; similar observations were also noted for reductions in LDL-cholesterol, total cholesterol, and triglycerides [
34]. In a retrospective analysis evaluating liraglutide treatment over 7.5 months, significant reductions in SBP from baseline (~7 mmHg;
P < 0.001) were observed, but the reductions were not found to be associated with body weight loss (
r = 0.22;
P = 0.24) [
35].
BP target achievement was low overall considering that 65 to 70 % of patients were taking antihypertensive treatments at baseline. Less than half of all patients achieved the SBP (<130 mmHg) and DBP (<80 mmHg) targets [
9]. One possible explanation for the low rates observed is that patients with type 2 diabetes mellitus are generally regarded as a population that is difficult to treat to guideline-recommended BP goals [
36,
37]. Similar SBP goal achievement rates were observed across several clinical trials in an analysis of exenatide once weekly, although an incremental benefit was observed with exenatide compared with other oral glucose-lowering therapies [
38]. Another explanation relates to the study design; there was no protocol requirement to standardize concomitant antihypertensive medications. Despite randomization, there was a disparity in the proportions of patients taking concomitant medications at baseline. It is interesting that BP reductions were greater in the exenatide twice-daily group despite less use of antihypertensive medications at baseline and a significantly lower proportion of patients receiving at least one new antihypertensive medication during the study.
Heart rate was decreased by less than 1 beat per minute in patients treated with exenatide twice daily and the lack of change in heart rate is not unexpected. In a pooled analysis of studies comparing exenatide once weekly with other glucose-lowering medications, early increases in heart rate were observed with all GLP-1RA therapies [
39]. At the endpoint observation, a clinically insignificant mean increase in heart rate of 0.9 beats per minute was observed in patients treated with exenatide twice daily, compared with increases of 2.8 and 2.6 beats per minute for the longer-acting exenatide once weekly and liraglutide once daily, respectively. Inhibited neurotransmission of vagal neurons, increased sympathetic activity, and endothelium-induced vasodilation are all proposed mechanisms for heart rate increases that are observed with GLP-1RA therapy [
40]. It is possible that shorter-acting exenatide twice daily has less impact on heart rate than longer-acting once-daily liraglutide or once-weekly exenatide because exenatide was present for short periods of time during the day and may not have been present when heart rate was measured [
41,
42].
Total cholesterol and LDL-cholesterol increased from baseline in both treatment groups during 36 months of add-on therapy with no significant between-group differences observed, while a decrease in triglycerides and an increase in HDL-cholesterol were observed in patients treated with add-on exenatide twice daily. At baseline, the mean values for LDL- (2.5 mmol/L) and HDL-cholesterol (1.3 mmol/L) in the study population were already within recommended targets [LDL-cholesterol <2.6 mmol/L; HDL-cholesterol >1.1 (male) and >1.3 (female) mmol/L]. More than half of randomized patients were taking lipid-lowering agents at baseline. As was seen with antihypertensive medication, fewer patients on exenatide received additional lipid-lowering medication during the course of the study. At least one long-term observational follow-up study of 82 weeks’ duration showed significant improvements from baseline in mean HDL-cholesterol and triglycerides with exenatide twice-daily therapy; improvements in total cholesterol and LDL-cholesterol were not found to be significant [
43]. There is a lack of long-term data on the effects of glimepiride on lipid profiles. At least one short-term study of glimepiride added on to metformin showed that glimepiride had a neutral effect on lipids overall [
44].
Despite the observed differences in biomarkers, no information is yet available on any differences in cardiovascular events between these therapies. Large retrospective database analyses and meta-analyses of randomized controlled trials have suggested either decreased rates or no increased risk of cardiovascular events in patients treated with GLP-1RAs [
45‐
47], despite evidence that GLP-1RA therapy improves lipid profiles and a wide variety of markers of endothelial and vascular health [
42,
48,
49]. Prospective clinical trials such as the ongoing Exenatide Study of Cardiovascular Event Lowering (EXSCEL) (ClinicalTrials.gov identifier: NCT01144338) are required to confirm these findings. However, some data suggest that sulfonylureas have negative effects on cardiovascular risk and mortality [
50‐
54], so a comparative cardiovascular outcomes trial between exenatide twice daily and a sulfonylurea would provide interesting results.
Limitations of this study include the open-label design, which could have led to bias due to expectations of effects on cardiovascular markers. Furthermore, the study design of EUREXA was such that patients who did not maintain glycemic control exited the study over time, thus responders may have been selected for.
Authors’ contributions
RS, BG, GS, BrG, JR-G, FD, AF, and JK were principal investigators for the primary study, contributed to the conception and design of the analysis, interpretation of data, and drafting of the manuscript, including critical revisions of subsequent versions. MZ contributed to the conception and design of the analysis, interpretation of data, and critical revisions of the manuscript. All authors agree to be accountable for the work. All authors read and approved the final manuscript.