Introduction

Based on accumulated clinical evidence and demonstration of safety and efficacy, metformin has become recognized as an established glucose-lowering agent for the initial treatment of type 2 diabetes (T2D) [1]. In the latest European Society of Cardiology (ESC) Guidelines on diabetes, pre-diabetes, and cardiovascular diseases developed in collaboration with the European Association for the Study of Diabetes (EASD), it was also recommended that metformin should be considered as first-line therapy in patients with T2D, especially in overweight patients without cardiovascular disease (CVD) and those with a moderate cardiovascular risk [2]. Recent cardiovascular outcome trials (CVOTs) on several classes of glucose-lowering agents, including sodium-glucose cotransporter 2 (SGLT2) inhibitors and glucagon-like peptide-1 receptor agonists, have shown that these agents reduced cardiovascular events and death in patients with T2D at high cardiovascular risk or with established CVD. This led to a recent groundbreaking revision of treatment guidelines to use either agents in these patients, independent of baseline use of metformin [1, 2]. Given the previous position of metformin in diabetes care, this was likely a marked paradigm shift in the selection of glucose-lowering agents.

In the CVOTs completed before the development of such a paradigm shift, almost all participants (i.e., more than two-thirds) received baseline metformin therapy in accordance with conventional treatment strategy for diabetes care and had the agents investigated in this study added to metformin therapy [3,4,5,6,7,8,9,10,11,12]. This raised the clinical question as to whether or not the cardiovascular benefits of the agents would be expected even in patients with T2D at high cardiovascular risk or those with CVD, independent of the effects of baseline metformin. However, subsequent analyses demonstrated that treatment with these agents consistently improved cardiovascular outcomes regardless of the baseline use of metformin [13,14,15,16,17]. These findings also support recent updated treatment guidelines for better outcomes especially in patients with T2D either at high cardiovascular risk or with CVD. However, only limited clinical data are currently available regarding differences in the effect of these agents on glycemic and non-glycemic parameters in this patient population. Identifying these differences may help to achieve optimal selection of glucose-lowering agents for secondary prevention of CVD in actual clinical settings.

The EMBLEM (Effect of Empagliflozin on Endothelial Function in Cardiovascular High Risk Diabetes Mellitus: Multi-Center Placebo-Controlled Double-Blind Randomized) trial had the primary aim of determining whether 24 weeks of empagliflozin treatment affected peripheral endothelial function in patients with T2D and established CVD [18, 19]. In that trial, about one-half of the participants did not receive metformin therapy at baseline. Therefore, the current post hoc analysis of the EMBLEM trial examined whether the effect of 24 weeks of empagliflozin treatment on glycemic and non-glycemic clinical parameters in patients with T2D and CVD differed according to the use of baseline metformin treatment.

Methods

Study design

The EMBLEM trial (UMIN000024502) was an investigator-initiated, multi-center, placebo-controlled, double-blinded, randomized-controlled trial undertaken in 16 centers in Japan. The details of the study design and primary results have been reported previously [18,19,20]. Briefly, eligible patients were ≥ 20 years old, with T2D, and a glycohemoglobin (HbA1c) level ranging from 6.0 and 10.0%, who were clinically stable without changes in T2D therapy for at least one month before consent, and had a previous history of at least one established CVD (coronary artery disease, stroke, peripheral artery disease, presence of known coronary artery stenosis (≥ 50%), or heart failure (HF) with a New York Heart Association classification I to III and clinically stable by the use of HF-medications for at least one month before consent). Key exclusion criteria were type 1 diabetes, a history of diabetic ketoacidosis or diabetic coma within the last 6 months, severe renal dysfunction (estimated glomerular filtration rate (eGFR) < 45 mL/min/1.73 m2 or undergoing dialysis), serious liver dysfunction, a history of atherosclerotic CVD within 3 months before consent, and prior use of a SGLT2 inhibitor within one month before consent.

The participants were assigned randomly to either 10 mg of daily empagliflozin or to placebo, using a web-based minimization system that balances for HbA1c (< 7.0 or ≥ 7.0%), age (< 65 or ≥ 65 years), systolic blood pressure (BP) (< 140 or ≥ 140 mmHg), and current smoking status at the time of screening. The participants underwent scheduled visits after 4, 12, and 24 weeks for dispensing of drugs and assessment of study endpoints. Although no specific goal of glycemic control was set in the EMBLEM trial, all participants were to be treated in accordance with the local treatment guidelines for T2D of the Japan Diabetes Society at that time. Each participant’s background medications, including glucose-lowering therapy, were in principle unchanged during the trial. However, if the therapeutic effect of the medications was insufficient, the addition of glucose-lowering agents other than SGLT2 inhibitors or an increased dosage of background medications were allowed at the judgement of the local investigator.

The ethical committees of the participating institutions approved the study protocol. Written, informed consent for participation in the study was obtained from all the subjects. This study was performed in accordance with the Helsinki Declaration of 1964 and its later amendments.

Outcome measures

The details of the original outcome measures in the EMBLEM trial have been reported previously [20]. The main endpoints in the post hoc analysis using data obtained from the trial were changes from baseline to week 24 in glycemic parameters (fasting plasma glucose [FPG], HbA1c, and glycoalbumin [GA]) and non-glycemic parameters, including systolic and diastolic BP, heart rate (HR), double product (systolic BP × HR), body mass index (BMI), N-terminal pro-brain natriuretic peptide (NT-proBNP), and high-sensitivity Troponin-I (hs-TnI). Of the laboratory markers, the assays of GA (SRL, Inc., Tokyo, Japan), NT-proBNP (SRL, Inc., Tokyo, Japan), and hs-TnI (Abbott Japan LLC, Tokyo, Japan) were performed at central laboratories. The post hoc analysis assessed the effect of empagliflozin on these variables according to the use or nonuse of baseline metformin.

Statistical analysis

All the analyses were conducted on the full analysis set, which included all participants who had received at least one dose of the study medication after randomization and who did not have any serious violation of the protocol. Baseline demographics and characteristics were expressed as numbers (percentages) for categorical variables and as means ± standard deviation for continuous variables. Data on NT-proBNP and hs-TnI were expressed as geometric mean (95% confidence interval [CI]), and the proportional changes from baseline to week 24 calculated based on a natural logarithmic scale. Inter-group differences and ratios were compared using Welch’s t tests for continuous variables or Fisher’s exact test for categorical variables. All statistical analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC, USA). A two-sided significance level of p < 0.05 was used for all assessments, with no adjustment for multiplicity being used in the analyses.

Results

Baseline characteristics

A detailed participant flow-chart and the overall baseline characteristics of the EMBLEM trial have been reported elsewhere [18, 19]. Briefly, of the 117 patients randomized, 105 patients (64.8 ± 10.4 years old; women 33 [31.4%]; HbA1c 7.2 ± 0.8%; diabetes duration 13.3 ± 11.1 years) were included in the full analysis set (empagliflozin group n = 52, placebo group n = 53). Overall, 69 (65.7%), 18 (17.1%), and 79 (75.2%) patients were taking an angiotensin-converting enzyme or angiotensin receptor blocker, diuretic, and statin therapy, respectively. The majority of patients (69.5%) were taking a dipeptidyl peptidase-4 (DPP-4) inhibitor. Twenty-four patients (22.9%) were taking one type of glucose-lowering medication, while 67 patients (63.8%) were taking ≥ 2 types of these medications.

At baseline, a total of 53 (50.5%) patients were receiving metformin therapy, with 25 (48.1%) in the empagliflozin group and 28 (52.8%) in the placebo group. As shown in Table 1, the baseline characteristics were almost similar between the randomization groups, irrespective of the use of metformin at baseline.

Table 1 Baseline characteristics stratified by the use of baseline metformin
Full size table

Changes in clinical parameters according to baseline metformin use

Changes from baseline to week 24 in clinical parameters according to the baseline use of metformin and randomization group are shown in Table 2. In patients receiving baseline metformin, the magnitude of reduction in the levels of systolic BP, double product, BMI, HbA1c, and GA were greater in patients treated with 24 weeks of empagliflozin, compared to those on placebo. In contrast, examination of the magnitude of changes in clinical parameters in metformin-naïve patients at baseline showed these were similar in the randomization groups, except for FPG. Irrespective of the use of baseline metformin, no significant randomization-based group ratios of proportional changes in NT-proBNP (metformin users, 1.11 [95% CI 0.78 to 1.58], p = 0.551; nonusers, 1.11 [95% CI 0.76 to 1.64], p = 0.579) and hs-TnI (metformin users, 0.93 [95% CI 0.74 to 1.16], p = 0.490; nonusers, 0.96 [95% CI 0.73 to 1.27], p = 0.788) were observed between the randomization groups.

Table 2 Changes from baseline to week 24 in clinical parameters, grouped according to baseline metformin use
Full size table

Effect of empagliflozin in metformin users and nonusers

In the group treated with empagliflozin, the changes in systolic BP and double product from baseline to week 24 were greater in patients receiving baseline metformin therapy than those in metformin-naïve patients. However, these differences were not statistically significant (Fig. 1A, D). The reductions in BMI were significantly greater in patients who received baseline metformin therapy, compared to those who did not (Fig. 1E). No significant differences were observed between baseline metformin users and nonusers for the effect of empagliflozin treatment on the other parameters; HR, diastolic BP, and glycemic parameters (FPG, HbA1c, and GA) (Fig. 1B, C, F–H). The effect of empagliflozin on NT-proBNP concentration was also similar between the baseline metformin users and nonusers (Fig. 1I), while the group ratio (baseline metformin users vs. nonusers) of proportional changes in the geometric mean of hs-TnI was 0.74 (95% CI 0.59 to 0.92, p = 0.009: Fig. 1J).

Fig. 1
figure 1

Change in clinical parameters from baseline to week 24 in patients treated with empagliflozin according to baseline use of metformin. Blue (red) indicates the group with (without) baseline metformin therapy. Values located at the bottom of each panel indicate the mean group difference (95% confidence interval) in the magnitude of change from baseline to week 24 in systolic blood pressure (A), diastolic blood pressure (B), heart rate (C), double product (systolic blood pressure × heart rate: D body mass index (E), fasting plasma glucose (F), glycohemoglobin (G), and glycoalbumin (H) or the mean group ratio (95% confidence interval) of the change ratio from baseline to 24 weeks in the geometric means of NT-proBNP (I) and hs-TnI (J). hs-TnI, high-sensitivity Troponin-I; NT-proBNP, N-terminal pro-brain natriuretic peptide

Full size image

Discussion

This post hoc analysis of the EMBLEM trial on patients with T2D and established CVD, showed that the effects of 24 weeks of empagliflozin treatment on non-glycemic parameters, such as BMI and hs-TnI level, were more apparent in patients who had received baseline metformin therapy compared to metformin-naïve patients. In contrast, the impact of empagliflozin treatment on standard glycemic parameters was similar in patients who had received baseline metformin therapy to those who had not. To our knowledge, this is the first study in patients with T2D and CVD that assessed whether the effects of empagliflozin on clinical parameters differ according to the use of baseline metformin. Given the recent recommendation for SGLT2 inhibitors to be considered as key drugs for improving outcomes in patients with T2D at high cardiovascular risk or with CVD, regardless of previous metformin use [1, 2], the findings of our study may provide clinicians with novel insights on the effects of empagliflozin on a range of clinical parameters used frequently in the daily care of diabetes and cardiovascular disease, according to the patient’s prior use of metformin.

In the landmark trial of the United Kingdom Prospective Diabetes Study (UKPDS), metformin, relative to diet alone or other glucose-lowering agents such as sulfonylureas and insulin, was shown to reduce the risk of cardiovascular events and mortality in obese patients with newly diagnosed T2D [21]. The study also demonstrated that this early treatment effect was maintained, as a legacy effect, for as long as 10 years after the study intervention [22]. Based on its high degree of safety and efficacy and low-cost, metformin has been recommended over two decades as the first-line glucose-lowering agent for treating T2D. However, this recommendation was made before the use of modern and established cardiovascular protective drugs, such as statins and renin-angiotensin aldosterone system inhibitors. Accordingly, the majority of participants in earlier and recent CVOTs on newer glucose-lowering agents, including SGLT2 inhibitors, were administered metformin at baseline. Nonetheless, previous meta-analyses have reported heterogeneous and controversial effects of metformin on CVD and mortality, due mainly to a lack of good evidence obtained using current standards of care and trial methodology [23,36]. These results might indicate that the baseline use of metformin helps to enhance and retain the immediate BW-reducing effect of SGLT2 inhibitors in the present study. Furthermore, combination therapy of a SGLT2 inhibitor and metformin might have efficiently promoted BW loss through intrinsic insulin saving [37, 38], selective fat mass reduction [39, 40], and mitigation of compensatory overeating induced by chronic administration of SGLT2 inhibitors [41, 42]. Collectively, this combination therapy may provide synergistic benefits in some non-glycemic parameters, and as a consequence deliver a comprehensive therapeutic approach for diabetes-related complications, beyond that provided by lowering of glucose levels.

Interestingly, our analysis of established markers of cardiac stress and damage showed that the baseline levels of NT-proBNP and hs-TnI in patients who had received baseline metformin appeared to be lower than those in nonusers, although this difference was not statistically significant. The prevalence of a previous history of HF was similar between patients with or without baseline metformin therapy. This finding that the use of baseline metformin therapy might be associated with lower levels of these cardiac markers may, in part, support the possibility that metformin has robust cardioprotective effects via multifaceted molecular mechanisms, such as AMP-activated protein kinase-dependent and -independent signaling pathways [43]. In fact, the latest guidelines still recommend metformin therapy should be continued or included in patients with T2D and HF [1, 2]. Furthermore, empagliflozin treatment was associated with significantly greater reductions in hs-TnI concentrations in patients with baseline metformin therapy, compared to that observed in metformin-naïve patients. This finding raises the possibility that the protective effect of this combination therapy on micro-myocardial damage may, at least in part, be enhanced by several cardiometabolic actions, including a modulation of impaired cardiac insulin signaling [44]. However, the clinical impact of SGLT2 inhibitors on those cardiac biomarkers remains controversial [45]. Further research is therefore warranted in order to better understand which markers are most suitable for clinical monitoring of the cardiovascular effects of SGLT2 inhibitors.

The present study had several limitations. First, it was a post hoc analysis that used data obtained from the EMBLEM trial that was designed primarily to assess the effect of empagliflozin treatment, relative to placebo, on peripheral endothelial function. Second, although about one-half of the participants in the study were metformin-naïve at baseline, the clinical reasons why these patients did not receive metformin are unknown. Hence, some confounding factors with metformin use, such as renal function and severity of HF may have partially affected the impact of empagliflozin treatment on our study endpoints in the patients with or without baseline metformin therapy. Third, the small number of patients in the subgroups may not have provided sufficient statistical power to detect true group differences in the clinical parameters examined in the study. Furthermore, adjustments of potential confounding factors at baseline and changes in the glucose-lowering agents during the trial were not carried out due to the small sample size and limited clinical information. Finally, because the EMBLEM trial included only Japanese patients with T2D and CVD, who were clinically stable and met the study inclusion and exclusion criteria, further research is needed to assess whether the present findings are applicable to other ethnicities and/or different patient populations.

Conclusion

This study in patients with T2D and CVD demonstrated that 24 weeks of empagliflozin treatment was associated with an improvement in glycemic control, irrespective of baseline use of metformin therapy. The effects of empagliflozin on reductions in BMI and hs-TnI were more apparent in patients who received baseline metformin, compared to that observed in metformin-naïve patients. Based on the most recent guidelines for T2D, especially for patients with increased cardiorenal risk, the clinical opportunity to prescribe SGLT2 inhibitors will likely increase, and further research is therefore needed to investigate the effect of these drugs on clinical parameters, taking into account the background situation of conventional glucose-lowering agents, such as metformin.