FormalPara Key Summary Points

Why carry out this study?

Empagliflozin is a sodium-glucose co-transporter 2 (SGLT2) inhibitor, originally developed for the treatment of adults with insufficiently controlled type 2 diabetes (T2D) as an adjunct to diet and exercise. Its use has further broadened to include patients with T2D and established cardiovascular disease to reduce the risk of death from cardiovascular disease and the treatment of patients with symptomatic chronic heart failure and chronic kidney disease.

While a large body of clinical trial data indicates that empagliflozin is well tolerated, several potential adverse events (AEs) remain of interest to clinicians.

The current meta-analysis provides an up-to-date evaluation of the safety of empagliflozin in a broad range of patients based on four large placebo-controlled clinical outcome trials, EMPA-REG OUTCOME, EMPEROR-Reduced, EMPEROR-Preserved, and EMPA-KIDNEY. This is the first meta-analysis of empagliflozin to include trials of all conditions for which empagliflozin is indicated.

This meta-analysis focused on patients receiving empagliflozin 10 mg and evaluated the occurrence of selected AEs in the overall population and in key subgroups of medical interest.

What has been learned from this study?

This analysis updates and extends the findings of previous pooled analyses and confirms current knowledge of the safety and tolerability of empagliflozin based on long-term data from 20,933 participants.

These findings were consistent across a broad range of participants, including patients with and without T2D, patients with established cardiovascular disease, patients with heart failure, and a wide range of patients with chronic kidney disease.

Introduction

Empagliflozin is a sodium-glucose co-transporter 2 (SGLT2) inhibitor that was initially developed for the improvement of glycemic control in patients with type 2 diabetes (T2D). Following further evaluation of empagliflozin in patients with T2D with established cardiovascular disease in the phase III cardiovascular outcome trial, EMPA-REG OUTCOME, a reduction in 3-point major adverse cardiovascular outcomes (3P-MACE) was demonstrated, primarily driven by a reduction in cardiovascular death [1]. More recently, three large clinical trials have demonstrated the benefits of empagliflozin in a broader patient population, irrespective of baseline cardiovascular risk or the presence of T2D [2,3,4]. These trials were completed across the spectrum of patients with heart failure (HF): Empagliflozin Outcome Trial in Patients with Chronic Heart Failure and a reduced ejection fraction (EMPEROR-Reduced) and a trial in patients with a preserved ejection fraction (EMPEROR-Preserved) [2, 4]. In addition, patients with chronic kidney disease (CKD) were specifically investigated (Study of Heart and Kidney Protection with Empagliflozin, EMPA-KIDNEY) [3]. In addition to using empagliflozin for adults with insufficiently controlled T2D, as an adjunct to diet and exercise, its use has further broadened to include the treatment of symptomatic chronic heart failure (HF) and CKD [5].

The safety profile of empagliflozin is well documented, based on the results of various phase I–III trials and the subsequent publication of several pooled analyses of these data [6,7,8,9,10]. The safety profile of a drug can be defined in terms of adverse reactions, which are undesired and harmful effects that result from the administration of the drug. Some of the commonly reported adverse reactions associated with SGLT2 inhibitors relate to their mode of action. Blockade of SGLT2 on the proximal tubules of the kidney induces the excretion of glucose and sodium in the urine, contributing to osmotic diuresis and a reduction in plasma volume [11]. These mechanisms may increase the likelihood of adverse reactions including genital mycotic infections, urinary tract infections (UTIs), and hypotension, although serious complications are infrequent [12]. Despite published safety data, several potential AEs remain of interest to clinicians [12, 13]. A comprehensive analysis of safety data for empagliflozin (10 mg or 25 mg once daily) in patients with T2D, published in 2020, showed that the risk of hypoglycemia was similar for empagliflozin and placebo, except when co-administered with insulin and/or a sulfonylurea [6]. Based on 16,480 patient-years of exposure to empagliflozin, the main adverse reactions included events consistent with genital infection for empagliflozin 10/25 mg versus placebo (3.54 vs. 0.95/100 patient-years, respectively). The incidence of events consistent with UTI was similar for the empagliflozin 10/25 mg and placebo groups (9.27 vs. 9.70/100 patient-years, respectively). The majority of events consistent with genital infection or UTI were non-serious, mild, or moderate in intensity and led to treatment discontinuation in < 1% of patients in the empagliflozin 10/25 mg and placebo groups. The frequency of events consistent with volume depletion was similar for empagliflozin 10/25 mg and placebo (3.1% vs. 3.0%, respectively); however, these events were slightly more frequent with empagliflozin 10/25 mg than placebo in older patients (aged 75 to < 85 years) (5.9% vs. 5.0%) and in patients receiving therapy with loop diuretics at baseline (9.8% vs. 7.4%). This analysis showed no increase in the incidences of safety topics of interest, including urinary tract carcinogenicity, renal impairment, Fournier’s gangrene, liver injury, pancreatitis, diabetic ketoacidosis, bone fractures, or lower limb amputation in the empagliflozin group versus placebo. A further meta-analysis evaluated the safety of empagliflozin in 15,081 patients with T2D and advanced CKD (defined as moderate to severe CKD [category G3–4]) [10]. No new safety concerns were identified in this high-risk population, and the overall rates of AEs were similar among patients who received empagliflozin and placebo, and across eGFR categories. In addition, rates of AEs of special interest were similar for patients who received empagliflozin versus placebo. An exception was the frequency of genital infections, which was higher in the empagliflozin 10/25 mg group compared with placebo (3.54 vs. 0.85/100 patient-years), with progressively lower incidence rates across CKD categories 3A (2.75/100 patient-years), 3B (1.78), and 4 (1.13), although the frequency of genital infections was greater for empagliflozin than placebo patients in all CKD categories.

The aim of the current meta-analysis was to provide an up-to-date evaluation of the safety of empagliflozin in a broad range of patients based on four large placebo-controlled clinical outcome trials, EMPA-REG OUTCOME, EMPEROR-Reduced, EMPEROR-Preserved, and EMPA-KIDNEY. Although the overall results and subgroup analyses have been published for each individual trial, each trial has limited power to quantify the effects of empagliflozin on safety outcomes. Therefore, this pooled analysis evaluated the effect of empagliflozin versus placebo on the occurrence of selected AEs in the overall population and in key subgroups of medical interest. The four included clinical trials studies encompass the group of patients with conditions for which empagliflozin is indicated. The long duration of empagliflozin exposure allows the assessment of events that occur with a low incidence. This analysis updates and extends the findings of previous pooled analyses of empagliflozin and is intentionally restricted to long-term outcome trials.

Methods

Participants

Data were pooled from four trials (Table 1) which included patients with T2D and high risk for cardiovascular events (EMPA-REG OUTCOME), patients with HF, with or without diabetes (EMPEROR-Reduced and EMPEROR-Preserved), and a broad population of patients with CKD, with or without diabetes (EMPA-KIDNEY). In EMPA-REG OUTCOME, patients were randomized to receive empagliflozin 10 mg or 25 mg or placebo in addition to standard of care. In the other three trials, only the 10 mg dose of empagliflozin was investigated. To avoid a potential indication-associated bias, this pooled analysis was restricted to empagliflozin 10 mg. Furthermore, since the safety profiles of empagliflozin 10 mg and 25 mg have already been shown to be similar in the EMPA-REG OUTCOME trial, as well as in the latest pooled safety analysis of empagliflozin [6, 14], inclusion of the 25 mg dose was not expected to add relevant further information to the analyses of the 10 mg dose. Therefore, only patients receiving empagliflozin 10 mg were included in the meta-analysis. In addition, patients with type 1 diabetes (10 patients from EMPEROR-Preserved and 68 patients from EMPA-KIDNEY) were excluded from the analyses to reflect the current indication.

Table 1 Overview of the four double-blind, randomized, placebo-controlled clinical trials included in the meta-analysis
Full size table

The ethics committee at each center approved the trials, and all patients provided written informed consent. All original trials were performed in accordance with the Declaration of Helsinki.

Assessments and Data Analyses

Safety and tolerability were assessed based on AEs reported in each trial. Since the safety profile of empagliflozin has been comprehensively evaluated in previous trials, the collection of safety data was streamlined in EMPA-KIDNEY. Therefore, the pooled analysis of the four currently available outcome trials was restricted to the information on AEs systematically collected and documented in EMPA-KIDNEY. The selection of safety endpoints and subgroups was based on the known risks for empagliflozin or safety topics for which the medical community has expressed interest for the class of SGLT2 inhibitors. If safety endpoints were not analyzed for the EMPEROR and EMPA-REG OUTCOME trials, corresponding safety endpoints were newly derived from individual patient data in these studies. For the calculation of eGFR values, the CKD-EPI formula was applied to individual patient data from EMPA-REG Outcome to be consistent with the calculations used in the other three trials.

AEs were coded according to preferred terms in the Medical Dictionary for Regulatory Activities (MedDRA) version 23.1. AEs were identified based on investigator-reported AEs using standardized MedDRA or customized BI MedDRA queries. For EMPA-KIDNEY, adjudicated events were considered if available. AEs of interest included serious UTI, serious pyelonephritis or urosepsis, serious genital infections, and serious acute kidney injury, severe hypoglycemia, ketoacidosis, serious volume depletion, bone fracture, and lower limb amputation. Non-serious AEs were not generally reported in EMPA-KIDNEY; therefore, only serious AEs were included in this meta-analysis, with the following exceptions: bone fractures, events leading to lower limb amputation, and severe hypoglycemia (defined as investigator-reported severe hypoglycemia—i.e., requiring assistance). Since lower limb amputations were not systematically recorded in EMPA-REG OUTCOME, this was assessed on the basis of a previous medical review of the AEs, concomitant therapy, and AE narratives. All AEs were analyzed overall and by T2D status and based on subgroups relevant to the AE of interest: age, sex, eGFR, HF, baseline blood pressure, peripheral artery disease at baseline, immunosuppressive therapy, and treatment at baseline (renin-angiotensin system inhibitors, diuretics, and antihypertensives). Although harmonized endpoints and subgroups were derived, the final databases of the individual studies were not modified, for example by adding or excluding data or by applying different inclusion or exclusion criteria. In general, the analyses were based on subgroup categorizations derived for the individual trials and corresponding trial data.

Analyses of AEs were based on participants who were dispensed trial medication. Treatment was evaluated as randomized. The safety analyses were based on the number of patients with AEs rather than the number of AEs. The AE analyses included data from the date of randomization to trial completion (date of last follow-up for safety) (intention-to-treat approach). Incidence rates for AEs were calculated per 100 patient-years. The 95% confidence interval (CI) for the incidence rate was based on the exact (Clopper-Pearson) CIs. Time at risk was defined as (date of onset of AE − randomization date + 1) for patients with event. Patients without event were considered at risk until trial completion. Pooled risk ratio estimates were calculated from the Cochran-Mantel-Haenszel procedure stratified by trial. The pooled risk difference and its 95% CI were calculated by pooling the risk differences of each trial across the trials. Studies were weighted according to their size (total number of patients). For the analysis of AEs, interpretation was based not only on the statistical test (indicated by the 95% CI) but also on overall knowledge of AEs commonly associated with empagliflozin. Formal statistical tests were not performed as the study was not powered to assess differences in safety findings or adjusted for multiple comparisons. A supplemental analysis was performed for patients who initiated chronic dialysis and continued trial medication while on dialysis. The occurrence of AEs, serious AEs, and AEs leading to treatment discontinuation was assessed from start of dialysis to trial completion.

Results

Patient Disposition, Exposure, and Baseline Characteristics

The analysis set included 20,933 patients, of whom 10,472 received empagliflozin 10 mg and 10,461 placebo. The median duration of follow-up was 2.1 years. The median exposure to the trial medication was 1.8 years. The total trial medication exposure was 19,727 patient-years in the empagliflozin group and 19,447 patient-years for placebo. Patient demographics and baseline characteristics were well matched between the empagliflozin and placebo groups (Table 2). In total, 61% of patients were aged > 65 years, 3.0% were aged ≥ 85 years, and 74% had renal disease at baseline.

Table 2 Patient demographics and baseline characteristics
Full size table

Overall Safety and Safety Topics of Interest

The percentages of patients with serious AEs, fatal AEs, and AEs leading to discontinuation were also similar for both groups (Table 3).

Table 3 Frequency of adverse events
Full size table

Serious UTI

The incidence of serious UTI was comparable between empagliflozin and placebo groups (0.78 vs. 0.76 events/100 patient-years, respectively; rate ratio, 1.03 [0.83, 1.27]) (Table 4). Female patients experienced a higher incidence of serious UTIs than male patients (Table 5). The event rates for serious UTIs were higher in female patients in the empagliflozin group versus placebo (rate ratio 1.33 [0.97, 1.82]) but not in male patients (rate ratio 0.81 [0.61, 1.09]). In the placebo group, patients with a history of HF also showed an increased incidence of serious UTI compared with individuals without HF (0.89 vs. 0.64 events/100 patient-years). A numerical increase was seen in the empagliflozin group versus placebo, with rate ratio of 1.18 (0.89, 1.56). In patients without a history of HF, the incidence of serious UTI was lower in those who received empagliflozin versus placebo (rate ratio 0.83 [0.60; 1.16]). There was no increased risk between empagliflozin and placebo for patients with T2D compared with patients without diabetes. The rates of serious UTI increased in the placebo group with decreasing eGFR; however, a similar pattern was not seen in the empagliflozin group. The event rates were comparable between empagliflozin and placebo groups across all eGFR categories in general.

Table 4 Incidence of selected AEs
Full size table
Table 5 Incidence of serious UTI and serious genital infection by subgroup
Full size table

Serious Pyelonephritis or Urosepsis

Similar rates of serious pyelonephritis or urosepsis were shown for empagliflozin versus placebo (0.28 vs. 0.25/100 patient-years, respectively; rate ratio, 1.09 [0.76, 1.56]) (Supplementary Material, Table S1). Subgroup analysis indicated an increased incidence with empagliflozin in women but not in men (rate ratios, 1.87 vs. 0.64 respectively). The risk of serious pyelonephritis or urosepsis was not consistently affected by empagliflozin; it was increased in the 75 to < 85 years age group for empagliflozin versus placebo (rate ratio 1.75 [0.93, 3.31]) but not in patients aged ≥ 85 years (rate ratio 0.90 [0.24, 3.36]).

Serious Genital Infection

The incidence of serious genital infection was not increased for patients who received empagliflozin versus placebo (rate ratio 0.62 [0.28, 1.37]) (Table 4). No risk increase for empagliflozin was indicated in any subgroup category (Supplementary Material, Table S1).

Serious Acute Kidney Injury

The risk of serious acute kidney injury was lower in the empagliflozin group versus placebo (0.73 [0.64, 0.83]) (Table 4). Although higher event rates were seen in patients with T2D, HF, worsening eGFR, and increasing age in both the empagliflozin and placebo groups, rates of serious acute kidney injury were consistently lower in the empagliflozin group versus placebo across all subgroups (Supplementary Material, Table S1).

Severe Hypoglycemia

The incidence of severe hypoglycemia was similar for patients in the empagliflozin and placebo groups (0.97 [0.78, 1.22]) (Table 4). No risk increase for empagliflozin was indicated in any subgroup category (Supplementary Material, Table S1).

Ketoacidosis

Ketoacidosis was reported in a small number of patients in the empagliflozin and placebo groups (15 vs. 7 patients; 0.07 events vs. 0.03 events/100 patient-years, respectively; rate ratio, 2.13 [0.87, 5.24]) (Table 6). Ketoacidosis occurred in only one patient without T2D in the empagliflozin group.

Table 6 Incidence of ketoacidosis and limb amputation by subgroup
Full size table

Serious Volume Depletion

Serious volume depletion was slightly more common in patients who received empagliflozin versus placebo (0.94 vs. 0.88 events/100 patient-years, respectively; rate ratio, 1.08 [0.88, 1.31]) (Table 4). The incidence of serious volume depletion increased with age. The highest incidence of volume depletion for empagliflozin versus placebo was among patients aged ≥ 85 years (15 vs. 6 patients; 2.56 vs. 1.00/100 patient-years; rate ratio, 2.39 [0.92, 6.21]) (Supplementary Material, Table S1).

Bone Fractures

The incidence of bone fracture was similar in the empagliflozin group versus placebo (2.03 vs. 1.93/100 patient-years; rate ratio, 1.05 [0.92, 1.20]) (Table 4). The subgroup analyses did not indicate an increased risk of bone fractures with empagliflozin (Supplementary Material, Table S1).

Lower Limb Amputation

The occurrence of lower limb amputation was similar among patients who received empagliflozin versus placebo (0.43 vs. 0.42 events/100 patient-years; rate ratio, 1.01 [0.76, 1.35]) (Table 4). For the subgroup of patients with body mass index (BMI) < 25 kg/m2, there was a numerical increase in the rate of lower limb amputations (rate ratio 2.12 [0.92, 4.89]) (Table 6). This finding is based on a relatively small number of patients with events (8 vs. 18 for placebo and empagliflozin, respectively) with consequently broad confidence intervals. This finding is likely related to the unusually high frequency of lower limb amputation in the empagliflozin 10 mg group with BMI < 25 kg/m2 of the EMPA-REG OUTCOME trial and is considered to be a chance finding. The analyses did not show a consistent trend across the four BMI categories evaluated (Table 6).

The remaining subgroup analyses do not indicate inconsistent findings among the individual subgroups.

Patients Who Initiated Chronic Dialysis

A total of 54 patients in the empagliflozin group and 83 who received placebo initiated chronic dialysis and continued with study treatment while on dialysis. Although this is not a randomized comparison, the results do not indicate an increase in AEs with empagliflozin regarding the frequency of investigator-defined drug-related AEs, AEs leading to treatment discontinuation, serious AEs, or fatal AEs in patients on dialysis (Table 7).

Table 7 Overall summary of AEs in patients on dialysis
Full size table

Discussion

This meta-analysis of pooled safety, based on more than 19,727 patient-years of exposure to empagliflozin, indicates no safety concerns with empagliflozin. This is the first meta-analysis of empagliflozin to include trials of all conditions for which empagliflozin is indicated and extends the findings of previous pooled analyses [6,7,8] by inclusion of a broad range of participants including patients with T2D and established cardiovascular disease, patients with HF, with or without diabetes, and a wide range of patients with CKD, with or without diabetes, enabling analyses in a range of subgroups. Since the assessment of non-serious (or not leading to treatment discontinuation) UTI and genital infections, acute kidney injury, and volume depletion was not included in the EMPA-KIDNEY trial, the present meta-analysis focused on serious AEs in these categories. However, data for non-serious AEs have already been published in patients with T2D [6] and HF [2, 4] and are in line with the results of the present analysis.

Consistent with previous experience [6,7,8], rates of serious UTI were higher in female patients compared with male patients, in both the placebo and empagliflozin groups. In addition, the rate of serious UTI was higher among women who received empagliflozin versus placebo. The incidence of serious pyelonephritis or urosepsis was not increased among male patients but there was a small increase for women who received empagliflozin versus placebo. With a low number of events overall, serious genital infections were not increased for empagliflozin versus placebo in any subgroup including sex.

The present meta-analysis showed that the occurrence of serious acute kidney injury was lower with empagliflozin therapy compared with placebo. This finding is consistent with the finding of a meta-analysis of SGLT2 inhibitors, which showed a 30% reduction in acute kidney injury among patients who received SGLT2 inhibitors [12]. A recent meta-analysis and systematic review of the effects of SGLT2 inhibition on kidney outcomes evaluated acute kidney injury as an efficacy outcome in terms of reduction in the number of patients with progression of kidney disease [9]. The meta-analysis showed a 23% reduction in acute kidney injury associated with SGLT2 inhibition, with similar reductions in patients with and without T2D. This outcome is considered to reflect the kidney-protective effects of empagliflozin [9].

Severe hypoglycemia (i.e., requiring assistance) is a concern for patients with T2D. The findings of the present meta-analysis indicate that this risk is not increased with empagliflozin. This finding is consistent with the results of previous pooled analyses of empagliflozin [6,7,8]. Since SGLT2 inhibitors reduce plasma glucose by increasing glucose excretion by the kidneys, an effect that is independent of insulin [15], the overall risk of hypoglycemia is low when these agents are used as monotherapy [16]. Diabetic ketoacidosis is a rare AE associated with SGLT2 inhibitor therapy that may occur with minimal increases in blood glucose [17]. The likelihood of ketoacidosis is increased among patients with established T2D who are being treated with insulin and can be potentiated by fasting, intercurrent illness, or surgical intervention. A large collaborative meta-analysis of available evidence from large SGLT2 inhibitor trials showed that patients without T2D are at especially low risk of ketoacidosis associated with SGLT2 inhibitor therapy [9]. This present meta-analysis showed a small absolute increase in the incidence of ketoacidosis among patients receiving empagliflozin versus placebo. Most of these events occurred in patients with T2D, with just one event in a patient without T2D who received empagliflozin.

Empagliflozin produces transient natriuresis and increases in urine volume [18], resulting in the potential for hypotension and volume depletion, particularly among elderly individuals [5]. The findings of this meta-analysis indicate that serious volume depletion was more frequent in the empagliflozin group compared with placebo, with an increased incidence with older age in the empagliflozin group (being greatest in patients aged ≥ 85 years).

Bone fracture has been an AE of interest following The Canagliflozin Cardiovascular Assessment Study (CANVAS), which reported an increase in bone fractures among patients who received canagliflozin versus placebo (15.4 vs. 11.9 participants with fracture/1000 patient-years; HR 1.26; 95% CI 1.04, 1.52) [19]. However, an increased risk of fractures was not observed in a subsequent trial of canagliflozin [20] and has not been observed with other SGLT2 inhibitors [12, 13]. Similarly, previous pooled analyses for empagliflozin have shown no association with treatment and an increased occurrence of bone fractures [6,7,8]. The present meta-analysis provides further support for these findings based on a median 2.1-year (110 week) follow-up period.

The CANVAS study also indicated a two-fold increase in risk of lower limb amputation (primarily of the metatarsal) among patients who received canagliflozin versus placebo [19]. As a result, the occurrence of amputations has specifically been evaluated in subsequent clinical trials of SGLT2 inhibitors. As with bone fractures, no increased risk of lower limb amputation was indicated in a subsequent trial of canagliflozin [20], with other SGLT2 inhibitors [12, 13], or in previous pooled analyses for empagliflozin [6,7,8]. These findings are further supported by the present meta-analysis.

Until recently, patients with CKD and eGFR < 30 ml/min/1.73 m2 and patients on dialysis have been underrepresented in trials of SGLT2 inhibitors. However, the EMPA-KIDNEY trial showed that empagliflozin was associated with preservation of kidney function across a wide range of eGFR levels, including levels as low as 20 ml/kg/1.73 m2, with no evidence of safety concerns. The present meta-analysis does not indicate that patients on dialysis are at increased risk of AEs if treated with empagliflozin. No new safety concerns were raised in the analyses of AEs by other subgroups, including age, BMI, presence of T2D, or concurrent HF. Thus, this meta-analysis does not indicate that empagliflozin is associated with an increased risk of AEs in subpopulations of interest, including patients with low eGFR, low BMI, elderly patients, and patients with or without comorbidities such as T2D and HF.

The strengths of this meta-analysis include the large number of participants and the long total exposure to treatment and follow-up period. Restricting the analyses to empagliflozin enabled an individual patient data approach and subgroup analysis to ensure the generation of high-quality data. This would not have been possible if a summary-based approach (aggregated data) had been used, as this would have required the availability of subgroup results across the trials. As with all meta-analyses, an important limitation of the present analysis is the inclusion of trials of different durations and methodologies. In addition, the analysis of lower limb amputations in the EMPA-REG OUTCOME study should be interpreted with caution in view of the manual retrieval and validation of these cases. Another limitation of the present meta-analysis is that only the AEs of interest included in EMPA-KIDNEY were analyzed. However, other AEs have been extensively analyzed in previous pooled analyses for empagliflozin [6,7,8]. Low patient numbers in some of the subgroups, including the group who started dialysis during the study, mean that the data collected were insufficient for generating conclusions and should, therefore, be interpreted with caution.

Conclusion

This meta-analysis of 20,933 participants of four large, randomized, placebo-controlled clinical trials of empagliflozin extends previous pooled safety analyses and confirms previous knowledge of the safety and tolerability of empagliflozin. These findings were consistent across a broad range of participants, including patients with and without T2D, patients with HF, and a wide range of patients with CKD, and various subgroups of specific interest. Furthermore, there was no evidence to indicate that patients on dialysis are at increased risk of AEs when treated with empagliflozin.