Cardiovascular Outcome in Patients Treated With SGLT2 Inhibitors for Heart Failure: A Meta-Analysis

Background: Sodium–glucose co-transporter 2 (SGLT2) inhibitors are an emerging class of glucose-lowering drugs that have become increasingly relevant for the treatment and prevention of heart failure (HF). Therefore, we aimed to investigate various SGLT2 inhibitors in patients with established HF at baseline and focused on the different types of HF. Methods: An extensive search of PubMed and Web of Science until January 2021 was done. Two reviewers, independently and in duplicate, applied the selection criteria. This meta-analysis was conducted according to the PRISMA guidelines. Data were pooled using a random-effects model. Randomized controlled trials (RCTs) of SGLT2 inhibitors vs. a comparator in patients with HF reporting clinical outcomes were included. The primary efficacy outcome was the composite of hospitalization for HF (HHF) or cardiovascular (CV) mortality. All-cause mortality, CV mortality, and HHF were considered as secondary endpoints. Subgroup analyses involving the status of diabetes, type of HF, administered type of SGLT2 inhibitor, sex, age, body mass index (BMI), estimated glomerular filtration rate (eGFR), cause of HF, and concomitant medication were performed. Results: Seventeen RCTs, comprising a total of 20,749 participants, were included (n = 10,848 treated with SGLT2 inhibitors and n = 9,901 treated with a comparator). Treatment with SGLT2 inhibitors in a HF population was associated with a 27% relative risk reduction (RRR) of HHF or CV mortality [risk ratio (RR) = 0.73, 95% CI = 0.68–0.78], 32% RRR of HHF (RR = 0.68, 95% CI = 0.62–074), 18% RRR of CV mortality (RR = 0.82, 95% CI = 0.73–0.91), and 17% RRR of all-cause mortality (RR = 0.83, 95% CI = 0.75–0.91). The effect of SGLT2 inhibitors on the primary endpoint was consistent among the different gliflozines. The effect of SGLT2 inhibitors on the primary endpoint was independent of underlying diabetes mellitus, age, sex, BMI, renal function, and HF type. Conclusions: SGLT2 inhibitors are associated with improved CV outcomes in patients with HF.

BACKGROUND Sodium-glucose co-transporter 2 (SGLT2) inhibitors are an arising drug class across antidiabetic therapeutics. During the last five years, large randomized trials have shown the cardioprotective effects of three SGLT2 inhibitorsempagliflozin, dapagliflozin, and canagliflozin-independently of the presence or absence of diabetes mellitus (DM) and within the first months after initiating the treatment (1)(2)(3)(4)(5). Recently, these benefits have also been demonstrated for sotagliflozin and ertugliflozin (6)(7)(8). The glucose-lowering effects of these agents are mediated through the inhibition of renal glucose reuptake in the proximal tubule of the nephron, which consequently leads to a decrease in blood glucose levels (9)(10)(11). However, it is assumed that the cardioprotective properties are of a different origin from the promoted urine glucose excretion. Although, a variety of hypotheses for SGLT2 inhibitor-induced benefits exist, the exact underlying mechanism is unclear (12). Aside from the impact on blood pressure and body weight, modulation of ion homeostasis and cellular processes are suggested (13). In particular, SGLT2 inhibitors have become progressively interesting for the treatment and prevention of heart failure (HF) (14), which has emerged as a global health issue. According to the most recent European Society of Cardiology (ESC) guidelines, HF is prevalent in 2% of the adult population and is more frequent in patients with atrial fibrillation (15, 16). Although, survival after onset of HF improved over the last 50 years (17), the 12-month all-cause mortality still occurs in about 7% of patients diagnosed with chronic HF and in 17% of patients suffering from acute HF (18)(19)(20)(21)(22).
Due to the promising results of the EMPA-REG OUTCOME, DECLARE-TIMI 58, CANVAS and CREDENCE trials regarding their cardiovascular (CV) outcomes in patient populations with and without HF, the next logical aim was to put focus on subgroups with an established HF at baseline, as in the DAPA-HF, EMPEROR-Reduced, and SOLOIST-WHF trials (1)(2)(3)(4)(5)(6). In this respect, we performed a systematic review and meta-analysis of randomized controlled trials (RCTs) that investigated SGLT2 inhibitors in patients with established HF regarding their clinical endpoints, with a particular focus on the type of HF.

METHODS
The following systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, as described previously (23)(24)(25)(26)(27)(28). We performed an extensive Abbreviations: SGLT2, sodium-glucose co-transporter 2; T2DM, type 2 diabetes mellitus; HF, heart failure; HHF, hospitalization for heart failure; CV, cardiovascular; RCT, randomized controlled trial; RR, relative risk; RRR, relative risk reduction; ARR, absolute risk reduction; EF, ejection fraction; HFpEF, HF with preserved ejection fraction; HFmrEF, HF with mid-range ejection fraction; HFrEF, HF with reduced ejection fraction; ARNI, angiotensin receptor neprilysin inhibitor; MRA, mineralocorticoid receptor antagonist; NNT, number needed to treat; CI, confidence interval; HR, hazard ratio; eGFR, estimated glomerular filtration rate; BMI, body mass index. search of PubMed and Web of Science, applying predefined search terms [(empagliflozin OR dapagliflozin OR canagliflozin OR ertugliflozin OR sotagliflozin) AND heart failure AND randomized controlled trial], until January 2021. The title and abstract of suspected relevant citations were screened for eligibility, and full-text was acquired for further evaluation if the citation was deemed pertinent. The references of the retrieved meta-analyses and reviews were also examined for additional trials.
All included studies had to be RCTs, regardless of sample size, and follow-up time, comparing SGLT2 inhibitors either to placebo or a comparator and evaluated clinical endpoints. The eligible patient population for our meta-analysis comprised patients with any diagnosed HF at baseline. Two reviewers (GMG and JMSM), independently and in duplicate, applied the selection criteria. The exclusion criteria were: non-RCTs, duplicate reports, ongoing studies, studies that included patients without HF, and studies that did not assess clinical endpoints.
The primary efficacy endpoint of our meta-analysis was a composite of CV mortality or hospitalization for HF (HHF). This primary composite endpoint of HHF or CV mortality was chosen based on a uniform definition of the primary endpoint used in the majority of HF trials, which was also used in the large RCTs included into our metaanalysis. CV mortality, all-cause mortality, and HHF were considered as our secondary endpoints. For further analysis of the data, we performed subgroup analyses involving the status of diabetes, type of HF, and type of SGLT2 inhibitors used.

Statistical Analysis
Variables are reported as number or percentages, as applicable. Risk ratios (RRs) were calculated from individual studies and pooled according to the inverse variance model with 95% confidence intervals (95% CI) and reported as relative risk reduction (RRR), as reported previously (28)(29)(30)(31)(32). The statistical inconsistency test (I 2 ) was used for the assessment of any heterogeneity between the studies. We used a randomeffects model for all analyses. The following sensitivity analyses were performed: (i) comparison of the results of fixed-vs. random-effects model; (ii) assessment of each study influence by successively deleting one by one to evaluate whether the pooled results of the meta-analysis change significantly; and (iii) introduction of the following subgroups: SGLT2 inhibitor type, DM vs. no DM, HF with reduced ejection fraction (HFrEF), HF with mid-range ejection fraction (HFmrEF), HF with preserved ejection fraction (HFpEF), sex, age, body mass index (BMI), estimated glomerular filtration rate (eGFR), cause of HF, and concomitant medication. We also calculated a number needed to treat (NNT) for the composite endpoint based on the mathematical formula: NNT = 1/absolute risk reduction (ARR). A two-tailed p-value of <0.05 was considered as significant. Review Manager (version 5.4; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration 2020) was used for statistical computations.

Outcomes
Primary Composite Outcome: Hospitalization for Heart Failure or Cardiovascular Death The composite outcome of HHF or CV death was regarded as our primary endpoint. Ten trials (2,4,6,8,34,35,(42)(43)(44)46) reported on the primary efficacy outcome. Overall, 17% of patients experienced HHF or CV death under treatment with SGLT2 inhibitors as compared to 23% in the control-group, resulting in a RRR of 27% (RR = 0.73, 95% CI = 0.68-0.78, p < 0.00001, I 2 = 0%) ( Figure 1A) and an ARR of 6%. This corresponds to a NNT of 17. For a group of 1,000 patients treated with SGLT2 inhibitors for HF for a mean time of 18 months, the composite endpoint of HHF or CV death could be prevented in 60 ( Table 2).

All-Cause Mortality
In patients treated with SGLT2 inhibitors, all-cause mortality was reported in 11% as compared to 13% in patients treated with placebo or a comparator. Consequently, the RRR to die from any cause was 17% (RR = 0.83, 95% CI = 0.75-0.91, p < 0.001, I 2 = 0%) ( Figure 1D) with an ARR of 2%, when assigned to SGLT2 inhibitors.
The differences between the individual subgroups were not deemed statistically significant (p = 0.69).

Hospitalization for Heart Failure
All five agents reached significant values for the prevention of HHF as compared to the control arm ( Figure 3B).

Subgroup Analyses for the Primary Composite Outcome According to Patients' Baseline Data
Sex Both men and women profited from treatment with SGLT2 inhibitors. As shown in Figure 5A, these benefits were even more pronounced in the female population, with a resulting RRR of 30% (RR = 0.70, 95% CI = 0.57-0.86, p = 0.0007, I 2 = 13%) as compared to a RRR of 26% in male patients (RR = 0.74, 95% CI = 0.65-0.84, p < 0.00001, I 2 = 25%).

BMI (kg/m 2 )
The magnitude of the effect for the prevention of HHF or CV mortality was similar in patients with a BMI <30 and a BMI ≥30

Period of Follow-Up (≤1.5 and >1.5 Years)
Based on the mean follow-up time of 18 months, we set the cutoff for the subgroup analysis at 1.5 years (≤1.5 and >1.5 years; Supplementary Figure 3). The shorter follow-up period (≤1.5 years) did not alter the beneficial effects negatively, which were observed under treatment with SGLT2 inhibitors, with a resulting RRR of 26% (RR = 0.74, 95% CI = 0.68-0.80, p < 0.00001, I 2 = 0%). In comparison, follow-up times longer than 1.5 years were associated with a RRR of 28% (RR = 0.72, 95% CI = 0.62-0.83, p < 0.00001, I 2 = 8%). Set side by side to the overall analysis for the composite endpoint with a RRR of 27%, the effects are of similar magnitude (RR = 0.73, 95% CI = 0.68-0.78, p < 0.00001,

Sensitivity Analysis
The results of the random and fixed effects were revealed to be similar (data not shown). Excluding single studies from the analyses did not change the direction and the magnitude of the effect. In patients who were treated with ARNIs, the use of sotagliflozin did not result in a reduction of the primary endpoint, but the use of dapagliflozin or empagliflozin was associated with a reduction of the primary endpoint.
The visual inspection of the funnel plot has shown minor asymmetry (data not shown).

DISCUSSION
The present meta-analysis investigated various RCTs of SGLT2 inhibitors regarding their clinical outcomes, with a particular focus on a patient population with established HF at baseline. Our results highlight that all SGLT2 inhibitors lower the risk of CV death, all-cause mortality, and HHF (1)(2)(3)5) in patients with underlying HF diagnosis. Hospitalization for HF emerged as the outcome, which was best reduced by SGLT2 inhibitors (RRR = 32%).
Previous meta-analyses have already reported on the outstanding ability of different SGLT2 inhibitors in decreasing the rates of hospitalization due to heart failure in participants with or without HF at baseline (50,51). In our analysis, we put focus on the different types of HF, including HFrEF (EF ≤40% and ≤45%), HFmrEF (EF 40-50%), HFpEF (EF >45% and >50%), and acute HF. We also included a subgroup of HF with unknown EF due to the fact that some studies did not classify between the different forms of HF, such as the EMPA-REG OUTCOME, CANVAS, CREDENCE, or VERTIS-CV post-hoc analysis (44)(45)(46)49). Furthermore, some studies solely focused on patients with HFrEF [EMPEROR-Reduced (34), DAPA-HF trial (2), and DEFINE-HF trial (39)] or used different left ventricular ejection fraction (LVEF) thresholds to distinguish between HFrEF and HFpEF [DECLARE-TIMI 58 post-hoc analysis, EF     < 45% (43); CANDLE, SOLOIST WHF, and SCORED trial, EF < 50% (6,8,41)]. In addition, the CANDLE trial did not assess the clinical safety endpoints with regard to the baseline LVEF (41). The reduction of HHF or CV mortality was of similar magnitude in most groups. Interestingly, our analysis show that SGLT2 inhibitors are also effective in HFmrEF and HFpEF, which is a novel finding as there is currently no effective therapy in this patient population. Nevertheless, the majority of the data for the HFmrEF and HFpEF groups were based on trials using sotagliflozin. Therefore, the question remains as to whether SGLT2 inhibitors display a class effect in this unique patient population.
The standard treatment approach for HFrEF currently includes angiotensin-converting enzyme (ACE) inhibitors, MRAs, and beta-blockers. ARNIs, I f -channel inhibitors, and diuretics are also often administered in HFrEF. To date, no pharmacotherapy ameliorating the prognosis in patients with HFpEF has been proven; therefore, treatment of the symptoms and comorbidities is the current approach (52)(53)(54). Although, some data indicate SGLT2 inhibitors as a sufficient therapy option for patients with HFpEF, results of large outcome trials, which investigate particularly participants with HFpEF, such as the EMPEROR-Preserved trial (Empagliflozin Outcome Trial in Patients with Chronic Heart Failure with Preserved Ejection Fraction; NCT03057951) or the DELIVER trial (Dapagliflozin Evaluation to Improve the Lives of Patients with Preserved Ejection Fraction Heart Failure; NCT03619213), are ultimately required to confirm these assumptions (55,56).
Heart failure is highly associated with comorbidities, which directly affect patients' mortality and morbidity. A survey from 2014 demonstrated that 74% of 3,226 patients with chronic HF suffer at least from one comorbidity, the most frequent being chronic kidney disease, anemia, and diabetes (57). The link between HF and diabetes has been reported previously (58), making medication treating both conditions even more appealing. Interestingly, a post-hoc analysis of the DAPA-HF trial (2) showed that dapagliflozin acted comparably efficiently in reducing the composite endpoint in individuals with or without diabetes (HR = 0.75 vs. 0.73, p-value for interaction = 0.83). According to these results, one can assume that dapagliflozin is applicable to patients with HFrEF, irrespective of their glycemic status (34). Of note is that dapagliflozin indeed reduced the glycated hemoglobin (HbA1c) in people with diabetes; however, it had no effect on HbA1c in patients without diagnosed diabetes (47). The safety and tolerability of dapagliflozin in people without diabetes were also suggested in the DEFINE-HF trial; nevertheless, this study was not powered for clinical outcomes, including a small sample size of only 263 patients (63% of which had T2DM) and a short followup period of only 12 weeks (39). The very recent EMPEROR-Reduced trial also demonstrated a similar efficacy in reducing the composite endpoints of CV mortality and HHF in the diabetic and non-diabetic subpopulations (HR = 0.72 vs. 0.78), leading to the assumption that empagliflozin is also beneficial in people with HFrEF independent of the presence of diabetes (34). Our meta-analysis further confirmed the similar efficacy profiles of canagliflozin, dapagliflozin, empagliflozin, and sotagliflozin for their composite endpoint, regardless of the diabetes status.
Cardiovascular mortality and all-cause mortality were also reduced in patients who were administered SGLT2 inhibitors as compared to the control group, with a RRR of 17-18%. Interestingly, when analyzed with regard to the type of SGLT2 inhibitor used (empagliflozin, dapagliflozin, canagliflozin, and sotagliflozin), the magnitude of the effect varied, which might be due to the differences in the studied populations. Interestingly, dapagliflozin reduced CV mortality in our meta-analysis in HF patients significantly, whereas, empagliflozin, canagliflozin, and sotagliflozin did not reach statistical significance. Importantly, a recent meta-analysis that investigated SGLT2 inhibitors for CV outcomes in patients with T2DM independently of HF diagnosis demonstrated a superiority of empagliflozin as compared to dapagliflozin and canagliflozin in preventing all-cause and CV mortality (59). However, these differences might be based on the differing study populations, with our analysis particularly focusing on patients with HF at baseline rather than on the diabetes status. Further trials addressing the question of superiority within the SGLT2 inhibitor class are warranted.
We would like to put a special emphasis on the very low NNT of 17 for the composite outcome, which we have shown for the SGLT2 inhibitors in our meta-analysis. For comparison, in 2014, the PARADIGM-HF trial demonstrated a NNT of 32 for the prevention of CV mortality in patients with HF when treated with ARNIs, which are nowadays indispensable for treating HF (60). Of note is that our subgroup analysis showed no benefit for the concomitant use of SGLT2 inhibitors and ARNIs. However, that outcome was solely generated by the missing advantage of sotagliflozin use in addition to ARNI treatment in the SOLOIST-WHF and SCORED trials (6,8), while the combination of dapagliflozin or empagliflozin with ARNIs resulted in beneficial effects, as demonstrated in a recent meta-analysis (61). Whether sotagliflozin constitutes as just an exception or whether these results were based on differences in the study population remains unclear. However, current data support the combined administration of these drug classes in patients with HF (62).
Hence, once more, our current meta-analysis emerges as highly relevant due to further confirming and demonstrating the outstanding properties of SGLT2 inhibitors in HF management in patients already on standard-of-care treatment.

STRENGTHS OF THIS META-ANALYSIS AND DIFFERENCES AS COMPARED TO OTHER META-ANALYSES
The most distinctive difference between our meta-analysis and other recent ones (63,64) is that this is the first meta-analysis investigating all current RCTs on SGLT2 inhibitors on HF with clinical endpoints, irrespective of the size of the trial. Furthermore, we included all available data of post-hoc analyses from large trials to elaborate the safety and efficacy of SGLT2 inhibitors in patients with established HF at baseline, irrespective of their diabetes status. In addition we also focused on subgroups, especially the different types of HF, such as HFrEF and HFpEF, which is also a novel aspect.

LIMITATIONS
The major source of limitation is the different follow-up periods of the included studies, ranging from a minimum of 60 days to a maximum of 4.2 years. Furthermore, in some trials, the presence of HF at baseline was assessed by medical history only rather than echocardiographic parameters or biomarkers; hence, there is a chance that some patients with present heart failure were undiagnosed. Majority of the trials examined the most common SGLT2 inhibitors empagliflozin, dapagliflozin, and canagliflozin; for this reason, the cardioprotective results for these were the most prominent compared to those of sotagliflozin and ertugliflozin.

CONCLUSION
In conclusion, treatment with SGLT2 inhibitors showed robust results in reducing the incidences of HHF, CV death, and allcause mortality in patients with underlying HF. Furthermore, the SGLT2 inhibitors appear to show a class effect.