1 Introduction

Heart failure (HF) is a common clinical syndrome that imposes substantial health and financial burdens worldwide [1, 2]. Patients with HF are characteristic for impaired left ventricular function, reduced quality of life, and increased mortality [3]. To date, multiple guideline-directed medical therapies have been proposed and applied to clinical practice, including β-blockers, renin–angiotensin–aldosterone system inhibitors, diuretics, and so forth [1]. Nevertheless, the burden of cardiovascular events in HF, particularly in HF with reduced ejection fraction (HFrEF; defined as left ventricular ejection fraction [LVEF] ≤ 40%), remains significant [4]. Therefore, a novel therapeutic strategy is required to improve symptoms, decrease mortality, and reduce hospitalization for patients with HF.

Dapagliflozin is a sodium-glucose co-transporter 2 inhibitor (SGLT2i) initially developed to treat type 2 diabetes mellitus (T2DM) by preventing glucose reabsorption in the proximal tubule and thereby lowering plasma glucose concentrations [5]. Recently, several clinical trials using dapagliflozin in patients with T2DM at elevated cardiovascular risk have demonstrated reduced rates of hospitalization for HF (HHF) [6,7,8]. In addition, several studies have suggested cardiovascular benefits with dapagliflozin in patients with HF regardless of the presence or absence of T2DM [9,10,11].

However, the overall effects of dapagliflozin in patients with HF remain poorly understood and are thus vigorously debated. Hence, a systematic review is required to summarize existing data and evaluate the protective effects of dapagliflozin against HF. We aimed to conduct a meta-analysis to comprehensively estimate the effects of dapagliflozin in patients with HF regardless of the presence or absence of T2DM.

2 Methods

2.1 Selection strategy

The literature search was conducted in the PubMed, Embase, Web of Science, and Cochrane Library electronic databases from inception to 15 February 2020. Detailed search strategies were used. For example, the following strategy was used in PubMed: (((sodium-glucose co-transporter [title/abstract]) OR SGLT2 [title/abstract]) OR dapagliflozin [title/abstract]) AND heart failure [title/abstract]. We also manually screened the reference lists of retrieved articles, previous meta-analyses, and reviewed articles to identify any related research.

2.2 Study design

Inclusion criteria were as follows: (1) randomized, double-blind, placebo-controlled clinical trials; (2) studies reporting efficacy or safety-related outcomes, including HHF, all-cause mortality, cardiovascular death, major adverse cardiovascular events (MACE; described as a composite of cardiovascular death, myocardial infarction, and ischemic stroke), and adverse events; and (3) full articles published in English in peer-reviewed journals. For multiple publications of the same trial, only the study containing the most data was included. Importantly, studies regarding the effects of dapagliflozin in patients with HF, regardless of LVEF, were considered, in order to capture the full spectrum of HF. We excluded animal studies, review articles, letters, case reports, articles lacking outcomes of interest/control group, and trials with follow-up of < 1 month). Two independent investigators (XD Zheng and XY Jiang) screened and reviewed article titles, abstracts, and full texts to assess eligibility. Any disagreements about eligibility were resolved via discussion with a third investigator (C Tang).

2.3 Data extraction and quality assessment

For each eligible study, data were extracted by two independent investigators (XD Zheng and Q Qu), with any discrepancy clarified through discussion with a third investigator (ZY Wang). Extracted characteristics included first author, publication year, follow-up period, drug dose, number of participants, patient demographic and clinical characteristics, and the following efficacy and safety outcomes: HHF, all-cause mortality, cardiovascular death, MACE, systolic blood pressure, body weight, glycated hemoglobin (HbA1c), and diverse adverse events.

We used the Cochrane Collaboration’s tool for assessing the risk of bias to assess the quality of the included studies [12] in terms of selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias. Two independent investigators (XD Zheng and ZY Wang) conducted this assessment, with any controversy settled by consensus with a third investigator (JY Sun).

2.4 Statistical Analysis

Categorical variables were presented as frequencies and risk ratios (RR) with corresponding 95% confidence intervals (CI), and continuous variables were presented as mean ± standard deviation. P values < 0.05 were considered statistically significant. The proportion of variance derived from between-study heterogeneity was quantified by I2, with a fixed-effects model applied when the I2 value was < 50% and a random-effects model adopted otherwise [13]. Baujat plot-based sensitivity analyses were performed to detect potential outliers and influence points that over-contributed to the overall heterogeneity and results [14], and influence analyses were conducted to identify whether any individual study distorted the pooled effects, as described by Viechtbauer and Cheung [15]. All analytical results were measured using Review Manager v5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark), and sensitivity analyses were performed using R software v3.6.3 (R Foundation for Statistical Computing, Vienna, Austria). The analysis was carried out according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines [16].

3 Results

3.1 Study selection and quality assessment

As shown in Fig. 1, a total of 2249 articles were identified via initial database search, of which 878 duplicates and 1323 irrelevant studies were excluded after title and abstract screening. After the full texts of the remaining 48 studies were reviewed, four articles with 6738 patients were ultimately included for further analysis [11, 17,18,19] (Table 1). Of note, 4226 (62.7%) participants were diagnosed with T2DM, whereas 5210 (77.3%) participants had concomitant reduced LVEF. All included studies were high-quality randomized controlled trials, and the corresponding risks of bias are presented in Fig. 2.

Fig. 1
figure 1

Flow chart of study selection

Full size image
Table 1 Studies included and the baseline characteristics of patients
Full size table
Fig. 2
figure 2

Risk of bias for the included studies. Green cells indicate a low risk of bias, red cells indicate a high risk, and empty cells indicate an unclear risk

Full size image

3.2 Efficacy outcomes assessment

HHF, all-cause mortality, cardiovascular mortality, and MACE were documented to assess efficacy outcomes with dapagliflozin. Importantly, all the included studies reported the effect of dapagliflozin on HHF, and a fixed-effects model was applied to evaluate the difference. As shown in Fig. 3a, dapagliflozin treatment was associated with significantly reduced HHF compared with placebo (RR 0.72; 95% CI 0.63–0.82; P < 0.00001; I2 = 28%). Similarly, four studies reported all-cause mortality, with results indicating that dapagliflozin significantly decreased all-cause mortality (RR 0.83; 95% CI 0.74–0.94; P = 0.004; I2 = 0%; Fig. 3b). Cardiovascular mortality (RR 0.86; 95% CI 0.74–0.99; P = 0.03; I2 = 0%; Fig. 3c) and MACE (RR 0.88; 95% CI 0.78–0.99; P = 0.03; I2 = 12%; Fig. 3d) were also reduced in patients with HF receiving dapagliflozin compared with those receiving placebo.

Fig. 3
figure 3

Forest plot of a hospitalization for heart failure, b all-cause mortality, c cardiovascular mortality, and d major adverse cardiovascular events. Comparison of hospitalization for heart failure, all-cause mortality, cardiovascular mortality, and major adverse cardiovascular events between dapagliflozin and placebo group. CI confidence interval

Full size image

3.3 Safety Outcomes Assessment

Multiple adverse events were evaluated, including hypoglycemia, volume depletion, and renal impairment, as were serious adverse events. Three studies enrolling 4751 patients reported no significant difference in hypoglycemia (RR 1.09; 95% CI 0.78–1.53; P = 0.60; I2 = 0%; Fig. 4a) or the occurrence of volume depletion (RR 1.10; 95% CI 0.91–1.34; P = 0.33; I2 = 0%; Fig. 4b) or renal impairment (RR 1.13; 95% CI 0.30–4.26; P = 0.86; I2 = 53%; Fig. 4c) between the two groups. Patients receiving dapagliflozin had a significantly lower incidence of serious adverse events than those receiving placebo (RR 0.90; 95% CI 0.84–0.96; P = 0.002; I2 = 0%; Fig. 4d).

Fig. 4
figure 4

Forest plot comparing a hypoglycemia, b volume depletion, c renal impairment, and d serious adverse events between dapagliflozin and placebo. CI confidence interval

Full size image

3.4 Assessments of Other Indicators

Other indicators analyzed included systolic blood pressure, body weight, and HbA1c. Three studies involving 4751 participants reported results for systolic blood pressure, with the overall value significantly lower with dapagliflozin than with placebo [mean difference (MD) − 1.00; 95% CI − 1.86 to − 0.15; P = 0.02; I2 = 0%; Fig. 5a]. Similarly, Fig. 5b illustrates a significant negative correlation between dapagliflozin treatment and body weight compared with placebo (MD − 1.72; 95% CI − 3.36 to − 0.08; P = 0.04; I2 = 87%), but no significant difference was noted in HbA1c (MD − 0.23; 95% CI − 0.52–0.07; P = 0.13; I2 = 90%; Fig. 5c).

Fig. 5
figure 5

Forest plot comparing a systolic blood pressure, b body weight, and c glycated hemoglobin between dapagliflozin and placebo. CI confidence interval, IV inverse variance, SD standard deviation

Full size image

3.5 Sensitivity analysis

No research fell in the top-right quadrant of the Baujat plot (Fig. 6a), suggesting a significant lack of any individual study driving the observed results. Moreover, Fig. 6b shows the influence analyses comparing weight with overall heterogeneity, which reveals that the pooled estimated outcomes were robust and did not depend on a single study.

Fig. 6
figure 6

a Baujat plot using the dapagliflozin fixed-effects model. The x-axis represents the squared Pearson resident of each study, and the y-axis represents the standardized squared difference between the model-estimated effect for each study. Studies located in the upper-right corner of the Baujat plot contribute greatly to the overall heterogeneity and results. b Influences of overall heterogeneity in model studies. Study 1–4 refers to the study by Kato et al. [11], Kosiborod et al. [19], McMurray et al. [9], and Nassif et al. [10], respectively

Full size image

4 Discussion

To the best of our knowledge, although previous studies have demonstrated cardiovascular protective effects with SGLT2i, this is the first meta-analysis to comprehensively evaluate the effects of dapagliflozin in patients with HF regardless of the presence or absence of T2DM. We systematically assessed 6738 patients with HF randomized to dapagliflozin or placebo from four randomized controlled trials. On one hand, the results indicated that dapagliflozin significantly reduced HHF, cardiovascular death, all-cause mortality, MACE, systolic blood pressure, and body weight. On the other hand, no significant differences were found in hypoglycemia, volume depletion, renal impairment, or serious adverse events. Generally, these findings reveal the efficacy and safety of dapagliflozin in patients with HF, regardless of the presence or absence of T2DM.

HF is a common syndrome that seriously threatens human health worldwide. Despite appreciable advances in therapeutic strategies, the cardiovascular outcomes, survival, and prognosis for patients with HF remain relatively poor, resulting in a persistently high risk of symptomatic progression and clinical deterioration. Therefore, a novel strategy against HF is required to complement current therapies as previously detailed. Additional cardiovascular protective effects beyond lowering plasma glucose have recently been reported for dapagliflozin [6,7,8], which has made it a research hotspot in HF treatment. This meta-analysis indicates that dapagliflozin can effectively reduce HHF, cardiovascular death, all-cause mortality, and MACE in patients with HF, findings that are consistent with those of previous studies [20,21,22,23]. Importantly, considering that more than three-quarters of the participants had reduced LVEF, these results may primarily reflect the effects of dapagliflozin in patients with HFrEF, which deserves paramount attention in clinical applications.

In terms of safety, this analysis showed no increased risk of hypoglycemia, volume depletion, renal impairment, or serious adverse events in patients receiving dapagliflozin. Dapagliflozin could inhibit renal glucose reabsorption, thus decreasing plasma glucose concentrations. Unlike other hypoglycemic agents, dapagliflozin can subsequently increase plasma glucagon concentrations and improve the production of endogenous glucose [24]. Therefore, although dapagliflozin could lower plasma glucose, it does not contribute to hypoglycemia in patients with HF without T2DM. Volume depletion is another major concern in patients receiving high doses of loop diuretics. Consistent with recent studies, this analysis showed no significant difference between dapagliflozin and placebo in volume depletion events [25, 26]. Interestingly, the vast majority of SGLT2i have been demonstrated as associated with a higher risk of volume depletion than placebo or other oral hypoglycemic agents [27, 28]. Furthermore, no significant reduction in cardiac biomarkers (natriuretic peptides) was reported in included studies, which further validates our results and contradicts the volume depletion hypothesis. Thus, we postulate that volume preservation can be partly explained by diuretic dose adjustments during trials; such a speculation was supported by the results of a recent clinical trial [29]. Regardless, these results suggest dapagliflozin has a reassuring safety profile.

Although several possible mechanisms for the cardiovascular protective benefits of SGLT2i have been proposed, the underlying biological mechanisms remain unclear. The protective effect was observed early after SGLT2i treatment, indicating that the possible mechanisms might be distinguished from other glucose-lowering therapies [9, 30]. In several clinical trials in patients with T2DM [31, 32], dapagliflozin lowered plasma volume and depressed systolic blood pressure without increasing heart rate, suggesting its potential as a promising therapeutic alternative in HF, particularly in HFrEF (e.g., β-blockers and bradycardia; renin–angiotensin–aldosterone system inhibitors and renal dysfunction/hyperkalemia). However, reduced systolic blood pressure could generate insufficient blood supply to the brain, heart, kidneys, and other organs and has been prominently correlated with an elevated risk of sudden cardiac death. Several clinical trials have corroborated the neutral or even detrimental role of antihypertensive drugs in established HF [33,34,35]. Hence, the clinical application of dapagliflozin might require combination with antihypotensive therapies. Of note, glycosuria could stimulate lipid oxidation to compensate for the loss of energy and further lead to weight loss and fat mass reduction [36, 37]. These changes may subsequently lighten both the preload and the afterload to the heart and eventually result in cardiovascular protective effects. Alongside a systemic hemodynamic and metabolic effect, direct cardiovascular effects have also been proposed as important drivers, including improved cardiac inflammation [38], oxidative stress [39], apoptosis [38], mitochondrial dysfunction [40], ionic dyshomeostasis [41], and so forth [42, 43].

Other studies have mainly evaluated the cardiovascular protection benefits of dapagliflozin in patients with T2DM with elevated cardiovascular risks, most of whom did not have established HF. Therefore, the cardiovascular benefits of dapagliflozin primarily corresponded to the prevention of HF. Clinical trials to evaluate protective effects in patients with HF, most of whom have reduced ejection fraction and T2DM, have only recently been conducted. The results from the DAPA-HF trial, which showed an association between dapagliflozin and improved HF-related prognosis regardless of baseline HbA1c, have led to dapagliflozin being recognized as a potential therapeutic strategy for HFrEF regardless of T2DM. In addition, several large ongoing trials are evaluating the clinical effects of SGLT2i in patients with HF with or without T2DM (NCT02653482, NCT03030235, NCT03057951, NCT03057977, and NCT03619213). These clinical trials may further demonstrate the effects of dapagliflozin in patients with HF, especially those with preserved ejection fraction.

Although this analysis adhered strictly to PRISMA guidelines, several limitations should be highlighted. First, the included studies did not provide enough relevant information to allow us to precisely define volume depletion, renal impairment, and serious adverse events. Second, the use of secondary outcomes (cardiovascular death, myocardial infarction, ischemic stroke) derived from DECLARE TIMI 58, a neutral trial on the original primary endpoint (MACE), may reduce the strength of this meta-analysis. There may also exist some selection bias induced by the inclusion criteria (e.g., exclusion of patients with symptomatic hypotension or systolic blood pressure < 95 mmHg at two of three measurements in visit 1 or 2 of the DAPA-HF trial) during the run-in phases of the included trials, which may cause unexpected signs of some estimators. Importantly, since 77.3% of participants involved in this meta-analysis had HFrEF, our findings might be mainly applicable to current HFrEF treatment. Moreover, since the included patients were characterized by a specific baseline (e.g., 62.7% of patients diagnosed with T2DM, 63.3% of patients with New York Heart Association class II disease, 73.1% males, and so forth), the generalizability of these findings to real-world HFrEF populations is debatable.

5 Conclusion

This meta-analysis predominantly strengthens the efficacy and safety of dapagliflozin in patients with HFrEF regardless of the presence or absence of T2DM, with some caveats. Importantly, whether benefits are observed regardless of LVEF remains to be established, and more randomized controlled trials are required to verify the effects of dapagliflozin in patients with HF.