Effect of Sodium-Glucose Cotransporter Type 2 Inhibitors on Clinical Outcomes in Patients With Cardiovascular Disease: A Meta-Analysis of Randomized Controlled Trials

Caiyun Zheng Fujian Medical University A liated Fuqing City Hospital Meimei Lin Fujian Medical University A liated Fuqing City Hospital Yan Chen Fujian Medical University A liated Fuqing City Hospital Haiting Xu Fujian Medical University A liated Fuqing City Hospital Lingqun Yan Fujian Medical University A liated Fuqing City Hospital Hengfen Dai (  hengfendai2011@163.com ) Fujian Medical University A liated Fuzhou First Hospital: Fuzhou First Hospital https://orcid.org/0000-0002-2037-4208

search strategies are detailed in Multimedia Appendix 1. The search was restricted to human studies.
References for the identi ed studies were also retrieved to identify studies that may be eligible. Only articles written in English were considered. There were no restrictions on the publication date.

Inclusion and exclusion criteria
Eligible patients included patients with cardiovascular disease, with or without T2DM. Patients with other age-related comorbidities were not excluded. Patients in the intervention group were treated with SGLT-2s (canagli ozin, dapagli ozin, empagli ozin, ertugli ozin, luseogli ozin, ipragli ozin, sotagli ozin, and tofogli ozin) for a period of time. Patients who received a placebo or drugs other than SGLT-2s were used as controls. The primary outcomes were all-cause mortality, cardiovascular death, and HHF. The secondary outcomes were adverse cardiovascular reactions, adverse reactions in endocrine, renal side adverse reactions, infection, and other adverse reactions.

Data extraction and quality assessment
The included studies reported all-cause mortality, cardiovascular mortality, and HHF. All literature titles and abstracts were initially screened for relevance to exclude studies that were not meaningful to our research by two reviewers. Drs. Zheng and Dai independently analysed the full text and extracted data from selected studies; divergence was settled by discussion and consensus or by third-party arbitration.
The risk assessment tool for Cochrane bias was used, and the included randomised controlled trials (RCTs) were assessed according to the literature assessment criteria in the Cochrane Systematic Review Manual [11].

Data synthesis and analysis
Meta-analyses were conducted using Cochrane's Review Manager (RevMan) version 5.3 (The Cochrane Collaboration, Copenhagen, Denmark) and R programming language, version 3.6.3 (R Foundation for Statistical Computing, Guangzhou, China). We used the associated relative risk (RR) with 95% con dence intervals (CIs) to evaluate clinical outcomes (e cacy and safety) in patients with cardiovascular disease receiving standard treatment with or without SGLT-2s. We assessed whether the results of the included studies were consistent. Chi-square tests were used to assess heterogeneity, and I 2 was used for quantitative analysis. Values of P ≥ 0.05 and I 2 ≤ 50% were not considered to represent heterogeneity, and xed-effect models were used. In contrast, P < 0.05 and I 2 > 50% indicated signi cant heterogeneity among the different studies. Sensitivity analysis was performed by successively deleting each study and reanalysing the dataset for all remaining studies. Publication bias was assessed using funnel plots and Egger's tests. P values less than 0.05 (two-tailed) were considered statistically signi cant, and we did not adjust for multiple testing.
Basic characteristics and quality evaluation of the literature In total, 42682 patients were included in 10 RCTs (25108 patients in the intervention group and 18574 patients in the control group). The detailed baseline characteristics of each study are presented in Table   S1. Quality evaluation was conducted using the Cochrane systematic evaluation method, and the included studies had a low risk of overall bias, as shown in Figure S1.

Primary outcomes
The summarised outcomes of our meta-analysis are presented in Table S2. To evaluate the primary outcomes, 10 trials were included in the meta-analysis. Estimates of primary outcomes for all-cause mortality, cardiovascular mortality, and hospitalisation for heart failure are shown in Figs. 2-4.

HHF
Nine articles included outcomes for HHF (24959 in the SGLT-2 group and 18436 in the placebo group).

Secondary outcomes
Adverse cardiovascular outcomes There were ve studies with adverse cardiovascular events, including four with myocardial infarction. The ve studies included 28983 patients (17753 in the SGLT-2 group and 11230 in the placebo group). Metaanalysis results showed that the incidence of myocardial infarction was signi cantly lower in patients receiving SGLT-2 inhibitors than in those receiving a placebo (RR: 0.90; 95% CI: 0.82-0.99; P = 0.03; I 2 = 32%). Five studies with stroke outcomes included 29248 patients (17869 in the SGLT-2 group and 11379 in the placebo group). Meta-analysis results indicated that there were no signi cant differences in the incidence of stroke between patients treated with SGLT-2 inhibitors and those treated with a placebo (RR: 0.98; 95% CI: 0.86-1.11; P = 0.72; I 2 = 21%; Fig. 5).

Adverse endocrine outcomes
In total, 30719 patients (18645 in the SGLT-2 group and 12074 in the placebo group) were included in the six studies on hypoglycaemia outcomes. Meta-analysis results showed that there were no signi cant differences in the incidence of hypoglycaemia between patients using SGLT-2s and those using the placebo (RR: 1.00; 95% CI: 0.96-1.05; P = 0.87; I 2 = 8%).
Additionally, 34796 patients (20674 in the SGLT-2 group and 14122 in the placebo group) were included in six studies of outcomes related to diabetic ketoacidosis. Meta-analysis results showed that the incidence of diabetic ketoacidosis in patients treated with SGLT-2s was signi cantly higher than that in patients treated with a placebo (RR: 3.65; 95% CI: 1.83-7.27; P = 0.0002; I 2 = 0%; Fig. 6).

Adverse renal outcomes
In the four trials, there were no signi cant differences in compound outcomes of acute renal failure between the SGLT-2 and placebo groups (RR: 1.00; 95% CI: 0.90-1.11; P = 0.99; I 2 = 84%). Sensitivity analysis was performed by sequentially deleting each study and reanalysing the datasets for all remaining studies (RR: 0.99, 95% CI: 0.90-1.10, P = 0.96; I 2 = 80.8%; Table S3). In total, 35124 patients with acute renal injury were included in seven articles (20850 in the SGLT-2 group and 14274 in the placebo group). Meta-analysis results showed that the incidence of acute renal injury in patients treated with SGLT-2s was signi cantly lower than that in patients treated with a placebo (RR: 0.80; 95% CI: 0.68-0.93; P = 0.004; I 2 = 0%; Fig. 7).

Adverse infection outcomes
Outcomes of infection, including urinary tract infection, male genital infection, and female genital infection, were evaluated. The results suggested that the incidence of infection in patients treated with SGLT-2s was signi cantly higher than that in patients treated with a placebo (P < 0.05). The results of urinary tract infection were assessed in four articles, which included a total of 25720 patients (16146 in the SGLT-2 group and 9574 in the placebo group). Meta-analysis results showed that the incidence of urinary tract infection in patients treated with SGLT-2 inhibitors was signi cantly higher than that in patients treated with a placebo (RR: 1.08; 95% CI: 1.01-1.15; P = 0.02; I 2 = 39%).
Results of male and female genital infections were included in the three articles, which involved a total of 25400 patients (15975 in the SGLT-2 group and 9425 in the placebo group). Meta-analysis results showed that the incidence of male and female genital infections in patients treated with SGLT-2s was signi cantly higher than that in patients treated with a placebo (male: RR: 3.35; 95% CI: 2.90-3.87; P < 0.00001; I 2 = 0%; female: RR: 3.85; 95% CI: 3.43-4.32; P < 0.00001; I 2 = 0%; Fig. 8).

Others
Five articles containing fracture outcomes included 34533 patients (20543 in the SGLT-2 group and 13990 in the placebo group). Meta-analysis results showed that there were no signi cant differences in fracture incidence between patients treated with SGLT-2s and those treated with a placebo (RR: 1.11; 95% CI: 0.99-1.23; P = 0.08; I 2 = 24%).
Four articles included outcomes of amputation involving 27513 patients (15856 in the SGLT-2 group and 11657 in the placebo group). Meta-analysis results showed that the incidence of amputation in patients treated with SGLT-2s was signi cantly higher than that in patients treated with a placebo (RR: 1.42; 95% CI: 1.18-1.71; P = 0.0002; I 2 = 32%).
Six articles included outcomes of volume depletion in 30719 patients (18645 in the SGLT-2 group and 12074 in the placebo group). Meta-analysis results showed that the volume depletion of patients treated with a placebo was signi cantly lower than that of patients treated with SGLT-2s (RR: 1.22; 95% CI: 1.11-1.33; P < 0.0001; I 2 = 45%; Fig. 9).

Subgroup analysis
Detailed forest plots displaying subgroup analyses are presented in the Supporting Information. For all outcomes in the overall cardiovascular disease population, no subgroup effect was observed upon strati cation by T2DM status (Figs. S2-S4). In a subgroup analysis of heart failure types, there were no signi cant differences in all-cause mortality, cardiovascular mortality, or HHF between the SGLT-2 and placebo groups among patients with heart failure with preserved ejection fraction (HFpEF; Figs. S5-S7). SGLT-2s reduced all-cause mortality, cardiovascular mortality, and HHF largely independent of drug type, with the exception of ertugli ozin and sotagli ozin (Figs. S8-S10). In terms of safety, only the subgroup analysis of ketoacidosis and acute renal injury was performed because of the limited number of included trials. In terms of ketoacidosis, canagli ozin and ertugli ozin showed signi cant differences compared with the placebo group, whereas the other drugs showed no signi cant differences. There were no signi cant differences in acute renal failure between patients administered dapagli ozin, canagli ozin, or ertugli ozin and patients administered a placebo (Fig. S11). Empagli ozin signi cantly reduced the number of acute renal failure events (Fig. S12).

Sensitivity analysis results
Sensitivity analysis was performed to compare all-cause mortality by sequentially removing one study at a time and observing the exaggerated effect this had on the overall results. As shown in Fig. 10, after each study was excluded, the pooled RR of the remaining included studies was approximately 0.87, with no signi cant change, suggesting that the results of this meta-analysis were stable and reliable. Removing any study did not affect the overall results (Fig. 10).

Publication bias
Using funnel plots drawn by Review Manager 5.3, we visually assessed the publication bias of all-cause mortality (Egger's test P = 0.3638, 95% CI -0.662-0.669). All evidence suggested that the probability of publication bias was low. Funnel plots were not generated for other comparisons because fewer than 10 studies were included (Fig. 11).

Discussion
To the best of our knowledge, this is the rst study comparing SGLT-2s with a placebo in comprehensive outcomes of primary e cacy and adverse events in cardiovascular diseases in patients with and without T2DM. Cardiovascular diseases include various conditions, such as common coronary artery disease, hypertension, dyslipidaemia, congenital heart disease, valvular disease, and arrhythmia. Accordingly, cardiovascular diseases are a serious threat to human health worldwide. Cardiovascular outcomes, survival, and prognosis in patients with cardiovascular diseases are relatively poor. In a previous metaanalysis, SGLT-2 treatment of patients with T2DM signi cantly reduced all-cause mortality, cardiovascular mortality, HHF, risk of HF, and renal failure [24][25][26][27][28]. Moreover, among participants with established cardiovascular disease, regardless of the presence or absence of T2DM, SGLT-2s signi cantly reduce mortality and major adverse cardiovascular events (MACEs) compared with the placebo, and the bene t for the prevention and treatment of HF and renal disease is consistent with that of earlier metaanalyses [29]. The same results were observed in the HFrEF population; however, there were no signi cant differences in patients with HFpEF and no signi cant differences between the two groups for most adverse events. The results for the risk of infection, amputation, and hypovolemia should be interpreted with caution.
SGLT-2s were superior to the placebo or other glucose-lowering drugs in terms of the pooled primary e cacy outcomes, including all-cause mortality, cardiovascular mortality, HHF, and renal bene ts from EMPA-REG OUTCOME [12,30], CANVAS-R [13,31], CANVAS [32,33], CREDENCE [34,35], DECLARE-TIMI 58 [36,37], and VERTIS-CV trials [20,38]. In the EMPA-REG OUTCOME trial, all eligible patients had established cardiovascular disease, and empagli ozin reduced the risk of all-cause mortality by 32%, cardiovascular mortality by 38%, and HHF by 35% compared with the placebo. At the same time, the study con rmed that the dose of empagli ozin did not affect the hazard ratios of cardiovascular outcomes. In the CANVAS and CANVAS-R trials, treatment resulted in reductions of 49% for all-cause mortality, 47% for cardiovascular mortality, 30% for HHF, and 22% for major adverse cardiovascular events. Patients in the canagli ozin group had a lower risk of cardiovascular mortality, HHF, and renal failure than those in the placebo group at a median follow-up of 2.62 years in the CREDENCE trial. In the DECLARE-TIMI 58 trial, dapagli ozin appeared to robustly reduce the risk of cardiovascular mortality or HHF and MACEs in patients with T2DM and previous myocardial infarction. In the VERTIS-CV trial involving patients with T2DM and established atherosclerotic cardiovascular disease, ertugli ozin was shown to be noninferior to the placebo with MACEs. However, the incidence of cardiovascular mortality or HHF did not differ signi cantly between the two groups. Owing to the diverse properties of different drugs, we cannot exclude the possibility that differences among the agents in this class may result in real differences in clinical outcomes. It is also possible that the effects of individuals are similar because the con dence interval of VERTIS CV overlaps with that of previous trials. Owing to limited literature on the HFpEF population, the role of SGLT-2s in patients with HFpEF remains unclear. In current research on HFpEF, only the subgroup analyses in the DECLARE-TIMI 58 and VERTIS-CV trials suggested a potential impact on HHF or cardiovascular mortality. Two ongoing clinical trials in patients with HFpEF will provide further evidence of its e cacy (NCT03057951 and NCT01297257).
In terms of safety, three available studies showed neutral effects on myocardial infarction and stroke in patients receiving SGLT-2s [39][40][41]. In an RCT and a nationwide cohort study, the results showed no increased risk of hypoglycaemia with SGLT-2 monotherapy [42,43]; however, SGLT-2s were associated with approximately twice the risk of diabetic ketoacidosis as DPP4 inhibitors [44]. Accordingly, clinicians should be cautious when combining SGLT-2s with other hypoglycaemic drugs. SGLT-2s reduced the risk of dialysis, transplantation, or death due to kidney disease in individuals with T2DM, with or without basic renal disease, and provided protection against acute kidney injury [45][46][47]. The use of SGLT-2s was not associated with an increased risk of fracture and amputation compared with other antidiabetics [44,[48][49][50]. Volume depletion is another major concern, not consistent with recent studies, which showed signi cant differences between SGLT-2s and other oral hypoglycaemic agents in volume depletion events [51][52][53], perhaps because of individualised dosing of SGLT-2s. Other potential side effects include urinary tract infection and male/female genital infection, which may be attributed to the selective inhibition of renal proximal tubule glucose reabsorption by SGLT-2s. According to the United States Food and Drug Administration, all manufacturers are required to add a warning of potential infection to the prescribing information and patient medication guide for all SGLT-2s in 2018. Although these adverse events should not mask the overall cardio-renal bene ts of SGLT-2s [54][55][56], individuals at risk of these complications should be monitored closely, and treatment should be reconsidered or discontinued if such complications occur.
Our study had some limitations. First, confounding factors, such as age, sex, regionalism, baseline haemoglobin A1c, estimated glomerular ltration rate, and exposure to cardiovascular disease-related drugs, as well as drug combinations and other potential factors, were di cult to control. Second, articles published in languages other than English were excluded. Third, few RCTs of certain SGLT-2 drugs in the HFpEF population have been published. Finally, this meta-analysis may be underpowered for comparison of long-term adverse events between SGLT-2s and the placebo owing to the different durations of follow up for the included RCTs. Therefore, additional studies are required to con rm our ndings.

Conclusion
This meta-analysis showed that initiation of SGLT-2s was associated with substantially lower allcause/cardiovascular mortality and lower HHF in patients with cardiovascular disease and in both diabetic and nondiabetic patients. However, no bene t was observed in patients with HFpEF. SGLT-2s increased the risk of amputation, infection, volume depletion, and diabetic ketoacidosis, but slowed the decline in renal function and myocardial infarction adverse outcomes compared with a placebo. These ndings have important implications for the future direction of SGLT-2 therapy in cardiovascular disease research, but more studies are needed to determine the mechanism of SGLT-2s in cardiovascular disease treatment, as well as large RCTs to demonstrate the clinical e cacy of SGLT-2s in a speci c cardiovascular disease. Abbreviations T2DM: type 2 diabetes; HHF: hospitalisation for heart failure; SGLT-2: sodium-glucose cotransporter type 2 inhibitor; HFrEF: heart failure with reduced ejection fraction; HFpEF: heart failure with preserved ejection fraction; RCT: randomised controlled trial; RR: relative risk; MACE: major adverse cardiovascular event.

Declarations Acknowledgements
None.
Authors' contributions CYZ conceived the meta-analysis. CYZ and HFD developed the research strategy and provided statistical expertise. HFD drafted the manuscript. All Authors contributed to the development of the selection criteria, risk of bias assessment strategy, and data extraction criteria. All Authors read, provided feedback, and approved the nal manuscript.

Funding
This work was supported by the Startup Fund for Scienti c Research of Fujian Medical University (2017XQ1206) and Fuzhou Science and Technology Project Funds (2018-S-105-2). The funders of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.
Availability of data and materials Not applicable.
Ethics approval and consent to participate Not applicable.