Systematic review and meta-analysis for prevention of cardiovascular complications using GLP-1 receptor agonists and SGLT-2 inhibitors in obese diabetic patients

Patients with type 2 diabetes mellitus (T2DM) and obesity are at high risk of developing cardiovascular disease (CVD). Both glucagon-like peptide-1 receptor agonists (GLP-1 RAs) and sodium-glucose cotransporter (SGLT-2) inhibitors have been shown to prevent CVD in T2DM patients. Additionally, the two drugs reduce body mass. However, it is unknown which drug is more effective at reducing the risk of CVD in such patients. We searched Medline, EMBASE, and Cochrane Library records to February 20, 2021 and performed a network meta-analysis to compare the efficacy with which the drugs reduced the risk of major adverse cardiovascular events (MACE). We included 102,728 patients in 12 studies containing data of obesity subgroup analyses. In T2DM patients with obesity, GLP-1 RAs significantly reduced the risk of MACE versus placebo (relative risk, RR [95% confidence interval, CI]: 0.88 [0.81–0.96]), whereas SGLT-2 inhibitors showed a tendency (RR [95% CI]: 0.91 [0.83–1.00]). In an indirect comparison, GLP-1 RAs were not associated with a significant difference in MACE compared with SGLT-2 inhibitors (RR [95% CI]: 0.97 [0.85–1.09]). Thus, GLP-1 RAs are effective at preventing MACE than placebo in T2DM patients with obesity, although further studies are warranted to conclude their superiority to SGLT-2 inhibitors.

www.nature.com/scientificreports/ Diabetes Association (ADA) recommends the use of GLP-1 RAs and SGLT-2 inhibitors as first-line treatments for patients with T2DM and established or a high risk of atherosclerotic CVD 15 . Furthermore, both GLP-1 RAs and SGLT-2 inhibitors have anti-obesity effects, in contrast to the effects of some other anti-glycemic agents, such as insulin, the use of which tends to be associated with increases in the body mass of patients. Therefore, these two drugs are also recommended for use in overweight patients with T2DM by the ADA 15 . However, no head-to-head trials have been conducted on the efficacy of these drugs for the prevention of CVD in patients with obesity and T2DM. Therefore, we conducted a network meta-analysis to compare the effects of GLP-1 RAs and SGLT-2 inhibitors to prevent CVD in patients with obesity and T2DM.

Results
Search results and included studies. Figure 1 shows a flow chart of the study selection procedure. We selected 4,188 studies through database searching and identified a further nine studies through the searches of the reference lists of these articles. After the removal of duplicates, we removed a further 3,097 reports after screening the titles and abstracts, and another 79 because of missing data for upon inspection of the full-text articles. Therefore, 12 studies remained for inclusion in our meta-analysis. Of these, five were placebo-controlled randomized controlled trials (RCTs) of SGLT-2 inhibitors (canagliflozin 16,17 , empagliflozin 13 , ertugliflozin 18 , and dapagliflozin 14 ) and seven were placebo-controlled RCTs of GLP-1 RAs (lixisenatide 19 , exenatide 20 , liraglutide 11 , albiglutide 21 , subcutaneous and oral semaglutide 22,23 , and dulaglutide 12 ). All the studies, except CREDENCE, set major adverse cardiovascular events (MACE) as the primary endpoint. Eleven studies 11,13,14,[16][17][18][19][20][21][22][23] defined the cutoff value of body mass index (BMI) for obesity as 30 kg/m 2 and one 12 defined this cut-off as 32 kg/m 2 .
Patient characteristics and study quality assessment. The  www.nature.com/scientificreports/ are summarized in Table 2. There were 102,728 participants in the included RCTs (55,786 in the GLP-1 RA trials and 46,942 in the SGLT-2 inhibitor trials). In all the studies the mean age of the participants was between 60 and 70, and more than half were male. The mean BMI of the participants was > 30 kg/m 2 and their mean glycohemoglobin (HbA1c) was 7.4%-8.7%. Their mean systolic and diastolic blood pressures were 130-140 mmHg and 76-80 mmHg, respectively, and the mean circulating low-density lipoprotein-cholesterol concentrations were < 100 mg/dL (2.58 mmol/L) in all the included studies. Figure 2 shows the risks of bias for the included studies. Almost all the included studies were categorized as "low risk. " www.nature.com/scientificreports/ Figure 3 shows the network plot. In patients with obesity and

Results of the sensitivity analyses.
We conducted sensitivity analyses to validate the initial analysis (Table 3). First, we performed an analysis that excluded the CREDENCE data because the primary outcome of this trial was not MACE, but rather renal outcomes. This analysis showed that GLP-1 RAs significantly reduced the risk of MACE versus placebo in participants with T2DM who did or did not have obesity (   . There were moderate to high heterogeneity among the studies in both analyses (I 2 = 30% and 55%, respectively). An indirect comparison of GLP-1 RAs and SGLT-2 inhibitors, administered daily or weekly, did not reveal significant differences in the risks of MACE in T2DM patients with obesity.

Discussion
We have described the first network meta-analysis to compare the use of GLP-1 RAs and SGLT-2 inhibitors to reduce cardiovascular risk in the presence and absence of obesity. We found that GLP-1 RAs are superior to placebo for the prevention of MACE in T2DM patients with and without obesity, whereas SGLT-2 inhibitors show a tendency but do not outperform placebo in T2DM patients with obesity. However, in T2DM patients with or without obesity, there is no significance in the risk reduction of MACE between GLP-1 RAs and SGLT-2 inhibitors. Sensitivity analyses generated similar findings, which supports the validity of our findings.
Recent meta-analyses have shown protective effects of GLP-1 RAs and SGLT-2 inhibitors against CVD in patients with T2DM [26][27][28] . However, no head-to-head trials have compared the use of these drugs for the prevention of MACE in patients with obesity and T2DM. In the present study, we cannot conclude the superiority between GLP-1 RAs and SGLT-2 inhibitors in the risk reduction of MACE in T2DM with obesity (Fig. 4). Previous meta-analysis has showed that GLP-1 RAs and SGLT-2 inhibitors may have different effects on each component of three-point MACE, which comprised cardiovascular death, myocardial infarction, and stroke 29 . Particularly, a recent network meta-analysis has reported that GLP-1 RAs inhibited the risk of stroke than SGLT-2 inhibitors, although the risk of cardiovascular death and myocardial infarction is comparable 30 . These factors may influence on our results because our study assessed these diseases together.
The present data indicate that GLP-1 RAs reduce the risk of MACE compared with placebo in T2DM patients with obesity (Fig. 4). Furthermore, the sensitivity analysis that was performed after the exclusion of the CRE-DENCE data revealed a significant reduction in the risk of cardiovascular events compared with placebo in patients taking GLP-1 RAs with obesity (Table 3). In obesity, greater secretion of adipocytokines by visceral fatty tissues worsens T2DM and CVD through the induction of insulin resistance, endothelial dysfunction, and hypercoagulability 5 . Therefore, we hypothesize that GLP-1 RAs reduce the incidence of CVD events in obese T2DM patients by reducing the activation of adipocytokine pathways. Notably, some RCTs have shown that exenatide, one of the GLP-1 RAs, reduces the intra-abdominal fat volume, adipocytokine concentrations, inflammation, and insulin resistance in obese patients with T2DM 31 . Furthermore, GLP-1 RAs have anti-obesity effects through the regulation of appetite via central GLP-1 receptors, which may also have contributed to the present findings 4,15 . In our sensitivity analyses, liraglutide and subcutaneous semaglutide, drugs reported to  www.nature.com/scientificreports/ reduce patients' body weight in large RCTs 24,25 , also show the risk reduction of MACE compared with placebo for obese T2DM participants (Table 3). However, further research is needed to confirm the mechanisms whereby GLP-1RAs protect against CVD in patients with obesity and T2DM. We did not identify a significant advantage of SGLT-2 inhibitors over placebo, although they showed a tendency in obese T2DM. To our knowledge, no previous trial has assessed the efficacy of SGLT-2 inhibitors with respect to a reduction in the risk of MACE in patients who were or were not obese. However, some previous Table 3. Details of the sensitivity analyses. SGLT2i, sodium-glucose cotransporter-2 inhibitor; GLP-1 RA, glucagon-like peptide-1 receptor agonist; RR, relative risk; CI, confidence interval. www.nature.com/scientificreports/ reviews have identified anti-obesity effects of SGLT-2 inhibitors 32,33 . Among the studies included in the present meta-analysis, only canagliflozin caused a significant reduction in the incidence of MACE in the obesity subgroup of CREDENCE and CANVAS, while other SGLT-2 inhibitors did not 13,14,[16][17][18] , and hypotheses have been proposed to explain this inconsistency in the effects of SGLT-2 inhibitors on the incidence of MACE 34 . In particular, SGLT-2 inhibitors differ in their selectivity for SGLT-2 and SGLT-1 35 . SGLT-1 is expressed in the renal proximal tubules, but distally to SGLT-2 8,36 . Studies of animal models of diabetes mellitus have shown that renal SGLT-2 and SGLT-1 expression is upregulated and glucose reabsorption is greater 37,38 . Because some previous studies have shown that renal SGLT-1 activation reduces urinary glucose excretion, SGLT-2 inhibitors with low selectivity, and therefore a relatively high affinity for SGLT-1, may be more effective at reducing cardiometabolic risk than those with high selectivity 33,38,39 . A previous network meta-analysis showed that the use of canagliflozin is associated with better improvements in cardiometabolic markers in T2DM patients than that of other SGLT-2 inhibitors 40 . In the present study, the fact that the SGLT-2/1 selectivity of canagliflozin is lower than that of other SGLT-2 inhibitors might have affected the results, especially in patients with obesity and T2DM 32,35,41 .
Since GLP-1 RAs and SGLT-2 inhibitors have different pharmacological targets, the combination therapy of the two drugs is expected to exert various benefits. In fact, recent studies have suggested that the combination therapy of the two drugs could have the better effects on body weight, metabolic parameters, and cardiovascular function in T2DM patients than monotherapy 42,43 . Moreover, SGLT-2 inhibitors have established evidence reducing the risk of worsening heart failure in patients with low ejection fraction regardless of T2DM 44 . Further investigations are warranted to clarify the cardiac protective effects of SGLT-2 inhibitors and how to select or combine the two drugs depending on the patients' CVD risks.
The present study had some limitations. First, we could not conduct sub-analyses with respect to factors affecting BMI, including age, sex, and ethnicity. Second, we observed heterogeneity in some of the sensitivity analyses. Third, we were not able to assess the change in the body mass of the participants in each RCT. Fourth, splitting each trial's data into obese and non-obese subgroups reduced the power and therefore reduced the likelihood of demonstrating a benefit.
In conclusion, GLP-1 RAs reduce the risk of MACE versus placebo in patients with T2DM and obesity, whereas SGLT-2 inhibitors tend to be statistically favorable but do not show a significant difference. Further studies are warranted to conclude GLP-1 RAs' superiority of cardiovascular protection to SGLT-2 inhibitors in T2DM patients with obesity.
Selection of studies. Two independent reviewers (KU and YK) blindly checked the titles and abstracts of the identified articles. Next, the eligibility of studies with respect to the inclusion and exclusion criteria was judged by the two authors. If there was disagreement, a third senior reviewer (TY) was consulted and the matter discussed until a consensus was achieved.
The inclusion criteria were as follows: (i) RCTs published as peer-reviewed articles; (ii) all participants were ≥ 18 years of age and had T2DM; (iii) comparison of treatment (GLP-1 RA or SGLT-2 inhibitor) with placebo; (iv) comparison of the risk of MACE between the two groups; and (v) MACE data recorded according to BMI. The exclusion criteria were as follows: (vi) experimental animal study and (vii) insufficient data to evaluate the RRs of MACE even after contacting the authors.

Data extraction and assessment of bias.
After selecting the studies, the two independent reviewers (KU and YK) extracted data regarding the onset of MACE in participants that underwent treatment with either class of drug or placebo groups from each report. A third senior reviewer (TY) was responsible for resolving any anomalies regarding the data or quality assessment. To identify bias in the RCTs, we used the Cochrane risk of bias assessment 47 . Outcomes. The primary endpoint was three-point MACE, which comprised cardiovascular death, myocardial infarction, and stroke.