Cardiovascular and renal outcomes of combined SGLT2 inhibitors and GLP-1 receptor agonists versus monotherapy in patients with type 2 diabetes mellitus: a network meta-analysis [Research]

Background: Whether using both sodium–glucose cotransporter 2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1RA) is more effective than using either alone in type 2 diabetes mellitus remains uncertain. We sought to assess cardiovascular and kidney outcomes of combined SGLT2i and GLP-1RA therapy versus monotherapy using network meta-analysis.

Methods: We identified randomized controlled trials (RCTs) of SGLT2i or GLP-1RA in patients with type 2 diabetes mellitus through a comprehensive search of MEDLINE, Embase, and the Cochrane Library until Aug. 15, 2025. We compared patients receiving SGLT2i, GLP-1RA, both SGLT2i and GLP-1RA, or neither medication in a systematic review and network meta-analysis. Primary outcomes were major adverse cardiovascular events (cardiovascular death, myocardial infarction, and stroke) and major adverse kidney events (decline in estimated glomerular filtration rate, kidney failure, and death due to kidney failure). Secondary outcomes included heart failure–related hospital admissions, serious adverse events, and hypoglycemia.

Results: We included 12 RCTs with 99 683 participants. The risk of major adverse cardiovascular events did not differ significantly with combined SGLT2i and GLP-1RA therapy compared with SGLT2i (risk ratio [RR] 0.86, 95% confidence interval [CI] 0.74 to 1.01; very low certainty) or GLP-1RA alone (RR 0.95, 95% CI 0.81 to 1.12; very low certainty). Similarly, the risk of major adverse kidney events did not differ significantly with combined therapy from SGLT2i (RR 1.05, 95% CI 0.74 to 1.49; very low certainty) or GLP-1RA (RR 0.86, 95% CI 0.60 to 1.21; very low certainty). However, combined therapy was associated with a lower risk of heart failure–related hospital admission than SGLT2i (RR 0.72, 95% CI 0.52 to 0.99; very low certainty) or GLP-1RA (RR 0.62, 95% CI 0.45 to 0.86; very low certainty). Risks of serious adverse events or hypoglycemia did not differ significantly between combined therapy and monotherapy.

Interpretation: Compared with SGLT2i or GLP-1RA alone, combined therapy did not significantly reduce the risk of major adverse cardiovascular or kidney events but was associated with lower risks of heart failure–related hospital admission. Randomized controlled trials directly comparing combined therapy with monotherapy are needed to clarify the comparative effectiveness of combined therapy in patients with type 2 diabetes mellitus.

Sodium–glucose cotransporter 2 inhibitors (SGLT2i) lower blood glucose by reducing renal glucose reabsorption, preventing glucose toxicity and oxidative stress in kidney cells. 1 They also enhance tubuloglomerular feedback by increasing sodium delivery to the macula densa, which lowers glomerular pressure and blood pressure, thereby protecting kidney function. 2 On the other hand, glucagon-like peptide-1 receptor agonists (GLP-1RA) promote insulin secretion and suppress glucagon release, while lowering blood glucose by reducing liver glucose production and slowing gastric emptying. 3

The combination of SGLT2i and GLP-1RA has garnered considerable attention for its potential to improve cardiovascular, kidney, and metabolic outcomes in patients with type 2 diabetes mellitus. 4 , 5 Recent evidence, including findings from a large cohort study, suggests that combining these 2 drug classes may provide superior benefits for metabolic, cardiovascular, and renal outcomes compared with monotherapy. 5 The distinct and complementary mechanisms of action of SGLT2i and GLP-1RA result in synergistic effects that reduce both macrovascular and microvascular complications, improve cardiovascular risk factors, and prevent diabetic nephropathy, making their combination an appealing therapeutic option. 6

Meta-analyses that have evaluated combined therapy versus monotherapy report varying results. A recent meta-analysis by the SMART-C (SGLT2i Meta-Analysis Cardio-Renal Trialists Consortium) collaboration found that SGLT2i consistently reduced the risk of adverse cardiovascular events, heart failure–related hospital admission (henceforth, hospitalized heart failure), and progression of chronic kidney disease, regardless of whether patients were receiving a GLP-1RA at baseline. 7 Similarly, another meta-analysis, which analyzed randomized controlled trials (RCTs) of GLP-1RA, evaluated cardiovascular and kidney outcomes in patients with and without concurrent SGLT2i use. 8 The study concluded that GLP-1RA provide consistent cardiovascular and renal benefits, irrespective of background SGLT2i therapy. 8

However, both meta-analyses included a relatively small number of trials and participants using combined therapy at baseline, which limited the generalizability of their findings. The comparisons were limited to pairwise analysis, which precluded further insights into the comparison between various treatment arms. To address these gaps, we sought to incorporate data from a wider range of RCTs in a network analysis to evaluate the cardiovascular, kidney, and safety outcomes of combined SGLT2i and GLP-1RA therapy versus monotherapy or placebo. By leveraging both direct and indirect comparisons, we aimed to provide more comprehensive evidence on the clinical efficacy and safety of combined therapy compared with SGLT2i or GLP-1RA monotherapy to guide clinical practice in real-world scenarios.

This systematic review and network meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 checklist, 9 along with the PRISMA extension for network meta-analyses (PRISMA-NMA). 10 The PRISMA-NMA checklist is included in Appendix 1, Supplementary Table 1, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.250369/tab-related-content . The protocol was prospectively registered in PROSPERO (CRD42024605727). We conducted a comprehensive search for RCTs in Ovid (Embase, MEDLINE) and the Cochrane Library without any date or language restrictions, with the final search completed on Aug. 15, 2025. Search terms included a combination of free text and medical subject heading (MeSH) terms. Two reviewers (M-H.C. and C-W.H.) independently screened citations based on predefined selection criteria. We examined the references of included articles and relevant systematic reviews for additional eligible studies. The full search strategies are provided in Appendix 1, Supplementary Table 2.

We included RCTs of adults (aged ≥ 18 yr) with type 2 diabetes who received GLP-1RA, with or without baseline use of SGLT2i, and those who received SGLT2i with or without baseline GLP-1RA. All included trials evaluated either GLP-1RA or SGLT2i monotherapy versus placebo. No trial directly compared combined SGLT2i and GLP-1RA therapy with placebo. We excluded trials that compared GLP-1RA or SGLT2i with other glucose-lowering medications, studies that did not specifically target the population with type 2 diabetes, trials in which the use of GLP-1RA and SGLT2i could not be distinguished, studies that did not report cardiovascular or kidney outcomes, and studies published in abstract form or without peer review.

The primary outcomes of the current study were major adverse cardiovascular events (the composite of cardiovascular death, myocardial infarction, and stroke) and major adverse kidney events (the composite of ≥ 40% or 50% decline in estimated glomerular filtration rate [eGFR], kidney failure [eGFR 2 , chronic dialysis, or kidney transplantation], or death due to kidney failure). Trials in which the components of these composite outcomes differed were not included for the corresponding outcome. Secondary outcomes included hospitalized heart failure, eGFR slope, serious adverse events, and hypoglycemia (blood glucose level ≤ 70 mg/dL or any severe episode requiring the assistance of another person). Serious adverse events were defined by individual trials. A detailed list of serious adverse events for each trial is provided in Appendix 1, Supplementary Table 3.

For each included study, we recorded study and patient characteristics, including the year of publication, the medications used in the intervention arm, sample size, mean age, sex, history of cardiovascular disease or heart failure, systolic blood pressure, body mass index, mean baseline eGFR, glycated hemoglobin, and the proportion of patients under renin-angiotensin system blockade at baseline. We recorded the number of patients with primary and secondary outcomes, the total number of patients in each group, and changes in eGFR during treatment. Trial-level sample sizes and the number of major adverse cardiovascular events and major adverse kidney events are provided in Appendix 1, Supplementary Table 4. We categorized patients into treatment groups (i.e., SGLT2i, GLP-1RA, or placebo) if they had been randomized to it or if they had been using it at baseline in an included study. Accordingly, patients randomized to the SGLT2i group who also used GLP-1RA at baseline or those randomized to the GLP-1RA group who also used SGLT2i at baseline were categorized as the combined therapy group. We considered patients taking one of the medications at baseline and subsequently randomized to placebo, as well as those not taking it and subsequently randomized to receive the medication, to be part of the same treatment group. We extracted data for the subgroups of patients who did or did not use SGLT2i or GLP-1RA at baseline separately. To avoid violating treatment randomization, in each trial, patients receiving the noninterventional medication at baseline (e.g., GLP-1RA in SGLT2i trials) were compared only with those who were also receiving it at baseline and vice versa. Two independent examiners (H.-Y.W., H.-C.P.) extracted data, which a third examiner (V.-C.W.) subsequently reviewed and adjudicated.

We evaluated the risk of bias in the included studies using Cochrane’s Risk-of-Bias tool 2.0 for RCTs, including random sequence generation, allocation concealment, blinding, loss of outcome data, and selective reporting of results. 11 Two investigators (V.-C.W., H.-C.P.) independently carried out the assessments. In instances where the reviewers disagreed on the risk-of-bias rating, a third reviewer (J.-Y.C.) was consulted for discussions to reach a consensus. Based on these assessments, we classified studies as low risk of bias, some concerns, or high risk of bias. We assessed the transitivity assumptions according to the clinical and methodological characteristics of the included studies, as well as by a comparison of the treatment effects among treatment arms. We proceeded with data synthesis if the transitivity assumptions were met, confirming validity of indirect comparisons.

We evaluated the certainty of evidence for each outcome across the network using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach. This approach assessed within-study risk of bias, inconsistency, indirectness, and publication bias for direct and indirect evidence in a stepwise manner, along with intransitivity, incoherence, and imprecision, providing a final grading for the network evidence categorized as high, moderate, low, or very low certainty. 12

In the networks of treatment comparisons, a node represented each treatment, and randomized comparisons between therapies were shown as links between nodes. We performed statistical analyses for all outcomes based on 4 nodes: combined therapy, GLP-1RA alone, SGLT2i alone, and placebo. We conducted frequentist network meta-analysis using random-effects models. We reported continuous data as mean differences (MDs) with 95% confidence intervals (95% CIs) and dichotomous data as risk ratios (RRs) with 95% CIs. We evaluated overall heterogeneity and inconsistency with I 2 statistics. We estimated the between-study variance τ 2 using the restricted maximum likelihood method and calculated the CIs based on the standard normal distribution. We performed sensitivity analyses based on the t distribution to make statistical inferences using multivariate random-effects meta-analysis. We assessed publication bias with funnel plots and Egger tests. We calculated P-scores to evaluate the relative ranking of the probability that a treatment modality was superior to others. These scores range from 0 to 1, with higher values indicating a greater likelihood that a treatment is superior to other treatments in the network. The league tables displayed the relative efficacy of all possible pairwise comparisons, with the point estimates presented as RRs along with their 95% CIs. We evaluated incoherence by a design-by-treatment interaction test and node-splitting analysis. Subgroup and dose–response analyses were not pre-specified and not conducted because of the inconsistent availability of arm-level data.

To ensure that treatment effects were not systematically different between SGLT2i and GLP-1RA trials, we performed proportional meta-analyses to assess between-group heterogeneity for pooled proportions of patients with outcomes from SGLT2i trials versus GLP-1RA trials. In the absence of significant heterogeneity, we deemed treatment effects consistent across trial designs, and we proceeded with a network meta-analysis.

All tests were 2-tailed, with the statistical significance level set at a p value less than 0.05. We conducted statistical analysis using R version 4.2.2 with packages netmeta (version 3.2–0), meta (version 8.2–0), metafor (version 4.8–0), and ggplot2 (version 3.5.2).

Ethics approval was not required for this study, as it was a systematic review and meta-analysis of previously published, publicly available aggregate data and did not involve individual patient-level data or human participants.