How SGLT2 inhibitors, incretin-based treatments, and finerenone compare in improving heart and kidney health

A comprehensive search of MEDLINE and the Cochrane Central Register of Controlled Trials (CENTRAL) was conducted from January 2023, the final search date of our previous analysis [10], through April 2025. A full search strategy is included in the supplementary appendix.

References were uploaded to Covidence electronically merged to eliminate duplicates [15]. Two investigators screened abstracts to determine if they met inclusion criteria. Inclusion criteria included RCTs that examined the effect of SGLT2i, incretin-based therapies (including GLP-1RAs and glucose-dependent insulinotropic polypeptide [GIP]/GLP-1 receptor agonists, excluding dipeptidyl peptidase-4 [DPP-4] inhibitors), or finerenone on one or more cardiorenal outcomes of interest in adults (age ≥ 18 years). DPP-4 inhibitors were excluded from this analysis because major cardiovascular outcome trials have demonstrated non-superiority compared with placebo for major adverse cardiovascular events [3]. Outcomes of interest included: cardiovascular (CV) mortality, all-cause mortality, time to first HF hospitalization or HF event (typically defined as hospitalization, urgent care visit, or oral diuretic intensification), non-fatal MI, non-fatal stroke, major adverse cardiovascular events (MACE), composite of time to first CV death or HF hospitalization/event, and kidney composite outcome. The kidney composite outcome generally includes a combination of decline in estimated glomerular filtration rate (eGFR), development of end-stage kidney disease, need for dialysis or kidney transplant, or death due to renal causes.

Exclusion criteria included studies with no placebo group, fewer than 500 patients, and with follow-up times less than 10 months. Two investigators independently reviewed full texts and extracted study data, with discrepancies resolved by consensus or, when needed, with adjudication by a third reviewer.

All analyses were stratified by subpopulations of interest to enable subgroup analyses and generate more clinically relevant hazard ratios (HRs). These included patients with T2D and ASCVD or high CV risk, CKD, HFrEF, HFpEF, HFpEF with obesity, post-MI, acute HF, and ASCVD with overweight/obesity without T2D.

Meta-analyses were conducted with random effects models using the meta package on RStudio [16]. Given that this review builds upon prior meta-analyses that employed the DerSimonian and Laird (DL) random-effects model, we selected the same model for consistency and comparability for the primary analyses. The Cochrane Handbook for meta-analyses notes that DL is prone to under-estimating between-study variance and may yield narrow confidence intervals (CI), and for which studies have demonstrated superior performance of alternative estimators [17]. Accordingly, we repeated all analyses with restricted maximum-likelihood (REML) estimation and applied the modified Hartung-Knapp (HK) adjustment. A flaw of the HK adjustment is that it can yield overly wide CIs when only a few studies are available and or occasionally overly narrow CIs when between-study variance is 0%. In such cases, there is no universally optimal solution, therefore Cochrane recommends doing sensitivity analysis with multiple statistical models [17]. Whenever there were three or fewer studies or between-study variance (i.e. tau-squared) was 0%, we additionally analysed the data using REML with Wald-type CIs to provide a more balanced estimate of uncertainty.

We conducted a network meta-analysis (NMA) to indirectly compare efficacy of treatment classes, using placebo as the reference. NMA was performed when at least two distinct treatment classes were represented for a given outcome within a defined population subgroup. A random-effects model was fitted using the REML estimator to account for between-study variance and treatment effect estimates were expressed as HRs with 95% CI with HK adjustment. Network diagrams were generated and treatment effects relative to placebo were summarized in forest plots. Because all included trials were placebo-controlled, comparisons between treatment classes were derived through indirect evidence using placebo as the common comparator. All analyses were conducted in RStudio using the netmeta package [18].

Forest plots displayed hazard ratios (HR) with 95% CIs, and a p-value less than 0.05 was considered statistically significant. Presence of heterogeneity was assessed using the chi-squared (Cochran's Q) test (α = 0.05), and its magnitude was quantified using the I² statistic, where I² < 25% indicated low, 25% to 49% indicated moderate, and 50% or greater indicated substantial heterogeneity. Between-study variance was quantified using tau-squared (τ²). Publication bias was assessed using funnel plots and Egger's regression test when more than 10 studies were included, with asymmetry defined as p < 0.05 suggesting potential small-study effects.

The Cochrane Risk of Bias 2.0 (RoB 2.0) tool was used to assess the methodological quality of included trials. The Grading of Recommendations, Assessment, Development and Evaluations (GRADE) framework was used to assess the certainty of evidence across the outcomes in the meta-analysis. Each outcome was independently assessed by two reviewers. The GRADE assessment within the 'impression' component was informed by multiple factors, including the precision, risk of bias, inconsistency, indirectness, and publication bias. Although multiple statistical models were used for sensitivity analyses, GRADE ratings were derived from results using the DL method to maintain consistency with the previous meta-analyses this work builds on.

A total of 1550 studies identified through searches of MEDLINE and CENTRAL were imported into Covidence for screening. An additional three studies were identified through a grey literature search. Additionally, we incorporated data from 22 trials included in the previous meta-analyses on which this review builds [9, 10], comprising 14 trials of SGLT2 inhibitors [19 –32] and 8 trials of GLP-1RAs [33 –40]. Following the removal of duplicates, 1282 unique records remained. After title and abstract screening, 1245 studies were excluded, and 37 proceeded to full-text review. Of these, 33 studies met the inclusion criteria and were selected for data extraction. The study selection process is shown in Fig. 1. Baseline information on the 11 newly included trials [7, 8, 41 –49] can be found in Supplementary Table S2. Additional data were obtained from five post-hoc analyses of the newly included trials [50 –55]. In total, this review included over 200,000 patients across 33 trials: 16 on SGLT2i, 14 on incretin-based therapies, and 3 on finerenone. A table displaying the full titles of all referenced trials, their corresponding acronyms, and subpopulations where trials or their corresponding post-hoc analyses were analyzed are available in Supplementary Appendix Table S1.

The risk of bias (RoB) assessments was completed for the 11 newly added studies using the Cochrane RoB 2.0 tool. Of these, 4 studies were assessed as having "some concerns" for overall risk of bias and 7 were rated as "low risk." The main contributors to "some concerns" were related to reporting of the outcome and measurement of the outcome domains. The cumulative GRADE assessment across the full dataset largely reflects moderate certainty across the majority of outcomes. The updated RoB and GRADE assessments for the new trials are presented in Supplemental Tables S3 and S4. RoB assessments for the other 22 included trials are reported in the prior publications on which this analysis builds [9, 10].

Figures 2, 3, and 4 display HRs and their 95% CIs, along with GRADE certainty, total number of studies, and total population size for each outcome of interest (CV mortality, all-cause mortality, HF hospitalization or HF event, non-fatal MI, non-fatal stroke, kidney composite outcome) within each subpopulation for incretin-based therapies, SGLT2i, and finerenone respectively.

Data for the DL and REML-modified HK models are presented when more than two data points were available for a given outcome. For outcomes informed by ≤ 3 studies or when τ² = 0%, REML-Wald results were additionally reported in the main figures. Data for all statistical models for all outcomes can be found in the supplementary appendix (Supplemental Figures-). S1 S62

Data for the overall population, which pools all available trial data, are presented in the supplementary appendix (Supplemental Figures–); however, these findings should be interpreted with caution given the heterogeneity in study populations. The overall population analysis included data from all eligible studies, with each trial represented once using data from the full study population. Subgroup-level data were not used to avoid double-counting participants. S1 S10

Data for composite outcomes such as CV mortality or HF hospitalization (Supplemental Figures,,,), CV mortality or HF hospitalization/event (Supplemental Figures,,,,,,), and MACE (Supplemental Figures,,,,,,), can be found in the supplementary appendix. If data for HF hospitalization/event were available, they were presented in the main text; if separate data for HF hospitalization alone were available for a given population, they were included in the supplement. (Supplemental Figures,,,,,). S4 S14 S24 S33 S3 S13 S23 S32 S42 S46 S60 S9 S19 S28 S38 S50 S57 S62 S6 S16 S25 S35 S55 S56

Absolute risk reductions per 1,000 patients, along with corresponding 95% CIs, are presented in Supplementary Table. Funnel plots assessing publication bias, with their corresponding Egger's regression test can be found in the supplementary appendix (Supplemental Figures-). S4 S89 S112

Figure 2 summarizes the efficacy of incretin-based therapies across various cardiorenal outcomes. Data is presented for the following subpopulations: (1) T2D with ASCVD/high CV risk, (2) CKD, (3) HFpEF and obesity, (4) ASCVD and overweight/obesity without T2D.

Figure 3 summarizes the efficacy of SGLT2i across various cardiorenal outcomes. Analyses were conducted on the following subpopulations: (1) T2D and ASCVD/high CV risk, (2) CKD, (3) HFpEF, (4) HFrEF, (5) post-MI, and (6) acute HF.

Finerenone was evaluated in two subpopulations: (1) individuals with T2D and CKD and (2) those with HFpEF. Figure 4 summarizes the effects of finerenone on cardiorenal outcomes across these groups.

A network meta-analysis (NMA) was conducted across three populations with available data: (1) T2D with ASCVD/high CV risk, (2) CKD, and (3) HFpEF. For the T2D with ASCVD/high CV risk and CKD NMAs, all three drug classes were included in the analysis. Finerenone data were incorporated into both analyses but primarily reflect patients with diabetic kidney disease (DKD) and should be interpreted cautiously. In the HFpEF NMA, all three classes were compared; however, the incretin-based therapy data predominantly represent patients with HFpEF and overweight/obesity, which should be considered when interpreting the results.

In T2D with ASCVD/high CV risk, SGLT2i reduced HF hospitalization/event when compared to incretin-based therapies (HR 0.78, 95% CI 0.70–0.87), with no significant difference when compared to finerenone (HR 0.90, 95% CI 0.75–1.07) or between incretin-based therapies and finerenone (HR 1.15, 95% CI 0.95–1.38). In T2D with ASCVD/high CV risk, SGLT2i reduced the kidney composite outcome when compared to incretin-based therapies (HR 0.84, 95% CI 0.73–0.96) and finerenone (HR 0.78, 95% CI 0.65–0.94), with no significant difference between incretin-based therapies and finerenone (HR 0.93, 95% CI 0.78–1.11).

In CKD, SGLT2i reduced HF hospitalization when compared to incretin-based therapies (HR 0.72, 95% CI 0.58–0.90), with no significant difference when compared to finerenone (HR 0.84, 95% CI 0.67–1.07) or between incretin-based therapies and finerenone (HR 1.17, 95% CI 0.88–1.56). In CKD, SGLT2i reduced the kidney composite outcome when compared to incretin-based therapies (HR 0.82, 95% CI 0.73–0.91) and finerenone (HR 0.81, 95% CI 0.71–0.92), with no significant difference between incretin-based therapies and finerenone (HR 0.99, 95% CI 0.87–1.12). No significant differences were observed for CV mortality, all-cause mortality, non-fatal MI, or non-fatal stroke in the T2D with ASCVD/high CV risk and CKD subpopulations.

In the HFpEF NMA, there were no significant differences between any of the drug classes for CV mortality, all-cause mortality, or HF hospitalization/event. Similarly, for the kidney composite outcome, no significant difference was observed between SGLT2i and finerenone, and no data were available for incretin-based therapies.

Forest plots, network diagrams, as well data for other composite outcomes (e.g., CV mortality or HF hospitalization, MACE) are presented in the Supplementary Appendix (Supplemental Figure-). S63 S88

This analysis examined the effects of incretin-based therapies on cardiorenal outcomes, including both GLP-1RA, and the dual GIP/GLP-1 receptor agonist tirzepatide. In the T2D population with ASCVD/high CV risk, several outcomes reached statistical significance, partly due to the inclusion of newer incretin-based therapy trials that shifted previously non-significant findings seen in earlier analyses [9, 11]. The addition of data from the FLOW and SOUL trials rendered the reduction in non-fatal MI statistically significant, a result not observed in earlier meta-analyses [9, 11]. In our analysis, there was a non-significant reduction in HF hospitalization among patients with T2D and ASCVD/high CV risk and became statistically significant with the inclusion of broader HF event data from the SOUL and STEP-HFpEF DM trials.

In the CKD population, beyond their renal benefits, incretin-based therapies also led to significant reductions in CV mortality, all-cause mortality, non-fatal MI, and MACE. This finding was in part driven by the addition of relatively newer data from the FLOW trial for CV mortality, all-cause mortality, and non-fatal MI, as well as the inclusion of the FLOW trial and a subset of the SELECT trial for MACE outcomes. This reinforces the growing evidence base supporting the cardioprotective effects of incretin-based therapies, especially in patients with diabetes and CKD. The 2024 Kidney Disease Improving Global Outcomes (KDIGO) guidelines recommend the use of GLP-1RAs in "adults with T2D and CKD who have not achieved individualized glycemic targets despite use of metformin and SGLT2 inhibitor treatment" [58]. The majority of GLP-1RA trials included in this meta-analysis explored outcomes in participants with CKD and concurrent T2D, with the exception of the SELECT trial. Notably, the SELECT trial demonstrated 31% reduction in MACE, suggesting a role for semaglutide in overweight or obese patients without T2D but with CKD. However, additional trials are warranted to evaluate long-term outcomes in non-diabetic patients with CKD of diverse etiologies before incretin-based therapies can be broadly recommended for the wider CKD population.

In the HFpEF population with obesity, incretin-based therapies were associated with a significant 40% reduction in time to first HF hospitalization/event. This finding was driven primarily by the SUMMIT trial, which contributed the greatest weight to the analysis (Supplemental Figure S43). In a post-hoc analysis of the SUMMIT trial [59], there was no statistical difference on the effect of tirzepatide on worsening HF event among patients with and without CKD, suggesting that the benefits of incretins in patients with HFpEF and obesity go beyond baseline CKD status. However, this finding should be interpreted in the context of the known limitations of estimating kidney function, as significant discrepancies between creatinine- and cystatin C–based eGFR can lead to misclassification of CKD status [60]. Ultimately, incretin-based therapies demonstrate promising benefits in patients with HFpEF with obesity, particularly in reducing the risk of HF hospitalization/event. This reinforces its therapeutic potential and warrants further investigation and consideration in future HFpEF management strategies in a subset of patients. However, additional evidence is needed to clarify their efficacy in non-diabetic populations, as most data to date are derived from cohorts with high rates of comorbid T2D.

Data on the effects of incretin-based therapies in individuals with ASCVD and overweight/obesity without T2D were limited to the SELECT trial. Among the outcomes of interest, semaglutide was associated with statistically significant reductions in all-cause mortality, CV mortality or HF hospitalization/event, non-fatal MI, MACE, and the composite kidney outcome [46].

Early results from the SURPASS-CVOT trial, which compared tirzepatide with dulaglutide in patients with T2D and ASCVD, demonstrated noninferiority for MACE, with preliminary evidence suggesting potential benefits in all-cause mortality, glycemic control, weight reduction, and kidney outcomes, with these findings pending full publication [61].

Across diverse patient populations, SGLT2i significantly reduced the risk of several outcomes. In subpopulations with T2D and ASCVD/high CV risk, as well as in those with CKD, SGLT2i lowered the risk of CV mortality, all-cause mortality, HF hospitalization/event, non-fatal MI, MACE, and the kidney composite outcome. Similar to previous analyses, no reduction in non-fatal stroke risk was observed in either the T2D or CKD subpopulations [10]. In the HFrEF subpopulation, significant reductions were seen in CV mortality, all-cause mortality, HF hospitalization/event, and the kidney composite outcome. While benefits were also observed in HFpEF populations, they were more modest, with a significant reduction primarily in time to first HF hospitalization/event.

In the post-MI population, pooled data from DAPA-MI and EMPACT-MI showed a significant reduction in time to first HF hospitalization, driven largely by EMPACT-MI, which contributed over 80% of the weight (Supplemental Figure S42). Notably, subgroup analysis from EMPACT-MI indicated that the benefit was confined to patients without baseline T2D, with no significant effect observed in those with diabetes [43]. The pooled analysis limited to DAPA-MI and the non-diabetic subgroup of EMPACT-MI remained significant, suggesting potential utility of SGLT2 inhibitors in non-diabetic post-MI patients.

Evidence supporting SGLT2i in the peri-acute HF setting was primarily derived from the SOLOIST-WHF trial, which demonstrated a significant reduction in total HF hospitalization events. The DAPA ACT HF-TIMI 68 trial was recently published after the date of our original search was therefore not included in this review. This trial found in-hospital initiation of dapagliflozin in patients hospitalized for acute HF had no statistically significant reduction in the risk of worsening HF events or CV death. However, a prespecified meta-analysis of SOLOIST-WHF, EMPULSE, and DAPA ACT HF-TIMI 68 demonstrated that in-hospital initiation of SGLT2i significantly reduced the risk of worsening HF or CV death by 29% and all-cause mortality by 43% [62].

Finerenone has demonstrated an improved safety and tolerability profile compared to traditional steroidal MRAs in specific patient populations [63, 64]. Of the three RCTs on finerenone included in our analysis, FIGARO-DKD and FIDELIO-DKD focused on populations with T2D and CKD. In contrast, the FINEARTS-HF trial evaluated patients with HF with a LVEF ≥ 40%. This heterogeneity limits the interpretability of pooled effect estimates, as the underlying pathophysiology and therapeutic goals differ between these groups.

In participants with T2D and CKD, finerenone led to a statistically significant reduction in HF hospitalization/event and in the kidney composite outcome. In a pooled analysis of all three studies involving patients with HF and LVEF ≥ 40%, finerenone was associated with a significant reduction in the composite outcome of CV mortality or HF hospitalization. In the analysis of a subpopulation from FIGARO-DKD and FIDELIO-DKD with HF and LVEF ≥ 40%, finerenone led to a significant reduction in HF hospitalization/event. In patients with HFmrEF or HFpEF, finerenone appears to confer clinical benefits across the spectrum of kidney risk. In a prespecified analysis of FINEARTS-HF, finerenone consistently reduced CV mortality and total HF events, and improved HF-related health status (as measured by Kansas City Cardiomyopathy Questionnaire-Total Symptom Score [KCCQ-TSS]), regardless of baseline KDIGO risk category [65].

A meta-analysis of the three finerenone trials, titled FINE-HEART, was published in 2024 and employed individual participant-level data, allowing for harmonized endpoint definitions and more granular time-to-event analyses [66]. Interestingly, it demonstrated a statistically significant 20% reduction in the risk of the kidney composite outcome, largely driven by FIDELIO-DKD and FIGARO-DKD [66]. In contrast, while our trial-level meta-analysis showed a consistent direction of effect, the pooled kidney composite outcome data across the three trials did not reach statistical significance. This discrepancy likely reflects differences in methodology, as FINE-HEART was able to apply consistent outcome definitions across trials and conduct more statistically powerful analyses. Nevertheless, these findings highlight finerenone's cardiorenal efficacy and reinforce its role in patients with overlapping cardiovascular, renal, and metabolic risk.

Our NMA provides a comparative overview of class-level efficacy across T2D with ASCVD/high CV risk, CKD, and HFpEF populations. SGLT2i demonstrated superior efficacy in reducing both HF hospitalization and kidney composite outcome compared to incretin-based therapies in T2D with ASCVD/high CV risk and CKD subgroups. SGLT2i also outperformed finerenone for these outcomes; however, this comparison should be interpreted with caution, as most finerenone data were derived from a broader diabetic kidney disease population. In contrast, no significant differences were observed between drug classes in the HFpEF population across CV mortality, all-cause mortality, or HF outcomes. The NMA findings should be interpreted with caution and viewed as hypothesis generating rather than definitive. While indirect NMA comparisons reached statistical significance for some outcomes, class-level estimates against placebo show overlapping confidence intervals, and head-to-head trial evidence is lacking. Nonetheless, these findings highlight the differential cardiorenal benefits of each medication class depending on the underlying comorbidity, reinforcing the need for phenotype-specific therapeutic strategies, and the importance of determining the optimal sequencing of therapy in patients eligible for dual or triple therapy.

Some limitations should be acknowledged when interpreting the findings of this review. Heterogeneity in study design, patient populations, and endpoint definitions must be considered when evaluating pooled results. Specifically, definitions of HF based on ejection fraction, laboratory cutoffs for CKD, and components of the kidney composite outcome varied across trials, which may impact the consistency and validity of subgroup analyses. In the case of the finerenone trials, the FINEARTS-HF trial differed significantly in its inclusion criteria compared to FIDELIO-DKD and FIGARO-DKD, enrolling a HFpEF population rather than patients with T2D and CKD. This clinical heterogeneity complicates pooled analyses, particularly for kidney outcomes, where the contribution of FINEARTS-HF may not be directly comparable to the other finerenone trials. Combining HF hospitalization data with broader HF event data introduces additional heterogeneity. While this approach broadens event capture, it may reduce the specificity of the outcome. A dedicated analysis focused solely on HF events could help clarify treatment effects more precisely. Additionally, studies were limited for certain subpopulations, particularly those post-MI, acute HF, and ASCVD and overweight/obesity without T2D. Similarly, HFpEF with obesity subpopulation was largely derived from STEP-HFpEF and SELECT, while the finerenone evidence came from relatively few studies. Subgroup findings based on limited number of trials should therefore be interpreted with caution. Subpopulation findings based on a limited number of trials should therefore be interpreted with caution, and in these cases, results can be considered exploratory rather than definitive. Next, this study was not prospectively registered, which limits transparency regarding pre-specified objectives and any post hoc modifications to the analytical approach. Additionally, meta-regression analyses could not be performed due to an insufficient number of trials for the majority of outcomes per subgroup, as such analyses are generally underpowered when conducted with fewer than ten studies [17]. Although a NMA was performed, certain subgroup comparisons are limited by heterogeneity in population definitions. Additionally, NMAs rely on assumptions of transitivity and consistency, which may not be fully satisfied given heterogeneity across trials. GRADE ratings were derived from DL estimates for consistency with prior meta-analyses, though this approach has recognized limitations and may overestimate certainty of evidence. Lastly, Finerenone data largely reflect the DKD population but were applied to both T2D and CKD subgroups, while incretin-based therapy data in HFpEF predominantly represent overweight or obesity with HFpEF populations. Lastly, publication bias was assessed using Egger's test and funnel plots for outcomes with ten or more studies; however, several subgroup and outcome analyses included fewer trials, precluding formal assessment, and thus residual publication bias cannot be excluded.

Future research should focus on underrepresented populations, including various non-diabetic populations, to better define the role of these therapies beyond traditional indications. Data suggest complementary benefits from dual combination therapy [67 –72]; however, randomized trial evidence on hard clinical endpoints, whether for dual or triple therapy, remains limited, highlighting an important area for future investigation. As the evidence base continues to expand, updated clinical guidelines are warranted to reflect the broader applicability of these agents in cardiorenal risk reduction.

Other recent trials have been published, and while not included in our analysis due different patient populations, may contribute to future standards of care. The STRIDE trial evaluated the effects of once-weekly semaglutide on walking capacity, and quality of life in patients with T2D and symptomatic peripheral artery disease (PAD) [73]. Semaglutide significantly improved maximum walking distance, pain-free walking distance, and vascular quality-of-life scores over 52 weeks, and reduced need for rescue interventions [73].​ While event rates were low, semaglutide was associated with a lower incidence of all-cause mortality (1% vs. 2%); however, this difference was not statistically significant [73]. To our knowledge this is the first large RCT to evaluate the effects of a GLP-1RA on functional outcomes in patients with symptomatic PAD, addressing a major therapeutic gap in this high-risk population. The DAPA-TAVI trial evaluated the effects of dapagliflozin patients with HF and aortic stenosis undergoing transcatheter aortic valve implantation (TAVI) [74]. Dapagliflozin significantly reduced the 1-year incidence of the composite of all-cause mortality or worsening HF compared to standard care, driven primarily by fewer HF events [74]. Although subjects in this trial met existing indications for use of SGLT2i, this trial provides important evidence for their use in the elderly post-TAVI population that has been underrepresented in prior cardiovascular outcome trials.