Clinical efficacy and safety of sodium-glucose cotransporter protein-2 (SGLT-2) inhibitor, glucagon-like peptide-1 (GLP-1) receptor agonist, and Finerenone in type 2 diabetes mellitus with non-dialysis chronic kidney disease: a network meta-analysis of randomized clinical trials

Two investigators (J. Guo and Y. Jiang) independently searched and identified relevant studies from various databases, including Cochrane Library, PubMed, EMBASE, Web of Science, CNKI, CQVIP database, and WanFang data, from inception to November 2023. To ensure comprehensive retrieval, a combination of subject words and free words was used. Key search terms included "SGLT-2 inhibitor", "GLP-1 receptor agonist", "Finerenone", "Type 2 diabetes mellitus", "chronic kidney disease", etc. The detailed search strategies for each database are described in Supplementary Appendix 2.

This trial included double-blind RCTs comparing SGLT-2 inhibitor, GLP-1 receptor agonist, and Finerenone or directly with placebo in adults with T2DM and non-dialysis CKD. Studies using other control drugs, studies with repeated publications and incomplete data, studies with eGFR<15 mL/min/1.73 m², studies published in languages other than Chinese or English, and studies using the drug within 3 months before screening were excluded.

The clinical outcomes assessed changes in Glycosylated hemoglobin (HbA1c), estimate glomerular filtration rate (eGFR), low-density lipoprotein cholesterol (LDL-C), systolic blood pressure (SBP), diastolic blood pressure (DBP), and body weight from baseline. Safety endpoints included any adverse events (any AE), urinary tract infections (UTI), Hypoglycemia and Acute Kidney Injury (AKI).

The search results were screened by two blinded independent researchers (J. Guo and Y. Jiang) according to the inclusion and exclusion criteria, and the abstracts of the remaining literatures were reviewed using EndNote20. Two researchers used standard data extraction tables for information extraction, judgment, and literature extraction information including: study author, publication year, intervention measures, outcomes, etc. In case of disagreement, a third researcher (W. Zhang) assisted in making a judgment.

The risk of bias assessment was conducted by 2 researchers (J. Guo and Y. Jiang) using the Cochrane Risk of Bias Assessment tool (RoB 2.0) (Sterne et al., 2019). Each study was classified as having low risk, some concerns or high risk of bias.

For outcome indicators, odds ratio (OR) was used for bicategorical variables, mean difference (MD) was used for continuous variables, and 95% confidence interval (95% CI) was used to represent statistical results. The findings were considered statistically significant if the 95%CI did not include the null value (0 for MD and 1 for OR). For each result were calculated using a random effects model. Bilateral P-values <0.05 were considered statistically significant. We evaluated the between-study heterogeneity using the I² statistic and its associated p-values. Specifically, I² values of 25%, 50%, and 75% were indicative of low, moderate, and high levels of statistical heterogeneity, respectively. Subsequently, subgroup analyses were conducted to investigate potential sources of this heterogeneity. STATA 17.0 was used for statistical analysis, evidence network and surface under the cumulative raking curve (SUCRA). We evaluated publication bias of articles using funnel plots and Egger's test. We examined the confidence of evidence using the CINeMA (Salanti et al., 2014) web application, which grades the confidence of the results as high, moderate, low, and very low.

A total of 4,929 articles were retrieved. After removing duplicates, 3,182 remained. Following the title and abstract review, 1719 were selected for full text reading. Ultimately, 39 studies involving 99,599 patients were included: 23 used SGLT-2 inhibitor, 4 used GLP-1 receptor agonist and 12 used Finerenone (Figure 1). Baseline characteristics were comparable between groups. The characteristics of the included studies are shown in Table 1. Figure 2, and Supplementary Appendix 3 shows the network. (Barnett et al., 2014; Haneda et al., 2016; Curovic et al., 2022; Yale et al., 2013; Chen, 2016; Yamout et al., 2014; Pollock et al., 2019; Cherney et al., 2016; Kohan et al., 2014; Wada et al., 2022a; Wada et al., 2022b; Takashima et al., 2018; Dagogo-Jack et al., 2021; Fioretto et al., 2018; Cherney et al., 2021; Allegretti et al., 2019; Cherney et al., 2023; Sivalingam et al., 2024; George et al., 2018; Gao, 2022; Wanner et al., 2018; Sarafidis et al., 2023; Cristian et al., 2020; Bakris et al., 2020; Bakris et al., 2015; Koya et al., 2023; Agarwal et al., 2022a; Agarwal et al., 2023; Mahaffey et al., 2019; Bhatt et al., 2021; Agarwal, et al., 2022b; Zhang et al., 2023; Rosas et al., 2023; Perakakis et al., 2024; Tuttle et al., 2022; Pitt et al., 2021; Bakris et al., 2020; Filippatos et al., 2021; Perkovic et al., 2019).

The RoB 2.0 was employed to evaluate the risk of bias in the 39 included studies, among which one study was classified as "high risk", eight studies were categorized as having "some concerns", and the remaining thirty studies were deemed to have a "low risk". The distribution of each category is presented in Figure 3, while the quality assessment for each individual study can be found in Supplementary Appendix 4.

The consistency test results show that the network has good consistency (P > 0.05), and the loop-specific method did not suggest any inconsistency between closed loops. In addition, the node-segmentation approach did not suggest statistical inconsistency for any outcome (). Supplementary Appendix 5

Assessment of outcome measures showed that 100% of the evidence was rated as low or very low (). The funnel plot and Egger's test indicated publication bias for eGFR (P = 0.007) (). Heterogeneity was observed in HbA1c, body weight, and eGFR. Subgroup analyses suggested that study country, medication timing, numbers of participants and drug category may contribute to this heterogeneity (). Supplementary Appendix 20 Supplementary Appendix 17 Supplementary Appendix 21

SGLT-2 inhibitors, GLP-1 receptor agonists, and Finereone are recommended as preferred pharmacotherapies for T2DM patients with CKD; however, the guidelines do not specify a clear hierarchy of preference among these options (Chinese Diabetes Society, 2025; American Diabetes Association Professional Practice Committee, 2023). Given the widespread clinical application of these drugs, we assert that comparing their clinical efficacy holds significant importance. Therefore, in the absence of direct comparative evidence, we conducted an indirect comparison to determine the clinical efficacy and safety of SGLT-2 inhibitors, GLP-1 receptor agonists and Finerenone in patients with T2DM and non-dialysis CKD. The study has identified several significant findings. The drug Semaglutide (a GLP-1 receptor agonist) appears to be the optimal choice among all drug classes for patients with high HbA1c and poor eGFR, when compared to the other two drugs. The efficacy of SGLT-2 inhibitors and GLP-1 receptor agonists in reducing body weight and HbA1c surpasses that of Finerenone. Canagliflozin, Empagliflozin (SGLT-2 inhibitors), and Semaglutide are considered the most suitable medications for patients with high body weight and elevated HbA1c levels, specially Canagliflozin. Specifically, Empagliflozin, Dapagliflozin and Canagliflozin (SGLT-2 inhibitors) demonstrate remarkable effectiveness in managing high SBP. The use of Empagliflozin and Canagliflozin may be considered for patients presenting with high DBP as the primary accompanying symptom. Liraglutide (a GLP-1 receptor agonist) may be prioritized for patients with elevated LDL-C levels. Although there is no obvious advantage in reducing the indicators and related risk factors in patients with T2DM with non-dialysis CKD, Finerenone has less damage to eGFR, and has a convincing reduction in the occurrence of Hypoglycemia events, and the incidence of adverse events is low, so it can still be used as the first choice for some patients.

The efficacy and safety of these drugs have been extensively deliberated. In an NMA comprising 816 randomized controlled trials with 471,038 participants, SGLT-2 inhibitors, GLP-1 receptor agonists, and Finerenone exhibited favorable outcomes in terms of reducing all-cause mortality and enhancing cardiorenal results among patients diagnosed with T2DM (Shi et al., 2023). 3 studies confirmed that SGLT-2 inhibitors have a positive effect on cardio-renal outcomes in T2DM patients with CKD compared with GLP-1 receptor agonists or Finerenone (Yamada et al., 2021; Nguyen et al., 2023; Apperloo et al., 2024). However, these studies either focused solely on T2DM patients or had a limited number of RCTs and study indicators, leaving the impact of SGLT-2 inhibitors, GLP-1 receptor agonists, and Finerenone on non-dialysis CKD patients with T2DM unclear. Therefore, we designed an NMA to evaluate the effects of several drugs on various clinically accessible indicators.

Benefits of most SGLT-2 inhibitors include renal safety, reduction in HbA1c, blood pressure, and body weight. Hyperglycemia is linked to an increase in eGFR as compensation (Thomas, 2014). Hypertension and overweight increase the risk of T2DM. Therefore, the optimal antidiabetic drug should not only have good glucose-lowering ability, but also be beneficial for body weight, blood pressure, and renal function (Leehey et al., 2015; Garvey, 2022; American Diabetes Association, 2020). The renal protective mechanism of SGLT-2 inhibitors may be dominated by direct effects on renal vessels. Empagliflozin decreases proximal tubular sodium reabsorption, increases distal sodium delivery to the macula densa, and activates glomerular feedback. This process reduces hyperfiltration and regulates vascularization (Wanner, 2017). In addition, SGLT-2 inhibitors can directly act on the kidney, reduce renal fibrosis by inhibiting oxidative stress in the kidney, (Woods et al., 2019), attenuate the increase of angiotensinogen, and reduce NLRP3 inflammasome activity (Yaribeygi et al., 2019), these mechanisms may explain how SGLT-2 inhibitors reduce blood pressure. The short-term use of SGLT-2 inhibitors may temporarily decrease eGFR, but it will gradually recover, indicating a long-term protective effect on eGFR (Barnett et al., 2014). This may be due to reduced uric acid, inhibited inflammatory response, and decreased vascular stiffness (Heerspink et al., 2016; Chilton et al., 2015). Although previous studies have suggested that angiotensin converting enzyme inhibitors (ACEI) or angiotensin II receptor blocker (ARB) are the preferred antihypertensive drugs for T2DM patients with CKD, they do not significantly reduce blood glucose levels and their effects remain limited (Bakris et al., 2000). The potential of using SGLT-2 inhibitors as the first choice after ACEI and ARB drugs, considering its high cardio-renal safety and hypoglycemic function, needs further discussion.

Unexpectedly, we found Canagliflozin, an SGLT-2 inhibitor agent, to be the most effective for weight reduction (SUCRA = 90.2%), superior to GLP-1 receptor agonists. The reason for this may be due to different principles of action. The hormone GLP-1, produced in cells lining the intestines, slows down digestion and reduces food intake by inhibiting neural activity in the brain (Drucker, 2016). SGLT-2 inhibitors can also lead to weight loss depending on the dosage, as excess glucose is eliminated from the body (Brown et al., 2019). At the same time, it can also accelerate fat burning by lipolysis and fatty acid oxidation (Vallon et al., 2017). Therefore, it can be hypothesized that the progressive increase in blood glucose levels and concurrent decrease in body weight may account for the superior weight reduction observed with SGLT-2 inhibitors compared to GLP-1 receptor agonists.

GLP-1 receptor agonists, especially Semaglutide and Liraglutide, showed the greatest advantage in reducing LDL-C (SUCRA = 100%) and HbA1c (SUCRA = 72.0% and 56.4%), and had a higher safety profile against eGFR (SUCRA = 79.5% and 53.4%). The management of LDL-C is crucial in preventing cardiovascular events caused by atherosclerotic plaque formation (Giglio et al., 2021). The existing NMA highlights the remarkable cardiovascular benefits of GLP-1 receptor agonists (Zhang Y. et al., 2022). However, our study suggests that these advantages may be attributed to specific pathways that lower LDL-C and reduce lipid deposition in the cardiovascular system. Nevertheless, further testing is required to confirm these pathways.

Finerenone is a novel non-steroidal mineralocorticoid receptor antagonist (MRA), and its safety in cardiorenal prognosis has been proved (Agarwal et al., 2022b; Zhang M-Z. et al., 2022), this may be due to the reduction of proteinuria and tissue inflammation and fibrosis (Wish and Pablo, 2022). Our study confirmed Finerenone's safety and highlighted its significant advantage in reducing SBP, while showing no notable effect on blood glucose reduction. The combination of Finerenone with SGLT-2 inhibitors or GLP-1 receptor agonists have been suggested to enhance anti-inflammatory, anti-oxidative stress, and endothelial protection effects (Lv et al., 2023). At the same time, SGLT-2 inhibitors inhibit sodium reabsorption, while Finerenone promotes sodium retention (Lv et al., 2023; Kidokoro et al., 2019). Whether the pathways of SGLT-2 inhibitors and Finerenone partially coincide, lead to water and salt metabolism disorders, and increase additional adverse reactions still needs more clinical trials to prove its safety.

Certain limitations should be noted when interpreting our study. The results of our study indicate a higher risk of urinary tract infections in patients treated with GLP-1 receptor agonists compared to those receiving SGLT-2 inhibitors therapy, contradicting previous research findings (Tanrıverdi et al., 2023; Drucker, 2016). The effect of GLP-1 receptor agonists on UTI occurrence was only reported in one literature, which included a small number of patients and specified clear inclusion criteria as "T2DM and albuminuria" (Sivalingam et al., 2024). The presence of proteinuria may influence UTI incidence. The eGFR safety profile of most GLP-1 receptor agonists are good, except for Exenatide. From April 2005 to August 2008, 78 cases of renal disease caused by Exenatide were reported to the FDA. Gastrointestinal adverse reactions are the most common side effects of GLP-1 receptor agonists, which can cause significant fluid loss and pre-renal acute failure (Bzowyckyj, 2020). At the same time, GLP-1 receptor agonists can simultaneously enhance renal sodium efficacy, leading to renal hypoperfusion and AKI development (Skov et al., 2013). Interestingly, our study does not support a higher likelihood of Exenatide causing eGFR damage (SUCRA = 56.4%), possibly due to its ability to reduce renal pathological material deposition. Animal experiments have demonstrated that Exenatide reduces inflammatory and apoptotic cell infiltration in the glomerulus of mice, as well as lipid content (Park et al., 2007). Activation of the GLP-1 receptor stimulates adenylyl cyclase, leading to increased cAMP (a key mediator of GLP-1-induced insulin secretion) production (Fujita et al., 2014). In conclusion, Exenatide remains a controversial drug and should be minimized in patients with severe renal insufficiency.

The advantages of this network meta-analysis are as follows: first, we included a large number of literatures with a large sample size and reliable data; secondly, on top of the existing analysis, we provided additional evidence to specifically analyze the advantages and disadvantages of different SGLT-2 inhibitors and GLP-1 receptor agonists and Finerenone in the treatment of T2DM patients with non-dialysis CKD to facilitate clinical decision making by physicians. Last but not least, the ADA recommends that individuals with T2DM and CKD should use SGLT-2 inhibitors or GLP-1 receptor agonists as their first choice, but it does not specify which medicine to use for patients with different clinical priorities (Committee, 2024). Our study identified the efficacy of both drugs in improving HbA1c and eGFR, while also providing more favorable evidence regarding the benefits and limitations of various drugs in other indicators such as LDL-C, SBP, DBP, as well as the safety profile of Finerenone.

Limitations of our NMA are largely driven by the available evidence. Firstly, it is acknowledged that the heterogeneity and inherent bias within the literature are objective realities, which may potentially compromise the accuracy of research outcomes. For instance, several included trials lacked baseline data, and variations existed in population characteristics, duration of pharmacological treatment, and follow-up periods. This may have led to some bias. Third, the dose of the study drug and the level of detail in our study were not considered. Finally, the literatures included in this paper provide only indirect comparisons between drugs, lacking direct comparison evidence, which affects the credibility of the results. More detailed studies are needed to supplement this aspect in the future.