Background
Diabetes mellitus (DM) is one of the major causes of death in humans and doubles the risk of cardiovascular disease (CVD) in the United States (US) [
1]. The 2008 US Food and Drug Administration antidiabetic drug guidance mandated that novel antihyperglycemic medications should demonstrate cardiovascular safety through large cardiovascular outcome trials (CVOTs) [
2]. Accordingly, CVOTs comparing dipeptidyl peptidase-4 inhibitors, sodium–glucose cotransporter-2 inhibitors (SGLT-2is), and glucagon-like peptide-1 receptor agonists (GLP-1RAs) with placebo on a background of standard of care have been conducted [
3,
4]. Among them, certain SGLT-2i and GLP-1RA compounds have shown not only safety but superiority in their effects on cardiovascular outcomes [
5].
Of note, there are some concerns about adverse events in SGLT-2i and GLP-1RA therapy in clinical practice. In the early days, many doctors were reluctant to use SGLT-2i in patients with reduced renal function or older patients, particularly because of concerns about adverse effects on the kidney. However, recent trials with SGLT-2i in patients with chronic kidney disease have proven significant efficacy on composite renal outcomes [
6‐
8]. Moreover, there was no difference in the cardiovascular benefits from SGLT-2i therapy between younger and older patients [
9]. Moreover, SGLT-2i therapy seemed to be more effective regarding composite cardiovascular outcomes in patients with preexisting CVD [
10,
11].
Physicians tend to avoid prescribing GLP-1RA to old people and those with low body mass index (BMI) because of its gastrointestinal adverse events [
12]. However, more evidence is needed to support this practice pattern. On the contrary, it was speculated that people with overweight or obesity respond well to GLP-1RA because the therapy can reduce weight. In addition, some GLP-1RA types, such as dulaglutide and liraglutide, showed benefits in composite renal outcomes in the CVOTs [
13,
14]. However, distinctive beneficial effects of GLP-1RA on cardiovascular outcomes in patients with obesity or those with renal impairment vs. those without these conditions have not been established yet.
Taken together, there is a clinical interest in determining patient factors that predict a differential therapeutic response to SGLT-2i and GLP-1RA (in the sense that subgroups are identified who respond better to one and worse to the alternative treatment).
To the best of our knowledge, no systematic approach has ever been taken to identify the patient characteristics related to the effectiveness of these two agents. Therefore, the current study was designed to compare the 3-point major adverse cardiovascular event (3P-MACE) risk reduction with SGLT-2i and GLP-1RA in subgroups of patients participating in the large CVOTs. Through this approach, we aimed to suggest recommendations for individualizing the choice between these two cardioprotective medications in patients with type 2 DM (T2DM).
Methods
Data sources and study selection
We conducted a systematic review and meta-analysis following the updated guidance in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA 2020) [
15]. Neither ethics approval nor patient consent was required for this analysis. The review was registered in PROSPERO (CRD42021235989).
Databases searched included MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials. The latest searches were conducted in December 2022, focusing on the period from January 2008 to December 2022. The search terms included the following keywords: ‘type 2 diabetes’ for the population; ‘sodium-glucose cotransporter 2 inhibitor’ and ‘glucagon-like peptide-1 receptor agonist’ (or the individual compounds in these classes) for the study intervention; ‘major adverse cardiovascular event’ and ‘cardiovascular event’ for outcomes. Searches were restricted to clinical trials and those reported in the English language.
We included trials using the following inclusion criteria: (1) evaluation of an antidiabetic agent compared with placebo; (2) report of 3P-MACE for the overall population and subgroups (grouped by patient characteristics); (3) a minimum number of 1,000 T2DM subjects enrolled; and (4) multinational randomized controlled trials (RCTs) with a wide representation of ethnic backgrounds.
Outcomes
The main outcomes were the relative and absolute risk reductions (RRR and ARR) for 3P-MACE by subgroup (based on patient characteristics presented at baseline) comparing SGLT-2i and GLP-1RA treatment, aiming to identify differential responses. The baseline characteristics included age, sex, BMI, initial HbA1c level, baseline renal function (estimated glomerular filtration rate, eGFR), presence of albuminuria, and preexisting CVD or heart failure (HF): 3P-MACE combined cardiovascular death, nonfatal myocardial infarction, and nonfatal stroke.
Data were extracted independently by two authors (M.S. and S.L.). The following data were extracted for the included studies: name of the trial, year of publication, number of patients, patient characteristics, comorbidity status, and concurrent medication. To check the population risks for CVDs, event rates of 3P-MACE were collected from each trial. The outcomes included the event rate, event number, and sample size of the group.
Data synthesis and analysis
Data were analyzed using R software (version 4.1.0; R Development Core Team, Vienna, Austria) with the ‘metafor’ package. Included trials were assessed for the risk of bias using the Cochrane risk of bias 2 tool. The I2 statistic was used to assess the overall heterogeneity of all the comparisons, with values of under 25%, 50%, and 75% corresponding to mild, moderate, and high heterogeneity, respectively.
ARR and RRR were calculated using the number of events and patient-years of observation [
16], and then used for pooled estimates and their 95% confidence intervals (CIs). If these data were not explicitly reported, they were approximated: event rate = [(number of people with the event [n]) / ([total person-years of observation] × 100]).
We performed a meta-regression analysis of the selected trials to assess the relationship between the patient characteristics at baseline (based on mean values, e.g., for BMI, or proportions with a characteristic of interest, e.g., patients with preexisting CVD) and the corresponding ARR and RRR for 3P-MACE. We also estimated the strength of the association of the characteristics by estimating R2. The ‘rma’ function was used to examine the effects of drug types as moderators. The outcomes among different baseline subgroups or therapies were compared using frequentist meta-analysis with random-effects models.
Quality of evidence
The Grading of Recommendations Assessment, Development and Evaluation (GRADE) method was used to assess the quality and strength of the evidence for each subgroup. Two authors (M.S. and S.L.) rated the quality of the evidence for each outcome independently. We used GRADEpro software (McMaster University and Evidence Prime Inc., Hamilton, Ontario, Canada) to generate evidence profile tables. An I2 value over 50% was regarded as an indication of serious inconsistency. Imprecision was assessed as serious if the reported subgroups were less than half and as very serious when only one trial reported the results. In terms of 3P-MACE, none of the meta-analyses were considered to have serious indirectness.
Discussion
In CVOTs, SGLT-2i and GLP-1RA therapies were significantly and similarly effective in 3P-MACE reduction, providing benefits of 13% and 15% RRR and 0.55 and 0.67 ARR (event/100 person-years of observation), respectively. In the meta-regression of 13 randomized placebo-controlled trials examining SGLT-2i and GLP-1RA treatments, there was a negative association between poorer renal dysfunction (decreased eGFR) and greater ARR for 3P-MACE. The presence of albuminuria was also linked to a greater ARR with SGLT-2i therapy. In the meta-analysis by eGFR subgroups, SGLT-2i therapy was more effective in reducing 3P-MACE in people with eGFR < 60 ml/min/1.73 m
2 compared with those with normal renal function (Fig.
3). At the same time, this difference was much less prominent with GLP-1RA therapy.
Both reduced eGFR and albuminuria are independently associated with a higher risk of cardiovascular events in people with T2DM [
22]. Recent studies on people with established chronic renal failure have proven that SGLT-2i effectively reduces cardiovascular fatality and a composite renal outcome [
6,
7]. Thus, in addition to the previously reported renal benefits of SGLT-2i therapy in patients with T2DM [
23‐
25], our data support the use of SGLT-2i therapy in this subgroup at high risk for both cardiovascular and renal complications.
It is obvious that both SGLT-2i and GLP-1RA therapy help reduce 3P-MACE in patients with renal impairment, who generally exhibit a high risk for CVD [
26]. Intriguingly, the ARR in people with normal renal function seemed to be larger with GLP-1RA therapy than with SGLT-2i therapy (–0.68 vs. − 0.17; P = 0.06; Fig.
3). However, it should be noted that GLP-1RA trials included a relatively narrow spectrum in the proportion of people with eGFR < 60 ml/min/1.73 m
2 than SGLT-2i trials (21.6–31.6% vs. 7.4–100%, P < 0.05). Thus, it might have been difficult to identify the different effects of GLP-1RA on 3P-MACE reduction according to the eGFR level. On the contrary, the lack of benefit observed in GLP-1RA therapy in patients with low eGFR may have resulted from insufficient statistical power caused by the narrow range of eGFR in the study participants [
27]. Taken together, our analysis data including the most available studies indicate that GLP-1RA therapy may be beneficial even for people with normal renal function, which is not the case for SGLT-2i therapy.
Liraglutide therapy in the LEADER trial reduced 3P-MACE significantly more in patients with eGFR < 60 ml/min/1.73 m
2 than in those with ≥ 60 ml/min/1.73 m
2 [
21], while other GLP-1RAs such as exenatide, albiglutide, and semaglutide tended to reduce 3P-MACE more in patients with normal eGFR [
19,
28,
29]. This finding suggests that individual GLP-1RAs might have varying effects on 3P-MACE reduction according to the baseline renal function, but data are not enough to draw a conclusion indicating a drug-specific efficacy of GLP-1RA rather than the class effect. More studies with participants with a wide spectrum of eGFR are needed.
Subjects with albuminuria at baseline showed a tendency of greater reduction in the 3P-MACE by both SGLT-2i and GLP-1RA therapies. Treatment with SGLT-2is is known to significantly reduce albuminuria [
30,
31], and reduction in albuminuria in the first year was associated with long-term cardiovascular benefits [
31]. Among SGLT-2i trials, sotagliflozin therapy in SCORED showed remarkable efficacy in 3P-MACE reduction in patients with albuminuria [
8]. Since the SCORED trial included only patients with low eGFR, with 65.0% of them had albuminuria, it can be speculated that the protective effect of SGLT-2is could be more pronounced in such patients [
32]. Only the LEADER trial reported the subgroup results by albuminuria, which hinders the gathering of important data regarding the potential influences of albuminuria on the cardioprotective effects of GLP-1RA class.
Regarding albuminuria, renin–angiotensin–aldosterone system (RAAS) blockers might be intertwined in response to SGLT-2i or GLP-1RA therapy. However, RAAS blockers were used in over 80% of all 13 trials without significant differences (Table
1). In addition, no associations between the use of RAAS blockers and the efficacy of SGLT-2i or GLP-1RAs on 3P-MACE were observed in the meta-regression analysis.
Other characteristics, including age, sex, BMI, HbA1c level at baseline, preexisting CVD, or preexisting HF status, did not significantly affect the effects of SGLT-2i and GLP-1RA therapies on the ARR or RRR of 3P-MACE. However, GLP-1RA therapy tended to be more effective in the ARR of 3P-MACE than SGLT-2i therapy, with borderline significance in the younger age group (Supplementary Figure S6). It is also noteworthy that, compared with the SGLT-2i therapy, GLP-1RA therapy tended to exhibit better efficacy for 3P-MACE reduction in people with uncontrolled diabetes vs. those with controlled diabetes (Table
2).
A previous study with a different approach reported that people with established CVD showed a greater reduction in 3P-MACE by GLP-1RA and SGLT-2i therapies compared with those with risk factors alone (difference in effect between patients with vs. without a history of CVD: P = 0.049) [
33]. However, that study included CVOTs up to June 2019. In our study, which included the CVOTs up to 2022, CVD was associated with a greater reduction in 3P-MACE by SGLT-2i and GLP-1RA. There was no statistically significant difference in the between-group analysis. Taken together, it is obvious that more trials are needed to confirm such effects of patient characteristics. Still, our data do not support the differential use of SGLT-2is and GLP-1RAs with the expectation to improve the prognosis by individualizing treatment considering these baseline characteristics.
The present analysis could be improved in a number of ways. First, the relatively small number of studies is likely to reduce the power needed to find significance between subgroups, particularly in meta-regression analysis [
34]. Thus, nonsignificant results cannot rule out the significant impact of certain characteristics on the efficacy of SGLT-2i or GLP-1RA therapy in 3P-MACE reduction. Second, studies that did not report the 3P-MACE results by subgroups were not included for the analysis, reducing the statistical power of the analysis. Some subgroups were limited in representation: e.g., studies with GLP-1RAs in patients with impaired renal function at baseline. Third, each compound was only tested against placebo, and there were no head-to-head comparison studies between SGLT-2is and GLP-1RAs regarding cardiovascular effects. Thus, all comparisons are indirect and may be confounded by differences in unidentified patients characteristics, such as concomitant medical therapy addressing hypertension, lipid abnormalities, and other conditions typically associated with T2DM. Nonetheless, we included all large CVOTs currently available, and we believe that the results obtained by meta-regression and meta-analysis provide meaningful information about the proper use of these two novel agents.
Conclusions
In the meta-regression and meta-analyses, SGLT-2i therapy was more effective in the ARR of 3P-MACE in patients with decreased renal function (significant) or albuminuria (trend) than in those without. In contrast, the beneficial effects of GLP-1RA therapy on the ARR of 3P-MACE did not differ for these subgroups. However, the effect of GLP-1RA therapy was more pronounced in patients with high HbA1c levels at baseline than in those without.
Both classes showed a significant positive association between the efficacy of the reduction of 3P-MACE and the proportion of patients with reduced renal function. SGLT-2i therapy was effective in the 3P-MACE in both groups of patients with normal and reduced eGFR, with greater efficacy in reduced eGFR. In contrast, the efficacy of GLP-1RA therapy of 3P-MACE was similar in both groups, but significant only in those with normal eGFR. Our findings support the use of SGLT-2i in patients with impaired renal function and GLP-1RA in patients with normal renal function for optimizing the differential prescription to prevent major adverse cardiovascular events.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.