Meta-analyses of Results From Randomized Outcome Trials Comparing Cardiovascular Effects of SGLT2is and GLP-1RAs in Asian Versus White Patients With and Without Type 2 Diabetes

Randomized trial results and their meta-analyses have demonstrated the cardiovascular (CV) benefits of novel antihyperglycemic medications for type 2 diabetes such as sodium-glucose cotransporter 2 inhibitors (SGLT2is) and glucagon-like peptide 1 receptor agonists (GLP-1RAs) (1,2). Furthermore, certain populations such as Asians, which form nearly 60% of the world’s population, including South Asians and East/Southeast Asians (here collectively categorized as Asian because of the way most trials categorize race), experience a greater burden of type 2 diabetes compared with Whites, although they vary markedly in their risk of atherosclerotic CV disease (3,4). Results from a recent meta-analysis pooling data for Asians with type 2 diabetes also indicated a significant reduction in major adverse CV events (MACE) with GLP-1RAs but not with SGLT2is, compared with placebo. However, this study did not directly compare Asians versus Whites and did not consider all relevant SGLT2i trials (5).

In the present report, we meta-analyzed summary hazard ratio (HR) estimates from CV outcome trials (CVOTs) of SGLT2is and GLP-1RAs for effect modification of the efficacy of study drugs by race (Asian vs. White) with regard to selected CV outcomes. Of the antihyperglycemic drug classes, SGLT2is and GLP-1RAs are the only ones to show consistent benefits in CV outcomes, and therefore, we set out to examine these two drug classes only.

A systematic review was performed in PubMed from 1 January 2015 to 8 December 2020 to identify published large randomized placebo-controlled trials of SGLT2is and GLP-1RAs (6). Our search strategy is detailed in our Supplementary Fig. 1, Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) diagram.

Inclusion criteria included CVOTs of SGLT2is and GLP-1RAs providing effect size estimates (HRs [95% CIs]) in the subgroups of interest (i.e., Asians and Whites) for the outcomes of 1) MACE in patients with type 2 diabetes, and 2) CV death/hospitalization for heart failure (HHF) in patients with HF and reduced ejection fraction (HFrEF).

HRs (95% CIs) for the outcomes of interest in Asians and Whites were extracted from each study by at least two researchers independently, adhering to PRISMA guidelines (7). Risk of bias was assessed using version 2 of the Cochrane risk-of-bias tool for randomized trials (8) (Supplementary Fig. 2).

We used the statistical software Stata/SE 16.0, using the Stata command “meta” to perform a random-effects model (DerSimonian and Laird method) meta-analysis (9). Our primary analyses compared CV outcomes reported by race of Asian and White, in addition to tests for group differences (P_interaction).

We identified 11 SGLT2i trials and 7 GLP-1RA trials (10–26) (Supplementary Fig. 1 PRISMA diagram).

Two SGLT2i trials (DECLARE-TIMI 58 and SOLOIST-WHF) were excluded because the HR (95% CI) was not available for the subgroups of Asian and White patients (18,19). In SOLOIST-WHF, in the Asian subgroup, the HR could not be estimated because of 0 events in one of the treatment groups (19). Two other SGLT2i trials in patients with chronic kidney disease (CREDENCE and DAPA-CKD) were excluded because the reported outcomes available for Asian versus White patients were not MACE or CV death/HHF (13,14). One SGLT2i trial (SCORED) was excluded because it had a different population (i.e., not HF) from the other two trials that reported the CV death/HHF outcome (15).

Six trials of SGLT2is were analyzed (Table 1); four diabetes trials (EMPA-REG OUTCOME, CANVAS Program [CANVAS and CANVAS-R], and VERTIS CV) reported MACE, and two HF trials (Dapa-HF and EMPEROR-Reduced) reported CV death/HHF (10–12,16–17).

In four SGLT2i outcome trials including 3,298 Asian and 20,258 White patients with type 2 diabetes, the HR (95% CI) for the MACE outcome in Asian versus White patients was 0.81 (0.57, 1.04) vs. 0.90 (0.80, 1.00), respectively (P_interaction = 0.46) (Fig. 1).

In two SGLT2i outcome trials including 1,788 Asian and 5,962 White patients with HF and reduced ejection fraction (HFrEF), the HR (95% CI) for CV death/HHF in Asian versus White patients was 0.60 (0.47, 0.74) vs. 0.82 (0.73, 0.92), respectively (P_interaction = 0.01) (Fig. 2).

One GLP-1RA trial (ELIXA) was excluded because the HR (95% CI) was not extractable for the subgroups of Asian and White patients (20).

Six trials of GLP-1RAs were analyzed (Table 1); LEADER, SUSTAIN-6, EXSCEL, HARMONY OUTCOMES, REWIND, and PIONEER 6 (21–26) reported MACE.

In six GLP-1RA trials with 4,195 Asian and 37,530 White patients, the HR (95% CI) for the MACE outcome in Asian versus White patients was 0.68 (0.53, 0.84) vs. 0.87 (0.81, 0.94), respectively (P_interaction = 0.03) (Fig. 3).

Results of clinical outcome trials of SGLT2is and GLP-1RAs comparing CV outcomes in Asians versus Whites suggest differential treatment effects of SGLT2is (in those with HFrEF) and GLP-1RAs by race, with greater benefits of both classes in Asians. Because the interaction between Asian race and outcome seems consistent across different groups of studies, particular for the GLP-1RA class, it seems unlikely that random variation in baseline characteristics between trial arms explains the results.

The recent meta-analysis by Singh and Singh (5) of three SGLT2i trials found no significant reduction in MACE, HHF, or CV death in Asians and postulated whether these results were due to low statistical power, underrepresentation of Asians, or true ethnic difference. However, there seems to be benefit in terms of reduction in risk of CV death/HHF when the two HFrEF trials are meta-analyzed. This suggests that SGLT2is lessen risk in Asians at least as well as in Whites and potentially better in patients with HFrEF. Perhaps more significantly, we show GLP-1RA–derived MACE benefits are significantly (P_interaction = 0.03) better in Asians (mean [95% CI] risk reduction 32% [16–47%]) than in Whites (13% [6–19%]), with remarkably lower HRs in Asians versus Whites in all six trials. That is, GLP-1RAs have an ∼2.5-fold larger relative risk reduction for the MACE end point in Asians compared with Whites.

These findings are all hypothesis generating, and several limitations should be considered before concluding that such therapies, particularly GLP-1RAs, lessen MACE risks more in Asians. Firstly, the Asian subpopulation is often derived from several countries that collectively have a heterogeneous racial makeup, including East, South East, and South Asian subgroups. Although all Asian subgroups are at elevated type 2 diabetes risk compared with Whites, their atherosclerotic CV disease risk can vary markedly, being higher in South Asians versus Whites but lower in other Asian subgroups (4). However, the findings are complex; we noted lower CV mortality in South Asians compared with Whites in a recent observational study of people with type 2 diabetes (27). Furthermore, because of a lack of breakdown of baseline characteristics by race in trials, it is not known to what extent Asian populations were treated or pretreated with standard preventative therapies in each of the trials examined. It is possible that some Asian subgroups were less likely to receive comparative preventive therapy compared with Whites, leading to an apparent accentuated benefit seen with SGLT2i and/or GLP-1RA therapy in this group, although even if this were true, such an observation would be clinically important.

Alternatively, there may be other genuine biological explanations for these differences that merit future investigation. For example, because diabetes onset is far earlier in Asian subgroups and occurs at lower BMI (28,29), it may be that MACE or other cardiac risks driven by glycemia-affected pathways are more pronounced in Asian groups and that such pathways are better targeted by novel diabetes therapies. This is speculative, however, and more work is needed to determine if differences are genuine and, if so, identify potential mechanisms, including effects on multiple risk pathways.

There may have been differences in the proportion of Asian versus White patients with type 2 diabetes and chronic kidney disease within the HF trials, although this information was not routinely available. A recent meta-analysis by Zannad et al. (30) of the two SGLT2i trials in patients with HFrEF showed no clear difference in outcome by either diabetes status or eGFR levels ≥ or <60 mL/min/1.73 m². Therefore, it seems unlikely that these characteristics explain any potential differences in CV death/HHF outcomes by ethnicity.

Our meta-analysis has several limitations. Detailed analyses of specific subpopulations of Asians are not possible from these trials because of limitations of relevant data collection. There was heterogeneity between the studies, particularly in the SGLT2i outcome trials, with I² ranging from 0% to >60%. That noted, there was far less heterogeneity within Asian and White groups in the analyses for which we report differences, namely for SGLT2is and HF and in the GLP-1RA analysis (I² up to maximum of 28.6%). Although not definitive, these findings lend some confidence to the findings we report. The primary outcomes varied among the five diabetes CVOTs versus two HF trials. Although trials reported data on Asians, as mentioned, data were lacking on more specific subgroups (e.g., South, East, and Southeast Asians). Definitions of race or ethnicity varied (reported by patients in VERTIS CV, EMPEROR-Reduced, LEADER, EXSCEL, HARMONY OUTCOMES, and REWIND; reported by investigators in Dapa-HF; reported by investigator inquiry of participants in CANVAS Program; and not clearly specified in EMPA-REG OUTCOME, SUSTAIN-6, or PIONEER 6). Our main goal, as predefined in PROSPERO, was to examine Asians versus Whites, and therefore, other races and ethnic groups were not analyzed. However, we feel that examining differences in other races and ethnic groups should form the basis of a separate report. Finally, individual patient–level data were not used for analyses, and therefore, we could not adjust for potential important differences such as duration of diabetes, baseline treatment, or baseline characteristics such as BMI or smoking history, which vary between groups and could have affected MACE and CV death/HHF outcomes.

Going forward, we suggest an individual participant meta-analysis would be informative when all trial data are made available, including looking at the risk of adverse effects by ethnic group. It might also be of interest to examine adherence to therapy and the proportion of people who complete a trial or drop out by ethnic group. For example, it might be that Asians recruited to CVOTs are more compliant with therapy than Whites. Such possibilities could also be assessed through secondary analyses.

In conclusion, there is an apparent greater benefit of GLP-1RA therapy in Asians compared with Whites across all types of GLP-1RAs so far tested. In addition, in Asians, SGLT2is have at least as good an effect on MACE and potentially better effect on CV death/HHF outcomes compared with Whites. For future trials to be more informative, Asian race needs to be recorded with greater granularity, rather than grouping by geographical region. In particular, Asian race (a grouping that now covers nearly 60% the world’s population) should be broken down into more specific groups. We also suggest a need for more outcome-specific trials in Asian countries to further explore and validate the findings for GLP-1RAs. The findings of these analyses are relevant to the design, data capture, and reporting of future CVOTs.

Acknowledgments. The authors thank Liz Coyle, University of Glasgow, for assistance with the preparation of this manuscript.

Funding. This work was supported by British Heart Foundation Centre of Research Excellence grant RE/18/6/34217 to N.S.

The funders had no role in the design or conduct of the study, collection, management, analysis, or interpretation of the data, preparation, review, or approval of the manuscript, or decision to submit the manuscript for publication.

Duality of Interest. M.M.Y.L. reports a research grant paid to his institution, the University of Glasgow, from Boehringer Ingelheim. N.G. reports receiving honoraria from AstraZeneca and consultancy fees from Sanofi, both unrelated to submitted work. D.K.M. reports receiving honoraria for clinical trial leadership from AstraZeneca, Boehringer Ingelheim, Eisai, Esperion, GlaxoSmithKline, Janssen, Lexicon, Merck, Novo Nordisk, CSL Behring, and Sanofi; consultancy fees from Afimmune, Applied Therapeutics, AstraZeneca, Boehringer Ingelheim, Lilly, Merck, Pfizer, Novo Nordisk, Metavant, and Sanofi outside the submitted work; and personal fees and nonfinancial support from Merck and nonfinancial support from Pfizer outside the submitted work. M.K.R. reports receiving research funding from Novo Nordisk, consultancy fees from Novo Nordisk and Roche Diabetes Care, and modest owning of shares in GlaxoSmithKline outside the submitted work. N.S. reports grant and personal fees from Boehringer Ingelheim and personal fees from Afimmune, Amgen, AstraZeneca, Eli Lilly, Merck Sharp & Dohme, Novartis, Novo Nordisk, Pfizer, and Sanofi outside the submitted work. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. M.M.Y.L., N.G., and N.S. extracted the data, performed the statistical analysis, interpreted the data, and drafted the manuscript. M.M.Y.L., N.G., and N.S. conceptualized and designed the study. D.K.M. and M.K.R. critically edited the manuscript. N.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.