Racial, ethnic and regional differences in the effect of sodium–glucose co-transporter 2 inhibitors and glucagon-like peptide 1 receptor agonists on cardiovascular and renal outcomes: a systematic review and meta-analysis of cardiovascular outcome trials

This systematic meta-analysis was based on a pre-defined protocol and registered in the PROSPERO prospective register of systematic reviews (CRD42023401734). It was conducted in line with PRISMA guidelines (Supplementary Appendix 1). We conducted a database search of MEDLINE, Embase and the Cochrane Library from inception to 7 July 2023 with no restrictions on language. The computer-based searches combined free texts and MeSH terms related to T2D, GLP1RAs or SGLT2-Is, and cardiovascular outcomes. A randomised control trial (RCT) design search filter was employed. Details of the search algorithm are presented in Supplementary Appendix 2. Titles and abstracts of all retrieved citations were initially screened by one author (SKK) to assess the potential for inclusion. The screening was conducted using Rayyan, a free online bibliographic tool that helps expedite the screening via a process of semi-automation while incorporating high levels of usability.²⁴ This was then followed by full text acquisition for detailed evaluation, which was independently conducted by two authors (SS and SKK). Disagreements regarding eligibility of an article/study were discussed and resolved by mutual agreement with involvement of all the authors. To acquire other articles that might have been missed by the search strategy, the reference lists of relevant studies and review articles were manually scanned and citing references were also checked in Web of Science.

Eligible studies were randomised, placebo-controlled CVOTs of SGLT2-Is or GLP1-RAs in patients with T2D, which have reported data on cardiovascular and/or renal outcomes by race/ethnicity and/or region of participants. The following studies were excluded: (i) studies that specifically enrolled only patients with known renal insufficiency or established renal parenchymal disease without T2D; (ii) non-CVOTs; (iii) studies that enrolled patients with type 1 diabetes; and (iv) studies that did not report any of the outcomes stratified by race/ethnicity.

A predesigned data collection form, used in previous related reviews around the same topic,^25–27 was adapted for data extraction. One experienced reviewer (SKK) initially extracted data from the eligible studies and a second experienced reviewer (SS) independently checked the extracted data for accuracy using the original articles. Any disagreements or inconsistencies were resolved by involvement of the other authors. Data were extracted on the following when available with stratification by race, ethnicity or region: study publication date, geographical location, patient characteristics (baseline age, sex, duration of T2D), study design characteristics (randomisation, allocation concealment, blinding, duration), intervention type and comparator groups, outcomes of interest and hazard ratios (HRs) with their 95% confidence intervals (CIs). A number of studies presented forest plots without the HRs and their 95% CIs; these estimates were extracted using Plotdigitizer, an online data extraction tool that allows users to extract data from images in numerical format (https://plotdigitizer.com/app↗).

The pre-specified primary outcomes were defined as (i) major adverse cardiovascular events (MACE) or composite of death from cardiovascular causes, non-fatal myocardial infarction, or non-fatal stroke (composite cardiovascular outcome); (ii) composite outcome of CVD death/heart failure (HF) hospitalisation; (iii) composite outcome of end-stage kidney disease, doubling of creatinine level or death from renal causes (composite renal outcome); and (iv) all-cause mortality. Pre-specified secondary outcomes were (i) individual components of MACE or composite cardiovascular outcome; (ii) CVD death; (iii) HF hospitalisation; (iv) individual components of composite renal outcome; (v) other cardiovascular and renal outcomes; and (vi) adverse events (AEs).

The Cochrane Collaboration’s risk of bias tool was used to evaluate the risk of bias for each of the included trials.²⁸ This tool evaluates seven possible sources of bias, which are random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other bias. For each individual component, studies were classified into low, unclear and high risk of bias. We also assessed the certainty of the body of evidence for each outcome using the Grading of Recommendations Assessment, Development and Evaluation (GRADEpro) tool (https://gdt.gradepro.org↗), based on study limitations, inconsistency of effect, imprecision, indirectness and publication bias.²⁹ The certainty of the evidence was rated at four levels: high, moderate, low and very low.

Figure 1 shows the study selection process. The search of relevant databases and manual scanning of reference lists of relevant studies identified 466 potentially relevant citations. After the initial screening of titles and abstracts, 31 articles remained for full text evaluation. Following a detailed evaluation, 11 articles were excluded because (i) no subgroup results by race/ethnicity or region were reported (n = 6) and (ii) the populations recruited were not relevant to review question (n = 5). The remaining 20 articles met the inclusion criteria and were included in the review.^{20–22,35–51}

The 20 articles comprised 14 unique double-blinded RCTs (seven involved the comparison of SGLT2-I with placebo and seven involved the comparison of GLP1-RA) that recruited patients with T2D and reported cardiovascular and/or renal outcomes by race, ethnicity or region (Table 1). Based on the available data, we stratified participants into the following groups and subgroups: (i) Race: White, Asian, Black and Other (e.g. American Indian, Alaska Native, Native Hawaiian, Multiracial); (ii) Ethnicity: Hispanic Latino and non-Hispanic Latino; and (iii) Region: North America, South/Central America, Europe (Eastern and Western), Asia, Australia-New Zealand (Pacific), Asia/Pacific and Africa. The proportion of participants enrolled in the trials ranged from 66.6% to 93.2% for White populations, 1.2% to 21.6% for Asian populations, 2.4% to 8.3% for Black populations and 0.9% to 23.1% for Other populations. In addition to T2D, patients had other co-existing conditions such as chronic kidney disease, high cardiovascular risk, atherosclerotic CVD or HF. All patients had been diagnosed with T2D (average duration ranged from 9.3 to 15.8 years) and were being managed on standard treatment therapies including background antihyperglycaemic and antihypertensive medications before inclusion into the trials. All the studies were conducted in multiple countries (ranging from 28 to 49 countries). The types of SGLT2-Is used included dapagliflozin, canagliflozin, empagliflozin, ertugliflozin and sotagliflozin and those for GLP1-RAs comprised albiglutide, dulaglutide, exenatide, liraglutide, lixisenatide, semaglutide and efpeglenatide. Non-compliance of trial medications was very low and ranged from 0.0% to 1.0% across trials that provided these data. The average age of participants ranged from 60 to 69 years. The mean/median duration of follow-up ranged from 9 months to 3.8 years. Using the Cochrane Collaboration tool, all trials demonstrated low risk of bias in all domains (Supplementary Appendix 3).

None of the trials reported AEs by race/ethnicity and/or region. However, a pre-specified post hoc publication involving the EMPA-REG OUTCOME trial conducted in patients of Asian race reported on AEs.⁴⁹ The results show that the percentages of Asian patients who had any AEs, serious AEs or AEs leading to discontinuation of study intervention were similar in the empagliflozin and placebo groups.

The GRADE working group recommends up to seven patient-important outcomes to be listed in the ‘summary of findings’ tables in systematic reviews.²⁹ We selected outcomes for assessment based on their importance (primary outcomes) and how frequently they were reported by studies. GRADE certainty of the evidence ranged from moderate to high (Supplementary Appendix 15).

In a pooled analysis of 14 unique randomised, placebo-controlled CVOTs of SGLT2-Is and GLP1-RAs in patients with T2D, we examined racial/ethnic and regional patterns in the effects of these agents on various cardiovascular and renal outcomes. Our findings indicate that SGLT2-Is and GLP1-RAs reduced the risk of MACE in White, Asian and Hispanic/Latino populations. SGLT2-Is reduced the risk of all other cardiovascular and renal outcomes in White populations. Similarly, for Asian populations, SGLT2-I reduced the risk of all other cardiovascular and renal outcomes except for the composite outcome of CVD death/HF hospitalisation. No significant beneficial effects were observed in Black populations for both interventions and for all the evaluated outcomes, except for HF hospitalisations in relation to SGLT2-Is. However, some of the effect estimates were similar to those of other racial groups, but they did not reach statistical significance and were imprecise. We did not observe any regional differences in the effects of SGLT2-Is on MACE. However, GLP1-RAs reduced the risk of MACE specifically in the European region, while there was no strong evidence of effects in other regions. Furthermore, SGLT2-Is demonstrated a risk reduction in the composite outcome of CVD death/HF hospitalisation in North America and Europe, but no strong evidence of effects was found in other regions (Europe, South/Central America, Asia/Pacific and Africa). It is important to note that the findings regarding CVD death, incident or worsening nephropathy, and the composite renal and CVD outcome were based on single studies. The certainty of the overall evidence ranged from moderate to high.

The current study cannot be directly compared with any previous studies on the topic, but a number of previous related reviews deserve mention and discussion. Cai et al. explored the racial/ethnic patterns in cardiovascular outcomes in trials of antidiabetic medications in people with T2D.⁵² A major strength of this review was the fact that they evaluated several antidiabetic medications in addition to SGLT2-Is and GLP1RAs, which included DPP-4 inhibitors, basal insulin, peroxisome proliferator-activated receptor antagonists and alpha glucosidase inhibitors. However, there were several relevant aspects that were not addressed. First, the review only focused on the outcome of MACE. Second, regional and some ethnic differences were not evaluated. Third, trials such as CANVAS (Canagliflozin Cardiovascular Assessment Study) and ELIXA (Evaluation of Lixisenatide in Acute Coronary Syndrome), which reported racial/ethnic and region-specific estimates, were not included in the review. Subsequent publications of the VERTIS CV (Evaluation of Ertugliflozin Efficacy and Safety Cardiovascular Outcomes) and EMPA-REG OUTCOME trials, which reported additional cardiovascular and renal outcomes by racial/ethnic and region, were also not included. Fourth, the review included trials such as the EMPEROR-Reduced (Empagliflozin Outcome Trial in Patients with Chronic Heart Failure and a Reduced Ejection Fraction) and DAPA-CKD (Dapagliflozin and Prevention of Adverse Outcomes in Chronic Kidney Disease), which were based in patients with and without T2D and did not specifically report racial/ethnic and region-specific estimates for T2D. Qiu et al. evaluated the effects of SGLT2-Is and GLP1-RAs on only the outcome of MACE by race, ethnicity and region based on a pooled analysis of 11 CVOTs; consistent with our results, they showed no evidence of an effect of SGLT2-Is and GLP1-RAs on the risk of MACE in Black populations.⁵³ Another recent meta-analysis of CVOTs by Lee et al. only compared the cardiovascular effects of SGLT2-Is and GLP1-RAs in Asian versus White populations with and without T2D.⁵⁴ The results suggested differential treatment effects of SGLT2-Is and GLP-1RAs by race (White and Asian populations), with greater benefits of both classes of drugs in Asians.

Compared to the study by Cai et al.,⁵² we only focused on SGLT2-Is and GLP1-RAs because among the newer therapeutic agents for the treatment of T2D, only members of these two classes have been shown to significantly reduce the risk of major cardiovascular events such as MACE^19–22; they have also been the focus of guideline recommendations.²³ Three of the DPP-4Is (saxagliptin, alogliptin and sitagliptin) have been demonstrated to have no effect on MACE^55–57; furthermore, the SAVOR-TIMI 53 (Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus-Thrombolysis in Myocardial Infarction 53) trial showed an increased risk of HF hospitalisations,⁵⁵ findings that were unexpected and very concerning. We evaluated a comprehensive panel of cardiovascular and renal outcomes in addition to safety outcomes as reported by the trials. In addition to race data, we also evaluated ethnic (Hispanic/Latino vs. non-Hispanic/Latino) and regional patterns in outcomes, which were reported by the majority of trials included. Finally, we only included trials that recruited patients with only T2D; hence, our results can be generalised to this specific population group.

Over the last decade, several landmark CVOTs have consistently demonstrated the cardiovascular and renal protective effects of SGLT2-Is and GLP1-RAs.^19–22 SGLT2-Is work by targeting renal tubular glucose reabsorption, thereby exerting glucose-lowering effects through increased urinary glucose excretion.⁵⁸ GLP1-RAs exert their glucose-lowering effects via glucose-dependent increase in insulin secretion, inhibition of glucagon secretion, reduction in postprandial glucose excretion and increase in satiety.⁵⁹ It has been reported that both classes of drugs may exert their cardiorenal protective effects independently of glucose control via their individual pleiotropic properties.⁵⁹

The main findings showed consistent reductions in risk of adverse cardiovascular and renal outcomes for White and Asian populations, with no evidence of beneficial effects for Black and Other populations. Though some of the effect estimates were similar for these populations, they were imprecise and did not reach statistical significance for Black populations. These findings may suggest inadequate power to demonstrate any effects, which is reflected in the low sample size and event rates for these populations. Indeed, these findings were based on results from subgroup analyses and the trials were rather designed and powered for the overall populations recruited. Furthermore, given the consistent nature of the significant lack of beneficial effects across the majority of outcomes for Black populations, other factors, apart from power issues, may also be at play here. It is well known that Black populations have a higher prevalence of major cardiovascular risk factors such as hypertension, dyslipidaemia, smoking, physical inactivity and other co-morbidities compared with White populations.⁶⁰ This suggests that Black populations already have a higher baseline risk for developing adverse cardiovascular and renal events. The majority of these factors require lifestyle modifications and other interventions and cannot be controlled by the trial interventions. None of the included trials reported a breakdown of baseline characteristics according to race/ethnicity, so it is difficult to know the baseline risk profile of the various groups. Differences in medication compliance and diabetes control across the various ethnic and racial groups in the trials might also account for these findings; however, none of the trials reported on these data according to race/ethnicity. Finally, there may be racial/ethnic differences in the pharmacokinetics, pharmacodynamics and safety of SGLT2-Is and GLP1-RAs; hence, their effects may differ. For example, it is well known that there are differences in the degree and rate of response of Black and White populations to various antihypertensive medications. For instance, diuretics and calcium channel blockers (CCBs) are more effective in older Black patients than beta blockers or ACE inhibitors, whereas White patients, especially younger individuals, appear to respond somewhat better to beta blockers and ACE inhibitors than to diuretics or CCBs.⁶¹ In a review to determine the efficacy of newer treatments in White and Asian populations with diabetes, the findings suggested that the glucose-lowering efficacy of SGLT2-Is, and to a lesser extent DPP-4Is, was greater in Asian populations compared with White populations; no difference was observed for GLP1-RAs.⁶² However, in a 24-week placebo-controlled, double-blind, parallel-group RCT to study the effects of empagliflozin in Black patients with T2D and hypertension, empagliflozin reduced glycated haemoglobin, body weight and blood pressure and it was well tolerated during the study period.⁶³ It is not known if these beneficial changes will translate to reductions in adverse cardiovascular and renal outcomes as patients were not followed up for these outcomes. The mechanistic pathways underlying the cardiorenal protective effects of SGLT2-Is and GLP1RAs and their potential racial/ethnic differences warrant investigation.

Given the well-documented evidence that Black and other ethnic minority populations are more likely to develop T2D and at a younger age, have lower rates of risk factor control and have higher lifetime risk of developing microvascular and macrovascular complications,^12–14,64 the consistent lack of benefits observed among Black populations is concerning. The global burden of T2D has increased substantially in recent decades and will continue to soar in the next few decades; this increase will be accompanied by a proportionate increase in morbidity and deaths attributable to the disease. This trend is likely to disproportionately impact on populations at high risk of developing the disease and with very little access to care and effective treatments. Newer and more effective therapeutic agents have been developed for the management of T2D, and guidelines suggest that they may be appropriate for all populations; however, the evidence seems to suggest that the beneficial effects are not consistent across all populations. As discussed, though there are many factors that could have contributed to the lack of evidence of beneficial effects for Black and other non-White populations, low statistical power due to small sample sizes of these populations may partly be responsible for the current findings. It is quite clear from the current data that some racial/ethnic groups such as Black populations were underrepresented in all the included trials. Data suggest that ethnic minorities are under-represented in CVOTs of T2D and that these trials may be characterised by recruitment bias.^65,66 Furthermore, reporting of baseline and trial characteristics and outcome data according to race/ethnicity is far from optimal in these trials. To echo Khunti et al.,⁶⁶ efforts should be made to improve reporting of race/ethnicity and improve diversity in trial recruitment. It is a fact that populations that are often under-represented in these trials are the ones who are more likely to receive the interventions being evaluated. Given that regulatory bodies now require the conduct of separate efficacy and safety studies for paediatric populations, we recommend that regulatory bodies propose a diversity recruitment threshold when such trials are being conducted to improve diversity. The long-term cardiorenal efficacy and safety of these antihyperglycaemic agents deserve specific investigation in these underrepresented populations to ascertain if the current findings are related to statistical power or to the specific mechanisms underlying the cardiorenal protective effects of these agents. In the absence of robust efficacy and safety data among different ethnic/racial groups, the use of these newer agents needs to be individualised based on patients’ unique circumstances.

KK has acted as a consultant, speaker or received grants for investigator-initiated studies from Astra Zeneca, Novartis, Novo Nordisk, Sanofi-Aventis, Lilly and Merck Sharp & Dohme, Boehringer Ingelheim, Bayer, Berlin-Chemie AG/Menarini Group, Janssen and Napp. SS reports receiving personal fees from Amgen, AstraZeneca, NAPP, Lilly, Merck Sharp & Dohme, Novartis, Novo Nordisk, Abbott, Roche, Sanofi-Aventis and Boehringer Ingelheim. Additionally, SS reports receiving grants from AstraZeneca, Sanofi-Aventis, Servier and Janssen. SKK has no conflicts of interest.

This study was supported with a research grant from Primary Care Diabetes Europe (PCDE). PCDE, as a society, has received sponsorship from Novo Nordisk, Eli Lilly and Roche Diagnostics, but the companies had no input in the study.

This is secondary research and all data used are publicly available. No ethics approval required.

SKK, KK, SS

Setor K Kunutsor: Study design, literature search, data collection, analysis, data interpretation and writing.

Kamlesh Khunti: Study design, data interpretation and writing.

Samuel Seidu: Study design, literature search, data collection, data interpretation and writing.

The authors are supported by the National Institute for Health and Care Research (NIHR) Applied Research Collaboration East Midlands (ARC EM), Centre for Ethnic Health Research and Leicester NIHR Biomedical Research Centre (BRC). The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.

The corresponding author 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. This study is based on data from published articles. The supplementary material are all included in this submission.

Not commisioned; peer reviewed by David Goldsmith and Julie Morris.