Academic Editor: Grigorios Korosoglou
Chronic kidney disease (CKD) and cardiovascular disease share common risk
factors such as hypertension, diabetes mellitus and dyslipidemia. Patients with
CKD carry a high burden of cardiovascular disease and may be excluded from
clinical trials on the basis of safety. There are an increasing number of
clinical trials which predefine sub-group analysis for CKD. This systematic
review with fixed-effect meta-analysis investigates glucose lowering therapy and
cardiovascular outcomes in relation to CKD. We included randomized controlled
trials (RCT) of glucose lowering treatments performed in adults (aged
Cardiac dysfunction in the presence of chronic kidney disease (CKD) is a well-established phenomenon. End-stage renal disease (ESRD) is associated with adverse cardiac events in up to 64% of patients undergoing hemodialysis. Developing evidence-based medicines for cardiovascular disease (CVD) through randomized, controlled trials (RCTs) in this population is a current priority [1, 2]. Cardiovascular death accounts for 50% of deaths in CKD patients representing an unmet clinical therapeutic need for safe and effective therapies in these patients [2]. Many of the risk factors of cardiac dysfunction and CKD are shared, including hypertension, diabetes mellitus, dyslipidemia, and coronary artery disease [3]. Additionally; anemia, hypovolemia, proteinuria and mineral metabolism abnormalities associated with CKD are associated with increased likelihood of adverse cardiovascular outcomes [3].
Diabetic nephropathy is the leading cause of CKD. Patients with CKD and diabetes carry a high burden of CVD [4]. The mainstay of reducing diabetes-associated cardiovascular (CV) death is through modification of underlying risk factors, including glycemic control, dyslipidemia, smoking, physical inactivity, and hypertension, which are shared risks in myocardial infarction (MI) [5] and stroke [6, 7]
Type 2 diabetes mellitus (T2DM) is also associated with cardiovascular disease, with a 2–3 times increased risk of cardiovascular disease [8]. Around 40% of T2DM patients develop diabetic kidney disease, which portends an even greater risk of adverse cardiovascular outcomes [9]. Randomized controlled trials in patients with diabetes mellitus have demonstrated the beneficial cardiovascular effects of antihyperglycemic agents such as sodium-glucose co-transporter inhibitors (SGLT2i), glucagon-like peptide-1 receptor (GLP-1R) agonists [10, 11]. Previous systematic reviews and meta-analysis have focused on the identification of RCTs involving patients with T2DM [12, 13], however cardiovascular outcomes in the context of CKD and ESRD patients are less well-established as CKD patients are frequently excluded from clinical trials, and only recently that focused trials have been designed to address this patient group [14]. Shared risk factors and the prevalence of concomitant CKD and CVD in clinical practice demonstrate the need for rigorous safety and efficacy data in this population.
This systematic review with meta-analysis investigates the data available on cardiovascular outcomes specifically in CKD/ESRD patients involved in clinical trials of glucose-lowering therapies.
This review focuses on cardiovascular outcomes in trials in chronic kidney
disease patients and was registered within the PROSPERO network (registration
number: CRD42021268563) [15]. An electronic database search was performed of
Embase & Ovid Medline up to April 2021 through the Ovid Gateway system in
addition to screening of review articles generated during this search. The search
was restricted to randomized controlled trials (RCT) performed in adults (aged
The MeSH terms of the search strategy were created following an initial scoping of existing reviews and primary study literature. These included ‘cardiovascular disease’, ‘chronic heart failure’, ‘chronic kidney disease’, ‘chronic renal disease’ and variants. These terms were then combined with Boolean operators in addition to limitations.
The search was designed to be sensitive to the tautological array of alternative
terminologies included in studies of CKD and CVD and was performed on 31st March
2021. Identified studies were de-duplicated prior to screening of title and
abstract to exclude irrelevant or ineligible articles. The manuscripts of
remaining articles were then reviewed. Studies were included if they met the
following conditions: primary RCT or subsequent subgroup analysis of larger RCT
in a population of patients with chronic kidney disease stages 1–5 with native
kidneys; Sample size greater than 300; Follow-up
All references were uploaded to Rayyan [16], a web-based computational reference manager. These were then independently screened in two stages. Initially through title and abstract, and then by full text. AK and RS reviewed eligibility. Discordance was adjudicated by AM. These were recorded in a PRISMA workflow and reported according to the PRISMA guidelines for systematic review and metanalysis [17] (Supplementary Table 1).
Data was extracted by AK then reviewed by RS and AM for discrepancies. Demographic data and the definitions of CVD and CKD used within the RCT were extracted from the studies or attached supplementary data. Baseline characteristics including presence of congestive heart failure, hypertension, diabetes, and cerebrovascular disease were also gathered. The definition of the primary outcome for each study was collated.
Hazard ratios were collected, and where not available calculated from reported
or supplementary data for primary composite outcomes, all-cause mortality, renal
and cardiac outcomes. Forest plots were created for primary outcome, all-cause
mortality, Fatal/Non-fatal MI and hospitalizations for heart failure (HHF) where
available with the aid of RevMan software [18]. Two-tailed statistical analysis
was performed with a significance threshold of p
Heterogeneity of studies was assessed prior to subgroup analysis with an I
Our initial search yielded 1561 abstracts. Sixty-seven studies were identified as RCTs with cardiovascular outcomes which involved CKD patients. Of these we identified 7 RCTs for inclusion which involved diabetes medications. The PRISMA workflow is shown in Fig. 1. These were CARMELINA [20], CREDENCE [21], DAPA-CKD [22], EMPEROR-REDUCED [23], HARMONY [24], LEADER [25] and SCORED [26]. A risk of bias assessment can be viewed in Fig. 2. Most studies had a low risk of bias, however SCORED had high bias in terms of its outcome analysis. This was due to the change in definition of primary outcome from time-to-event to number of events. The study also used investigator-reported outcomes for the analysis.
PRISMA [17] workflow of included studies. *Consider, if feasible to do so, reporting the number of records identified from each database or register searched (rather than the total number across all databases/registers). **If automation tools were used, indicate how many records were excluded by a human and how many were excluded by automation tools.
Risk of bias summary of included trials. Reviewers’ judgements for risk of bias within each included study.
Four studies involved SGLT2i, two involved GLP-1R agonists, and one assessed a DPP-4 inhibitor versus placebo in patients with chronic renal failure. A total of 47,879 participants were enrolled across all 7 trials. Trial participants were predominantly male and aged greater than 60 years of age. Only one study, SCORED, included patients with CKD stage 5, whereas other studies only included stages 1–4. A summary of the included studies can be viewed in Table 1 (Ref. [20, 21, 22, 23, 24, 25, 26]) alongside the intervention details.
Parent trial | CARMELINA [20] | CREDENCE [21] | DAPA-CKD [22] | EMPEROR-REDUCED [23] | HARMONY [24] | LEADER [25] | SCORED [26] |
Subgroup | Secondary analysis of T2DM with coexisting kidney disease | Patient with type 2 diabetes and CKD with prior history of HF | - | Pre-specified analysis of EMPORER-REDUCED | - | - | - |
Author | Rosenstock et al. | Mahaffey et al. | Heerspink et al. | Packer et al. | Hernandez et al. | Marso et al. | Bhatt et al. |
Year of study | 2019 | 2019 | 2020 | 2020 | 2018 | 2016 | 2021 |
Drug action | DPP-4 inhibitor | SGLT2 inhibitor | SGLT2 inhibitor | SGLT2 inhibitor | GLP-1R agonist | GLP-1R agonist | SGLT2 inhibitor |
Intervention | Linagliptin | Canagliflozin | Dapagliflozin | Empagliflozin | Abiglutide | Liraglutide | Sotagliflozin |
Number of participants | 6979 | 4401 | 4304 | 3730 | 9463 | 9340 | 10584 |
Mean age (years) | 65.9 | 63 | 61.9 | 67.2 | 64.1 | 64.3 | 69** |
Male, (%) | 62.9 | 66.1 | 66.9 | 76.5 | 70 | 64.3 | 55.7** |
CKD stage* | CKD1-4 | CKD2-4 | CKD2-4 | CKD1-3b | CKD1-3b | CKD 1-4 | CKD 3-5 |
Mean. eGFR (mL/min/1.73 m |
54.6 | 56.2 | 43.1 | 61.8 | 79.1 | - | 44.4 |
Median UACR (mg/g) | 162 | 927 | 949.5 | - | - | - | 74 |
Diabetes, (%) | 100 | 100 | 67.5 | 49.8 | 100 | 100 | 100 |
Stroke/Cerebrovascular disease, n (%) | - | - | - | - | 17 | 16.1 | 8.9 |
Atrial fibrillation, (%) | 9.6 | - | - | 35.6 | 8 | - | - |
Hypertension, (%) | 91 | 96.8 | 72.4 | 86 | 90 | - | |
Ischemic heart disease, (%) | 58.5 | - | - | - | 47 | - | - |
NYHA class at baseline | - | 0–III | 0–III | II–IV | - | II–III | 0–IV |
Prevalence of HF at baseline, (%) | 26.8 | 14.8 | 10.9 | 100 | 20 | - | 31 |
Other treatment | Standard care | Standard care | ACEI or ARB 4 weeks prior | Standard care | Standard care | Standard care | Standard care |
RAAS-inhibitor, (%) | 81.1 | 69.0 | - | 69.7 | - | - | 88.5 |
ACE-inhibitor (%) | - | - | 31.5 | - | 49.0 | 51.0 | 38.3 |
Angiotensin receptor blocker (%) | - | - | 66.7 | - | 33.0 | 31.9 | 49.0 |
Angiotensin receptor-neprilysin inhibitor (%) | - | - | - | 19.5 | - | - | 1.2 |
Mineralcorticoid-receptor antagonist (%) | - | - | - | 71.3 | - | - | 15.0 |
Beta-blocker, (%) | 59.5 | 40.2 | - | 94.7 | - | 55.5 | 62.5 |
Diuretic, (%) | 54.9 | 46.7 | 43.7 | - | - | 41.8 | 65.2 |
Statin, (%) | 71.9 | 69.0 | 64.9 | - | - | 72.2 | - |
Antithrombotic*** (%) | 62.2 | 59.6 | - | - | 77.0 | 67.8 | - |
Mean Follow-up (months) | 26.4 | 31.44 | 28.8 | 16 | 19.2 | 45.6 | 16 |
*CKD stages: Stage 1 with normal or high GFR (GFR ** Median value reported. *** Anticoagulation or antiplatelet agent, including aspirin. SGLT2i, sodium-glucose cotransporter-2 inhibitor; GLP-1R, glucagon-like peptide-1 receptor; DPP4, Dipeptidyl-peptidase 4; CKD, Chronic Kidney disease; eGFR, estimated Glomerular Filtration Rate; UACR, Urinary albumin to creatinine ratio; NYHA, New York Heart Association functional classification; RAAS, Renin-Aldosterone-Angiotensin System; ACE, angiotensin converting enzyme; ACEI, Angiotensin coverting enzyme inhibitor; ARB, angiotensin receptor blocker. |
Included trial demographics were assessed for heterogeneity in addition to the
degree of renal impairment. Heterogeneity assessment from the SCORED required
estimation of mean and standard deviations from median and interquartile ranges
provided. Mean eGFR was unable to be calculated for LEADER as the supplementary
appendix grouped eGFR by class. These were normal, mild, moderate, and severe
with 2.5% of liraglutide and 2.3% of placebo patients categorized as severe
impairment (eGFR
Between all trials, age yielded a Chi
SCORED was unfortunately prematurely terminated due to financial reasons related to the coronavirus pandemic [26]. Due to this, the primary endpoint definition was altered to include total number of events rather than a time-to-event analysis.
Table 2 lists outcomes for each individual trial. Confidence intervals were calculated based on this information in Table 3 and corresponding p-values provided.
Parent trial | CARMELINA | CREDENCE | DAPA-CKD | EMPORER-REDUCED | HARMONY | LEADER | SCORED | |||||||
Arm | Linagliptin | Placebo | Canagliflozin | Placebo | Dapagliflozin | Placebo | Empagliflozin | Placebo | Albiglutide | Placebo | Liraglutide | Placebo | Sotagliflozin | Placebo |
Participants, N | 3494 | 3485 | 2202 | 2199 | 2152 | 2152 | 1863 | 1867 | 4731 | 4732 | 4668 | 4672 | 5292 | 5292 |
Primary composite outcome | Time to first occurrence of CV death, non-fatal MI or non-fatal stroke | Composite. of ESRD (dialysis, transplantation, or a sustained estimated. GFR of |
Composite of sustained decline in the eGFR of at least 50%, ESRD, or death from renal or CV causes | Composite of adjudicated CV death or hospitalization for heart failure, analyzed as the time to the first event | First occurrence of CV death, myocardial infarction, or stroke | First occurrence of death from CV causes, non-fatal myocardial infarction, or non-fatal stroke | Total number of deaths from CV causes, hospitalizations for HF and urgent visits for HF | |||||||
Primary composite outcome, n (%) | 434 (12.4)* | 420 (12.1)* | 245 (11.1)* | 339 (15.4)* | 198 (9.2)* | 312 (14.5)* | 361 (19.4)* | 461 (24.7)* | 338 (7)* | 428 (9)* | 607 (13)* | 696 (14.9)* | 400 (7.6)* | 530 (10.0)* |
All-cause mortality, n (%) | 367 (10.5) | 10.7 | 167 (7.6) | 201 (9.1) | 101 (4.7)* | 146 (6.8)* | 250 (13.4) | 265 (14.2) | 196 (4.0) | 205 (4.0) | 383 (8.2)* | 449 (9.6)* | 246 (4.7) | 246 (4.7) |
Cardiovascular mortality, n (%) | 221 (7.3) | 225 (7.6) | 110 (5.0) | 140 (6.4) | 65 (3.0) | 80 (3.7) | 186 (10) | 202 (10.8) | 122 (2.6) | 130 (2.7) | 219 (4.7)* | 280 (6.0)* | 155 (2.9) | 170 (3.2) |
Fatal/non-fatal Myocardial infarction, n (%) | 165 (4.7) | 146 (4.2) | 83 (3.8) | 95 (4.3) | 32 (1.5) | 45 (2.1) | - | - | 181 (4.0)* | 240 (5.0)* | 294 (6.3)* | 341 (7.3)* | - | - |
Stroke (fatal/non-fatal), n (%) | 81 (2.3) | 88 (2.5) | 62 (2.4) | 80 (3.0) | 26 (1.2) | 30 (1.4) | - | - | 94 (2.0) | 108 (2.3) | 173 (3.7) | 201 (4.3) | - | - |
Progression to ESRD disease, n (%) | 63 (1.8) | 64 (1.8) | 117 (5.3) | 165 (7.5) | 110 (5.1) | 161 (7.5) | - | - | - | - | - | - | - | - |
Doubling of baseline creatinine/change in eGFR, n (%) | - | - | 119 (5.4) | 187 (8.5) | 112 (5.2) | 200 (9.3) | - | - | - | - | - | - | 37 (0.7) | 52 (1.0) |
Hospitalisation for heart failure, n (%) | 209 (6.0) | 226 (6.5) | 89 (4.0) | 141 (6.4) | - | - | 388 (20.8)* | 553 (29.6)* | - | - | 218 (4.7) | 248 (5.3) | 245 (3.5)* | 360 (5.1)* |
Change in eGFR = decrease ESRD, End-stage renal disease; Egfr, estimated glomerular filtration rate. |
Outcome | Drug class | Trial | Risk ratio [95% CI] | p-value |
Primary composite outcome | DPP-4 inhibitor | CARMELINA | 1.03 [0.91, 1.17] | |
Test for overall effect: Z = 0.47 (p = 0.64) | ||||
SGLT2 inhibitor | CREDENCE | 0.72 [0.62, 0.84] | 0.00001 | |
DAPA-CKD | 0.63 [0.54, 0.75] | |||
EMPORER-REDUCED | 0.78 [0.69, 0.89] | |||
SCORED | 0.75 [0.67, 0.85] | |||
Total | 0.73 [0.68, 0.79] | N/A | ||
Test for overall effect: Z = 8.80 (p | ||||
GLP-1R agonist | LEADER | 0.87 [0.79, 0.97] | 0.01 | |
HARMONY | 0.79 [0.69, 0.91] | |||
Total | 0.84 [0.78,0.91 | N/A | ||
Test for overall effect: Z = 4.17 (p | ||||
All-cause mortality | DPP-4 inhibitor | CARMELINA | 0.98 [0.82, 1.17] | 0.74 |
Test for overall effect: Z = 0.27 (p = 0.79) | ||||
SGLT2 inhibitor | CREDENCE | 0.83 [0.68, 1.01] | 0.61 | |
DAPA-CKD | 0.81 [0.59, 1.12] | 0.004 | ||
EMPORER-REDUCED | 0.92 [0.76, 1.11] | N/A | ||
SCORED | 0.91 [0.74, 1.13] | N/A | ||
Total | 0.89 [0.81, 0.98] | N/A | ||
Test for overall effect: Z = 2.43 (p = 0.02) | ||||
GLP-1R agonist | LEADER | 0.78 [0.66, 0.93] | 0.02 | |
HARMONY | 0.94 [0.74, 1.20] | 0.644 | ||
Total | 0.89 [0.80, 0.99] | N/A | ||
Test for overall effect: Z = 2.21 (p = 0.03) | ||||
Cardiovascular mortality | DPP-4 inhibitor | CARMELINA | 0.98 [0.82, 1.17] | 0.63 |
Test for overall effect: Z = 0.22 (p = 0.82) | ||||
SGLT2 inhibitor | CREDENCE | 0.78 [0.62, 1.00] | 0.86 | |
DAPA-CKD | 0.81 [0.59, 1.12] | N/A | ||
EMPORER-REDUCED | 0.92 [0.76, 1.11] | N/A | ||
SCORED | 0.91 [0.74, 1.13] | 0.35 | ||
Total | 0.90 [0.79, 1.02] | N/A | ||
Test for overall effect: Z = 2.35 (p = 0.02) | ||||
GLP-1R agonist | LEADER | 0.78 [0.66, 0.93] | 0.007 | |
HARMONY | 0.94 [0.74, 1.20] | 0.578 | ||
Total | 0.83 [0.72, 0.96] | N/A | ||
Test for overall effect: Z = 2.56 (p = 0.01) | ||||
Fatal/Non-fatal MI | DPP-4 inhibitor | CARMELINA | 1.13 [0.91, 1.40] | 0.3 |
Test for overall effect: Z = 1.08 (p = 0.28) | ||||
SGLT2 inhibitor | CREDENCE | 0.87 [0.65, 1.16] | 0.37 | |
DAPA-CKD | 0.71 [0.45, 1.11] | N/A | ||
EMPORER-REDUCED | - | - | ||
SCORED | - | - | ||
Total | 0.82 [0.64, 1.05] | N/A | ||
Test for overall effect: Z = 1.60 (p = 0.11) | ||||
GLP-1R agonist | LEADER | 0.86 [0.74, 1.00] | 0.046 | |
HARMONY | 0.75 [0.62, 0.91] | 0.003 | ||
Total | 0.82 [0.73, 0.92] | N/A | ||
Test for overall effect: Z = 3.35 (p = 0.0008) | ||||
Hospitalizations for heart failure | DPP-4 inhibitor | CARMELINA | 0.92 [0.77, 1.11] | 0.26 |
Test for overall effect: Z = 0.87 (p = 0.38) | ||||
SGLT2 inhibitor | CREDENCE | 0.63 [0.49, 0.82] | 0.98 | |
DAPA-CKD | - | - | ||
EMPORER-REDUCED | 0.70 [0.63, 0.79] | |||
SCORED | 0.68 [0.58, 0.80] | |||
Total | 0.69 [0.63, 0.75] | N/A | ||
Test for overall effect: Z = 8.50 (p | ||||
GLP-1R agonist | LEADER | 0.88 [0.74, 1.05] | 0.14 | |
HARMONY | - | - | ||
Test for overall effect: Z = 1.42 (p = 0.16) | ||||
SGLT2i, sodium-glucose cotransporter-2 inhibitor; GLP-1R, glucagon-like peptide-1 receptor; DPP4, Dipeptidyl-peptidase 4; CI, Confidence interval; N/A, not applicable; -, no data available. |
All primary composite outcomes included CV death. Across all studies, interventions were associated with statistically significant reductions in the primary outcome compared with placebo (Table 3). CREDENCE and DAPA-CKD had primary composite outcomes which included renal outcomes as well as cardiovascular outcomes. All remaining studies shared a similar primary outcome of 3-point major adverse cardiac events (MACE) involving CV death, MI and stroke.
All-cause mortality was reduced in DAPA-CKD (6.8% vs 4.7% p = 0.004) and LEADER (9.6% vs 8.2% p = 0.02) (Fig. 3A).
Forest plots of primary composite outcomes. (A) All-cause mortality. (B) Fatal/Non-fatal MI. (C) Cardiovascular Mortality. (D) Hospitalization for heart failure.
There were no between-group differences in mortality in CARMELINA, CREDENCE, or HARMONY. p-values were not provided by EMPORER-REDUCED and SCORED.
The effect size of SGLT2i (Z = 2.43 [p = 0.03]) and GLP-1R agonists (Z = 2.21 [p = 0.03]) was similar in reducing mortality.
Data were unavailable for fatal/non-fatal MI in EMPORER-REDUCED and SCORED.
HARMONY (4.0% vs 5.0% p = 0.003) and LEADER (6.3% vs 7.3% p
= 0.046) had a statically significant reduction in fatal/non-fatal MI (Table 2).
Overall, GLP-1R agonists had the greatest effect in the prevention of
fatal/non-fatal MI (Z = 3.35 [p = 0.0008]) when compared to DPP4i (Z =
1.08 [p = 0.28]) and SGLT2i (1.60 [p = 0.11]) (Fig. 3B). There
were no between group differences observed between CREDENCE and DAPA-CKD in
prevention of MI (Chi
CV mortality reduction was significantly reduced in LEADER (6.3% vs 7.3% p = 0.007).
SGLT2i (Z = 2.35 [p = 0.02]) and GLP-1R agonists (Z = 2.56 [p
= 0.01]) had a similar effect size in reducing CV mortality (Fig. 3C).
Additionally, there did not appear to be between group differences in this
reduction in SGLT2i (Chi
Data were not available for HHF in DAPA-CKD and HARMONY.
HHF were decreased across remaining trials and significantly reduced in
EMPORER-REDUCED (20.8% vs 29.6% p
An initial sensitivity analysis was performed for studies which included 3-point
MACE as the primary outcome (Table 4). This only affected the SGLT2i group, but
did not lead to any significant change (Z = 5.91 [p
Outcome | Number of participants | Risk ratio [95% CI] | Effect size (Z) | p-value | I | |
SGLT2 inhibitor | Placebo | |||||
All-cause mortality | 7494 | 7491 | 0.92 [0.81, 1.05] | 1.21 | 0.23 | 0.49 |
Cardiovascular mortality | 7494 | 7491 | 0.85 [0.73, 1.00] | 1.92 | 0.05 | 0.00 |
Fatal/Non-Fatal MI * | 2202 | 2199 | 0.87 [0.65, 1.16] | 0.93 | 0.35 | N/A |
Hospitalisations for heart failure | 7494 | 7491 | 0.67 [0.58, 0.76] | 5.91 | p |
0.00 |
SGLT2i, sodium-glucose cotranporter 2 inhibitor; *, data only available for CREDENCE; N/A, not applicable. |
The populations of DAPA-CKD and EMPORER-REDUCED varied substantially from other studies as there were individuals without diabetes included in the trial designs. Sensitivity analysis of all-cause mortality, fatal/non-fatal MI, CV mortality and HHF were performed for SGLT2i excluding these trials (Table 4). Reductions in CV mortality and HHF remained significant for SGLT2i.
A further sensitivity analysis was performed excluding SCORED from the SGLT2i
group due to the high risk of reporting bias on outcomes. All-cause mortality (Z
= 2.94 [p = 0.003]), CV mortality (Z = 2.26 [p = 0.02]) and HHF
(Z = 7.05 [p
Begg’s funnel plots were created for the CV outcomes of CV mortality (Fig. 4A), Fatal/non-fatal MI (Fig. 4B) and HHF (Fig. 4C). Fatal/non-fatal MI showed slight asymmetry from the CARMELINA trial, however due to five trials being included the power to detect asymmetry was reduced.
Risk of publication bias assessment. Funnel plots to assess risk of bias in (A) Cardiovascular mortality, (B) Fatal/Non-Fatal myocardial infarction and (C) Hospitalisations for heart failure.
Our systematic review yielded three main classes of diabetic drugs which have had their cardiovascular outcomes studied in large-scale clinical trials. These were a DPP-4 inhibitor, SGLT2i, and GLP-1R agonists. In patients with CKD, SGLT2i and GLP1R agonists were associated with a reduction in cardiovascular death and all-cause death, and SGLT2i, GLP1R agonists and DPP4 inhibitors were associated with a reduction in hospitalization for heart failure.
DPP-4 inhibitors and GLP-1R agonists are incretin-based therapies. GLP-1 is a peptide hormone with multiple actions including insulin secretion stimulation, inhibition of gastric emptying and appetite suppression [27].
Ordinarily, GLP-1 exerts its effects through the GLP-1 receptor. GLP-1 can be cleaved by DPP-4 and neutral endopeptidase 24.11 into smaller fragments which can also exert their effects on surrounding tissues although this is not well understood [27]. DPP-4 is largely responsible for the inactivation of GLP-1 [27].
There are several overlaps between the actions of incretin-based therapies due to their effects on different aspects of signaling pathways. However, several contrasting actions do exist. Notably, GLP-1R agonists have been shown to improve left ventricular (LV) function and heart rate and decrease body weight and blood pressure [27]. DPP-4 inhibitors, on the contrary, have little effect on blood pressure and none on body weight, but have modest effect on improving LV function [27].
At the centre of modern therapy in CKD and T2DM is the modification of underlying risk factors including glycemic control, hypertension and dyslipidemia. In 2021, the multicentre study ‘Nephropathy in Diabetes type 2 (NID-2)’ showed intensive management of these three factors in diabetic nephropathy patients with albuminuria reduced the incidence of MI and stroke [28].
Heart failure is the predominant cardiovascular complication in CKD. The presence of HF before the commencement of renal replacement therapy is a significant independent predictor of mortality, with estimated average survival reduced to three years from five when compared to those without HF. More than half of patients with HF have renal impairment, which increases the likelihood of premature mortality [29]. The management of HF and concomitant CKD or ESRD is complicated by having to balance the nephrotoxicity of mainstay HF therapies and the cardiovascular risks associated with hemodialysis.
Evidence-based medical therapy for HF includes renin-angiotensin-aldosterone (RAAS) inibition, beta-blocke therapy, and manangement of risk factors, including smoking, hypertension, obesity, glycemic control, proteinuria, lipid-lowering therapy, and anemia [30, 31]. The current guidelines for guided-medical therapy for HF with reduced ejection fraction (HFrEF) have recently been updated to include SGLT2i with or without T2DM, in addition to angiotensin-converting enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs), angiotensin receptor-neprilysin inhibitors (ARNIs), beta-blockers, diuretics and aldosterone antagonists as treatments for HF [32].
SGLT2i have exhibited improved glycemic control, renal and cardiovascular outcomes in large-scale clinical trials. Their mechanism of action is through inhibition of glucose reabsorption in the proximal convoluted tubule in the nephron, leading to increased glucose excretion in the urine [9]. The exact mechanism of how SGLT2i improve cardiovascular outcomes is not well established. However, it is hypothesized that a combination and interplay of several factors, including improved glycemic control, reduction in systemic blood pressure, improved endothelial function, and reduced systemic inflammation, contribute to this [9].
CARMELINA [20] evaluated the DPP-4 inhibitor Linagliptin, the only DPP4i RCT identified in our inclusion criteria. This study showed similar cardiovascular outcomes in placebo and intervention groups and no between-group differences in our other selected outcomes. While Linagliptin did not improve these outcomes compared to placebo, it is promising that there were no apparent adverse effects of introducing linagliptin [20]. The selected study population for CARMELINA was at high risk of negative CV and renal outcomes, which may not be indicative of the clinical population. Additionally, the follow-up time of 2.2 years may not have been sufficient to establish further progression to ESRD [20].
CREDENCE [21] was stopped early based on recommendations from its Independent
Data Monitoring Committee (IDMC) when a planned interim analysis showed the trial
had achieved pre-specified criteria for the primary composite endpoint. The
planned duration of the trial was 5 years, however it was was stopped after 2.5
years. There was a 34% relative reduction in its primary composite outcome when
compared to placebo. Moreover, there was significant improvement in
cardiovascular outcomes, including a 39% reduction in hospitalizations for heart
failure [21]. This was one of the first studies to include patients with
significant diabetic kidney disease, in patients with eGFR of 30–90
mL/min/1.73 m
DAPA-CKD included CKD patients both with and without diabetic kidney disease
[22]. Dapagliflozin has become the first SGLT2 inhibitor to be approved for CKD
regardless of diabetes status due to its strong performance in reducing the
decline of eGFR, hospitalization for HF, and cardiovascular death in CKD patients
[36]. The DAPA-CKD trial was discontinued by the data monitoring committee when
an interim analysis showed a statistically significant effect of dapagliflozin on
the primary composite outcome, that would not be expected to change with
additional enrolment. These patients had a lower risk of a decline in eGFR
EMPORER-REDUCED focused on the renal and cardiovascular outcomes of patients with reduced ejection fraction heart failure. Some data were available for patients included with CKD, however not all of this could be related to specifically designed CKD trials. However, the data that was available showed that Empagliflozin has some role in the slowing of eGFR decline and some reduction in all-cause mortality, though the reduction in CV mortality is slim. Furthermore, secondary outcomes were treated in a stepwise hierarchical manner to reduce multiplicity. The mean follow-up period for EMPEROR-REDUCED was relatively short at 1.3 years. Further follow-up studies would be required to ascertain if this improvement is long-standing [23].
EMPORER-REDUCED and DAPA-CKD were the only trials included which had participants enrolled both with and without CKD. The findings of these trials provide the possibility that SGLT2i may have some benefit in CKD patients without diabetes, though subgroup analysis of these patients would be required as well as long-term follow up due to the differing pathology.
SCORED was the only study to evaluate an SGLT2i in the presence of varying
albuminuria [26]. Other SGLT2i studies defined the presence of albuminuria as a
urinary albumin to creatinine ratio of
Liraglutide and Albiglutide are GLP-1R agonists which were investigated in the LEADER and HARMONY [22, 21] trials. Liraglutide is DPP-4 resistant human GLP-1 analogue, whilst Albiglutide is a GLP-1 dimer fused to albumin with a position 2 replacement of alanine to glycine making it resistant to DPP4 degradation [37].
LEADER intervention group had a sustained improvement in cardiovascular outcomes over the course of the trial and a reduction in cardiovascular death as well as reduced hospitalization for heart failure [25]. The study however selected patients with high risk of CV events for inclusion, and so the translation of benefit to those with low risk may not be to the extent seen in the clinical trial. Moreover, as the trial was terminated early the data is restricted to 3.5 year follow-up rather than the anticipated 5-year period [25]. Additionally, outcomes of fatal/non-fatal stroke and MI were not pre-specified outcomes of the study and exploratory outcome p-values were not adjusted for multiplicity, hence these results may not be confirmatory [25]. Further trials with pre-specified analysis would be required to establish this finding fully.
HARMONY results exhibited reduced CV events and improved glycemic control in patients given Albiglutide [24]. HARMONY did not collect data on urinary albumin so inferences of renal safety in CKD should be treated with some caution. Like EMPEROR-REDUCED, the short mean follow-up of 2.2 years may not provide sufficient detail of long-term safety outcomes in CKD patients. Though LEADER and HARMONY had improved CV outcomes, other GLP-1R agonists have not shown improvement [38, 39]. This effect may not be synonymous with all GLP-1R agonists.
This meta-analysis finds that glucose lowering therapies may be beneficial for CV outcomes in CKD patients, however there are some limitations. Notably, EMPORER-REDUCED and DAPA-CKD included participants both with and without T2DM. As data in these were not clearly separated out to account for this difference the data was excluded within a sensitivity analysis.
Moreover, individual studies had differening eligibility criteria including baseline CV risk, influencing cardiovascular outcome data between study populations. Reporting of events across studies was also inconsistent, with some outcome data unavailable.
In summary, large, randomized control trials of oral antihyperglycemic agents in
patients with chronic renal failure have demonstrated some improvement in
cardiovascular outcomes compared with placebo. There is encouraging evidence from
existing trial data that SGLT2i and GLP1R agonists are safe in patients with
chronic renal failure. SGLT2i have significantly lower rates of all-cause
mortality and HHF. GLP1R agonists appear to reduce fatal/non-fatal MI, but across
the class have differing effects on other CV outcomes. Additionally, long-term
follow-up data is required with wider inclusion of patients with eGFR
AK, RS and AM designed the review and performed data extraction and analysis. KM and CB provided advice on study design and interpretation of data. All authors contributed to editorial changes to the manuscript. All authors read and approved the final manuscript.
Not applicable.
We would like to express our gratitude to all individuals who helped in our investigation and to the peer-reviewers for their suggestions.
Professor Colin Berry is supported by research funding from the British Heart Foundation (PG/17/25/32884; RE/13/5/30177; RE/18/6/34217). Dr Andrew Morrow is supported by research funding from the Medical Research Council (MR/S018905/1). Dr Kenneth Mangion is supported by research funding from the Chief Scientist Office (COV/LTE/20/10, COV/GLA/20/05).
Professor Colin Berry is employed by the University of Glasgow which holds consultancy and research agreements for his work with companies that have commercial interests in the diagnosis and treatment of angina. The companies include Abbott Vascular, Astra Zeneca, Boehringer Ingelheim, GSK, HeartFlow, Menarini, Novartis, and Siemens Healthcare. The other authors do not have any potential conflicts of interest.