Background
In recent years, many studies have found that a variety of antidiabetic drugs exert significant protective action on the cardiovascular system, a mechanism that may be partially independent of hypoglycemic effects. This undoubtedly sounds like exciting news for diabetic patients, especially those with cardiovascular disease, although the cardiovascular protective mechanism is not clear. The sodium glucose cotransporter type 2 (SGLT-2) inhibitor empagliflozin has been demonstrated to reduce cardiovascular mortality by 38% and heart failure (HF) hospitalizations by 35% in patients with type 2 diabetes (T2DM) in the EMPA-REG OUTCOMES clinical trial [
1,
2]. An early nonrandomized pilot study showed improved left ventricular function when glucagon-like peptide-1 (GLP-1) agonists were infused in patients with acute myocardial infarction and HF [
3]. Two small randomized controlled trials showed that GLP-1 agonist infusion exerted a positive effect on patients with ischemic heart disease [
4,
5]. Traditional hypoglycemic drugs such as metformin (MET) have also been found to have a positive effect on cardiovascular protection [
6‐
8]. A 2019 meta-analysis found that MET reduced cardiovascular mortality and the incidence of cardiovascular events in diabetic patients [
9]. These results suggest that we should prioritize drugs that have some cardiovascular protective effects when choosing a treatment for diabetes.
The mitigation of left ventricular (LV) remodeling was paralleled by improvements in LV systolic performance. If ventricular remodeling can be delayed, the process of HF can be fundamentally delayed. Recently, a study found that empagliflozin-treated pigs showed higher LV ejection fraction (LVEF) and significantly greater contractile reserve than control animals [
10]. Empagliflozin improved adverse anatomic LV remodeling, enhanced left ventricular systolic function, and inhibited neurohormonal activation. Liraglutide [
11], a GLP-1 agonist, slightly increased LVEF in patients with ST-segment elevation myocardial ischemia (STEMI) who underwent direct percutaneous coronary intervention. Alogliptin, a DPP-4 inhibitor, improves coronary flow reserve (CFR) and LVEF in patients with T2DM with coronary artery disease (CAD), and the improvement in CFR was associated with increased LV systolic function [
12]. Ventricular remodeling is an important determinant of patient morbidity and long-term prognosis [
13]. Although many antidiabetic drugs have the effect of reducing cardiovascular death and adverse cardiovascular events, it is still unclear which antidiabetic drugs can improve ventricular remodeling and fundamentally delay the process of HF. If drugs can be found to improve ventricular remodeling, it will be of great significance for patients with T2DM with cardiovascular disease (CVD).
The purpose of this network meta-analysis is to compare the efficacy of SGLT-2 inhibitors, DPP-4 inhibitors, GLP-1 agonists, MET, sulfonylurea (SU) and thiazolidinediones (TZDs) in improving LV remodeling in patients with T2DM and/or CVD.
Methods
Protocol and guidance
This systematic review and network meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta analyses (PRISMA) guidelines [
14].
Eligibility criteria
We included randomized controlled trials with parallel group or crossover designs in this network meta-analysis, as well as a small number of cohort studies. All trials had to include treatment with one of the following 6 drugs or multiple drugs: MET, GLP-1 agonists, DPP-4 inhibitors, SGLT-2 inhibitors, TZDs and SU. The control group was treated with placebo or one of the 6 drugs. Study participants were either type 2 diabetic patients with or without CVD or patients with CVD alone. The outcome of the included studies must contain at least one of the following 6 cardiac function and structure measures: LVEF, LV end-diastolic diameter (LVEDD), LV end-systolic diameter (LVESD), LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LV mass index (LVMI), early diastolic velocity (e′), early diastolic to late diastolic velocities ratio (E/A) and mitral inflow E velocity to tissue Doppler e′ ratio (E/e′).
We searched articles published before October 18, 2019, regardless of language or data, in 4 electronic databases: PubMed, EMBASE, Cochrane Library and Web of Science. The articles were selected by manual screening.
The following terms were used in the search: ventricular remodeling OR cardiac reverse remodeling OR CRR OR cardiac remodeling OR left ventricular remodeling OR left ventricular dysfunction OR LVD OR ejection fraction OR EF OR left ventricular ejection fraction OR LVEF OR end-diastolic volume OR EDV OR end-diastolic dimension OR EDD OR end-systolic volume OR ESV OR end-systolic dimension OR ESD OR LVEDD OR left ventricular end-diastolic dimension OR LVEDV OR left ventricular end-diastolic volume OR LVESD OR left ventricular end-systolic dimension OR LVESV OR left ventricular end-systolic volume OR left ventricular diameter OR left ventricular volume OR left ventricular mass index OR LVMI OR left atrial volume OR LAV OR left atrial volume index OR LAVI) AND (Dipeptidyl peptidase-4 inhibitors OR DPP-4 inhibitors OR Sodium-Glucose Cotransporter-2 Inhibitors OR SGLT-2 inhibitors OR Glucagon-like peptide-1 agonists OR GLP-1 agonists OR exenatide OR Lyxumia OR liraglutide OR Saxenda OR Tanzeum OR albiglutide OR Trulicity OR dulaglutide OR canagliflozin OR dapagliflozin OR empagliflozin OR ertugliflozin OR ipragliflozin OR luseogliflozin OR tofogliflozin OR sitagliptin OR vildagliptin OR saxagliptin OR alogliptin OR linagliptin OR gemigliptin OR teneligliptin OR metformin OR sulfonylureas OR glibenclamide OR glyburide OR glybenzcyclamide OR gliquidone OR Gliclazide OR glipizide OR Glucotrol OR gliclazide OR Gliclarizonaide OR glimepiride OR Amaryl OR acarbose OR Precose OR air conditioningarbose OR miglitol OR Glyset OR voglibose OR Basen OR repaglinide OR nateglinide OR mitiglinide.
Study selection
Two methodologically trained independent reviewers screened titles and abstracts to determine whether they met the eligibility criteria. The reviewers read the full text and extracted relevant data after consensus was reached. Any differences were resolved through discussion and arbitration, if necessary, by a third reviewer. The reasons for inclusion or exclusion are recorded in detail. Case reports, letters and minutes of meetings were excluded. The PRISMA flow diagram was used to summarize the study selection processes.
Two investigators used a predefined data extraction sheet to independently extract data from each included study, such as authors, publication year, study design, population, subject ages, intervention, male sex, sample size, grouping and number of people in the group, baseline and endpoint data, including counts and effect estimates (mean ± SD), country, follow-up months, title, and conclusion. The third investigator independently reviewed the data to ensure accuracy. If no data in digital format were available, we used the free software Plot Digitizer to estimate data from the graphs.
Definition of outcomes
The outcome of this meta-analysis was the difference in the mean change in echocardiographic parameters between the treatment group and control group. The echocardiographic parameters included LVEF, LVEDD, LVESD, LVEDV, LVESV, LVMI, e′, E/A and E/e′.
Statistical analysis
We used the network meta-analysis approach to evaluate the comparative effect by combining direct and indirect evidence of all relevant treatment effects. To visualize network geometry and node connectivity, we summarized the geometry of the evidence network using network plots. We conducted a network meta-analysis of the comparative efficacy using a multivariate random-effects (restricted maximum likelihood estimation) meta-analysis model. For all treatment comparisons, we present summary mean differences and 95% confidence intervals. To obtain treatment hierarchies, we used a parametric bootstrap procedure with 5000 resamples to compute ranking probabilities. Mean rankings as well as surface under the cumulative ranking curve (SUCRA) values were computed for each treatment. We checked the consistency of the network using local and global inconsistency tests. The local inconsistency test evaluates the loop inconsistency of all the triangle loops on the network. Global inconsistency is a goodness-of-fit test. If any relevant sources of bias were found, we performed sensitivity analyses. All analyses were conducted in Stata/SE, version 14.
Assessment of risk of bias in individual studies
Each study was evaluated using the Cochrane tool. Potential sources of bias include random sequence generation, allocation concealment, blinding of participants and staff, blinding of outcome assessors, incomplete outcome data, and selective reporting. Each trial received a study level score of low, high, or unclear risk of bias for each domain. Two authors independently conducted this assessment, and discrepancies were resolved by consensus.
Assessment of small study effects
To evaluate the presence of small study effects, we visually inspected comparison-adjusted funnel plots for each outcome. We produced funnel plots for all comparisons concerning the difference in LVEF change between treatment and placebo.
Discussion
In this network meta-analysis, 48 studies were included, comprising a total sample size of 4790 subjects. The 48 studies included 36 RCTs and 12 cohort studies. We found that compared with placebo, GLP-1 agonists increased LVEF and decreased LVESV and E/e′, and SGLT-2 inhibitors decreased LVEDD and E/e′. These results suggested that GLP-1 agonists and SGLT-2 inhibitors could improve ventricular remodeling.
Remodeling is an important determinant of patient morbidity and long-term outcomes. Adverse anatomical remodeling occurs at several levels, including anatomical, metabolic, and neurohormonal remodeling. Anatomical remodeling is characterized by LV dilatation, hypertrophy, and geometrical remodeling (the heart becomes more spherical).
Our meta-analysis revealed that GLP-1 agonist treatment triggered an increase in the mean change in LVEF% of 2.04% and a decrease in mean change in LVESV of 4.39 ml compared with placebo treatment. GLP-1 agonists were also demonstrated to significantly reduce E/e. Our subgroup analysis suggested that GLP-1 agonists had a better effect on diabetic patients with CVD than patients with CVD alone. GLP-1 agonists [
61] are an incretin hormone secreted mainly by intestinal L-cells in response to the presence of nutrients. GLP-1 agonists mimic the effects of the native GLP-1 receptor, which increases insulin secretion, inhibits glucagon secretion, increases satiety and slows gastric emptying [
33,
62]. However, the mechanism by which GLP-1 agonists exert their cardiovascular protective action, especially to improve ventricular remodeling, is still unclear. It was reported that liraglutide, one kind of GLP-1 agonist, inhibited angiotensin II and pressure overload-induced cardiac remodeling by regulating PI3K/Akt1 and AMPKα signaling [
63]. Wang et al. [
64] also found that the cardiac protection of GLP-1 agonists might be dependent on inhibition of oxidative stress through the mammalian target of rapamycin complex 1/p70 ribosomal protein S6 kinase pathway. In addition to their weight loss and glucose-lowering effects, GLP-1 agonists have been shown to protect the heart during acute ischemia and improve mitochondrial function, microvascular function, and myocardial glucose uptake in experimental animal models of heart failure [
41,
65]. Giblett et al. [
61] found that GLP-1 agonists are present in left ventricular cardiomyocytes but are not expressed in vascular tissue, so GLP-1 agonists may have direct ventricular effects and mediate secondary vasodilation through ventricular artery interactions. In addition, GLP-1 agonists increase natriuresis [
17], reduce blood pressure, reduce inflammation, reduce ischemic injury, increase heart rate, increase plaque stabilization and decrease smooth muscle proliferation. It has been suggested that the positive effects of GLP-1 agonists on cardiovascular disease may be the result of a direct action on the arteriosclerotic process [
66]. All these processes may leave an imprint on GLP-1 agonists’ role in the improvement of ventricular remodeling.
The results of our study revealed that SGLT-2 inhibitors could significantly reduce LVEDD but could not exert a significant effect on LVEDV. However, SGLT-2 inhibitors could significantly reduce E/e′ and improve diastolic function of the left ventricle. Some studies suggested that SGLT-2 inhibitors could reduce oxidative stress [
66], thereby improving arteriosclerosis and endothelial dysfunction. Experimental data in obese and diabetic mice demonstrated that the SGLT-2 inhibitor empagliflozin significantly ameliorated cardiac fibrosis, coronary arterial thickening, and cardiac macrophage infiltration, suggesting a direct cardiac effect along with an attenuation of oxidative stress on the myocardium [
67]. Previous studies have indicated that SGLT1 receptors are predominantly expressed in the human intestine, and the higher selectivity of SGLT1 receptors could lower the variations in postprandial blood glucose, which might help to reduce heart failure risk. These factors collectively could play a crucial role in reducing the vasculopathy burden on the heart. Other potential beneficial mechanisms, for instance, improved arterial compliance and so on, have also been postulated [
67,
68]. Another study found that SGLT-2 inhibitors in addition to tofogliflozin administration had a favorable effect on left ventricular systolic and diastolic function in patients with T2DM [
65,
69‐
71]. A network meta-analysis of 91 randomized trials by Yang et al. also found that in terms of heart failure risk, sodium-glucose cotransporters 2 were the most favorable option among all classes of antidiabetic medications [
72]. Although SGLT-2 inhibitors seemed to reduce the risk of heart failure, Shao et al. considered that dapagliflozin might have greater effects on heart failure reduction compared to empagliflozin [
73]. This study has shown the effect of SGLT-2 inhibitors on improving LVEDD but not on improving LVEDV. There is a need for much more research in the future due to the paucity of studies on SGLT-2 inhibitors and involved patients.
Of note, in 2018, a network meta-analysis involving 236 studies and 176,310 subjects found that GLP-1 agonists and SGLT-2 inhibitors were significantly associated with lower cardiovascular mortality than were the control treatments, and SGLT-2 inhibitors were associated with a reduction in heart failure events and myocardial infarction. This study is consistent with our results.
Although MET did not exert an effect on ventricular remodeling in our conclusion, a number of experimental and clinical studies have demonstrated that MET had a beneficial effect on lipids, atherosclerotic thrombosis, inflammation, endothelial function, oxidative stress, and antiproliferative and neuroprotective properties. Based on these findings, the recently published guidelines of the American Diabetes Association and the European Association for the Study of Diabetes recommended [
74] either SGLT-2 inhibitors or GLP-1 agonists in patients with T2DM who are unable to achieve their target level of glycemic control with MET. Based on our results, MET treatment in combination with GLP-1 agonists or SGLT-2 inhibitors may also be a good choice for type 2 diabetic patients with cardiovascular disease.
Our results suggested that DPP-4 inhibitors exerted a significant negative impact on LVEDV. Zheng et al. [
75] also found in their meta-analysis that the use of DPP-4 inhibitors was not associated with lower mortality than placebo or no treatment and that the use of SGLT-2 inhibitors or GLP-1 agonists was associated with lower mortality than DPP-4 inhibitors. Studies have found that DPP-4 inhibitor therapy did not increase the overall risk of major cardiovascular and renal outcomes but increased the hospitalization rate for heart failure [
76]. Uncleaved brain natriuretic peptides, which are known to be substrates of the enzyme DPP-4, might be associated with decompensated HF [
77]. Moreover, an increase in LVEDV should result in an increase in the ejection fraction according to the Frank-Starling law. However, our study found that although LVEDV increased with the use of DPP-4 inhibitors, LVEF did not increase. McMurray et al. found that vildagliptin had no major effect on LVEF but did lead to an increase in left ventricular volumes with type 2 diabetes and heart failure [
38]. This finding was consistent with our results. A lack of EF increase suggests a negative impact of DPP-4 inhibitors on myocardial contractility. Studies have reported that DPP-4 inhibition is accompanied by increases in myocardial cAMP, which are related to potentiation of endogenous GLP-1. The increases in cAMP may exacerbate the clinical course of heart failure [
78,
79]. DPP-4 inhibition might exacerbate the clinical course of heart failure via pathways of SDF-1 (stromal cell-derived factor 1)/CaMKII (Ca
2+/calmodulin-dependent protein kinase II). These factors collectively might damage cardiomyocytes, which may be the reason why LVEDV increased, but LVEF did not [
80]. Therefore, this result suggests that we should be cautious about the use of DPP-4 inhibitors in type 2 diabetic patients with cardiovascular disease because DPP-4 inhibitors may have adverse effects on cardiac function.
Limitations
First, this paper included studies that covered three types of patients: one was type 2 diabetic patients, another was cardiovascular patients, and the third was patients with comorbidities of the first two. This may lead to between-studies heterogeneity and exert a certain impact on the combined results. Second, given that the sample size of the included studies ranged from 15 to 559, there was a lack of controlled clinical trials with a large sample size to conduct a more powerful demonstration of our outcome. Third, the variability of the ventricular structural changes estimated by echocardiography may exaggerate or ignore the therapeutic effect, leading to between-studies heterogeneity.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.