Effects of Newer Antidiabetic Drugs on Endothelial Function and Arterial Stiffness: A Systematic Review and Meta-Analysis

Background Newer antidiabetic drugs, i.e., dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and glucagon-like peptide-1 receptor agonists (GLP-1 RAs) may exert distinct cardiovascular effects. We sought to explore their impact on vascular function. Methods Published literature was systematically searched up to January 2018 for clinical studies assessing the effects of DPP-4 inhibitors, GLP-1 RAs, and SGLT-2 inhibitors on endothelial function and arterial stiffness, assessed by flow-mediated dilation (FMD) of the brachial artery and pulse wave velocity (PWV), respectively. For each eligible study, we used the mean difference (MD) with 95% confidence intervals (CIs) for FMD and PWV. The pooled MD for FMD and PWV were calculated by using a random-effect model. The presence of heterogeneity among studies was evaluated by the I2 statistic. Results A total of 26 eligible studies (n = 668 patients) were included in the present meta-analysis. Among newer antidiabetic drugs, only SGLT-2 inhibitors significantly improved FMD (pooled MD 1.14%, 95% CI: 0.18 to 1.73, p = 0.016), but not DPP-4 inhibitors (pooled MD = 0.86%, 95% CI: -0.15 to 1.86, p = 0.095) or GLP-1 RA (pooled MD = 2.37%, 95% CI: -0.51 to 5.25, p = 0.107). Both GLP-1 RA (pooled MD = −1.97, 95% CI: -2.65 to -1.30, p < 0.001) and, to a lesser extent, DPP-4 inhibitors (pooled MD = -0.18, 95% CI: -0.30 to -0.07, p = 0.002) significantly decreased PWV. Conclusions Newer antidiabetic drugs differentially affect endothelial function and arterial stiffness, as assessed by FMD and PWV, respectively. These findings could explain the distinct effects of these drugs on cardiovascular risk of patients with type 2 diabetes.


Introduction
Type 2 diabetes (T2D) is a chronic disease affecting 8.3% of the adult population worldwide, with a rising prevalence that renders its tackling a global challenge [1]. Patients with T2D are at high cardiovascular disease (CVD) risk [2,3] and are characterized by micro-and macrovascular dysfunction which is of multifactorial origin [4,5].
The safety and effects of newly licensed antidiabetic drugs on the cardiovascular system represent important clinical issues [6,7]. Recent evidence from clinical trials suggests that newer antidiabetic drugs can not only exert glycemiclowering properties but also decrease CVD risk [8,9]. In this context, sodium-glucose cotransporter-2 (SGLT-2) inhibitors, i.e., empagliflozin in the Empagliflozin Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients-Removing Excess Glucose (EMPA-REG OUTCOME) study [8] and canagliflozin in the Canagliflozin Cardiovascular Assessment Study [10], significantly reduced the rates of CVD events, hospitalization for heart failure (HF), CVD, and total mortality, as well as improved kidney function in T2D patients with established CVD. Similar beneficial effects were reported for liraglutide, an once-daily glucagon-like peptide-1 receptor agonist (GLP-1 RA), and for semaglutide, an once-weekly GLP-1 RA, both of which reduced CVD morbidity and mortality (but not hospitalization for HF) in T2D patients with established CVD, in the Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) trial [9] and the Trial to Evaluate Cardiovascular and Other Long-term Outcomes with Semaglutide in Subjects with Type 2 Diabetes (SUSTAIN-6) [11], respectively. In contrast, lixisenatide once daily and exenatide once weekly did not affect CVD risk in the Evaluation of Lixisenatide in Acute Coronary Syndrome (ELIXA) trial [12] and the Exenatide Study of Cardiovascular Event Lowering (EXSCEL) [13], respectively. Furthermore, dipeptidyl peptidase-4 (DPP-4) inhibitors seem to exert neutral effects on CVD risk as shown for alogliptin in the Examination of Cardiovascular Outcomes with Alogliptin versus Standard of Care (EXAMINE) trial [14] and for sitagliptin in the Trial Evaluating Cardiovascular Outcomes with Sitagliptin (TECOS) [15]. Saxagliptin was reported to increase the rate of hospitalization for HF [16] in the Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR)-Thrombolysis in Myocardial Infarction (TIMI) 53 trial. Despite this evidence provided by large randomized clinical trials, the mechanisms by which antidiabetic drugs can affect CVD risk remain not entirely clear.
Vascular dysfunction is one of the initial steps in the atherosclerotic process [17,18]. Endothelial function and arterial stiffness [17,19] are two widely used indices of vascular function, which both offer prognostic information on the risk of CVD events in T2D patients [19]. Improvement of these indices represents one of the mechanisms by which drugs with established CVD benefits, such as statins, exert their effects [20,21]. Currently, it remains unknown how newer antidiabetic drugs may affect vascular function as studies have yielded conflicting results.
We conducted a systematic review of the literature, followed by a meta-analysis, to investigate the effects of newer antidiabetic drugs, i.e., DPP-4 inhibitors, GLP-1 RAs, and SGLT-2 inhibitors, on vascular function as assessed by flow-mediated dilation (FMD) of the brachial artery and pulse wave velocity (PWV). The PRISMA flow chart for the study is presented in Figure 1.

Study Eligibility.
Studies were eligible if they were fulllength publications in peer-reviewed journals reporting on (a) endothelium-dependent vasodilatory response by FMD and/or (b) arterial stiffness assessed by carotid-femoral, carotid-radial, or brachial-ankle PWV. Studies need also to be either double-blind, placebo-controlled, randomized clinical trials or observational studies assessing the outcomes of interest before and after treatment with a newer antidiabetic drug. No restriction criteria were imposed with regard to the size of the population studied. Analysis did not include studies assessing endothelial function or arterial stiffness by other markers. Only human studies were included in the analysis, whereas review articles were excluded.
2.3. Extraction of Data. Literature search, selection of studies, and extraction of data were performed independently by two investigators (KB and AA). Means and standard deviations (SD) of FMD and PWV as well as their changes following drug treatment were recorded from cumulative published data. In studies reporting standard error, SD was calculated using the equation SD = standard error * √n. In studies reporting median and interquartile range, mean and SD were calculated and used in further analyses [22]. In those studies where extraction of cumulative statistics could not be reliably performed based on the full-length publication, a communication with the corresponding author was attempted to provide summary statistics [23-27].

Statistical Analysis.
For each eligible study, we used the mean difference (MD) with 95% confidence intervals (CIs) for endothelium-dependent vasodilation and arterial stiffness as summary statistics (MD pre-and posttreatment). The difference in mean ± SD for FMD and PWV was included in the quantitative synthesis to explore the pooled MD after treatment with any of the studied antidiabetic drugs; the results were presented in respective forest plots. Subgroup analysis was performed for different drug classes. The presence of heterogeneity was evaluated by the I 2 statistic. A randomeffect model was used to obtain pooled MD and 95% CIs. Results were considered statistically significant at two-tailed p value < 0.05. The findings of the meta-analysis were also confirmed in leave-one-out sensitivity analysis. The quality of published studies was assessed by the modified version of the Downs and Black questionnaire [28]. The effect of publication bias on pooled estimates was explored by the Duval and Tweedie nonparametric "trim and fill" method and the construction of relevant funnel plots. All statistical analyses were performed by STATA software version 13.0 (StataCorp LLC, College Station, Texas, US).

Results
3.1. Qualitative Summary. The literature search identified 30 studies for potential inclusion in the present meta-analysis. Certain identified studies that were originally included in the qualitative synthesis had to be excluded from the quantitative synthesis due to not fulfilling the eligibility criteria. One study was excluded because it assessed endothelial function by reactive hyperemia peripheral arterial tonometry [29] and one because it used a 24 h approach to measure PWV [30]. Also, two studies had to be excluded due to inability to extract reliable summary statistics for the outcomes of interest [26,27] (Figure 1). Therefore, a total of 26 studies (n = 668 patients) were finally included in the quantitative synthesis. The details of the individual studies included in the meta-analysis are summarized in Table 1.
All studies were published since 2012. The sample sizes of the studies ranged from 10 to 51 patients. The mean followup time was 152 days after initiation of treatment. The majority of studies were randomized controlled trials [23-25, 31-48], but some of them were uncontrolled or singlearm observational studies [49][50][51][52][53]. All studies performed the ultrasound-based technique to assess brachial artery FMD and PWV to assess arterial stiffness. are summarized in Figure 2. Overall, 16 studies investigated the effects of DPP-4 inhibitors on FMD. In the pooled meta-analysis, the effects of DPP-4 inhibitors on FMD was not statistically significant (pooled MD = 0.86%, 95% CI: -0.15 to 1.86, p = 0 095). But there was significant heterogeneity between studies (I 2 = 87 8%, p < 0 0001). In leaveone-out sensitivity analysis, the results were unchanged ( Table 2); only by excluding the study of Ayaori et al. [31] was a significant effect of DPP-4 on FMD observed (pooled MD = 1 25%, 95% CI: 0.24 to 2.27, p = 0 015), but heterogeneity among studies remained significant. Even within the group of DPP-4 inhibitors, no specific agent was associated with significant improvements in FMD.

Effects of Newer Antidiabetic Drugs on Arterial Stiffness.
The effects of newer antidiabetic drugs on PWV are summarized in Figure 3. Marked heterogeneity was observed among analyzed studies (I 2 = 80 7%, p < 0 001). A total of 4 studies (n = 132 patients) investigated the effects of DPP-4 inhibitors on arterial stiffness, which were associated with a significant reduction in PWV (pooled MD = −0 18, 95% CI: -0.30 to -0.07, p = 0 002). A similar effect was identified for GLP-1 RAs (2 studies; n = 62 patients), which were associated with  For SGLT-2 inhibitors, only one study reported a reduction in PWV, and therefore, no safe conclusions can be drawn.

Quality Assessment of Included Studies and Publication
Bias. The quality of published studies was assessed by the modified Downs and Black checklist [28]. The quality of published studies is considered moderate with a mean modified Downs and Black score of 23.1. To explore publication bias, we constructed a funnel plot of published studies for the effect size of antidiabetic drugs on the primary endpoint of our study, i.e., endothelial function assessed by FMD. The funnel plot was symmetric, suggesting the absence of publication bias (Figure 4).

Discussion
In the present systematic review, we sought to explore the effects of newer antidiabetic drugs, namely, SGLT-2 inhibitors, DPP-4 inhibitors, and GLP-1 RAs, on vascular function. We hypothesized that the distinct profile of each antidiabetic drug class could be also associated with differences in their vascular effects. The systematic review of the published literature showed that evidence in this field is modest, based mainly on small randomized clinical trials with significant heterogeneity. In this context, published studies supported a beneficial effect of SGLT-2 on FMD, which seems to not be shared by GLP-1 RAs or DPP-4 inhibitors. Accordingly, evidence suggested a reduction in PWV by both DPP-4 inhibitors and GLP-1 RAs. These findings are potentially   important as they suggest a different impact of newer antidiabetic drugs on vascular function, which could be linked with their distinct effects on CVD risk. However, these results should be interpreted with caution given the modest quality of evidence in the published literature and the significant heterogeneity between studies. DPP-4 inhibitors are an antidiabetic drug class on which there is abundant clinical experience, since they have been marketed for over a decade (since 2006). Large randomized clinical trials in the field are consistent in their findings and support a neutral effect of DPP-4 inhibitors on CVD outcomes. However, there is evidence that saxagliptin may increase the risk for HF hospitalization [16]. Recent metaanalyses have also found that DPP-4 inhibitors do not affect the risk for CVD mortality and stroke [54]. Data from the small clinical studies included in the present meta-analysis indicate a marginal effect of DPP-4 drugs on FMD and a significant reduction in PWV. These small effects could be related to the glucose-lowering properties of DPP-4 inhibition but may not be commonly shared by all agents of the DPP-4 drug class. This is an interest finding which (a) confirms the safety profile of this drug class and (b) may explain the neutral effect of certain DPP-4 drugs on CVD outcomes.
For SGLT-2 inhibitors, evidence suggests a beneficial effect of these agents on CVD risk and mortality in T2D patients with established CVD. Evidence from randomized clinical trials suggests that empagliflozin and canagliflozin significantly reduce the CVD morbidity, all-cause mortality, and CVD mortality as well as HF hospitalization and nephropathy development or progression [8]. Similar effects have also been reported for canagliflozin [10]. These benefits could be related to glucose-lowering as well as to reductions in blood pressure, weight, and serum uric acid and to improvements in oxidative stress, glomerular hyperfiltration, albuminuria, arterial stiffness, plasma lipids, sympathetic nervous system activity, myocardial oxygen consumption, and cardiac workload [55]. Our findings complement these SGLT-2 inhibitor actions, suggesting also a significant improvement in FMD. Concerning the impact of SGLT-2 inhibitors on arterial wave velocity, it should be noted that despite the ample evidence on the impact of this class of drugs on natriuresis and blood pressure [8,10], more data are required to assess the effects of these drugs on PWV, as there are only few published data on this topic [46].
Our findings also agree with the published evidence from large randomized clinical trials on the effects of liraglutide [9] and semaglutide [11] on CVD outcomes in T2D patients. In the present meta-analysis, GLP-1 RAs significantly decreased PWV in T2D patients, although they did not affect FMD. Since the effects of these drugs on CVD risk is still debatable [12,56], it remains to be seen whether their impact on vascular function may play a role in determining their CVD effects. The limitations of the existing studies in the field should be noted. The published evidence is modest and mainly based on small-sized randomized clinical studies, some of which were uncontrolled. Furthermore, studies were significantly heterogeneous, and therefore, the results of the present meta-analysis should be interpreted cautiously. The number of the published studies in this field did not allow for subgroup analysis per drug class or for between-agent comparisons within the same drug class. Furthermore, based on the published literature, we cannot conclude whether beneficial effects of SGLT-2 inhibitors, DDP-4 inhibitors, and GLP-1 RAs are due to direct glucose-lowering effects or to indirect effects driven by the modulation of other cardiovascular risk factors such as body weight loss and arterial blood pressure modification [8][9][10][11]57]. More data is also warranted for the effect of SGLT-2 inhibitors on arterial stiffness as well as endothelial function as there are limited published studies reporting on their effects.

Conclusion
In conclusion, the present meta-analysis suggests that the published literature in the field of newer antidiabetic drugs and vascular function is of modest quality and characterized by significant heterogeneity among studies. Overall, both DPP-4 inhibitors and GLP-1 RAs were shown to significantly   decrease PWV without affecting FMD. In contrast, SGLT-2 inhibitors significantly improved FMD, but concrete data on their effects on PWV is still missing. Whether these distinct properties of newer antidiabetic drugs, in relation to their effects on endothelial function and arterial stiffness, may explain their differential effects on CVD risk remains to be elucidated in future studies.

Conflicts of Interest
The authors declare that they have no conflicts of interest.