Cardiovascular Disease and Type 2 Diabetes: Has the Dawn of a New Era Arrived?

The major benefit of reducing plasma glucose levels in T2D is prevention of long-term microvascular complications and, to lesser extent, macrovascular complications. Individuals with T2D have two- to threefold greater risk of CV events compared with subjects without diabetes, and CVD is responsible for ∼80% of the mortality in T2D (1). Hyperglycemia is a weak risk factor for CVD (5,11), and interventions (2–4) focused on reducing plasma glucose have failed to significantly reduce CV risk and mortality, particularly in secondary prevention trials. Moreover, in the United Kingdom Prospective Diabetes Study (UKPDS) (11) and Veterans Affairs Diabetes Trial (VADT) (12), it took 10 years to observe the CV benefit associated with improved glycemic control. Most individuals with T2D manifest moderate to severe insulin resistance, which is associated with multiple CV risk factors (obesity, dyslipidemia, hypertension, endothelial dysfunction, procoagulant state). This cluster of CV/metabolic disturbances is known as insulin resistance (metabolic) syndrome and is a principal factor responsible for increased CV risk in T2D (13,14). A multifactorial intervention that improves CV risk factors has been shown to reduce CV events and mortality in T2D (15). Further, the molecular mechanisms responsible for insulin resistance directly contribute to pathogenesis of atherosclerosis, independent of associated metabolic abnormalities (13,14). Thus, obese individuals without diabetes but with insulin resistance syndrome manifest a similarly increased risk for CVD compared with T2D patients, supporting the concept that hyperglycemia is not the major risk factor for CVD (16). Consequently, it is not surprising that lowering blood pressure and improving the lipid profile lead to greater reduction in CVD risk than lowering plasma glucose in T2D. Consistent with this, antidiabetes agents like insulin (17), sulfonylureas (18,19), and dipeptidyl peptidase 4 (DPP-4) inhibitors (20–22), which lower plasma glucose without affecting insulin resistance or other metabolic abnormalities associated with insulin resistance syndrome, do not lower CVD risk and mortality in T2D. Conversely, pioglitazone, which improves insulin sensitivity and multiple components of insulin resistance syndrome, i.e., blood pressure, lipids, and endothelial dysfunction (23), exerts a favorable effect on CVD risk in T2D, independent of its glucose-lowering action (9,10). In PROactive (9) pioglitazone lowered MACE (CV death, nonfatal myocardial infarction [MI], nonfatal stroke), which was the main secondary end point, by 16% (P = 0.027), and in IRIS (9) pioglitazone reduced the incidence of recurrent stroke and MI by 24% in insulin-resistant individuals without diabetes who had suffered a recent transient ischemic attack or stroke.

LEADER (6) and SUSTAIN-6 (7) examined the effect of once-daily liraglutide and once-weekly semaglutide on CV risk (Supplementary Table 1). In LEADER (6), 9,340 T2D patients at high CVD risk (82% with prior CV event) were randomized to liraglutide, 1.8 mg/day, or placebo for a mean of 3.8 years. Investigators were blinded to the study intervention and instructed to maintain HbA_1c <7.0% with any antidiabetes medication except a GLP-1 RA or DPP-4 inhibitor. The primary outcome was 3-point MACE. Liraglutide-treated patients experienced a 13% reduction in MACE, which was driven by a 22% reduction in CV mortality (P = 0.007). Nonfatal MI was decreased by 12% (P = 0.11), while nonfatal stroke was reduced by 11% (P = 0.30).

In SUSTAIN-6 (phase 3 trial) (7), 3,297 T2D patients at high risk for CVD were randomized to semaglutide, 0.5 and 1 mg/week, or placebo and followed for 104 weeks. This study was designed to demonstrate noninferiority, which accounts for the smaller number of subjects. The primary outcome was 3-point MACE. Most participants (83%) had established CVD and the remaining 17% were >60 years of age with multiple CV risk factors that were well controlled. Investigators were instructed to lower HbA_1c to <7.0% according to local guidelines without using incretin-based therapies. Subjects receiving semaglutide experienced greater HbA_1c reduction than those on placebo (1.4% vs. 0.4%). This difference (1.0%) in HbA_1c was considerably greater than in LEADER (0.4%). Weight loss (4 kg) and systolic blood pressure reduction (3 mmHg) were twice as great in SUSTAIN-6 versus LEADER. In SUSTAIN-6, the primary outcome (3-point MACE) was reduced by 26%, and that decrease was driven primarily by a 39% reduction in nonfatal stroke (P = 0.04) and a 26% reduction in nonfatal MI (P = 0.12); CV mortality was not decreased (hazard ratio [HR] = 0.98). Of note, the placebo group had significantly more revascularization procedures than the semaglutide group, which could have reduced future deaths; it may be that the similar overall CV death rate between the two groups resulted from the increased number of revascularization procedures in the control group.

Two aspects of LEADER and SUSTAIN-6 deserve emphasis: 1) patients at higher CVD risk benefited more from GLP-1 RA treatment, and 2) the benefit of liraglutide and semaglutide was evident on top of optimal control of traditional CV risk factors. In a meta-analysis of the two long-acting GLP-1 RAs (Fig. 1), 3-point MACE was decreased by 15%, with similar and significant benefit for all three components: nonfatal stroke, nonfatal MI, and mortality by 18%, 16%, and 13%, respectively.

However, there are some concerns about the generalizability of findings in LEADER. In the 23% of participants without a prior CV event, there was no reduction in MACE (HR = 1.20, P not significant). There was more sulfonylurea and insulin use in the placebo group. In SUSTAIN-6, the number of participants was one-third of that in LEADER, the follow-up was short (two years), the decrement in HbA_1c (1.0%) was much greater in the treatment group than the placebo group (although hyperglycemia is not considered to be a major risk factor for CVD), and the incidence of serious eye complications (vitreous hemorrhage, blindness, and photocoagulation) was significantly increased.

In PROactive, 5,238 patients with T2D with established CVD (population similar to EMPA-REG OUTCOME) were randomized to pioglitazone or placebo plus standard of care for CV risk factors and glycemic control (10). The 3-point MACE, the “main secondary end point,” was significantly reduced (HR = 0.84, P = 0.027). The primary end point (3-point MACE plus coronary and leg revascularization) did not reach statistical significance (HR = 0.90, P = 0.09); however, it is now well recognized that peripheral vascular disease is refractory to antihypertensive, lipid-lowering, and glucose-lowering therapy (24,25). Further, by preserving people from death, MI, and stroke, pioglitazone would make more people available for leg revascularization (26).

In IRIS (9), 3,876 insulin resistant (HOMA-IR >3.0) nondiabetic individuals with a recent (within 6 months) ischemic stroke or transient ischemic attack were randomized to pioglitazone or placebo for 4.8 years. Subjects receiving pioglitazone experienced a 24% reduction in fatal/nonfatal stroke plus MI (HR 0.76, P = 0.007); mortality was reduced slightly (by 7%) but not significantly.

Meta-analysis was performed to examine the combined effect of the treatments compared with the placebo-treated control group. Outcomes were expressed as risk ratios and combined risk difference (with 95% CI) and were calculated using a fixed effect model. To examine the appropriateness of the model, Cochran's Q was calculated to measure inconsistency between studies and I² was calculated to describe the percentage of variation. CMA statistical package was used for the analysis. Differences were considered significant at P < 0.05 (Fig. 2 and Supplementary Table 1).

In a meta-analysis of PROactive plus IRIS (Fig. 2), pioglitazone reduced 3-point MACE by 18%, with the main effect being driven by an 18% reduction in stroke and a 26% reduction in MI, while total mortality decreased, but not significantly, by 8%.

In EMPA-REG OUTCOME, empagliflozin caused a 14% reduction (P = 0.04 for superiority) in 3-point MACE in 7,020 patients with T2D with established CVD over 3.1 years (Supplementary Table 1). Several outcomes were surprising and different from LEADER, SUSTAIN-6, PROactive, and IRIS. First, the primary outcome was driven by a robust 38% reduction in CV mortality. Second, there was a striking disconnect between the three MACE components (Fig. 3): 1) for nonfatal MI, HR (0.87) decreased slightly but not significantly (P = 0.22); 2) for nonfatal stroke, HR (1.24) increased slightly but not significantly (P = 0.22); 3) for CV death, HR (0.62) decreased markedly and significantly by 38% (P = 0.001). Third, unlike in LEADER, SUSTAIN-6, and PROactive, separation between the empagliflozin and placebo curves occurred very early, such that reduction in the primary outcome was evident at 3 months after starting treatment.

Although investigators were instructed to maintain HbA_1c <7.0% in all trials, subjects treated with the active drug experienced significantly lower HbA_1c than patients receiving placebo (0.3% in EMPA-REG OUTCOME, 1.0% in SUSTAIN-6, 0.4% in LEADER, 0.5% in PROactive). However, it is unlikely that such HbA_1c differences over 2–4 years can explain the difference in primary outcome (MACE). First, hyperglycemia is a weak risk factor for CV disease. Intensive glycemic control failed to decrease CV events in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) (2), Action in Diabetes and Vascular Disease: Preterax and Diamicron MR Controlled Evaluation (ADVANCE) (3), and VADT (4) studies in T2D patients with long-standing diabetes duration, and in UKPDS (5,11) and VADT (12) it took 10 years to demonstrate a small (∼10%), though significant, reduction in CV events in newly diagnosed individuals. The beneficial CV effects of empagliflozin, liraglutide, and pioglitazone were evident after 3, 18, and 24 months, respectively. More conclusively, IRIS participants were normoglycemic, making it unlikely that improved glucose control was a contributory factor to the reduction in MI and stroke. The HbA_1c difference between semaglutide and placebo in SUSTAIN-6 was clinically meaningful (1.0%), but similar HbA_1c reductions in ACCORD and ADVANCE had no benefit on MACE. Of note, in both LEADER and SUSTAIN-6 the majority of treatment intensification in the placebo group was done with insulin and sulfonylureas, which have been associated with increased atherosclerotic CV events in some studies (14,18,19), although no randomized control trials have demonstrated such an adverse effect.

Reduction in CV events in LEADER, SUSTAIN-6, IRIS, and PROactive started at 1–2 years after initiation of therapy and widened thereafter. This time course is reminiscent of interventions that slow atherosclerosis, e.g., statins and blood pressure–lowering therapy. Similar to results with statins and antihypertensive medications, MI and stroke were reduced in SUSTAIN-6, LEADER, and PROactive, although the magnitude of reduction varied (Figs. 1 and2); the effect on mortality varied most among the three studies. Pioglitazone improves multiple CV risk factors (blood pressure, triglycerides, plasminogen activator inhibitor 1, endothelial dysfunction, insulin resistance, visceral fat) (23) and has documented antiatherogenic effects in preclinical and clinical studies (9,10,23,27,28). Thus, the beneficial effect of pioglitazone in PROactive and IRIS is likely the result of this thiazolidinedione’s antiatherogenic effect.

GLP-1 RAs also improve many CVD risk factors (obesity, hypertension, dyslipidemia, inflammation, visceral fat) in T2D patients. However, the magnitude of improvement in CV risk factors in SUSTAIN-6 and LEADER was modest (2.3 kg weight loss and 1.2 mmHg decrease in systolic blood pressure in LEADER; 3.6–4.9 kg weight loss and 3.4–5.4 mmHg decrease in systolic blood pressure in SUSTAIN-6). LDL cholesterol was not reported in either study. It is unlikely that these changes in CV risk factors can explain the 13% (LEADER) and 26% (SUSTAIN-6) reduction in primary outcome. In ADVANCE, a greater reduction in systolic blood pressure (5.6/2.2 mmHg) was associated with a nonsignificant 8% reduction in MI, stroke, and CV death. Similarly, it is unlikely that weight loss was the principal factor responsible for reduction in the primary outcome. In Look AHEAD: Action for Health in Diabetes, a mean weight loss of 4 kg in patients with T2D (twice that in liraglutide-treated patients in LEADER), as well as modest reductions in HbA_1c, blood pressure, and triglycerides, was associated with a nonsignificant (5%) reduction in MI, stroke, and CV death.

Insulin resistance is a core defect in T2D and is associated with multiple metabolic and CV risk factors, which collectively are known as insulin resistance (metabolic) syndrome (14,16). Furthermore, the molecular etiology of insulin resistance promotes the development of atherosclerosis (14). It follows that interventions which improve insulin sensitivity might reduce CV events in T2D patients. Although neither insulin resistance nor surrogate markers of insulin resistance were measured in any of the CV outcome trials except IRIS, all active interventions (GLP-1 RAs, sodium–glucose cotransporter 2 inhibitors [SGLT2i], thiazolidinediones) are known to improve insulin sensitivity in T2D. Pioglitazone is a powerful insulin sensitizer in skeletal muscle, liver, and adipocytes (23), and in IRIS (9) HOMA-IR decreased by 24% (P < 0.001). GLP-1 RAs do not have a direct insulin-sensitizing effect, but they promote weight loss, which is associated with enhanced insulin action. Lastly, treatment with dapagliflozin for as little as 2 weeks modestly increases insulin-mediated glucose disposal secondary to reversal of glucotoxicity (29). Thus, improved insulin sensitivity could have contributed to the reduction in MACE in LEADER, SUSTAIN-6, EMPA-REG OUTCOME, PROactive, and IRIS.

GLP-1 receptors are expressed in the myocardium and vasculature (30), and GLP-1 and GLP-1 RAs can directly affect CV function via multiple mechanisms: 1) direct action on the myocardium; 2) direct effect on blood vessels to increase nitric oxide production, cause vasodilation, and increase blood flow; 3) direct effect on atherosclerotic plaque formation; and 4) change in autonomic nervous system balance (31–33). With respect to the latter, GLP-1 RAs have been shown to stimulate the parasympathetic nervous system (34). This could explain the increase in heart rate, as well as the cardioprotective effect, seen with this class of antidiabetes drugs. Each of these GLP-1 actions potentially could have contributed to decreased CV events in SUSTAIN-6 and LEADER. If GLP-1 RAs increase coronary blood flow, especially in patients with existing heart disease, this effect could reduce ischemia, infarct size, and risk of arrhythmias. A recent study in subjects with normal glucose tolerance demonstrated that GLP-1 infusion augments blood flow in small vessels in skeletal muscle and heart (35), and increased blood flow in small coronary vessels after myocardial ischemia has been shown to predict increased survival and reduced infarct size after MI (36).

Multiple studies in animals and humans (37) have demonstrated a cardioprotective effect of GLP-1 and GLP-1 RAs on myocardial function after ischemic injury. These benefits include decreased infarct size, increased coronary blood flow, improved left ventricular (LV) function, decreased LV filling pressure, and increased survival. Although the cellular/molecular mechanisms of these GLP-1 actions are not fully understood, similar effects of liraglutide and semaglutide on the heart in LEADER and SUSTAIN-6 could have contributed to the decrease in CV events and death. Studies in animals have demonstrated that the postischemic beneficial effects of GLP-1 on the heart are preserved in animals lacking the GLP-1 receptor, suggesting a GLP-1 receptor–independent mechanism (38). In humans, beneficial effects of GLP-1 RAs on LV function and filling pressure have been observed with exenatide (37), liraglutide (39), and native GLP-1 (40).

Lastly, GLP-1 RAs can reduce CV events by slowing the atherosclerotic process. Studies in experimental animals have demonstrated that liraglutide accelerates endothelial recovery after injury (41), retards atheroma formation in apolipoprotein E knockout mice (42), exerts anti-inflammatory effects on the vasculature, and inhibits reactive oxygen species formation and platelet aggregation (43). Collectively, these actions of GLP-1 RAs could slow the atherosclerotic process.

The beneficial effect of empagliflozin on CV events differs from that of pioglitazone and GLP-1 RAs with respect to both the time course and the individual components benefited (i.e., MI vs. stroke vs. mortality). The beneficial effect of empagliflozin on MACE was driven by a robust reduction in CV mortality, was rapid in onset, and was associated with a marked decrease in hospitalization for heart failure. Empagliflozin did not significantly reduce the risk of MI, while stroke risk increased slightly. As previously reviewed (44), the impressive reductions in mortality and heart failure most likely are explained by the rapid and simultaneous reductions in blood pressure (afterload reduction), intravascular volume (preload reduction), and arterial stiffness—i.e., hemodynamic effects—and not by slowing of the atherosclerotic process. Consistent with this, a recent preliminary study demonstrated that empagliflozin treatment for 3 months reduced LV mass index and improved diastolic dysfunction (45). Other factors suggested to account for the beneficial CV effects in EMPA-REG OUTCOME have been reviewed and include increased circulating ketone levels, reduced uric acid, and increased angiotensin (1–7) and angiotensin type 2 receptor activity, among others (44). All of these proposed mechanisms focus on nonatherosclerotic processes.

Because the beneficial CV effects of empagliflozin most likely are hemodynamically mediated, while those of GLP-1 RAs and pioglitazone represent a direct action on the vasculature (plus improved CV risk factors) to retard atherogenesis, it is plausible that combination therapy with empagliflozin plus pioglitazone and/or a GLP-1 RA will exert an additive, even synergistic, CV benefit (Fig. 4). Empagliflozin profoundly reduced CV mortality, whereas pioglitazone and GLP-1 RAs primarily reduced the risk of nonfatal MI and nonfatal stroke, so addition of empagliflozin to pioglitazone or a GLP-1 RA may produce a robust reduction in all three MACE components. Well-designed large, randomized, placebo-controlled studies should be performed to examine whether combination therapy with an SGLT2i, GLP-1 RA, and/or pioglitazone can produce an additive effect to further reduce CV events compared with monotherapy with these agents.

Combination therapy with an SGLT2i or GLP-1 RA with pioglitazone has other potential benefits. Pioglitazone improves diastolic dysfunction, enhances myocardial insulin sensitivity and reduces myocardial fat content (46), and decreases blood pressure (23). However, these CV benefits can be offset by the drug’s sodium retentive effect on the kidney. Because SGLT2i (47) and, to lesser extent, GLP-1 RAs (31) exert a natriuretic effect, the renal salt retentive effect of pioglitazone will be negated. On the other hand, it is possible that the renal sodium retentive effect of pioglitazone could reduce some of the hemodynamic benefits produced by the volume-depleting effect of the SGLT2i. Since the side effects (fluid retention and weight gain) of pioglitazone are dose related (48), we do not recommend doses in excess of 30 mg/day. Pioglitazone also promotes fat weight gain by stimulation of hypothalamic appetite centers (49). However, the fat weight gain primarily represents a cosmetic concern because the greater the weight gain, the more the decrement in HbA_1c and the greater the improvements in β-cell function and insulin sensitivity (50,51). Furthermore, pioglitazone reduces CV events (9,10,28). The weight gain can be negated by combining pioglitazone with an SGLTi or GLP-1 RA or both (52). Combination therapy with an SGLT2i and GLP-1 RA is especially effective in reducing body weight (53). The only drug shown to conclusively reduce liver fat and reverse biopsy-proven nonalcoholic steatohepatitis (NASH) is pioglitazone (54). Both GLP-1 RAs (55) and SGLT2i (56) reduce visceral (hepatic) fat, making combination therapy with any two or three of these drugs an attractive option for preventing/treating NASH and nonalcoholic fatty liver disease (NAFLD), and we recommend that such a study be carried out. In EMPA-REG OUTCOME, empagliflozin reduced the composite end point of renal disease by 39% (57). Although less well appreciated, thiazolidinediones also prevent diabetic nephropathy in diabetic animal models (58), and liraglutide in LEADER significantly reduced the composite renal outcome, although this was primarily due to its effect to decrease proteinuria (6). Thus, combination therapy with any of these three classes of antidiabetes medications may provide an additive renal protective effect. Pioglitazone is a potent insulin-sensitizing agent (23) and markedly enhances and preserves β-cell function (50,51,59). SGLT2i cause a modest improvement in insulin sensitivity (by 33%) and major improvement in β-cell function (by 217%) (29,60). GLP-1 RAs exert a powerful effect to increase β-cell function (52,61–63) and indirectly improve insulin sensitivity by promoting weight loss (62). Thus, SGLT2i, GLP-1 RAs, and pioglitazone represent a triad of antidiabetes medications that, when used in combination, may provide additive effects in preventing CV complications, promoting weight loss, preserving renal function, preventing NASH/NAFLD, and improving β-cell function and insulin sensitivity. Patients treated with GLP-1 RAs plus pioglitazone should have an eye exam before initiating combination therapy with these two agents because of an increased incidence of eye complications in SUSTAIN-6 and rare cases of macular edema with pioglitazone use.

Although the increase in stroke in EMPA-REG OUTCOME did not achieve statistical significance, it nevertheless presents a worry. The mechanisms responsible for any such increased stroke risk, despite decreased blood pressure, are unclear. Since both pioglitazone (9,10) and GLP-1 RAs (6,7) significantly reduce the incidence of stroke in T2D patients, it is possible that addition of a GLP-1 RA or pioglitazone to empagliflozin will prevent any increase in stroke risk. Although results of the Dapagliflozin Effect on Cardiovascular Events (DECLARE) study (due in 2019) and the CANagliflozin cardioVascular Assessment Study (CANVAS)/Study of the Effects of Canagliflozin on Renal Endpoints in Adult Subjects With Type 2 Diabetes Mellitus (CANVAS-R) (due in 2017) are not available, a recent meta-analysis (64) suggests that all three SGLT2i will exert similar effects on CV end points and congestive heart failure.

It is not possible at this time to determine whether the CV risk reduction with liraglutide (LEADER) and semaglutide (SUSTAIN-6) is a class effect or a specific effect inherent to each individual agent. It also is difficult to determine whether other SGLT2i will exert a similar benefit on CV mortality as empagliflozin, although published data with dapagliflozin and canagliflozin suggest that this will be the case (64). Ongoing CV outcome studies with other GLP-1 RAs (exenatide, dulaglutide) and other SGLT2i (dapagliflozin, canagliflozin) will provide an answer to this question. Previous CV outcome studies reported a neutral effect on MACE of other antidiabetes agents that act via the incretin axis (lixisenatide, alogliptin, saxagliptin, and sitagliptin) (20–22). However, the ability of DPP-4 inhibitors to raise circulating GLP-1 levels is modest, whereas GLP-1 RAs achieve much higher plasma GLP-1 levels (>100 pmol/L) than DPP-4 inhibitors (∼30 pmol/L). Lixisenatide has a short half-life (∼4 h). Thus, patients are uncovered during most of the day, and this could explain its lack of CV benefit. Consistent with this, the reductions in body weight and blood pressure were smaller in ELIXA (Evaluation of Cardiovascular Outcomes in Patients With Type 2 Diabetes After Acute Coronary Syndrome During Treatment With AVE0010 [Lixisenatide]) compared with LEADER (0.6 vs. 2.4 kg and 0.8 vs. 1.4 mmHg). Studies in experimental animals have demonstrated that some of the cardioprotective effect afforded by GLP-1 is independent of the GLP-1 receptor (38). It is possible, therefore, that the cardioprotective action of GLP-1 RAs is structure dependent, and lixisenatide has only ∼50% sequence homology with native GLP-1. Lastly, the study design and patient population in ELIXA differed significantly from LEADER. Patients in ELIXA were recruited because they had acute coronary syndrome and the primary end point was the composite of CV death, MI, stroke, and unstable angina (MACE-4).

Participants in LEADER, SUSTAIN-6, EMPA-REG OUTCOME, and PROactive had T2D and >80% had a previous CV event. It is not possible to determine whether T2D patients without established CVD and who are earlier in the natural history of their disease will similarly benefit from treatment with GLP-1 RAs, SGLT2i, or pioglitazone. It can be argued that, because pioglitazone and GLP-1 RAs slow atherosclerosis and because T2D patients are at high risk for atherosclerotic complications, these agents are likely to reduce CV events in individuals with T2D without, as well as with, established CVD. Previous studies have demonstrated that pioglitazone improves surrogate (anatomical) markers of CVD (carotid intima-media thickness and coronary atheroma volume) in T2D patients without established CVD (27,28). However, it also could be argued that since the reduction in CV events produced by GLP-1 RAs, pioglitazone, and empagliflozin was seen in patients with a prior CV event, the antidiabetes medications would not be effective or would be less effective in T2D patients without evidence of CVD. Therefore, at this stage, if considering only cardiovascular risk, evidence-based medicine dictates that pioglitazone, semaglutide, liraglutide, and/or empagliflozin should be considered to reduce CVD risk in T2D patients with established CVD, as in the published CV outcome trials. An exception is pioglitazone, which significantly reduced the incidence of stroke/MI in insulin-resistant individuals without diabetes.

The primary goals of T2D management are to 1) improve glycemic control to prevent microvascular complications and 2) normalize CVD risk factors to reduce CV events and CV mortality. Hyperglycemia is the principal risk factor for microvascular complications, and a decrease in HbA_1c, by whatever means, reduces the risk of eye, kidney, and nerve complications. Review of CV outcome trials in this Perspective demonstrates that liraglutide, semaglutide, pioglitazone, and empagliflozin reduce macrovascular events by 14–26%, independent of their glucose-lowering effect and improvement in traditional CV risk factors (Fig. 4). Although not established in T2D individuals with lower CV risk, we propose that long-acting GLP-1 RAs (liraglutide and, if approved by the U.S. Food and Drug Administration, semaglutide), SGLT2i (empagliflozin until the results of CANVAS and DECLARE become available), and pioglitazone should be given preferential consideration, along with metformin, over other agents that similarly lower HbA_1c but have not been shown to reduce CV risk. Although metformin is recommended as first-line therapy by the American Diabetes Association and was reported to reduce CV events in UKPDS (5,11), the number of subjects (n = 342) was small and would not meet the standards of current CV outcome trials. On the other hand, it could be argued that the ability to demonstrate a difference with such a small sample size speaks to the significance of the results.

Funding. This work was supported by a grant to R.A.D. from the Foundation for the National Institutes of Health (NIH-DK-24092-42). R.A.D.’s salary is supported in part by the South Texas Veterans Health Care System, Audie Murphy Division.

Duality of Interest. R.A.D. is on the advisory boards of AstraZeneca, Novo Nordisk, Janssen, Intarcia, and Boehringer Ingelheim; receives research support from Bristol-Myers Squibb, Boehringer Ingelheim, Takeda, Janssen, and AstraZeneca; and is on the speakers’ bureaus of Novo Nordisk and AstraZeneca. S.D.P. has served on the scientific board and received honoraria for consulting fees from AstraZeneca, Boehringer Ingelheim, Eli Lilly and Co., GlaxoSmithKline, Merck & Co., Novartis Pharmaceuticals, Novo Nordisk, Sanofi, Servier, and Takeda Pharmaceuticals. R.C. has received research grants and has been an advisor for Amgen, Boehringer Ingelheim, Sanofi, Merck Sharp & Dohme, Bristol-Myers Squibb, AstraZeneca, and Eli Lilly and Co. R.E.J.R. has received speaker fees, consultancy fees, and educational sponsorships from Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.