Cardiovascular effects of antiobesity drugs: are the new medicines all the same?

Abstract

Waiting for a definite answer from well-designed randomized prospective clinical trials, the impact of the new antiobesity drugs -liraglutide, bupropion/naltrexone, phentermine/topiramate and lorcaserin- on cardiovascular outcomes remains uncertain. What has been learned from previous experience with older medicines is that antiobesity drugs may influence cardiovascular health not only causing weight reduction but also through direct actions on the cardiovascular system. Therefore, in the present review, we examine what is known, mainly from preclinical investigations, about the cardiovascular pharmacology of the new antiobesity medicines with the aim of highlighting potential mechanistic differences. We will show that the two active substances of the bupropion/naltrexone combination both exert beneficial and unwanted cardiovascular effects. Indeed, bupropion exerts anti-inflammatory effects but at the same time it does increase heart rate and blood pressure by potentiating catecholaminergic neurotransmission, whereas naltrexone reduces TLR4-dependent inflammation and has potential protective effects in stroke but also impairs cardiac adaption to ischemia and the beneficial opioid protective effects mediated in the endothelium. On the contrary, with the only exception of a small increase in heat rate, liraglutide only exerts favorable cardiovascular effects by protecting myocardium and brain from ischemic damage, improving heart contractility, lowering blood pressure and reducing atherogenesis. As far as the phentermine/topiramate combination is concerned, no direct cardiovascular beneficial effect is expected for phentermine (as this drug is an amphetamine derivative), whereas topiramate may exert cardioprotective and neuroprotective effects in ischemia and anti-inflammatory and antiatherogenic actions. Finally, lorcaserin, a selective 5HT_2C receptor agonist, does not seem to exert significant direct effects on the cardiovascular system though at very high concentrations this drug may also interact with other serotonin receptor subtypes and exert unwanted cardiovascular effects. In conclusion, the final effect of the new antiobesity drugs on cardiovascular outcomes will be a balance between possible (but still unproved) beneficial effects of weight loss and “mixed” weight-independent drug-specific effects. Therefore comparative studies will be required to establish which one of the new medicines is more appropriate in patients with specific cardiovascular diseases.

Introduction

The approval of four new antiobesity medicines -liraglutide, bupropion/naltrexone, topiramate/phentermine and lorcaserin- has been enthusiastically welcomed because they could lower the body weight in patients not responding to diet and physical exercise. Achieving a good control of weight is considered a primary objective of obesity treatment as high BMI values are associated with an increase in all-cause mortality [1] that can be reduced by weight loss [2]. The role of cardiovascular disorders in this scenario is still controversial. In fact, while obesity is associated with a higher prevalence of hypertension and other major cardiovascular risk factors, and coronary heart disease and stroke are among the main causes of death in obese patients [1], there is also some evidence that moderate obesity could, instead, reduce the severity of cardiac failure and coronary artery disease, the so-called “obesity paradox” [3]. Despite the positive results of some studies, the clinical evidence accumulated so far does not show any clear benefit that body weight reduction improves cardiovascular outcomes [4].

When dealing with antiobesity drugs the situation turns out to be even more complicated because these medicines may exert direct effects on the cardiovascular system independent from their anorexigenic activity. The relevance of the problem dramatically emerged with medicines such as sibutramine that was withdrawn from the market because of dangerous effects on the cardiovascular system. The lesson learned was that antiobesity drugs are not necessarily beneficial for cardiovascular health as, in some circumstances, they can also be very dangerous. Therefore, in the present review, we will examine what is known about the cardiovascular pharmacology of these new medicines with the aim of highlighting potential mechanistic differences in their expected cardiovascular effects. We will show that, though all of the drugs lower body weight, the new medicines are very different from each other as far as their cardiovascular effects are concerned. This suggests that their final effect on cardiovascular outcomes will be a balance between possible (but still unproved) beneficial effects of weight loss and “mixed” weight-independent drug-specific effects. Therefore comparative studies are eagerly awaited in order to establish which one of the new therapies is more appropriate in patients with specific cardiovascular diseases.

Methods

To prepare the present review we used the Pub Med (https://www.ncbi.nlm.nih.gov/pubmed) and Semantics Scholar (www.semanticscholar.org) search engines by crossing the drug names liraglutide, bupropion/naltrexone, topiramate/phentermine and lorcaserin with the following keywords: cardiovascular, heart, myocardial infarction, stroke, blood pressure, hypertension, plasma lipids, atherogenesis, endothelial dysfunction and endothelium. No specific limit was put in the time interval to be covered by our search.

Bupropion-Naltrexone

The Bupropion/Naltrexone combination (BUP/NAL) is approved for use in obesity both in the US and in the EU. Bupropion and Naltrexone both possess central anorexigenic properties, though quite weak in the case of naltrexone. They have been combined in a single medicine to take advantage of their synergism as naltrexone may antagonize compensatory opioidergic mechanisms that limit bupropion efficacy. As discussed in the next sections, both Bupropion and Naltrexone may exert relevant cardiovascular effects independent of body weight decrease.

The question whether BUP/NAL is safe for the cardiovascular system or even decreases cardiovascular risk continues to be open even though evidence from a randomized clinical trial, the LIGHT trial, has already been published [5]. This trial was, indeed, prematurely stopped by the FDA because of serious confidentiality breaches that led to the disclosure of preliminary findings suggesting a positive effect on cardiovascular outcomes. Nevertheless, the results of the LIGHT trial including those collected at the time 25 and 50% of the planned major cardiovascular events (MACE) (death from cardiovascular causes, nonfatal myocardial infarction, or nonfatal stroke) had occurred and when the study was stopped, were published in 2016 [5]. Whereas the results at 25% MACE actually suggested some cardiovascular benefit from the medicine, those collected at 50% of MACE only showed non-inferiority vs placebo at the limit of 2.0 but not of 1.4, as planned (Table 1). Data at 64% MACE, when the study was closed, indicated a worse safety profile with a higher prevalence of nonfatal stroke in the BUP/NAL. As the trial was not completed and given the different results of the analyses performed at different time points, no definite conclusion can be drawn on cardiovascular safety of this drug combination.

Table 1.

Clinical trials with new antiobesity drugs on cardiovascular outcomes.

	LIGHTa			LEADER trial			SCALE TRIALSb			CAMELLIA–TIMI 61
Treatment (n)	Bup/Nal (4455)	Placebo (4450)	HR (CI)	Liraglutide (4668)	Placebo (4672)	HR (CI)	Liraglutide (3384)	Placebo (2036)	HR (CI)	Lorcaserin (6000)	Placebo (6000)	HR (CI)
MACE (n, %)	90 (2.0)	102 (2.3)	0.88 (0.57–1.34)	608 (13)	694* (14.9)	0.87 (0.78–0.97)	8 (0.002)	10 (0.04)	0.42 (0.17–1.08)	364 (6.1)	369 (6.9)	0.99 (0.85–1.14)
CV Death (n, %)	17 (0.4)	34 (0.8)	0.50 (0.21–1.19)	219 (4.7)	278* (6.0)2	0.78 (0.66–0.93)	2 (0.0006)	2 (0.001)	0.56 (0.08–4.08)	90 (1.5)	86 (1.4)	1.04 (0.78–1.40)
Nonfatal Stroke (n, %)	21 (0.5)	19 (0.4)	1.10 (0.44–2.78)	159 (3.4)	177 (3.8)	0.89 (0.72–1.11)	3 (0.0009)	3 (0.001)	0.47 (0.09–2.32)	84 (1.4)	98 (1.6)	0.86 (0.64–1.15)
Nonfatal MI (n, %)	54 (1.2)	54 (1.2)	1.00 (0.57–1.75)	281 (6.0)	317 (6.8)	0.88 (0.75–1.03)	4 (0.001)	5 (0.002)	0.44 (0.12–1.68)	225 (3.8)	227 (3.8)	0.99 (0.82–1.19)

Note that inclusion criteria of the mentioned studies were different. Specifically, patients with obesity and high cardiovascular risk were enrolled in the LIGHT and the CAMELLIA-TIMI 61 trials, whereas the LEADER trial was performed in obese and non-obese diabetic patients with high cardiovascular risk and the SCALE TRIALS in patients with obesity not selected for a specific severity of the cardiovascular risk.

MACE major adverse cardiovascular events, MI myocardial infarction, Bup/Nal Bupropion/Naltrexone.

*P < 0.05

^aOnly the results of the 50% interim analysis are reported here because they are the last data collected at a scheduled time point before the study was closed by FDA for confidentiality breaches. The cited paper (ref. [5].) also reports the data collected at the previous time point (25% of MACE) and at the time the study was stopped (64% of MACE-unscheduled time point). See text for more details.

^bThe data here reported are from the merged results of the four SCALE trials as reported in ref. [52].

Bupropion

Bupropion is an atypical antidepressant that blocks presynaptic recapture of dopamine and noradrenaline hence enhancing dopaminergic and noradrenergic neurotransmissions both centrally and peripherally, also including the cardiovascular system. Even though bupropion has a mechanism of action similar to cocaine it is considered safer for the cardiovascular system because of its lower potency both on dopamine and on noradrenaline transporters. Nevertheless, significant cardiovascular effects have been described for this drug as detailed in the following paragraphs.

Cardiac effects

Bupropion exerts positive inotropic effects as demonstrated by its ability to increase the contractile response elicited in vitro by noradrenaline in rat ventricular strips [6] and by electrical stimulation in strips from human atrial trabeculae [7]. Bupropion also weakly blocks IKs (hERG) with an IC₅₀ of 34 μM, and QT prolongation but not torsades de points has been observed during acute intoxication with this drug [8]. In addition, bupropion widens the QRS complex and delays action potential propagation possibly by decreasing the electrical coupling among cardiomyocytes [8]. The increased response to noradrenaline and the conduction disturbances caused by bupropion could imply an increased risk of myocardial ischemia and arrhythmias similar to what happens with cocaine. Sparse reports of these adverse events after bupropion ingestion have been published and according to the French Pharmacovigilance Database 10.1% of all patients taking bupropion suffered ischemic cardiac events [9].

Effect on blood pressure

Bupropion increased the pressor response to intravenously-injected noradrenaline in anesthetized rats [6]. In humans, acute treatment with rapidly acting oral bupropion caused a small but significant increase in arterial blood pressure both when used for smoking cessation and for the treatment of depression [10]. Evidence that bupropion could affect blood pressure was obtained also in volunteers with mild hypertension treated with the extended release formulation of this drug for 4 weeks [11]. In most of cases a bupropion-induced hypertensive effect is considered not clinically relevant even though the drug may be responsible for hypertensive crisis with tachycardia in the setting of overdosage [12].

Effect on plasma lipids and atherogenesis

Recent evidence shows that bupropion like other antidepressants reduces TNFα release and has anti-inflammatory properties [13]. Whether these anti-inflammatory effects could impact on the development of atherosclerotic plaques is still unknown. Tsai et al. [14] showed that in monocytic THP-1 cells stimulated in vitro with LPS, bupropion suppressed the release of the interferon-γ-inducible protein 10 (IP-10), a cytokine associated with the development of atherosclerotic lesions. However, this drug did not decrease intima/media thickness ratio or enhance acetylcholine-induced relaxation in the thoracic aorta from rats on a high fat diet [15], and no association between markers of subclinical atherosclerosis and the use bupropion or other antidepressants was observed in the 10-year follow-up of the Multi-Ethnic Study of Atherosclerosis (MESA) study [16].

Naltrexone

Naltrexone is a non-selective opioid receptor blocker with about tenfold higher affinity for μ than for κ or δ receptors. This effect is stereospecifically exerted only by the (−) enantiomer whereas the (+) enantiomer is ineffective. A wealth of data shows that endogenous opioids and opiate drugs affect the cardiovascular system acting both on the blood vessels and on the heart, which have been shown to express opioid receptors [17], and, on the brain [18]. The scenario that emerges from these studies is that endogenous opiates have a role in the adaption to acute cardiovascular events that could be attenuated by the use of their antagonists such as naltrexone though with uncertain clinical consequences.

Effect on the heart

Endogenous opioids protect the myocardium from ischemic damage and reduce the occurrence of ischemia-induced arrhythmias [19]. This effect of opioids was originally observed in the context of the so called ischemic preconditioning, i.e., in the cardioprotection against major ischemia afforded by repeated episodes of subthreshold ischemic episodes. More specifically, it was shown that preconditioning is abolished by the non-specific opioid antagonist naloxone and reproduced by repeated morphine administration [20]. Likewise opioids have a role in ischemic postconditioning [21], i.e., the protection against reperfusion damage afforded by brief bouts of ischemia delivered after a major myocardial ischemia. Evidence has been reported that that endogenous opioid cardioprotection towards ischemic damage also occurs in humans [22].

Opioid preconditioning largely depends on the activation of cardiac opioid receptors by endogenous opioids, both circulating and locally synthesized in the ischemic myocardium, even though the activation of central opioidergic pathways may contribute as well [23]. The direct cardiac preconditioning effect of opioids is exerted through δ receptors, whose activation has been unequivocally shown to be protective, and κ opioid receptors whose role has been more controversial [24]. Conversely, μ opioid receptors do not seem to be implicated in preconditioning by direct cardiac effect but they could cooperate with δ and κ opioid receptors in mediating the effects of preconditioning that are exerted in the CNS [23]. Opioids are implicated also in other forms of cardioprotection that could be abrogated by naltrexone such as the cardioprotection elicited by the chronic exposure to moderate hypoxia [25]. In addition, through the stimulation of κ opioid receptors endogenous opioids mediate both remote ischemic pre- and post-conditioning, i.e., the reduction of myocardial ischemic damage induced by brief episodes of ischemia delivered to peripheral organs (e.g., a limb) respectively before [26] or after an episode of major myocardial ischemia [27].

Opioids also play a role in chronic heart failure. In particular, in the failing heart there is a significant increase in the cardiac expression of endogenous cardiac opioids and of δ and κ and, in some experimental models of heart failure, also of µ opioid receptors [28]. The functional consequences of the activation of the opioidergic system in heart failure are complex and may vary with the receptors involved and the animal species considered. In the normal heart, opioids decrease cardiomyocyte contractility and this effect may be enhanced in heart failure as demonstrated in a hamster model of hypertrophic cardiomyopathy [29]. Moreover, the opioid antagonists naloxone and naltrexone have beneficial hemodynamic effects and improve the baroreflex sensitivity in dogs with experimental heart failure [30]. While these data suggest that opioid antagonists could have beneficial effects in heart failure it has also to be considered that endogenous opioids may exert cardioprotective effects in the failing myocardium [28] that could be lost upon opioid receptor blockade. It is presently unclear how these findings from preclinical models translate to the evolution of heart failure in humans. In fact, whereas in patients with chronic heart failure, high plasma concentrations of proenkephalins are predictive of cardiovascular-related hospitalization and death [31], in a very small series of patients with NYHA class III/IV chronic heart failure the acute administration of naloxone did not modify hemodynamic parameters or clinical symptoms [32]. Therefore, in the absence of larger studies the effect of opioid antagonists and, in particular, of naltrexone in chronic heart failure remains uncertain.

Effect in stroke receptors

Opioid antagonists may protect the brain from ischemic damage. This was initially suggested by the report of two human patients with stroke who experienced a significant clinical improvement after the administration of naloxone [33], and has been confirmed in experimental animal models [34]. Importantly, protective effects were demonstrated also for other opioid antagonists including naltrexone [35]. The mechanism responsible for ischemic neuroprotection by opioid antagonists is still uncertain. The decrease in blood brain permeabilization and edema [36], the preservation of cerebral blood flow, an antiseizure effect [37], an antiapoptotic effect mediated by the inhibition of the NIK/IKKα/NF-κB pathway [38] may all have a role. A major determinant of opioid antagonist-dependent neuroprotection is, however, the ability of these compounds to reduce microglial activation and neuroinflammation [39]. These antinflammatory effects are largely independent from the blockade of opioid receptors and may involve the non-stereospecifical inhibition of Toll-like receptors 4 as demonstrated for naloxone and naltrexone [40]. Whether circulating naltrexone in patients treated with BUP/NAL could exert any protective effect against stroke is at the moment undetermined also considering that clinical trials with naloxone in stroke patients showed no obvious benefit.

Effect on blood pressure and on endothelium

Endogenous opioids regulate arterial blood pressure through central and peripheral mechanisms by causing vasodilation and blunting baroreceptor reflexes [17, 41]. These effects have been documented in experimental animals and in humans and are believed to be quite small in physiological conditions. Some evidence has been reported that an opioidergic dysfunction could be involved in the genesis of arterial hypertension as well although the question remains open [41]. On the contrary, opioidergic activation is a major mechanism keeping blood pressure low in shock [42]. Opioid antagonists may reverse the effect on blood pressure of endogenous opiates [43] and an increase in arterial blood pressure is mentioned among the possible unwanted effects in the prescribing information for naltrexone.

Most of the direct vasodilating effects of opioids depend on pharmacological actions exerted on the endothelium. The presence of μ3 receptors and their coupling to nitric oxide production and vasodilation has been demonstrated in endothelial cells [44]. In addition, κ opioid receptor stimulation with the selective agonist U50,488H causes pulmonary artery vasodilation in rats by inducing NO generation [45]. The effect of opioid agonists on NO synthesis and release could be part of a more complex beneficial role of these peptides in endothelial dysfunction. It has been demonstrated, indeed, that U50,488H protects the aortic endothelium from the oxidative damage due to oxidized lipoprotein accumulation and preserves NO synthesis in rats fed with a high fat diet, by activating the PI3K/Akt signaling pathway [46]. This drug also normalizes vascular tone and reduces thickness and fibrosis of the aortas of rats made diabetic with streptozotocin [47]. Endomorphins, which preferentially bind μ opioid receptors, prevent the decrease in eNOS expression induced in HUVEC cells by the exposure to high glucose concentrations [48] or ox-LDL [49] and inhibit lipid droplet accumulation and foamy cell transformation in monocyte/macrophages cultured in vitro by decreasing the expression of CD36 scavenger receptors [50].

In conclusion, bupropion and naltrexone are expected to exert both positive and detrimental cardiovascular effects. In fact, bupropion, which may increase heart rate and blood pressure by potentiating catecholaminergic neurotransmission, also has beneficial antinflammatory effects, whereas naltrexone is expected to impair cardiac adaption to ischemia and the beneficial protective effects mediated by endogenous opioid in the endothelium, could have a protective effect in stroke and reduce TLR4-dependent inflammation.

Liraglutide

Before being approved for the treatment of obesity, liraglutide has been used for years in diabetic patients (though at a lower dosage) and, therefore, a wealth of data has been collected on its cardiovascular effects. In addition, a controlled, randomized trial, the LEADER trial, performed to assess its safety in people with diabetes with high cardiovascular risk, showed a 13% decrease in MACE [51] (Table 1). A post-hoc analysis of the SCALE trials, the clinical studies that led to liraglutide approval as an antiobesity drug, showed no detrimental effect and a possible benefit on MACE also in people with obesity [52] (Table 1).

The evidence that GLP1 receptors (GLP1R) are also expressed in the cardiovascular system and in the kidney suggests that GLP1 agonists like liraglutide can exert direct effects on the cardiovascular system [53]. In the next sections we will briefly review the evidence suggesting that liragutide and other GLP1 agonists exert direct anti-ischemic effects on the heart and on the brain, affect blood pressure and heart rate, reduce endothelial dysfunction and have antiatherogenic effects. The effect of GLP1 agonists are not necessarily identical to those of native GLP1_7-36 because this peptide partially acts in a GLP1R- independent manner through its degradation product GLP1_9–36, which is not a GLP1R agonist [54]. The scenario is further complicated by the evidence that important differences could also exist among different GLP1 agonists with some drugs showing clearly positive effects on cardiovascular outcomes (liraglutide, semaglutide and albiglutide) and others not (extended release exenatide and lixisenatide) [55].

Effect in myocardial ischemia and heart failure

The first hint that GLP1 could exert important cardioprotective effects in heart ischemia came from a clinical study in humans that showed a significant improvement in myocardial function in patients with acute myocardial infarction who received the infusion of this hormone after angioplasty [56]. Thereafter, GLP1-induced reductions in infarct size, preservation of contractile function or both were demonstrated in animal models of myocardial infarction [57]. Beneficial effects after acute myocardial infarction were observed also with GLP1 analogs including liraglutide both in experimental animals [58] and in human patients [59]. GLP1 may protect the myocardium from ischemic damage both when it is given prior to ischemia, i.e., as a preconditioning mimetic, or after ischemia at reperfusion [60]. Interestingly, a role has been proposed for GLP1 also in remote pre- or per-conditioning i.e., in the protection elicited when a short-lasting ischemia is induced at distal sites, generally at a limb, before or during the induction of myocardial ischemia [61]. GLP1 protective effects are also observed when experimental ischemia is delivered to the isolated heart or primary cardiomyocyte cultures hence suggesting that they are, at least in part, directly exerted in the heart [62]. Multiple pathways seem to be involved in GLP1-dependent cardioprotection. A first mechanism is the inhibition of the intrinsic apoptosis pathway through the activation of the reperfusion injury salvage kinases, ERKs and PI3K with their downstream effectors mTOR and p70s6k [62]. Recent evidence suggests that liraglutide can also prevent intrinsic apoptosis by enhancing mitophagy through a SIRT-1-Parkin dependent pathway [63] and by promoting the expression of the smooth endoplasmic reticulum Ca²⁺ pump through a PI3K/Akt pathway hence lowering the intracellular Ca²⁺ overload that occurs in ischemia and contributes to mitochondrial dysfunction [64].

The cardioprotective effects of GLP1 and, surprisingly, liraglutide are partly independent from GLP1R activation, being still observed in GLP1R ko mice and partially reproduced by GLP1_9–36 [54]. The molecular mechanism of these GLP1R-independent responses remains uncertain as no alternative receptor for GLP1_9–36 has been identified so far. Part of these cardioprotective effects could also be exerted indirectly through an action on the vasculature leading to coronary vasodilation [54]. Evidence has been reported that GLP1_7–36 and its metabolite GLP1_9–36 dilate coronary arteries in vivo in humans, a mechanism that can contribute to its protective effect in myocardial ischemia [65]. Finally, the ability of GLP1 and its analogs to promote new vessel formation may also have a role in cardioprotection by promoting tissue repair. Indeed, an increase in new blood vessel formation in the ischemic myocardium has been observed in rats after the intracardiac injection of liraglutide-coated nanoparticles [66]. A recent meta-analysis of the clinical studies on liraglutide in patients with myocardial infarction showed an improvement in left ventricular ejection fraction and a reduction in high-sensitivity C reactive protein, but no change in cardiac death, major adverse cardiovascular events, repeated revascularization or recurrence of MI [67].

GLP1 also exerts positive inotropic effects as indicated by the increase in left ventricular developed pressure in isolated mouse hearts exposed to GLP1_7–36 [54] and, on the opposite, by the impaired contractile response to inotropic agents observed in GLP1R ko mice [68]. In the rat heart GLP1_7–36 inotropic effects depend on the activation of GLP1R and are not replicated by GLP1_19–36 [54]. Experiments performed in isolated human atrial strips showed that exenatide increases the force of isometric contraction through a GLP1/PKA/phospholamban-dependent pathway [69]. The evidence that GLP1 improves left ventricular function in dogs with pacing-induced cardiomyopathy suggested that GLP1 and its analogs could be clinically useful also in heart failure [70]. In mice, liraglutide reduces the contractile dysfunction, inflammation, fibrosis and cardiomyocyte apoptosis induced by obesity and diabetes suggesting that it could improve diabetic cardiomyopathy [71]. These effects are independent from body weight reduction and involve the activation of AMPK and autophagy [71] and the inhibition of endoplasmic reticulum stress-induced apoptosis [72]. In addition, exendin-4 prevents the apoptosis induced by saturated fatty acids, an important mechanism involved in the pathogenesis of diabetic cardiomyopathy [73]. Whether the beneficial effects of GLP1 and its analogs observed in experimental animals also occur in patients with heart failure is still a matter of controversy. In fact, whilst an improvement in hemodynamic parameters has been observed in some studies [74] it has not been confirmed in others including the two randomized trials LIVE and FIGHT [75, 76].

Effect in stroke

Liraglutide and other GLP1 analogs decrease brain damage in experimental brain ischemia in rodents [77] and increase the survival of cultured neurons exposed to hypoxia in vitro [78]. Among the mechanisms that could account for this neuroprotective effect an important role is played by the enhanced expression of anti-apoptotic proteins of the Bcl₂ family through the activation of the PI3K/AKT and MAPK pathways and the concomitant decrease of the pro-apoptotic proteins Bax and Bad [79]. A relevant consequence of the enhanced Bcl₂ expression is the decrease in mitochondrial ROS generation, which also contributes to neuroprotection [77, 79]. By preserving mitochondrial integrity liraglutide also modulates autophagy in neurons [80]. Enhanced reparative angiogenesis can also have a role [81]. Evidence for the clinical relevance in humans of the neuroprotective effect of liraglutide is still limited and in the LEADER study the prevalence of stroke considered in isolation (i.e., not as part of MACE) was not significantly decreased by liraglutide [51].

Effect on arterial blood pressure

GLP1 and GLP1R agonists including liraglutide decrease arterial blood pressure in hypertensive rats [82] and in human patients with diabetes or metabolic syndrome [83]. In randomized clinical trials liraglutide caused a decrease of about 3 mmHg in systolic blood pressure, which tended to disappear approximately one year from the beginning of the treatment, whereas it had no effect on diastolic blood pressure [83]. The decrease in blood pressure in humans is accompanied by a slight increase in heart rate, which was also observed in rodents and depends both on a direct action on the heart and on the activation of the sympathetic system [84]. Because GLP1 vasorelaxant properties were also observed in vitro in rings from isolated blood vessels it was proposed that GLP1 could directly act on the vasculature as also suggested by the presence of GLP1R on endothelial and vascular smooth muscle cells [54]. The evidence that GLP1-induced vasorelaxation was abrogated by NOS inhibitors [85] or when the vessels were denuded of their endothelium [86] suggested that GLP1, which is known to induce the expression of eNOS in endothelial cells, was acting by enhancing NO release from the endothelium. The blood pressure lowering effect of GLP1 may be part of a more general improvement of endothelial dysfunction [87]. Additional mechanisms can also be in place considering that in other experimental systems GLP1 was still effective after removing the endothelium. In particular, it can relax vascular smooth muscle cells through the PKA-dependent activation of KATP channels [88]. These effects were also replicated by GLP1_9–36 suggesting that GLP1R was not involved [54, 88]. Contrasting data have been reported about the ability of degradation-resistant GLP1 analogs to trigger GLP1R-independent vasodilation [88]. Liraglutide was ineffective in relaxing endothelium denuded aortic rings in vitro that, instead, was relaxed by the coronary perfusate from the heart of mice treated with this drug [89]. Because liraglutide lowered the blood pressure in GPL1R^+/+ but not in GLP1R^−/− mice and this effect was abolished by the atrial natriuretic peptide (ANP) antagonist anantin, it was concluded that this GLP1 analog lowers blood pressure indirectly by inducing ANP synthesis in the atria through a GLP1R/cAMP/cAMP-activated guanine nucleotide exchange factor pathway [89].

Effect on atherogenesis

GLP1 and its analogs exert anti-atherogenic effects by simultaneously acting on vascular smooth muscle cells, endothelial cells and foamy cells. Endothelial cells are the main target of liraglutide anti-atherogenic effects. GLPR1 agonists preserve viability in endothelial cells exposed to free radicals and prevent their senescence [90], exert anti-inflammatory effects, increase eNOS expression and NO release and decrease the expression of adhesion molecules hence reducing monocyte margination [91]. Notably also TNFα-induced PAI-1 expression is inhibited and this could contribute to decrease the risk of thrombus formation [91]. Importantly, GLP1 agonists improve endothelial dysfunction also in human patients with diabetes [92]. Liraglutide and other GLP1R agonists reduce monocyte margination, their accumulation in the atherosclerotic plaque, their uptake of ox-LDL and foamy cell transformation and the release of proinflammatory cytokines [93]. These effects could be partly dependent on the ability to affect the phenotype of infiltrating cells by enhancing the prevalence of M2 anti-inflammatory relative to M1 pro-inflammatory macrophages [94]. The stimulation of GLP1R in vascular smooth muscle cells elicits antiproliferative and anti-inflammatory effects. In effect, the treatment with exendin-4 reduced myointimal thickening in mice with endothelial denudation injury of the femoral artery [95]. In addition, this GLP1 analog also reduced the proliferation of aortic vascular smooth muscle cells cultured in vitro in the presence of PDGF [95] or angiotensin II [96]. This antiproliferative effect depends on AMPK activation and contributes to the antiatherogenic effect of liraglutide in ApoE deficient mice [96].

In conclusion, with the only exception of a small increase in heat rate, liraglutide exerts favorable cardiovascular effects by protecting myocardium and brain from ischemic damage, improving heart contractility, lowering blood pressure and reducing atherogenesis.

Phentermine/Topiramate

Phentermine (PHEN) and Topiramate (TPM), two drugs both lowering body weight with different mechanisms, have been combined in a single medicine, fixed dose PHEN/TPM that was approved by FDA in 2012. Limited data are available on the cardiovascular safety of the PHEN/TPM in human patients. The Phase III Registration trial CONQUER and EQUIP and the extension study SEQUEL showed that PHEN/TPM decreased several cardiovascular risk factors including plasma glucose, LDL and triglycerides, systolic and diastolic blood pressure and, obviously, body weight. No prospective randomized controlled trial specifically designed to assess hard cardiovascular endpoints, such as MACE, has been published so far.

Recently, a large retrospective cohort study of real-world US insurance billing data compared the occurrence of MACE in a group of patients taking either TPM, or PHEN or fixed dose PHEN/TPM, before, during and after the end of the treatment and showed that MACE prevalence was lower when patients were taking fixed dose PHEN/TPM or PHEN [97]. By contrast, TPM use was associated with an increase in MACE occurrence [97]. However, this was an uncontrolled, retrospective study with the important limitation of the very short duration of drug exposure (on average 2.1 years). In addition, the safety in patients with cardiovascular diseases was not specifically assessed.

In the following sections we review the evidence that supports the hypothesis that both PHEN and TPM may affect cardiovascular risk independently from their effects on body weight.

Phentermine

Amphetamines are prototypical cardiotoxic agents and the concept that they can induce a whole array of cardiovascular unwanted effects is so well established that it will not be examined in detail in the present review. The point here is instead that PHEN, which is a substituted amphetamine, has been claimed to be less cardiotoxic than the other molecules of this family [98]. Its lesser potency has been proposed as an explanation of this possible higher tolerability [98]. Whilst the few and small randomized trials published so far suggest that PHEN is safe for the cardiovascular system, a few cases of myocardial infarction, stroke, and ventricular arrhythmias [98] have been reported in patients taking this drug. Therefore, the question of PHEN cardiovascular safety remains open in the absence of solid evidence from randomized clinical trials specifically designed to address this issue.

Topiramate

TPM acts on multiple targets blocking voltage-gated ion channels, glutamate AMPA receptors and carbonic anhydrase and activating GABA_A receptors. Even though most of TPM antiobesity effects are exerted centrally through the modulation of hypothalamic neuronal pathways controlling feeding behavior, TPM also acts on its targets peripherally and these peripheral actions may be relevant for cardiovascular risk reduction. More specifically, preclinical evidence suggests that TPM could exert protective effects in ischemic disorders such as stroke and myocardial infarction, decrease atherogenesis, improve insulin sensitivity and preserve endothelial integrity in diabetes.

As with other antiepileptic drugs, TPM reduces the severity of neuronal damage in in vitro [99] and in vivo [100] experimental brain ischemia. The blockade of voltage-gated Na⁺ channels and of AMPA receptors is an obvious explanation of TPM-induced neuroprotection because the activation of these ion channels represents a major death mechanism in the ischemic brain. However, more recently, evidence emerged of an additional mechanism depending on the anti-inflammatory effects. TPM, indeed, dose-dependently decreases NFkB activation, the levels of myeloperoxidase and the release of inflammatory cytokines in the brain of rats undergoing experimental subarachnoid hemorrhage [101]. TPM anti-inflammatory effects are not limited to ischemic brain damage as they have been documented also in experimental liver [102] and, notably, cardiac ischemia [103]. TPM acts directly on white blood cells through the modulation of their intrinsic GABAergic system [104]. In effect, monocyte/macrophages and T-lymphocytes synthetize GABA and express GABAergic receptors [104, 105] and GABAergic drugs may affect the evolution of autoimmune diseases such as autoimmune encephalomyelitis by acting on these cells [105]. In mice TPM decreased the cardiac damage induced by the ligation of the coronary artery, reduced the prevalence of heart rupture and increased survival by reducing the percentage of M1/Ly-6C^high macrophages, that have proinflammatory and tissue destroying activities, at the same time increasing tissue repairing M2/Ly-6C^low macrophages [103]. By modulating the intrinsic GABAergic system in macrophages TPM reduced lipid droplet accumulation and foamy cell transformation in macrophages derived by human monocytes exposed in vitro to oxidized LDL [106]. This effect was dependent on the decrease of the expression of SR-A, CD36 and LOX-1, which are responsible for LDL uptake by macrophages, and on the increase in the expression of ABCA1, ABCG1 and SR-BI, which, conversely, promote cholesterol efflux. Importantly, TPM also caused a decrease in p38MAPK and NF-κB pathway activation and in TNFα production suggesting that it could also reduce inflammation in atherosclerotic plaques. In contradiction to this hypothesis, TPM did not induce any decrease in foamy cell accumulation or atherosclerotic plaque formation in high fat fed apoE-deficient mice [107]. However, these negative results could have been determined by the extremely high plasma lipid levels that are attained in high fat fed apoE-deficient mice. Interestingly, despite the lack of effect on atherosclerotic plaque formation TPM markedly reduced glomerular lipidosis and the related kidney damage.

By blocking carbonic anhydrase TPM protects the blood brain barrier from hyperglycemia-induced damage preventing astrocyte and pericyte loss and the loosening of tight and adherens junctions [108]. Specifically, TPM reduces the generation of HCO3⁻ which is required for the conversion of pyruvate to oxaloacetate that will be further metabolized in the Krebs cycle with free radical generation.

In conclusion, while no direct cardiovascular beneficial effect is expected for PHEN due to its amphetamine-like properties, TPM may exert cardioprotective and neuroprotective effects in ischemia and anti-inflammatory and antiatherogenic actions.

Lorcaserin

The synthesis of the selective 5HT_2C agonist lorcaserin was the result of a program aiming to obtain a drug that could activate 5HT_2C receptor-dependent anorexigenic pathways in the hypothalamus without stimulating other serotonin receptors (mainly of the 5HT_2B subtype) which may induce serious unwanted cardiovascular effects. Clinical evidence that lorcasein is, indeed, safe for the cardiovascular system has been reported only at the end of 2018 with the publication of the results of the CAMELLIA–TIMI 61 trial [109] (Table 1). In this study no difference in MACE occurrence was observed in patients with obesity and established atherosclerotic cardiovascular disease or multiple cardiovascular risk factors randomly assigned to receive either lorcaserin or placebo. Of note, while this study showed that lorcaserin is safe it did not report any evidence of a reduction of cardiovascular outcomes.

Though remarkably more selective than other 5HT_2C agonists, at doses higher than 20 mg/day lorcaserin may still exert effects on 5HT_2A and 5HT_2B receptors  since the Ki for these receptor subtypes is only about 100-fold higher than for 5HT_2C receptors. 5HT_2C receptors are mainly expressed in the central nervous system [110] and therefore, the cardiovascular effects of 5HT_2C stimulation, if any, are exerted centrally. 5HT_2C agonists decrease the activity of neurons in the nucleus of tractus solitarius which receive the afferent inputs of cardiac baroreflex arc [111]. In addition, in unstressed rats the intracerebroventricular injection of the 5HT_2C agonist mCPP caused an increase in mean arterial blood pressure that is initially accompanied by bradycardia and, then, by tachycardia [112]. This effect was prevented by the pretreatment with the 5-HT_2C receptor antagonist SDZ SER082 suggesting a role of central 5HT_2C receptors in the control of heart rate and blood pressure. However, mCPP is far from being specific [113] and SDZ SER082 also blocks 5HT_2B receptors [114]; these considerations cast doubts on the real involvement of 5HT₂ receptors in the aforementioned effect also considering that no cardiovascular abnormality has been observed in 5HT_2C receptor ko mice [115]. Although 5HT_2C receptors have been traditionally considered as not present the cardiovascular system, a few studies suggested that 5HT_2C receptor activation could induce vasoconstriction in specific vascular beds such as the perfused hindquarter in anesthetized rats [116] or the mesenteric vascular bed [117]. While these results were obtained with agonists whose specificity could be questioned, Zhang et al. [118] showed that in rats that developed pulmonary hypertension because of exposure to chronic hypoxia the expression of 5HT_2C receptors evaluated with Western blots, immunocytochemistry and RT-PCR did increase in smooth muscle cells of the pulmonary artery. Importantly, siRNA against these receptors lowered the proliferation of pulmonary smooth muscle cells in vitro. Likewise, Morán et al. [119] supported their hypothesis that mCPP-induced vasoconstriction in the in situ autoperfused kidney from normotensive rats was due to the activation of 5HT_2C receptors by demonstrating the expression of this receptor type in renal arteries by Western blot experiments. If confirmed, these results could deserve some attention because they suggest that lorcaserin could exert unwanted vasoconstricting activity in critical vascular beds such as the renal and pulmonary ones.

In conclusion, lorcaserin does not seem to exert significant direct effects on the cardiovascular system though at very high concentrations it could lose some of its specificity and start acting on 5HT_2A and 5HT_2B receptors, which, conversely, have a relevant role in cardiovascular pathophysiology.

Conclusions

We performed a review of the literature on the cardiovascular effects that are exerted independently from the decrease in body weight by the active substances contained in the four recently approved antiobesity medicines. Relevant differences emerged among these drugs since some of them show unequivocal cardioprotective or antiatherogenic effects whereas others could detrimentally impact on cardiovascular risk (Table 2). Therefore their final effect on cardiovascular risk will be the combined result of the benefits of body weight lowering and of  drug-specific effects. In this perspective, future comparative studies also considering gender differences [120] are mandatory to establish whether any of the antiobesity drugs do offer advantages as compared with the others and, importantly whether in terms of precision medicine prescription, specific subgroups of patients should better be treated with one particular antiobesity medicine rather than with another.

Table 2.

Summary of the cardiovascular effects of the new antiobesity drugs.

	Bupropion	Naltrexone	Liraglutide	Topiramate	Phentermine	Lorcaserin
Effect on myocardial ischemic damage	↑ (9)	↑a	↓ (56–67)	↓ (103)	↑ b	n.a.
Effect on myocardial contractility	↑ (6,7)	↑ (28,30–32)	↑ (54–76)	n.a.	↑ b	n.a
Risk of arrhythmias or conduction disturbances	↑ (8,9)	↑a	↑ (53)	n.a.	↑ b	n.a.
Effect on ischemic brain damage	n.a.	↓ (33–40)	↓ (77–81)	↓ (99–101)	↑ b	n.a.
Effect on arterial blood pressure	↑ (6,10–12)	↑ (42,43)	↓ (82–89)	n.a.	↑ b	↑ (?) (112–119)
Effect on endothelial dysfunction	n.a.	↑ a	↓ (90–92)	↓ (106–107)	n.a.	n.a.
Effect on inflammation in the atherosclerotic plaque	↑ (13–15)	n.a.	↓ (94)	↓ (102–105)	n.a.	n.a.
Effect on foamy cell formation	n.a.	↑ a	↓ (93)	↓ (106–107)	n.a.	n.a.

^aExpected as a consequence of the reversal of physiological effects of endogenous opioids but not directly demonstrated.

^bExpected as a class effect of amphetamines although limited evidence suggests that phentermine could be safer than other amphetamines.

An important note of caution is that many of the studies that we reviewed were performed in animal models or in human patients without obesity and therefore they could not be immediately translatable to patients with obesity. Accordingly, it is important to mention that cardiovascular disorders could be different from a pathophysiological point of view in patients with and without obesity as observed, for instance, in the case of arterial hypertension.

Author contributions

The authors’ responsibilities were as follows: MC, AC, and FG: were responsible for the concept of this paper and drafted the manuscript; GM, LB, SS and AC: provided a critical review of the paper. OPERA Group members participated to the revision of the manuscript. All authors and OPERA Group Members contributed to and agreed on the final version of the manuscript.

Compliance with ethical standards

Conflict of interest

AC received lecture fees from Eli Lilly, Novo Nordisk, Sanofi Aventis, Astra Zeneca, Bruno farmaceutici, Roche. The remaining authors have nothing to disclose.