Cardiovascular impact of drugs used in the treatment of diabetes

Abstract

The International Diabetes Federation predicts that by 2035 10% of the population of the world will have been diagnosed with diabetes, raising serious concerns over the resulting elevated morbidity and mortality as well as the impact on health care budgets. It is also well recognized that cardiovascular disease is the primary cause of the high morbidity and mortality associated with diabetes, raising the concern that appropriate drug therapy should not only correct metabolic dysfunction, but also protect the cardiovascular system from the effects of, in particular, the epigenetic changes that result from hyperglycaemia. A number of new classes of drugs for the treatment of diabetes have been introduced in the past decade, providing the opportunity to optimize treatment; however, comparative information of the cardiovascular benefits, or risks, of the newer drugs versus older therapies such as metformin is variable. This review, in addition to summarizing the cellular basis for the therapeutic action of these drugs, addresses the evidence for their cardiovascular benefits and risks. A particular focus is provided on metformin as it is the first choice drug for most patients with type 2 diabetes.

Introduction

The global impact of obesity and diabetes continues to increase and negatively affect morbidity, mortality and health care budgets. Reports in 2011 from the International Diabetes Federation [IDF, 2013] stated that there was an estimated 285 million people worldwide who had already been diagnosed with diabetes and that the worldwide prevalence of diabetes has truly reached pandemic levels [Chen et al. 2011; Zimmet, 2011]. Furthermore, increases in the prevalence of diabetes for all age groups worldwide are predicted, with total numbers to reach approximately 450 million in 2030 amounting to 7% of the population of the world. The most recent estimate predicts 592 million in 2035 or approximately 10% of the total population [Chen et al. 2011; Zimmet, 2011; IDF, 2013]. A serious concern is for developing countries where it is predicted there will be a 69% increase in the number of adults with diabetes compared with so-called developed countries where a 20% increase is anticipated [Shaw et al. 2010]. Type 2 diabetes (T2D) accounts for approximately 90% of all cases of adult-onset diabetes and the increase in the prevalence of T2D and associated insulin resistance can be linked to a rise in obesity resulting from a combination of lifestyle changes and genetic susceptibility [Daousi et al. 2006].

A natural consequence of an increased incidence in diabetes is the high likelihood that this will be accompanied by a marked rise in the occurrence of cardiovascular disease [Nolan et al. 2011]. Although the aetiology of vascular dysfunction in diabetes has been extensively investigated we still have not optimized the therapeutic management of diabetes such that the cardiovascular system is appropriately protected. It is also very important to emphasize that diabetes-associated vascular complications result in 75% of the deaths linked to diabetes [Grundy et al. 2002]. In addition there is a two- to fourfold increase in the incidence of coronary artery disease, a 10-fold increase in peripheral vascular diseases and, overall, a three- to fourfold higher mortality rate when the data are compared with those from people without diabetes [Grundy et al. 2002]. Essentially diabetes is a disease that has a profound negative effect on the health of the human vascular system and clearly a better understanding of the cellular basis of diabetes-associated cardiovascular disease is an important goal. To date, diabetes management has largely focused on the control of hyperglycaemia; however, since the rising burden of this disease is mainly correlated to its vascular complications, it is also important to consider how best to protect cardiovascular function and, therefore, not only focus on the reduction of plasma glucose [Schalkwijk and Stehouwer, 2005; Fowler, 2008; Defronzo, 2009]. We now recognize that advances in our understanding of the pathophysiology of diabetic cardiovascular disease will facilitate the discovery of potential new treatment regimens. Since the use of drugs for the treatment of diabetes is frequently a major component of the management of diabetes, it is also important that both the beneficial and potentially negative cardiovascular effects of these same drugs are fully recognized [Ledford, 2013; Singh et al. 2013].

Diabetes, hyperglycaemia, glucose toxicity and cardiovascular disease

The results from the 1991 United Kingdom Prospective Diabetes Study (UKPDS), with 5102 subjects with T2D, and the 1993 Diabetes Control and Complications Trial (DCCT), with 1441 subjects with type 1 diabetes (T1D) helped establish the dogma that the most beneficial approach to the management for patients with diabetes is tight glycaemic control [DCCT, 1993; UKPDS, 1991, 1998]. Such conclusions have been reemphasized as reflected by a report by Standl and colleagues indicating that the incidence of cardiovascular events associated with diabetes is reduced by 10–15% per 1% reduction in absolute glycated haemoglobin (HbA_1c) [Standl et al. 2009]; the level of HBA_1c being a measure of the average impact of plasma glucose over the previous 4–12 weeks. In addition, the Emerging Risk Factors Collaboration (ERFC) study, a collaborative meta-analysis of 102 prospective studies, made the observation that the preexistence of diabetes more than doubled the risk of vascular disease [ERFC et al. 2010]. Furthermore, the ERFC study of 820,900 participants also concluded that a 50 year old with diabetes died approximately 6 years earlier than a person without diabetes and that diabetes was moderately associated (but not necessarily causality) with death from certain cancers, such as liver, pancreas, bladder, breast, colorectal, lung and ovary as well as infectious diseases, degenerative disorders and other diseases [ERFC, 2010]. Nonetheless, a ‘glucocentric’ approach to the treatment of diabetes has been delivered a setback with the results, released in 2007, from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study of 10,251 patients. The intensified glucose lowering arm of the ACCORD study with 5128 patients with a targeted HBA_1c of less than 6% was discontinued when it became apparent that such an intensive regimen resulted in a significantly higher all-cause risk of death of 22% and a 35% increase in cardiovascular mortality [ACCORD, 2008]. A contribution to the increase in mortality associated with the intensified glucose-lowering arm of the ACCORD study may be linked to an increase in the frequency of hypoglycaemia-related events and associated endothelial-vascular dysfunction [Wang et al. 2012]. Clearly, in order to improve the clinical management of diabetes, a better understanding of the impact of both hypoglycaemia and hyperglycaemia on the progression of endothelial dysfunction and vascular disease is required [Ding and Triggle, 2010; Basha et al. 2012; Triggle and Ding, 2010, 2011; Triggle et al. 2012].

Insulin, insulin resistance, hyperglycaemia, oxidative stress and endothelial dysfunction

Insulin was first discovered in 1921 by Banting, Best and McLeod [Best and Scott, 1923] and the importance of insulin in mediating multiple cellular effects is now well recognized. Insulin, released from pancreatic β cells, mediates its anabolic and metabolic cellular effects via trans-phosphorylation activation of its plasma membrane heterotetramer receptor. Activation of the insulin receptor catalyzes the tyrosine phosphorylation of a number of substrates, including the insulin receptor substrate (IRS1-4) proteins, initiates activation of the phosphatidylinositol 3 (PI-3) kinase pathway that, in turn, activates protein kinases, notably Akt (protein kinase B) and atypical protein kinase C. Via the phosphorylation of an adapter protein, APS, insulin also regulates Glut4 translocation and the regulation of glucose transport into skeletal, cardiac muscle and adipose tissue with approximately 80% of glucose disposal going into skeletal muscle [Saltiel and Kahn, 2001]. Once released from pancreatic β cells, the rate-limiting step for the blood-borne insulin to reach its target tissues and mediate its hormonal effects is transport across the endothelium. King and Johnson demonstrated that insulin can be transported into endothelial cells by an insulin receptor-mediated process [King and Johnson, 1985]. Insulin is internalized into capillary and arterial endothelial cells following the activation of insulin receptors on the endothelial cell by a nitric oxide (NO)-independent process that requires PI-3-kinase and mitogen-activated protein kinase (MAPK)/ERK signalling [Jialal et al. 1984; Hachiya et al. 1988; Brunner and Wascher, 1998; Genders et al. 2013]. Insulin uptake is dependent on the expression of caveolin 1 and is reduced by the inflammatory cytokines, interleukin 6 (IL-6) and tumour necrosis factor α (TNFα) [Wang et al. 2011]. Insulin resistance is associated with endothelial dysfunction and with a reduced delivery of insulin across the endothelium (Mather et al. 2013).

Insulin plays an important role in the regulation of vascular tone and, within the physiological concentration range, promotes vasodilatation [Baron, 1994; Steinberg et al. 1994]. Insulin-mediated regulation of vascular function is endothelium dependent via PI-3-kinase/Akt activation of endothelial NO synthase (NOS) and the generation of NO [Dimmeler et al. 1999; Montagnani et al. 2002; Clark et al. 2003]. NO facilitates capillary recruitment and increased blood flow in the microcirculation [Clark et al. 2003; Vincent et al. 2003]. Mice in which the IRS-2 has been genetically ablated (IRS-2^−/−) and become insulin resistant and T2D after about 10 weeks of age demonstrated impaired endothelium-dependent, but not endothelium-independent, vasodilatation and are prone to injury-induced atherosclerosis [Kubota et al. 2003]. Furthermore, insulin-mediated eNOS phosphorylation is deficient in endothelial cells from IRS-2^−/− mice and insulin-induced capillary recruitment, insulin delivery and glucose uptake by skeletal muscle are all reduced [Kubota et al. 2011]. Interestingly, this vasodilator action of insulin is also blunted in obesity as well as in the presence of insulin resistance [Laakso et al. 1990; Del Turco et al. 2007]. Insulin resistance is also associated with the reduction in the availability of endothelium-derived NO and with an increase in the synthesis of the highly potent and endothelium-derived vasoconstrictor peptide, endothelin 1 [Boulanger and Lüscher, 1990; Irving et al. 2001]. Potenza and colleagues demonstrated the link between insulin resistance, activation of the Ras/MAPK pathway and impairment of the PI-3-kinase-dependent signalling, resulting in an imbalance in the production of the potent vasoconstrictor, endothelin, versus the generation of the vasodilator NO [Potenza et al. 2005]. Insulin signalling in the heart is also altered in insulin resistance with hyperinsulinaemia and diabetes and is associated with cardiac hypertrophy, possibly indirectly via the angiotensin system, angiotensin II expression and activation of the MEK-ERK1/2 pathway [Samuelsson et al. 2006; Bertrand et al. 2008].

Endothelial dysfunction, as reflected by a reduction in endothelium-dependent vasodilatation (EDV), is an early indicator of the development of cardiovascular disease and is closely associated with hyperglycaemia and the development of microvascular disease in both T1D and T2D. Endothelial dysfunction contributes to the reduction in blood flow and capillary recruitment and, in the presence of insulin resistance, both insulin and glucose delivery to tissues are exacerbated [Clark et al. 2003; Kim et al. 2006]. In addition, obesity may, independent of insulin resistance, contribute to the development of endothelial dysfunction as has been reported from studies of the effects of direct measurements of leg blood flow induced by the injection of the endothelium-dependent vasodilator, methacholine, into the femoral artery [Steinberg et al. 1996]. Of significance was that the response to sodium nitroprusside, an endothelium-independent vasodilator, was normal, supporting the argument that endothelial dysfunction precedes vascular smooth muscle dysfunction [Steinberg et al. 1996]. In the study by Steinberg and colleagues EDV was reduced by about 40–50% in patients with obesity compared with lean patients regardless of whether they were diabetic. Of additional importance was the finding that although low physiologic levels of insulin enhanced blow flow in the control lean group, higher levels of insulin were ineffective in the group with obesity (×2 insulin dose) and diabetic (×30) [Steinberg et al. 1996].

Under normoglycaemic conditions endothelial cells are predominantly glycolytic [Dobrina and Rossi, 1983]; however, during periods of hyperglycaemia the increase in intracellular glucose results in mitochondria generating excessive amounts of superoxide with the result that glyceraldehyde-3-phosphate dehydrogenase is inhibited and glucose is shunted into other pathways: the polyol and the hexosamine pathways, increased formation of advanced glycation end products (AGEs) and their receptors (RAGEs), and activation of protein kinase C isoforms [Du et al. 2000; Giacco and Brownlee, 2010]. Oxidative stress is also responsible for the uncoupling of eNOS from the dimeric NO-generating enzyme to the monomeric superoxide generating form [Aljofan and Ding, 2010; Triggle et al. 2012]. A rapid increase in blood levels of glucose, as triggered by a 75 g oral glucose load, is sufficient to reduce blood flow, as determined by flow-mediated vasodilatation, even in subjects without diabetes, but the impact is greater in those with impaired glucose tolerance or in those with diabetes [Kawano et al. 1999]. In addition, it has been reported that the effects of an oral glucose load on blood flow reduction can be prevented by the active form of the key cofactor for eNOS, 6R-tetrahydrobiopterin (BH₄), but not the inactive 6S-BH₄ [Ihlemann et al. 2003]. The reduction in the bioavailability of NO has profound negative effects on vascular health, resulting in the development of atherogenesis [Li and Förstermann, 2013]. Furthermore, frequent fluctuations, or transients, in plasma glucose levels may have a greater long-term effect on endothelial function and the development of vascular disease than sustained hyperglycaemia [Mah and Bruno, 2012]. Glucose transients have also been linked to epigenetic changes and the development of hyperglycaemic memory that has been linked to changes in promoter function of, for instance, nuclear factor κB [El-Osta et al. 2008]. Hyperglycaemia, possibly linked to vascular dysfunction, has also been associated with impaired memory and may therefore contribute to the development of dementia [Kerti et al. 2013]. Elevated glucose and its subsequent metabolism and the generation of superoxide inhibits vascular potassium channels and thus promotes vasoconstriction [Rainbow et al. 2006]. Acute hyperglycaemia, most likely via mitochondria-derived superoxide, results in an increase in the release of inflammatory cytokines, such as the interleukins, IL-6, IL-8, IL-18, as well as TNFα, which have been linked to the development of insulin resistance and monocyte adhesion (IL-8) [Esposito et al. 2002; Marette, 2002; Srinivasan et al. 2003; Zhou et al. 2011]. In conclusion, in the treatment of diabetes the minimal goal of drug therapy should be to maintain normoglycaemia and avoid frequent fluctuations in blood glucose, thus protecting the cardiovascular system from glucose toxicity.

The risk of hypoglycaemia

As reflected by the results of the ACCORD study, hypoglycaemia is a potential risk in the treatment of T2D. Severe hypoglycaemia is most frequently associated with TID, but the incidence increases in T2D as the use of insulin therapy increases. Apart from insulin, hypoglycaemia is particularly associated with the use of sulfonylurea and meglitinide oral hypoglycaemic drugs, but not with metformin. Severe hypoglycaemia is also a very serious issue and can lead to an increased risk of dementia, cardiovascular events and death [Brunton, 2012].

Biomarkers for cardiovascular disease and diabetes

The availability of reliable and readily obtainable biomarkers, preferably a panel of biomarkers, for the assessment of the progression, or regression, of cardiovascular disease in patients with diabetes would be an obvious benefit. Advances have been made, but not to the extent that there is agreement [Herder et al. 2014]. Since the early stages of cardiovascular disease are associated with the development of changes in the endothelium, it is tempting to use measures of endothelial function as surrogate markers. Thus, measurements of flow-mediated dilatation, C-reactive protein, vascular cell adhesion molecules, the endogenous NOS inhibitor, asymmetric dimethylarginine (ADMA), von Willebrand factor, microRNAs and other putative biomarkers have all been discussed [Tousoulis et al. 2013]. Levels of adiponectin, an adipose tissue-derived hormone, are inversely correlated with inflammation, obesity, insulin resistance, diabetes and cardiovascular risk as well as mortality and thus represents another potentially useful biomarker of metabolic disease burden [Mather and Goldberg, 2014]

Diabetes drug therapy and cardiovascular complications

The use of drugs for the treatment of diabetes may itself increase the risk from cardiovascular complications as a result of direct side effects related to the drug’s molecular actions or indirect effects that result from changes in the metabolic profile of the patient. For the latter, for instance, the risk of hypoglycaemia is potentially serious. Such risks may not have been detected, or fully appreciated, during drug development, but were realized after meta-analysis of multiple clinical trial data. Thus, it is very important that health care workers weigh such risks versus established benefits when recommending the most appropriate therapeutic regimen. As a result of postmarketing data revealing cardiovascular problems that became evident with the thiazolidinedione (TZD) rosiglitazone, the US Food and Drug Administration (FDA), since 2008, has required clinical trial evidence attesting to the cardiovascular risk of new drugs.

A summary of the cardiovascular benefits as well as major side effects and cardiovascular risks of the drugs used for the treatment of diabetes are summarised in Table 1.

Table 1.

Drugs used for the treatment of diabetes: cardiovascular benefits and significant side effects.

Drug	Cardiovascular benefits	Significant side effects
Metformin	Indirect via maintaining glycaemic control without hypoglycaemia (does not promote insulin release); improved lipoprotein metabolism; direct/indirect protection of endothelial/vascular function; weight neutral/weight reduction	Lactic acidosis, very rareGI, minor
Sulfonylureas	Enhanced insulin release; reduction of hyperglycaemia	Hypoglycaemia
		Increased cardiovascular risk
Thiazolidinediones	Enhanced sensitivity to insulin, reduced hepatic gluconeogenesis, pleiotropic actions to reduce inflammation, fat redistribution to subcutaneous sites and reduction in lipotoxicity; reduction in BP	Weight gain, oedema, bone fractures, cardiovascular risk? Cancer?
Glitazars	Insufficient data regarding relative risk	Cardiovascular risk (drugs withdrawn)
Incretins	Promote insulin release, pleiotropic protective effects on pancreatic β cells, improved endothelial function	GI, nausea, pancreatitis, pancreatic cancer?
DPP-4 inhibitors	Enhance insulin release via GLP-1 thereby reduce blood glucose and glucagon, direct and indirect effects on endothelium	Headache, nausea, hypersensitivity,Unknown effects due to nonspecificity of DPP-4 inhibition.
α-Glucosidase inhibitors	Slow postprandial increases in blood glucose, weight loss, lower triglycerides	GI
SGLT2 inhibitors	Improved insulin sensitivity, reduced blood pressure	Polyuria, restrict use in renal insufficiency; lack of information as only recently introduced
Insulin	Improved glucose disposal, improved blood flow; anti-inflammatory action (?)	Hypoglycaemia; promotes cardiac hypertrophy, increase in all-cause mortality? Cancer?

BP, blood pressure; DPP-4, dipeptidyl peptidase 4; GI, gastrointestinal; GLP-1, glucagon-like peptide 1; SGLT2, sodium glucose cotransporter 2.

Drug classes and drugs currently in use for the treatment of T1D and T2D

Figure 1 summarizes the site of action, cardiovascular effects and significant side effects of the drugs used for the treatment of diabetes.

Figure 1.

Summary of important sites of action and side effects and cardiovascular actions of drugs used in the treatment of diabetes.

DPP-4, dipeptidyl peptidase 4; eNOS, endothelial nitric oxide synthase; GI, gastrointestinal; GLP-1, glucagon-like peptide 1; SGLT2, sodium glucose cotransporter 2; TZD, thiazolidinedione.

The biguanide, metformin

The biguanide, metformin (dimethylbiguanide), was first introduced for use in the treatment of T2D in the UK in the late 1950s and today metformin is considered to be the first-choice drug and the ‘gold standard’ for most people with T2D [Halimi, 2006; Bosi, 2009]. Metformin is also the most widely prescribed as well as one of only two oral hypoglycaemic agents listed on the World Health Organization (WHO) Model List of Essential Medicines [WHO, 2013]. For this reason, a particular focus is based on metformin in this review article.

It has been estimated that the worldwide annual number of people receiving prescriptions for metformin is in excess of 120 million [Bosi, 2009]. Metformin is a synthetic oral hypoglycaemic drug with a chemistry linked to the guanidines, such as isoamylene guanidine, found in French lilac, Galega officinalis (or Italian fitch also known as Goat’s rue), a remedy known since medieval times for the relief of the excessive urination associated with diabetes [Witters, 2001]. The benefits of metformin include not only its hypoglycaemic efficacy with minimal side effects other than gastrointestinal-related issues, but also its ability to improve insulin sensitivity, thus improving glucose utilization as well as lipid metabolism. In addition, the use of metformin, unlike a number of drugs used in the treatment of diabetes, is either weight neutral or associated with weight loss [Glueck et al. 2001].

Metformin has been the subject of a number of recent reviews but, despite being in clinical use for over 50 years, the precise cellular modes of action remain unclear [Kirpichnikov et al. 2002; Halimi, 2006; Viollet et al. 2012; Bosi, 2009; Rena et al. 2013; Rojas and Gomes, 2013; Pernicova and Korbonits, 2014].

Pharmacokinetic properties of metformin

An appreciation of the pharmacokinetic properties of metformin is important in order to better understand the cellular basis for the therapeutic effects of metformin. Metformin has a bioavailability of between 30% and 60%, a plasma half life of approximately 2 h and an elimination half life of between 4 and 8.7 h; is not measurably metabolized, it is not bound to plasma protein and it is excreted unchanged in the urine with elimination characteristics consistent with a three-compartment model [Pentikäinen et al. 1979; Tucker et al. 1981]. Although a strong base and highly ionized molecule at physiological pH, metformin has a volume of distribution, V_D, considerably greater than total body water, indicating that some accumulation into tissues must occur. The gastrointestinal tract serves as a major site of accumulation and this may explain the frequency of gastrointestinal side effects that are associated with the use of metformin [Wilcock and Bailey, 1994]. Erythrocytes may also serve as a potential deep compartment for metformin [Lalau and Lacroix, 2003]. Nonetheless, the very brief plasma and elimination half lives of metformin make it very unlikely that significant tissue accumulation of the drug will occur with a standard three times daily therapeutic regimen [Sirtori et al. 1978].

Based on the pharmacokinetic properties of metformin and assuming 50% bioavailability and a V_D of 100 liters, the peak concentration of approximately 10 µM metformin should theoretically be achieved following an oral dose of 500 mg. Frid and colleagues also indicate that, not surprisingly given that metformin is excreted unchanged, the glomerular filtration rate (GFR) will also influence serum metformin levels [Frid et al. 2010]. Tucker and colleagues report the peak plasma concentration determined in normal controls following an oral dose of 500 mg is approximately 15 µM and for an intravenous dose 80 µM [Tucker et al. 1981]. It is important to keep these peak concentrations in mind when critiquing the published literature describing the in vitro effects of metformin. The discrepancy between the blood levels obtained for intravenous versus oral can be explained by slow absorption from the gastrointestinal tract, which has been estimated at approximately 6 h to achieve total maximal absorption from one dose. Absorption is thus the rate-limiting step for determining the therapeutic efficacy of metformin. Metformin can be transported into cells via organic cation transporters (OCTs) also known as solute carriers, of which there are 22 members, as well as the plasma membrane monoamine transporter, also known as hENT4 [Nies et al. 2011]. The accumulation of metformin into tissue, notably the gastrointestinal tract enterocytes, erythrocytes, hepatic and endothelial cells, via OCTs may also account for the comparatively short plasma half life of metformin [Tucker et al. 1981]. OCTs, specifically the organic cation transporter, OCT1, abundant in the liver and in the intestine, and OCT2, which is abundant in the kidney, are involved in the rapid renal clearance of metformin via tubular secretion [Shu et al. 2008]. Polymorphisms in OCTs may also affect the absorption from the gastrointestinal tract, hepatic uptake as well as the renal clearance of metformin, and thus result in patient-to-patient variations in the therapeutic efficacy of metformin [Wang et al. 2002; Shu et al. 2008; Graham et al. 2011; Yue et al. 2013].

Beneficial effects of metformin on cardiovascular function

In addition to its hepatic-mediated hypoglycaemic actions in human subjects, the cardiovascular benefits of metformin are also demonstrated in the prospective, randomized, multicentre UKPDS study of patients with T2DM [UKPDS, 1998]. The reduced cardiovascular morbidity and mortality benefits of treatment of patients with T2D with metformin has been confirmed in other reports; for instance Johnson and colleagues [Johnson et al. 2005]. Metformin also has a positive effect on the lipid profile [Glueck et al. 2001; Eleftheriadou et al. 2008]. In addition, metformin also improves endothelial function via an action that seems to be unrelated to its hypoglycaemic actions and is the only antidiabetic drug with a proven therapeutic efficacy to reduce the cardiovascular complications of diabetes [Ajjan and Grant, 2006; de Jager et al. 2005; Johnson et al. 2005; Arunachalam et al. 2014]. Metformin has also been shown to improve endothelial function in patients with T2DM [Mather et al. 2001; Vitale et al. 2005].

Side effects associated with use of metformin

The side effects associated with metformin use are usually minor with gastrointestinal problems the most common and, as noted, gastrointestinal side effects may reflect the accumulation of the drug in the small intestine [Wilcock and Bailey, 1994]. An association with lactic acidosis, possibly linked to putative effects of metformin in mitochondria to inhibit Complex 1, limit use in patients with kidney and liver dysfunction, although the Cochrane reports indicate that the risk of developing lactic acidosis with metformin treatment is exceedingly low and less than that in nonmetformin-treated patients with diabetes: 4.3 versus 5.4 cases per 100,000 patient-years respectively [Halimi, 2006; Salpeter et al. 2010]. Frid and colleagues make the valid observation that at GFRs greater than 30 ml/min/1.73 m² serum metformin levels are unlikely to rise above 20 µM, whereas lactic acidosis has been associated with metformin concentrations greater than 200 µM [Frid et al. 2010].

The gastrointestinal side effects of metformin, although considered fairly minor, can also result in folate malabsorption problems and vitamin B deficiency. Such deficiencies can result in elevated homocysteine levels and an elevated risk of cardiovascular disease secondary to endothelial dysfunction. Elevated homocysteine is an established risk factor for cardiovascular disease [Tousoulis et al. 2012]. However, folate supplementation can offset increases in homocysteine levels and, furthermore, the effects of metformin, at least over a 16-week treatment period, have been reported to be quite modest [Wulffelé et al. 2003; Aghamohammadi et al. 2011]. Furthermore, the gastrointestinal side effects are dose related and usually more evident at initiation of therapy with metformin.

Cellular basis for the oral hypoglycaemic action of metformin

Metformin reduces hepatic gluconeogenesis, decreases glycogenolysis, facilitates glucose disposal into peripheral tissues, reduces fatty acid oxidation and also protects the cardiovascular system from the negative effects of hyperglycaemia [Kirpichnikov et al. 2002]. Metformin may also delay or decrease glucose absorption from the gastrointestinal tract; however, this has been disputed [Czyzyk et al. 1968; Cuber et al. 1994]. Despite over 50 years of research, the precise mode of action of metformin remains controversial and, in fact, there may be several sites of action that are determined by concentration/dose-dependent effects as well as tissue selectivity and the relative expression levels of drug transporters. It is thus very important, particularly with respect to analyzing data from ex vivo studies, to determine whether the concentrations of the drug that were studied can be related to therapeutic levels of metformin.

Metformin has been shown to inhibit gene expression of the rate-limiting enzymes for hepatic gluconeogenesis, namely enzyme phosphenolpyruvate carboxykinase and glucose-6-phosphatase [Minassian et al. 1998; Yuan et al. 2002]. Adenosine monophosphate (AMP)-activated protein kinase (AMPK) is the frequently cited cellular target for metformin with the primary effects mediated by AMPK being the inhibition of hepatic gluconeogenesis and lipogenesis, and also the enhancement of insulin-regulated GLUT-4-mediated glucose transport into striated muscle and adipocytes [Owen et al. 2000; Zhou et al. 2001; Yang and Holman, 2006; Viollet et al. 2012; Fullerton et al. 2013]. AMPK, a heterotrimeric enzyme comprising α subunit and regulatory β and γ subunits with at least 12 potential combinations, serves as a metabolic energy sensor and is activated when adenosine triphosphate (ATP) levels decrease and AMP rises as a result of, for instance, exercise or cellular stress, with subsequent allosteric activation as a result of AMP binding to the γ subunit. AMPK can also be activated by the upstream tumour suppressor kinase, liver kinase B1 (LKB1), as well as calcium/calmodulin-dependent protein kinase β [Hardie et al. 2012; Hardie, 2013].

It has also been reported that metformin inhibits mitochondria respiration via an action on complex 1 [El-Mir et al. 2000; Owen et al. 2000]. In the study by El-Mir and colleagues a concentration-dependent effect of metformin on mitochondria respiration in rat hepatocytes, with a half maximal inhibitory concentration of 1 mM, was reported, but only in intact and not permeabilized cells. Furthermore, the inhibitory action of metformin was observed in functionally isolated complex 1 [nicotinamide adenine dinucleotide (NADH), quinone oxidoreductase]. Collectively, these data led to the conclusion that metformin inhibited mitochondria respiration via a cell membrane receptor-mediated signalling pathway [El-Mir et al. 2000]. A direct mitochondrial action may also explain the ability of metformin to inhibit glucose-induced endothelial cell death via inhibiting the opening of permeability transition pore; in this regard the effects of metformin are comparable to those of the antioxidant N-acetyl-cysteine [Detaille et al. 2005]. The increase in the AMP:ATP ratio that would be predicted to result from this ‘mild inhibitory’ effect of metformin might also be the basis of the role of AMPK in mediating the cellular effects of metformin [Hardie, 2007; Hawley et al. 2010]. However, caution is required as the inhibitory effects of metformin require a minimal concentration of 100 µM and, in some protocols up to 10 mM metformin was used [Owen et al. 2000]. In addition, the ability of metformin to bind to mitochondria is rather poor and only about 2% of that reported for another biguanide, phenformin, which was withdrawn from clinical use due to a high incidence of lactic acidosis [Schäfer, 1980]. Furthermore, Wollen and Bailey reported that whereas 10 µM metformin, together with 10 nM insulin, did not lower mitochondrial ATP in rat hepatocytes, evidence of decreased oxidative phosphorylation (an increased NADH/NAD⁺) was observed with 10 mM metformin [Wollen and Bailey, 1988]. The data from Wilcock and colleagues detailing the distribution of [14C]metformin in mice also indicated that 78% of the accumulation of [14C]metformin in the liver was associated with the cytosolic fraction and only 7–10% with mitochondria [Wilcock et al. 1991]. These data are clearly not supportive of a hypothesis that the therapeutic effects of metformin can be explained by a mitochondrial action that requires high micromolar concentrations unless, of course, metformin can accumulate to appropriately high concentrations in mitochondria. However, the latter has not been reported in the literature.

The cellular basis for the cardiovascular protective effects of metformin

As already noted, the cellular basis for the therapeutic hypoglycaemic actions of metformin has been attributed to the enhancement of AMPK-mediated suppression of hepatic gluconeogenesis and enhancement of insulin-regulated GLUT-4-mediated glucose transport into striated muscle and adipocytes [Zhou et al. 2001]. Improving glycaemic control would reduce the impact of hyperglycaemia on endothelial dysfunction and the development of microvascular disease.

Activation of AMPK might also be a contributing factor to the beneficial effects of metformin on the endothelium as AMPK also phosphorylates and activates eNOS [Chen et al. 2009]. However, Isoda and colleagues reported that phosphorylation of AMPK by metformin in human saphenous vein smooth muscle and endothelial cells required concentrations greater than 200 µM [Isoda et al. 2006]. Furthermore, in cell-free assays, AMPK is activated by the AMP analogue, 5-aminoimidazole-4-carboxamide riboside, a known activator of AMPK, but is not activated by metformin, thus suggesting that the effects of metformin on AMPK are indirect [Hawley et al. 2002, 2003]. Another apparent paradox is that although the kinase upstream of AMPK, LKB1, seems to be required for the activation of AMPK, it is not itself activated by biguanides such as phenformin, which is a close relative of metformin [Lizcano et al. 2004; Sakamoto et al. 2004]. Another question concerning the role of AMPK also arises because mice that are genetically deficient in AMPKα1 and also carry a liver-specific deletion of AMPKα2 still demonstrate a hypoglycaemic response to metformin [Foretz et al. 2010]. Thus, it seems unlikely that either AMPK or LKB1 are directly activated by metformin, at least in the therapeutic concentration range. Arunachalaum and colleagues have reported that metformin, via an insulin-independent mechanism that requires the ‘longevity gene’, SIRT1, protects microvascular endothelial cells against hyperglycaemia-induced oxidative stress and promotion of senescence, apoptosis and dysfunction [Arunachalaum et al. 2014]. Of interest is that high doses of metformin, with serum concentrations of approximately 500 μM that provide anti-inflammatory effects and protection against oxidative stress, have been reported to extend the lifespan and improve health parameters, ‘healthspan’, of middle-age mice by approximately 6% [Martin-Montalvo et al. 2013].

Another cell signalling pathway that might also be involved in mediating the cellular effects of metformin is through changes in cyclic AMP as an increase in AMP will inhibit adenylate cyclase and therefore suppress glucagon-mediated gluconeogenesis in the liver [Miller et al. 2013].

In summary, further studies of the cellular sites of action of metformin are required, particularly to determine the basis for its cardiovascular protective action, as although activation of AMPK may play a role in mediating the effects of metformin it is unlikely to be the direct target, or at least, not the sole target and other targets should be considered [Hardie, 2013].

The sulfonyureas

The discovery of sulfonylureas as hypoglycaemic drugs came about by serendipity as result of the use of a new sulphonamide antibiotic, p-amino-benzene-sulfamido-isopropyl-thiodiazole (2254 RP), by Dr Marcel Janbon at the Clinic of the Montpellier Medical School during an outbreak of typhoid in 1942. One of the adverse reactions was a large drop in blood glucose. This observation was pursued by a physiologist, Dr Auguste-Louis Loubatières, and ultimately resulted in the development of the sulfonylurea oral hypoglycaemic drugs for T2D [Loubatières-Mariani, 2007]. Carbutamide, shortly followed by tolbutamide, was the first analogue to be used for its hypoglycaemic actions with apparent early controversy as to who first described its synthesis [Kleinsorge, 1998].

The sulfonylureas inhibit the opening of K_ATP channels resulting in the depolarization of pancreatic β cells, a subsequent increase in calcium influx and intracellular free calcium levels and, thus, the promotion of the release of insulin. K_ATP channels are ubiquitous and are expressed in mitochondria (mitoK_ATP) as well as the plasma membrane (sarcoK_ATP). K_ATP channels also play an important function in protecting against myocardial ischemia, ‘ischaemic preconditioning’, and hence there are concerns over an increase in cardiovascular risk with the use of this class of oral hypoglycaemic drug.

The sulfonylureas inhibit the K_ATP channel by binding to the sulfonylurea receptor, SUR, component of the channel that, in turn, regulates the associated K_ir6.x (either K_ir6.1 or K_ir6.2) potassium channel. The K_ATP channel is a hetero-octomeric complex composed of two types of subunits: four inward-rectifying potassium channel pore-forming (Kir6.x) subunits and four high-affinity SUR1 subunits. The SUR protein is a member of the ATP-binding cassette (ABC) gene family homologues. The SUR protein senses the intracellular ATP:ADP ratio and hence the metabolic status of the cell that reflects glucose metabolism; higher ATP closes the channel and enhances insulin release. There are three isoforms of SUR: SUR1, encoded by the ABCC8 gene, and SUR2A and SUR2B, which are splice variants of the ABCC9 gene. The β-cell K_ATP channel is composed of four Kir6.2 subunits and four SUR1 subunits. Cardiac K_ATP channels are also composed of four Kir6.2, but the SUR protein is represented by four SUR2A subunits. Vascular K_ATP channels, which are also regulated by cell metabolism, are composed of K_ir6.1 and SUR2B. Thus, ideally an oral hypoglycaemic sulfonylurea drug should selectively target only the SUR1 protein and the pancreatic β cell K_ATP channel. Although absolute selectivity does not exist, glibenclamide does show relatively good selectivity, but can inhibit cardiac K_ATP channels within the anticipated therapeutic concentration range; selectivity for repaglinide, a non-sulonylurea KATP channel inhibitor and so-called meglitinide, is less [Nagashima et al. 2004; Stephan et al. 2006].

Cardiovascular safety of the sulfonylureas

Serious concerns have been expressed about the cardiovascular safety of the short-acting sulfonylurea, tolbutamide, and the following statement made: ‘With no evidence of efficacy and a definite possibility of toxicity, the investigators concluded that the safety of the patients still receiving tolbutamide therapy required its discontinuance and that the factual basis for this decision needed to be communicated to the biomedical scientific community’ [Cornfield, 1971: 1676]. See also The University Group Diabetes Program (UGDP, 1970)].

In 1971 the existence of the K_ATP channel was not known, nor, of course, the role of the K_ATP channel in preconditioning. Concerns regarding the link between hypoglycaemia and cardiovascular risk were also not apparent until the report of the UKPDS in 1998. Recurrent hypoglycaemia is another major morbidity that is linked to sulfonylureas and the risk with glibenclamide is higher than with other sulfonylureas [Gangji et al. 2007]. A large retrospective analysis of over 27,000 patients treated with sulfonylureas versus metformin has also stressed an increase in cardiovascular risk for those treated with sulfonylureas versus metformin alone [Pantalone et al. 2012]. The addition of a sulfonylurea to a drug regimen initially based on monotherapy with metformin is common practice. A MEDLINE database analysis for the period January 1966 through July 2007 with data reviewed from almost 300 sources for the combination therapy metformin plus sulfonylurea significantly increased the relative risk (RR) of the composite endpoint of cardiovascular hospitalization or mortality (RR 1.29) versus control group therapy with diet, or monotherapy with metformin or a sulfonylurea [Rao et al. 2008]. It should be noted that the combination of a sulfonylurea with metformin has been the accepted second step in the management of patients with T2D and such patients may therefore have a more severe form of T2D. The recent availability of agents such as the incretins and gliptins is changing prescribing practice [Cakirca et al. 2014]

The cardiovascular safety of sulfonylureas continues to be debated, particularly with respect to monotherapy with sulfonylureas versus other drugs [Hemmingsen et al. 2013; Monami et al. 2013].

As for metformin, retrospective analysis of clinical data suggests that treatment with at least some sulfonylureas, namely gliciazide and tolbutamide, may reduce mortality associated with cancer [Bo et al. 2013]. The putative protective actions of drugs used for the treatment of diabetes against cancer will not be further discussed in this review.

The meglitinides (glinides)

The meglitinides, first introduced in 1997 with repaglinide and followed by nateglinide and mitaglinides, are oral hypoglycaemic agents that inhibit K_ATP channels and thus serve as insulin secretalogues. They share comparable side effects, notably hypoglycaemia and weight gain, with the sulfonylurea drugs. They differ from the sulfonylureas in having a faster onset and shorter duration of action, thus making them suitable for postprandial control of blood glucose.

The TZDs or glitazones

The glitazones are activators of the nuclear peroxisome proliferator-activated receptor γ (PPARγ). PPARγ is expressed in adipose tissue, pancreatic β cells, muscle, liver, endothelial cells and also macrophages. Activation of PPARγ results in increased GLUT4 expression, enhanced sensitivity to insulin, reduced hepatic gluconeogenesis and therefore reduce blood glucose and HbA_1c levels. However, the glitazones do not enhance insulin secretion. The glitazones also reduce lipotoxicity and redistribute fat from muscle, liver and pancreatic β cells to subcutaneous depots. Glitazones also enhance the expression of PPARγ cofactor 1, an important regulator of mitochondria biogenesis, and thus modulate energy homeostasis [Puigserver and Spiegelman, 2003]. In addition, likely secondary to their effects on improved insulin sensitivity and glucose homeostasis, the glitazones also reduce blood pressure and microalbuminuria [de Rivas et al. 2005; Defronzo et al. 2013a]. Thus, the beneficial effects of glitazones can be linked to their effects on insulin sensitivity and the reduction of lipotoxicity and glucotoxicity with improved β cell function. The ACT NOW Study (Actos Now for the prevention of diabetes) was designed to determine if pioglitazone can prevent/delay progression to diabetes in high-risk impaired glucose tolerance subjects and also to define the mechanisms that mediate the beneficial effects. The study concluded that the primary mechanism whereby pioglitazone prevents diabetes and improves glucose tolerance is via the preservation and improvement of pancreatic β cell function [Defronzo et al. 2013b]. Rosiglitazone and pioglitazone are the two glitazones currently remaining in therapeutic use. They are orally effective and have half lives from 3 h up to 24 h respectively, with pioglitazone having active metabolites.

Beneficial effects of glitazones on cardiovascular function

Data from the PROactive Study (PROspective pioglitAzone Clinical Trial In macroVascular Events) indicate that patients with diabetes who are at high cardiovascular risk and are treated with pioglitazone show improvement in insulin sensitivity, a reduction in the need for insulin, and improved cardiovascular outcome [Dormandy et al. 2005]. The beneficial effects of the glitazones, notably pioglitazone, on cardiovascular function have also been linked to their pleiotropic effects, such as the reduction of inflammatory markers and, on the basis of apparent differing effects on cardiovascular outcomes, pioglitazone is the preferred choice over rosiglitazone, particularly when low doses are administered [Defronzo et al. 2013a; Zou and Hu, 2013]. Glitazones, like metformin, do not promote insulin release and have been described as ‘insulin secretion decelerators’ and, together with their actions to reduce energy production, as already discussed for metformin, may be the basis for their beneficial effects in the treatment of T2D [Lamontagne et al. 2013].

Cardiovascular and other safety issues with the glitazones

Troglitazone was withdrawn from clinical use in the UK in 1997 due to hepatotoxicity and elsewhere shortly after. However, studies with rosiglitazone and pioglitazone indicate that hepatotoxicity is not a class effect [Lebovitz, 2002]. Despite the positive effects of the glitazones to lower plasma triglycerides and raise high-density lipoprotein, low-density lipoprotein levels are also raised. Furthermore, glitazones reduce leptin levels and thus promote weight gain in the form of redistribution of fat to subcutaneous depots. The glitazones also have negative effects on cardiac function due to oedema, which is seen in about 5% of patients and therefore may precipitate heart failure. Therefore, glitazones are contraindicated for use in patients with New York Heart Association class III and IV heart failure. There is a greater concern with rosiglitazone than with pioglitazone and a retrospective analysis of data from more than 225,000 patients [Graham et al. 2010] concluded that the use of rosiglitazone was associated with an increased risk of stroke, heart failure and all-cause mortality, as well as an increased risk of the combination of acute myocardial infarction, stroke, heart failure or all-cause mortality, in patients aged 65 years or older. On the basis that the risks outweigh the benefits, rosiglitazone use was restricted in the USA to patients with T2D who were not effectively treated with other medications. Rosiglitazone was suspended in the European market and in some other countries such as New Zealand [Bourg and Phillips, 2012]. In late 2013 the FDA lifted this restriction on the basis of the results from the RECORD clinical trial (Rosiglitazone Evaluated for Cardiovascular Outcomes and Regulation of Glycemia in Diabetes), which failed to reproduce the results from the 2007 meta-analysis and indicated no elevated risk of heart attack or death in patients being treated with rosiglitazone versus diabetes medications [Mahaffey et al. 2013]. Nonetheless vigilance is needed as concerns over cardiovascular safety issues with pioglitazone were raised as a result of the PROactive study of patients with diabetes, which indicated an 11% increased risk of congestive heart failure (CHF) [Dorkhan et al. 2009]. However, an earlier retrospective cohort analysis of more than 16,000 patients concluded that despite the increase in peripheral oedema the use of glitazones decreased mortality [Masoudi et al. 2005].

There are also differences in the safety profiles of the two glitazones in use with respect to the potential for drug interactions as the oxidative metabolism for each drug occurs by distinct cytochrome pathways: pioglitazone involves cytochrome P450 (CYP) 3A4 and CYP 2C8, whereas rosiglitazone is principally metabolized by CYP 2C8. Since CYP 3A4 is involved in the metabolism of well over 150 drugs, the potential for drug interactions with pioglitazone is far greater than with rosiglitazone. Furthermore, pioglitazone use has been associated with an increased risk of bladder cancer and this has resulted in 2011 in the withdrawal from use in France and restricted use elsewhere (Germany) or appropriate warnings (FDA). Tumours of the bladder were noted in the PROactive study [Dormandy et al. 2005], but at that time were not considered to be a result of treatment with pioglitazone. A number of more recent reports, including a large retrospective cohort study by Mamatami and colleagues [Mamtani et al. 2012], support a link, particularly with long-term treatment of more than 5 years. To further add to the risk benefits for the use of glitazones is their association with doubling the risk of bone fractures, particularly hip and wrist, notably in postmenopausal women with diabetes who are treated with either pioglitazone or rosiglitazone, compared with other oral glycaemic agents [Meier et al. 2008; Betteridge, 2011]. The risk of bone fractures is most likely a class effect and related to PPARγ regulation to reduce osteoblast activity and bone formation.

Glitazars

Glitizars, a combination of a PPARα agonist and a PPARγ agonist, have a pharmacological profile similar to combining a fibrate with a glitazone. Two agents, muraglitazar and tesaglitazar, were introduced into clinical trials, but discontinued in 2006 due to elevated risk of CHF and reduced renal function.

The incretins: glucagon-like peptide 1

The development of exenatide, approved in 2005, and liraglutide, approved in 2009, and other analogues more recently, was based on the discovery of two gut hormones that were found to have an ‘incretin effect’. The ‘incretin effect’ defined as the ability to elicit a higher insulin release to an oral glucose load (as, for instance, would be administered as a test for impaired glucose tolerance) compared with that resulting from an equivalent intravenous glucose load. Glucose-dependent insulinotropic polypeptide (GIP) is secreted from the K cells in the duodenum and jejunum, and glucagon-like peptide 1 (GLP-1) is secreted from the L cells of the distal ileum and colon. It was the pioneering work of the Canadian scientist, John Brown, and his team that first discovered and described GIP that stimulated further research into the role of incretins in the regulation of glucose homeostasis [Brown and Dryburgh, 1971; Kim and Egan, 2008]. Not only do the incretins promote insulin release via the activation of their specific G-protein coupled receptors (GPCRs) to enhance cyclic AMP in the β cell and, in turn, enhance insulin release, but activation of the GLP-1 GPCR also results in the inhibition of glucagon release most likely via paracrine actions of insulin [Komatsu et al. 1989; Kim and Egan, 2008]. In T2D the insulinotropic actions of GIP are deficient and thus only GLP-1 has been pursued as a pharmacological target [Kim and Egan, 2008].

Cardiovascular benefits of GLP-1 mimetics

A significant advantage of the incretin analogues relates to their pleiotropic actions on the pancreas where they help maintain β cell function and the biphasic release of insulin. In 2005 exenatide, the first GLP-1 mimetic approved for the treatment of T2D, is a synthetic version of exendin 4 isolated from the saliva of the Gila monster lizard. Exenatide and liraglutide slow the emptying of the stomach and decrease body weight, but also may have a positive inotropic action. GLP-1 has been shown to have a beneficial effect on endothelial function that is negated by cotreatment with the sulfonylurea, glyburide (USAN) (glibenclamide, INN) [Basu et al. 2007]. GLP-1 infusion improved left ventricular function in patients with severe systolic dysfunction [Nikolaidis et al. 2004]. Since GLP-1 receptors are also found on vascular cells, monocytes and macrophages it has been suggested that the protective effects of GLP-1 analogues and dipeptidyl peptidase 4 (DPP-4) inhibitors on cardiovascular function involves direct actions to reduce oxidative stress, vascular inflammation and improve endothelial function [Nikolaidis et al. 2004; Oeseburg et al. 2010; Mita and Watada, 2012]. The ability of GLP-1 to directly affect vascular function may, however, be vessel and species dependent. The rat aorta responds to GLP-1 with a dose-dependent relaxation at a threshold of 10 picomolar via an endothelium-independent and NOS inhibitor (L-NAME), cyclooxygenase (indomethacin) and hydrogen peroxide (catalase)-insensitive mechanism involving an elevation of cyclic AMP and the opening of K_ATP channels [Green et al. 2008]. The response to GLP-1 receptor activation and the resultant vasodilatation was sensitive to the GLP-1 receptor antagonist, exendin(9-39) [Green et al. 2008]. Although, rat pulmonary and femoral arteries also relax in response to GLP-1, the maximum vasorelaxation response is no more than 30% and is endothelium dependent in the rat pulmonary artery, but endothelium independent and L-NAME insensitive in the rat femoral artery [Golpon et al. 2001; Nyström et al. 2005].

GLP-1 has a very short half life of less than 30 min, but synthetic derivatives have longer half lives. Nonetheless a disadvantage in the use of GLP-1 peptide analogues is that they require frequent subcutaneous injection. Exenatide has a relatively short half life of approximately 3 h, but longer for liraglutide, which has a half life of 11–15 h. The requirement for subcutaneous injection raises the issue of patient compliance; however, a once weekly injection formulation has been developed and was approved in 2012. An orally effective formulation of liraglutide, NN9924, has also been developed and is currently undergoing clinical evaluation. Concerns have recently been raised over a link between the use of incretins and pancreatitis; however, based on a joint evaluation by the FDA and their European equivalent, the European Medicines Agency (EMA), the data were not conclusive, but the need is recognized for an ongoing evaluation [Egan et al. 2014].

DPP-4 inhibitors

The FDA approved the first DPP-4 inhibitor, sitagliptin, in 2006. The primary therapeutic benefit is assumed to be via the increase in incretin levels (the glucagon-like peptides, GLP and GLP-1) subsequent to the inhibition of DPP-4. DPP-4, also known as CD26, is a 766-amino-acid cell surface membrane-associated, serine-type protease enzyme which is widely, if not ubiquitously, distributed in numerous tissues, including adipocytes, endothelial cells, kidney, intestine, liver, spleen, placenta, adrenal glands and lymphocytes [Abbott et al. 1994]. DPP-4 is a multifunctional protein and thus inhibition of DPP-4 may have many diverse effects [Boonacker and Van Noorden, 2003]. DPP-4 expression is increased in adipocytes from subjects with obesity and the release of DPP-4 is correlated with adipocyte size and, furthermore, DPP-4 has been shown to impair insulin signalling [Lamers et al. 2011].

Potential cardiovascular benefits of DPP-4 inhibitors

DPP-4 is not specific for GLP-1 and also plays a role in the degradation of other peptides wherein the penultimate amino acid is either a proline or alanine. Peptides that are substrates for DPP-4 include a number of chemokines and the neuropeptide and endothelium-dependent vasodilator, substance P [Mentlein and Struckhoff, 1989]. Thus, DPP-4 inhibitors, via the inhibition of the breakdown of vasodilator neuropeptides, may play an important role in enhancing blood flow and reducing the negative impact of hyperglycaemia on vascular function. In addition, DPP-4 inhibition also has been shown both in vitro and in vivo to enhance endothelial cell growth, thus suggesting that DPP-4 inhibitors, in addition to their action to enhance the effects of GLP-1, may also have direct actions to reverse the deleterious effects of diabetes on vascular function [Takasawa et al. 2010]. Van Poppel and colleagues reported that 4 weeks of treatment with the DPP-4 inhibitor, vildagliptin, improved acetylcholine-mediated EDV [Van Poppel et al. 2011]. Matsubara and colleagues reported that the DPP-4 inhibitor, des-fluoro-sitagliptin, improved acetylcholine-mediated EDV and reduced atherosclerotic lesion development in apolipoprotein E deficient mice via the augmentation of GLP-1-mediated effects in endothelial cells and macrophages [Matsubara et al. 2012]. DPP-4, a serine proteinase, will also activate proteinase-activated receptors (PARs) via the cleavage of the extracellular amino terminus of the PAR, thus exposing the amino-terminal ‘tethered ligand’ domain that binds to and activates the cleaved PAR. PAR1, 2, 3 and 4 are targets for thrombin, trypsin, matrix metalloproteinases and other proteinases, and are involved in a multitude of physiological functions. They have also been linked to inflammation and pathophysiological cellular processes [Ramachandran and Holleneberg, 2008; Gieseler et al. 2013]. Activation of PAR2 in endothelial cells promotes EDV via NO and non-NO-mediated cell signalling [McGuire et al. 2002]. Thus, inhibition of DPP-4 could, indirectly, impair EDV. The acute exposure of endothelial cells in culture to alogliptin enhances eNOS and Akt phosphorylation (Ser¹¹⁷⁷ and Thr⁴⁷³ respectively) [Shah et al. 2011]. Thus, although stated as ‘cardiovascular neutral’, there are reports of favourable effects in animal models of ischaemia and reperfusion, and they may also modulate blood pressure. Thus, pharmacological inhibition and also genetic ablation of DPP-4 are cardiovascular protective in ischaemia/reperfusion protocols and post myocardial infarction in mice [Ku et al. 2011; Sauvé et al. 2010]. In part, the beneficial effects of DPP-4 inhibitors may involve the inhibition of AGE-mediated increase in oxidative stress via the direct inhibition of NADPH-oxidase or xanthine oxidase [Kröller-Schön et al. 2012; Ishibashi et al. 2013]; however, this may not be a class effect, but limited to linagliptin [Kröller-Schön et al. 2012]. Based on studies with Zucker diabetic fatty rat and human umbilical vein endothelial cells (HUVECs) it has been argued that in vivo DPP-4 inhibition will enhance GLP-1-mediated increases in cellular cyclic AMP levels and, via cyclic AMP response element-binding protein transcription of antioxidant genes, such as HO-1 [Oesburg et al. 2010]. An anti-inflammatory action of DPP-4 inhibitors can also be linked to a reduction in the production of pro-inflammatory cytokines [Reinhold et al. 1993].

In conclusion, an accumulation of data indicates that inhibition of DPP-4, in addition to enhancing the effects of both endogenously released GLP-1 and exogenously administered GLP-1, has actions unrelated to inhibition of the enzymatic breakdown of GLP-1 [Fadini and Avogaro, 2011; Zhong et al. 2013]. However, negative effects on endothelium function have also been noted. In addition to the potential reduction of PAR2-mediated vasodilatation, other negative effects have been described. Thus, Ayaori et al. [2013] reported that a 6-week treatment of men with T2D with either sitagliptin or voglibose resulted in an attenuation of flow-mediated vasodilatation in the brachial artery. Furthermore, DPP-4 inhibits the polymerization of fibrin monomers and thus inhibition of DPP-4 may be prothrombotic [Mentlein and Heymann, 1982]. The antithrombotic properties of DPP-4 and the fact that the protease serves as an immobilized anticoagulant on endothelial cells have also been emphasized in a study by Krijnen and colleagues. They reported that the expression and activity of DPP-4 are decreased following myocardial infarction in humans and this results in a prothrombogenic phenotype [Krijnen et al. 2012]. A number of clinical trials are underway with DPP-4 inhibitors and the results should shed further light on their effects on cardiovascular function and cardiovascular risk factors [Scheen 2013a, 2013b].

The DPP-4 inhibitor, vildagliptin, as an add-on therapy with metformin administered to a group of 68 patients with T2D has been shown to result in a favourable cardiovascular response as measured by a reduction in serum ADMA levels. However, no difference was noted in levels of C-reactive protein or fibrinogen [Cakirca et al. 2014].

α-Glucosidase inhibitors

Acarbose was the first such inhibitor approved by the FDA in 1995 and miglitol in 1996. Acarbose and related drugs are orally effective and should be taken together with meals. Predictably the most frequent side effects are gastrointestinal related, such as flatulence and diarrhoea; however, hypoglycaemia can occur and requires treatment with a monosaccharide such as glucose. There have been rare cases of liver toxicity reported following several months of therapy with acarbose. The primary action of these compounds is to inhibit the α-glucosidase enzymes (maltase, isomaltase, sucrase, glucoamylase) that hydrolyze oligosaccharides and disaccharides to glucose. These enzymes are associated with the small intestine and thus, when inhibited, there is a reduction of glucose availability and absorption.

Cardiovascular benefits of α-glucosidase inhibitors

By slowing glucose absorption and reducing plasma glucose levels, particularly postprandial glucose, α-glucosidase inhibitors should reduce cardiovascular risk in patients with diabetes and this has been shown for the inhibitor, acarbose, with additional benefits in terms of weight loss, blood pressure reduction and lowering of triglycerides [Hanefeld and Schaper, 2008; Hanefeld et al. 2008].

Sodium glucose cotransporter 2 inhibitors

Sodium glucose cotransporter 2 (SGLT2) inhibitors are the newest group of drugs to be used for the treatment of diabetes and, by binding to the SGLT2 in the proximal tubule of the kidney, block glucose reabsorption. The development of SGLT2 inhibitors was based on studies with the flavinoid, phlorizen, that occurs in several fruit trees, including apple and cherry, and the bark of the pear tree. Phlorizen is an inhibitor of the widely expressed SGLT1 as well as SGLT2. Phlorizen has been shown to correct hyperglycaemia and improve insulin sensitivity in an animal model of diabetes [Rossetti et al. 1987]. Phlorizen has not been developed for clinical use, presumably because of assumed toxicity and nonspecificity of action on the SGLT transporter family [Ehrenkranz et al. 2005]. The advantage of SGLT2 inhibitors is that expression of SGLT2 is highly specific for the kidney and thus, theoretically, should not affect glucose transport in other tissues [Chen et al. 2010]. Canagliflozin was approved in the USA in 2013 and dapagliflozin was approved in Europe in 2012 and in the USA in early 2014. Other ‘flozins’ are under development. The use of SGLT2 inhibitors is associated with improved insulin sensitivity, reduction in blood pressure and weight loss; however, inhibiting SGLT2 can, potentially, result in dehydration and, it has also been reported, enhance endogenous glucose production; the latter, presumably, as a compensatory effect to the excretion of glucose in the urine [Merovci et al. 2014].

Cardiovascular impact of SGLT2 inhibitors

Correction of hyperglycaemia, weight loss and reduction in blood pressure suggest a positive impact on cardiovascular health; however, effects on other cardiovascular risk factors such as lipid profile are less clear and insufficient data are available to make a conclusion at this time [Guthrie, 2013]. The effectiveness of the flozins depends on adequate renal function and since chronic kidney disease is a feature of the advanced stages of T2D, caution in their use in patients with renal impairment is required.

Insulin

Treatment with insulin is essential for patients with T1D and, with the progression of β-cell dysfunction in T2D, is ultimately required for many subjects with T2D. Insulin facilitates the storage and synthesis of carbohydrates, lipids and protein. From a physiological perspective and the regulation of glycaemia, insulin functions as a physiological antagonist of glucagon; however, insulin has multiple effects in the body, including effects on gene expression. Today there are a variety of insulin preparations available from the ultra-rapid-acting lispro, with onset within 15 min and duration of up to 5 h, to ultra-long-acting degludec (approved in Europe but not yet by FDA) with a duration of action of 40 h. The major side effects of insulin use are hypoglycaemia and weight gain. As reflected by the results of the ACCORD study, hypoglycaemia is a potential risk in the treatment of T2D. Severe hypoglycaemia is most frequently associated with T1D, but the incidence increases in T2D as the use of insulin therapy increases. Hypoglycaemia, as already discussed, is also associated with the use of sulfonylurea and meglitinide oral hypoglycaemic drugs, but not with metformin. Severe hypoglycaemia is a very serious issue and can lead to an increased risk of dementia, cardiovascular events and death [Brunton, 2012].

An important question regarding insulin therapy in patients with T2D with insulin resistance is whether high levels of insulin, or the use of long-acting insulins, contributes to cardiovascular morbidity? In this regard it is of interest to note that the FDA has suspended approval of the ultra-long-acting insulin, degludec, subject to provision of cardiovascular safety data. Based on the data from the DCCT and UKPDS studies, we know that maintaining near normal HbA_1c decreases mortality in patients with T1D and T2D. The recently completed 6-year duration ORIGIN trial with the long-acting insulin, glargine, enrolled over 12,000 patients and was designed to maintain basal insulin and help normalize plasma glucose levels [ORIGIN et al. 2012]. The results demonstrated a neutral effect on cardiovascular outcomes and also cancers, a reduction in the incidence of diabetes, but a positive link to weight gain and hypoglycaemia. An association of insulin use with cancer has also been reported. In one study, a meta-analysis of over 40 studies indicated an increased risk for some forms of cancer (breast), but not others (colon) [Karlstad et al. 2013]. Others have suggested that the risk of cancer associated with exogenous insulin use increases in a cumulative fashion whereas monotherapy with metformin, compared with a sulfonylurea, lowered risk [Bowker et al. 2010]. The cardiovascular and neurological risk potential of hypoglycaemia resulting from insulin use are well recognized, but the longer-term effect of hyperinsulinaemia on cardiovascular function remains unclear. The proproliferative effect of insulin on vascular smooth cells in culture has been known for almost 45 years and a positive link between hyperinsulinaemia and lipid abnormalities, atherosclerosis and hypertension reported [Stout and Vallance-Owen, 1969; Stout et al. 1975; Stout 1990; Muis et al. 2005]. For instance, the results from a study of the effects of hyperinsulinaemia in the absence and presence of high concentrations of the free fatty acid, palmitate, on primary cell cultures of human coronary arteries demonstrated that palmitate inhibited the effects of insulin on Akt signalling, but promoted the mitogenic actions of insulin, thus suggesting that in a metabolic syndrome and diabetic milieu, hyperinsulinaemia would be proatherogenic [Cersosimo et al. 2012]. However, a retrospective study of 70 patients with insulinomas failed to link hyperinsulinaemia to lipid abnormalities and hypertension was only present in five of the patients who remained hypertensive after removal of the insulinoma [Leonetti et al. 1993]. In addition, Sawicki and colleagues failed to link hyperinsulinaemia to hypertension in patients with insulinomas and a meta-analysis of 12 published studies of patients with hyperinsulinaemia reported only a weak positive link to cardiovascular disease [Ruige et al. 1998]. Furthermore, there is a substantial body of data indicating that insulin has an anti-inflammatory action that might therefore offset proatherogenic effects [Hyan et al. 2011]. In its physiological concentration range, insulin is anti-inflammatory, but at supraphysiological concentrations insulin is pro-inflammatory [Krieger-Brauer et al. 1992, 1997]. However, intensive insulin therapy administered to patients in intensive care units has also been shown to normalize hyperglycaemia, lower IL-6 and raise levels of the antioxidant enzyme superoxide dismutase that dismutases superoxide anions into oxygen and hydrogen peroxide [Wiryana, 2009]. For the latter study it is, of course, important to recognize that the benefits of intensive insulin use that were described were in an acute setting and the effects of high insulin levels and insulin resistance in the chronic setting may have a different end result. Thus, it is argued that insulin resistance serves a protective role and guards insulin-sensitive tissues against glucose overload, glucose toxicity and, by enhancing glucose uptake in the presence of high lipid concentrations, glucolipotoxicity [Nolan et al. 2011, 2013]. Recent reports also raise concern, linking insulin use with an increase in mortality. Thus, in a retrospective analysis of over 84,000 patients with T2D, treatment with exogenous insulin was associated with an increased risk of diabetes-related complications and all-cause mortality, as well as cancer [Currie et al. 2013].

Summary

There is an increasing number of drugs available for the treatment of diabetes and therapeutic choices can be made on the basis of an expanding knowledge base of the cardiovascular benefits versus potential cardiovascular risks of the available drugs. Recognition that treatment should not only address blood glucose levels and HbA_1c is important for the reduction of the morbidity and mortality associated with diabetes.