Thiazide Diuretics and β-Blockers in the Treatment of Hypertension in Diabetes Mellitus
Thomas D. Giles, MD, Heart and Vascular Institute, Tulane University School of Medicine, 109 Holly Drive, New Orleans, LA 70005
Although advocated as “Step 1” in the early Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC) reports, thiazide diuretics and β-blockers have more recently received less enthusiasm compared with newer agents, and their role in the treatment of diabetic patients has been scrutinized due to the potential for adverse effects on glucose and lipid levels and the physiologic response to hypoglycemia. There are, in fact, limited data on the relative efficacy of these agents in reducing cardiovascular outcomes in patients with type 2 diabetes mellitus.1 Further difficulties arise from differing pharmacologic effects of the drugs within each class, with resultant clinically important differences in antihypertensive and metabolic profiles of individual drugs, thus rendering comparison with other classes of antihypertensive agents even more difficult. Just one example is the recent review suggesting that hydrochlorothiazide doses up to 25 mg daily are no different from placebo.2
In reviewing available data, several studies that compared different drug classes have reported diabetic subgroups separately. In the Swedish Trial in Old Patients With Hypertension-2 (STOP Hypertension-2), blood pressure (BP) was decreased similarly in all treatment groups.3 The primary combined end point of fatal stroke, fatal myocardial infarction, and other fatal cardiovascular disease occurred in 221 of 2213 patients in the conventional drug (atenolol metoprolol, pindolol, or hydrochlorothiazide/amiloride) group (19.8 events per 1000 patient-years) and in 438 of 4401 in the newer drug (enalapril, lisinopril, felodipine, or isradipine) group (19.8 per 1000 patient-years; relative risk (RR), 0.99 [95% confidence interval (CI), 0.84–1.16]; P=.89). The combined end point of fatal and nonfatal stroke, fatal and nonfatal myocardial infarction, and other cardiovascular mortality occurred in 460 patients taking conventional drugs and in 887 taking newer drugs (RR, 0.96; CI, 0.86–1.08; P=.49). There was a significant decrease in myocardial infarction in patients on angiotensin-converting enzyme (ACE) inhibitors compared with patients treated with the dihydropyridine calcium channel blockers (DCCBs), but not compared with β-blockers and diuretics. In the International Nifedipine Study(GITS), no difference was found for diabetic patients between therapy based on nifedipine relative to co-amilozide, a thiazide/potassium-sparing diuretic combination.4 The Nordic Diltiazem (NORDIL) trial compared treatment based on non-DCCB (diltiazem) with β-blocker/diuretic-based treatment.5 This study reported a significantly lower risk of stroke for patients treated with diltiazem-based therapy compared with the β-blocker/diuretic-therapy group, but a nonsignificant trend toward higher rates of myocardial infarction, cardiovascular death, and congestive heart failure in the diltiazem group. There were no differences in combined cardiovascular events or mortality. In none of these studies did results differ for participants with diabetes.
Evidence from retrospective studies suggests the possibility of increased cardiovascular mortality in diabetic patients receiving diuretics.6 These studies were not randomized, and significant baseline differences between patients receiving diuretics and the patients not receiving them may have existed. Furthermore, these studies were based on data collected in the 1970s, when high-dose diuretic treatment was the norm. The Systolic Hypertension in the Elderly Program (SHEP) study previously described showed that low-dose thiazide treatment of systolic hypertension in older diabetic patients reduced the cardiovascular event rate by 34% compared with placebo. The absolute risk reduction was twice as great for diabetic patients vs nondiabetic patients.7 However, most of this reduction in morbidity was secondary to decrease in stroke, prompting many to speculate that lowering BP might be beneficial for stroke reduction but not for prevention of myocardial infarction. To avoid the adverse effects of high-dose diuretics, it became popular to use “low-dose diuretics.” However, studies of low-dose diuretics were responsible for the data that proved them to be noneffective.2
Thiazides may not be effective in patients who have significantly decreased renal function (ie, glomerular filtration rate <60 mL/min). The effects of thiazide diuretics on the progression of early or advanced diabetic nephropathy have not been studied in large randomized clinical trials. Thiazides have been reported to have variable effects on insulin sensitivity. When compared with a verapamil-trandolapril combination, an atenolol-chlorthalidone combination at a similar level of BP reduction aggravated insulin resistance (insulin sensitivity index, from 0.8 ± 0.2 to 0.3 ± 0.1 × 10−4/min/unit/mL), increased serum triglyceride levels, and decreased high-density lipoprotein cholesterol and plasma potassium levels. The reduction in potassium suggests that the chlorthalidone component significantly contributed to the observed metabolic abnormalities.8
The potential for an increase in cardiovascular risk of diuretic therapy was first seriously raised by the results of the Multiple Risk Factor Intervention Trial (MRFIT), a primary prevention study that included a subgroup of 3600 men with abnormalities in baseline electrocardiography (ECG). Excess coronary artery disease mortality was observed among hypertensive men with ECG abnormalities at rest in the special intervention group who were prescribed diuretic drugs (RR estimated as 3.34 among men with baseline ECG abnormalities at rest and as 0.95 among men without such abnormalities) and was manifested chiefly as sudden death. The association between ECG abnormalities at rest and the coronary artery disease mortality rate among hypertensive men was independent of the baseline level of BP or of the findings on exercise ECG. Analysis of cumulative mortality data following antihypertensive regimens that included high dosages of diuretics revealed an association between ECG abnormalities at rest and diuretic treatment that related to adverse outcome. When the dosages of the diuretic were lowered, this trend was reversed.9,10 Stroke volume index and cardiac index were systematically lower during the entire period of treatment in SI receiving higher average doses of thiazide diuretics.11
The potential negative effect of diuretics was recently focused on by the publication of the Avoiding Cardiovascular Events Through Combination Therapy in Patients Living With Systolic Hypertension (ACCOMPLISH) trial.12 This was an outcomes study comparing an ACE inhibitor plus a diuretic (hydrochlorothiazide) with the same ACE inhibitor plus the DCCB amlodipine, titrated to an identical level of BP reduction, on a composite end point of death from cardiovascular causes, nonfatal myocardial infarction, nonfatal stroke, hospitalization for angina, resuscitation after sudden cardiac arrest, and coronary revascularization. The study was discontinued early because the outcomes of the amlodipine arm were significantly superior to the outcomes of the diuretic group. While one could reason that the results were a reflection of the beneficial qualities of amlodipine, one could just as easily reason that the difference in outcomes was due to a negative effect of the diuretic hydrochlorothiazide.
The publication of the results of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial calls attention to the issue of diuretic-induced hypokalemia in the treatment of hypertension in diabetes.13 The stated purpose of this trial was to determine whether lower systolic BP (<120 mm Hg) produced better outcomes than a less-intense systolic BP treatment target (<140 mm Hg) in patients with type 2 diabetes mellitus at high risk for cardiovascular events. BP reduction between groups was 14.2/6.1 mm Hg, a reduction that would have been predicted to reduce stroke by 38% and coronary heart disease by 24%. Surprisingly, treatment to the lower BP target did not improve the primary composite end point. However, there was a positive signal for stroke. The annual rate for stroke was 0.32% and 0.53% annually in the lower and higher BP groups, respectively (P<.01). In fact, stroke is the most sensitive target organ damage associated with hypertension.
Thus, the negative outcome of ACCORD is the result of the lack of a beneficial effect of lowering BP on the ischemic heart disease events included in the composite end point. The intensively treated group received roughly twice as much diuretic as the less-intensively treated group. In the intensive treatment group, diuretics were used in 83% and 89% of patients at 12 months and at the last visit, respectively, while in the standard care group, the usage was 52% and 56%. Thus, at 12 months, diuretic use was 33% higher in the intensive arm. This diuretic usage could account for the greater incidence of hypokalemia seen in the intensive treatment group (P=.01), reflecting the observations noted in the MRFIT trial. Analysis of data from the Systolic Hypertension in the Elderly Program (SHEP) trial14 suggests that increased risk associated with this degree of hypokalemia would essentially offset the projected benefit from BP reduction on ischemic heart disease events achieved in the ACCORD trial. This observation is similar to the lack of benefit seen with chlorthalidone shown in the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT).15
β-Blockers are the fourth most commonly prescribed drug class in the United States. Despite their less than desirable side effect profile, the “traditional”β-blockers—atenolol and metoprolol—have been considered first-line therapy in the management of hypertension by every report of the JNC, including JNC 7.16 However, in comparison with other antihypertensive drugs, atenolol was not found to be as effective as other antihypertensive drugs.
β-Blockers do offer at least one advantage over other antihypertensive drugs: they are particularly effective in preventing ischemic cardiovascular end points and decreasing the incidence of sudden cardiac death. Moreover, at least certain β-blockers have been found to be effective in the treatment of congestive heart failure. These attributes have been even more prominent in patients with diabetes mellitus. In the β-Blocker Heart Attack Trial (BHAT), propranolol was found to be more protective in the reduction of recurrent myocardial infarctions in the diabetes subgroup than in those without diabetes.17 Atenolol-based therapy was at least equivalent to ACE inhibitor–based therapy in improving cardiovascular and metabolic outcomes in the United Kingdom Prospective Diabetes Study (UKPDS) when examining the importance of achieving tight BP control in patients with type 2 diabetes mellitus.18
The results of the Anglo Scandinavian Cardiac Outcomes Trial (ASCOT) are helpful in choosing a particular β-blocker.19 In this study, conducted in a large cohort of relatively high-risk middle-aged hypertensive patients, a comparison of atenolol (usually paired with a thiazide) with the calcium channel blocker (CCB) amlodipine (usually paired with perindopril) revealed that major cardiac and stroke outcomes clearly favored the CCB/ACE inhibitor treatment arm. The trial was terminated prematurely owing to excess mortality in the patients randomized to the β-blockers. A clue to the cause of this unexpected observation is present in the ASCOT substudy, the Conduit Artery Function Evaluation (CAFÉ) trial, in which noninvasive techniques were used to compare atenolol- and amlodipine-based therapies on central aortic and peripheral BP.20 Although the two treatment arms had similar BP-lowering effects when measured in the brachial artery, the β-blockers were significantly less effective than the CCBs in reducing central pressure. Since both cardiac and stroke events are directly affected by the central BP, this difference in central aortic BP reduction is particularly important and may provide an explanation for the apparent failure of traditional β-blockers to reduce major clinical events to the same extent as other antihypertensive therapies, for example the Losartan Intervention for Endpoint Reduction (LIFE) trial. In the LIFE trial, the results favored the use of the ARB. Interestingly, it was the reduction in stroke that was responsible for the major effect on the composite end point.21
The third-generation β-blockers are characterized by the ability to produce arterial dilation and reduce peripheral vascular resistance in addition to blocking the β receptor. These effects are particularly important in the selection of a β-blocker for the diabetic patient. Two such drugs—carvedilol and nebivolol—are of particular interest in this situation. Carvedilol is a nonselective β-blocker with α1-receptor blockade. α1-Receptors mediate vasoconstriction in peripheral vessels, regulate blood flow to the kidneys, and have been linked to conditions such as myocardial hypertrophy; thus, the α1-blocking effects can modulate these pathophysiologic processes and increase blood flow and reduce hypertrophy. Nebivolol has been shown to have higher β1-selectivity than other β-blockers and produces endothelial-dependent vasodilation with increased bioavailability of nitric oxide (NO).22 Nebivolol is a racemic mixture of d-nebivolol (+SRRR) and l-nebivolol (−RSSS); both enantiomers appear to vasodilate equally.23 The l-enantiomer is responsible for the increased bioavailability of NO while the d-enantiomer is primarily responsible for blockade of the β-receptor. The exact mechanism for the increased bioavailability is not completely known but may involve the β3-receptor subtype and modulation of asymmetric dimethylarginine (ADMA).24
The third-generation β-blockers offer distinct benefits in diabetes. While the older β-blockers (propranolol, atenolol, and metoprolol) decrease insulin sensitivity, recent metabolic studies report that third-generation vasodilating β-blockers have beneficial effects on insulin sensitivity as well as on atherogenic risk factors and endothelial function.
The Glycemic Effects in Diabetes Mellitus/Carvedilol-Metoprolol Comparison in Hypertensives (GEMINI) trial, a double-blind randomized trial involving 1235 participants at 205 sites, studied the metabolic effects of carvedilol vs metoprolol in patients with type 2 diabetes and hypertension.25 Patients were randomized to receive metoprolol (50 mg twice a day, titrated up to 200 mg twice a day) or carvedilol (6.25 mg twice a day, titrated up to 25 mg twice a day) for 35 weeks. Similar BP levels were achieved in the two groups. Mean hemoglobin A1c increased significantly in the metoprolol group (0.15%; P<.001) but not in the carvedilol group (0.02%; P=.65), and insulin sensitivity improved significantly with carvedilol (−9.1%, P=.004) but not metoprolol (−2.0%, P=.48). Progression to microalbuminuria was less frequent with carvedilol than with metoprolol (6.4% vs 10.3%; P=.04). GEMINI is the first large-scale randomized trial to evaluate the addition of β-blockers to ACE inhibitors or angiotensin II receptor blockers to achieve the recommended BP target of <130/80 mm Hg for patients with type 2 diabetes. In these high cardiovascular–risk patients, carvedilol successfully achieved the BP goal while maintaining glycemic control.
Nebivolol is a novel third-generation, cardioselective (β1-adrenoceptor) lipophilic, vasodilatory β-blocker and produces endothelium-dependent vasodilation mediated through the l-arginine/NO) pathway.22–24 Nebivolol is the most cardioselective β-blocker when assessed in human myocardium. The hemodynamic effects of nebivolol differ from those of older β-blockers. In hypertensive patients, nebivolol administration decreased peripheral vascular resistance and preserved cardiac output, stroke volume, and left ventricular function.26 Nebivolol, unlike atenolol, has demonstrated efficacy in reducing central aortic pressure,26 and unlike metoprolol, increased insulin sensitivity in patients with hypertension.27
Hypertension Treatment Guidelines and Third-Generation β-Blockers
The latest edition of the guidelines of the European Society of the Cardiology/European Society of Hypertension, after considerable evaluation of all available data, continues to include the β-blockers among the other major drug classes as first-line considerations.28 Of note, however, these guidelines emphasize the positive attributes of nebivolol and carvedilol and recommend that they be the agents of choice. Likewise, the American Association of Clinical Endocrinologists, while recommending that β-blockers be strongly considered as additive therapy in diabetic patients already receiving an ACE inhibitor or an angiotensin receptor blocker, emphasize a preference for the vasodilatory agents.29
Recommendations for Diuretics and β-Blockers in the Treatment of Diabetes
At the present time, it does not seem prudent to utilize thiazide diuretics as first- or even second-line drugs in hypertensive diabetic patients. Although chlorthalidone has been determined to be a more potent antihypertensive agent than hydrochlorothiazide and to perhaps result in better clinical results, chlorthalidone was the agent used in SHEP and ALLHAT and failed to demonstrate an improved outcome. However, for patients requiring ≥3 drugs, thiazides may be useful. If they are used, great care must be given to monitoring blood potassium concentration. If there are no problems with potassium retention, treatment with the mineralocorticoid receptor antagonists spironolactone and eplerenone or potassium-sparing diuretics such as amiloride or triampterene should be considered as an alternative. In some situations, even a loop diuretic may be appropriate, particularly for patients with a reduced estimated glomerular filtration rate (<60 mL/min). β-Blockers should continue to be used for the compelling indications listed in JNC 7, ischemic heart disease, and heart failure. The benefits of β-blockade in reducing cardiovascular mortality, in particular sudden death, and morbidity may be achieved while avoiding the unwanted side effects on glucose metabolism, sexual function, fatigue, and cold extremities by using carvedilol or nebivolol. The hemodynamic profile and placebo-like adverse effect profile of nebivolol makes this drug very attractive for this purpose.