Diuretics

Loop and thiazide diuretics

Loop and thiazide diuretics have been the mainstay of treatment for symptomatic heart failure [ ], relieving symptoms and improving cardiovascular hemodynamics. However, despite their widespread use, they have not been shown to improve survival in patients with heart failure. As it is not feasible to conduct such a trial in patients with pulmonary edema due to heart failure, the place of diuretic therapy in the management of heart failure appears secure.

Aldosterone receptor antagonists

Although they are widely used in the management of heart failure, loop and thiazide diuretics have not been shown to prolong survival. However, spironolactone and eplerenone must be added to the list of medications that offer improved survival for patients with heart failure.

In the Randomized Aldactone Evaluation Study (RALES) in 1663 patients with New York Heart Association (NYHA) class III (70%) or IV (30%) symptoms and an ejection fraction less than 35%, the addition of spironolactone 25 mg/day to conventional treatment (an ACE inhibitor, a loop diuretic, in most cases digoxin, and in 11% a beta-blocker) for an average of 24 months lowered the risk of all-cause mortality by 30% (from 46% to 35%), death from progressive heart failure, and sudden death [ ]. There were similar reductions in hospital admissions for worsening heart failure and for all cardiac causes. The magnitude of the overall effect was similar and additional to the proven benefit from ACE inhibition in severe heart failure.

In the Eplerenone Post-Acute Myocardial Infarction Heart Failure Efficacy and Survival Study (EPHESUS), in 6632 patients with an acute myocardial infarction complicated by left ventricular dysfunction and heart failure, the addition of eplerenone 25–50 mg/day to optimal medical therapy significantly reduced all-cause mortality by 15% and cardiovascular mortality by 17% over a mean follow-up period of 16 months; hospitalization rates were also reduced [ ].

Thiazide and loop diuretics

Various factors influence the choice of a diuretic in patients with renal insufficiency. First, it is widely believed that thiazide diuretics are ineffective once renal function falls below a creatinine clearance of 40–50 ml/minute. Diuretics need to enter the renal tubule to reach their luminal sites of action. In renal insufficiency, higher diuretic doses are required to overcome disease-related impediments to drug delivery to their sites of action. In the case of the loop diuretics the common practice is to titrate the dose of the diuretic until the desired response occurs [ ]. No such titration generally occurs with thiazide-type diuretics; consequently they are likely to fail, although for no reason other than their not being titrated to a truly effective dose. The true basis for “failure” of thiazide-type diuretics in patients with renal insufficiency resides in the fact that these diuretics are not of sufficient maximal efficacy to produce adequate volume control in these typically volume-expanded patients [ ].

However, resistance to loop diuretics can occur by various mechanisms [ ]. These include poor adherence to therapy, poor absorption, progressive worsening of heart failure, excess volume loss, renal insufficiency, secondary hyperaldosteronism, and hypertrophy of the tubular cells of the distal nephron. Resistance due to inadequate drug absorption—either its speed or extent—is common with furosemide, which is poorly absorbed [ ]. Once recognized, this hurdle to response can be overcome by using loop diuretics that are predictably well absorbed, such as bumetanide and torasemide or by giving intravenous furosemide [ ].

If loop diuretics fail to produce the desired diuretic response, combination diuretic therapy can be considered by adding a thiazide or a thiazide-like diuretic, such as metolazone. Such combinations are generally quite successful in both advanced congestive heart failure and late-stage chronic renal insufficiency, although excessive diuresis is a constant risk with such combinations. An excess diuresis with combination loop plus thiazide diuretic therapy is best managed by temporarily withdrawing both diuretics. Generally, diuretic doses are reduced when therapy is resumed [ ].

An alternative in the diuretic-resistant patient is the use of continuous infusions of loop diuretics rather than bolus diuretic therapy. Such infusions can also be given with a small volume of hypertonic saline, with good effect [ ]. The reasons why continuous infusions of loop diuretics work when bolus doses have failed may relate to a more efficient time-course of diuretic delivery and/or less activation of the renin–angiotensin system [ ]. Furosemide and torasemide may be the safest loop diuretics to be given as infusions, in that infusion of bumetanide has been associated with severe musculoskeletal symptoms [ ].

The importance of volume control in patients with renal insufficiency extends beyond its effect on blood pressure. Accordingly, the addition of hydrochlorothiazide can overcome the blunting by a high sodium intake of the therapeutic efficacy of ACE inhibition on proteinuria [ ]. This presumably relates to volume-related activation of the renin–angiotensin system.

Finally, loop diuretics reduce the metabolic demand of tubular cells, reducing oxygen requirements and thereby, in theory, increasing resistance to ischemic insults and perhaps other toxic circumstances. This property has been advanced as the basis for using diuretics in acute renal insufficiency. Although an attractive hypothesis to date, there is no compelling evidence to suggest any benefit from loop diuretics in established acute renal insufficiency. Alternatively, loop diuretics can convert oliguric to non-oliguric acute renal insufficiency, thereby easing the fluid restriction that would otherwise be necessary in such patients [ ].

Aldosterone receptor antagonists

It is no longer appropriate to consider the endocrine or paracrine properties of aldosterone as being restricted to the classical target cells. Hemodynamic and humoral actions of aldosterone have important clinical implications for the pathogenesis of progressive renal disease, and may therefore affect future antihypertensive strategies. Initially, one might anticipate that the adverse effects of aldosterone could be attenuated merely by blocking aldosterone release with either an ACE inhibitor or angiotensin-II receptor antagonists. However, this appears not to be the case. Several investigators have now shown that ACE inhibitors acutely reduce aldosterone concentrations, but that with continued use this suppression fades. Thus, the presumption that ACE inhibitors would suppress the production of both aldosterone and angiotensin-II was incorrect. So, although ACE inhibitors and angiotensin-II receptor antagonists are individually very effective in retarding disease progression, additional benefit may be realized with a concurrent aldosterone receptor antagonist. As observed in clinical studies of congestive heart failure, as well as in animals with renal disease, antagonism of aldosterone receptors protects against end-organ damage through a combination of both hemodynamic and direct cellular actions [ ].

An important consideration regarding the feasibility of aldosterone receptor antagonist therapy in chronic renal insufficiency is the risk of provoking hyperkalemia. Many patients with chronic renal insufficiency are already taking an ACE inhibitor or an angiotensin-II receptor antagonist, with the attendant risk of hyperkalemia. Despite such concerns, the results of the RALES and EPHESUS trials have been reassuring [ ]. In those studies, patients taking an ACE inhibitor who were randomized to spironolactone or eplerenone 25–50 mg/day had only small increases in potassium concentrations. Although the differences between those who took aldosterone receptor antagonists and those who took placebo were statistically significant, the mean increases were not clinically important, and serious hyperkalemia was uncommon. However, in clinical practice the risk of hyperkalemia may be greater [ ], and close laboratory monitoring and judicious use of these drugs is necessary to minimize the risk.

Although it is an effective aldosterone receptor antagonist, spironolactone is limited by its tendency to cause undesirable sexual adverse effects. At standard doses, impotence and gynecomastia can occur in men, and premenopausal women can have menstrual disturbances. These adverse effects, caused by the binding of spironolactone to progesterone and androgen receptors, are substantial causes of drug withdrawal. In the RALES study there was a 10% incidence of gynecomastia or breast pain in men, compared with 1% with placebo, and significantly more patients discontinued treatment (2% versus 0.2%). Although troublesome, these adverse effects are reversible and dose-related. The advent of selective aldosterone receptor antagonists, such as eplerenone, should reduce these adverse effects and thereby improve patient compliance. In EPHESUS there was no increase in the incidence of gynecomastia, breast pain, or impotence in men or menstrual irregularities in women who took eplerenone.

Cardiac dysrhythmias

Changes in potassium metabolism supposedly cause electrical instability in the heart, cardiac dysrhythmias, and increased mortality; replacement of potassium has been said to eliminate the risk of dysrhythmias [ ]. Mild hypokalemia might be expected to cause dysrhythmias in patients with serious organic heart disease (cardiomegaly, an abnormal electrocardiogram, frequent ventricular extra beats before treatment). However, the evidence suggests that hypokalemia after myocardial infarction is not the cause of dysrhythmias, but that both are the result of excess catecholamines. Furthermore, although hypertensive patients with left ventricular hypertrophy have an increased frequency of ventricular dysrhythmias, extra beats do not increase in frequency during diuretic treatment, even in the face of profound hypokalemia [ ].

Early evidence linking thiazide-induced hypokalemia with dysrhythmias and sudden death was indirect and tenuous at best [ , ]. One study suggested that diuretics are not responsible for the relation between hypokalemia and ventricular fibrillation in acute myocardial infarction [ ]. Chronic preoperative hypokalemia due to diuretics was not a risk factor for intraoperative dysrhythmias [ ]. Two large studies using 24-hour electrocardiographic monitoring failed to show a relation between diuretic-induced hypokalemia and ventricular dysrhythmias [ , ].

However, a retrospective analysis of 6797 patients with ejection fractions below 0.36 enrolled in the Studies Of Left Ventricular Dysfunction (SOLVD) was conducted to assess the relation between diuretic use at baseline and the subsequent risk of dysrhythmic death [ ]. Patients who were taking a diuretic at baseline (n = 2901) were significantly more likely to have such an event than those not taking a diuretic (n = 2896): 3.1 versus 1.7 per 1000 person-years. On univariate analysis and after controlling for important co-variates, the relation remained significant (relative risks 1.85 and 1.37 respectively). However, the association was seen only with non-potassium-sparing diuretics (n = 2495; relative risk 1.33); for potassium-sparing diuretics, alone or in combination with a non-potassium-sparing diuretic (n = 406), the relative risk was 0.90.

These data suggest that diuretic-induced potassium disturbances can cause fatal dysrhythmias in patients with left ventricular systolic dysfunction. SOLVD were not randomized trials of the risk of dysrhythmic death caused by diuretics. On average, patients retaking diuretics not only had lower serum potassium concentrations, but were also older, had more severe heart failure and were more likely to be taking antidysrhythmic drugs at baseline, although they had fewer indicators of ischemic heart disease. Even controlling for bias in multivariate analysis does not exclude the influence of unrecognized confounders. It is unknown whether diuretics were continued or changed during the 3 years of the trial. Thus, it remains uncertain that diuretic therapy is related to a risk of sudden dysrhythmic death in patients with heart failure.

Diuretic-induced hypokalemia is undoubtedly associated with a risk of serious ventricular dysrhythmias if diuretics are co-administered with drugs that prolong the QT interval [ ]. Diuretics increase the risk of torsade de pointes during antidysrhythmic drug therapy, independent of serum potassium concentration [ ]. The list of cardiac and non-cardiac drugs that prolong the QT interval continues to lengthen [ ]. It includes various antidysrhythmic drugs, including ibutilide, almokalant, and dofetilide; antimicrobial drugs, including clarithromycin, clindamycin, co-trimoxazole, pentamidine, imidazoles (such as ketoconazole), some fluoroquinolones, and antimalarial drugs (quinine, halofantrine); histamine H1 receptor antagonists (terfenadine and astemizole); the antidepressant zimeldine; antipsychotic drugs (pimozide and sertindole); tricyclic/tetracyclic antidepressants; and cisapride. Particular care should be taken to avoid diuretic-induced hypokalemia when any of these agents is co-prescribed.

Coronary heart disease

There is no valid evidence that diuretics contribute to myocardial infarction, sudden death, or a failure of antihypertensive treatment or other risk factor interventions to prevent coronary deaths [ ]. An association between diuretics and sudden death has been suggested only in selected subset analyses, which allow no valid conclusions. Even in subjects with electrocardiographic abnormalities before treatment, there is no sound or consistent evidence to support the suggestion that diuretics predispose to sudden death.

Two retrospective case-control studies have reported an increased risk of cardiac arrest in hypertensive patients treated with thiazide-type diuretics [ , ]. The risk was less among those treated with low-dose thiazides (equivalent to hydrochlorothiazide 25 mg/day) or with thiazides plus potassium-sparing drugs. The case-control design lacks one of the major advantages of randomized clinical trials, a tendency to equalize unknown, but important, differences between the comparison groups. Underadjustment is usual in case-control studies; for example, patients with more severe hypertension were more likely to have been treated with higher doses and less likely to have received potassium-sparing drugs because of renal dysfunction. Failure to randomize treatment tends to exaggerate differences between groups. Data from the prospective observational Gothenburg study suggested that metabolic changes during long-term treatment with antihypertensive drugs (predominantly beta-blockers and thiazides) are not associated with increased risk of coronary heart disease [ ].

Eyes

In 1077 patients with intraocular pressures of 22–29 mmHg, normal visual fields, and normal optic discs, who were followed every 6 months for 5 years in the European Glaucoma Prevention Study, a multivariate analysis showed that the use of diuretics (HR = 2.41, 95% CI = 1.12, 5.19) was one of several factors associated with the development of open-angle glaucoma [ ]. Systemic hypertension was not associated with the conversion to open-angle glaucoma. Information concerning the use of any medications was obtained at baseline and every 6 months throughout the study. However, particular classes of diuretics were not separately cited. This contributory effect of diuretics may be explained by the play of chance alone or by as yet unknown detrimental effects of diuretics on the retinal ganglion cells. A possible reduction in ocular perfusion pressure induced by a diuretic-related reduction in systemic pressure may be an additional mechanism. In this study, diuretics were more often used in combination with other antihypertensive drugs, particularly in the patients who developed open-angle glaucoma. Blood pressure readings were not obtained during the study, and so it could not be determined whether there was hypotension.

Hearing

Sensorineural hearing loss is an important adverse outcome in survivors of neonatal intensive care illnesses, particularly those with persistent pulmonary hypertension of the newborn. The relations between ototoxic drugs and 4-year sensorineural hearing loss have been assessed in a prospective, longitudinal outcome study in near-term and term survivors of severe neonatal respiratory failure who were enrolled in the Canadian arm of the Neonatal Inhaled Nitric Oxide Study [ ]. A combination of loop diuretic use for more than 14 days and an average dose of neuromuscular blocker greater than 0.96 mg/kg/day contributed to sensorineural hearing loss among survivors (OR = 5.2; 95% CI = 1.6, 17). Cumulative dose and duration of diuretic use and overlap of diuretic use with neuromuscular blockers, aminoglycosides, and vancomycin were individually linked to sensorineural hearing loss. These studies implicate loop diuretics and neuromuscular blockers individually and possibly synergistically in sensorineural hearing loss.

Impaired glucose tolerance and insulin resistance

There is little doubt that high dosages of diuretics carry an appreciable risk of impairing diabetic control in patients with established diabetes mellitus. However, their role in causing de novo glucose intolerance is not clear [ ]. Long-acting diuretics are more likely to alter glucose metabolism. Impaired glucose tolerance is a relatively rare complication with loop diuretics, although isolated cases of non-ketotic hyperglycemia in diabetics have been described [ ].

The effects of thiazide-type diuretics on carbohydrate tolerance cannot be ignored [ ]. There is a definite relation between diuretic treatment, impaired glucose tolerance, and biochemical diabetes, and a possible relation with insulin resistance [ ]. It is well established that the effect of thiazides on blood glucose is dose-related, probably linearly, while the antihypertensive effect has little relation to dose [ ]. There is relatively little information on the time-course; numerous short-term studies have shown that the blood glucose concentration increases in 4–8 weeks [ ]. The evidence that current low dosages impair glucose tolerance in the long term is not entirely consistent, perhaps because of differences between studies in dosages, diuretics used, durations of treatment, and types of patient [ , ]. In SHEP, low-dosage chlortalidone in elderly patients for 3 years resulted in a non-significant excess of diabetes (8.6% versus 7.5%) compared with placebo [ ]. The apparent differences between diuretics may be due to comparisons of dosages that are not equivalent [ ]. Important differences between individual diuretics will be established only when their complete dose–response relations for metabolic variables and blood pressure have been defined [ ].

In contrast to the wealth of evidence on impaired glucose tolerance in diabetics, sound clinical trials of the effect on insulin resistance are difficult to find, considering the amount of comment and speculation on the topic [ , ]. It is not known whether insulin resistance is completely or even partly responsible for the changes in glucose tolerance that occur during long-term thiazide treatment; impaired insulin secretion may also have a role [ ]. Hypokalemia or potassium depletion may contribute to impaired glucose tolerance, by inhibiting insulin secretion rather than by causing insulin resistance, but is not the only or even the main cause of impaired glucose tolerance during long-term diuretic treatment [ , ]. The routine use of potassium-sparing diuretics with relatively low dosages of thiazides does not prevent impaired glucose tolerance.

Diuretics worsen metabolic control in established diabetes, but it is not known whether this adversely affects prognosis [ ]. Disturbances of carbohydrate homeostasis have been detected by detailed biochemical testing, but their clinical importance is uncertain [ ]. The major clinical trials have not shown a major risk of diabetes mellitus. The incidence of diabetes mellitus in diuretic-treated subjects is only about 1%, even when large dosages are used [ ]. In ALLHAT, among patients classified as non-diabetic at baseline, the incidence of diabetes after more than 4 years was 12% with chlortalidone compared with 9.8% (amlodipine) and 8.1% (lisinopril). Despite these trends, there was no excess of cardiovascular events or mortality from chlortalidone in the entire population or among patients with diabetes. Although these data are reassuring, observational data suggest that diuretic-induced new-onset diabetes carries an increased risk of cardiovascular morbidity and mortality, but that this may take 10–15 years to become fully apparent [ ].

Since changes in glucose balance after diuretics tend to be reversible on withdrawal, measures of carbohydrate homeostasis should be assessed after several months of thiazide treatment to detect those few patients who experience significant glucose intolerance [ ]. With this approach, the small risk of diabetes mellitus secondary to diuretic therapy can be minimized.

Hyperuricemia

Most diuretics cause hyperuricemia. Increased reabsorption of uric acid (along with other solutes) in the proximal tubule as a consequence of volume depletion is one reason; however, diuretics also compete with uric acid for excretory transport mechanisms. There is a small increased risk of acute gout in susceptible subjects [ ]. In the large outcome trials, about 3–5% of subjects treated with diuretics for hypertension developed clinical gout [ ]. In those with acute gout during diuretic treatment, attacks were more strongly related to loop diuretics than to thiazides [ ]. Gout was significantly associated with obesity and a high alcohol intake in the subgroup taking only a thiazide diuretic. About 40% of cases of acute gout may have been prevented by avoiding thiazides in those 20% of men who weighed over 90 kg and/or who consumed more than 56 units of alcohol per week.

Well-conducted studies have shown that diuretic-induced changes in serum uric acid are dose-related [ , ]. In low-dosage regimens, as currently recommended, alterations are minor, and other than the risk of gout the long-term consequences of an increased serum uric acid are unknown.

The issues of whether hyperuricemia is an independent risk factor for cardiovascular disease and the clinical relevance of the rise in serum uric acid caused by diuretic treatment are controversial [ ]. In the Systolic Hypertension in the Elderly Program (SHEP), diuretic-based treatment in 4327 men and women, aged 60 years or more, with isolated systolic hypertension was associated with significant reduction in cardiovascular events [ ]. Serum uric acid independently predicted cardiovascular events in these patients. The benefit of active treatment was not affected by baseline serum uric acid. After randomization, however, an increase in serum uric acid of less than 0.06 mmol/l (median change) in the active treatment group was associated with a hazard ratio (HR) of 0.58 (CI = 0.37, 0.92) for coronary heart disease compared with those whose serum uric acid rose by 0.6 mmol/l or more. This was despite a slight but significantly greater reduction in both systolic and diastolic blood pressures in the latter. Those with serum uric acid increases of 0.6 mmol/l or more in the active group had a similar risk of coronary events as those in the placebo group.

This analysis of the SHEP database confirms the findings of a systematic worksite hypertension program [ ]. In 7978 treated patients with mild to moderate hypertension, cardiovascular disease was significantly associated with serum uric acid (HR = 1.22; CI = 1.11, 1.35), controlling for known cardiovascular risk factors. The cardioprotective effect of diuretics increased from 31% to 38% after adjustment for serum uric acid.

These observations suggest that persistent elevation of serum uric acid during diuretic-based antihypertensive therapy may detract from the benefit of blood pressure reduction. However, the relation between serum uric acid and cardiovascular disease was independent of the effects of diuretics. Furthermore, low-dose thiazide regimens have a smaller impact on serum uric acid [ ].