Should a Diuretic Always Be the First Choice in Patients with Essential Hypertension? The Case for No

Diuretic-Induced Stimulation of the Renin System

Diuretics invariably stimulate renin secretion via the dual mechanism of intrarenal baroreceptor activation and the increase in sodium load presented to the macula densa. This effect may seem trivial, but it is not always fully appreciated that in the many (50%) hypertensive patients who have a “normal” renin profile, the activity of the system is already inappropriately high at baseline in relation to the levels of BP, and this is even more so in the 20 to 25% who have a “high” renin profile (6). Thus, in approximately three of four patients with essential hypertension, the diuretic-induced stimulation of renin secretion results in circulating levels of angiotensin II (Ang II) and of aldosterone, which are abnormally high. An excess of Ang II and of aldosterone, either systemically or locally generated, may be detrimental because, in addition to their well-recognized vasoconstrictive and sodium-retaining actions, they have a number of other negative effects. These include the potentiation of the sympathetic nervous system both centrally and peripherally (7), the stimulation of cardiac myocytes and fibroblasts that favor the development of cardiac hypertrophy (8), the multiple hemodynamic and nonhemodynamic actions that accelerate the progression of renal injury (9), the ability of increasing the generation of the reactive oxygen species in the vessel wall (10), and, finally, the multiple intracellular signaling pathways that interfere with insulin action (11). Thus, it was proposed recently that Ang II may have a pivotal role in amplifying the oxidative stress brought about by a number of risk factors, including high cholesterol, diabetes, smoking, and high BP itself (12).

Obviously, one may wonder whether the activation of all of these deleterious actions actually has a relevant impact on the clinical outcome of patients. In this respect, it worth recalling that already in the early 1990s, Alderman et al. (13) showed that a high renin profile is in itself a risk factor; indeed, they reported that in a large cohort of hypertensive patients who were stratified according to baseline renin levels but had similar levels of glucose, cholesterol, and smoking, those with a high renin profile had a significantly greater incidence of myocardial infarction with respect to those with normal or low renin profile. These findings were more recently confirmed and extended by the same investigators (14).

The idea that the activation of the renin system is harmful for the cardiovascular system is reinforced by the consideration that its antagonism has been proved to be protective in a number of clinical conditions (Table 1), some of which are not even associated with high BP. Also with respect to the selection of the antihypertensive drugs, needless to say that among those recommended as first choice by the recent European and American guidelines, diuretics are the only one to stimulate the renin system, all of the others being either suppressive or neutral.

One very practical reason for not preferring the renin-stimulating drugs is that the reactive rise in Ang II largely blunts the antihypertensive effects of these agents. For instance, Schmidt et al. (15) showed that the long-term reduction of systolic and diastolic BP achievable with the combination of 25 mg/d hydrochlorothiazide with 80 mg of the angiotensin receptor antagonist valsartan is twofold that obtained with the same dose of hydrochlorothiazide administered alone.

Diuretic-Induced Metabolic Alterations

Apart from the inherent limitations of their antihypertensive effects, diuretics can cause several metabolic alterations that may prove to be disadvantageous, particularly in the long term, the most relevant of which is the disturbance in glucose metabolism (16,17). The results of numerous large intervention trials support this concept. The Intervention as a Goal in Hypertension Treatment Trial (18) demonstrated that among hypertensive patients who were treated with a combination of hydrochlorothiazide and amiloride, the incidence of new case of diabetes was significantly higher than in those who were treated with the calcium antagonist nifedipine; patients of the former group also had a higher increment of serum uric acid and of lipids than the latter group. Similar results were observed in the Captopril Prevention Project (CAPPP), in which a converting enzyme inhibitor was compared with a conventional therapy that included diuretics (19), and in the ALLHAT itself (1), in which the incidence of new cases of diabetes at the end of the study was 11.6, 9.8, and 8.1% in the chlorthalidone, amlodipine, and lisinopril groups, respectively. This drawback of diuretics has been stressed further by the results of the recent Antihypertensive Treatment and Lipid Profile in a North of Sweden Efficacy Evaluation (ALPINE) Study (20), in which newly detected hypertensive patients who were treated for 1 yr with 25 mg/d hydrochlorothiazide alone or in combination with atenolol had a significantly greater incidence of new cases of diabetes and of metabolic syndrome than a comparison group that was treated with 16 mg/d candesartan alone or in combination with the calcium antagonist felodipine. Moreover, overall, the diuretic-treated group developed some degree of insulin resistance in response to an oral glucose tolerance test, which did not occur in the candesartan-treated group. This alteration was associated with those less evident but also statistically significant in triglycerides and HDL. Along this line in another recent investigation, Grassi et al. (21) showed that hydrochlorothiazide and candesartan administered to obese hypertensive patients in doses similar to those used in the ALPINE Study caused significant and similar reduction in BP; however, the treatment with the Ang II receptor antagonist reduced the muscle sympathetic nerve activity measured by microneurographic recording and increased the insulin sensitivity, expressed as the ratio between insulin and glucose levels during an oral glucose load, whereas the thiazide did not significantly affect the peripheral sympathetic activity and worsened insulin sensitivity.

Again, the question that arises is whether these alterations are clinically relevant, because in the 5 yr of follow-up of the ALLHAT, the greater incidence of new cases of diabetes observed in diuretic-treated patients was not associated with a greater incidence of major cardiovascular events (1). However, 5 yr time may be not enough to disclose the negative consequences of the diuretic-induced metabolic alterations because the incubation period for vascular disease may be very long (22). In this respect, a previous long-term study from Sweden showed that in hypertensive patients who developed new-onset diabetes, the coronary risk was increased by 48% in comparison with those who did not developed it (23). Also, Dunder et al. (24) showed that even small increments of blood glucose are the best predictor of subsequent risk for myocardial infarction in middle-aged hypertensive patients. Even Coutinho et al. (25) reported that a small increase of fasting glucose within the normal range (from 75 to 110 mg/dl) can increase the cardiovascular events by 33% over 12 yr. Alderman et al. (26) in a 6.3-yr follow-up study observed that in hypertensive diabetic patients who were frequent users of diuretics, the incidence of cardiovascular events was greater than in moderate or rare users. Finally, in a very recent study, Verdecchia et al. (27) in a large cohort of hypertensive patients found that the antihypertensive treatment included a diuretic in 53.5% of those who developed new cases of diabetes versus 30.4% of those in whom diabetes did not developed. Moreover, after adjustment for several confounders, the relative risk of events in the group with new diabetes and in the group with diabetes at entry was similarly increased when compared with patients who were persistently free of diabetes, indicating that the treatment-induced diabetes portends a risk not dissimilar from that of previously known diabetes.

Long-term diuretic treatment may also cause hypokalemia, which in the ALLHAT occurred in 12.7% of patients who were receiving thiazide-based treatment against 2.6 and 1.5%, respectively, in patients who were receiving amlodipine- and lisinopril-based treatment. Obviously, the cardiac risk associated with hypokalemia depends on the degree of serum potassium reduction, and it is held that the hypokalemia that ensues from the doses of diuretics commonly used should not do any arm (28); yet in the Systolic Hypertension in the Elderly Program (29), in which all patients were receiving a diuretic-based treatment, those with serum potassium levels <3.5 mEq/L had no benefit in terms of cardiovascular events despite decreases in BP similar to those observed in patients with ≥3.5 mEq/L, suggesting that even mild hypokalemia may counterbalance, in some way, the beneficial effect of BP lowering.

Compliance and Cost

It is well known that the compliance to diuretic treatment is poor and that after 1 yr of therapy, only one third of patients stay on this drug in comparison with 50 to 60% of those who are treated with the more modern drugs, namely the angiotensin-converting enzyme inhibitors and the Ang II receptor antagonists. This may be because of the side effects that are associated with the use of this class of drug. In the ALPINE Study (20), the percentage of patients who had serious side effects that led to the change in therapy was almost double that of those who were treated with candesartan. It is also plausible that negative effects on sexual functions contribute the to the low tolerability of diuretics. Indeed, in the Treatment of Mild Hypertension Study (30), it was reported that after 2 yr of follow-up, 17.1% of male patients who were receiving chlorthalidone were complaining of some degree of sexual disturbance against 8.3 and 9.7%, respectively, of those who were receiving amlodipine and enalapril.

The frequency of side effects with the resultant need to change therapy is known to have a major impact on the cost in that it brings about the expenses for further investigations, occasional hospital admissions, and additional drugs (e.g., potassium supplement, antidiabetic agent) (31); moreover, the change in therapy frequently leads patients to stop whatever antihypertensive treatment they are receiving. Thus, the application of a more expensive treatment at the beginning of therapy may imply a substantial and continued economical saving in the long term, whereas the opposite may turn out to be true for some inexpensive drugs (32).

As a final consideration, it should be said that the debate on the first-choice drug may seem nowadays somehow out of date considering that the vast majority of the hypertensive patients require a combination of at least two drugs to control BP; moreover, even the recent JNC report recognizes not less than 15 special or compelling clinical conditions that justify the use of antihypertensive drugs other than diuretics as first choice (33) (Table 2). Thus, as it has been stated by Jones and Hall (34) in their recent editorial commentary to the JNC guidelines on hypertension, it seems wise to conclude that “the key emphasis should be on the fact that lowering BP is more important than the choice of the antihypertensive agent.”