Beta-blockers are a prime example documenting a dissociation of the surrogate endpoint from the real endpoint: despite having a "beneficial" effect on blood pressure (surrogate endpoint), they fail to affect the real endpoints, i.e., heart attack, stroke, and cardiovascular and all-cause morbidity and mortality. This would indicate that, at present, millions of elderly hypertensive patients are needlessly exposed to the cost, inconvenience, and adverse effects of beta-blockers even though they will never harvest any benefits. Even investigators who, time and again, have recommended beta-blockers as first-line therapy in hypertension have admitted these agents are inefficient at preventing heart attacks in hypertensive patients, regardless of their age. Thus, Psaty et al state: "Perhaps the most interesting finding from the beta-blocker component of the meta-analysis is the fact that . . . beta-blockers do not appear to prevent coronary events in the primary prevention trials in patients with high blood pressure."8 It is ironic that the studies that clearly documented the inefficacy of the beta-blockers in preventing cardiovascular events provided the fundament upon which the recommendations of the JNC VI were built. Perhaps of even more concern in the MRC study in the elderly is that beta-blockers had a higher risk of cardiovascular events compared with diuretics, even after the difference was adjusted for the decrease in arterial blood pressure.6 Thus, for any given fall in arterial pressure, patients on diuretics fared better than those on beta-blockers.7 Obviously, this indicates either that lowering blood pressure by beta-blockade confers an ill effect on the cardiovascular system that overrides the beneficial effects of the decrease in pressure or that lowering blood pressure with a diuretic confers a specific benefit irrespective of the decrease in blood pressure.
In all prospective studies in which beta-blockers were implied to reduce morbidity and mortality, they were used in combination with a diuretic in the majority of patients. In the Swedish Trial in Old Patients (STOP) more than two thirds of the patients received combination therapy, and no information was provided regarding the effects of beta-blockers or diuretics in monotherapy.9
In the Systolic Hypertension in the Elderly Program (SHEP), only 21% of patients received atenolol, all in combination with a diuretic.10
In the study of Coope and Warrender, which demonstrated a significant reduction in the rate of strokes, 70% of patients in the treatment group received atenolol and 60% received bendrofluazide, although all of them were initially started on atenolol.11
Coope and Warrender clearly state: "Since patients were not randomized to treatment groups, it is impossible to compare response to the beta-blockers and the diuretics."11
It is hard to believe that these studies were considered to be ironclad scientific information documenting that beta-blockers reduce morbidity and mortality in hypertension. One could as well conclude that tonic water causes cirrhosis of the liver from a study in which the majority of patients in the active treatment arm were on gin and tonic, some on gin alone and some on tonic water alone, and no attempt was made to separately assess the effect of the individual ingredients. The studies of Coope and Warrender,11 SHEP,10 and STOP9 do not allow us to conclude that either beta-blockers alone or the addition of beta-blockers to the diuretic regimen did, indeed, significantly impact morbidity and mortality. To the contrary, the MRC study in the elderly allows us to conclude that beta-blocker-based therapy is distinctly inferior to diuretic-based therapy and is not different from placebo. Given this and the not-so-benign side-effect profile of beta-blockers, can we really blame practicing physicians for not following guidelines ?
In contrast to beta-blocker-based therapy, numerous prospective randomized trials have documented that diuretic-based therapy is effective in reducing morbidity and mortality in hypertensive patients.4 If anything, the benefits of diuretic therapy have been shown to be more marked in the elderly than in younger patients. The effect of diuretics is particularly pronounced with regard to reduction of the risk for stroke and somewhat less impressive with regard to reduction of the risk of coronary heart disease. However, of particular concern for many years, and even decades, is the possibility that this pharmacological intervention could adversely affect the risk for extracardiovascular diseases. Indeed, the very recent meta-analysis suggesting that long-term diuretic therapy could increase the risk for RCC is of distinct concern.5
Case-control and cohort studies
In 9 case control studies done in the past decade, an association between RCC and diuretic therapy was documented (odds ratio, 1.55; confidence interval, 1.42 to 1.71; P < 0.00001) (Fig. 2).5
Equally, in 3 cohort studies, in a total study population in excess of one million, patients who were taking diuretics had about a 2-fold higher risk of RCC than patients who were not on diuretic therapy.5
In most studies, women were found to have a higher risk of diuretic-associated RCC than men (odds ratio, 2.01 vs. 1.69). In 3 studies in which this was examined, the risk of RCC increased with duration of diuretic therapy (cumulative dose). The association between diuretic use and RCC was also found in normotensive subjects who took diuretics for another reason, and it persisted even when corrected for the presence of hypertension in the majority of the studies.
It is unlikely that hypertension itself was the common denominator accounting for the association between RCC and diuretic therapy. Indeed, it seems more likely that hypertension is merely an innocent bystander. There are no plausible clinical, biochemical or pathophysiological reasons why hypertension, per se, should be a risk for RCC. However, the longstanding presence of hypertension in any patient is intrinsically linked to diuretic use. Most patients will not remember taking a diuretic because numerous fixed combinations containing diuretics are presently on the market. As most clinicians will easily concede, it is next to impossible to perform a "correction" for the presence of hypertension (that is, to retrospectively separate hypertensive patients who used diuretics during the past 20-30 years from those who did not), as has been attempted in many studies. It seems more likely that the common denominator linking hypertension to RCC is, indeed, diuretic use. Regardless, in 5 out of the 9 case control studies, the risk of RCC with diuretic use persisted and remained significant even after adjustment for potentially confounding cofactors.
Hypothetical carcinogenic mechanism
Perhaps one of the most convincing arguments for the connection between RCC and diuretic use is the fact that RCC arises from the renal tubular cell, the main target of the diuretic’s pharmacologic effect. Conceivably, the chronic chemical bombardment of this cell over years or decades may have a low-grade carcinogenic effect.
Hydrochlorothiazide is a cyclic imide and can be converted in the stomach to a mutagenic nitroso derivative that is excreted in the kidneys.12,13 Diuretics have been associated with both nephropathy and renal cell tumors in animals.14,15 The thiazide diuretics cause massive degenerative changes and cell death in the distal tubule in rats.16 After thiazide exposure, these cells looked like tumor cells and exhibited markers of tumor cells.16
Renal cell carcinoma is a relatively rare malignancy that occurs two to three times more often in men than in women. The fact that most studies in our meta-analysis document women to be at a higher risk than men with regard to diuretic-induced RCC suggests the presence of a hormonal mechanism. Indeed, estrogens have been shown to enhance the thiazide effect in the distal tubule of ovariectomized rats.17 This effect could possibly account for the inverse gender predominance with regard to diuretic-associated RCC. In addition, although the use of diuretics has declined over the past decade, women still use 2 to 3 times more diuretic therapy than men do, possibly because women have a greater tendency for edema than men.18
Lack of evidence of RCC in prospective randomized trials
Carcinogenicity of diuretic therapy is low and certainly less than that of smoking for lung cancer. If one had to design a prospective randomized trial proving that smoking caused lung cancer, a study duration of at least 1 decade, but preferably 2, would be required. Given the comparatively weak carcinogenicity of diuretic therapy, it probably would take longer to document a difference with regard to RCC. Therefore, it is hardly surprising that in none of the prospective randomized trials, duration of which is usually < 5 years, was an excess of RCC found.
Diuretics were introduced into medicine in 1958. Since it probably takes more than 20 years of diuretic exposure to significantly increase the risk of RCC, we are only now seeing this association. Of note, the incidence of RCC has increased by 43% over the past 15 years.19
True risk-to-benefit ratio
Several epidemiologic studies, such as SHEP and MRC, allow us to estimate the true risk-to-benefit ratio of diuretic therapy in hypertension. It can be estimated that diuretic therapy for 1 case of RCC will prevent 20 to 40 strokes, 3 to 28 heart attacks, and 4 to 18 deaths in the general population.20 In the elderly, in which diuretics are particularly efficacious, the risk-to-benefit ratio may look even better. However, in middle-aged women, only 6 strokes, 2 heart attacks, and no deaths are prevented for 1 case of RCC.20 The actual risk-to-benefit ratio would clearly argue against the use of diuretics in this age and gender group.
We believe that younger and middle-aged women, therefore, probably should no longer be treated with diuretics for hypertension because they potentially will be exposed to these drugs for several decades, they have a well-known tendency to overuse diuretics, they are less protected by diuretics against cardiovascular morbidity and mortality than men, and their risk of diuretic-associated RCC is higher than that in men. In contrast, in patients with congestive heart failure and other forms of edema, the low-grade carcinogenicity of diuretic therapy can possibly be disregarded because their life expectancy is relatively short, and they are unlikely to live long enough for the cumulative diuretic dose to reach the threshold of carcinogenicity.
Both diuretics and beta-blockers have been used to treat essential hypertension for more than 3 decades. Both of these drug classes have impressive safety records that are unparalleled by other drugs. Despite this, no prospective randomized study has shown that beta-blockers, either in monotherapy or when added to diuretic therapy, diminish cardiovascular morbidity and mortality. Quite to the contrary, our recent meta-analysis in the elderly reported little if any, benefits of beta-blocker therapy when compared with placebo or other therapy, although blood pressure was lowered by the beta-blockers.
The inefficacy of beta-blockers may come from their unfavorable effects on systemic hemodynamics and on other pathophysiologic findings in the hypertensive patient, such as arterial stiffness, hypertensive heart disease, kidney disease, and cerebrovascular disease. In addition, comorbid conditions often present in the elderly, such as chronic obstructive pulmonary disease, peripheral vascular disease, diabetes, depression, and erectile dysfunction, are relative contraindications to the use of beta-blockers. It is ironic that the very same studies that demonstrate the inefficacy of beta-blockers in the elderly were used as an argument to promote them to a preferred status.
Recent data showing low-rate carcinogenicity for RCC with diuretic therapy must be seen in proper context. In the elderly, the cardiovascular benefits of diuretics clearly outweigh the low-grade risk of RCC. However, in younger patients, particularly in women, diuretics probably should no longer be used as initial antihypertensive therapy. In view of the unparalleled safety and efficacy of diuretics, no conclusions should be drawn with regard to safety and efficacy of other antihypertensive drugs.
In conclusion, sweeping recommendations for the use of beta-blockers and diuretics as "preferred" therapeutic strategies are inappropriate in hypertension, as is usually the case in medicine, a more sophisticated approach is needed.