Portal de Periódicos da CAPES

The ALLHAT trial

2003; Lippincott Williams & Wilkins; Volume: 21; Issue: 2 Linguagem: Inglês

10.1097/00004872-200302000-00003

1473-5598

Robert Fagard,

Hormonal Regulation and Hypertension

It is well-known that high blood pressure increases the incidence of fatal and non-fatal cardiovascular events such as stroke, coronary heart disease (CHD) and heart failure (CHF). However, effective and tolerable antihypertensive agents have only become available in the second half of the last century. These drugs have then been used as tools in randomized controlled outcome trials investigating whether treatment of hypertension would reduce the incidence of cardiovascular complications and of mortality. A large number of individual studies have been performed, first in patients with systolic-diastolic hypertension [1–14] and subsequently in patients with isolated systolic hypertension [15–17]. Meta-analyses of these studies clearly showed that all-cause and cardiovascular mortality, non-fatal and fatal stroke and CHD, and aggregate cardiovascular events were significantly reduced in both types of hypertension [18–21]. Diuretics have been used as first-line treatment in most of these trials [1–11,13–15] and are undoubtedly the best studied of the older drug classes. Some trials used a beta-blocker [10,12–14] or a dihydropyridine calcium channel blocker (CCB) [16,17] as initial therapy. No outcome trials have currently been published in patients with uncomplicated mild (grade 1) systolic hypertension or in the very elderly, but a meta-analysis revealed a cardiovascular benefit in patients with hypertension who were aged at least 80 years [22]. When newer classes of antihypertensive drugs became available, proper placebo-controlled outcome trials were not performed for most of them, either because of a lack of financial support or because ethical considerations precluded their organization. The research question was rephrased and several outcome trials were carried out in which new antihypertensive agents were compared with old or conventional therapy. Most of these trials have been included in a prospectively designed meta-analysis of such randomized trials in the framework of the Blood Pressure Lowering Treatment Trialists' Collaboration (BPLTTC) [23], the most comprehensive and authoritative source of information at the time of the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) publication. A comparison of 8097 patients randomized to angiotensin-converting enzyme (ACE)-inhibitor-based therapy with 8064 patients randomized to diuretic-based or beta-blocker-based therapy in three trials [24–27] revealed that there were no differences with regard to total and cardiovascular mortality, all major cardiovascular events and cause-specific events (i.e. stroke, CHD and CHF). Similar results were obtained when 11 685 patients randomized to dihydropyridine or non-dihydropyridine CCB-based-therapy were compared with 11 769 patients randomized to diuretic-based or beta-blocker-based therapy in five trials [25,28–32] except that the incidence of stroke was lower by 13% [95% confidence interval (CI): –2, –23] and the incidence of CHD higher by 12% (95% CI: 0, 26) with CCBs than with traditional therapy. In some of these trials, the first-line conventional drug was either a diuretic [28,29,31,32] or a beta-blocker [26,27], in others the choice of a diuretic or a beta-blocker was left to the discretion of the local investigators [24,25,30]. Such was the historical context when the results of the ALLHAT trial became available, a randomized double-blind multicentre clinical trial, in which treatment with, respectively, the dihydropyridine CCB amlodipine or with the ACE inhibitor lisinopril was compared with treatment with the diuretic chlorthalidone in hypertensive patients aged 55 years or older with at least one other risk factor [33]. The doxazosin arm of the trial had been terminated earlier because of a 25% higher incidence of cardiovascular events in participants assigned to the doxazonin group compared to those assigned to the diuretic, and the low likelihood of obtaining a significant difference for the primary outcome by the end of the trial [34]. Whereas meta-analyses provide more precise estimates of the magnitude of the effects under study and increase the statistical power, they cannot replace large individual studies [35]. The number of participants in the three arms of the ALLHAT study (n = 33 357) is only slightly less than in the trials comparing new and old drug classes in the BPLTTC overview (n = 37 402). The primary outcome (i.e. combined fatal CHD and non-fatal myocardial infarction) was not significantly different between amlodipine and chlorthalidone in ALLHAT (relative risk 0.98; P = 0.65). Therefore, ALLHAT does not confirm the 12% higher incidence with CCB-based therapy than with conventional therapy for similarly defined CHD in the BPLTTC overview. An important difference is that conventional therapy in previous trials could include diuretics or beta-blockers. Whereas the incidence of stroke was 13% less with CCBs than with conventional therapy in the BPLTTC overview, there was a tendency for a 7% lower incidence with amlodipine than with chlorthalidone in ALLHAT (P = 0.28). A true advantage of CCBs can therefore not be excluded. The overview furthermore showed a non-significant 12% higher incidence of CHF, mainly defined as CHF causing death or requiring hospital admission, with CCB-based therapy than with traditional therapy. The difference between the CCB and the diuretic was more impressive in ALLHAT and amounted to 35% (P < 0.001). This suggests that diuretic-based treatment is particularly beneficial in the prevention of CHF. However, another factor should be considered. In ALLHAT, 90% of the patients were on antihypertensive drugs before randomization and these were continued until participants received randomized study drug. Details on prior medication were not reported [33,36], but this could mean that a number of patients on diuretics were suddenly switched to the CCB (or to the ACE inhibitor), which may have caused fluid retention and contributed to CHF. It is obvious from the Kaplan–Meier curves [33] that the difference in incidence of CHF occurred almost immediately after randomization and that the curves ran parallel at least up to the fourth year of follow-up, after which the interpretation is hampered by the smaller numbers. Finally, there were no differences in mortality between CCB and conventional therapy in ALLHAT and in the BPLTTC analysis. The relative risk for CHD with ACE inhibitor-based therapy versus conventional therapy was 1.00 in the BPLTTC overview and 0.99 in ALLHAT, which leaves little doubt that ACE inhibitors are not superior to conventional therapy for this outcome. Whereas the two types of treatment did not significantly differ with regard to stroke in the overview of previous trials, the incidence of stroke was significantly greater by 15% with lisinopril than with chlorthalidone in ALLHAT (P = 0.02). However, the relative risk was 1.00 in non-Blacks, in accordance with the large majority of patients in the trials of the BPLTTC trial being white. The results on fatal/hospitalized CHF are discordant in that ALLHAT favours the diuretic and the BPLTTC overview the ACE inhibitors, but none of the relative risks was statistically significant. The superiority of chlorthalidone for all CHF in ALLHAT requires further clarification in the light of the prevailing therapy at the time of randomization. Finally, all-cause mortality was similar with both types of therapy in ALLHAT and in the overview of previous trials. The diuretic chlorthalidone was never significantly inferior to the CCB amlodipine or the ACE-inhibitor lisinopril in ALLHAT, and this is striking in view of the well-known metabolic disturbances, such as a decrease in serum potassium and increases in total cholesterol and fasting glucose. This could mean that the results with the diuretic might have been better if these side-effects could have been avoided or that diuretic-induced metabolic disturbances do not impact on prognosis. In the randomized, double-blind, placebo-controlled trial in elderly patients with systolic-diastolic hypertension by the European Working Party on High Blood Pressure in the Elderly, the first-line drug was a combination of a thiazide diuretic (hydrochlorothiazide) and a potassium-sparing agent (triamterene) [11]. Slight changes of serum potassium, blood sugar and of glucose tolerance were noted [37]. It is currently unknown whether prognosis would be better with a combination diuretic than with low-dose chlorthalidone as used in ALLHAT. Furthermore, the ALLHAT investigators emphasized that the results do not support reports of an increased risk of cancer, gastrointestinal bleeding or all-cause mortality with CCBs. This confirms the observations in the double-blind Systolic Hypertension in Europe trial in which older patients with systolic hypertension were randomized to the dihydropyridine CCB nitrendipine or placebo [16]. Approximately one decade ago, it was reported that ACE inhibitors are superior compared to other classes of antihypertensive drugs with regard to the regression of left ventricular hypertrophy and this fueled the expectation of a better outcome. However, this claim was based on meta-analyses of mainly single-drug studies or on inappropriate meta-analyses of comparative studies [38,39]. By contrast, when the comparative design of the individual studies was respected while meta-analysing the available material, there was no evidence of superiority of ACE inhibitors [40], and this conclusion is not contradicted by subsequent studies in which diuretics, beta-blockers, CCBs and/or ACE inhibitors have been directly compared. The major strengths of ALLHAT are its large size, the fact that the population is broadly representative of hypertensive patients over 55 years of age and the large representation of subgroups, such as patients with diabetes, in whom the overall results were not different from those of non-diabetics. However, a number of limitations have to be considered. As mentioned previously, 90% of the patients were on antihypertensive therapy at the time of randomization, and the possible impact of the sudden switch to study drugs should be examined. Whereas selection criteria for the trial included stage 1 or stage 2 hypertension, actual blood pressure without treatment is not known in the majority of the participants. For obvious reasons, step 2 and step 3 drugs were mostly limited to atenolol, clonidine, reserpine and hydralazine, which do not belong to the four first-line classes. Close to half of the randomized patients received additional therapy, but details were not reported. The fact that current practice and guidelines [41–43] would rather favour combinations of ALLHAT first-line drugs, as well as beta-blockers, should be considered in the interpretation of the results. Addition of atenolol may have been less effective in the lisinopril group than in the other two groups [44]. ALLHAT furthermore illustrates that it appears to be very difficult to achieve the same blood pressure control when different drugs or treatment regimens are compared in large outcome trials. Statistical adjustment for achieved blood pressure can at least partly account for this problem, but the impact of blood pressure differences on outcome cannot be ruled out completely. Finally, the current analysis of the ALLHAT trial is by intention-to-treat. As in other trials, the number of patients on study drugs decreased with time, and a certain number of patients were receiving drug classes from the other treatment arms. These percentages differed among treatment groups and may have influenced the results. Although the intention-to-treat analysis is certainly the most important one, a secondary on-treatment analysis could contribute to the interpretation of the results. There is little doubt that ALLHAT is a milestone in hypertension research, particularly with regard to the choice for initial pharmacotherapy. The main result that neither the CCB amlodipine nor the ACE inhibitor lisinopril is superior to the diuretic chlorthalidone in preventing major coronary and other cardiovascular events, is important and should lead to reconsideration of the declining role of diuretics in the treatment of hypertension [45,46]. Current guidelines for the management of hypertension [41–43] list a number of considerations in the selection of initial therapy, such as the strength of the evidence for the reduction of cardiovascular morbidity and mortality, the cardiovascular risk profile of the individual patient, target organ damage, coexisting diseases that may be beneficially or adversely affected by the antihypertensive agent, the possibility of interactions with drugs used for other conditions in the patient and the cost of therapy. The results from ALLHAT can only strengthen the prominent role of a diuretic as initial drug therapy, provided there are no contra-indications to diuretics or no specific indications for other types of drugs. Finally, many patients with hypertension need two or more drugs for adequate blood pressure control and diuretics should be considered as add-on therapy when treatment was started with a drug from another class. Acknowledgement The author gratefully acknowledges the secretarial assistance of N. Ausseloos.