Portal de Periódicos da CAPES

The ALLHAT study

2003; Lippincott Williams & Wilkins; Volume: 21; Issue: 3 Linguagem: Inglês

10.1097/00004872-200303000-00001

1473-5598

Stevo Julius,

Hormonal Regulation and Hypertension

The recent results of the ALLHAT study [1] are an important contribution to the therapeutics of hypertension. However, the early conclusions in the press, and comments by some opinion leaders that ALLHAT proved that diuretics ought to be the firstline treatment in all patients with hypertension, are surprising. There is nothing in ALLHAT to support such a wide generalization. Let us start with the obvious: ALLHAT did not investigate initiation of treatment. In 90% of patients, ALLHAT evaluated the effect of 'roll over' from previous treatment to a well defined and, by design, restrictive treatment scheme; a scheme that excluded what is usually carried out in clinical medicine. Diuretics could not be combined with angiotensin-converting enzyme (ACE) inhibitors and a calcium antagonist could not be added to the other two treatment regimens. How this relates to starting treatment, and what should be the next step when the first step fails, is beyond me. Nevertheless, it is predictable that the ALLHAT results will be neatly ensconced as the first mandatory step in future guidelines. I understand the economics of this, but not the science. However, the above is a minor concern compared to the fact that proponents of generalization chose to ignore what kind of population has been studied in ALLHAT. Is it really possible to project findings from high risk-advanced hypertension to the population of mild hypertension (stage 1, to be guideline-wise correct), the population in which we are presumed to start treating hypertension? In my opinion, such a generalization is not tenable on epidemiological, physiological and clinical grounds. Epidemiologically, the largest proportion of all patients with hypertension have stage 1 hypertension and, in a decade of follow-up, 70% of hypertension-related mortality occurred among patients who at baseline had stage 1 hypertension [2]. From the population-risk standpoint, these are the most important patients. How is it then that ALLHAT, and most current outcome trials comparing different drugs, opted to study high-risk populations? It is not because the investigators are unaware of the importance of mild hypertension; the choice is rooted in calculations of sample sizes needed to prove the point. In their infancy, outcome studies in hypertension could compare the effect of blood pressure-lowering-drugs with placebo. The differences from placebo were so large that, even in mild hypertension [3], it was possible to demonstrate the superiority of treatment in relatively small sample sizes. Present studies seek to show outcome differences when the blood pressure is lowered to a similar degree with different antihypertensive agents. Obviously in such a setting, only small difference can be expected and a large number of events are needed to show the difference. Thus, high-risk patients in whom many events are expected are chosen for study. The euphemism of 'high risk' stands for hypertensive patients with advanced vascular disease, target organ damage and metabolic abnormalities. Many epidemiological studies have shown that each added risk factor greatly enhances the effect of blood pressure on mortality and morbidity. Blood pressure lowering in such patients has an overwhelming effect. In the UKPD study [4], control of blood pressure had a larger impact on outcomes than glucose control. Recent studies [5,6] have demonstrated that patients with diabetes benefit more from blood pressure lowering than non-diabetic patients. In the LIFE study [7], the advantage of losartan over atenolol was particularly well demonstrated in atherosclerotic patients with isolated systolic hypertension. Current drug comparison studies are caught in a conundrum. High-risk patients are needed to complete the study, but the higher the risk, the more impressive is the blood pressure lowering effect, and the harder it is to show subtle differences between various drugs. It is not surprising that, akin to other trials which preceded it [8–10], ALLHAT failed to find between-drugs differences in major outcomes. The results of the LIFE study are a refreshing exception and, hopefully, other forthcoming trials of angiotensin receptor blocking agents will confirm that it is possible to show differences even in patients in whom blood pressure lowering has such an overwhelming positive effect. Physiologically, there is a huge difference between mild and high-risk hypertension. Most patients with borderline hypertension have no target organ damage but, as we have shown in the Tecumseh study [11], in addition to a slight pressure elevation, they are also overweight, have higher cholesterol, triglycerides, insulin, blood sugar and haematocrit levels. Their high-density lipoprotein levels are decreased. At the very outset, hypertension is a syndrome of multiple cardiovascular risk factors and a patient's long-term outcome will depend on how these risk factors, including blood pressure, are managed and how they evolve with time. Clinically, an individual with stage 1 hypertension is middle-aged and has no visible cardiovascular complications. A patient with high-risk hypertension is old and has multiple abnormalities. In stage 1 hypertension, the goal is to prevent the development of the high-risk state; in a high-risk individual, the goal is to postpone life-threatening manifestations of an existing underlying disease. Obviously, a younger patient will be longer under the care of a physician than his high-risk counterpart. It is now clear that diuretics and β-blockers have a negative effect on insulin sensitivity [12] and it is a well proven that, compared to traditional drugs, ACE inhibitors and angiotensin blocking agents decrease the incidence of new-onset diabetes [6,13]. In the ALLHAT study, among those patients treated with chlorthalidone, the blood glucose increased from baseline to 2 years and remained increased at 4 years. At 2 and 4 years, the fasting glucose and the percentage of individuals with glucose ≥ 126 mg/dcl were significantly higher in the chlorthalidone compared to lisinopril group. A similar difference was seen between chlorthalidone and amlodipine groups at 2 years but, at 4 years, the difference was only marginally significant. A practicing physician must be aware that the ALLHAT study does not provide any clues about the management of stage 1 hypertension (i.e. the type of hypertension he is most likely to see in his practice). The question is whether he is willing to expose a younger patient to 30 years of treatment with an antihypertensive drug, which favours acceleration of cardiovascular risk. I certainly am not prepared to do so. Finally, opinion leaders should not endorse the illogical conclusion that the ALLHAT results justify a primary use of diuretics in all patients with hypertension. An outcome trial comparing various drugs in stage 1 hypertension might require huge numbers of patients, be expensive and necessitate a superb multinational organization, but it is not impossible. In this age of evidence-based medicine, we should state the obvious: that there is no evidence of how to treat mild hypertension. We need an ALLHAT-like trial in stage 1 hypertension and, accordingly, should relentlessly push for the realization of such a trial.