The Return on INVEST

A decade ago, practitioners who prescribed antihypertensive therapy, and their patients, were transfixed by a passionate and sometimes unseemly public debate over whether dihydropyridine calcium antagonists (not the verapamil type) were safe.¹ Aware that the available scientific data were insufficient, contending parties, although agreeing on little else, were all willing to submit their positions to the test of rigorous clinical trials. The missing data now have been provided.² Disappointingly, while some concerns about calcium antagonists have been laid to rest, their role in antihypertensive therapy remains controversial. Disputants who had thrived on extrapolation from observational data now disagree about how to interpret and apply the abundance of clinical trial data.

The INVEST trial,³ reported in this issue of THE JOURNAL, is the latest addition to a growing database. This trial was designed to "compare mortality and morbidity outcomes in patients with hypertension and CAD [coronary artery disease] treated with a calcium antagonist strategy [verapamil] or a non–calcium antagonist strategy [atenolol]."³ The study design reflects the importance of an active control group and equivalent achievement of blood pressure levels. The study also applied the evidence that angiotensin-converting enzyme (ACE) inhibition was beneficial for patients with renal failure, diabetes, and heart failure.⁴ The 22 576 INVEST study participants had hypertension (ie, mildly elevated systolic blood pressure) complicated by evidence of coronary artery disease, and were older than patients enrolled in most earlier, placebo controlled trials of participants with uncomplicated hypertension. Most had been treated with antihypertensive agents and approximately one third had diabetes at baseline. In sum, participants were at high global risk for cardiovascular disease events.

Patients were randomly assigned to receive verapamil followed by trandolapril, or atenolol followed by hydrochlorothiazide to achieve conventional blood pressure goals. In addition, regardless of randomization, all patients with diabetes, renal impairment, or heart failure (more than a third of those enrolled) were prescribed trandolapril. Any other agents could then be added to achieve a target systolic blood pressure of less than 140 mm Hg and diastolic blood pressure of less than 90 mm Hg (<130 and <85 mm Hg, respectively, for patients with diabetes or renal impairment).

By design, physicians and patients were aware of their treatment and able to alter it. Despite these circumstances, the investigators achieved about an 80% adherence to randomly assigned therapy throughout the study. In addition, when study and nonstudy drugs were combined, most patients in the verapamil group (82.0%) and the atenolol group (71.6%) received the ACE inhibitor. Moreover, combining study and nonstudy drugs, 63.0% and 81.4% of the 2 groups, respectively, also were prescribed a diuretic. Physician discretion played a role in drug allocation. While 35.5% of patients in the atenolol group met the requirement for trandolapril, 71.6% received an ACE inhibitor. Thus, most patients in the verapamil group received trandolapril (this was part of the study design to achieve blood pressure control), but patients in the atenolol group were given an ACE inhibitor in virtually equal measure, possibly for different reasons. Because the indications for ACE inhibition may have differed between the 2 groups, patients with different characteristics may have been selected for this therapy. As a result, the protection against bias generally afforded by blinded randomization may have been breached. In fact, as it turned out, INVEST was actually a comparison of 2 three-drug combinations. In these circumstances it is virtually impossible to draw conclusions about the contribution of any single agent.

Throughout the study, blood pressure levels were equal in the 2 groups and were less than 140 mm Hg (systolic) and 90 mm Hg (diastolic) for most patients. The primary outcome, the composite of all-cause mortality plus nonfatal stroke or myocardial infarction, was equally divided over the mean follow-up of 2.7 years, with 1119 (9.93%) events in the verapamil strategy group and 1150 (10.17%) in the atenolol strategy group (relative risk, 0.98; 95% confidence interval, 0.90-1.06). Individual event rates and results by demographic, geographic, and clinical subgroups were consistent with the overall findings. Of note, patients with diabetes were at significantly higher risk of events than those without, although treatment group did not affect outcomes. A possible interaction of drug class and incidence of new diabetes cannot be reasonably assessed in this complex, open-label study. Adverse effects were infrequent and, with expected differences, roughly equivalent in the 2 study groups.

The results of INVEST do not exist in a vacuum. Other large clinical trials⁵ and several meta-analyses^{6 - 8} uniformly support the notion that a diuretic (used by >70% of participants in INVEST) is at least equivalent in all outcomes, and superior in some, when compared with other antihypertensive classes. Similarly, in a meta-analysis (not including verapamil),⁶ ACE inhibitors have been found to be superior to calcium antagonists in preventing coronary heart disease and heart failure. In addition, a recent clinical trial in hypertensive patients with left ventricular enlargement found that losartan, an angiotensin-receptor blocker, was superior to atenolol.⁹ In short, perhaps the message is that atenolol and verapamil may be equally suitable additions for patients already taking both a diuretic and an ACE inhibitor.

In addition, the rate of cardiovascular events in INVEST was high—about 4% per year. Extrapolating from placebo controlled trials, if these patients had been denied antihypertensive therapy, roughly 30% more events would have occurred. Nevertheless, most cardiovascular disease events that might have naturally occurred happened despite conventional blood pressure control.

Better cardiovascular disease prevention could be achieved, both in studies such as this and in practice. The categorical format of INVEST, like its many antihypertensive predecessors, and as is often the case in practice, reflects a narrow and biologically inappropriate view of how to maximize cardiovascular disease prevention. The issue is not the level of any particular risk factor, but the level of risk. Global (or absolute) risk denotes the sum total of individual contributing factors (some known, some postulated, and others yet to be identified) that determine the likelihood of cardiovascular events. Global risk should be the basis for patient stratification to assess risk, and the basis for determining the potential benefit to be gained by intervention. Not all risk factors (age, for example) can be modified. But there are multiple avenues for effective intervention. Patients at high global risk, such as INVEST participants, deserve application of all the tools available—lipid lowering, aspirin, blood pressure reduction, and blockade of the renin-angiotensin system at least.¹⁰ Each of these interventions should be applied, not because a particular level of an individual risk factor has been exceeded, but because reduction of each risk factor is associated with a lower global risk of events. An important caveat to this approach, however, is the absence of direct clinical trial data in hypertension treatment to confirm the safety and benefit of systolic blood pressure lower than 140 mm Hg.¹¹

INVEST focused on blood pressure thresholds. There was substantial, but probably less than optimal blood pressure reduction, and most patients had the benefit of the renin-angiotensin system blockade. On the other hand, only slightly more than half received aspirin, and less than half were treated with a statin; this is less than optimal and is particularly surprising in a clinical trial. These observations highlight the need to abandon categorical approaches to prevention—both in research and in practice. Participants in cardiovascular disease prevention trials, or in routine treatment, should receive all the established treatments for the condition being studied. Only tools not yet proven are fair game for controlled clinical trials.

However, INVEST was not designed to address global risk, but rather sought to draw distinctions between individual antihypertensive drugs. Its strategy was not likely to be productive. Blood pressure is a physical sign that is the expressed consequence of a heterogeneous set of underlying pathophysiological phenomena. Better and more effective blood pressure treatment will likely only come through the study of pharmacological agents addressing these underlying mechanisms, rather than by searching for the universally optimal antihypertensive agent. For example, some drugs have greater effects on volume, and others on vasoconstriction. The activity of the renin system appears to monitor these dimensions of the hemodynamic milieu. Presumably, interventions targeted to those alternate mechanisms would best be compared with knowledge of the relative contribution of these mechanisms in the patients under study. For example, more precise phenotyping of participants prior to randomization might be a fruitful research strategy.^{12 - 13} More effective antihypertensive therapy is likely to flow from an ability to understand and target the underlying mechanisms responsible for the elevated blood pressure and its attendant vascular disease.

Despite the large scale and extensive follow-up of INVEST, its design, which permitted rational manipulation of therapy, and the dominating application of diuretics and ACE inhibitors, have combined to attenuate distinctions between drug therapies. Under these study conditions, it did not seem to matter whether the calcium antagonist, verapamil, or the β-blocker, atenolol, were included in a treatment regimen. Thus, when global risk assessment determines that blood pressure should be reduced, the answer remains that diuretics should usually come first, with blockade of the renin-angiotensin system a close second.^{5 ,14}