Prevention of Heart Failure With Chlorthalidone in ALLHAT: Placing the Results Into Perspective

Perhaps the most intriguing finding of the Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) has been the lower rates of heart failure events observed in the chlorthalidone group compared to each of the three other treatment arms (amlodipine, lisinopril, and doxazosin). ¹ , ² Because patients with a history of heart failure or known low ejection fraction were excluded from the study, one would expect that the majority of these cases represented new‐onset heart failure. Notably, the Kaplan–Meier event curves for patients hospitalized for or dying of heart failure begin to separate immediately after randomization, with relative risk (RR) of 2.22 (95% confidence interval [CI], 1.69–2.91; P<.001) vs chlorthalidone in the first year for hospitalized or fatal heart failure with amlodipine and of 2.08 (95% CI, 1.58–2.74; P<.001) with lisinopril vs chlorthalidone. In subsequent years, there was no significant difference in the number of patients experiencing these heart failure events between lisinopril and chlorthalidone (RR, 0.96; 95% CI, 0.85–1.10; P=.58), and the RR for amlodipine vs chlorthalidone was much lower but remained statistically significant (1.27; 95% CI, 1.10–1.46; P<.001). An excess of heart failure events was also seen in the doxazosin arm, which was one of the findings that led to its early discontinuation. ²

What Is the Explanation for the Lower Heart Failure Event Rate With Chlorthalidone?

These higher rates of heart failure events in each of the nondiuretic treatment arms were unexpected, particularly in the case of lisinopril because angiotensin‐converting enzyme inhibitors have been shown to prevent the onset of heart failure and both prevent hospitalizations and prolong survival. ³ , ⁴ Several explanations have been proffered. Since the initial analyses were based on clinician judgment, one possibility was that heart failure was misdiagnosed in patients with edema and fluid retention, which are known side effects of dihydropyridine calcium blockers and alpha‐1 receptor blockers. However, an extensive validation study that utilized adjudication criteria that excluded edema confirmed the lower rates of heart failure events in the chlorthalidone group. ²

Since 90% of the ALLHAT patients had been taking antihypertensive medication prior to entry, another plausible explanation was that these findings may have been influenced by their prior treatment or its withdrawal at the time of study drug initiation. In particular, many patients were taking diuretics at doses that may have been sufficient to treat previously undiagnosed heart failure, which then became evident when they were discontinued. The marked early excess of heart failure events in the nondiuretic groups is consistent with this explanation. However, the extensive analyses of the relation of heart failure events to patients’ preceding antihypertensive therapy by Grimm and colleagues ⁵ in this issue of The Journal of Clinical Hypertension go a long way toward refuting this hypothesis. Although the lack of data on prior medications in 37% of patients is a limitation, the relatively small proportion (39%) of patients previously receiving diuretics argues against this being a dominant explanation. The lack of significant interaction between pretreatment diuretic use and 1‐year heart failure outcomes is also quite convincing, although the small numbers of events in each treatment group suggest that these interaction analyses may not be adequately powered to detect some effect on the subsequent appearance of heart failure.

What, then, is the explanation for the lower rate of heart failure in the chlorthalidone group? Several possibilities come to mind. First, although lisinopril may have been anticipated to be the most effective agent in preventing heart failure, as noted above it was associated with a higher rate than chlorthalidone, particularly in the first year. ¹ , ⁵ A possible explanation for this was the lesser reduction in both systolic and diastolic blood pressure, particularly in this initial treatment period. This probably was the result of the ALLHAT protocol, which specified use of atenolol, an agent with substantial mechanistic overlap and less additive blood pressure reduction when combined with angiotensin‐converting enzyme inhibitors. In high‐risk hypertensive patients, higher blood pressure, especially systolic pressure, is associated with a higher rate of heart failure. ⁴ Similarly, both amlodipine and doxazosin produced lesser systolic blood pressure reductions. ⁶ , ⁷

Calcium channel blockers, particularly dihydropyridines, have been associated with higher rates of heart failure in several trials. ⁸ A likely explanation for this is the fluid retention that is associated with these agents. In this context, it is noteworthy that patients with preserved ejection fractions may be more sensitive to fluid retention as a trigger for decompensation, and these constituted the majority of patients in whom heart failure developed in ALLHAT. ⁹ Alpha‐1 receptor antagonists such as doxazosin also cause fluid retention, and doxazosin was associated with the greatest excess of new‐onset heart failure. ⁷

Putting the ALLHAT Results Into Perspective

Much has been written concerning the lower rate of heart failure with chlorthalidone as the first‐line agent in ALLHAT. ¹⁰ Certainly, heart failure is an important outcome that should be avoided if possible. Indeed, patients in whom heart failure developed in ALLHAT had substantially higher subsequent mortality rates. ⁹ However, the occurrence of heart failure was not the primary or even a secondary end point in ALLHAT, but rather one of several components of the “combined cardiovascular disease” secondary outcome, itself the third of 7 secondary outcomes and indeed only 1 of 7 components of this end point. This is not meant to deny the importance of heart failure events in ALLHAT, but an important principle in designing and interpreting clinical trials is to avoid the potential for false‐positive findings by performing multiple analyses, particularly in a post hoc manner. Current practice, especially in the regulatory environment, is to consider positive secondary end points in a trial that has not achieved its primary end point as hypothesis‐generating rather than definitive. ALLHAT did not demonstrate superiority of chlorthalidone over the other treatments for its primary outcome of fatal coronary heart disease (CHD) or nonfatal myocardial infarction. Nor did it demonstrate superiority for its first 2 secondary end points: all‐cause mortality and combined CHD (fatal CHD, nonfatal myocardial infarction, coronary revascularization, or hospitalized angina). Chlorthalidone did reduce stroke, the third‐ranked secondary end point, compared with lisinopril but not compared with amlodipine, which was associated with a nonsignificant trend toward a lower stroke rate than chlorthalidone.

In contrast to the comparisons with amlodipine and lisinopril, chlorthalidone was clearly superior to doxazosin in preventing major outcomes including cardiovascular mortality, stroke, heart failure events, hospitalized angina, and coronary revascularization. ⁷

What Is the Take‐Home Message of ALLHAT?

ALLHAT is a critically important trial. It has clearly demonstrated that one thiazide‐type diuretic, chlorthalidone, is a very effective first‐line antihypertensive agent and ideally should reverse the historical mistake of blaming diuretics for the lack of a positive outcome in the Multiple Risk Factor Intervention Trial (MRFIT), which was attributed to chlorthalidone (used in dosages up to 100 mg/d rather than the 25‐ or 50‐mg doses in ALLHAT) based on an ill‐conceived post hoc analysis in a small subgroup of patients. ¹¹ However, despite the most recent report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure recommending thiazide diuretics as the initial first‐line agent or part of a first‐line combination regimen for most hypertensive patients, ¹² arguments by some leading hypertension experts, ⁸ , ¹⁰ and the apparent but modest impact of the ALLHAT results on patterns of first line antihypertensive medication use, ¹³ it is unlikely that diuretics will return to their earlier place as the “standard” first‐line agent for hypertension in either the general population or in high‐risk patients such as those enrolled in ALLHAT. Furthermore, to the extent that diuretics are used as initial therapy, often the agents and doses commonly employed have not been demonstrated to improve major outcomes.

While many will, with some justification, attribute the decline of diuretic use to the successful marketing of more expensive agents that have become available over the past 25 years, including calcium channel blockers, angiotensin‐converting enzyme inhibitors, and angiotensin receptor blockers, there are strong arguments for the use of more individualized selection. Even more importantly, recent trials have provided convincing evidence that combinations including these newer agents can have more favorable effects on major outcomes than diuretic‐based regimens. In the Anglo‐Scandinavian Cardiac Outcomes Trial (ASCOT), ¹⁴ the combination of amlodipine and perindopril significantly reduced several major outcomes compared to the combination of bendroflumethiazide and atenolol. These outcomes included fatal and nonfatal myocardial infarction, all‐cause and cardiovascular mortality, and fatal and nonfatal stroke. Fatal and nonfatal heart failure events were modestly decreased. More recently, the Avoiding Cardiovascular Events Through Combination Therapy in Patients Living With Systolic Hypertension (ACCOMPLISH) trial demonstrated that the combination of amlodipine and benazepril reduced a variety of cardiovascular end points when compared with hydrochlorothiazide plus benazepril. ¹⁵ These include the primary end point of cardiovascular death and a composite of other cardiovascular events, but most importantly, a 21% reduction in the composite of cardiovascular death, nonfatal myocardial infarction, and nonfatal stroke, despite the fact that excellent and equivalent blood pressure control was achieved in both groups. Importantly, and in distinction from the amlodipine and lisinopril arms of ALLHAT, the amlodipine/benazepril arm did not display an excess of heart failure hospitalizations while producing a tremendous reduction in irreversible cardiovascular end points.

ALLHAT investigators and other experts have argued that the ACCOMPLISH results reflect the use of the wrong diuretic in the wrong dose (hydrochlorothiazide 12.5 mg titrated to 25 mg). ¹⁰ This is a valid concern, since unlike chlorthalidone, hydrochlorothiazide has not yielded major outcome benefits in large outcome trials. However, this message has been confounded by the frequent substitution of the class “thiazide diuretics” for chlorthalidone. Perhaps this reflects the recognition that few physicians are familiar with chlorthalidone and very few use it. In North America, most combination antihypertensives include hydrochlorothiazide as the diuretic component. In other regions, bendroflumethiazide is popular. Importantly, all of the agents involved in these trials are or soon will become available as generics, so price differences need not preclude the use of newer agents that have proven effective in clinical trials.

Thus, rather than continue to argue about which pharmacologic classes should be considered first‐line therapy antihypertensive agents, we should celebrate our increasing ability to prevent the major outcomes afflicting the growing hypertensive population and focus on identifying and treating them.

Disclosures:  ALLHAT investigator and member of the heart failure writing group. Consultant to Novartis Pharmaceuticals and member of the ACCOMPLISH Steering Committee.