Toleration of High Doses of Angiotensin-Converting Enzyme Inhibitors in Patients With Chronic Heart Failure: Results From the ATLAS Trial

Background  Treatment with angiotensin-converting enzyme (ACE) inhibitors reduces mortality and morbidity in patients with chronic heart failure (CHF), but most affected patients are not receiving these agents or are being treated with doses lower than those found to be efficacious in trials, primarily because of concerns about the safety and tolerability of these agents, especially at the recommended doses. The present study examines the safety and tolerability of high- compared with low-dose lisinopril in CHF.

Methods  The Assessment of Lisinopril and Survival study was a multicenter, randomized, double-blind trial in which patients with or without previous ACE inhibitor treatment were stabilized receiving medium-dose lisinopril (12.5 or 15.0 mg once daily [OD]) for 2 to 4 weeks and then randomized to high- (35.0 or 32.5 mg OD) or low-dose (5.0 or 2.5 mg OD) groups. Patients with New York Heart Association classes II to IV CHF and left ventricular ejection fractions of no greater than 0.30 (n = 3164) were randomized and followed up for a median of 46 months. We examined the occurrence of adverse events and the need for discontinuation and dose reduction during treatment, with a focus on hypotension and renal dysfunction.

Results  Of 405 patients not previously receiving an ACE inhibitor, doses in only 4.2% could not be titrated to the medium doses required for randomization because of symptoms possibly related to hypotension (2.0%) or because of renal dysfunction or hyperkalemia (2.3%). Doses in more than 90% of randomized patients in the high- and low-dose groups were titrated to their assigned target, and the mean doses of blinded medication in both groups remained similar throughout the study. Withdrawals occurred in 27.1% of the high- and 30.7% of the low-dose groups. Subgroups presumed to be at higher risk for ACE inhibitor intolerance (blood pressure, <120 mm Hg; creatinine, ≥132.6 µmol/L [≥1.5 mg/dL]; age, ≥70 years; and patients with diabetes) generally tolerated the high-dose strategy.

Conclusions  These findings demonstrate that ACE inhibitor therapy in most patients with CHF can be successfully titrated to and maintained at high doses, and that more aggressive use of these agents is warranted.

DURING THE past 15 years, a series of studies and large clinical trials have established the efficacy of angiotensin-converting enzyme (ACE) inhibitors in patients with chronic heart failure (CHF) and left ventricular systolic dysfunction,^1-3 as well as in those with asymptomatic⁴ and postmyocardial infarction left ventricular systolic dysfunction.^5,6 These trials demonstrated that ACE inhibitors prolong survival, prevent hospitalizations, improve clinical status, and prevent progressive left ventricular dysfunction. Despite the overwhelming evidence supporting the routine administration of ACE inhibitors and recommendations by professional organizations and governmental authorities for their use,^7-10 a substantial number of appropriate patients remain untreated,^11-16 and more are receiving low doses of ACE inhibitors that have never been demonstrated to be effective in clinical studies.^13,16-18 The most frequently observed and cited reasons for not using ACE inhibitors in general, and the higher doses used in clinical trials in particular, are concerns about excessive reductions in blood pressure, renal dysfunction, and safety in older patients.^{12,13,15,16,19}

The Assessment of Treatment with Lisinopril and Survival (ATLAS) trial demonstrated that treatment with high doses of the ACE inhibitor lisinopril (target dose, 32.5-35.0 mg once daily [OD]), when compared with the lower doses often prescribed (2.5-5.0 mg OD), was associated with a trend toward reduced mortality (−7.9%; 96.1% confidence interval [CI], −17.6% to 3.0%; P = .13) and a significantly lower risk for death or hospitalization for any cause (−11.6%; 95% CI, −4.5% to −18.2%; P = .002).²⁰ The present study addresses in depth the second critical dimension of this issue, ie, whether a strategy of using these high doses can be implemented with acceptable safety and tolerability.

The ATLAS trial commenced recruitment on October 1, 1992 and terminated on September 15, 1997. Patients were recruited from 287 centers in 19 countries in North America, Europe, and Australia.^20,21 The primary inclusion criteria for this trial were the presence of symptomatic CHF (New York Heart Association class II-IV) with diuretic treatment for at least 2 months and a left ventricular ejection fraction (EF) of no greater than 0.30. To qualify for enrollment, patients with class II symptoms must have had a hospitalization or emergency department visit for decompensated CHF within the previous 6 months. Patients of all ages were eligible whether or not they were receiving ACE inhibitors at the time of entry. Other medications for heart failure, including digitalis, vasodilators, and β-blockers, were permitted but not required.

Major exclusion criteria included myocardial infarction, unstable angina, or cardiac surgery within the previous 2 months; clinical instability as indicated by treatment with intravenous positive inotropic drugs or mechanical or ventilatory assistance within 48 hours; current listing for cardiac transplantation; symptomatic hypotension; serum creatinine level of greater than 221.0 µmol/L (2.5 mg/dL); or the presence of a noncardiac condition with an expected survival shorter than the study period. In addition, patients who were receiving long-term treatment with nonsteroidal anti-inflammatory agents (other than aspirin) or who had known intolerance of ACE inhibitors were excluded.

The protocol was approved by the ethics committees of all participating centers, and written informed consent was obtained from all patients.

The study design is illustrated in Figure 1 and consisted of an open-label lisinopril tolerability phase of 4 weeks, a 4-week double-blind titration period, and a double-blind chronic maintenance phase. Patients who were not taking ACE inhibitors at the time of entry started open-label lisinopril therapy at a dosage of 2.5 or 5.0 mg OD for 2 weeks. The low dose was selected by the investigator as the smallest tablet size available in their country that was considered effective; it was 5.0 mg in 85% of patients. If the low dose was tolerated, an additional 10 mg of open-label lisinopril was added for 2 more weeks. In patients previously receiving maintainence therapy using an ACE inhibitor, lisinopril dosage was converted to 12.5 or 15.0 mg (the lowest dose considered effective plus 10 mg) and maintained for 4 weeks. Blood pressure and renal function were monitored at 2 and 4 weeks thereafter.

After this 4-week, open-label tolerability phase, patients were randomized equally to low- (LD) or high-dose (HD) groups. The 5.0- or 2.5-mg minimal dose of open-label lisinopril was continued in all patients. The 10-mg open-label lisinopril dose was discontinued and replaced with 2 double-blind 10-mg tablets of active lisinopril or placebo. If tolerated, a third 10-mg tablet of active drug or placebo was added after 2 weeks, providing a total active dose of 2.5 or 5.0 mg in the LD group and 32.5 or 35.0 mg in the HD group. These target doses were continued for the remainder of the study, although doses of blinded medication could be adjusted at the investigator's discretion. Blood pressure and renal function were monitored after 2 and 4 weeks of this titration phase and every 3 to 6 months during the maintenance phase.

The primary efficacy end point was all-cause mortality; the major secondary end point was combined risk for death or hospitalization for any reason, both analyzed as time to first event. These findings are described in detail elsewhere.²⁰

The primary safety and tolerability analyses focused on changes in blood pressure and related symptoms and on alterations in renal function and serum potassium levels that are generally considered dose related. For this purpose, in addition to hypotension and postural hypotension, the adverse events reported as dizziness, vasodilatation, syncope, and shock were grouped together as a single category, hypotension/dizziness. Adverse events considered probably or possibly related to the renal effects of ACE inhibitors included kidney function abnormality, kidney failure, uremia, increased creatinine level, increased nonprotein nitrogen level, and hyperkalemia; these were grouped together as renal dysfunction/hyperkalemia. In addition, complete information was collected on other side effects and adverse events.

These safety data, as well as the primary efficacy end points, were monitored periodically by an independent Data and Safety Monitoring Board, but this board did not find it necessary to modify the conduct of the study, and the protocol was completed as planned when the last patient randomized into the study had been followed up for a minimum of 36 months.

A total of 3793 patients signed consent forms and entered the trial, and their disposition is shown in Figure 2. Of these, 3178 were receiving an ACE inhibitor at baseline, 405 were not, and previous ACE inhibitor treatment was not recorded in the remaining 210 (who were not randomized and therefore did not have complete baseline information). Of the 3793 patients who entered, 3164 eventually underwent randomization and 629 did not. The largest subgroup of nonrandomized patients (209) were individuals who signed consent to have an EF determination but did not have a qualifying EF measurement. These patients also constituted most of the subjects in whom previous ACE inhibitor treatment status was not known. The remainder were not randomized for a variety of administrative reasons (62 protocol deviations; 55 with <80% medication compliance during the tolerability phase; 43 with consent withdrawn; and 29 others) or because of adverse events (231 patients) that will be discussed below.

Table 1 presents the baseline characteristics of the 3164 patients who entered the trial and underwent randomization and the 420 patients with qualifying EF values who received lisinopril during the open-label tolerability phase but did not undergo randomization. The randomized and nonrandomized patients were similar in most of their baseline characteristics, with the exception that a higher proportion of the nonrandomized group were in New York Heart Association class IV (14.0% vs 7.1%). Patients randomized to HD and LD groups did not differ in any baseline characteristic.

The prerandomization titration and stabilization period provides information about the tolerability of low- to medium-dose lisinopril in the 405 previously ACE-naive subjects and in the 3178 patients who had received previous ACE inhibitor treatment, most of whom (approximately 75%) received relatively low doses (eg, total daily doses of lisinopril, ≤10 mg; of enalapril maleate, ≤10 mg; of captopril, ≤75 mg; or the equivalent). A total of 5 patients (all previously received an ACE inhibitor) discontinued due to rash, angioedema, or unspecified allergic reactions. Ten (5 who had and 5 who had not previously used ACE inhibitors) discontinued due to cough. A total of 111 patients (3.1%) discontinued due to adverse reactions in the categories of hypotension/dizziness and renal dysfunction/hyperkalemia that were considered possibly related to ACE inhibitor therapy (Table 2). For the patients who did not receive an ACE inhibitor before entry, these adverse reactions are subdivided into whether they occurred at the first or second open-label dose. For either class of adverse effects, 4.2% of patients who did not receive a previous ACE inhibitor (1.2% at the 2.5- or 5.0-mg dose level and 3.1% at the 12.5- or 15.0-mg dose level) and 3.0% of patients who did withdrew because of such events. The reasons were divided approximately equally between effects possibly related to both categories of adverse effects. These figures provide a maximum estimate of drug-related discontinuations in these categories, since there was no concomitant placebo-treated control group to provide an indication of the frequency of discontinuations for these reasons unrelated to ACE inhibitor therapy.

Table 3 presents information on the mean dose of randomized blinded medication taken by patients in the LD and HD groups and the proportions of patients in each group who continued to receive the target dose and who discontinued the study drug. Lisinopril or placebo doses in more than 90% of patients in both groups were titrated to the target of 35.0 or 32.5 mg. Thereafter, although the mean dose of study medication (blinded lisinopril or blinded placebo) declined gradually in both groups, the mean dose of blinded medication and the proportion of patients receiving the target dose remained essentially identical in both groups throughout the study. Indeed, the proportion of patients not taking randomized drug tended to be lower in the HD group. Over time, a growing number of patients in both groups received open-label ACE inhibitors instead of or in addition to the blinded drug. This in large part was the result of hospitalizations for worsening CHF, and occurred somewhat more frequently in patients randomized to the LD lisinopril group (35.2% vs 29.7% in the HD group; P = .002).

Figure 3 illustrates the mean values for blood pressure and serum creatinine level in the 3164 randomized patients from the time of study entry (prebaseline) to the 4-year follow-up point. The mean difference in systolic and diastolic blood pressure between the LD and HD lisinopril groups was 3 to 4 mm Hg. Serum creatinine levels differed by approximately 8.8 µmol/L (0.1 mg/dL) between the groups.

Adverse events reported as serious occurred in 83.3% of the LD and 79.0% of the HD groups. Most of these were in the category of worsening heart failure, coronary ischemia, sudden death, or arrhythmias, rather than known adverse effects of ACE inhibitors. Withdrawals due to adverse events occurred in 17.0% of the HD and 18.0% of the LD groups. Table 4 lists the most frequent adverse events that were classified by the investigator as serious or that led to withdrawal of blinded study medication. Overall, these occurred in similar proportions of both groups and were dominated by events that appear to be more closely related to the patients' underlying cardiac conditions than to ACE inhibitor therapy. The low incidence of cough, which probably reflected most patients having been exposed previously to ACE inhibitors or at least tolerating doses of up to 12.5 or 15.0 mg of lisinopril, did not appear to be increased at higher dosages.

The potentially dose-related adverse effects of hypotension/dizziness and renal dysfunction/hyperkalemia occurred somewhat more frequently in the HD group, but the number of patients in whom study drug was withdrawn for these adverse effects was small and quite similar in both groups. Furthermore, many of these adverse effects did not lead to a change in the dose of the study medication. Thus, for dizziness, 32.3% and 26.1% of the HD and LD groups, respectively, had dose adjustments; after hypotension, these figures were 56.7% and 46.3%; for rising creatinine levels, they were 49.2% and 43.2%; and for hyperkalemia, they were 44.3% and 33.3%.

Table 5 presents data on the number of patients in several subgroups often thought to be at higher risk for dose-related adverse effects of ACE inhibitors, including patients with prerandomization systolic blood pressure of less than 120 mm Hg, patients with prerandomization serum creatinine level of at least 132.6 µmol/L (1.5 mg/dL), patients aged at least 70 years at randomization, and patients with diabetes requiring hypoglycemic therapy. These high-risk subgroups tended to have a higher frequency of hypotension/dizziness in the HD and LD groups, but the relative increase in serious events appeared to be greater in the HD group. Nonetheless, even in these 4 high-risk groups, the absolute increase in the number of events classified as serious was generally less than 5% in the HD compared with the LD group, and the absolute increase in the frequency of withdrawal was not more than 2.5% in the HD compared with the LD group.

The incidence of renal dysfunction/hyperkalemia was also somewhat higher in patients in these high-risk subgroups in both treatment groups, but the occurrence of adverse effects that were classified as serious or requiring withdrawal were not consistently higher in the HD lisinopril group compared with the LD group, suggesting that ACE inhibitor–induced renal dysfunction may be more related to patient factors than ACE inhibitor dose.

Although more than a decade has passed since the first trials demonstrating that ACE inhibitors prolong survival¹ and improve symptoms in patients with CHF,²² physicians continue to underuse these agents.^11-16 In some studies, up to 50% of patients considered appropriate for ACE inhibitor therapy are not receiving these drugs, and most of those receiving ACE inhibitors are receiving doses that have never been shown to be effective in clinical trials. The most frequently cited reasons for these practices are concern about the possibility of adverse effects related to hypotension and renal dysfunction.^12,13,15-17

To our knowledge, the ATLAS trial provides the first large experience that addresses these concerns directly. The ATLAS trial showed that patients whose lisinopril dose was titrated to a daily target of 32.5 or 35.0 mg exhibited a trend toward improved survival and highly significant reductions in the end point of death or hospitalization due to any cause and in hospitalizations for heart failure. The results of the present safety and tolerability analyses should serve to reassure physicians that even the high doses of lisinopril used in this trial are generally well tolerated.

This conclusion is supported by a number of findings. First, there was no between-group difference in the proportion of patients in whom, once having demonstrated tolerability of medium doses of lisinopril, the study drug doses could be titrated and maintained on the target. Mean doses of the blinded study drugs were nearly identical throughout the trial, indicating that dose reductions were not deemed more necessary by the investigators for patients receiving the HD than in those given the LD therapy. The proportion of patients withdrawing from therapy was, in fact, higher in the LD than in the HD group (27.1% vs 30.7%, respectively). Second, for the group as a whole, the mean differences in blood pressure and renal function between patients randomized to the HD and LD groups were quite small. Third, although the occurrence of adverse events possibly related to hypotension or renal dysfunction (as defined previously) was greater in the HD group, these differences were relatively small for events classified as serious or those that led to withdrawal. Last, even when patients usually considered to be at particularly high risk for these complications (eg, older patients or those with lower pretreatment blood pressures, elevated serum creatinine levels, or diabetes) were examined, the occurrence of serious events was relatively low. Indeed, serious renal dysfunction did not appear to be dose related, and hypotension rarely required withdrawal of lisinopril therapy, although dose reductions were frequent.

One important limitation to these safety and tolerability analyses should be emphasized. Of the 3584 patients who entered the baseline phase with qualifying EF values, 88.8% were already taking ACE inhibitors, although most, around 75%, were receiving relatively low total daily doses. The proportions of patients in the participating institutions in whom ACE inhibitor therapy previously had been initiated and subsequently withdrawn because of adverse effects is unknown, but generally only a small percentage of patients with CHF do not tolerate some dose of an ACE inhibitor. More helpful is the experience of the 405 patients who had not previously received ACE inhibitors and who were entered in the ATLAS trial. Only 4.2% of these did not enter the trial because of hypotension or renal dysfunction, and this figure may be an overestimate of the number attributable to the effects of the drug, since there was no corresponding control group to provide an estimate of the spontaneous occurrence of these adverse events. These observations should reassure physicians who are concerned about hypotension-related and renal adverse events with ACE inhibitors.

Patients entering clinical trials often differ from the overall patient population in being more compliant and motivated and in having fewer serious comorbid conditions. In addition, supervision during the trial may be closer than that in ordinary practice, although in this study visits occurred only at 3-month intervals after titration upward, and laboratory tests were mandated after 3 months and then yearly—an intensity of follow-up quite consistent with usual practice. Therefore, it is possible that similar degrees of tolerability may not be achieved in the nontrial setting.

These findings should not be interpreted as indicating that all patients with CHF will tolerate high doses of ACE inhibitors. Rather, ATLAS results demonstrate that a strategy of titrating to and maintaining higher ACE inhibitor doses than are used currently in clinical practice will be successful in most patients and certainly in most of those who tolerate low doses. This is the case even in patients with lower blood pressure, baseline increases in serum creatinine levels, and diabetes and who are aged at least 70 years, each of which is associated with less use of ACE inhibitors in clinical practice. Although hypotension will occur in some patients, this is usually not serious and generally does not require withdrawal of the ACE inhibitor therapy. This new information on ACE inhibitor tolerability, coupled with the lack of evidence that low doses of ACE inhibitors improve morbidity and mortality and with the additional reduction in the composite of death and hospitalizations at the high doses used in ATLAS, provide a clear message that higher ACE inhibitor doses can and should be used in most patients.

Accepted for publication July 20, 2000.

The ATLAS trial study was supported by a grant from Zeneca Pharmaceuticals, Macclesfield, England.

Corresponding author: Barry M. Massie, MD, Cardiology Division (111C), Veterans Affairs Hospital, 4150 Clement St, San Francisco, CA 94121 (e-mail: Barry.Massie@med.va.gov). Reprints are not available from the authors.