Up-Titrating Angiotensin-Converting Enzyme Inhibitors in Heart Failure: Evidence and Challenges

This article refers to ‘Similar clinical benefits from below-target and target dose enalapril in patients with heart failure in the SOLVD Treatment trial’ by P.H. Lam et al., published in this issue on pages 359–369. Angiotensin-converting enzyme inhibitors (ACE-I) reduce mortality and morbidity in patients with heart failure with reduced ejection fraction (HFrEF), whereas, at present, there are no convincing data to support their use in heart failure with preserved ejection fraction (HFpEF).1-5 In the current European Society of Cardiology (ESC) guidelines on heart failure (HF), ACE-I [and angiotensin receptor blockers (ARBs) if ACE-I are not tolerated due to cough) have an IA recommendation in patients with symptomatic and asymptomatic HF in order to reduce the risk of HF hospitalization and death.6 In clinical trials, the treatment arms were titrated toward target doses and, therefore, guidelines recommend the up-titration of ACE-I and ARBs to the ‘maximum tolerated dose’.6 However, not much is known about the incremental benefit of higher doses, whether below or above the target doses in trials. Early dose-randomized trials demonstrated greater haemodynamic and symptomatic benefits with high vs. low ACE-I doses.7-9 Later, two dose-randomized trials in HFrEF suggested greater benefit in terms of mortality/morbidity with higher dose. The ATLAS trial, randomizing patients to higher (32.5–35 mg daily) vs. dramatically lower doses (2.5–5.0 mg daily) of lisinopril resulted in a significant 12% reduced risk of the composite of death or hospitalization for any reason, mainly driven by the 24% reduction in HF hospitalization, but no effect was observed on mortality.10 Additionally, side effects (i.e. dizziness, worsening renal failure) were more frequent in patients receiving high vs. low lisinopril doses.10 In HEAAL, 150 mg vs. 50 mg daily of losartan resulted in a reduction in the composite of death or hospitalization for HF.11 In contrast, there was no effect of ACE-I up-titration in the randomized NETWORK trial, with similar death, HF hospitalization, or HF worsening rates in HFrEF patients receiving 2.5 vs. 5 mg vs. 10 mg twice daily of enalapril.12 The SOLVD trial was designed to achieve and test enalapril (20 mg daily) vs. placebo, but at the end of the study the mean enalapril dose was only 11 mg daily and more than half of the patients enrolled failed to reach the target dose of the study drug. Thus, in this issue of the Journal, Lam et al.13 hypothesize that the beneficial effects of enalapril in the SOLVD trial may not depend on the dose achieved, and report a post-hoc analysis of this trial assessing the associations between below-target (5–10 mg daily) vs. target dose (20 mg daily) enalapril and outcomes in patients with HFrEF. In particular, they show that both target dose enalapril vs. target dose placebo and below-target enalapril vs. below-target placebo achieved a reduction in the risk of all-cause mortality by 10% (absolute risk reduction of 5%), and of HF hospitalization (by 30% and 21%, respectively), leading to a reduction in the risk of the composite of mortality and HF hospitalization (by 30% and 22%, respectively). However, when target and below-target enalapril doses were directly compared, no differences in the risk of all-cause death and HF hospitalization were observed. As with any post-hoc analysis, and in particular from a trial of more than 25 years ago, the results and applicability to contemporary clinical practice should be viewed with caution. More importantly, since there was no randomization between dosing protocols, the term ‘benefit’ in the title is misleading, since the observational analysis can only prove an association. This association is exquisitely sensitive to unmeasured confounding and selection bias. Indeed, the well-established fact that healthier patients tolerate higher doses of HF medications has discouraged investigators from even attempting to assess the role of doses in trials that did not randomize doses.14 Lam et al. nevertheless did so. Patients receiving enalapril target and below-target doses were similar regarding demographic and clinical characteristics and concomitant treatments, except for blood pressure and serum creatinine levels that were higher and lower, respectively, in patients receiving high vs. low drug dose, as expected. Even after adjustments, one would expect an association between higher doses and more favourable outcomes. The fact that this was not observed does lend some support to the interpretation that dosing may not be as important as generally believed. Observational data can serve descriptive real-world purposes. However, if the aim is real-world applicability, data are better obtained from generalizable registry or large-scale cohort settings, rather than from randomized trial cohorts. These observational findings from the SOLVD trial confirm the randomized data from the NETWORK trial, which was designed to assess differences in outcome among different enalapril doses, and observed no differences in mortality and hospitalization risk between high and low drug doses.12 However, in NETWORK, only 1532 patients were treated for only 6 months; thus the limited exposure to the drug and sample size could have prevented any difference in the occurrence of clinical events among treatment groups.12 Thus, we weigh the randomized ATLAS and HEAAL trials in favour of higher doses vs. the randomized but underpowered NETWORK and now this observational post-hoc analysis from SOLVD not in favour of higher doses, and we conclude that there is no reason to change current guideline recommendations of attempting to reach target doses, since there is no clear evidence supporting that. Optimal dosing remains a critical issue in many clinical and future trial design settings. In real-world settings, there is great difficulty to reach target doses15 and side effects related to ACE-I up-titration may prevent the possibility of starting therapies that have been shown to unquestionably reduce mortality (i.e. beta-blockers, mineralocorticoid receptor antagonists, sacubitril/valsartan). When, for example, aiming for recovery after left ventricular assist device implantation, supra-target doses have been suggested to enhance reverse remodelling.16 And if higher doses are better than lower doses in ATLAS and HEAAL, why are not even higher doses better? If patients are treated with supra-target doses of ACE-I (e.g. enalapril 40 mg daily) under the non-proven assumption that higher doses are better, are they then candidates for sacubitril/valsartan, which was superior vs. enalapril but only at the dose of 20 mg daily? And if reaching target doses is limited by, for example, worsening renal function and hyperkalaemia, is there a reason to test a strategy of using novel potassium binders to enable use of higher doses of renin–angiotensin–aldosterone system inhibitors? And importantly, if doses are not relevant, this would permit busy clinicians caring for frail patients with real and perceived symptomatic side effects, or with hypotension and borderline renal function and potassium levels, to back off the aggressive up-titration of HF medications. In conclusion, the available randomized data on dosing suggest that higher doses are better and clinicians should strive for target doses as recommended and specified in guidelines.6 The present observational post-hoc analysis from SOLVD raises intriguing challenges but does not change this long-standing consensus. Conflict of interest: none declared.