PARADIGM-HF: Does Dose Matter?

This article refers to ‘Efficacy of sacubitril/valsartan vs. enalapril at lower than target doses in heart failure with reduced ejection fraction: the PARADIGM-HF trial’ by O. Vardeny et al., published in this issue on pages 1228–1234. The PARADIGM-HF (Prospective Comparison of ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure) trial showed unequivocal superiority of sacubitril–valsartan (target dose 200 mg twice daily) over enalapril (target dose 10 mg twice daily) in clinically stable patients with heart failure, a reduced ejection fraction (HFrEF) and mostly mild symptoms who were initially able to tolerate the protocol-specified target doses of each agent.1 The trial was designed to satisfy regulators' requirements and was not meant to be a practical blueprint for a clinical care plan. Like most clinical trials, it should be regarded as an aid to clinical judgement rather than a precise set of instructions. Clinicians must extrapolate evidence from clinical trials to individual patients and judge what is in their patients' best interests rather than merely decide whether the patient fulfils the trial entry criteria. A very conservative stance on dose, as in recent European Society of Cardiology (ESC) guidelines on sacubitril/valsartan,2 may apply to rather few patients; a more liberal stance, as suggested by the European Medicines Agency or National Institute for Health & Care Excellence may apply to many.3 Who is right? Does dose matter? PARADIGM-HF required all patients to tolerate pre-specified doses of both medications. It did not set out to prove that the target dose of either agent was optimal. Indeed, the concept of a single universal target dose appropriate for all patients is implausible. Plasma concentrations of medicines will depend on absorption, body size and composition, and the rate of clearance by the kidney, liver, and other mechanisms, and not just the dose ingested. For many patients, higher doses will be tolerated and might be more effective; for others they will not be tolerated and might be deleterious. Lower doses will usually be better tolerated in the short term but might offer less long-term protection from disease progression.4 Identifying the optimal dose for an individual patient is difficult. Clearly a zero-dose must be ineffective, although not necessarily suboptimal for some individuals, and there will be a dose above which the medicine is not tolerated or lethal. Randomized clinical trials comparing higher with lower doses of ACE inhibitors5 or ARBs6 failed to show substantial differences, although perhaps because higher doses were poorly tolerated amongst the sickest patients (Figure 1). This confounds analysis by intention-to-treat because there may be little difference in the dose actually ingested amongst the sickest patients who are most likely to have events. Also, dose comparison trials recruit patients younger and with fewer co-morbidities than in clinical practice and often exclude those who are unable to tolerate substantial doses of medication. With beta-blockers or ivabradine, heart rate achieved,7 and, for mineralocorticoid antagonists (MRAs), serum potassium8 may be better targets than guideline-recommended doses; other biomarkers, including the ratio of biologically active to propeptide fragments of natriuretic peptides for sacubitril–valsartan, might also provide useful guidance. It is sobering to think that after 200 years of digoxin and 50 years of cardiovascular aspirin we are still uncertain of the optimal dose of either, which could still prove to be zero. Optimal dosing is difficult to study and to achieve. In a post-hoc analysis, Vardeny and colleagues9 now show that many patients who had initially tolerated target doses of each agent in PARADIGM-HF subsequently had doses reduced either temporarily or permanently; some had to discontinue medication altogether. Information on adherence, which is difficult to measure robustly, is not reported, but a proportion of patients will have taken substantially less medication than prescribed, even amongst those who down-titrated their dose.10 Although reductions in dose were associated with a worse outcome regardless of assigned therapy, the superiority of sacubitril–valsartan over enalapril persisted in such patients. This analysis was not pre-specified, dose reductions will not have occurred at random, and the reasons for dose reduction probably differed between agents. Why were differences still observed in the subpopulation taking lower doses? The obvious explanation is that even at lower doses, sacubitril–valsartan is superior to enalapril. Alternatively, there may have been a persisting ‘legacy effect’ from a period of treatment at higher doses.11 Intolerance to enalapril might also identify a sicker population than does intolerance to sacubitril–valsartan. Accordingly, this analysis should be extrapolated with caution to other clinical populations or settings. Patients with heart failure who are unable to tolerate target doses of disease-modifying therapies usually have a worse prognosis,12 as confirmed in PARADIGM-HF. This may be because low blood pressure and worse renal function lead to a lethal therapeutic triad (Figure 1). High doses of medication will be less likely to be prescribed for, or tolerated by, patients with low blood pressure or glomerular filtration rate (GFR), who have a worse prognosis; reduction or cessation of therapy may lead to loss of therapeutic benefit; excessive lowering of blood pressure or GFR may neutralize or reverse the benefits of the therapy, if taken. Despite previous failures with other agents, further randomized controlled trials should be considered to determine the optimal dose of sacubitril–valsartan. Lower doses might have fewer side effects and may retain full therapeutic potency and could even be superior to higher doses if they exert less reduction in blood pressure. Of course, lower doses might be less effective and higher doses than those used in PARADIGM-HF might be even better. Ultimately, a particular dose of sacubitril–valsartan will not be optimal for all patients, which makes the design of studies comparing doses difficult to do and to interpret. PARADIGM-HF also does not exclude the possibility that higher doses of ACE inhibitors would have been as effective as sacubitril–valsartan. In the ATLAS (Assessment of Treatment with Lisinopril and Survival) trial, those with milder symptoms who were more likely to remain on assigned therapy (placebo or 30 mg of lisinopril on an open-label background of 2.5–5.0 mg of lisinopril), higher doses were associated with a significantly lower mortality.5 This was a post-hoc, subgroup analysis and therefore does not constitute robust evidence of greater efficacy with higher doses of ACE inhibitors. Recently, intense inhibition of the renin–angiotensin–aldosterone system (RAAS) with a combination of enalapril and aliskiren (a renin inhibitor) on a background of substantial MRA use was shown not to be superior to enalapril alone at a dose of 10 mg b.i.d.,13 providing strong, contemporary evidence of the limits of intense RAAS inhibition. However, discussions about the hypothetical outcomes of hypothetical trials are of little practical value to the patient or doctor faced with making a decision today. It will take years before the results of any dose-ranging study could be known. Who would fund such a trial? The obvious answer is health services, since if they are willing to pay for a medication then they should ensure that they are using it to best effect. In the absence of such information, this analysis of PARADIGM-HF9 provides the best available information to guide decisions. Does dose matter? One dose of any effective medicine certainly matters and that dose is ‘zero’. Avoid it whenever possible! Educate individual patients about the doses of medication you think might be best for them. Maybe once they know what the targets are, they might help you achieve them. After all, it's in their own interests. Conflict of interest: The University of Hull has received research grants from Novartis, and J.