Ivabradine for heart failure: regulatory differences in Europe and United States

Ivabradine, an inhibitor of the cardiac pacemaker I_f current that slows the heart rate, is approved by both the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) for the treatment of heart failure (HF). However, the approved prescribing information differs slightly between the two agencies. The FDA¹ restricts the indication to patients with a resting heart rate ≥ 70 b.p.m. while the EMA² restricts ivabradine to patients whose heart rate is ≥75 b.p.m. The FDA adopted the entry criterion for the pivotal SHIFT trial supporting approval while the EMA based its indication on the greater reductions in endpoints in the subgroup with the higher heart rate, noting also that the latter subgroup also had a safety profile similar to the overall safety profile. The FDA summary review of ivabradine includes a series of unpublished quintile analyses that show the impact of heart rate upon outcomes.³ SHIFT randomized 6558 HF patients with left ventricular ejection fraction ≤ 0.35 and heart rate ≥ 70 to ivabradine or placebo.⁴ The SHIFT patients otherwise received standard care including ACE inhibitors, angiotensin receptor blockers, mineralocorticoid receptor antagonists, beta blockers, and loop diuretics. SHIFT was successful for its primary endpoint, a composite of cardiovascular death and HF hospitalization [hazard ratio 0.82, 95% confidence interval (CI) 0.75–0.90, P < 0.0001]. Figure 1 shows outcomes by baseline heart rate quintiles for the SHIFT trial and pertinent subgroups. Apparently, the benefit for all patients appeared to be predominantly at the higher baseline heart rates for HF hospitalization. Furthermore, the FDA analyses suggest that other factors, including ischaemic aetiology and loop diuretic use, interact with heart rate. For non-ischaemic aetiology (Figure 1B), the evidence suggests that the HF hospitalization benefit is independent of heart rate while the cardiovascular death benefit remains uncertain. For ischaemic aetiology loop, diuretic use appears to play a role: for ischaemic patients not receiving a loop diuretic at baseline (Figure 1C) suggests that the hospitalization benefit is uncertain and cardiovascular mortality may be worsened; for ischaemic patients receiving a loop diuretic at baseline (Figure 1D) suggests that the benefits are for both cardiovascular death and hospitalizations at the higher heart rates. The detailed analyses regarding loop diuretics are included in the FDA review³ and will be reported elsewhere. These latter reports also confirm that loop diuretic use, rather than HF severity (which may be associated with loop diuretic use), interacts with ivabradine because neither left ventricular ejection fraction nor New York Class Association class interacts significantly with it. Based on these analyses, the FDA reviewer proposed an indication differentiated by aetiology. For patients with non-ischaemic aetiology HF, he proposed using the SHIFT criterion of heart rate > 70. However, for patients with ischaemic aetiology HF, he proposed ivabradine use only in patients with heart rate ≥ 75 and taking a loop diuretic. For the SHIFT trial population, he calculated the various endpoint risk ratios for the indicated and excluded subgroups for his proposal compared to the EMA’s simpler exclusion criterion of heart rate ≥ 75 alone. We show the risk ratios in Figure 2.

Figure 1

Outcomes by baseline heart rate quintile in the SHIFT trial. CI, confidence interval; CV, cardiovascular; HF, heart failure; HR, heart rate; RR, risk ratio (ivabradine/placebo).

Open in new tabDownload slide

Figure 2

Endpoint risk ratios for the FDA reviewer and EMA indicated and excluded subgroups in the SHIFT trial. CI, confidence interval; CVD, cardiovascular death; died, all-cause mortality; HFH, heart failure hospitalization; PEP, primary endpoint (CVD and HFH), RR, risk ratio (ivabradine/placebo).

Open in new tabDownload slide

While for the SHIFT trial populations both indication criteria discriminate well between the subgroups that benefited from ivabradine use and those that did not, the FDA reviewer indication discriminates better with greater benefit in the indicated subgroup and greater detriment in the excluded subgroup. The critical question for these analyses is whether we should accept subgroup analyses of clinical trials. The EMA, for SHIFT, accepted the subgroup analysis that patients with heart rate ≥ 75 fare better, and hence the latter criterion is more appropriate for the indication than the SHIFT entry criterion of heart rate ≥ 70. We agree with the EMA’s judgement, justified by the above analyses and some similar analyses performed by the SHIFT investigators and others.^5–7 We suggest that, whenever a criterion is based on a rigid, binary cut-off of a continuous variable like heart rate, it is implicitly prespecified that the appropriateness of that cut-off will be scrutinized when the study is analysed.

It is more problematic to accept all of the FDA reviewer’s proposals. Whether one accepts or rejects them depends upon whether one believes that the only efficacy questions that can be answered in a clinical trial involve the primary endpoint analysis or a rigidly specified chain preserving an alpha of 0.05. While the latter restrictions minimize the chance of committing a type I error, they do not allow for maximizing the clinical utility of the thousands of data points that comprise a typical large cardiovascular outcome trial. Acceptance also depends upon how much one trusts patterns of findings, such as the heart rate quintile analyses, rather than isolated P-values.

So what should be the take-home messages for the clinician? The least controversial is that we should be more aggressive in our use of ivabradine particularly when patients are compatible with the EMA indication (except ischaemic patients not on a loop diuretic) or the FDA indication in patients with non-ischaemic aetiology. The estimated risk reductions for all-cause mortality (EMA 17% and FDA reviewer 20%) and HF hospitalization (EMA 26% and FDA reviewer 29%) are substantial. In fact, these benefits are comparable to those reported for sacubitril/valsartan, which has been hailed as a breakthrough in HF therapy.⁸ We should also be more aggressive in our use of ivabradine in ischaemic patients on a loop diuretic.^9,10 While staying within the indication, we should at least achieve the indicated benefits and, if the ivabradine-loop diuretic interaction is real, the achieved benefits will be greater. We should be cognizant that loop diuretic use is likely a two-edged sword.⁸ While diuretics are needed to relieve congestion, they also likely increase the risk of arrhythmic death. That loop diuretic use should be closely monitored may be one of the hidden lessons of the ivabradine trial experience. Finally, even if loop diuretic use is a better measure of HF severity than either of the latter two standard measures, it appears to be the best measure for determining who with ischaemic heart disease benefits most from ivabradine.

Acknowledgements

Special thanks to the dedicated FDA and EMA medical reviewers, without whom such analyses would not be possible.

Conflict of interest: none declared.

References