Marc Pfeffer presented the results of PARADISE-MI at ACC last week. As discussed previously on CVCTCardiobrief, sacubitril/valsartan did not reduce the primary outcome compared to placebo in people post-MI without previous heart failure.
The primary outcome of cardiovascular death, heart failure hospitalization or outpatient heart failure occurred in 11.9% of those randomised to sacubitril/valsartan and 13.2% of those randomised to placebo (p=0.17).
It is certainly tempting to highlight the “numerical trend” towards positivity here, especially given our community longstanding diatribe against dichotomizing trials by p<0.05. But the issue at hand is whether to use ramipril or the new drug and I think anyone can see that there is no substantial advantage of sacubitril/valsartan over ramipril.
Pfeffer, on the other hand, found the results “encouraging”. He took a surprisingly positive stance on the neutral results in his statement that “Although we did not significantly reduce our primary endpoint, there were consistent findings that sacubitril/valsartan could represent an incremental improvement over ramipril”.
This appears to be based on 2 analyses: (1) a comparison of total primary events (i.e. counting all heart failure hospitalizations, not just the first), and (2) a comparison of the original primary outcome but with investigator event-adjudication. Pfeffer emphasized numerical trends toward a sacubitril/valsartan advantage. In one analysis, the benefit became most marked when investigator-reported outpatient heart failure events were added to investigator-reported heart failure hospitalizations (HR 0.69 (95% CI 0.54-0.88), P=0.003). All these analyses went beyond the primary endpoint analysis, and will undoubtedly be much debated. Giving consideration to non-adjudicated endpoints is a topic of heated discussion among trialists. Accounting for outpatient worsening of heart failure events (i.e. heart failure without hospitalization) is gaining some support, and this approach is not without merit. However, no matter how interesting, giving prominence to these exploratory analyses may be misleading, giving the impression that the trial might as well have been positive. It is likely that journals and peer-reviewers of the main publication will keep the message true to the pre-specified analyses.
No doubt the PARADISE-MI primary outcome results are unwelcome results for the investigators and for Novartis. Faiez Zannad comments: “There may be reasons for these results being below expectations, including the low risk profile of the patients, as a result of contemporary drug and revascularization therapy and, which most patients received in the trial. Indeed, in PARADISE-MI, the event rate was lower than in previous post MI with heart failure/low EF trials, which sought added benefit over ACE-inhibitor therapy (VALIANT, OPTIMAAL, EPHESUS). Nevertheless, it may be concluded from PARADISE-MI that neprilysin inhibition, in the dose achieved, as much as angiotensin receptor blockers (VALIANT, OPTIMAAL) does not add much to the benefit of RAS-inhibitors in these post MI patients. Interestingly, only 40% of patients in PARADISE-MI were taking a mineralocorticoid receptor antagonist, which is unfortunate. Most patients in PARADISE MI were eligible for such therapy, since mean eGFR was 72ml/min/1.73m2. The mineralocorticoid receptor antagonist, eplerenone, provided substantial benefit on top of RAS-inhibitors as shown in EPHESUS.”