COMMENTARY

PARADISE-MI Makes Me Question the Benefits of Sacubitril/Valsartan

John Mandrola, MD

Disclosures

May 16, 2021

The PARADISE-MI trial comparing sacubitril/valsartan to ramipril in patients with left ventricular dysfunction after myocardial infarction (MI) and clinical heart failure led off the late-breaking clinical trial session at the American College of Cardiology (ACC) 2021 Scientific Session. Organizers tend to reserve this slot for the biggest newsmakers.

And if the angiotensin neprilysin inhibitor had been associated with a statistically significant reduction of the primary endpoint of cardiovascular death or heart failure events, it would have been big news.

But PARADISE-MI did not meet that goal.

Patients treated with sacubitril/valsartan experienced 6.7 events per 100 patient-years compared with 7.4 events per 100 patient-years for ramipril-treated patients. The 0.7 absolute difference translated to a hazard ratio of 0.90, with 95% CIs ranging from 0.78 to 1.04.

The slide depicting a separation of the Kaplan-Meier curves featured a truncated y-axis from 0 to 0.20. Had the authors used a y-axis from 0 to 1, the curves would have been close to superimposable.

The P value of .17 is less than 2 standard deviations from the null; this indicates that if there were no difference in these drugs, the chance of seeing these results is nearly 1 in 5. That is not strong evidence; the accepted (albeit arbitrary) standard for "significance" is 1 in 20 or .05.

The lower rate of all-cause death in the sacubitril/valsartan arm also did not come close to reaching statistical significance (P = .16), and neither did the first three key secondary endpoints.

The PARADISE-MI Presentation

We were then shown results of the trial with different ways of measuring outcomes—one using total events and another using investigator-reported events.

Recall that when scientists design experiments, they define a primary endpoint to measure. PARADISE-MI authors chose time to first event. That turned out to be nonsignificant.

When the authors analyzed the data using total events, patients treated with sacubitril/valsartan experienced a 21% relative reduction in events, with a 95% CI ranging from 0.65 to 0.97 (P =.02). The analysis using only investigator-adjudicated primary events yielded a 15% reduction; the CI ranged from 0.75 to 0.96 (P = .01).

Adverse events overall did not differ, although 182 more patients (6%) in the sacubitril/valsartan arm had hypotension.

The first bullet point of the conclusion slide stayed true to the primary endpoint: "sacubitril/valsartan did not result in a significantly lower rate of CV death, heart failure hospitalization or outpatient heart failure requiring treatment."

That should have been it. But it wasn't.

The second bullet point on the summary slide read: "pre-specified observations of reductions in both the investigator reports of the primary composite as well as in the total (recurrent) adjudicated events support incremental clinical benefits of sacubitril/valsartan."

While presenter Marc Pfeffer, MD, mentioned that the nonsignificant primary endpoint renders other findings hypothesis-generating, the second bullet point, with its causal language, remains for all to see and screenshot and tweet.

The last slide of the presentation showed the declining death rates in clinical trials of patients after MI over the past 2 decades. This is surely due to better background care. The point is that lower events overall make it harder to show an incremental benefit for new therapies. This is bad news for industry but good news for patients.

Specific Take on PARADISE-MI

Spin in medical research is defined by language that detracts from primary endpoints that are not statistically significant. I am afraid there was spin during this late-breaking session.

I asked clinical trial expert, Sanjay Kaul, from Cedars-Sinai in Los Angeles, California, to comment on the presentation of the "positive" analyses of total primary outcome events and investigator-reported outcomes in the setting of nonsignificant primary and key secondary endpoints. By email, Kaul wrote that these analyses were not actionable for regulatory approval or informing clinical practice. He added that "strictly speaking, P values assigned to these outcomes are meaningless."

We should not mistake Kaul's comments as technicalities. PARADISE-MI enrolled high-risk patients after MI. The median ejection fraction was 36%, and the inclusion criteria required both clinical evidence of heart failure and 1 of 8 risk-enhancing features, such as diabetes, atrial fibrillation, or age 70 years or older.

Despite these factors, sacubitril/valsartan did not significantly outperform the generic and inexpensive angiotensin-converting enzyme inhibitor ramipril. I added the modifier "significantly" because I don't want to completely discount the possibility of a small reduction in the time-to-event analysis.

But it's not enough of an advantage.

Sacubitril/valsartan costs vastly more than generic ramipril. Cost matters because money spent on one thing means less is available for others. Think disparities in care.

We should also not ignore the fact that more patients in the sacubitril/valsartan arm had hypotension. This number would surely be higher outside the confines of a trial.
 

General Comments on Angiotensin Neprilysin Inhibition

I am beginning to wonder about this drug. Sacubitril/valsartan has been studied in three major clinical outcomes trials and missed statistical significance in two of them. In PARAGON-HF, sacubitril/valsartan did not reach statistical significance vs valsartan alone in patients with heart failure and a preserved ejection fraction.

The landmark PARADIGM-HF trial comparing sacubitril/valsartan to enalapril in patients with heart failure due to a reduced ejection fraction reported a remarkable 4.7–percentage point absolute risk reduction in the primary endpoint of cardiovascular death and heart failure hospitalization, as well as a significant reduction in overall death.

These data easily passed regulatory review and changed guidelines. Over time, the data, combined with marketing, ensconced this drug as something of a super-drug. In my office, stories emerged that it made patients feel dramatically better. Some of my colleagues began calling it "the big E."

Yet numerous authors had published critical appraisal of PARADIGM-HF. Ahn and Prasad questioned the unequal drug run-in periods, the dosing of enalapril, and the fairness of comparing two drugs (sacubitril and valsartan) against one (enalapril). A Brazilian group also raised the issue of inadequate enalapril dosing.

These criticisms seem stronger in light of the last two nonsignificant trials of sacubitril/valsartan.

A proponent of angiotensin neprilysin inhibition might dismiss the miss in PARAGON-HF because (1) patients with heart failure with preserved ejection fraction are a diverse group and (2) it barely missed with a P value of .06.

But the decidedly mediocre result in PARADISE-MI is harder to dismiss because the patients enrolled in this trial were (somewhat) similar to those enrolled in PARADIGM-HF: similar age, the median left ventricular ejection fraction is within 6 percentage points, and the patients had clinical heart failure.

If you believe sacubitril/valsartan is hugely beneficial over standard renin-angiotensin blockade in patients with heart failure with reduced ejection fraction, why wouldn't it confer similar benefits to patients with new-onset heart failure due to left ventricular systolic dysfunction after MI?

Perhaps the outsized benefit seen in PARADIGM-HF was an outlier. Large effect sizes often don't replicate in subsequent studies. Or maybe the explanation for the misses in PARAGON-HF and PARADISE-MI is the use of fairer comparators: valsartan in equal doses and ramipril, respectively.

Conclusion

PARADISE-MI does not convince me that sacubitril/valsartan should be used in patients with left ventricular dysfunction after MI.

And the three outcomes trials, taken together, make me wonder whether we have overestimated the true effect of angiotensin neprilysin inhibition.

Many other cardiac drugs—β-blockers, renin-angiotensin blockers, and sodium glucose cotransporter-2 inhibitors, for instance—have demonstrated statistically robust benefit in multiple trials.

Perhaps another lesson from PARADISE-MI is that new drugs for common conditions ought to require two confirmatory trials before widespread clinical acceptance.

Kaul reports having consulted for Norvartis. 

John Mandrola practices cardiac electrophysiology in Louisville, Kentucky, and is a writer and podcaster for Medscape. He espouses a conservative approach to medical practice. He participates in clinical research and writes often about the state of medical evidence. 

Follow John Mandrola on Twitter

Follow theheart.org | Medscape Cardiology on Twitter

Follow Medscape on Facebook, Twitter, Instagram, and YouTube

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....

Recommendations