So far as I can tell the only problem with PARADIGM-HF is that the results are so good that it’s boring. Anyone interested can reasonably assume that what they hear or read about PARADIGM-HF — and cardiologists will be seeing and hearing an awful lot about it — will be overwhelmingly positive.
Briefly, the trial did everything its sponsor and its investigators (led by Milton Packer and John McMurray) hoped. It met all its major endpoints in all the subgroups without raising any sort of a safety signal. If the findings are confirmed after a rigorous FDA review, then will likely have a blockbuster on its hand.
All the excitement is over a new drug still known only by its number, LCZ696. It consists of the well-known angiotensin receptor blocker valsartan and a novel compound, the neprilysin inhibitor sacubitiril. Inhibition of neprilysin is intended to held reduce the neurohormonal activation that helps drive common heart failure (HF) processes like vasoconstriction, sodium retention, and remodeling. This specific combination was designed, according to the authors, to avoid angioedema, a side effect which doomed a similar earlier drug, omapatrilat, which combined an ACE inhibitor and a neprilysin inhibitor.
Results of the trial were presented today at the European Society of Cardiology meeting in Barcelona and published simultaneously in the New England Journal of Medicine.
A total of 8,442 HF patients with reduced ejection fraction (EF) were randomized to enalapril or LCZ696. The study was stopped early, though enrollment in the trial had been completed, following an interim analysis performed by the data and safety monitoring committee, which found “that the prespecified stopping boundary for an overwhelming benefit had been crossed.”
After a mean followup of 27 months, LCZ696 was superior to enalapril in most of the major outcomes. There were significant reductions in the primary endpoint (the combination of cardiovascular death or hospitalization for HF), the individual components of the endpoint, all-cause mortality, and HF symptoms:
- The primary endpoint was reduced by 20%, from 26.5% in the enalapril group to 21.8% in the LCZ696 group (HR 0.80, CI 0.73-0.87, p<0.001)
- Cardiovascular death was also reduced by 20%, from 16.5% to 13.3% (HR 0.80, 0.71-0.89, p<0.001
- Hospitalizations were reduced by 21%, from 15.6% to 12.8% (HR 0.79, 0.71-0.89, p<0.001)
- Deaths (from any cause) were reduced by 16% from 19.8% to 17%
There was also a highly significant improvement in quality of life but there were no significant differences in the onset of new atrial fibrillation or in renal function.
The authors calculated that over the course of the trial 21 patients would need to have been treated with LCZ696 instead of enalapril to prevent one primary outcome event. The number needed to treat to prevent one cardiovascular death was 32.
LCZ696 was effective across the entire spectrum of prespecified subgroups, including age, race, sex, geographic region, NYHA class, diabetes status, and prior drug use. The benefits of LCZ696 were displayed early in the trial, and occurred in a population well treated with other proven therapies including beta-blockers and mineralocorticoid-receptor antagonists.
Overall LCZ696 was well tolerated. A total of 17.8% of LCZ696 patients and 19.8% of enalapril patients discontinued treatment during the course of the study. Significantly more LCZ696 patients developed hypotension, though this did not normally require drug discontinuation. By contrast, enalapril was associated with significantly more cough and increased serum creatinine and potassium levels.
Angioedema occurred in 19 LCZ696 patients and 10 enalapril patients. No patients had their airway compromised. Three LCZ696 patients and one enalapril patient required hospitalization for angioedema.
The authors wrote that the benefits of LCZ696 were “highly significant and clinically important, even though the drug was compared with a dose of enalapril that has been shown to reduce mortality, as compared with placebo.” The study “was designed to provide persuasive evidence to support the replacement of ACE inhibitors or ARBs with LCZ696 in the management of chronic heart failure. The trial was devised to show an advantage with respect to cardiovascular mortality alone, which was the primary determinant of the sample size and for which a statistically compelling effect was required to stop the trial early because of a benefit.”
In an accompanying editorial, Mariell Jessup said the results should be broadly applicable to the large population of people with mild to moderately severe HF, since the patients studied in the trial were similar to those in previous large, influential trials. She pointed out that although the systolic blood pressure on treatment was lower by 3.2 mm Hg in the LCZ696 arm, the investigators “propose that this difference in blood pressure was not a determinant of the salutary benefits of LCZ696.”
The trial, Jessup concluded, “may well represent a new threshold of hope for patients with heart failure” and “may apply to a wide spectrum of patients, even those who are currently receiving the best possible therapy.”
Uncharacteristically, Sanjay Kaul was quite upbeat: “At long, long last, a winner!” He said the trial “appears to have met all endpoints.” It was “reasonable” to truncate the trial since it was based on “a statistically persuasive effect on CV death as well as the primary endpoint.” Since the reduction in cardiovascular deaths was evident across subgroups, the drug “can be potentially recommended across a broad swath of heart failure patients.”
Kaul — remembering past instances (most notably ATLAS ACS-TIMI 51) of seemingly successful trials that were then torpedoed at the FDA — did insert one note of caution, but remained optimistic: “Unless there are issues with regards to trial conduct and data ascertainment (which can only be addressed during rigorous FDA review), I think we might at long, long last have a winner in this therapeutic space!”
The Co-Principal Investigator
In an interview, Packer said that the trial “turned out better than we thought,” though he also stated, emphatically, that the investigators always had ambitious goals for a trial with an active treatment for a comparator. “We said upfront that we wanted to look for a 15% incremental effect — above and beyond enalapril. We chose that number because enalapril had a 15% effect over placebo and our goal was to double the effect. We thought if we came in with an extraordinary p value it would be convincing.”
Packer said that he believes the trial results are generalizable to the larger HF population in the U.S. with low EFs. Asked about the apparent low rate of ICDs in the study, at least compared to the U.S. rate, Packer explained that this sort of problem is inevitable in a large international trial in 47 countries, where ICD usage is often lower than in the U.S. He pointed out that LCZ696 was equally effective in the U.S. and elsewhere. Packer also agreed with Jessup that patients in the trial were not younger than in previous trials. Unlike the heart failure with preserved ejection fraction (HFPEF) population, which is largely composed of older patients, the low EF population is fairly represented in PARADIGM-HF, Packer said.
Clinical trial experts will likely focus their questions on the possible effects of stopping the trial early, thereby perhaps accentuating a temporary finding and preventing a natural regression to the mean. Packer pointed to several lines of evidence suggesting that this was not the case. For one, the effect size was constant throughout the trial. There was no temporary bubble that coincided with a single view by the DSMB. In addition, said Packer, the effect size has actually grown larger since the trial was stopped in March. Finally, the statistical power was robust because, despite the early termination, the number of cardiovascular deaths at the time was actually larger than the original target that had been used to determine the trial’s sample size.
Packer said he was also confident the trial would withstand the scrutiny of the FDA. “We look forward to the FDA digging in to their heart’s content.” He said he’s not seen any sign of the types of problems that the FDA has uncovered in the past. The very high statistical significance of the trial bolsters his confidence. “If you’re sitting on a p value of .03 you need to worry. With our p value we have a very big buffer.” (Clinical trial junkies will appreciate Packer’s pride in the success of the authors in persuading the NEJM editors to go beyond the “p<0.001” for the primary endpoint and to include in brackets the full p value with 7 zeroes after the decimal point.)
Finally, Packer talked about the commercial impact of LCZ696, which is likely to be expensive compared to the generic ACE inhibitors and ARBs that it is meant to replace. “If you have a drug that saves lives, keeps people out of the hospital, and reduces symptoms, and is better than the cornerstone of therapy, and in patients who are already receiving the other best therapies, what is that worth? That is not for me to say, but it has to be worth something.”