Clinicians managing patients with reduced ejection fraction heart failure (HFrEF) may look at findings from trials such as DAPA-HF, EMPEROR Reduced, and PIONEER HF and ask themselves: “Which drug should I use—an SGLT2 inhibitor or ARNI?”
“Both,” said Milton Packer, MD, of Baylor University Medical Center in Dallas and Imperial College in London. Packer based his response on findings from a subgroup analysis of data from EMPEROR Reduced that analyzed outcomes in a subgroup of patients treated with empagliflozin who also received sacubitril/valsartan versus empagliflozin-treated patients who did not receive the ARNI. He presented results from the subgroup analysis at a late-breaking clinical trials session at the Heart Failure Society of America (HFSA) virtual meeting.
EMPEROR Reduced randomized 3,730 patients with chronic HFrEF (defined as an ejection fraction of 40% or less) and NYHA Class II-IV Heart Failure to receive 10 mg of empagliflozin daily or placebo. All patients were “very well-treated,” Packer noted. “They were receiving appropriate medical therapy (beta-blockers, ACEi/ARBs/ARNI, MRA) and about 30% had device therapy.”
Packer reported the main findings of the trial at the ESC Virtual Congress in late August.
Half of the patients had a history of diabetes and more than 70% had LVEF of 30% or less. Importantly, about 20% of the patients were receiving sacubitril/valsartan, which prompted Packer and colleagues to look at the outcomes in patients taking both an SGLT2 inhibitor and ARNI.
“We have three primary endpoints listed in a hierarchical manner: the primary endpoint was cardiovascular death and time to first heart failure hospitalization. The second of the hierarchies was all heart failure hospitalizations — first and recurrent. The third of the hierarchies was the slope of decline in estimated glomerular filtration rate (GFR),” Packer explained.
In the subgroup of patients taking sacubitril/valsartan and empagliflozin, “we reached significance in each of the subgroups,” he said.
For example, looking at the topline results from the entire trial, 462 patients in the placebo arm of EMPEROR Reduced had a primary endpoint event versus 361 patients randomized to empagliflozin, which resulted in an HR of 0.75, or a 25% reduction in relative risk (95% CI 0.65-0.86, P <0.0001).
In the subset analysis, 340 empagliflozin patients and 387 placebo patients were also on neprilysin inhibition, and the HR for a primary endpoint event was 0.64 (95 CI 0.45-0.89). There were 3,003 EMPEROR Reduced patients who were not on neprilysin inhibition, and the HR for a primary endpoint event was 0.77 (95% CI 0.66-0.90) favoring the empagliflozin-treated patients. The P value for treatment by neprilysin interaction was 0.31.
Considering overall heart failure hospitalizations, the HR was 0.65 (95% CI 0.42-1.00) favoring empagliflozin, and for the third hierarchical endpoint, slope of GFR decline, the HR was 0.50 (95% CI 0.32-0.77), again favoring empagliflozin.
In terms of adverse events, 19.9% of neprilysin inhibitor patients in the placebo group and 15.9% of those taking empagliflozin and a neprilysin inhibitor had an adverse event leading to discontinuation of the study drug. The adverse event rate generally was low and favored neprilysin plus empagliflozin, with the exception of symptomatic hypotension—there were two more cases in the neprilysin plus empagliflozin group.
Packer concluded that “background use of neprilysin inhibition (with sacubitril/valsartan) did not diminish (and may have enhanced) the efficacy or safety of empagliflozin,” but he noted that “all interaction P values were statistically non-significant.”
Harvard statistician Brian Claggett, PhD, an associate scientist at Brigham and Women’s Hospital, served as the discussant for Packer’s presentation and he pushed back against Packer’s “may have enhanced” conclusion.
Claggett said that because the interaction P value was non-significant, it “should probably not be used to make a strong claim … non-significant P value means that we failed to find a dramatic signal;” however, that does not mean that the there is no treatment effect.
When interpreting trial results, Claggett said he has generally found that “too much emphasis has been placed on point estimates and not enough emphasis on confidence intervals. Point estimates describe the past — what happened in a specific finite population. The confidence intervals describe what’s going to happen to those future patients.”
For example, he said that the true value of neprilysin plus empagliflozin is probably not the point estimate presented by the HR — say 36% — but rather somewhere between an 11% reduction and a 55% reduction based on the 95% CI 0.45-0.89.
In an interview with BreakingMED, Packer noted that he and Claggett have had the point estimate versus confidence interval debate many times, but when presenting evidence from clinical trials it must be presented in ways that clinicians understand.
“Do you think that a clinician is going to prescribe a treatment if we say, well it could reduce the risk by 11% or possibly as much as 55%? No, they wouldn’t. We need to present findings in a way that is understandable to the clinicians,” he said.
In the case of ARNI—sacubitril/valsartan—and SGLT2i, there is a wealth of evidence to support benefit, with P values that are not a near thing. “There are 9 randomized trials of SGLT2i in reduced ejection fraction heart failure and all of them are showing a 30% benefit.”
In an email, to BreakingMED, Packer wrote:
“Brian [Claggett] emphasized the confidence intervals rather than our point estimates in a single trial, but in truth, the confidence intervals should be interpreted across trials, and in the specific instance of SGLT2 inhibitors, the amount of evidence is so large that the confidence intervals actually converge on the point estimates. So, I am very confident with the point estimates that I showed.
“One additional point: Brian focused on one endpoint, but in my presentation, I showed many endpoints — all moving in the same direction — in favor of combining SGLT2 inhibitors and neprilysin inhibitors,” he added.
Be aware that this activity discusses findings from a post-hoc subset analysis that requires confirmation in a prospective, randomized trial.
Note that this analysis suggests that sacubitril/valsartan does not appear to diminish the benefit of empagliflozin in HFrEF patients.
Peggy Peck, Editor-in-Chief, BreakingMED™
The EMPEROR Reduced trial was funded by Boehringer Ingleheim and Eli Lilly.
Packer disclosed consultant agreements with Amarin, Amgen, AstraZeneca, Akcea, Abbvie, Boehringer Ingelheim, Daiichi Sankyo, Johnson & Johnson, Lilly, Novartis, NovoNordisk, Pfizer, Sanofi, Synthetic Biologics, and Theravance.
Claggett has received consultancy fees from Boehringer Ingelheim, Gilead, and AOBiome.
Cat ID: 3
Topic ID: 74,3,3,707,127,192,925,706