Sacubitril/Valsartan in HFpEF: A Trial That Narrowly Missed but Didn't Really Fail
An ejection fraction of 48% sits well inside the exact subgroup where the pivotal HFpEF trial found its clearest benefit — even though the trial's own headline result missed statistical significance by a narrow margin.
J.R., a 72-year-old woman, spent nearly four decades in public education, the last fifteen as a high school principal, before retiring to spend more time with her grandchildren and the garden she'd neglected for years. She was diagnosed with HFpEF eight months ago after a hospitalization for volume overload, with an echocardiogram showing an ejection fraction of 48% and an elevated NT-proBNP consistent with structural heart disease. She remains symptomatic — NYHA class II, occasional ankle swelling, shortness of breath with two flights of stairs — on a diuretic and well-controlled blood pressure. Her cardiologist is considering adding sacubitril/valsartan, and the conversation has turned to a trial whose headline result is easy to misread as a failure.
PARAGON-HF enrolled patients with an ejection fraction of 45% or higher and compared sacubitril/ valsartan against valsartan alone. Its primary composite endpoint — total heart failure hospitalizations and cardiovascular death — narrowly missed statistical significance, rate ratio 0.87, 95% confidence interval 0.75 to 1.01, p=0.059. But a pre-specified subgroup analysis, splitting patients by the trial's own median ejection fraction of 57%, found a real, statistically significant benefit below that median — rate ratio 0.78, 95% confidence interval 0.64 to 0.95 — with no comparable signal above it. J.R.'s EF of 48% sits well inside that below-median subgroup, closer to the trial's lower boundary than to its high-normal edge. The 2022 AHA/ ACC/HFSA heart failure guideline responded to exactly this pattern with a Class 2b recommendation for sacubitril/valsartan in HFpEF — a real, if intentionally modest, endorsement built specifically around the subgroup finding rather than the missed primary endpoint alone.
Heart failure clinic, HFpEF therapy escalation
I'd start sacubitril/valsartan. The primary endpoint missed narrowly, but the pre-specified below- median-EF subgroup found a real benefit, and J.R.'s EF of 48% sits well inside that subgroup, not at the trial's high-normal edge where the signal was weakest.
I'm not saying the primary-endpoint miss doesn't matter — I'm saying it matters less for a patient who sits in exactly the subgroup where the trial's own strongest signal was found.
A narrowly-missed primary endpoint is still a missed primary endpoint. Even pre-specified subgroup findings carry a real risk of chance results, and I'd be cautious about treating a secondary analysis as equivalent in strength to a positive primary trial.
I'm not disputing that this particular subgroup finding might be real — I'm flagging that the methodology itself deserves the caution regardless of whether it turns out to be right here.
The 2022 guideline already did this weighing for us — a Class 2b recommendation built specifically around the subgroup pattern, not despite it. This isn't a fresh judgment call being made from scratch at her bedside; the committee that wrote the guideline already balanced the primary-endpoint miss against the subgroup signal and reached a real, if modest, recommendation.
Agreed: start sacubitril/valsartan at a low dose, titrate as tolerated, and reassess symptoms, blood pressure, and renal function in four weeks.
Not agreed, and carried forward explicitly rather than smoothed over:
Maintains that subgroup-based prescribing deserves ongoing scrutiny as a general practice — accepted today's decision as reasonable given the guideline's own endorsement, not a retraction of the broader methodological concern.
All three voices agreed this wouldn't necessarily indicate the wrong decision was made, given the trial's own modest effect size — but would prompt a broader reassessment of her HFpEF management rather than simply continuing unchanged.
The Heart Failure Cardiologist's subgroup-matching argument was accepted by both other voices as the practical basis for today's decision — not as proof the drug will help her, but as the best available reason to expect it might, given where her own numbers actually fall.