Author’s Reply to Stability or Latent Risk? Interpreting MRA Withdrawal in Recovered Heart Failure

Dear Editor,

We sincerely thank Birgün and colleagues for their thoughtful interest in our With-HF trial evaluating spironolactone (mineralocorticoid receptor antagonist, MRA) withdrawal in patients with heart failure with improved ejection fraction (HFiEF).¹⁾ Their comments appropriately highlight important limitations and the need for careful interpretation. We agree that our study has clear constraints, and we also agree with the central message of their correspondence: our trial is hypothesis-generating and does not justify definitive conclusions regarding routine MRA discontinuation. Rather, it provides an initial signal that selective withdrawal of a single agent (MRA), while maintaining other foundational therapies, may be feasible in a carefully selected subset, and it underscores the need for larger and longer-term randomized studies.

Below, we address their main points.

SMALL SAMPLE SIZE AND SHORT-TERM FOLLOW-UP

We fully acknowledge that the modest sample size (62 randomized) and the short follow-up (6-month visit window) limit statistical power and the ability to detect late relapse or infrequent adverse outcomes. This was an inherent aspect of our pilot design, intended to test feasibility and provide preliminary safety signals to inform the design of a definitive, adequately powered trial. Importantly, much of the existing withdrawal literature that informs current concerns has similarly relied on relatively small, pilot-scale cohorts—most notably TRED-HF, where withdrawal of all heart failure therapies was associated with a high relapse rate within months.²⁾

In addition, we are currently preparing extended follow-up analyses of this cohort; while these data are not yet peer-reviewed, they will help address the concern that short-term stability may not fully capture latent risk.

OPEN-LABEL NATURE AND RISK OF BIAS

We agree that the open-label design is a limitation. However, our primary and key secondary endpoints were chosen specifically to reduce susceptibility to expectation or treatment-intensity bias: left ventricular ejection fraction (LVEF) and other echocardiographic measures were analyzed by an independent echocardiographer blinded to clinical information, and N-terminal pro-B-type natriuretic peptide (NT-proBNP) was used as an objective biomarker endpoint. While open-label designs can influence subjective outcomes or discretionary care, our principal outcomes were objectively measured and centrally structured by protocol follow-up. In addition, while capturing objective clinical events or structural and functional cardiac changes following withdrawal is important, assessing changes in patients’ quality of life and exercise capacity is also a crucial component. In particular, potential bias related to open-label design when evaluating quality-of-life outcomes should be carefully considered in future studies.

PATIENT SELECTION AND GENERALIZABILITY (INCLUDING ISCHEMIC ETIOLOGY AND “ASYMPTOMATIC” PROFILE)

We agree that our findings should not be generalized to all patients labeled HFiEF in routine practice. Our cohort largely reflected a clinically stable population (predominantly New York Heart Association [NYHA] class I, with exclusion of NYHA III–IV), and a high proportion received contemporary background therapy including angiotensin receptor neprilysin inhibitor. Accordingly, our results are most applicable to patients whose HFiEF represents meaningful recovery in both LVEF and symptoms, and we do not advocate withdrawal in patients with persistent symptoms, residual congestion, or ongoing instability.

At the same time, we would note that our cohort was not restricted to purely “idealized” recovery phenotypes. For example, we included patients with ischemic etiology (14.5%)—a group not represented in some landmark withdrawal work such as TRED-HF trial, which focused on recovered dilated cardiomyopathy. We therefore agree with the correspondents’ call for broader, pragmatic trials spanning diverse etiologies and risk profiles, but we would emphasize that our study population was intentionally aligned with the clinical scenario in which de-escalation is most commonly contemplated: patients with both improved LVEF and improved clinical status.

PLEIOTROPIC EFFECTS OF MINERALOCORTICOID RECEPTOR ANTAGONISTS AND POTENTIAL PROGNOSTIC TRADE-OFFS

We agree that MRAs may confer benefits beyond short-term changes in LVEF, including antifibrotic and antiarrhythmic effects, and that these effects may not be fully captured by echocardiographic stability and NT-proBNP over 6 months. This is precisely why we view our results as a starting point rather than a practice-changing conclusion. That said, clinical decision-making in stable HFiEF also requires balancing potential long-term benefits against known adverse effects (e.g., hyperkalemia, hyponatremia, renal dysfunction) and patient burden, and—crucially—recognizing that current guideline recommendations to continue GDMT in HFiEF are necessarily conservative because high-quality evidence for safe de-escalation strategies remains limited.³⁾

CONCLUSION

In summary, we appreciate the correspondents’ careful framing: short-term stability does not guarantee absence of latent risk, and definitive evidence requires larger, longer-term randomized trials enriched with biomarkers, arrhythmic outcomes, and patient-centered endpoints. Our pilot trial should be interpreted as evidence that selective, single-agent withdrawal (MRA)—while maintaining other foundational therapies—may be feasible in a carefully selected, clinically stable HFiEF subgroup, contrasting with the high relapse rates observed when all heart failure therapies are withdrawn. We hope these data will encourage the field to move beyond “all-or-none” withdrawal and toward rigorous evaluation of staged, individualized de-escalation strategies with close monitoring.