STEPwise De-escalation and Optimizing Withdrawal of ARNI and SGLT2i in Normalized Heart Failure

Overview

This is a multicenter, randomized, open-label pilot study to evaluate whether stepwise withdrawal of two heart failure medications-angiotensin receptor-neprilysin inhibitor (ARNI) and sodium-glucose cotransporter-2 inhibitor (SGLT2i)-is safe in patients with Heart Failure with improved Ejection Fraction (HFimpEF) whose underlying structural cause (valvular heart disease or ischemic cardiomyopathy) has been completely corrected by surgery or intervention. Eighty adult patients (40 per arm) whose left ventricular ejection fraction (LVEF) has recovered to 50% or higher and whose NT-proBNP is below 250 ng/L will be randomized 1:1 to either (1) stepwise withdrawal of ARNI followed by SGLT2i over one month under close echocardiographic monitoring, or (2) continuation of their current guideline-directed medical therapy. The primary outcome is the change in LVEF at 12 months, with non-inferiority of the withdrawal strategy declared if the LVEF decline is within 5 percentage points of the continuation arm. Secondary outcomes include cardiovascular death, heart failure hospitalization, NT-proBNP, quality of life (KCCQ-12), 6-minute walk distance, and adverse events. Results from this pilot will inform the design and sample size of a subsequent definitive non-inferiority trial and may provide initial evidence to guide deprescribing decisions in clinical practice.

Background The 4-pillar guideline-directed medical therapy (GDMT) for heart failure-comprising ARNI, SGLT2i, mineralocorticoid receptor antagonists (MRA), and beta-blockers-is the established standard of care for patients with reduced ejection fraction. However, evidence guiding medication withdrawal in patients whose ejection fraction has normalized after correction of a reversible structural cause (HFimpEF) is lacking. The TRED-HF trial demonstrated that withdrawing GDMT in patients with idiopathic or familial dilated cardiomyopathy led to relapse in approximately 40% within six months. In contrast, WITHDRAW-AF, which enrolled patients whose tachycardia-induced cardiomyopathy resolved after successful catheter ablation, observed only 8.3% relapse, and CATHEDRAL-HF, which retained beta-blockers while withdrawing other agents, reported approximately 10% relapse. These findings suggest that selective, stepwise withdrawal in patients with fully corrected underlying disease may be safer than indiscriminate discontinuation. Rationale for a Pilot Trial This pilot study addresses three objectives prior to a definitive non-inferiority trial: (1) primary safety assessment of a stepwise withdrawal protocol after structural correction; (2) feasibility of multicenter enrollment and follow-up; and (3) estimation of event rates and variance of the primary outcome to inform sample-size calculation for the subsequent confirmatory trial. Study Design Multicenter, randomized, open-label, parallel-group pilot trial. Eighty patients will be randomized 1:1 to stepwise withdrawal or continuation, stratified by baseline NT-proBNP (≤50, 51-125, 126-250 ng/L) using a web-based central randomization system (Sealed Envelope). Five centers in the Republic of Korea will participate. Intervention Stepwise Withdrawal Arm: ARNI is discontinued first. After one month of clinical and echocardiographic assessment, SGLT2i is discontinued only if there is no echocardiographic deterioration (LVEF drop \<10 percentage points) and NT-proBNP remains ≤250 ng/L. MRA and beta-blockers, if present, are continued. Continuation Arm: All current heart-failure medications, including ARNI and SGLT2i, are maintained per standard practice. Safety Monitoring Two key safeguards are implemented: (1) a strictly sequential, stepwise withdrawal that prevents abrupt neurohormonal activation; and (2) mandatory transthoracic echocardiography at 1 and 3 months post-withdrawal to detect early subclinical deterioration before symptomatic relapse. This intensive echocardiographic surveillance distinguishes the protocol from prior trials, which relied primarily on serum biomarkers. Patients meeting predefined deterioration criteria (LVEF decline ≥10 percentage points from baseline, LVEF \<40%, heart failure hospitalization, or emergency-department visit) are immediately reinstated on full GDMT and recorded as protocol withdrawal events. An independent Data and Safety Monitoring Board reviews safety data periodically. Endpoints Primary Endpoint: Change in LVEF from baseline to 12 months. Non-inferiority is declared if the lower bound of the one-sided 95% confidence interval for the between-group difference exceeds -5 percentage points in the per-protocol population, with ITT analysis as a supportive analysis. Secondary Endpoints: Composite of cardiovascular death and heart failure hospitalization at 12 months; absolute LVEF change; NT-proBNP change at 3 and 12 months; NYHA functional class change; KCCQ-12 quality-of-life score; 6-minute walk distance; incidence and severity of adverse events; and rate and timing of medication reinstatement. Eligibility Adults (≥19 years) with a prior LVEF ≤40% that has recovered to ≥50% following complete surgical or interventional correction of valvular heart disease (mitral regurgitation, aortic stenosis, aortic regurgitation) or ischemic cardiomyopathy (PCI or CABG); NT-proBNP \<250 ng/L at enrollment; receiving both ARNI and SGLT2i for at least three months; and no heart-failure hospitalization within the prior six months. Major exclusions include irreversible cardiomyopathy, incomplete revascularization, residual moderate or greater valvular regurgitation, eGFR \<30 mL/min/1.73 m², symptomatic hypotension or bradycardia, pregnancy, and limited life expectancy. Follow-up Scheduled visits at 1, 3, 6, 9, and 12 months. Assessments include vital signs, NT-proBNP, renal function, electrolytes, echocardiography (baseline, 1, 3, 6, 12 months), ECG, KCCQ-12, 6-minute walk test, and adverse event capture. Statistical Analysis The pilot sample size of 40 per arm (total 80) is based on precision estimates for the primary outcome: assuming a 10% event rate, this yields approximately ±9.3% precision per arm and ±6.6% overall (Wilson method), which is adequate for variance estimation. A 15% drop-out rate is anticipated. All analyses will be conducted with SPSS and R. Expected Contribution By generating the first prospective evidence on stepwise ARNI/SGLT2i withdrawal in patients with structurally corrected HFimpEF, this pilot trial aims to lay the methodological foundation for a definitive non-inferiority trial and ultimately to inform evidence-based deprescribing guidelines for this growing patient population.