From Recovery to Relapse: A Systematic Review and Meta-Analysis of Guideline-Directed and Targeted Therapy De-Escalation and Phenotypic Variability in Heart Failure with Improved Ejection Fraction

(1) Background: Heart failure with improved ejection fraction (HFimpEF) is a heterogeneous state of remission rather than a cure; the safety of withdrawing guideline-directed medical therapy (GDMT) after left ventricular ejection fraction (LVEF) recovery remains uncertain. (2) Methods: PubMed/MEDLINE, Cochrane CENTRAL and the Cochrane Database of Systematic Reviews were searched to 9 November 2025, with citation searching. Forty-four studies were retained: 38 contributing extractable outcome data (approximately 6665 patients) and six unselected cohorts (17,855 patients) used descriptively. (3) Results: In an exploratory pooling of two rhythm control trials, cardiovascular death was less frequent (RR 0.65, 95% CI 0.50–0.84), whereas the estimate for worsening heart failure hospitalization (RR 0.75, 95% CI 0.63–0.90) was sensitive to the pooling model. An indirect cross-study comparison suggested differences in relapse after withdrawal but cannot establish causality. In the complete recovery subgroup, study-defined adverse events occurred in 27/140 patients after withdrawal versus 2/76 with continuation; differing designs and definitions limit interpretation. The withdrawal-specific predictive performance of biomarkers remains unvalidated. (4) Conclusions: Current evidence is insufficient to establish the long-term safety of GDMT withdrawal in any etiological phenotype. Recovery evidence must be distinguished from direct withdrawal evidence, and adequately powered, phenotype-stratified randomized withdrawal trials are needed.