Predictors of left ventricular ejection fraction recovery after guideline-directed medical therapy in patients with newly diagnosed dilated cardiomyopathy and baseline LVEF ≤35%

Background Patients with newly diagnosed dilated cardiomyopathy (DCM) and a baseline left ventricular ejection fraction (LVEF) ≤35% generally have a poor prognosis. Early identification of patients who are unlikely to recover LVEF after guideline-directed medical therapy (GDMT) is clinically important for risk stratification, closer follow-up, repeated LVEF assessment, and timely reassessment of guideline-based device therapy eligibility. Methods This retrospective cohort study enrolled 198 patients with newly diagnosed dilated cardiomyopathy (DCM) and baseline left ventricular ejection fraction (LVEF) ≤35%, all of whom received standardized guideline-directed medical therapy (GDMT) for 3∼6 months. Baseline and follow-up echocardiographic, electrocardiographic, clinical, and laboratory data were collected. The primary outcome was recovery of LVEF to 35% after treatment. The secondary outcome was the first occurrence of cardiovascular death or rehospitalization for heart failure during follow-up. Multivariable Firth Logistic regression was used to identify independent predictors and to develop a prediction model, which was internally validated using the area under the receiver operating characteristic curve (ROC AUC), the Hosmer–Lemeshow goodness-of-fit test, and decision curve analysis. Cox proportional hazards regression was used to evaluate predictors of long-term prognosis, and time-dependent ROC analysis was performed. Results Multivariable Firth logistic regression showed that absence of left bundle branch block (LBBB) and higher apolipoprotein A1 (ApoA1) levels were independently associated with LVEF recovery to 35%. The presence of LBBB was strongly associated with a lower likelihood of LVEF recovery (OR = 0.02, 95% CI 0.00–0.13, P 0.001), whereas higher ApoA1 levels were associated with a greater likelihood of recovery (OR = 1.40 per 10-unit increase, 95% CI 1.06–1.86, P = 0.019). A model integrating LBBB, ApoA1, and age demonstrated good discrimination, with ROC AUCs of 0.93 in the training set and 0.88 in the validation set, good calibration, and positive net benefit on decision curve analysis. Conclusions The model may help with early risk stratification, closer surveillance, repeated LVEF assessment, and timely reassessment of guideline-based ICD eligibility. However, external validation is required before routine clinical application.