ABSTRACT NUMBER: ESOC2026A2392 EARLY AEROBIC FITNESS TRAINING AFTER STROKE (EARLY-FIT TRIAL): FEASIBILITY AND SAFETY OF VERY EARLY AEROBIC TRAINING IN CARDIOMETABOLIC HIGH-RISK STROKE PATIENTS

Abstract Background and aims Early aerobic training after stroke is recommended to improve physical fitness and functional recovery. However, the PHYS-STROKE trial yielded negative results and raised safety concerns in cardiometabolic high-risk patients. Early-Fit investigates the feasibility and safety of very early, progressive aerobic training in a selected high-risk population and explores glycaemic variability as a mechanistic marker of cardiometabolic stress. Methods Early-Fit (DRKS00032233) is a randomized controlled feasibility study enrolling patients undergoing in-patient neurological rehabilitation within 3–21 days after stroke. Key inclusion criteria were prediabetes/diabetes, and/or atrial fibrillation, and the ability to take at least one step with assistance. Patients were randomized to treadmill-based training (target ≥6 sessions) or usual care. Progressive training intensity was prescribed using heart rate reserve. Continuous glucose monitoring was applied in both groups. We report feasibility and safety outcomes. Results Between 11/2023 and 10/2025, 40 patients were randomized at a mean of 9 days after stroke. Mean (SD) age was 79.2(6) years and median [IQR] NIHSS was 5[3-6]. Median completed sessions were 6 [4.5–6], with 34% discontinued and target heart rate achieved in 75% of sessions. The study identified frequent intercurrent infections, technical constraints affecting cardiopulmonary exercise testing, and restricted length of stay in early rehabilitation as key factors of intervention delivery. No intervention-related safety signals were observed. Conclusions Early-Fit demonstrates that very early aerobic training in cardiometabolic high-risk stroke patients can be feasible and safe but is strongly influenced by intercurrent medical events and rehabilitation settings. Findings define key prerequisites for future trial design. Conflict of interest All authors have nothing to disclose