Autonomic impairment in advanced heart-failure patients revealed by nonlinear heart rate variability measured during exercise

Advanced heart failure is characterized by profound autonomic dysfunction, yet resting measures of heart rate variability (HRV) provide limited insight into autonomic adaptability under physiological stress. We investigated whether graded exercise unmasks alterations in cardiac autonomic regulation in patients with advanced heart failure using both conventional and information-theoretic HRV measures. 77 patients with end-stage heart failure with reduced ejection fraction underwent resting and cycle-ergometer exercise testing with continuous electrocardiographic recording. HRV was assessed at rest, during early exercise and at peak exercise using time- and frequency-domain indices alongside entropy-based measures of cardiac rhythm complexity. Clinical follow-up was performed for a composite adverse outcome, defined as cardiovascular death, comprising cardiac-related death, heart transplantation or left ventricular assist device implantation. During graded exercise, conventional HRV measures showed the expected workload-dependent attenuation, whereas entropy-based metrics revealed a progressive increase in signal complexity and a reduction in heartbeat predictability. Notably, during the early phase of exercise, patients who subsequently experienced adverse outcomes (n = 17) exhibited lower conditional entropy and higher self-entropy compared with survivors, while no differences were observed at rest or peak exercise. In survival analyses, self-entropy measured during early exercise remained independently associated with adverse outcomes after adjustment for clinical and exercise-derived variables. These findings indicate that early exercise represents a sensitive physiological window for revealing impaired autonomic adaptability in advanced heart failure. Information-theoretic analysis of HRV during submaximal exercise may provide complementary insights into cardiac autonomic regulation that are not captured by conventional resting or peak exercise measures.