RESUMEN
OBJECTIVES: Neurologic outcomes of patients under venoarterial extracorporeal membrane oxygenation (VA-ECMO) may be worsened by secondary insults of systemic origin. We aimed to assess whether sepsis, commonly observed during ECMO support, is associated with brain injury and outcomes. DESIGN: Single-center cohort study of the "exposed-non-exposed" type on consecutive adult patients treated by VA-ECMO. SETTING: Medical ICU of a university hospital, France, 2013-2020. PATIENTS: Patients with sepsis at the time of VA-ECMO cannulation ("sepsis" group) were compared with patients without sepsis ("no sepsis" group). The primary outcome measure was poor functional outcome at 90 days, defined by a score greater than or equal to 4 on the modified Rankin scale (mRS), indicating severe disability or death. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: A total of 196 patients were included ("sepsis," n = 128; "no sepsis," n = 68), of whom 87 (44.4%) had presented cardiac arrest before VA-ECMO cannulation. A poor functional outcome (mRS ≥ 4) was observed in 99 of 128 patients (77.3%) of the "sepsis" group and 46 of 68 patients (67.6%) of the "no sepsis" group (adjusted logistic regression odds ratio (OR) 1.21, 95% CI, 0.58-2.47; inverse probability of treatment weighting (IPTW) OR 1.24; 95% CI, 0.79-1.95). Subsequent analyses performed according to pre-ECMO cardiac arrest status suggested that sepsis was independently associated with poorer functional outcomes in the subgroup of patients who had experienced pre-ECMO cardiac arrest (adjusted logistic regression OR 3.44; 95% CI, 1.06-11.40; IPTW OR 3.52; 95% CI, 1.68-7.73), whereas no such association was observed in patients without pre-ECMO cardiac arrest (adjusted logistic regression OR 0.69; 95% CI, 0.27-1.69; IPTW OR 0.76; 95% CI, 0.42-1.35). Compared with the "no sepsis" group, "sepsis" patients presented a significant increase in S100 calcium-binding protein beta concentrations at day 1 (0.94 µg/L vs. 0.52 µg/L, p = 0.03), and more frequent EEG alterations (i.e., severe slowing, discontinuous background, and a lower prevalence of sleep patterns), suggesting brain injury. CONCLUSION: We observed a detrimental role of sepsis on neurologic outcomes in the subgroup of patients who had experienced pre-ECMO cardiac arrest, but not in other patients.
