078 - Cerebral oxygen saturation as a predictor of acute brain injury in neonates on ECMO

Publication Number: 78.4291

Kyusang Yoo, Asan Medical Center Children’s Hospital, Seould, Seoul-t'ukpyolsi, Republic of Korea; Soo Hyun Kim, Asan Medical Center, Seoul, Seoul-t'ukpyolsi, Republic of Korea; Keon Hee Seol, Asan medical center, Seoul, Seoul-t'ukpyolsi, Republic of Korea; Jung-il Kwak, Asan medical center, Seoul, Seoul-t'ukpyolsi, Republic of Korea; Taegyeong Kim, Seoul Asan medical center, Seoul, Seoul-t'ukpyolsi, Republic of Korea; Bo Kyeong Jin, University of Ulsan College of Medicine, Seoul, Korea, Seoul, Seoul-t'ukpyolsi, Republic of Korea; Joo Hyung Roh, Asan medical center, Songpa-gu, Seoul-t'ukpyolsi, Republic of Korea; Ha Na Lee, Asan medical center, children’s hospital, Gangdong-gu, Seoul-t'ukpyolsi, Republic of Korea; Jiyoon Jeong, Asan Medical Center, Seoul, Seoul-t'ukpyolsi, Republic of Korea; Chae Young Kim, children hospital, seoul, Seoul-t'ukpyolsi, Republic of Korea; Euiseok Jung, Asan Medical Center, Seoul, Seoul-t'ukpyolsi, Republic of Korea; Byong Sop Lee, Asan Medical Center, Seoul, Seoul-t'ukpyolsi, Republic of Korea

Background: Neonates undergoing extracorporeal membrane oxygenation (ECMO) have a higher incidence of brain injury, necessitating continuous cerebral monitoring. Near-infrared spectroscopy (NIRS) offers non-invasive, continuous evaluation of cerebral oxygenation, potentially aiding in the early detection of acute brain injury.

Objective: This study aims to evaluate the utility of cerebral NIRS monitoring in predicting moderate to severe brain injury in neonates treated with ECMO.

Design/Methods: We retrospectively reviewed medical records from a single tertiary center. Neonates admitted to the NICU from November 2018 to June 2024 who received ECMO and NIRS monitoring were included. Patients who experienced cardiac arrest prior to ECMO were excluded. Brain MRI and ultrasonography were performed to evaluate brain injury as needed. Infants were categorized into two groups based on neuroimaging scores: normal to mild brain injury vs. moderate to severe brain injury. Regional cerebral oxygen saturation (rScO2) data were compared between these groups.

Results: Thirty-three infants were included in the study, with a median gestational age and birth weight of 38.1 weeks and 2938 grams, respectively. The mortality rate was 33.8%. The average duration of ECMO support was 15.2 ± 8.0 days, with ECMO initiated at a median age of 2 days. Thirteen infants (39.4%) were classified as having moderate to severe brain injury. This group had significantly lower minimum rScO2 values (20 vs. 46, p=0.016) and a higher hypoxia burden, defined as the percentage of time with rScO2 below 50% (407 vs. 6, p=0.034). They also had longer durations of rScO2 decline greater than 20% from baseline (80 vs. 20 hrs, p=0.014). ROC analysis identified optimal cutoffs for key parameters, including a minimum rScO2 cutoff of 22.5% (AUC 0.75), a duration of rScO2 < 50% at 42 hours (AUC 0.727), and a hypoxia burden of 305.5 (AUC 0.719), demonstrating their predictive value for brain injury outcomes.

Conclusion(s): Cerebral NIRS monitoring effectively predicts moderate to severe brain injury in neonates undergoing ECMO. Key indicators include significantly lower minimum rScO2 values, higher hypoxia burden, and longer durations of rScO2 decline. These findings support the use of NIRS for early detection and risk assessment in this population.