Early Pan-Immune-Inflammation Value Trajectories are Associated with Unfavorable Neurological Outcomes After Extracorporeal Cardiopulmonary Resuscitation.

Background: Neurological outcome is key to extracorporeal cardiopulmonary resuscitation (ECPR) benefit. However, the prognostic relevance of early PIV dynamics after ECPR remains poorly characterized. Methods: We conducted an exploratory retrospective single-center study of 107 ECPR patients with at least three valid complete blood count (CBC) measurements within 24 h of extracorporeal membrane oxygenation (ECMO) pump-on. The primary outcome was an unfavorable neurological outcome at hospital discharge, defined as Cerebral Performance Category (CPC) 3-5. Latent class mixed models (LCMMs) were used to identify early pan-immune-inflammation value (PIV) trajectory groups, and multivariable logistic regression evaluated their association with the outcome. Incremental predictive performance was explored using leakage-controlled nested cross-validation in which trajectory modeling was repeated within each outer training fold. Results: Among 107 included patients, 53 had an unfavorable neurological outcome at hospital discharge. Two distinct early PIV trajectory patterns were identified in this selected cohort: a high-baseline decreasing class (n=68) and a low-baseline rising class (n=39). The low-baseline rising class had a significantly higher rate of unfavorable neurological outcome than the high-baseline decreasing class (69.2% vs 38.2%, P=0.004). In the fully adjusted model, the low-baseline rising trajectory was associated with significantly higher odds of unfavorable neurological outcome compared with the high-baseline decreasing trajectory (odds ratio 5.70, 95% confidence interval 1.73-18.81, P=0.004). In an internal exploratory comparison, adding the trajectory class to the clinical risk model increased the area under the receiver operating characteristic curve from 0.746 to 0.812, but the bootstrap 95% CI for the AUC increment crossed zero. Conclusion: Two exploratory early PIV trajectory groups were identified in this selected cohort. The low-baseline rising pattern was associated with a higher probability of CPC-defined unfavorable neurological outcome at hospital discharge in this selected single-center cohort. Although trajectory information numerically improved internal predictive performance, the incremental improvement was uncertain and requires external validation.