Predictive survival models have been proposed to esti-mate survival rates in patients with refractory cardio-genic shock considered for veno-arterial extracorporeal membrane oxygenation (VA-ECMO) [1, 2]. Notably, the Survival after VA-ECMO (SAVE) score stands out as a potential tool to optimize resource allocation, enable risk-adjusted comparisons among healthcare facilities, and assist clinicians in identifying patients who may likely benefit from ECMO. Despite the emergence of AI-driven ECMO survival scores [3], current models do not consider patient immune status, thus raising concerns regarding their applicability to immunocompromised patients. In a preplanned ancillary analysis conducted within a cohort of 177 immunocompromised patients undergoing VA-ECMO [4], we aimed to evaluate the per-formance of the SAVE relative to common intensive care unit (ICU) severity scores within this specific population.For each patient, we recalculated the SAVE score, the Simplified Acute Physiology Score (SAPS II), the Acute Physiology and Chronic Health Evaluation (APACHE) II score, and the Sequential Organ Failure Assessment (SOFA) score, at ICU admission or upon cannulation. Lacking pre-ECMO peak inspiratory pressure data, a zero value was assigned for this variable in the SAVE score calculation. We assessed the discriminatory power of each score for predicting 90-day survival using the area under the receiver operating characteristic curves (AUROC) and compared them via the De Long test. Cali-bration and prediction accuracy were further evaluated through the Hosmer–Lemeshow test and the Brier score, respectively. The study followed TRIPOD recommenda-tions for prediction model development