Abstract
Background/Objectives: Preoperative hypoalbuminemia is a known risk factor for adverse outcomes in cardiac surgery, but its role in patients undergoing HeartMate 3 (HM3) left ventricular assist device (LVAD) implantation is unclear. This study evaluated the association between albumin levels and postoperative outcomes, aiming to define a clinically meaningful cut-off for risk stratification. Methods: We retrospectively analyzed 205 adult patients who underwent HM3 implantation at a single center from June 2014 to December 2023. Receiver operating characteristic (ROC) analysis identified an optimal pre-implant albumin cut-off of <32 g/L. This threshold, derived using the maximal Youden Index, provided a sensitivity of 52.1%, specificity of 71.6%, and an AUC of 0.64 (95% CI 0.56-0.71), with internal bootstrapping validation confirming model stability, and calibration demonstrating good agreement between predicted and observed outcomes. Kaplan-Meier analysis assessed freedom from hemocompatibility-related adverse events (HRAEs) and survival. Cox proportional hazards models evaluated albumin and other variables as independent risk factors for HRAEs. Results: Patients with pre-implant albumin <32 g/L had higher rates of HRAEs, including stroke (24.9% vs. 8.4%, p = 0.004) and bleeding (38.1% vs. 23.2%, p = 0.012). Freedom from HRAEs was significantly lower in the hypoalbuminemia group (45.2% vs. 69.8%, p < 0.001) and competing risk-adjusted cumulative incidence for HRAE was higher, but did not reach statistical significance (p = 0.11), one-year HRAE-free survival was also reduced (68.5% vs. 85.7%, p = 0.03). In multivariable analysis, low albumin (HR 0.56, 95% CI 0.33-0.93, p = 0.026) and temporary right ventricular assist device (RVAD) support (HR 3.32, 95% CI 2.05-5.39, p < 0.001) were independent predictors of HRAEs. Conclusions: Low preoperative albumin is independently associated with increased HRAEs and reduced one-year survival after HM3 implantation. Compared with the traditional 35 g/L threshold, the ROC-derived 32 g/L cut-off offered superior balance between sensitivity and specificity, underscoring its clinical utility. Albumin may serve as a simple, pragmatic, and cost-effective biomarker for preoperative risk assessment and optimization.