Intravascular hemolysis and clinical outcomes in continuous-flow LVAD patients: A single-center retrospective analysis.

OBJECTIVE: To determine the incidence of intravascular hemolysis in patients implanted with continuous-flow left ventricular assist devices (LVADs), to describe longitudinal changes in hemolysis-related laboratory markers (LDH, total/direct bilirubin, INR), and to assess their diagnostic value for pump thrombosis and cerebrovascular events. METHODS: We retrospectively analyzed 100 end-stage heart failure patients (2014-2015). To ensure consistent longitudinal analysis, 59 long-term survivors (42 HeartWare HVAD, 17 HeartMate II) were included. For each patient, a dynamic LDH increase ratio (peak/preoperative baseline) was calculated. The diagnostic performance of the criterion "LDH >250 U/L and ⩾4-fold increase from baseline" was evaluated. Cox regression, ROC analysis, and Kaplan-Meier survival analysis were performed. RESULTS: During 1-year follow-up, 17 patients experienced thrombotic or cerebrovascular events (8 confirmed pump thrombosis). While absolute LDH values at fixed time points were not discriminative, the dynamic LDH increase ratio was significantly higher in the thrombosis group (p = 0.0102), with an AUC of 0.840. Preoperative LDH was a significant risk factor in Cox regression (HR: 1.005, p = 0.009), reflecting baseline disease severity. The ⩾4-fold LDH increase criterion demonstrated a Negative Predictive Value (NPV) of 97.6%, indicating excellent rule-out performance. No significant difference was found between device types (p > 0.05), supporting a device-independent biomarker dynamic. CONCLUSION: Baseline LDH alone is not a reliable predictor of pump thrombosis. In contrast, a ⩾4-fold dynamic rise in LDH relative to the patient's own baseline provides excellent negative predictive value. This dynamic surge serves as a practical, device-independent decision-support tool to rule out pump thrombosis with high confidence in all cf-LVAD patients.