48-hour lactate-to-albumin ratio predicts 180-day mortality after microaxial flow pump support in cardiogenic shock

Introduction Cardiogenic shock (CS) remains associated with high mortality despite increasing use of microaxial flow pump (mAFP) support, with or without veno-arterial extracorporeal membrane oxygenation (VA-ECMO). Biomarkers measured at initial assessment may not capture early response to MCS support. The lactate-to-albumin ratio (LAR) integrates metabolic stress and inflammatory burden and has shown prognostic value in several critical illness populations; however, its value as a serial marker during mAFP support in CS is uncertain. Purpose To evaluate the prognostic utility of serial LAR after mAFP initiation and to identify the optimal time point for predicting 180-day outcomes in CS. Methods The UNLOADERS-PVAD registry is a multi-center observational study in Japan. Between April 2018 and December 2023, 501 patients received mAFP support; 32 who died within 48 hours were excluded, leaving 469 patients for analysis. LAR was calculated as lactate/albumin. Serial LAR was assessed at pre-support and at 24, 48, and 72 hours after mAFP initiation. The primary endpoint was 180-day all-cause mortality. Discrimination was assessed using receiver operating characteristic (ROC) analysis and area under the curve (AUC). Kaplan–Meier analyses were used for 180-day all-cause mortality. Independent associations were evaluated using stepwise Cox proportional hazards regression adjusting for demographics, etiology and extracorporeal cardiopulmonary resuscitation (E-CPR), support strategy (mAFP alone vs mAFP with VA-ECMO), and severity indices (vasoactive–inotropic score, APACHE II, SAVE score, and pre-MCS SCAI stage). Results Among 469 patients, LAR measured at 48 hours demonstrated the highest discriminative ability for 180-day mortality among serial measurements (AUC 0.739 vs 0.657 at pre-support, 0.705 at 24 hours, and 0.712 at 72 hours). When patients were stratified by 48-hour LAR quartiles, 180-day survival decreased stepwise with increasing quartiles (log-rank p<0.001). In stepwise Cox models with comprehensive adjustment up to the final model incorporating support strategy and multiple severity scores, 48-hour LAR remained independently associated with 180-day mortality (hazard ratio per 0.1 increase 1.051, 95% confidence interval 1.039–1.064; p<0.001). Conclusion Serial LAR assessment after mAFP initiation identifies prognostic separation, with 48-hour LAR providing the strongest prediction of 180-day mortality. This simple composite marker may serve as an early treatment-response biomarker to refine risk stratification in mechanically supported CS.