Dynamics in kidney function markers as predictors of post-ICU survival: an innovative approach to enhance marker accuracy

Introduction and aims The variability of physiological biomarkers is emerging as an independent indicator of health across various contexts. In intensive care unit (ICU) patients, high variability in serum creatinine, regardless of absolute values, has been associated with higher mortality. This study investigates the dynamics of blood and urine kidney function markers in ICU patients. Methods In a post hoc analysis of the FROG-ICU cohort, focusing on patients with blood and urine samples at ICU admission and discharge, we investigate the relationships among kidney function markers, their dynamics between the two time points, and their association with kidney outcomes and survival. As a post hoc , exploratory analysis, all findings should be considered hypothesis-generating and require prospective validation. Results At ICU admission, only serum Proenkephalin (PENK) values demonstrated a weak association with short-term post-ICU survival. In contrast, urine peptidome score CKD273 and cystatin C levels measured at ICU discharge were strongly associated with one-month survival. The differences in CKD273, Galectin-3, and plasma neutrophil gelatinase-associated lipocalin (NGAL) between admission and discharge were the most predictive of one-month survival, independent of baseline values. Furthermore, the slopes of CKD273 and plasma NGAL trajectories were prognostic. Absolute dynamic changes in urine NGAL were also linked to poorer short-term survival. Long-term post-ICU survival showed patterns similar to short-term outcomes with respect to biomarker differences during the ICU stay. When examining absolute changes, pronounced dynamics in PENK across the ICU course were associated with worse one-year survival. Conclusion Both the magnitude and dynamics of kidney function markers may serve as independent predictors of short and long-term post-ICU mortality. As a post hoc exploratory study, this study is hypothesis-generating and requires prospective validation. It highlights the importance of tracking biomarker fluctuations in ICU patients rather than relying solely on static measurements.