FO066SERUM CYSTATIN C AN ACCURATE EARLY KIDNEY FUCTION MARKER IN INTENSIVE CARE UNIT SETTING

Abstract

Introduction and Aims: Accurate and early diagnosis ofacutekidney injury (AKI) in critically ill patients is challenging. Serum cystatin C (SCysC) performance as a diagnostic tool for AKI in ICU is still unknown.Our aim was to determine the prediction ability of SCysC for AKI diagnosis in a longitudinal analysis of daily data of critically ill patients and additionally, to determine whether fluid balance changed SCysC accuracy for AKI diagnosis and staging. Methods: In a prospective, observational study of 128 consecutive ICU adult patients, the daily performance of serum creatinine (SCr) and SCysC was assessed. Based on baseline SCr and its evolution during ICU stay, all patients were adjudicated to one of four groups: Normal Kidney Function (NF), Stable Chronic Kidney Disease (sCKD according to K/DOQI), Transitory azotemia (TAz) or AKI (according to KDIGO classification). Alive patients were followed after hospital discharge up to 4 years.To estimate the impact of SCysC on AKI diagnosis, a longitudinal analysis was performed with a population-average model estimated through generalized estimating equations (GEE) with logit link and exchangeable working correlation structure. Results: We included 128 patients with median age 60.5 years (IQR:43-73), 61.7% male, and 85.2% Non-black. Main comorbidities were cardiovascular disease (54%), diabetes (27%) and chronic obstructive lung disease (16%) with no significant difference between groups. During ICU stay 41% of patients developed AKI; 15% TAz; 3% remained as sCKD, and 41% NF. Median baseline eGFR by group were 76.2; 92.3, 49.3 and 102mL/min/1.73m2, respectively. The highest APACHEII score corresponded to AKI group (31 (IQR25-37.5)).The time course of SCysC was significantly higher in patients who developed AKI with median values 2.2mg/L (IQR:1.7-2.8) and the discriminative ability corresponded to AUC-ROC: 0.91 (CI 0.86-0.96). In longitudinal GEE analysis SCysC predicted AKI at 1; 2; or even 3 days before AKI-KDIGO criteria, decreasing its adjusted Odds with time (5.2 CI:2.9-9.4; 3.6 CI:2.0-6.2; 1.7 CI:1.2-2.3 respectively) p<0.001 for all. A relative longitudinal analysis of SCysC was applied and by Generalized additive model (GAM) the cut-off point of Relative SCysC corresponded to 25%. In a multivariable model, a higher than 25% change of Relative SCysC, was associated with 2-fold increase of AKI development. Regarding fluid balance influence on SCysC performance no significant changes were found, whereas SCr and AKI classification and staging were significantly influenced by fluid overload (p<0.001).The kidney function decreased markedly in AKI patients through time, comparing with other groups, being the median eGFR at discharge 42.6 (IQR 30.3-66.6) and 38.5 (IQR 14.4-65.6) mL/min/1.73 m2 at the last follow-up (p<0.001). Conclusions: This study ascertains a longitudinal approach for early AKI diagnosis based on daily SCysC, demonstrating a high discriminative and prediction ability. We provide the first prospective results with a cut point of the relative changes of SCysC that can be useful in clinical practice: changes higher than 25%, between two consecutive days could be a good tool for early detection of AKI in ICU. Finally, the effect of fluid balance did not change the SCysC ability for AKI diagnosis.