Electronic supplementary material (ICM-2009-01010)
Material and methods
Biochemical analysis
Samples of blood and urine were collected twice daily until discharge from the ICU or earlier if renal replacement therapy was initiated. Samples were immediately centrifuged at 2000 rpm at 4˚C for 10 min. The supernatant plasma and urine were aliquoted into cryovials and stored at -80˚C. Plasma and urine samples were analyzed at the department of Clinical Chemistry, UppsalaUniversityHospital, Uppsala, Sweden. Plasma was analyzed for NGAL, procalcitonin (PCT), C-reactive protein (CRP), myeloperoxidase (MPO) and cystatin C. Urine was analyzed for NGAL, creatinine, cystatin C and α1-microglobulin.
NGAL in plasma (pNGAL) and urine (uNGAL) were measured by the polyclonal antibody based RIA[1-2]. The intra- and inter-assay coefficients of variation were less than 6% and 10%, respectively. Urine creatinine was measured on the Architect® instrument (Abbott Laboratories) and served to correct the urine levels of NGAL for variations in urine dilutions. Urine levels of HNL/NGAL are consistently expressed in ng/mg creatinine in this paper. Expected normal pNGAL level was <73.5 ng/mL[1]. The upper 97.5 percentile of 101 healthy controls (141 ng/mg creatinine) was used as reference limit for uNGAL (Venge P and Xu SY, unpublished).
Plasma MPO was measured by enzyme-linked immunosorbent assay (Diagnostics development, Uppsala, Sweden). Plasma PCT, CRP and cystatin C as well as urine cystatin C and α1-microglobulin were measured at the Department of Clinical Chemistry, Uppsala University Hospital, Sweden. Urine cystatin C and α1-microglobulin were corrected for urine creatinine and were expressed in mg/g creatinine. Expected normal plasma levels were <55.4 ng/mL for MPO, <0.05 ng/mL for PCT, <5 mg/L for CRP, <1.55 mg/L (>50 years of age) and <1.2 mg/L (<50 years of age) for cystatin C and <100 μmol/L (men) and <90 μmol/L (women) for creatinine. Expected normal urine levels were <6.2 mg/g creatinine for cystatin C and <6.2 mg/g creatinine for α1-microglobulin.
Statistical analysis
Data were analyzed using STATA® for Windows version 10.1 software (Stata Corp., College Station, TX, USA). Continuous variables were expressed as median and interquartile range and categorical variables as n (%). The Chi-square test and Kruskal-Wallis test (for multiple groups) and Fisher’s exact test and Mann-Whitney test (for two groups) were used for comparison between categorical and continuous values respectively. The diagnostic characteristics of pNGAL and uNGAL in predicting AKI were assessed by calculation of the area under the receiver operating characteristic curve (AUC-ROC). AUC-ROC analysis was performed by (1) comparing AKI-patients with all non-AKI patients and by (2) comparing AKI-patients with those non-AKI patients with septic shock. The pNGAL and uNGAL concentrations corresponding to the optimal combination of sensitivity and specificity were used as cut-off levels for calculation of AUC-ROC estimates at various timepoints before AKI was first diagnosed. An AUC-ROC value above 0.90 was considered excellent, a value of 0.80 to 0.89 good, a value of 0.70 to 0.79 fair, a value of 0.60 to 0.69 poor and a value below 0.59 useless. A p-value less than 0.05 were considered statistically significant.
References
1.Xu SY, Petersson CG, Carlson M, Venge P (1994) The development of an assay for human neutrophil lipocalin (HNL)--to be used as a specific marker of neutrophil activity in vivo and vitro. J Immunol Methods 171: 245-252
2.Cai L, Borowiec J, Xu S, Han W, Venge P (2009) Assays of urine levels of HNL/NGAL in patients undergoing cardiac surgery and the impact of antibody configuration on their clinical performances. Clin Chim Acta 403: 121-125