Changes in the concentrations of creatinine, cystatin C and NGAL in patients with acute paraquat self-poisoning

Abstract

An increase in creatinine > 3 μmol/L/h has been suggested to predict death in patients with paraquat self-poisoning and the value of other plasma biomarkers of acute kidney injury has not been assessed. The aim of this study was to validate the predictive value of serial creatinine concentrations and to study the utility of cystatin C and neutrophil gelatinase-associated lipocalin (NGAL) as predictors of outcome in patients with acute paraquat poisoning. The rate of change of creatinine (dCr/dt) and cystatin C (dCyC/dt) concentrations were compared between survivors and deaths. Receiver-operating characteristic (ROC) curves were constructed to determine the best threshold for predicting death. Paraquat was detected in 20 patients and 7 of these died between 18 h and 20 days post-ingestion. The dCr/dt ROC curve had an area of 0.93 and the cut-off was > 4.3 μmol/L/h (sensitivity 100%, specificity 85%, likelihood ratio 7). The dCyC/dt ROC curve had an area of 0.97 and the cutoff was > 0.009 mg/L/h (sensitivity 100%, specificity 91%, likelihood ratio 11). NGAL did not separate survivors from deaths. Death due to acute paraquat poisoning is associated with changes in creatinine and cystatin concentrations. Further validation of these measurements is needed before they can be adopted in guiding intensive treatments.

Fig. 1

(a) Serial plasma creatinine concentrations in patients with acute paraquat poisoning, relative to outcome. (b) Serial plasma cystatin C concentrations in patients with acute paraquat poisoning, relative to outcome. (c) Serial plasma NGAL concentrations in patients with acute paraquat poisoning, relative to outcome. Values less than the reference range (<60 ng/mL) are plotted at 30 ng/mL for illustrative purposes. () Reference range; () death; () alive.

Fig. 2

(a) Receiver-operator characteristic curve of dCr/dt for predicting death using linear regression of data within 75 h of poisoning. (b) Receiver-operator characteristic curve of dCyC/dt for predicting death using linear regression data within 75 h of poisoning.

Fig. 3

(a) Influence of the time of the second sample on the estimated rate of change of the concentration of creatinine (dCr/dt, upper graph) and cystatin C (dCyC/dt, lower graph) in patients who died. Arrow depicts the last value below the cut-off threshold determined by the ROC. (b) Influence of the time of the second sample on the estimated rate of change of the concentration of creatinine (dCr/dt, upper graph) and cystatin C (dCyC/dt, lower graph) in patients who survived.

Fig. 4

Serial changes in paraquat, creatinine and cystatin C concentrations for three patients with fatal outcome. The mean rates of change of creatinine and cystatin C concentrations in each individual are shown, (–■–) Paraquat; (--▴--) creatinine; (··○··) cystatin C.

Fig. 5

Relationship between dCr/dt and dCyC/dt in patients with acute paraquat poisoning.