Enhancing the performance of GFR prediction equations in children by combining a cystatin C-based with a creatinine-based equation

Introduction: Estimates of glomerular filtration rate (GFR) are considered the best overall indices of kidney function. Exceptional dietary intake or muscle mass hampers the use of creatinine-based estimating equations, whereas cystatin C-based equations are affected by steroid therapy. Combining a creatinine-based with a cystatin C-based equation may improve prediction of GFR in children. Methods: In 331 children, GFR was estimated using the creatinine-based Schwartz equation (eGFRcrea) and the cystatin C-based Filler equation (eGFRcys). In addition GFR was measured by single-injection inulin clearance as gold standard. In case of agreement between the estimates (defined as eGFRcys within 30% of eGFRcrea) the arithmetic mean (eGFRmean) of the two was used to predict GFR. If not, the estimate not affected by known confounders such as muscle dystrophy, malignancy or steroid therapy was used. Results: In 226 children (68.3%), eGFRmean was used because of good agreement between eGFRcys and eGFRcrea. It outperformed eGFRcys and eGFRcrea with 80.7% vs. respectively 76.1% and 75.5% of measurements within the 30% limits of measured GFR. In the remaining 105 children (31.7%) with considerable difference between eGFRcrea and eGFRcys, the latter performed better in children with cancer (64.3% vs. 38.1%) and severe wasting (71.4% vs. 57.1%). eGFRcrea outperformed eGFRcys in children receiving steroids (75% vs. 25%). When comparing bias and precision, similar results were obtained. Conclusions: If there is no considerable difference between eGFRcrea and eGFRcys their mean is the best prediction of GFR in children. Strongly discrepant results should prompt a search for known confounders to choose whether to use eGFRcrea or eGFRcys.

creatinine - cystatin C - glomerular filtration rate - kidney function