New Model for Estimating Glomerular Filtration Rate and Accurately Dosing Carboplatin in Cancer Patients


Key Points

  • A new model for estimating glomerular filtration rate was more accurate than published models compared with 51Cr-EDTA GFR.
  • The new model reduced the incidence of carboplatin dosing error compared with published models.

Janowitz et al have developed a new model for estimating glomerular filtration rate (GFR) that improves accuracy compared with chromium-51 (51Cr) EDTA excretion measurements and accuracy of carboplatin dosing compared with other published models in cancer patients. Their findings were reported in the Journal of Clinical Oncology.

Study Details

In the study, data on age, sex, height, weight, serum creatinine concentrations, and results for GFR from 51Cr EDTA excretion measurements (51Cr-EDTA GFR) were obtained from 2,471 white patients aged ≥ 18 years with histologically confirmed cancer diagnoses at the Cambridge University Hospital NHS Trust. A new multivariable linear model for GFR was developed using statistical regression analysis in development, internal validation, and external validation sets.

51Cr-EDTA GFR was compared with estimated GFR (eGFR) from seven published models and the new model using the root-mean-squared-error and median residual. A comparison of carboplatin dosing accuracy was performed on the basis of an absolute percentage error > 20%.

Performance of New Model

The final coefficients of the new model consisted of intercept, age, body surface area (BSA), ln creatinine (Cre), ln(Cre)2, ln(Cre)3, male sex (variable = 1 for male, 0 for female), interaction of age and male sex, and interaction of age and body surface area.

The new model improved eGFR accuracy (root-mean-squared-error = 15.00 mL/min, 95% confidence interval [CI] = 14.12–16.00 mL/min) compared with all published models. The BSA-adjusted chronic kidney disease epidemiology model was the most accurate published model (root-mean-squared-error = 16.30 mL/min, 95% CI = 15.34–17.38 mL/min). The new model reduced the proportion of patients with carboplatin dose absolute percentage error > 20% to 14.17% vs 18.62% for the BSA-adjusted chronic kidney disease epidemiology and 25.51% for the Cockcroft-Gault formula.

The investigators concluded: “In a large data set from patients with cancer, BSA-adjusted [chronic kidney disease epidemiology model] is the most accurate published model to predict GFR. The new model improves this estimation and may present a new standard of care.”

The model is available as an online tool at

The work was supported by the Wellcome Trust Translational Medicine and Therapeutics Programme, the University of Cambridge, and the National Institute of Health Research Cambridge Biomedical Research Centre.

Tobias Janowitz, MB, BChir, PhD, of Cambridge Research UK, University of Cambridge, is the corresponding author of the Journal of Clinical Oncology article. 

The content in this post has not been reviewed by the American Society of Clinical Oncology, Inc. (ASCO®) and does not necessarily reflect the ideas and opinions of ASCO®.