Agreement of glomerular filtration rate estimation equations for chemotherapy dosing in cancer patients at a tertiary referral hospital in Sub-Saharan Africa.

Abstract

Introduction: Narrow therapeutic indices of chemotherapeutic agents necessitate precise dosing to ensure efficacy and minimize nephrotoxicity. Due to the complexity of directly measuring Glomerular filtration rate (GFR), renal dosing is usually based on GFR estimation equations. The Cockcroft-Gault formula remains the most widely used equation in cancer patients, despite the availability of more precise kidney function estimation equations. Therefore, the aim of the study was to assess the agreement between Cr and cystatin-C (CysC) based GFR estimation equations and GFR estimated by Cockcroft-Gault for appropriate chemotherapy dosing in cancer patients undergoing assessment for first-Line chemotherapy at an oncology unit of St. Paul's Hospital Millennium Medical College in Ethiopia.

Methods: A cross-sectional study was conducted among 136 adult cancer patients scheduled to initiate chemotherapy at the hospital between November 1, 2021, and April 30, 2022. GFR was calculated using 12 different GFR estimation equations to be compared with Cockcroft-Gault; MDRD, MDRD adjusted for ethnic factor (MDRDef), the 2009 CKD-EPI calculated based on serum creatinine (CKD-EPI 2009 Cr), the 2009 CKD-EPI Cr adjusted for ethnic factor (CKD-EPI 2009 Cref), the 2012 CKD-EPI calculated based on serum Cystatin C (CKD-EPI 2012 CysC), the 2021 CKD-EPI calculated based on serum creatinine and Cystatin C (CKD-EPI 2021 Cr-CysC), the 2021 CKD-EPI (CKD-EPI 2021), FAS calculated based on serum creatinine (FAS Cr)¸ FAS Cr adjusted for African coefficient (FAS Craf), FAS calculated based on serum Cystatin C (FAS CysC), FAS calculated based on serum creatinine and Cystatin C (FAS Cr-CysC), and FAS Cr-CysC adjusted for African coefficient (FAS Cr-CysCaf). To assess the level of agreement, bias (mean error/ME), precision, and accuracy (root-mean squared error/ RMSE) were analyzed for each equation, where for all measurements a value closer to 0 indicates minimal bias, high precision, and high accuracy demonstrating good agreement with Cockcroft-Gault. To confirm the significance of the recorded levels of agreement, a one-sample t-test, a Bland-Altman plot, and a linear regression analysis were performed step by step for variables which proved to have statistical agreement, where a p-value > 0.05 and the presence of heteroscedasticity indicates a non-significant difference and hence the presence of good agreement.

Results: The GFR estimation equations revealed variation, with some methods underestimating and others overestimating GFR. However, only four equations showed potential agreement with Cockcroft-Gault based on a one-sample t-test: MDRD, CKD-EPI 2009 Cr, CKD-EPI 2021, FAS Cr, and FAS Cr-CysCaf. Among these, CKD-EPI 2009 Cr exhibited the least bias (ME = 0.72 ml/min, 95% CI: -67.66, 69.10 ml/min), while FAS Cr-CysCaf demonstrated the highest precision (SD = 33.92) and accuracy (RMSE = 34.53). However, further analysis using Bland-Altman plots and linear regression to confirm agreement revealed no agreement between any of the formulas and Cockcroft-Gault.

Conclusion: The study revealed that the most recent and accurate GFR estimation equations that are recommended to be used in cancer patients did not show agreement with Cockcroft-Gault. This suggests that current GFR estimation practices in cancer patients might be inaccurate, potentially leading to improper chemotherapy dosing and poorer patient outcomes.