Accounting for individual body surface area does improve estimation of absolute GFR using the CKD-EPI equation

Abstract

<h3>Aims</h3> To compare estimated glomerular filtration rate (eGFR) according to the CKD-EPI equation, with (CKD-EPI, mL/min/1.73m²) and without body surface area (BSA)-normalisation (CKD-EPI_;BSA, mL/min), against measured Tc-DTPA GFR (mL/min). <h3>Methods</h3> The CKD-EPI and CKD-EPI_;BSA equations were compared in 222 individuals with Tc-DTPA GFR for bias, proportion within 30% of GFR (P30) and area under the receiver-operator curve (ROC) for detecting GFR<90mL/min. In 80 oncology patients and 78 obese subjects, we also evaluated concordance in relation to carboplatin dosing. <h3>Results</h3> Chi-square indicated differences in P30s between CKD-EPI_;BSA and CKD-EPI: in those with BMI ≥30kg/m² (32%), in those with BMI>25.0–29.9kg/m² (18%) and in those with BMI ≥18.5–25.0kg/m² (2%) (<i>p</i><0.0001). The ROC area under curve (AUC) for CKD-EPI_;BSA equation to detect GFR<90mL/min (0.85) and>125mL/min (0.81) was greater than for the CKD-EPI (0.80 and 0.71, respectively). Concordance for carboplatin dosing using the CKD-EPI_;BSA equation was 71% and 56.% by the CKD-EPI equation (<i>p</i>=0.07). <h3>Conclusion</h3> Estimation of absolute Tc-DTPA GFR using the CKD-EPI equation was improved by removal of BSA normalisation, reflected by higher proportion of results within 30% of GFR and less underestimations of GFR, with implications for drug dosing.