Reliability of Glomerular Filtration Rate Estimated by Creatinine-Based Formulas in Moderate to Severe Proteinuria.

BACKGROUND Creatinine-based Glomerular Filtration rate (GFR) formulas introduce a substantial bias in GFR estimations in patients with frank nephrotic syndrome. The bias and accuracy of creatinine-based GFR estimates (eGFR) in patients with non-nephrotic proteinuria need better characterization. METHODS We utilized data from the Ramipril in non-diabetic renal failure (REIN 1) and REIN 2 trials involving non-diabetic chronic kidney disease (CKD) patients with proteinuria to compare eGFRs derived from the CKD Epidemiology Consortium (CKD-EPI)formulas (with and without race), and the European Kidney Function Consortium (EKFC) equations with iohexol clearance (a gold-standard GFR measure, measured glomerular filtration rate [mGFR]). Bias was defined as the median difference between eGFR and mGFR, while accuracy was assessed using P30 and P15 metrics, which represent the percentage of eGFR values within ±30% and ±15% of mGFR, respectively. RESULTS The median bias of the three formulas being compared did not differ, being minimal and in a strict range (0.04 to 0.05 ml/ml/min/1.73m2) in the REIN 1 study and (-0.04 to -0.03 ml/min/1.73 m2) in the REIN 2 study. These findings were confirmed in analyses stratified by age and mGFR. The global accuracy of the three formulas regarding P30% showed sufficient accuracy (P30 >75%) in REIN 1 and all strata in REIN 2, but the mGFR stratum <15 ml/min/1.73m2. CONCLUSION The CKD-EPI (with and without race), and EKFC equations show no significant bias and sufficient accuracy in patients with proteinuria. These formulas can be safely applied to non-diabetic CKD patients with moderate to severe proteinuria.