Assessment of GFR Estimation Methods in Patients With Obesity.

Abstract

Obesity is an independent risk factor for chronic kidney disease, and accurate estimation of the glomerular filtration rate (GFR) is crucial. However, limited data are available on the performance of the European Kidney Function Consortium (EKFC) equation in individuals with overweight or obesity. To evaluate the performance of the EKFC equation by comparing its estimated GFR (eGFR) to values obtained from the 2021 Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation and machine learning (ML) models. We compared the models using measured GFR (mGFR, obtained via plasma iohexol clearance) as a reference standard in a cohort of patients with overweight or obesity. We calculated the median bias (eGFR minus mGFR) and the concordance correlation coefficient for each method. The non-indexed EKFC equation exhibited the highest median bias (-4.80 mL/min, 95% CI: -7.49 to -2.92), particularly in patients with overweight (-7.81 mL/min, 95% CI: -12.87 to -4.80) and obesity (-4.57 mL/min, 95% CI: -8.57 to -1.72), with lower bias in healthy BMI individuals (-2.50 mL/min, 95% CI: -6.85 to 2.23). The EKFC equation showed lower concordance with mGFR than both the CKD-EPI 2021 equation and ML models and was the least accurate for CKD stage classification in patients with obesity, particularly those with hyperfiltration. In patients with overweight or obesity, the EKFC equation performed worse than the ML-based models and the CKD-EPI 2021 equation despite limitations of the latter in European populations. Our findings highlight the need for more accurate GFR estimation in this population, optimization of these models, and their validation in larger, more diverse cohorts.