Endocrine Abstracts

Background: Diabetes mellitus (DM) is the leading cause of chronic kidney disease (CKD) worldwide. In clinical practice, kidney function is routinely assessed through glomerular filtration rate (GFR) estimated by equations. However, the accuracy of current equations has been questioned for people with diabetes mellitus (DM).

Aim: To evaluate the performance of the European Kidney Function Consortium (EKFC), the Full Age Spectrum (FAS) and the 2021 Chronic Kidney Disease Epidemiology Collaboration (2021 CKD-EPI) equations based on serum creatinine to estimate GFR in healthy and type 2 DM participants.

Methods: This cross-sectional study evaluated three creatinine-based equations in two different populations: healthy adults (eligibility criteria: BMI ≤30 kg/m² ) and people with type 2 DM (eligibility criteria: mGFR>60 ml/min/1.73 m² ). GFR calculated by the equations was compared with measured GFR (mGFR) by the plasma clearance of ⁵¹ Cr-EDTA as the reference method. The performance of the equations was assessed using Bland-Altman plots, bias and P30 accuracy (defined as the percentage of GFR estimations falling within 30% of mGFR values).

Results: We included 100 healthy adults (aged 39±15 years, 67% women), with mean mGFR, 2021 CKD-EPI, FAS and EKFC of 112±20 109±14, 102±18 and 104±19 ml/min/1.73 m², respectively (P=0.109 between mGFR and 2021 CKD-EPI; P<0.001 for the other equations). Overall, the smallest bias was found for 2021 CKD-EPI (5 ml/min/1.73 m² ). All equations presented an acceptable P30 accuracy (92% [CI 95% 86-96] for 2021 CKD-EPI, 89% [82-94] FAS and 87% [79-93] EKFC;P=0.348) in the healthy participants. We also included 122 people with type 2 DM (aged 61±10 years, 55% women) in our study. Comparing with mean mGFR (100±28 ml/min/1.73 m² ), 2021 CKD-EPI, FAS and EKFC equations underestimated GFR (mean 86±20, 79±24, and 77±19 ml/min/1.73 m², respectively; P<0.001 for all equations). No equation achieved optimal P30 accuracy in participants with DM, but 2021 CKD-EPI tended to perform more accurately (74% [CI 95% 65-81] vs. FAS 66% [58-74],P=0.082; vs. EKFC 64% [56-72],P=0.01). The Bland Altman plots showed positive bias for all equations, with larger biases found in participants with type 2 DM, demonstrating that the underestimation was more pronounced in the presence of diabetes.

Conclusion: In healthy adults, 2021 CKD-EPI, FAS and EKFC are suitable to estimate GFR. However, none of the three equations reached an optimal performance for estimating GFR in participants with type 2 DM. There is still a necessity of improving GFR estimation in people with diabetes.