Background  
Accurate estimation of glomerular filtration rate (GFR) is pivotal in managing diabetic kidney disease.

Aim  
To assess the performance of modification of diet in renal disease (MDRD), and chronic kidney disease epidemiology collaboration (CKD-EPI) estimated GFR for diagnosing clinically significant GFR cut-points in adults with diabetes.

Method  
Data from 1487 participants (86% with type 2 diabetes) with 2703 measurements of plasma clearance of (99m)Tc-diethylenetriamine-pentaacetic acid ((99m)Tc-DTPA) as the reference test was analysed. Area under receiver operating characteristic curve (AUC) of MDRD and CKD-EPI equations as well as their sensitivity, specificity, and AUC at clinically significant cut-points were in investigated in GFR &lt;15mL/min/1.73m(2) (threshold for CKD stage5), GFR&lt;30 (threshold for CKD stage4), GFR&lt;45 (threshold for CKD stage3b), GFR&lt;60 (threshold for CKD stage3a), GFR&lt;90 (threshold for CKD stage2) and GFR≥120mL/min/1.73m(2) (signifying a state of hyperfiltration). The above-mentioned cut-points were calibrated using Youden index.

Results  
Mean (SD) (99m)Tc-DTPA measured GFR was 66 (25)ml/min/1.73m(2). The proportion of participants at different measured GFR (ml/min/1.73m(2)) thresholds was &lt;15: 0.5%, &lt;30: 7%, &lt;45: 22%, &lt;60:43 %, &lt;90: 83%, and ≥120: 3%. The general AUC ranged between 0.87-0.98. However, AUCs at cut points were &lt;0.82 in both equations. Increasing CKD stage 5, 4, and 3 cut-point and decreasing CKD stage 2 and hyperfiltration cut-point in MDRD increased sensitivities by approximately 15-60%. Similarly in CKD-EPI, increasing CKD stage 5, 4, 3, and 2 cut-point and decreasing hyperfiltration cut-point resulted in 1-65% increase in sensitivities (Table).

Table. Sensitivity, specificity, and AUC of established clinically relevant GFRs and suggested GFR cut-points for maximising the performance of the MDRD and CKD-EPI equations

[ Table Removed ]

Discussion  
When assessed in continuously, MDRD and CKD-EPI have excellent discriminatory power for identifying (99m)Tc-DTPA cut-points. However, their sensitivity for determining clinically relevant GFR cut points as determined by direct measurements of GFR was low. There was a major discordance between GFR cut points for clinically relevant CKD stages and those that optimised the performance of estimates of GFR. This work highlights the need to improve current methods of estimating GFR in people with diabetes.



Calibrating estimated glomerular filtration rate cut points in adults with diabetes