To evaluate the effectiveness of estimated glomerular filtration rate (eGFR) equations in patients with type 2 diabetic kidney disease.

A total of 224 patients with type 2 diabetes mellitus (male 133, female 91, average age (60±13) y) who were treated during October 2007 and December 2010 in the Third Affiliated Hospital of SUN Yat-sen University were enrolled in this study. Serum creatinine (Scr) and cystadin-C (cysC) were tested. eGFR was calculated with abbreviated Modification of Diet in Renal Disease (MDRD) equation, Chinese equation, Cockcroft-Gault equation, Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation, Ruijin equation, EPI-Cys1 equation, EPI-Cys2 equation and MacIsaac equation, respectively, and compared with ^99mTc-DTPA-GFR (sGFR) by Bland-Altman analysis, Pearson correlation and receiver operator characteristic (ROC) curve.

(1) In Bland-Altman analysis, consistency of EPI-Cys1, EPI-Cys2 and MacIsaac equations with sGFR was better than other equations, and all eGFRs were above the limit point. The slopes of EPI-Cys1, EPI-Cys2 and MacIsaac equations were 0.168, 0.183, and 0.186, respectively. The intercept of MacIsaac equation was smaller. (2) Bias of EPI-Cys1, EPI-Cys2 and MacIsaac equations were smaller than others. 30% or 50% accuracy rate of these three equations was above 60% and 85%. EPI-Cys1 and EPI-Cys2 equations underestimated actual GFR and other Scr-based equations overestimated GFR. (3) As to diagnosis of diabetic kidney disease, EPI-Cys1 and EPI-Cys2 equations were more sensitive (94%) and accurate (>88%). The cut-point of MacIsaac and Ruijin equations were 62.4 and 64.9 ml·min^-1·1.73 m^-2, closer to 60 ml·min^-1·1.73 m^-2.

All the equations shows bias in estimating actual GFR. Compared with the equations induced by Scr, CysC-based equations may be more accurate and efficient. All the equations should be amended when applying to patients with type 2 diabetes mellitus.

Diabetes mellitus, type 2; Diabetic kidney disease; Glomerular filtration rate