How Reliable Is Estimation of Glomerular Filtration Rate at Diagnosis of Type 2 Diabetes?

OBJECTIVE—The Cockcroft-Gault (CG) and Modification of Diet in Renal Disease (MDRD) equations previously have been recommended to estimate glomerular filtration rate (GFR). We compared both estimates with true GFR, measured by the isotopic 51Cr-EDTA method, in newly diagnosed, treatment-naïve subjects with type 2 diabetes. RESEARCH DESIGN AND METHODS—A total of 292 mainly normoalbuminuric (241 of 292) subjects were recruited. Subjects were classified as having mild renal impairment (group 1, GFR <90 ml/min per 1.73 m2) or normal renal function (group 2, GFR ≥90 ml/min per 1.73 m2). Estimated GFR (eGFR) was calculated by the CG and MDRD equations. Blood samples drawn at 44, 120, 180, and 240 min after administration of 1 MBq of 51Cr-EDTA were used to measure isotopic GFR (iGFR). RESULTS—For subjects in group 1, mean (±SD) iGFR was 83.8 ± 4.3 ml/min per 1.73 m2. eGFR was 78.0 ± 16.5 or 73.7 ± 12.0 ml/min per 1.73 m2 using CG and MDRD equations, respectively. Ninety-five percent CIs for method bias were –11.1 to −0.6 using CG and –14.4 to –7.0 using MDRD. Ninety-five percent limits of agreement (mean bias ± 2 SD) were −37.2 to 25.6 and −33.1 to 11.7, respectively. In group 2, iGFR was 119.4 ± 20.3 ml/min per 1.73 m2. eGFR was 104.4 ± 26.3 or 92.3 ± 18.7 ml/min per 1.73 m2 using CG and MDRD equations, respectively. Ninety-five percent CIs for method bias were –17.4 to –12.5 using CG and –29.1 to –25.1 using MDRD. Ninety-five percent limits of agreement were −54.4 to 24.4 and −59.5 to 5.3, respectively. CONCLUSIONS—In newly diagnosed type 2 diabetic patients, particularly those with a GFR ≥90 ml/min per 1.73 m2, both CG and MDRD equations significantly underestimate iGFR. This highlights a limitation in the use of eGFR in the majority of diabetic subjects outside the setting of chronic kidney disease.