Prediction of relative glomerular filtration rate in adults:: New improved equations based on Swedish Caucasians and standardized plasma-creatinine assays

Objective. To evaluate newly developed equations predicting relative glomerular filtration rate (GFR) in adult Swedish Caucasians and to compare with the Modification of Diet in Renal Disease (MDRD) and Mayo Clinic equations using enzymatic and zero-calibrated plasma creatinine assays. Material and methods. GFR was measured with iohexol clearance adjusted to 1.73 m(2). One population sample (n=436/Lund) was used to derive an equation based on plasma-creatinine/age/gender, and a second with the addition of lean body mass (LBM). Both equations were validated in a separate sample (n=414/Malmo). The coefficients of the equations were eventually fine-tuned using all 850 patients and yielding Lund-Malmo equations without (LM) and with LBM-term (LMLBM). Their performance was compared with the MDRDCC (conventional creatinine calibration), MDRDIDMS (isotope dilution mass spectroscopy traceable calibration) and Mayo Clinic equations. Results. The Lund equations performed similarly in both samples. In the combined set, the Mayo Clinic/MDRDCC resulted in +19.0/+10.2% median bias, while bias for the other equations was < 10%. LMLBM had the highest accuracy (86% of estimates within 30% of measured GFR), significantly (p < 0.001) better than for MDRDIDMS (80%). In men with BMI < 20 kg/m(2), MDRDIDMS/LM had +46%/+19% median bias. MDRDIDMS also overestimated GFR by 22%/14% in men/women above 80 years of age. The LMLBM equation had < 10% bias irrespective of BMI, age or GFR except for a 15% negative bias at GFR >= 90 mL/min/1.73 m(2). Conclusion. The newly developed Lund-Malmo equations for GFR estimation performed better than the MDRDIDMS and Mayo Clinic equations in a Swedish Caucasian sample. Inclusion of an LBM term improved performance markedly in certain subgroups.