RESUMO
BACKGROUND/AIM: Cystatin C (Cys C) is an endogenous marker of glomerular filtration rate (GFR) unaffected by body composition. The aim of the present study was to assess the utility of Cys C-based GFR prediction equations (Hoek, Larsson and Stevens) and creatinine (modification of diet in renal disease-isotope dilution mass spectrometry--MDRD-IDMS, and Cockcroft-Gault--CG) compared with 51Cr-EDTA. METHODS: This study was carried out in 40 Caucasian older patients with advanced age (> or = 60) and chronic kidney disease stages 3-4. To assess the utility of prediction equations in relation to body composition, we measured lean mass (LM) with densitometry (DXA). Pearson's, Bland-Altman and Lin's coefficient (Rc) were used to study accuracy and precision. RESULTS: 51Cr-EDTA was 36.9 +/- 9.2 ml/min/1.73 m2 (22-60). Cys C levels were 2.2 +/- 0.8 mg/l (r = 0.085; p = 0.662 LM) and creatinine 2.8 +/- 1.1 mg/dl (r = 0.427; p = 0.021 LM). The most accurate equations were the Hoek, Larsson and Stevens formulae, with a bias of -0.2 (Rc 0.48), -2.9 (Rc 0.44) and 2.6 ml/min/1.73 m2 (Rc 0.58). The biases obtained with MDRD-IDMS and CG were -14.6 (Rc 0.35) and -12.5 (Rc 0.40). All correlations among biases obtained with creatinine-based formulae and LM were negative and statistically significant (p < 0.05). CONCLUSIONS: The results show superiority of Cys C-based GFR formulae over the MDRD-IDMS and CG equations. This significant underestimation obtained with conventional prediction equations was directly related to the influence of LM.