A Comparison of GFR Estimating Formulae Based Upon S-cystatin C and S-creatinine and a Combination of the Two

Background: Current recommendations (KDIGO and NKF-K/DOQI) are that patients with chronic kidney diseases (CKD) should be classified in stages 1-5 based on GFR. A serum creatinine-based prediction equation (abbreviated MDRD formula) can be used to estimate GFR (eGFR). Cystatin C has been proposed as an alternative filtration marker to creatinine. We present validation of currently used formulae for eGFR based upon s-creatinine and s-cystatin C and we compare two different methods for the determination of cystatin C.

Methods: S-cystatin C and s-creatinine were measured in 644 patients referred for determination of GFR by plasma clearance of iohexol during the period 1 June 2004 to 31 December 2005. S-cystatin C was determined by turbidimetry using two different reagents (DAKO A/S and Gentian A/S). The 644 patients were divided into two groups. Group 1 was used to calculate own eGFR-formulae based on s-cystatin C (Orebro-cyst). Group 2 was used to validate the formulae. Three creatinine-based equations (Cockcroft-Gault, MDRD and Jelliffe) and seven cystatin C-based (Larsson, Hoek, Filler, leBricon, Grubb and Orebro-cyst DAKO, Gentian) were evaluated. Evaluation was done according to the recommendations by K/DOQI.

Results: In the test sample (group 2) mean GFR (iohexol clearance) was 50.4 ml/min/1.73 m(2) (range 12-150)-mean s-cystatin C (DAKO AS) was 1.63 mg/l and mean s-cystatin C (Gentian AS) 1.92 mg/l. The s-cystatin C concentrations obtained by the Gentian method were approximately 10% lower than the DAKO method within the normal GFR range but were approximately 40% higher within the low GFR range. Bias for the creatinine-based equations was in the range -0.9 to 5.9 ml/min/1.73 m(2) and for the cystatin C-based equations in range -2.4 to 7.9 ml/min/ 1.73 m(2). Accuracy within 30% ranged from 68.6 to 80.4% and 54.0 to 82.9%, respectively. By combining both, an accuracy within 30% for 87.0% could be reached (MDRD/cystatin C by Gentian). Overall the patients were correctly classified for the different stages of CKD in 62.1-64.0% for the creatinine-based equations, 61.5-72.0% for the cystatin C-based equations and 70.2-73.9% for the combination.

Conclusion: Estimating GFR using formulae based on s-creatinine or s-cystatin C alone was equally accurate according to the NKF K/DOQI guidelines. A formula that combines both provided a greater accuracy. If Cystatin C, which is clearly more expensive, is used, the choice of the cystatin C determination method and an adjusted prediction equation is essential. Use of the IDMS-traceable MDRD seems to yield the best cost-benefit ratio for routine practice.