Performance of Serum Cystatin-C Versus Serum Creatinine in Subjects With Type 1 Diabetes

Glomerular filtration rate (GFR) is considered the best marker of renal function, and its estimation using inulin or ⁵¹Cr-EDTA is considered the “gold standard.” Such determination is inconvenient and often replaced by an estimate of creatinine clearance (CrCl) derived from the formula of Cockroft and Gault (CG). Serum cystatin-C was recently proposed as a reliable alternative routine marker of GFR in pediatric, adult, and elderly subjects (1–4). Although Dharnidharka et al. (5) thouroughly review the issue of comparing cystatin-C and creatinine in relation to reference GFR measurements, few reports deal with the comparison of serum creatinine and cystatin-C in diabetic patients (5–9). Odozze et al. (6) suggested that serum cystatin was not superior to creatinine for estimating GFR in type 1 and type 2 diabetic patients with early renal impairment, while others clearly hint toward cystatin-C measurement being a more sensitive and specific GFR marker in type 2 diabetic subjects with normal or slightly reduced GFRs (5–8). Mussap et al. (9) found that cystatin-C was more accurate than creatinine in discriminating diabetic subjects according to ⁵¹Cr-EDTA GFR.

In the November 2002 issue of Diabetes Care, Tan et al. (10) reported the clinical usefulness of cystatin-C for GFR estimation in subjects with type 1 diabetes using iohexol clearance as the “gold standard” GFR measurement and discriminant ratio (DR) methodology (10,11). Using the same DR approach (11), we assessed, from intra- and intersubject variability, the performance of cystatin-C (particle-enhanced immunonephelometric method, N Latex Cystatin-C; Dade Behring) in 46 subjects with type 1 diabetes spanning a wide range of kidney function, as compared with that of serum creatinine (8). Age and diabetes duration were 45 ± 16 and 18 ± 12 years, respectively, BMI 24 ± 3 kg/m², and HbA_1c 8.7 ± 1.5%. Median CrCl estimated by the CG formula was 92 ml/min (range: 15–149; 25–75th percentile: 81–110), adjusted to a body surface area of 1.73 m², with normo- (CG 70–120 ml/min), hyper- (CG >120), and hypofiltration (CG <70) present in 63, 20, and 17%, respectively. Serum creatinine levels were 1.01 ± 0.73 and 1.02 ± 1.00 mg/dl and cystatin-C 1.00 ± 0.67 and 0.98 ± 0.73 mg/l on days 1 and 2, respectively. A close linear relationship was observed between means of duplicates for creatinine and cystatin-C (Pearson product-moment correlation 0.97). The DR (ratio of the underlying between-subject to within-subject SD) was 3.92 for creatinine and 9.09 for cystatin-C (P < 0.0001), implying superior discriminating ability for cystatin-C. Once adjusted for attenuation, measured Pearson product-moment correlation increased to 1.00. The unbiased linear regression equation between methods had a slope of 0.83 and an intercept at 0.16. We conclude that serum cystatin-C better discriminates among a population of type 1 diabetic patients with regard to their estimated GFR, as compared with conventional serum creatinine measurement, and recommend its routine use for estimating kidney function in this population.