Abstract
BACKGROUND: We compared values of baseline serum cystatin C (SCysC), serum creatinine (SCr), and measured glomerular filtration rate (mGFR) for predicting end-stage renal disease (ESRD) in patients with type 2 diabetes and elevated albuminuria. STUDY DESIGN: Observational longitudinal study. SETTING & PARTICIPANTS: Pima Indians with type 2 diabetes and elevated albumin-creatinine ratio (ACR ≥30 mg/g). PREDICTORS: Baseline SCysC, SCr, and mGFR. OUTCOMES & MEASUREMENTS: Individuals were followed up from their first examination with diabetes and ACR ≥30 mg/g until December 2010, onset of ESRD, or death, whichever came first. Incidence rates adjusted for age and sex were computed by Mantel-Haenszel stratification. The abilities of SCysC, SCr, and mGFR values to predict ESRD were compared with receiver operating characteristic curves. RESULTS: Of 234 Pima Indians with a mean age of 42.8 years who were followed up for a median of 10.7 (range, 0.6-21.3) years, 68 (29%) developed ESRD. The incidence of ESRD was significantly higher in patients in the lowest versus highest tertile of 1/SCysC (incidence rate ratio, 2.43; 95% CI, 1.31-4.50). By contrast, mGFR and 1/SCr had J-shaped associations with ESRD. In unadjusted analyses, 1/SCysC had the highest area under the receiver operating characteristic curve (AUROC; 0.719 ± 0.035) and mGFR had the lowest (0.585 ± 0.042; P < 0.001); the AUROC for 1/SCr was intermediate (0.672 ± 0.040; P = 0.1 and P = 0.03 vs 1/SCysC and mGFR, respectively). In analyses adjusted for age, sex, diabetes duration, height, weight, hemoglobin A1c level, and ACR, 1/SCysC had the highest AUROC (0.845 ± 0.026). Models with mGFR or 1/SCr alone had similar AUROCs (P = 0.9) and both were lower than the model with 1/SCysC alone (P = 0.02 and P = 0.03, respectively). LIMITATIONS: The predictive values of the filtration markers are limited to the extent that their precision is based on a single measurement. CONCLUSIONS: SCysC level was a better predictor of ESRD than mGFR or SCr level in Pima Indians with type 2 diabetes and elevated albuminuria.