Abstract
OBJECTIVE: This method comparison study, conducted at the peritoneal dialysis (PD) outpatient clinic of the Department of Renal Medicine, Aarhus University Hospital, Denmark, set out to evaluate the accuracy and reproducibility of methods for estimating glomerular filtration rate (GFR) based on endogenous markers in PD patients. PATIENTS: The 12 consecutive patients included in the study were examined twice while in a stable condition. All patients finished the study. Inclusion criteria were age 18 years or older, ability to collect 24-hour urine, and urine production greater than 300 mL in 24 hours. MAIN OUTCOME MEASURES: The methods for estimating GFR using endogenous markers included the average of urinary clearances of creatinine and urea [U-Cl(crea-urea)] and two equations using the serum concentration of cystatin C [eGFR(CysC)]. The resulting GFR estimates were compared with those obtained using urinary and corrected plasma clearances of (51)Cr-EDTA [U-Cl(EDTA) and cP-Cl(EDTA)], the corrected plasma clearance being plasma clearance minus dialysate clearance. RESULTS: Compared with the U-Cl(EDTA), the U-Cl(crea-urea) GFR estimate was 12% higher [95% confidence limits (CL): 3%, 21%]. Although significantly different (p = 0.01), the latter two methods showed the best agreement. The estimates obtained using the eGFR(CysC) methods were skewed from y = x compared with the estimates obtained using other methods, indicating strong bias, probably because of extrarenal elimination. The cP-Cl(EDTA) estimate was 34% (95% CL: 26%, 42%), higher than the U-Cl(EDTA) estimate (p < 0.001). The reproducibility (coefficients of variation) differed significantly between methods: cP-Cl(EDTA), 7%; U-Cl(EDTA), 14%; U-Cl(crea-urea), 18%; and both eGFR(CysC) methods, 3%. CONCLUSIONS: In PD patients, GFR may be estimated as U-Cl(crea-urea) when complete urine collection is performed, taking into account an overestimation of approximately 12%. The available equations for eGFR(CysC) seem to be inaccurate; further development and validation is desirable. Omitting the eGFR(CysC) methods, cP-Cl(EDTA) was the most reproducible method and might be useful in certain situations.