Estimation of the glomerular filtration rate (GFR) is the most widely used test of renal function, reflecting the relative mass of functional renal tissue and thus the number of functioning nephrons. Methods based on measurement of exogenous substances such as inulin, 51Cr-EDTA, 99mTc-diethylenetriaminepentaacetic acid, and iohexol are accurate but too complex and laborious for routine clinical use; thus, measurement of endogenous blood substances is common practice. Plasma or serum creatinine and its renal clearance are the approaches most commonly used despite their acknowledged unreliability. Cystatin C, a small basic protein, has been proposed as a better marker than creatinine. Recently, the value of cystatin C was thoroughly reviewed in this Journal (1), and according to this review and a new metaanalysis (2), most studies have concluded that cystatin C is superior to plasma creatinine, whereas several authors have concluded that cystatin C provides no advantage. One purpose of the present study was to clarify possible reasons for the earlier, partly conflicting results. A recently published guideline from the National Kidney Foundation (3) recommended that GFR be estimated from prediction equations taking into account the serum creatinine concentration and some or all of the following variables: age, gender, race, and body size. We therefore also compared cystatin C with GFRs calculated by the Cockcroft–Gault (4) and the MDRD(5) formulas. We studied 112 patients (55 men and 57 women) for whom 51Cr-EDTA clearance had been requested. The mean age of the patients was 57.0 years (range, 17–89 years). Body mass index (BMI) was 15.2–42.4 kg/m2, and 51Cr-EDTA clearance was 5–109 mL · min−1 · (1.73 m2)−1. The most common diagnoses were diabetic nephropathy (n = 27), rheumatoid arthritis-related diseases (n = 20), and chronic glomerulonephritis (n = 20; for more details, see the Data Supplement that accompanies the online version of this Technical Brief at http://www.clinchem.org/content/vol49/issue7/). A detailed medical history was obtained from all participants. Approval was obtained from the Ethical Committee of Tampere University Hospital. The study was in accordance with the ethical standards of the Helsinki Declaration. Plasma 51Cr-EDTA clearance was assessed by the single-injection method and blood samples drawn at 0, 90, and 180 min (6). A part of the time 0 sample was used to determine plasma creatinine and cystatin C. Plasma cystatin C was measured turbidimetrically (7) on a Hitachi 704 instrument, and creatinine was determined enzymatically (8) on the same instrument. The reference intervals for plasma cystatin C using this method were 0.5–1.3 mg/L for ages 3–50 years and <1.5 mg/L for age >50 years. Enzymatically determined creatinine values were slightly lower than those measured with the Jaffe procedure; the reference intervals were 50–95 μmol/L for women and 55–105 μmol/L for men (8). Statistical analysis was performed with SPSS 11.0 for Windows software (SPSS Inc.) on a microcomputer. Plasma concentrations of creatinine and cystatin C are inversely related to GFR. Application of the reciprocals of the measured concentrations linearized this curvilinear relationship. Linear regression analysis was used to predict GFR values (dependent variable), using plasma cystatin C (in fact, 1/cystatin C), plasma creatinine (1/creatinine), GFRCG (calculated by the Cockcroft–Gault formula), and GFRMDRD (calculated by the MDRD formula) as predictors (independent variables) in the different subgroups. Diagnostic accuracies were determined using the GraphROC program (9). All tests describing GFR correlated well with plasma 51Cr-EDTA clearance. The correlation coefficients (cystatin C, r = 0.917; creatinine, r = 0.884; GFRCG, r = 0.914; GFRMDRD, r = 0.935) did not differ statistically from each other. GFR values calculated by the Cockcroft–Gault formula were ∼10% higher [regression line: y = 1.11x − 1.76 mL · min−1 · (1.73 m2)−1] than the plasma 51Cr-EDTA clearance values. The MDRD formula gave values very similar to those for plasma 51Cr-EDTA clearance (regression line: y = 1.01x − 3.32 mL · min−1 · (1.73 m2)−1. The comparisons between the diagnostic accuracies are presented in Fig. 1 . The diagnostic accuracy of cystatin C was superior to that of creatinine (P = 0.042) when results in all 112 patients were considered. The difference was also evident (P = 0.035) when only individuals with normal or moderately impaired kidney function [GFR > 40 mL · min−1 · (1.73 m2)−1] were studied. The difference between diagnostic accuracies disappeared if BMI was normal (20–25 kg/m2; P = 0.478). In the patient group studied, there were only two individuals with abnormal BMI (<20 or >25 kg/m2) and with normal GFR [≥80 mL · min−1 · (1.73 m2)−1]. It was therefore impossible to examine this group with the ROC curve. The linear regression analysis showed, however, that among individuals with abnormal BMI, cystatin C seemed to be a better predictor of GFR than the plasma creatinine value (see the online Data Supplement). If plasma creatinine values were corrected for age, gender, race, and body size with use of the Cockcroft–Gault or MDRD formula, GFRCG and GFRMDRD predicted kidney function as well as cystatin C did (Fig. 1 ). Table 1 describes how well the tests classified patients as having normal or reduced GFR as measured with 51Cr-EDTA clearance. Ability of the tests to correctly classify individuals as having normal or reduced GFR. Differentiation limit of creatinine: 95 μmol/L for women, 105 μmol/L for men. Differentiation limit of cystatin C: 1.30 mg/L for individuals ≤50 years, 1.50 mg/L for those >50 years. Differentiation limit of GFRMDRD and GFRCG: 80 mL/min. Ability of the tests to correctly classify individuals as having normal or reduced GFR. Differentiation limit of creatinine: 95 μmol/L for women, 105 μmol/L for men. Differentiation limit of cystatin C: 1.30 mg/L for individuals ≤50 years, 1.50 mg/L for those >50 years. Differentiation limit of GFRMDRD and GFRCG: 80 mL/min. Nonparametric ROC plots for the diagnostic accuracy of cystatin C (CysC), creatinine (Cre), GFRCG, and GFRMDRD in distinguishing between normal and reduced GFR [51Cr-EDTA clearance <80 mL · min−1 · (1.73 m2)−1]. (A), All participants (n = 112). CysC area = 0.9728 (SE, 0.0173); Cre area = 0.9316 (SE, 0.0246); GFRCG area = 0.9908 (SE, 0.0071); GFRMDRD area = 0.9786 (SE, 0.0105). All P values are listed in the online Data Supplement. (B), individuals with 51Cr-EDTA clearance > 40 mL · min−1 · (1.73 m2)−1 (n = 34). CysC area = 0.8812 (SE, 0.0647); Cre area = 0.7063 (SE, 0.0943); GFRCG area = 0.9604 (SE, 0.0296); GFRMDRD area = 0.9083 (SE, 0.0552). All P values are listed in the online Data Supplement. Many investigators have found that the reciprocal of cystatin C correlates more closely with GFR than does that of creatinine (1)(2)(10). Deinum and Derkx (11) criticized such comparisons because they underestimate the value of plasma creatinine. They stress that algorithms that take into account creatinine’s dependence on age and body mass should be used in these comparisons. The results presented here support their view. One essential, still unanswered question, however, is whether measurement of cystatin C instead of creatinine leads to other or earlier clinical decisions, with better outcome for the patient. Answers will require more longitudinal studies (12) in diverse patient groups. The K/DOQI recommendation also emphasizes the importance of accurate measurement of plasma or serum creatinine (3). The enzymatic method we used (8) seems to be highly specific and gives results very similar to those obtained with a HPLC method (13). This might be one reason for the good correlations between 51Cr-EDTA clearance and GFRCG and GFRMDRD in this study. We conclude that cystatin C is superior to enzymatically measured creatinine as an estimator of GFR only when the BMI is abnormal or when the GFR is normal or moderately impaired. When GFR estimates based on plasma creatinine were improved by taking into account age, body size, or gender, the estimates GFRCG and GFRMDRD were very similar to those based on the cystatin C determinations.
No takes yet. Share an insight, caveat, or question.
Harmoinen et al. (2003) studied this question.
Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context: