Clin Chem Lab Med 2005;43(9):953 957 2005 by Walter de Gruyter Berlin New York. DOI 10.1515/CCLM.2005.163 2005/147 Short Communication Evaluation of renal function in intensive care: plasma cystatin C vs. creatinine and derived glomerular filtration rate estimates Thierry Le Bricon 1, *, Isabelle Leblanc 2, Mourad Benlakehal 1,Cécile Gay-Bellile 1, Danielle Erlich 3 and Said Boudaoud 2 1 Laboratoire de Biochimie A, 2 Service de Réanimation chirurgicale, 3 Laboratoire de Biochimie B, Hôpital St-Louis (AP-HP), Paris, France Abstract Plasma cystatin C, a new marker of glomerular filtration rate (GFR), was prospectively evaluated in surgical intensive care. Cystatin C was measured (immunonephelometry, Dade-Behring) in 10 patients selected to cover a full range of GFR (phase I) and in 28 unselected consecutive patients followed for 5 days post-admission (phase II). Results were compared with 51 Cr-EDTA clearance (phase I only), plasma creatinine (kinetic Jaffe, Roche), 24-h or estimated by Cockcroft and Gault (CG) creatinine clearance (CrCl), and modified diet in renal disease (MDRD)-estimated GFR. In phase I, the highest correlation with 51 Cr-EDTA clearance (22 198 ml/min) was noted for CG CrCl (r 2 : 0.883, p-0.001). During phase II followup, 24-h CrCl could not be calculated in 25% of daily evaluations. Cystatin C correlated with creatinine (0.856, p-0.0001) and CG CrCl with MDRD GFR (0.926, p-0.0001) in renal failure (10 78 ml/min, ns60). There was a q40% (p-0.001) median difference between cystatin C and creatinine (as a%of upper normal cut-off). Sensitivity/specificity to detect a -80 ml/min CG CrCl was 88/97% for cystatin C vs. 48/100% for creatinine (laboratory cut-off). In patients with normal and stable renal function (ns14), day-today intra-individual variation was 7.4% for cystatin C (vs. 10.6% for creatinine). In intensive care unit surgical adult patients, CG CrCl provides an easy and cost-effective estimate of GFR. Superior to creatinine, plasma cystatin C can be measured in selected patients where CG CrCl is known to be inaccurate. Keywords: biochemical marker; intensive care; renal failure. Sensitive and reliable recognition of renal failure is of primary importance in hospitalized patients, especial- *Corresponding author: T. Le Bricon, Laboratoire de Biochimie A, Hôpital St-Louis (AP-HP), 1 avenue Claude Vellefaux, 75010 Paris, France Phone: q33-1-4249-4621, Fax: q33-1-4249-9247, E-mail: thierry.le-bricon@sls.ap-hop-paris.fr ly during critical illness. The convenience of assessing glomerular filtration rate (GFR) by a single blood determination, available in less than 30 min and on a 24-h basis, has contributed to creatinine success, despite emphasised limitations (1, 2). Recent guidelines on chronic kidney disease (3), however, recommend using GFR prediction equations, rather than creatinine alone or 24-h creatinine clearance. Cystatin C, a 13.3-kDa cystein proteinase inhibitor, is a new marker of GFR. It possesses some advantages over the laboratory standard creatinine wreviewed in Ref. (4)x. Produced by all nucleated cells, this cationic protein is freely filtered by the glomerular membrane before being entirely catabolised by proximal tubular cells. Cystatin C blood concentration is less dependent on factors other than renal function (especially gender, muscle mass and age) than creatinine (5, 6). A recent meta-analysis suggests that cystatin C is superior to creatinine for the detection of impaired GFR in cross-sectional studies (7). This is especially true in children, the elderly and subjects with reduced muscle mass. Surprisingly, limited data are available on the validity of cystatin C as a GFR marker in intensive care (8, 9). Plasma cystatin C was prospectively evaluated as a marker of renal function in a group of adult patients admitted to our surgical intensive care unit (ICU). Results were compared with plasma creatinine, creatinine clearance (CrCl) measured on 24 h or estimated by Cockcroft and Gault (CG) (10), and GFR estimated by the abbreviated modified diet in renal disease (MDRD) equation (11). In a first 2-month phase, 10 subjects admitted to our surgical ICU (age: 49"6 years, male/female: 8/2, total body weight: 69"4 kg, Simplified Acute Physiology Score (SAPS) II score: 34"53) were selected to cover a full range of GFR based on their day 1 post-admission plasma creatinine (48 198 mmol/l) (exclusion criteria: haemodialysis). Initial diagnosis was: postoperative follow-up (ns7), cranial trauma (ns1), acute meningitis (ns1) and acute pancreatitis (ns1). On day 2, plasma creatinine and cystatin C, 24-h CrCl, MDRD-estimated GFR and CG CrCl were evaluated for agreement with 51 Cr-EDTA clearance. In a second 4-month phase, 28 unselected consecutive patients (age: 55"17 years, male/female: 14/14, total body weight: 68"12 kg, SAPS: 30"12, hospital stay: 19"17 days) were included. Initial diagnosis was: post-operative follow-up (ns15, including sepsis: ns8), pneumonia (ns2), hemothorax (ns3), acute pulmonary oedema (ns4), cranial trauma (ns1), cardiac arrest (ns1), mediastinal infection (ns1), and
954 Le Bricon et al.: Markers of renal failure in intensive care Article in press - uncorrected proof gangrene (ns1). They were monitored on a daily basis with the above markers (except 51 Cr-EDTA clearance), starting on day 1 after admission and for 5 consecutive days (exclusion criteria: haemodialysis). This study was in accordance with the ethical standards of the Helsinki declaration of 1975, revised in 1983. Seven ml of venous blood were drawn at 08:00 am in a Vacutainer tube (Becton Dickinson, Le Pont-de- Claix, France) with lithium heparinate as an anticoagulant and centrifuged (1500=g at 208C for 15 min). Samples from 24-h urine were obtained in 25-mL Monovette tubes (Sarstedt, Orsay, France) and centrifuged for 5 min at 1500=g at q48c. All markers were measured on fresh samples, except cystatin C (storage at y208c before analysis within 1 month). Plasma cystatin C was measured by an immunonephelometric assay (Dade-Behring, Paris La Défense, France) as previously described in detail (12 14). Our laboratory reference interval is 0.46 0.94 mg/l regardless of age and gender. Creatinine (plasma, urine) was assayed by a Jaffe compensated kinetic technique (Hitachi 747, Roche, Meylan, France) (14). CG CrCl (ml/min) was calculated according to the formula established by Cockcroft and Gault (10) and adapted for the use of plasma creatinine in mmol/l: CG CrCls(140-age)=(total body weight)=1.23 for male or 1.04 for female/plasma Cr, with total body weight in kg. Abbreviated MDRD-estimated GFR (ml/ min/1.73 m 2 ) was calculated using the published formula (11). Our laboratory reference intervals (adults) for plasma creatinine are 60 100 mmol/l for females and 64 109 mmol/l for males, and 80 120 ml/min for GFR regardless of age and gender (13, 14). In phase I, GFR was estimated using a Brochner-Mortensen corrected one-compartment model (15). The disappearance of radioactivity was measured from three timed plasma samples (T1, T2, T3) after i.v. injection of a single dose of 51 Cr-labeled EDTA (3 mci/kg total body weight). T2 and T3 sampling times were adapted to each patient according to radioactivity measured at T1 (3 h post-injection): T2 at 4 or 5 h, and T3 at5or24h. Results are presented as median and range. To correlate GFR markers, linear regression (in phase I) or Spearman rank order correlation (in phase II) were selected, as appropriate, after checking for Gaussian or non-gaussian distribution (Kolmogorov-Smirnov test) (Sigmastat software, Jandel Scientific, Corte Madera, USA). The difference between markers was tested by analysis of variance (ANOVA) on ranks. A p-value less than 0.05 was considered statistically significant for all tests. In phase I, day 2 post-admission 51 Cr-EDTA clearance ranged from 22 to 198 ml/min (median: 107 ml/min, patients with renal failure: ns5). It correlated with 24-h CrCl (r 2 s0.552, p-0.05), 1/plasma creatinine (0.686, p-0.01), MDRD-estimated GFR (0.712, p-0.01), 1/plasma cystatin C (0.755, ps0.001), and CG CrCl (0.883, p-0.001). There was no significant difference between 51 Cr-EDTA and creatinine GFR estimates (data not shown). Specificity to detect a -80 ml/min 51 Cr-EDTA clearance was 100% for all markers; sensitivity was 60% for creatinine (two false negatives), 80% for cystatin C, 24-h CrCl, and MDRDestimated GFR (one false negative), and 100% for CG CrCl. In phase II, 13 patients had reduced CG CrCl at day 1 post-admission, but without requiring extra-renal epuration (median: 52 ml/min, range: 10 78 ml/min); 15 patients had normal estimated GFR (119 ml/min; 84 155 ml/min). In 35 out of 140 daily evaluations, 24-h CrCl could not be calculated (urine missing). In patients with normal CG CrCl, no correlation was found between GFR markers (data not shown), except between MDRD-estimated GFR and CG CrCl (0.534, p-0.001). In those with reduced CG CrCl (ns60), plasma GFR markers significantly correlated (0.856, p-0.0001), as did MDRD-estimated GFR and CG CrCl (0.926, p-0.0001). In this sub-group of patients, there was a q40% median difference (p-0.001) between plasma cystatin C and creatinine (as a % of upper normal cut-off) (Bland-Altman analysis, Figure 1). There was also a significant q8 ml/min median difference between MDRD-estimated GFR and CG CrCl (p-0.05, range: y13 to q35 ml/min). In 13 samples (nine patients, male/female: 2/7), estimated GFR ranged from 80 to 102 ml/min/1.73 m 2 by the MDRD formula. Ten of them had elevated cystatin C (0.97 2.31 mg/l), but normal creatinine. By receiver-operating characteristic (ROC) curve analysis, cystatin C displayed a higher diagnostic performance than creatinine to detect a CG CrCl -80 ml/min (Figure 2). Efficacy (sensitivity/specificity) was 88/97% for plasma cystatin C and 48/100% for creatinine using our laboratory cutoff. One patient with normal CG CrCl at day 1 postadmission (95 ml/min) developed an acute renal failure at day 3 (53 ml/min) evidenced by the MDRD GFR (46 ml/min), creatinine (145 mmol/l) and cystatin C (1.14 mg/l). Two patients with moderately decreased CG CrCl at day 1 (74, 78 ml/min), but normal MDRD-estimated GFR (80, 99 ml/min) and creatinine (data not shown) returned to normal values at day 2 (cystatin C: 0.97 to 0.78 mg/l, and 0.88 to 0.78 mg/l). In patients with normal and stable renal function (ns14), day-to-day intra-individual variation was 10.6% for creatinine vs. 7.4% for cystatin C. In ICU patients, limitations of blood creatinine alone as a marker of GFR are particularly prominent due to frequently reduced muscle mass (muscle wasting, malnutrition) (1, 2, 8, 9). GFR might be thus greatly overestimated, potentially compromising the management of renal failure in these patients. A 24-h urine sample would provide additional information, but collection is frequently impracticable (recent admission, anuria) or unreliable (loss of urine). Here, it contributed to 25% of missing 24-h CrCl results in the follow-up phase II study (not mentioning inaccurate records of urinary volume). When available, 24-h CrCl displayed the highest intra-individual variation (CVs34%) of all studied markers in patients with normal and stable renal function. It also possessed the poorest correlation with 51 Cr-labeled EDTA clearance, a surrogate GFR marker with excellent agreement with the gold-standard inulin clearance. Measuring
Le Bricon et al.: Markers of renal failure in intensive care 955 Figure 1 Difference between plasma cystatin C and creatinine in patients with renal failure. Creatinine clearance (CrCl) was estimated from plasma creatinine using the formula developed by Cockcroft and Gault (10). Over the 5-day follow-up period, impaired GFR (-80 ml/min) was observed in ns60 out of 134 samples. Results are expressed in % of upper normal cut-off. Dashed line: median value, dotted lines: 25th 75th percentiles. GFR by the present 51 Cr-EDTA technique is almost impracticable in ICU patients and costly, at least on a large scale. Sepsis is also associated with an intravascular fluid deficit due to vasodilatation, venous pooling and capillary leakage. In some patients, large ascites and/or other extra-capillary fluids might have altered the distribution volume of 51 Cr EDTA. Alternatives include the use of GFR prediction equations (3) and/or new better markers than creatinine, such as cystatin C. In the present study, we selected two GFR prediction equations: the abbreviated MDRD (11) and the Cockcroft and Gault developed in 1976 (10). They are well adapted to ICU because of their simplicity (no 24-h urine collection), requiring the measure of a single blood parameter (creatinine), age, gender, and total body weight (10) or ethnicity (11). In the present study, the MDRD-estimated GFR and CG CrCl highly correlated (0.926, p-0.0001) in patients with renal failure (correlation was weak in those with normal renal function). Results obtained with the MDRD equation were significantly higher than those of the CG formula, but the median difference was small (8 ml/min). It should be recalled that the MDRD GFR estimate is normalised to body surface (11), whereas CG CrCl is not (10). Interestingly, discordant GFR results (decreased by CG, confirmed by elevated cystatin C, but normal by MDRD) were more prominent in females. The Cockcroft and Gault formula (10) here displayed the highest correlation with 51 Cr-EDTA Figure 2 Diagnostic performance of plasma markers for the detection of renal failure. Efficacy of plasma cystatin C and creatinine to detect an impaired Cockcroft and Gault estimated CrCl during follow-up was evaluated by ROC analysis.
956 Le Bricon et al.: Markers of renal failure in intensive care Article in press - uncorrected proof clearance and sensitivity to detect an estimated GFR -80 ml/min (100%, no false negatives). These results should be, however, cautiously interpreted due to the limited number of patients (ns10). The most complex of MDRD equations (which includes serum urea nitrogen and albumin concentrations) estimated GFR more accurately than CG CrCl in patients with chronic renal disease (16); it was not, however, validated in normal renal function and in ICU patients. Additional studies are needed to compare available GFR prediction equations during critical illness. Alternative analytes (such as cystatin C) and derived GFR prediction equations may be more accurate than CG or MDRD formulas, at least in some cases. Creatinine blood concentration is dependent on changes in renal tubular metabolism (during cimetidin therapy, for example) and creatine content of the diet (vegetarians, use of creatine supplements). Creatinine measurement in plasma or serum is also subject to analytical difficulties (interference, calibration of Jaffe colorimetric assays). Cystatin C is less dependent on these factors, thus resulting in a lower inter-individual variability than creatinine in healthy subjects (17). Since it includes total body weight as a variable, the CG equation is inaccurate in ICU patients with unusual body composition, such as during obesity or severe malnutrition, and in the presence of ascites or extra-capillary fluids. The use of ideal body weight instead of total body weight would prevent GFR overestimation in obese patients and those with large oedema, but it requires the measurement of height. Cystatin C sensitivity to detect CG CrCl -80 ml/min was here clearly superior to creatinine (88 vs. 48%, laboratory reference values). Similar results were observed by ROC analysis (92 vs. 73% at 90% specificity, cut-off: 0.92 mg/l for cystatin C, 78 mmol/l for creatinine). This phenomenon has been observed in various clinical conditions and GFR cutoffs (70 80 ml/min) (18). In a small group of medical ICU patients (ns14), the ability of cystatin C to detect a GFR below 80 ml/min/1.73 m 2 was better than that of creatinine (8). In another recent study (9), serum cystatin C was as good as creatinine in detecting acute renal failure in ICU patients. In our sub-group of patients with impaired GFR, the q35% increase in cystatin C concentration (vs. upper normal cut-off) estimated the 48% decrease of CG CrCl more accurately than creatinine (y9%). The latter was particularly unresponsive in the 50 80 ml/min range (y19%, vs. q11% for cystatin C), as previously noted (13, 14, 18). Interestingly, intra-individual variability was in the 10% range for plasma markers and even slightly lower for cystatin C, an important issue for its routine use (19). A higher intra-individual variability was previously reported in healthy subjects, suggesting that it might be unsuitable for the longitudinal follow-up of GFR (17). Since an estimation of GFR (in ml/min) is more readily interpretable than a cystatin C concentration, mathematical GFR estimations have been proposed (14, 19, 20); their validity needs to be confirmed in ICU patients. Some aspects of the routine use of cystatin C as a GFR marker are still problematic (high cost vs. creatinine) or deserve further studies wreviewed in Ref. (4)x. In a study involving a large number of subjects (6), serum cystatin C concentration has been shown to be related to some degree to age, weight, and height. The thyroid status and hormone therapy (21), inflammation (6) or use of steroids (22) might also influence cystatin C levels. The issues are particularly relevant to ICU patients. For example, critical illness is often associated with low thyroid-stimulating hormone (TSH), reduced thyroid hormone secretion and changes in their peripheral metabolism, resulting in the socalled low triiodothyronine (T 3 ) syndrome wwith normal or decreased thyrotropin (T 4 )x (23). In ICU surgical adult patients, CG CrCl provides an easy and cost-effective estimate of GFR. Superior to creatinine, plasma cystatin C can be measured in selected patients where CG CrCl is known to be inaccurate, such as in obesity, large extra-capillary fluids and severe malnutrition. References 1. Perrone RD, Madias NE, Levey AS. Serum creatinine as an index of renal function: new insights into old concepts. Clin Chem 1992;38:1933 53. 2. Gowans E, Fraser CG. Biological variation of serum and urine creatinine and creatinine clearance: ramifications for interpretation of results and patient care. Ann Clin Biochem 1988;25:259 63. 3. Levey AS, Coresh J, Balk E, Kausz AT, Levin A, Steffes MW, et al. National kidney foundation practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Ann Intern Med 2003;139:137 47. 4. Filler G, Bökenkamp A, Hofmann W, Le Bricon T, Martínez-Brú C, Grubb A. Cystatin C as a marker of GFR history, indications, and future research. Clin Biochem 2005;38:1 8. 5. Galteau MM, Guyon M, Gueguen R, Siest G. Determination of serum cystatin C: biological variation and reference values. Clin Chem Lab Med 2001;39:850 7. 6. Knight EL, Verhave JC, Spiegelman D, Hillege HL, de Zeeuw D, Curhan GC, et al. Factors influencing serum cystatin C levels other than renal function and the impact on renal function measurement. Kidney Int 2004;65: 1416 21. 7. Dharnidharka VR, Kwon C, Stevens G. Serum cystatin C is superior to serum creatinine as a marker of kidney function: a meta-analysis. Am J Kidney Dis 2002;40: 221 6. 8. Delanaye P, Lambermont B, Chapelle JP, Gielen J, Gerard P, Rorive G. Plasmatic cystatin C for the estimation of glomerular filtration rate in intensive care units. Int Care Med 2004;30:980 3. 9. Ahlstrom A, Tallgren M, Peltonen S, Pettila V. Evolution and predictive power of serum cystatin C in acute renal failure. Clin Nephrol 2004;62:344 50. 10. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31 4. 11. Levey A, Greene T, Kusek J, Beck GJ, MDRD study group. A simplified equation to predict glomerular filtration rate from serum creatinine wabstractx. J Am Soc Nephrol 2000;11:155A. 12. Finney H, Newman DJ, Gruber W, Merle P, Price CP. Initial evaluation of cystatin C measurement by particleenhanced immunonephelometry on the Behring nephelometer systems (BNA, BNII). Clin Chem 1997;43: 1016 22.
Le Bricon et al.: Markers of renal failure in intensive care 957 13. Le Bricon T, Thervet E, Benlakehal M, Bousquet B, Legendre C, Erlich D. Changes in plasma cystatin C after renal transplantation and acute rejection in adults. Clin Chem 1999;45:2243 9. 14. Le Bricon T, Thervet E, Froissart M, Benlakehal M, Bousquet B, Legendre C, et al. Plasma cystatin C is superior to 24-h creatinine clearance and plasma creatinine for estimation of glomerular filtration rate 3 months after kidney transplantation. Clin Chem 2000;46:1206 7. 15. Brochner-Mortensen J. A simple method for the determination of glomerular filtration rate. Scand J Clin Lab Invest 1972;30:271 4. 16. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of diet in renal disease study group. Ann Intern Med 1999;130:461 70. 17. Keevil BG, Kilpatrick ES, Nichols SP, Maylor PW. Biological variation of cystatin C: implications for the assessment of glomerular filtration rate. Clin Chem 1998;44: 1535 9. 18. Laterza OF, Price CP, Scott MG. Cystatin C: an improved estimator of glomerular filtration rate? Clin Chem 2002; 48:699 707. 19. Podracka L, Feber J, Lepage N, Filler G. Intra-individual variation of cystatin C and creatinine in pediatric solid organ transplant recipients. Pediatr Transplant 2005;9: 28 32. 20. Larsson A, Malm J, Grubb A, Hansson LO. Calculation of glomerular filtration rate expressed in ml/min from plasma cystatin C values in mg/l. Scand J Clin Lab Invest 2004;64:25 30. 21. Den Hollander JG, Wulkan RW, Mantel MJ, Berhout A. Is cystatin C a marker of glomerular filtration rate in thyroid dysfunction? Clin Chem 2003;49:1558 9. 22. Cimerman N, Brguljan PM, Krasovec M, Suskovic S, Kos J. Serum cystatin C, a potent inhibitor of cysteine proteinases, is elevated in asthmatic patients. Clin Chim Acta 2000;300:83 95. 23. Sumita S, Ujike Y, Namiki A, Watanabe H, Kawamata M, Watanabe A, et al. Suppression of the thyrotropin response to thyrotropin-releasing hormone and its association with severity of critical illness. Crit Care Med 1994;22:1603 9. Received April 27, 2005, accepted July 7, 2005