Estimating GFR from serum creatinine concentration: Pitfalls of GFR-estimating equations

Abstract

Related Article, p. 463 Related Article, p. 463 ASSESSMENT OF glomerular filtration rate (GFR), the most frequently performed test of kidney function, allows evaluation of the severity of kidney failure, facilitates drug dosing, and assists in evaluating uremic symptoms.1Levey A.S. Measurement of renal function in chronic renal disease.Kidney Int. 1990; 38: 167-184Crossref PubMed Scopus (472) Google Scholar Although GFR can be measured directly by using parenteral administration of inulin, iohexol, or iothalamate, direct measurements commonly are not made because of cost, inconvenience, and nonavailability of these tests to the vast majority of patients with chronic kidney disease (CKD). However, GFR often is estimated from serum creatinine level and anthropometric and clinical characteristics of patients. Creatinine clearance is directly proportional to creatinine generation and inversely proportional to serum creatinine concentration. In practice, the broadly used Cockcroft-Gault formula2Cockcroft D.W. Gault M.H. Prediction of creatinine clearance from serum creatinine.Nephron. 1976; 16: 31-41Crossref PubMed Scopus (12909) Google Scholar uses serum creatinine concentration, age, weight, and sex to estimate creatinine clearance. It is based on observations, reported by Cockcroft and Gault2Cockcroft D.W. Gault M.H. Prediction of creatinine clearance from serum creatinine.Nephron. 1976; 16: 31-41Crossref PubMed Scopus (12909) Google Scholar in 249 hospitalized patients aged 18 to 92 years, that the generation of creatinine assessed by means of 24-hour collection of urine is directly proportional to weight, decreases with age, and is lower in women. The correlation coefficient between their formula and creatinine clearance was as good as the correlation coefficient between duplicate measurements of creatinine clearance. Thus, measurement of creatinine clearance using timed urine collections does not provide more accurate estimates of GFR than prediction equations. There are several pitfalls of this simple-to-use formula. First, body weight is an imperfect reflection of creatinine generation because increased body weight is associated more commonly with an increase in body fat or body water, edematous disorders, rather than an increase in muscle mass.3Rolin III, H.A. Hall P.M. Wei R. Inaccuracy of estimated creatinine clearance for prediction of iothalamate glomerular filtration rate.Am J Kidney Dis. 1984; 4: 48-54PubMed Google Scholar Second, a variety of wasting illnesses, liver cirrhosis, and limb amputations often are associated with a lower rate of creatinine generation than would be reflected by body weight alone. Third, the formula is calibrated to creatinine clearance as the reference standard, not the direct measurement of GFR. Because creatinine is secreted in the early stages of CKD, measurement of creatinine clearance may substantially overestimate the true GFR, especially in patients with CKD stage 2.4Shemesh O. Golbetz H. Kriss J.P. Myers B.D. Limitations of creatinine as a filtration marker in glomerulopathic patients.Kidney Int. 1985; 28: 830-838Crossref PubMed Scopus (989) Google Scholar A low-protein diet can reduce tubular creatinine secretion and creatinine generation, whereas antihypertensive therapy can reduce tubular secretion without influencing GFR.5Modification of Diet in Renal Disease Study GroupEffects of diet and antihypertensive therapy on creatinine clearance and serum creatinine concentration in the Modification of Diet in Renal Disease study.J Am Soc Nephrol. 1996; 7: 556-565PubMed Google Scholar Fourth, serum creatinine also is influenced by creatinine intake, such that a cooked-meat meal can substantially increase serum creatinine concentration without affecting creatinine clearance.6Mayersohn M. Conrad K.A. Achari R. The influence of a cooked meat meal on creatinine plasma concentration and creatinine clearance.Br J Clin Pharmacol. 1983; 15: 227-230Crossref PubMed Scopus (96) Google Scholar Finally, drugs and other chemicals can interfere with creatinine secretion or estimation.7Levey A.S. Perrone R.D. Madias N.E. Serum creatinine and renal function.Annu Rev Med. 1988; 39: 465-490Crossref PubMed Scopus (410) Google Scholar Recognizing these limitations of the Cockcroft-Gault and other creatinine clearance-based formulas, Levey et al8Levey A.S. Bosch J.P. Lewis J.B. Greene T. Rogers N. Roth D. Modification of Diet in Renal Disease Study GroupA more accurate method to estimate glomerular filtration rate from serum creatinine A new prediction equation.Ann Intern Med. 1999; 130: 461-470Crossref PubMed Scopus (12781) Google Scholar developed an estimating equation for GFR from knowledge of the serum creatinine concentration and other clinical characteristics, using directly measured GFR in 1,628 participants with CKD enrolled in the baseline phase of the Modification of Diet in Renal Disease (MDRD) study. Because the MDRD formula is not calibrated to creatinine clearance, but GFR, it addresses one shortcoming of the Cockcroft-Gault equation. Furthermore, the MDRD formula does not use body weight in estimating GFR. Although blood urea nitrogen and serum albumin concentrations also were found to correlate with GFR (MDRD equation 7), contributions of these two variables to the final equation are so small that Manjunath et al9Manjunath G. Sarnak M.J. Levey A.S. Prediction equations to estimate glomerular filtration rate An update.Curr Opin Nephrol Hypertens. 2001; 10: 785-792Crossref PubMed Scopus (246) Google Scholar recommended a simpler 4-component formula that includes just age, sex, race, and serum creatinine concentration. The abbreviated MDRD formula has been championed by the National Kidney Foundation-Kidney Disease Outcomes Quality Initiative to stage CKD.10Levey A.S. Coresh J. Balk E. et al.National Kidney Foundation Practice Guidelines for Chronic Kidney Disease Evaluation, classification, and stratification.Ann Intern Med. 2003; 139 (Erratum in Ann Intern Med 139:605, 2003): 137-147Crossref PubMed Scopus (3602) Google Scholar This formula also has been used to assess the prevalence of CKD in the general population.11Coresh J. Astor B.C. Greene T. Eknoyan G. Levey A.S. Prevalence of chronic kidney disease and decreased kidney function in the adult US population Third National Health and Nutrition Examination Survey.Am J Kidney Dis. 2003; 41: 1-12Abstract Full Text Full Text PDF PubMed Scopus (2275) Google Scholar Thus, performance characteristics of this formula are of substantial public health importance. Accordingly, interest has emerged in judging the accuracy and precision of this formula. To judge the adequacy of this formula, an adequate sample size over a wide range of GFR values and calibration of the serum creatinine measurement to the reference standard are required.12Coresh J. Astor B.C. McQuillan G. et al.Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate.Am J Kidney Dis. 2002; 39: 920-929Abstract Full Text Full Text PDF PubMed Scopus (613) Google Scholar Furthermore, a reference standard for GFR similar to that used in the MDRD study would be required. In this issue of the American Journal of Kidney Diseases, Zuo et al13Zuo L. Ma Y.-C. Zhou Y.-H. Wang M. Wang H.-Y. Xu G.-B. Application of GFR-estimating equations in Chinese patients with chronic kidney disease.Am J Kidney Dis. 2005; 45: 463-472Abstract Full Text Full Text PDF PubMed Scopus (197) Google Scholar report performance characteristics of this formula in Chinese patients. They studied 261 Chinese patients with stable CKD with a wide variety of causes of kidney disease.13Zuo L. Ma Y.-C. Zhou Y.-H. Wang M. Wang H.-Y. Xu G.-B. Application of GFR-estimating equations in Chinese patients with chronic kidney disease.Am J Kidney Dis. 2005; 45: 463-472Abstract Full Text Full Text PDF PubMed Scopus (197) Google Scholar The proportion of patients with diabetes mellitus is not specifically reported, and a larger than usual percentage of patients had obstructive uropathy (18.8%). The distribution of patients is between 30 and 90 patients in each of the 5 stages of K/DOQI categories. The investigators were careful to exclude people with edema, cachexia, and amputation, thus obviating the pitfalls associated with estimation of GFR from serum creatinine level. An appropriate reference standard was used to measure GFR, and performance of the MDRD equation 7, abbreviated MDRD equation, and Cockcroft-Gault formulas was evaluated. The method of measuring serum creatinine concentration (the Jaffé kinetic reaction) was the same as that used in the MDRD study. Because the Hitachi autoanalyzer used in this study consistently overestimates serum creatinine compared with measurements made for the MDRD study, an adjustment for this error was made. The major findings of the study are that the MDRD equation underestimates GFR in patients with CKD stage 1, whereas it overestimates GFR in those with CKD stages 4 and 5. Approximately 25% of results estimated using either MDRD equation, but only 20% of results estimated using the Cockcroft-Gault equation, were more than 50% off the true GFR value. By comparison, in a white population, 29% of results estimated using MDRD equation 7, 18% using the abbreviated MDRD equation, and only 12% using the Cockcroft-Gault equation were more than 50% off the measured GFR value.14Bostom A.G. Kronenberg F. Ritz E. Predictive performance of renal function equations for patients with chronic kidney disease and normal serum creatinine levels.J Am Soc Nephrol. 2002; 13: 2140-2144Crossref PubMed Scopus (326) Google Scholar It is not surprising that the formula based on serum creatinine concentration did not perform well. First, performance of the MDRD formula is dependent on ethnicity. Because only a small number of people of Chinese descent were used in the original validation of the formula, it is no surprise that the formula did not perform well in this population. Second, because creatinine is secreted into the proximal tubule in early stages of CKD, it is expected that serum creatinine would change little in early stages of CKD.15Bauer J.H. Brooks C.S. Burch R.N. Clinical appraisal of creatinine clearance as a measurement of glomerular filtration rate.Am J Kidney Dis. 1982; 2: 337-346PubMed Scopus (208) Google Scholar Because the MDRD study did not have enough patients with relatively well-preserved GFR, it was not calibrated to this level of kidney function. Even if MDRD GFR was measured in patients with early kidney disease, it is unlikely to improve the precision and accuracy of the GFR estimate because of tubular secretion of creatinine in patients with early CKD. In patients with normal serum creatinine levels, absolute GFR14Bostom A.G. Kronenberg F. Ritz E. Predictive performance of renal function equations for patients with chronic kidney disease and normal serum creatinine levels.J Am Soc Nephrol. 2002; 13: 2140-2144Crossref PubMed Scopus (326) Google Scholar, 16Nielsen S. Rehling M. Schmitz A. Mogensen C.E. Validity of rapid estimation of glomerular filtration rate in type 2 diabetic patients with normal renal function.Nephrol Dial Transplant. 1999; 14: 615-619Crossref PubMed Scopus (31) Google Scholar or change in GFR16Nielsen S. Rehling M. Schmitz A. Mogensen C.E. Validity of rapid estimation of glomerular filtration rate in type 2 diabetic patients with normal renal function.Nephrol Dial Transplant. 1999; 14: 615-619Crossref PubMed Scopus (31) Google Scholar cannot be assessed reliably from GFR-estimating equations. What alternatives exist for the clinician in situations in which the MDRD equation does not perform well? In patients with early-stage CKD, an emerging GFR marker is serum cystatin C concentration.17Mussap M. Dalla V.M. Fioretto P. et al.Cystatin C is a more sensitive marker than creatinine for the estimation of GFR in type 2 diabetic patients.Kidney Int. 2002; 61: 1453-1461Crossref PubMed Scopus (223) Google Scholar, 18Risch L. Blumberg A. Huber A. Rapid and accurate assessment of glomerular filtration rate in patients with renal transplants using serum cystatin C.Nephrol Dial Transplant. 1999; 14: 1991-1996Crossref PubMed Scopus (118) Google Scholar Cystatin C is not secreted in patients with early stages of CKD and may be more sensitive in detecting early-stage CKD. Another attractive technique is to use oral cimetidine to block tubular creatinine secretion and estimate creatinine clearance using the Cockcroft-Gault formula19Ixkes M.C. Koopman M.G. Van Acker B.A. Weber J.A. Arisz L. Cimetidine improves GFR-estimation by the Cockcroft and Gault formula.Clin Nephrol. 1997; 47: 229-236PubMed Google Scholar or measure renal creatinine clearance.20Van Acker B.A. Koomen G.C. Koopman M.G. de Waart D.R. Arisz L. Creatinine clearance during cimetidine administration for measurement of glomerular filtration rate.Lancet. 1992; 340: 1326-1329Abstract Full Text PDF PubMed Scopus (138) Google Scholar Results of creatinine clearance measured with 800 mg of cimetidine administered every 8 hours for 3 doses are close to those obtained using a standard GFR marker.19Ixkes M.C. Koopman M.G. Van Acker B.A. Weber J.A. Arisz L. Cimetidine improves GFR-estimation by the Cockcroft and Gault formula.Clin Nephrol. 1997; 47: 229-236PubMed Google Scholar Using an enzymatic assay instead of a colorimetric serum creatinine assay further improves results of the cimetidine-assisted GFR assessment.21Kemperman F.A. Silberbusch J. Slaats E.H. et al.Glomerular filtration rate estimation from plasma creatinine after inhibition of tubular secretion Relevance of the creatinine assay.Nephrol Dial Transplant. 1999; 14: 1247-1251Crossref PubMed Scopus (28) Google Scholar In patients with advanced kidney disease, urea is reabsorbed by the renal tubule. The arithmetic mean of renal urea and renal creatinine clearance are a good approximation of kidney function in such an advanced state of kidney disease,22Lavender S. Hilton P.J. The measurement of glomerular filtration-rate in renal disease.Lancet. 1969; 2: 1216-1219Abstract PubMed Google Scholar in which the MDRD equation overestimates GFR. From the clinical perspective, users of this formula must recognize that the MDRD formula, as well as other creatinine clearance formulas, is only an estimate of GFR. Because creatinine has inherent errors as a marker of glomerular function, all formulas that rely on serum creatinine will share these errors in GFR. If estimates in individual patients are made, wide margins of error are likely to be recognized if GFR is measured simultaneously. In other words, the prediction interval for the estimate in an individual is likely to be wide. Although precision in the MDRD formula may be sufficient to accurately stage a patient with respect to CKD, following up the progression of kidney disease with the MDRD formula will have the same shortcomings that are shared by measurements of serum creatinine concentration.4Shemesh O. Golbetz H. Kriss J.P. Myers B.D. Limitations of creatinine as a filtration marker in glomerulopathic patients.Kidney Int. 1985; 28: 830-838Crossref PubMed Scopus (989) Google Scholar, 23Levey A.S. Berg R.L. Gassman J.J. Hall P.M. Walker W.G. Modification of Diet in Renal Disease (MDRD) Study GroupCreatinine filtration, secretion and excretion during progressive renal disease.Kidney Int Suppl. 1989; 27: S73-S80PubMed Google Scholar, 24Walser M. Drew H.H. LaFrance N.D. Creatinine measurements often yield false estimates of progression in chronic renal failure.Kidney Int. 1988; 34: 412-418Crossref PubMed Scopus (114) Google Scholar However, from the study of Zuo et al,13Zuo L. Ma Y.-C. Zhou Y.-H. Wang M. Wang H.-Y. Xu G.-B. Application of GFR-estimating equations in Chinese patients with chronic kidney disease.Am J Kidney Dis. 2005; 45: 463-472Abstract Full Text Full Text PDF PubMed Scopus (197) Google Scholar it appears that the most convenient way to estimate GFR in Chinese patients is to use GFR estimated by means of the Cockcroft-Gault formula without the gender correction. Such estimation can be performed at the bedside by multiplying the Cockcroft-Gault creatinine clearance by 0.84 and can facilitate clinical decision making in Chinese patients. The MDRD formula in Chinese patients was not superior to the Cockcroft-Gault method; therefore, in my opinion, the more cumbersome-to-use MDRD formula is not necessary in Chinese patients. Even in the larger MDRD study, when the Cockcroft-Gault formula was corrected for overestimation of GFR by multiplying the result by 0.84, it predicted 86.6% of the variance of the logarithm of GFR compared with the MDRD formula, which predicted 90.3% of the variance. Thus, if access to a computer is not available, “back-of-the-envelope” calculations using the Cockcroft-Gault formula are worth performing.