Comparing Equations for Kidney Function
Study Shows MDRD Study Equation More Accurate for Drug Dosing Estimates
By Stuart Zehner
Labs have used the Cockcroft-Gault equation for decades to estimate glomerular filtration rate in patients with chronic kidney disease. The more recently developed Modification of Diet in Renal Disease Study equation also uses serum creatinine concentration as a way to estimate GFR. However, the formula has been modified to be used with serum creatinine methods that have been calibrated to isotope-dilution mass spectrometry, and it is widely considered to provide more accurate estimates of kidney function. A recently published study compares the use of these two equations for estimating kidney function and drug dosing recommendations. This issue of Strategies examines those findings.
Kidney diseases often alter the pharmacokinetic properties of many medications, requiring clinicians to estimate kidney function using either the Cockcroft-Gault (CG) or Modification of Diet in Renal Disease (MDRD) Study equation. Since the publication of Guidance for Industry: Pharmacokinetics in Patients With Impaired Renal Function by FDA in 1998, pharmaceutical companies have relied on the CG equation when developing drug dosing guidelines, despite the fact that the majority of labs now report estimated glomerular filtration rate (eGFR) using the MDRD Study equation. Although previous studies have compared drug dosing recommendations based on the CG and MDRD Study equations, first-of-its-kind research compared estimated kidney function and hypothetical drug dosing recommendations calculated using both these equations against an accurate, measured GFR in a clinically diverse population (Am J Kidney Dis 2009;54(1):33-42).
The team of researchers, led by Lesley A. Stevens, MD, assistant professor of medicine at Tufts University School of Medicine, used a data set obtained from the Chronic Kidney Disease Epidemiology Collaboration, pooling participants from six research studies and four clinical populations across a wide range of GFRs and clinical characteristics. In total, 5,504 participants were included in the study population whose GFRs were measured using urinary clearance of I
125-iothalamate. Kidney function was then estimated using three equations: MDRD Study equation; CG; and CG equation adjusted for ideal body weight (CGIBW). All creatinine results were corrected to be standardized to an isotope-dilution mass spectrometry reference procedure.
Participants were assigned to one of four kidney function categories (>80, 50 to 80, 30 to 49, or <30mL/min) based on their measured GFR and kidney function estimates obtained from the three equations. Drug dosage recommendations were determined for 15 medications calculated using measured GFR and the three estimates.
The mean measured GFR for all participants was 75+44mL/min; eGFR calculated from the MDRD Study equation, CG, and CGIBW were 69+38, 75+42, and 62+36 mL/min, respectively. The MDRD Study equation showed the highest level of concordance with measured GFR and assigned kidney function categories at 78%; the CGIBW showed the lowest level of concordance at 66%.
Significantly, the equations differed from the measured GFR in distinct ways. The CG equation assigned a higher kidney function category compared with the measured GFR in 16% of participants, compared to 5% for the CGIBW and 8% for the MDRD Study equation. CGIBW assigned a lower kidney function category in 29% of participants, compared to 12% for the CG equation and 14% for the MDRD Study equation. Overall, the MDRD Study equation showed the greatest agreement with measured GFR for all subgroups analyzed.
The MDRD Study equation was also 88% concordant for specific drug dosing recommendations when compared to measured GFR. The CG equation was 85% concordant and the CGIBW equation was 82% concordant for drug dosing recommendations. In the same manner that the CG equation assigned higher kidney function categories, the CG equation was more likely to recommend greater drug dosages, and the CGIBW was more likely to recommend lower drug dosages.
Although the study confirms the accuracy of the MDRD Study equation for estimating GFR, it was not without its unexpected results. "What’s surprising about the findings is that the correction for ideal body weight for the Cockcroft-Gault equation, which has traditionally been assumed to be a better estimate of kidney function, did substantially worse compared to measured GFR," said Stevens. "The CG equation was developed compared to a gold standard measurement of creatinine clearance, which overestimates GFR due to the tubular secretion of creatinine. Even adjusting for the difference, the CG equation is much less accurate because weight can mean a lot of different things—muscle, fat—but only muscle is related to creatinine. The CGIBW equation probably overadjusts for weight related to fat."
The study is also important because it showed that the CG equation, which the FDA recommends pharmaceutical companies use when developing drug dosing guidelines, is not the most accurate equation for that purpose. "The most significant finding is that using the MDRD Study equation to estimate drug dosages is equivalent to, if not slightly better than, the CG equation that has been used for drug labeling," said Greg Miller, PhD, professor of pathology at Virginia Commonwealth University and chair of the Laboratory Working Group of the National Kidney Disease Education Program (NKDEP). "Drugs are distributed with labeling that says, ‘Do your dosage estimates based on the CG equations,’ and in many cases it doesn’t allow for the use of other equations." Miller was not involved with this study.
Both Stevens and Miller agree that it is time to consider adopting other equations shown to be more accurate for estimating GFR. "The CG equation was developed many years ago using a creatinine method that is no longer available and from a very small study. It was designed to estimate creatinine clearance, not GFR," explained Miller. "The CG equation was not really intended to quantify GFR; it was intended to measure kidney function. MDRD was developed in the late 1990s and intentionally developed to estimate GFR, so it provides a much more accurate assessment of actual GFR."
Even with an estimated 70% of clinical laboratories reporting eGFR using the MDRD Study equation, according to the study’s authors, more accurate equations have already been developed. "There is a new equation, the CKD-EPI equation, which improves upon the MDRD Study equation by including a more diverse data set and different transformations of the variables to better capture the relationship," Stevens noted. "However, even this equation can’t overcome the limitations of creatinine, because there is always some individual variability in creatinine generation. As clinicians we need to understand the importance of confirmatory tests whenever we think the estimates are incorrect, or when it’s important to get accurate values such as in drug dosing for medications with narrow therapeutic indexes and high degrees of toxicity."
Miller agrees, although he suggests caution in adopting new equations for GFR. "The MDRD Study equation has been shown repeatedly to be a more accurate equation for GFR at less than 60 mL/min/1.73m2, but it has already been replaced by the CKD-EPI equation. The jury is still out if the new
equation will be adopted by laboratories, but it appears to be more accurate and will allow numeric values to be reported that are greater than 60 mL/min/1.73m2."
Ultimately, clinicians need to be aware of the most accurate methods currently available to estimate GFR and be able to use them. "FDA is developing a draft guidance that uses the MDRD Study equation for drug dosing," said Stevens. "One recommendation to make would be not to restrict it to specific equations, but to always be able to use the best estimate. We should always be able to use the most accurate tool."
Based on the current study, Stevens, Miller, and the Laboratory Working Group of NKDEP have prepared a new set of recommendations for use of kidney function estimates for drug dosing that was expected to be published on the NKDEP website in late August 2009.
Disclosure: Lesley Stevens receives research grant support from Gilead, Inc