Estimating Glomerular Filtration Rate (eGFR)

Christine Collier, PhD, FCACB

Chronic Kidney Disease (CKD) is a major public health problem with an incidence that is growing by 10% yearly. Major efforts are underway to identify patients at an earlier stage in their disease process in the hopes that medical intervention will halt the natural progression to end-stage disease (dialysis/transplant). In 2002, the National Kidney Foundation (NKF) in the States issued new a new 5 stage diagnostic system and practice guidelines through its Kidney Disease Outcomes Quality Initiative (KDOQI) recommending the routine reporting by clinical laboratories of an estimation of GFR (glomerular filtration rate, eGFR) as a new index of renal function that can be used as an aid in the diagnosis, monitoring and management of CKD (http://www.kidney.org/professionals/kdoqi/index.cfm). There are strong voices for and against this recommendation. On the debate side, the lack of definitive evidence, of full validation in a variety of patient populations and applications, and the limitations of current creatinine analyses have been emphasized by professionals concerned that changes are being proposed too quickly. Others respond that this disease is not being recognized early enough, that opportunities for prevention are being lost and that routine reporting of eGFR will correct this problem. Admittedly not perfect, they argue at this stage something is better than nothing. This divergence in opinion has resulted in a considerable amount of confusion as to whether or not clinical laboratories in Canada should be routinely reporting eGFR.

In a proactive initiative, the province of British Columbia undertook a provincial initiative in 2003 “standardize” the creatinine assays across the province by distributing challenge samples (human serum) with known (ID/GC/MS) reference values to determine individual laboratory regression parameters for any laboratory with a total error (TE) of more than 2%. There is an ongoing program in place that monitors the standardization for significant shifts in calibration. Before standardization, 50% of the labs could achieve a TE of 10% or less for the measurement of creatinine at a critical decision level for stage 3 renal disease (approximately 60 umol/L). After standardization, 90% of the labs could achieve this level of performance. The BC initiative was a collaboration between laboratory professionals, nephrologists, pharmacists and family physicians. This government funded project has also provided educational support and provincial treatment guidelines for chronic kidney disease to clinicians (http://www.healthservices.gov.bc.ca/msp/protoguides/gps/ckd.pdf).

In Alberta, the Alberta Kidney Disease Network (www.akdn.info) recommended implementation of eGFR(MDRD) throughout Alberta. Richard Krause (Calgary) and Laurel Thorlacius (Edmonton) have been assisting with this initiative which was implemented for outpatients only in November 2004. The Calgary Laboratory services decided to move all methods to a single enzymatic method as part of this initiative. A prospective one year trial on the effect of eGFR reporting on treatment is being performed in Calgary: family practice physicians will get either the current comment with low eGFR results, or a more detailed comment outlining recommended follow-up tests and drug therapy. Laurel will be sharing the Alberta experience at the meeting in Orlando this summer.

In Ontario, an AD Hoc OSCC committee met last summer to review the literature and consult with both nephrologists and family physicians in the province. A Clinical Laboratory Practice information sheet has been published on the website (www.clinicalchemistry.on.ca - see publications). Its purpose is to summarize the current status of the issues, and to provide recommendations for implementation if local collaborations indicate that this reporting is desired. In Ottawa, Don Greenway is working with a local nephrologist, Dr. Ayub Akbari, on several research projects addressing the efficacy of eGFR reporting. In Kingston, I am working with our family physicians who are representing the Ontario College of Family Physicians to evaluate the practical challenges of eGFR implementation.

In Quebec, some laboratories have moved towards implementation on an individual basis and others have implemented it for only nephrologists. Quebec City is performing an inter-laboratory comparison of serum creatinines. Further East, Ed Randall reports that they have been providing eGFR for about a year. And, Kent Dooley said that the Canadian Coalition for Quality in Laboratory Medicine (CCQLM) has identified eGFR as a topic for potential future guidelines.

In May this year, David Seccombe and I attended an Ad Hoc committee meeting at the annual meeting of the Canadian Society of Nephrologists (CSN) in Calgary. The committee is charged with updating the 1999 CMAJ Referral Guidelines for Patients with Elevated Creatinine. We were able to provide insight into several major laboratory issues during the discussion, facilitating a better understanding of what could and should be achievable in Canada. The new recommendations are expected to take a year to pull together, but we will have the opportunity to contribute, which is great a step forward!

The points below provide a synopsis of what current Canadian practice has implemented, as well as expectations for the next year or so.

• The evidence for the NKF and the CSN guidelines are at the level of expert opinion. It is believed that as the major risk factors for CKD are hypertension, diabetes, and heart disease, and that these diseases now all have effective treatments (drugs, etc) that will at least delay their progression, that early detection and risk factor intervention for CKD is a potentially effective strategy. Evidence indicates that the majority of patients with CKD actually die of cardiovascular disease before they go on to end-stage renal disease. However, the number of patients requiring dialysis is still growing at an alarming rate which health care systems will be unable to sustain. Another interesting statistic is that patients who first consult a nephrologist less than 3 months before requiring dialysis have a poorer prognosis than those with more adequate preparation time.
  
  
• The “Modification of Diet in Renal Disease” (MDRD) trial provided a new calculation “estimating GFR” that does not require the patient’s weight for its calculation. There are already modifications of this calculation, with the one proposed for use in the States requiring age and gender. Many USA laboratories report two results: one for Caucasians and one for African Americans (result is multiplied by 1.21). In Canada, it is preferable to report a single result, leaving the issue of body mass or race part of interpretation (and hence an educational issue).
  
  
• The MDRD equation was developed on the basis of creatinines performed on a Beckman CX3 analyzer. This Jaffe, alkaline picrate method, is known to have a positive bias, which fortuitously negated the negative bias of urine creatinines in the measurement of creatinine clearance. Recent studies have chronicled expected inter-laboratory variation due to both bias and imprecision. Early on it was suggested that laboratories “standardize” their creatinine results to the “Cleveland Clinic” where the study was originally performed on a Beckman CX3. Cleveland Clinic provided the opportunity for laboratories to send 100 patient samples for correlation. This was requested by some nephrologists because one paper suggested use of the intercept to correct the creatinine by a constant amount prior to eGFR calculation (serum creatinine results are not corrected themselves). One Canadian laboratory sent samples at the request of their nephrologists for the personal use by nephrologists on their patients. However, in an analysis of this correlation data, they noted linear regression was slightly more robust than the use of a constant correction alone. The Cleveland Clinic is no longer providing this service as the MDRD formula is being normalized for ID/GC/MS standardized creatinine test results and the NIH-NKDEP has set up a Laboratory Working group that is addressing the need to standardize creatinine measurements with the instrument manufacturers.
  
  
• In a March 2005 study, Miller et al reported on a single fresh frozen CAP (College of American Pathologist) challenge sample. They found that between manufacturer differences were more significant than the difference between methods themselves. The most frequently reported system in their data set had a bias that would contribute a 12% error to the calculation of eGFR; and, it is interesting to note that at the end of March this year, Ortho Clinical Diagnostics changed the calibration on the Vitros DT 60s to address its positive bias compared to other Vitroses. In Ontario, QMPLS has acknowledged a consistent bias between Vitros and non-Vitros methods by separate group analysis. The average positive bias of 10% at low concentrations is known by Ortho Clinical Diagnostics and related to the original decision to provide a creatinine clearance consistent with expected iodothalmate studies (due to the cancellation of errors mentioned above).
  
  
• In theory, the Vitros and Beckman (CX3) methods should both currently have positive biases, but give correct eGFR results with the current formula. Other assays that are negatively biased to these methods would result in eGFRs that are elevated using the current MDRD equation (which is based on CX3 creatinine). In the next couple of months, a revised MDRD equation will be published. It will have been determined based on creatinines measured by isotope dilution mass spectrometry (IDMS). When this is available, it is expected that manufacturers will change their serum creatinine calibration methods to be traceable to the IDMS Reference Method for creatinine. It has been estimated that this will take 2-3 years to be fully implemented. Laboratories with Beckman Cx3s or Vitroses are expected not to notice a major change in their eGFR results; other methods that are already IDMS referenced or inherently lower will have their eGFR results decrease slightly with implementation of the revised MDRD formula. The eGFR interpretation decision points will not vary as they are tied to iodothalamte clearance studies. Imprecision of the assays will not be affected by standardization and may remain unacceptably high, necessitating the call for improved performance similar to the cholesterol story. This may be particularly so for those methods that do not perform well at the lower near normal concentrations of creatinine (70 –120 umoles/L) as optimal performance over this concentration range is needed to support the accurate classification of patients with Stage 3 renal disease.
  
  
• It should be noted that the MDRD equation internally accounts for average body surface area. The inputs provided are serum creatinine, age and gender, but the units are in ml/min/1.73m2. (Some reports and discussions may have dropped the full units, which adds to the confusion). Hence if a patient has an atypical body size either due to high or low muscle mass or extra adipose tissue, interpretation of eGFR is not straight forward, and consultation with a nephrologist might be considered.
  
  
• At the present time, many physicians and pharmacists use the Cockcroft-Gault (CG) equation (serum creatinine, age, gender, weight) to calculate drug dosages. It is preferable to err on the side of under dosing as the dose can be adjusted slowly upwards. Apparently the eGFR-MDRD may tend towards underdosing compared to the CG calculation. The eGFR-MDRD equation remains to be fully validated for drug dosing. This is a major concern for some opponents of implementation as it may introduce a significant risk issue. However, several pharmacy groups point out that eGFR is just one of the tools in considering a drug dose, and as such, there is not much difference between the various calculations. A common question about eGFR-MDRD is “what is the difference between the calculations?” – ie “is the new calculation better”? The short answer is “maybe”, but actually there is no short answer as it depends to some extent on who is reading the literature. Our nephrologists consider these and many of the previous equations equivalent clinically. But it should be pointed out that neither is actually as close to the expected iodothalmate clearance as a chemist would expect, with 90% of eGFR results within 30% of their true iodothalmate value!
  
  
• It is interesting to consider the concept that “creatinines are insensitive”, but that eGFR will be useful when the latter is essentially the former modified for age and gender. In reconciling this, two things need to be appreciated: that the indirect relationship between serum creatinine and GFR will always contribute to a numbers game; and that serum creatinines should not have a population reference range. An excel spreedsheet to model and graph the correlations between serum creatinine (x) and eGFR (y) (using different equations) can be easily set up. The effect of gender and age can be estimated, and the extremes of body weight can be compared. An important consideration is the upper reporting limit for eGFR. The NKF guidelines recommend that values >60 ml/min/1.73m2 not be reported until such time as the calibration of the creatinine measurement systems have been standardized and the MDRD formula modified for the use of standardized creatinine test results. British Columbia has standardized their creatinine measurements to ID/GC/MS and are currently reporting eGFR values from 5 to 150 ml/min/1.73m2. It is likely that BC will decrease their upper limit of reporting to 120 ml/min/1.73m2 - aligning BC with Ontario. However, even with standardization, the indirect and exponential relationship between creatinine and eGFR will always produce a wider absolute 95% confidence interval for the eGFR, which can only be balanced by better precision in the creatinine assays and repeat patient testing.
  
  
• Although it is well known that GFR decreases with age, age and gender stratified reference ranges for serum creatinine in adults are not common. Hilde Vandenberghe at Gamma-Dynacare was able to pull a tremendous population reference range review by decade. The data was not very illuminating with the upper tail on the graphs just increasing with each decade. Mike McNeely has suggested a solution that seems appropriate: develop calculated reference ranges backwards from eGFR for each decade and gender using the stage 3 cutoff of 60 ml/min/1.73m2 (CKD is diagnosed when eGFR is below this for 3 months).
  
  
• For eGFR in children, the Schwartz formula is still used. I am not aware that automatic calculation in children has been recommended. BC Children’s Hospital has not implemented anything at this time, and Nathalie Lepage at CHEO in Ottawa presented a poster in this area at the CSCC in London, if anyone wants more information.
  
  
• Biological variation, once again has shown its importance in clinical chemistry. Callum Fraser’s book on this subject uses serum creatinines as the example of a test that should not legitimately be compared to a population reference range. This is because, like total CK, its within individual variation is much smaller than between individual variation. A person’s creatinine could double and still be within the reference range. Serial testing is the answer for eGFR. A real change in value (RCV) is 20% for serum creatinine, with a highly significant change occurring at 30% (based on commonly achieved analytical variation). For glomerular filtration rate, a significant annual change is a confirmed10% change.
  
  
• It should be noted that eGFR is meant to be automatically calculated and reported whenever serum creatinine is ordered. This may produce challenges for inclusion of more informative interpretive comments on reports, especially cumulative reports. In Alberta eGFR is being reported only for outpatients, emphasizing the point that eGFR is only valid in stable patients. In acute disease, serum creatinine is the test of choice.
  

eGFR is an “estimate” by definition, and as mentioned above is only an approximation of the iodothalamate measurement. “Estimating” is one of the calculated risks that clinicians contend with daily in their practice. Some feel this is a risk we should not encourage, and others point out that the burden of referrals on nephrologists will be overwhelming. Hopefully, definitive evidence-based research will be helpful in the future. Until then, emphasis on serial eGFR monitoring is important as repeat testing is an effective strategy to improve reliability. It may discourage over-interpretation of single results as well as avoiding the misdiagnosis of patients with inherently low but stable eGFRs. This will probably be an active field in clinical biochemistry over the next couple of years. In the interim, eGFR should help the general practitioner identify patients at risk at an earlier stage in their disease process providing them with an opportunity for prevention through the appropriate management of associated risk factors.

David and I would be pleased to chat about eGFR and its implementation if anyone has further questions.

Christine Collier
July 2005

[Back to top] [OSCC Home]