May 2, 2006 Presentation: Reporting Estimated GFR from Serum Creatinine: Recommendations from the Laboratory Working Group of the National Kidney Disease Education Program
Welcome to AACC’s Expert Access Live Online program
Reporting Estimated GFR from Serum Creatinine: Recommendations from the Laboratory Working Group of the National Kidney Disease Education Program
This month's expert is W. Gregory Miller, PhD, DABCC, FACB. View the presentation and direct your questions to our online expert. AACC would like to thank Bayer HealthCare Diagnostics for making this program possible.
In regards to MDRD equation limitations, talk more about validation is underway for additional ethnic groups, patient groups, and individuals with normal renal function
W. Gregory Miller, PhD, DABCC, FACB: There are studies underway that will complete during 2006-7 and will evaluate the applicability of eGFR in several populations of particular interest including diabetics, ethnic groups and people with more normal renal function than were represented in the original MDRD study. The NKDEP web site will be updated with information as it is available and provide the literature citations. A summary of current information is being prepared and will be available on the NKDEP website in the next month.
We are using the kinetic alkaline picrate reaction on the Beckman Coulter UniCel DxC 800 instrument to analyze creatinine. That method is not calibrated to the IDMS standard. Which MDRD equation should we use? We are a hospital-based laboratory with inpatients and an outreach program. For whom should the eGFR be calculated, i.e. inpatients, outpatients, normals, pediatrics, elderly (age limits)?
W. Gregory Miller, PhD, DABCC, FACB: Creatinine methods that are not calibrated to produce results that are traceable to IDMS reference methods should use the original MDRD equation called “conventional calibration equation” on the web site. This equation was developed and validated using creatinine results from a Beckman CX3 analyzer at the Cleveland Clinic that was the central laboratory for the MDRD study. Most other creatinine methods that use conventional calibration (i.e. the calibration that has been in use for many years) produce results that have relatively small biases when compared to the Beckman CX3 results (see Clin Chem 2006; 52: 5-18). Consequently, the original equation is appropriate. It is recommended to report eGFR with all creatinine results for patients 18 years and older. The primary goal of the NKDEP is to identify persons who have chronic kidney disease, and providing the eGFR allows the physician to better interpret the creatinine value. Although the MDRD study did not include persons older than age 70, there is no evidence the equation will not perform well in older patients. An eGFR value should not be reported for persons with normal or moderately impaired kidney function because the bias and imprecision of creatinine results in the range observed for these persons causes an excessively large confidence interval around the calculated result, and the equation is less accurate when the kidney function is less impaired. The NKDEP recommendation is to report >60 mL/min/1.73m2 when the calculated value is >60. Reporting eGFR for hospital inpatients is a decision for the laboratory director. The accuracy of the MDRD equation is poorer for inpatients and is not suitable for acute renal failure patients. The Laboratory Working Group report in Clin Chem 2006;52:5-18 includes details on these issues.
We are currently reporting the MDRD GFR on all of our serum creatinines. We use Beckman Coulter instrumentation for measurement of creatinine. Has this manufacturer standardized their creatinine to the isotope dilution mass spec. method that you recommend? Also, our PharmD's want us to put a qualifying comment on all those over 65 years of age, saying the equation has some limitations in this patient population. Do you agree?
W. Gregory Miller, PhD, DABCC, FACB: To my knowledge, Beckman has not changed the calibration of its systems to be traceable to IDMS. The NIST reference material for serum creatinine is expected to be available in mid 2006 and will enable most IVD manufacturers to initiate standardization of their measurement systems. Although the MDRD study did not include persons older than age 70, there is no evidence the equation will not perform well in older patients. It is not necessary to add a qualifying comment to results on older patients.
Are there IDMS-Traceable versions of the MDRD6 and MDRD7 equations or will they be available soon? These equations tend to be used more in clinical research than the abbreviated MDRD equation.
W. Gregory Miller, PhD, DABCC, FACB: Only the 4 parameter MDRD equation has been updated for methods that produce results that are traceable to IDMS. The 4 parameter equation is recommended because it requires fewer laboratory results, is essentially as accurate as the 6 parameter equation, but is easier to implement.
Is the eGFR different from GFR used to calculate pharmaceutical drugs? Are they inter changable?
W. Gregory Miller, PhD, DABCC, FACB: The MDRD equation estimates GFR. The Cockcroft-Gault equation estimates creatinine clearance. The values are not interchangeable. Clinical guidelines used by pharmacists to adjust drug dosages for patients with impaired kidney function are usually based on the Cockcroft-Gault estimating equation for creatinine clearance or on the serum creatinine result. The Cockcroft-Gault equation was developed using serum creatinine methods that had conventional calibration and were biased about 10-20% higher than results by IDMS. When laboratories introduce serum creatinine methods with calibration traceable to IDMS, a patient’s creatinine results will become lower, in most cases, and the estimated GFR using any of the estimating equations or algorithms derived from older creatinine methods will become higher by roughly the same percentage. Depending on the analytical system used by the clinical laboratory, and how that method calibrates urine creatinine measurements, the creatinine clearance calculated from serum and urine measurements will either be unaffected by the recalibration of creatinine or may increase slightly. For drug dosing purposes, the NKDEP does not recommend utilization of the MDRD equation at this time because the clinical impact on drug dose adjustment has not been compared between current practice and the MDRD equation. Pharmacists should continue to use their current drug dosing methods. Following implementation of serum creatinine methods with calibration traceable to IDMS, the clinical laboratory should notify the pharmacy and drug prescribes to inform them of the expected magnitude of change in serum creatinine values, and whether the creatinine clearance measured from serum and urine will be affected by the change. Clinical providers will need to compensate for the magnitude of change in creatinine values when applying the algorithms for drug dose adjustment.
The recommendations/guidelines are to use two decimal places for calculating eGFR. Does that apply to reporting of creatinines in mg/dL as well? Thanks.
W. Gregory Miller, PhD, DABCC, FACB: The recommendation is to use two decimals (e.g. 0.95 mg/dL) or the nearest whole number for SI units (e.g. 89 umol/L) for calculating eGFR. The reason for this recommendation is to reduce any rounding influence on the uncertainty in the calculated value. For reporting, it is not necessary to use two decimals. However, many LIS systems do not have the capability to use a different number of decimals for a calculation and a reported value; consequently, reporting two decimals may be necessary for this reason.
Do current LISs have the capability of doing the MDRD eGFR calculation? If so: (1) Can an LIS report out TWO results; one for African-Americans AND one for non-African-Americans from one creatinine result? (2) Can the LIS suppress a result if patient is less than 18 years old? (3) Can the LIS calculate and report gender-specific results (and give the two results stated above only for that gender)? (4) Can the LIS report out the result as >60mL/min/1.73m2 when appropriate.
W. Gregory Miller, PhD, DABCC, FACB: Some LIS systems have the capability to perform all these actions. It is necessary to contact individual LIS vendors to determine if a given system can support the calculations and reporting recommendations. If an LIS cannot support the calculation and reporting requirements, then it will not be possible for that laboratory to report eGFR at this time. Note, that the mathematics can be performed using logarithms for some systems that may not support fractional exponentiation but do support logarithmic functions.
I understand that the MDRD equation was based on mainly fasting blood samples for creatinine measurement. Is it possible that using post-prandial samples in routine practice will result in artefactual underestimation of GFR?
Wishaw, Lanarkshire, Scotland.
W. Gregory Miller, PhD, DABCC, FACB: Creatinine concentration is generally not affected by postprandial conditions, so a nonfasting sample can be used. Some dietary conditions can affect the creatinine concentration. For example, a vegetarian diet may cause a small decrease in creatinine level; regular ingestion of creatine supplements can cause an increase in creatinine level; creatinine may be transiently increased following ingestion of large amounts of meat.
Should we be using a standardized creatinine for the GFR calculation? Should we be reporting GFR on any patient with a creatinine test? What disclaimers should we use for certain patient groups?
W. Gregory Miller, PhD, DABCC, FACB: Creatinine methods that are not calibrated to produce results that are traceable to IDMS reference methods should use the original MDRD equation called “conventional calibration equation” on the web site. This equation was developed and validated using creatinine results from a Beckman CX3 analyzer at the Cleveland Clinic that was the central laboratory for the MDRD study. Most other creatinine methods that use conventional calibration (i.e. the calibration that has been in use for many years) produce results that have relatively small biases when compared to the Beckman CX3 results (see Clin Chem 2006; 52: 5-18). Consequently, the original equation is appropriate. It is important to coordinate using the IDMS traceable MDRD equation (see the NKDEP web site) with use of a creatinine method that has results calibrated to be traceable to IDMS. It is also important for laboratories to inform clinical providers and pharmacists that creatinine results with an IDMS traceable method will be lower, in most cases, and to provide the relationship between the IDMS traceable method and the conventionally calibrated method(s) it is replacing. This information is important for adjustment of drug dosages for patients with impaired kidney function and was discussed in more detail in another response. The creatinine standardization program is primarily directed to IVD manufacturers who will change their method’s calibration to produce results that are traceable to the IDMS reference method. It is considered much more efficient and sustainable to have IVD manufacturers make calibration changes than to expect individual laboratories to adjust the calibration of their methods. A new NIST reference material for serum creatinine is expected to be available in mid 2006 and will enable most IVD manufacturers to initiate standardization of their measurement systems. The transition to standardized creatinine methods is expected to take 1-2 years to be completed because new product labeling will be required, and inventories currently in the field will need to be used and replaced with the revised calibration systems. The questions about reporting eGFR and disclaimers for certain types of patients were answered in a previous response.
Are there reagent manufacturers that have already "standardized"? If so, is there a listing of which manufacturers have gone to a "standardized" calibration? If not, what is the expected time line?
W. Gregory Miller, PhD, DABCC, FACB: Roche has standardized many of its creatine procedures to have results traceable to IDMS. Refer to the frequently asked questions on the NKDEP Laboratory Professionals web site: Roche Enzymatic Creatinine method is traceable to IDMS; Roche Global Creatinine Jaffe methods are traceable to IDMS; Roche US Creatinine Jaffe methods are directly traceable to SRM 914a and indirectly traceable to IDMS through a Masterlot.
Do you have any information on the time lines for the different vendors (Bayer, Dade, Roche, OCD etc) of serum creatinine assays to recalibrate their methods to the IDMS traceble material. We have a large network of labs, using different instrumentation, that have been harmonized to provide consistent serum creatinine results and report the conventional MDRD GFR on lab reports. The re-calibration to IDMS requires that we co-ordinate that change of calibration, (and calculation) over multiple methods and laboratories to provide the least confusing transition. Thank you.
Edmonton, Alberta, Canada
W. Gregory Miller, PhD, DABCC, FACB: The timeline for individual IVD manufacturers to introduce standardization of calibration to IDMS would need to be provided by them. Some IVD manufacturers may be currently working to establish traceability using direct comparisons to a reference laboratory that offers IDMS methods (see the Joint Committee for Traceability in Laboratory Medicine web site http://www.bipm.fr/en/committees/jc/jctlm/jctlm-db/). A new NIST reference material for serum creatinine is expected to be available in mid 2006 that will enable most IVD manufacturers to initiate standardization of their measurement systems. The transition to standardized creatinine methods is expected to take 1-2 years to be completed because new product labeling will be required, and inventories currently in the field will need to be used and replaced with the revised calibration systems.
What is your opinion of the Cystatin C test?
W. Gregory Miller, PhD, DABCC, FACB: Cystatin C is a promising new test for renal function. It is not widely available, is more expensive than serum creatinine, and has method specific reference intervals. It is likely to be useful as a follow-up test for kidney disease and for use when a patient's basal creatinine production is expected to be very abnormal, such as patients of extreme body size or muscle mass (e.g., morbidly obese, severely malnourished, amputees, paraplegics or other muscle-wasting diseases) or with unusual dietary intake (e.g., vegetarian, creatine supplements). It is also likely to be useful for pediatric patients. Equations have been developed to estimate GFR from serum cystatin C (e.g. Grubb et al. Clin Chem 2005;51:1420-1431).
Have you done any validations or gathered information on what affect pregnancy has on an estGFR? Do the same guidelines apply? Thank you.
W. Gregory Miller, PhD, DABCC, FACB: Serum creatinine is moderately lowered in a person with a normal pregnancy. Consequently, it is not recommended to use the MDRD equation for pregnant women. Note that serum creatinine remains a useful indicator for impaired renal function as a complication of pregnancy (e.g. pre-eclampsia, eclampsia).
Most routine labs perform creatinine determination by compensated or non-compensated Jaffe reaction. The relevant calibrators bear a lower value for the uncompensated reaction and a higher one for the compensated reaction, with different settings. Roche uses nowdays the compensated value only. Olympus declares traceability of the calibrator to NIST SRM 909b material, but uses a lower value for the uncompensated and a higher value for the compensated reaction, with a B-factor (-0.2 mg/dL)in the settings. Do you consider the compensated Jaffe reaction as traceable to the IDMS reference method procedure? How is it possible for a calibrator to have different values, both declared traceable to the same SRM?
W. Gregory Miller, PhD, DABCC, FACB: The terminology “compensated” and “uncompensated” is not clearly defined and is used in different contexts by different IVD manufacturers. The product calibrators used by a given IVD manufacturer for a given measurement procedure should be used as instructed only for that specific measurement system. The NKDEP Laboratory Working Group is developing a creatinine standardization program that will provide tools and recommendations to achieve traceability for patient results to an IDMS reference measurement procedure. Briefly, there are two strategies that can be used. One is to use clinical samples that have been measured by a routine method and a IDMS reference measurement procedure as the basis for traceability. The second is to use a commutable serum reference material that has been validated to be commutable with clinical samples and which has been value assigned using a IDMS reference measurement procedure. The Joint Committee for Traceability in Laboratory Medicine web site (http://www.bipm.fr/en/committees/jc/jctlm/jctlm-db/) lists three reference measurement procedure laboratories; but does not list any commutable serum reference materials. A new NIST serum reference material for creatinine (SRM 967) is expected to be available in mid 2006 that will enable most IVD manufacturers to initiate standardization of their measurement systems. The NIST SRM is currently undergoing value assignment and commutability validation for many routine methods.
Clearly systematic standardization of calibration is a laudable goal. However, given the known and substantial interference potential of Jaffe methods on individual patients, shouldn't there be a major effort to require more specific creatinine methods as well?
Chapel Hill, NC
W. Gregory Miller, PhD, DABCC, FACB: The creatinine standardization program is intended to reduce the average bias to low enough levels that, in combination with an acceptably low imprecision, the total error of a serum creatinine value will be adequate for calculating eGFR. The measurement limitations of method non-specificity will remain, and can contribute to an inappropriate eGFR when a patient has an interfering substance present. IVD manufacturers should review the specificity of their creatinine methods and make improvements if necessary. Clinical laboratories should consider the specificity when selecting methods to use.
Has the laboratory working group addressed the problem of appropriate reference ranges to use in conjunction with the estimated GFR in different age/sex/race populations?
W. Gregory Miller, PhD, DABCC, FACB: The eGFR is used to provide a more appropriate interpretation of the serum creatinine than is possible with a traditional reference interval. The equations takes into account the age/sex/race of the patient. The MDRD equation result is normalized to a 1.73 m2 body surface area to allow interpretation of results for a large percentage of patients. Separate reference ranges for the eGFR are not necessary.
Our lab has not yet implemented the GFR calculation, but are very close to going live with it. You mentioned that the eGFR was not as reliable in inpatients. We were planning to report the eGFR with all creatinines. Do you know have an idea what the general practice is for labs, whether they report only on outpatients and specific orders on inpatients? It becomes difficult sometimes to make this distinction in lab computer systems.
W. Gregory Miller, PhD, DABCC, FACB: There is no aggregate information available on how different organizations implement eGFR reporting. There are clinically legitimate reasons to report eGFR only on selected patient populations. The lab director will need to make those judgments in consultation with clinical providers. The NKDEP recommends reporting eGFR with all serum creatinine results because the primary goal is to identify persons with chronic kidney disease, it can be difficult for some computer systems to determine under what conditions the result should be reported, and clinical providers are generally adept at identifying clinical situations when an eGFR is not reliably calculated from creatinine.
In an earlier answer, you stated that that eGFR should be reported on everyone older than 18 years, but later on not to report for those with normal or mild impaired kidney function. Please clarify.
W. Gregory Miller, PhD, DABCC, FACB: If the calculated eGFR is 61 mL/min/1.73 m2 or greater; the laboratory should report as >60 mL/min/1.73 m2.
Is there early data on the applicability of the MDRD equation to other ethnic groups (Chinese, Asaian Americans etc).
Iowa City, IA
W. Gregory Miller, PhD, DABCC, FACB: no
We have a Bayer Advia 1650 which had updated its creatinine assay. Which GFR calculation should be used? Also, if creatinine is reported in mg/dl, what calculation should be used? And a separate question: do you know if Bayer systems have the capability of automatically calculating GFR or if they plan to update their software so that this will be done?
W. Gregory Miller, PhD, DABCC, FACB: If a method IS NOT calibrated to provide results traceable to an IDMS reference method, the original conventional calibration MDRD equation should be used. If a method IS calibrated to provide results traceable to an IDMS reference method, the IDMS-traceable MDRD equation should be used. The IVD manufacturer will need to be contacted regarding the calibration traceability of their method and the calculating capability of their system. Keep in mind that not all LIS interfaces can send the necessary sex/age/race information to an instrument.
What is today the most practical way for a laboratory to check their creatinine results to confirm the presence or absence of a bias against IDMS values?
W. Gregory Miller, PhD, DABCC, FACB: Contact the IVD manufacturer of the measurement system and ask for the claim regarding traceability.
Have you encountered any problems with physician response since reporting the EGFR to doctors who only requested serum creatinine?
Trenton, New Jersey
W. Gregory Miller, PhD, DABCC, FACB: My laboratory has been reporting eGFR for a year according to the recommendations of the NKDEP. The service was requested by the chairs of nephrology and cardiology. The clinical staff has been pleased with the information.
It seems that use of eGFR for identification and monitoring of renal disease has become a standard of care. Does the traditional creat. clearance calc. with 24h urine still have any role to play? Are labs justified in no longer offering this assay? Are there specific situations in which use of the traditional CC would be preferred?
W. Gregory Miller, PhD, DABCC, FACB: For patients without compromised creatinine production, the eGFR is more accurate that a measured creatinine clearance. However, a measured creatinine clearance should continue to be offered. For adults, a measured creatinine clearance is appropriate when the patient's basal creatinine production is expected to be very abnormal, such as patients of extreme body size or muscle mass (e.g., morbidly obese, severely malnourished, amputees, paraplegics or other muscle-wasting diseases) or with unusual dietary intake (e.g., vegetarian, creatine supplements). A measured creatinine clearance is also recommended for pediatric patients when drug dose adjustments are necessary.
You stated that the NKDEP recommendation is to report >60 mL/min/1.73m2 when the calculated value is >60.Could calculated values above 60 be used to monitor reduced kidney function over time? Wouldn't this be better to diagnose a problem sooner?
Trenton, New Jersey
W. Gregory Miller, PhD, DABCC, FACB: It is recommended to NOT report eGFR values >60 mL/min/1.73 m2 because the combination of bias and imprecision for the typically low creatinine values in this range produce an excessively large confidence interval that compromises the reliability of the eGFR. It is hoped that the usefulness of results in the range from 60-90 mL/min/1.73 m2 can be improved by the creatinine standardization program.
The factor 1.73 is accepted as "average" body surface area; I've seen this value for my entire 13 yr career. Is this still an accurate "average" considering the ever increasing percentage of obesity in industrialized nations?
W. Gregory Miller, PhD, DABCC, FACB: The value 1.73 m2 for body surface area provides a basis for normalization of results to allow comparison to common interpretation criteria. It does not actually represent the average body surface area of an adult.
We currently do not report eGFR, and are reluctant to begin this practice, however, we have a new nephrology group that is demanding it to be done. We also offer Cystatin C testing in house. What is your suggestion for using these two tools so that the most accurate diagnosis/treatment can be achieved?
W. Gregory Miller, PhD, DABCC, FACB: The main goal of the NKDEP is to identify people with chronic kidney disease so they can be put into a treatment program to slow or prevent progression of the disease. Because serum creatinine is a commonly ordered test, it is useful to identify patients with kidney disease. Patients with eGFR approximately equal to 60 mL/min/1.73 m2 have serum creatinine values between 1.0-1.5 mg/dL depending on their age/sex/race. Many clinicians have difficulty interpreting creatinine values in the near normal range as consistent with significant kidney disease. The eGFR provides a very useful interpretation of the serum creatinine and allows confident identification of patients with chronic renal disease. Cystatin C, and urinary albumin, are good tests to use for follow up and confirmation of a low eGFR.