AACC Annual Meeting – Chicago, IL – July 22, 2009
National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)
National Kidney Disease Education Program (NKDEP) Laboratory Working Group Meeting
Participants: Greg Miller, Edward Ashwood, Lorin Bachmann, David Bunk, Christa Colbbaert, John Eckfeldt, James Fleming, Matthew Gnezda, Neil Greenberg, Glen Hortin, Yoshihisa Itoh, Chandra Jain, Graham Jones, Anthony Killeen, David Lacher, John Lieske, Leigh Ann Milburn, Andrew Narva, Eileen Newman, Mauro Panteghini, Max Robinowitz, Mary Robinson, Heinz Schimmel,George Schwartz,Lesley Stevens,Reba Wright, Ingrid Zegers
Summary of Action Items:
- Greg Miller will facilitate obtaining the urine albumin reference material being developed by Yoshi Itoh for use in the Urine Albumin Method Harmonization Study being lead by Lori Bachmann.
- Tony Killeen, John Lieske, and Glen Hortin volunteered to work with David Bunk to develop specifications for a proposal for the National Institute of Standards and Technology reference material for urine creatinine and albumin. Once the project proposal is developed it will be circulated to the LWG by email for comments and suggestions.
- Lesley Stevens agreed to prepare a first draft of information about the new CKD-EPI equation for circulation to the LWG in the next couple of months. The information will eventually be posted on the NKDEP website.
- Several investigations to obtain information needed to clarify serum creatinine methods' specificity and urine albumin measurement issues will continue under the leadership of the respective task force chairs.
Part 1: Serum/plasma/blood creatinine, eGFR, and cystatin C
1. CKD-EPI Equation for eGFR, Lesley Stevens, Tufts University
Lesley Stevens provided a review of a new eGFR equation that developed in the CKD-EPI Study and was published in the Annals of Internal Medicine in May, 2009.
Background: This equation was developed from the NIH-funded CKD-EPI study and is intended for use in general clinical practice. The MDRD study had 1,628 participants with predominantly non-diabetic CKD and included age, sex, and race as surrogates for non-GFR determinants. The equation has reasonable accuracy in the CKD population, but there is a systematic underestimation of measured GFR at higher levels and there is imprecision throughout the GFR range. Estimates > 60 mL/min/1.73m2 are not reported and there is a potential for false positive diagnosis of CKD in patients with measured GFR < 60 mL/min/1.73m2.
CKD-EPI Study: The goal of the CKD-EPI study was to address deficiencies of MDRD equation. CKD-EPI used data that were already collected in clinical studies and includes individuals with and without kidney disease in a wide range of measured GFR. Study populations were used if they had more than 250 participants, measured GFR, quality control data, and ability to calibrate creatinine to IDMS traceability. There were 10 studies, 2/3 for development of the equation and 1/3 for internal validation, and a separate set of 16 studies was used for external validation. Multiple models using new transformations of the original 4 variables included in the MDRD equation and additional variables (diabetes, transplant, weight) were included in the analyses during development. The equation was validated in 16 studies using a pre-specified algorithm. It should be noted that all data sets included a reasonable proportion of participants without kidney disease as shown by the mean GFR, which was 68 mL/min/1.73m2 versus 40 mL/min/1.73m2 in the MDRD Study. The final equation has the same 4 variables as the MDRD equation (creatinine, age, race, and sex), but they are used differently. Specifically, the creatinine variable is the spline of the log of creatinine with sex-specific knots versus the log in the MDRD equation, and the age variable is linear versus log.
Results: The CKD-EPI equation had a substantial reduction in bias measured and estimated GFR as compared to the MDRD equation. When these equations are applied to the NHANES Study population, the CKD-EPI equation gave a higher median eGFR and classified about 20% of the population at a higher stage of renal failure and 0.6% at a lower stage than the MDRD equation. In summary, the new equation will have less false positives in the 30-59 range, be able to detect mild decline in the 60-89 range, and be able to determine eGFR in the normal range >90. From the laboratory point of view, the equation is similar to the MDRD and could be implemented easily. The implication from the clinician point of view is the value of being able to report a more accurate value in the range near 60.
Next steps: Implementation will take some time. The renal community will need to use online calculators to evaluate the new equation for prognosis and improved care. The equation will need to be pilot-tested in a few centers. NKDEP could have a task force to develop preliminary education materials.
- A question was asked about the effect of lean body mass on the equation, and Lesley replied that results from studies that looked at lean body mass are variable and this is likely because it is difficult to measure lean body mass. It is important to provide guidance regarding when the equation will and will not work well.
- There was discussion about how this would/could be adopted by NKDEP. Andy Narva stated that the consensus at NIDDK and CDC is to allow a period of validation and peer-review as was done with the MDRD. This discussion is part of the process, and NKDEP will move forward when the time is right. There are only so many educational opportunities with NKDEP's target audience, and the Program is still trying to promote the concept of an estimated GFR, standardizing creatinine measurement, and using urine albumin for identifying kidney failure.
- Tony Killeen noted that adoption of this equation is predicated on labs using IDMS-traceable creatinine methods. Neil Greenberg commented that the conversion to IDMS-traceable creatinine methods and reporting estimated GFR has been a slow process. There are still calibration issues even with assays claiming to use IDMS-traceable calibration. (Note: Greg Miller later shared during the Manufacturers Forum that all major IVD manufacturers will have completed conversion to IDMS-traceable calibration by the end of 2009.)
- Neil Greenberg commented that imprecision of some creatinine methods at values corresponding to normal eGFRs may be a limiting factor.
- Graham Jones commented that he would like to use this equation right away. He has tested it with his own population and it works well, but implementing this in the U.S. would help implementation in Australia and other countries. It would be good to establish a timeline. Lesley does not feel that further validation is really required and offered to help with implementation in Australia.
- David Lacher feels that based on his experience with trying to roll out new lab tests, one needs to be proactive with the main users (e. g., nephrologists). If this equation is dramatically better, then we should adopt as soon as we can.
- Jim Fleming agrees that we should adopt the new equation as soon as possible, because it is better validated than the MDRD equation was. It solves several problems: 1) there is pressure to provide values >60; 2) CMS requires reporting of ICD-9 codes for CKD stages 1 and 2, which could be differentiated using this new equation.
- Neil Greenberg cautions that even when all labs are using IDMS-traceable methods, there needs to be communication about the limitations of estimated GFR. Andy Narva agreed that it is necessary to educate providers about the limitations of GFR estimation.
- John Eckfeldt asked if ethnic groups other than Caucasians and blacks were included in the CKD-EPI Study. Lesley Stevens commented that Asians were included but that there was not enough data. New data are needed to address this issue.
2. Serum Creatinine Specificity Study—Progress Report, Neil Greenberg, Ortho-Clinical Diagnostics, Inc.
Background: Calibration traceability to IDMS reference methodology does not change the limitations of creatinine results caused by the influence of interfering substances. The NKDEP LWG and IFCC WG-GFRA plan to develop recommendations on specificity requirements and performance characteristics for creatinine measurement procedures. However, there is very little comparative data looking at current creatinine methods, and there are still unresolved concerns with various drugs, endogenous substances, ketoacidosis, hyerbilirubinemia, hemoglobin, protein, and albumin. The protein effect seen especially with Jaffe procedures is an example of this; low protein/albumin in pediatric and hospitalized patients results in too large of a correction and thus an under-estimation of kidney damage.
NKDEP/IFCC Creatinine Specificity Study: A study was designed to evaluate method specificity using unaltered patient samples from several pathological groups (e. g., diabetes, CKD, liver disease, high and low protein syndromes). Also, there were supplemental samples spiked for certain volatile interferents (e. g., ascorbate, acetoacetate). These samples were assayed by eight different commercial methods (including four enzymatic and four Jaffe) and were performed by the manufacturer in their own facilities. An ID-LC MS/MS reference method developed by Neil Dalton's laboratory at Evilina Childrens Hospital in London was used to determine "truth."
Results: Preliminary data show that the reference method obtains results very close to the certificate values for SRM 967 material, calibration bias may still be significant with some routine methods, and significant bilirubin (4+ icteric samples) interference occurs with both (some) Jaffe and (some) enzymatic methods. Data analyses are still in progress and a final report is anticipated to be available in the next six to nine months.
- George Schwartz asked if hemolyzed specimens were used in the study; the answer is "yes."
- Mauro Panteghini commented that calibration bias seen at the ends of the calibration range is still problematic and asked if the methods used in the study were checked against the list of methods used in the commutability study performed by John Eckfeldt. Neil Greenberg replied that the commutability study did not evaluate calibration bias, but rather whether or not the SRM 967 material was representative of the performance of patient samples.
- David Lacher commented that the statisticians need to correct for co-interferences in the samples. Neil responded that they may not be able to do that due to minimal sample volumes that did not allow measurement of most other analytes in the samples.
- Graham Jones commented that any recommendations need to consider that some methods, especially Jaffe methods, do not have good precision at low levels.
- Christa Colbbaert asked if there will be enough samples with low protein concentration to evaluate this effect on the results. Neil replied that they did get samples with low protein levels, but they have not examined that data yet.
3. CKD and Drug Dosing Educational Statement for Providers, Lesley Stevens, Chair, NKDEP Committee to Develop Educational Advisory on Drug Dosing in CKD
Lesley Stevens presented information from a draft document for recommendations on drug dosing. (A draft of this document was provided prior to the meeting.) The draft discusses the following 6 categories: estimation of kidney function for medication dosage prescriptions in adults, NKDEP's recommended approach to drug dosing, impact of IDMS-standardized creatinine values, the MDRD Study equation, the Cockcroft-Gault equation, and limitations of any serum creatinine-based estimates. The basic question is: Can the eGFR based on standardized creatinine be used to determine drug dosing? To date, pharmacokinetic (PK) study results and FDA labels are dependent on the specific serum creatinine method used resulting in inconsistent translation of the labels to practice. In the future, because of standardization of creatinine methods, the PK study results and FDA labels can be independent of the serum creatinine methods. However, there is a problem with the vast number of drugs that have already been approved. It is possible that the FDA or drug manufacturers will choose to re-label based on studies performed using standardized creatinine methods, but not all drugs can be re-labeled. We need data on the impact of creatinine calibration for each estimating equation, simulations to assess differences among equations in recommended drug dosages, and simulations to assess differences in prediction of toxicities and outcomes.
The CKD-EPI Study "Development" data set was used to determine impact of different GFR estimating equations on drug dosage by doing a simulation on drug dosing recommendations for 15 medications. The study concluded that the MDRD Study equation had the highest rate of concordance with measured GFR, and for specific drug dosing, concordance rates among equations was high, with lower concordance for drugs with a greater number of dosing levels. (The results of this study have been published: Stevens et al. Am J Kidney Dis. 2009; 54(1):33-42.)
The NKDEP recommendations for drugs that have been proposed in the draft are:
- Utilize eGFR or eCrCl for dosing drugs
- If using eGFR in very large or very small patients, multiply the reported eGFR by the body surface area in order to obtain eGFR in units of mL/min
- Perform confirmatory tests (e. g., measured CrCl or measured GFR using exogenous filtration markers)
- With drugs with a narrow therapeutic index
- In individuals in whom eGFR and eCrCl provide different estimates of kidney function
- In individuals where any estimates based on creatinine are likely to be inaccurate
- Greg Miller commented that the intention is to replace the recommendations currently on the website, so everyone should be certain that their concerns are addressed and to provide supportive or contradictory comments. Lesley agreed to take comments by e-mail for discussion or incorporation into the recommendations.
- Graham Jones asked if we should provide guidance regarding what constitutes very large and very small body sizes in males and females. Lesley stated that it is difficult to provide guidance when there is none available in the literature. While this is not optimum, we provide general opinions for clinicians who have the ultimate responsibility.
- John Eckfeldt suggested a minor change on page four to change "Cockcroft-Gault equation cannot be re-expressed for IDMS-traceable creatinine values" to "... cannot be accurately re-expressed..." because it may be possible within an institution, since they usually know the impact by validating new methods.
4. Update: Cystatin C Standardization, Greg Miller for Anders Grubb (Chair, IFCC WG-SCC)
University Hospital, Lund, Sweden
Greg Miller reported for Anders Grubb, who was not able to attend. The project has successfully prepared a secondary reference material, which is pooled human serum to which a recombinant cystatin product has been added. The value assignment is complete, and they are beginning the commutability validation process. All methods used to evaluate the material gave results that are in close agreement. It is estimated that the product will be available in 2010. This group is planning a large study to develop a more universal equation once the standardization is in place.
5. Performance of CKiD Pediatric Equation with Different Cystatin C Methods, George Schwartz
University of Rochester Medical Center
Background: The original Schwartz equation used height/serum creatinine to estimate GFR. Blood urea nitrogen (BUN) tends to vary inversely with GFR and urea clearance represents approximately one to two thirds of the GFR. Cystatin C is a new marker and is inversely proportional to GFR. The purpose of this new study is to use these variables to develop a new, more precise and accurate equation for estimating GFR in children.
Study: There were 349 pediatric CKD patients; generally developmentally immature, mostly male, and stunted which is a stigmata of children with kidney disease. GFR was measured by iohexol disappearance. (The results of this study have been published: Schwartz et al, J Am Soc Nephrol. 2009; 20:629-637.) The cystatin variable in the equation was investigated further by comparing the Dade Behring PENIA (nephelometric) method to the DAKO PETIA (turbidometric) method.
Results: Univariate correlations are higher when cystatin C is measured by Dade-Behring method. Cystatin C and height/serum creatinine describe the entire variability and BUN used in the original publication is not important with a change in cystatin C method. A new equation is being developed.
Issues for NKDEP-LWG: Standardization of creatinine at low concentrations is critical for pediatric needs. Standardization of cystatin C methodology is required for generalized application of GFR estimating formulas; and current new pediatric estimating equations are valid only in the range of iohexol GFR (15-80 mL/min per 1.73 m2).
- Max Robinowitz requested a breakdown of the pediatric group kidney disease: approximately 20% have glomerular nephritis, and 80% have urological disease.
- John Eckfeldt asked for an expansion on use of urine versus serum for calibration of cystatin C. Dade Behring uses cystatin C derived from urine. Ingrid Zegers commented that the IFCC material is a recombinant material. John commented that there is an issue with the Dade-Behring calibration which has drifted down about 15% over the past decade with the same calibration scheme. He suspects it is related to the material used for calibration. The internal control provided in the kit is made with the same material, so one does not see the drift with those controls. Also, a change in Dade-Behring instrumentation has an effect on the calibration. So, the equation may vary depending on when the assay of cystatin C was performed.
- George Schwartz commented that there is concern about synthetic material; if it is not glycosylated it may be recognized differently by the antibodies used in the assay. Ingrid Zegers stated that the IFCC-developed material behaves similarly to serum, and that if a problem exists it may show up in the commutability studies about to commence.
- Yoshi Itoh commented that cystatin C protein is not stable in urine and is easily polymerized when stored for a long time, resulting in altered immunoreactivity. George stated that cystatin C is a small protein that is completely reabsorbed by the proximal tubule, so it is seen in nephritic syndrome when the proximal tubule is overwhelmed or not functioning properly. Yoshi Itoh added that it is easily degraded.
- Glen Hortin asked if there were patients in the study group who were on steroids and if this could affect the cystatin C production. George replied that there were only a small number of patients treated with steroids; this was not investigated as the explanation for outliers because the problem was vaster than could be explained by the small number taking steroids.
Part 2: Urine albumin and creatinine (Refer to written summaries sent with agenda)
1. Urine Albumin Biological Variability and Albumin Adsorption Study, Mary Robinson, Centers for Disease Control and Prevention
Study Design: Phase 1 is the container study; Phase 2 is the biologic/pathologic variability study.
Containers were made of the following materials: polypropylene (PP), polystyrene (PS), polyethylene (PE), polyethylene terephthalate (PET), and PL; and sample cups were made of PE and PS; circular discs punched from containers and cups were used; the discs were spiked with 125I-alb loaded on discs and rinsed; the raw data presented have not been statistically analyzed; all sets of containers showed the same pattern for pH and albumin concentration; PE and PE-terephthalate were lowest albumin binders.
Summary: All six materials have greater adsorption at higher albumin concentrations and longer exposure; adsorption at lower pH is greater for most materials; addition of Triton X and Tween significantly reduced the binding; adsorption in sample cups was essentially the same for PE and PS; and addition of Triton X and Tween reduced the adsorption.
- Max Robinowitz asked whether or not an increase in the concentration of albumin in the urine would be observed if one adjusted a urine from pH 4 to pH 8. Mary responded that they are using purified serum albumin and not urine, so she does not know the answer.
- Mary showed an additional slide that indicates the addition of detergent-reduced adsorption by almost 50% in the high albumin concentration samples, but not in the low albumin samples. Glen Hortin stated that based on a small study, some detergents have matrix effect in the immunoassay and may effect how aggregates form.
- Yoshi Itoh added that they have confirmed the same things and according to preliminary data, almost no albumin is adsorbed when using hydrophilic materials.
- Mary observed that the problem is most likely with sample cups and not the collection containers. Yoshi agrees with this observation.
- Greg Miller commented that more thought is needed regarding how to address these issues before the biological variability study is carried out.
- Heinz Schimmel commented there have been differences in the adsorption of stored DNA when using polypropylene vials from different manufacturers; also there is variation observed between polished and unpolished surfaces. This may or may not apply to albumin.
2. Urine Albumin Method Harmonization Study, Lori Bachmann, Virginia Commonwealth University
There was not a presentation about this study, but comments on the basic approach given in the handout would be appreciated. The project is expected to start in January 2010.
3. Urine Albumin Molecular Forms Study, Glen Hortin, University of Florida
A summary of the proposal is found in the outline. Comments are appreciated.
4. Urine Albumin Reference Material (JSCC), Yoshi Itoh, Asahikawa Medical College, Hokkaido, Japan
- Greg Miller asked if there is a timeline estimate for submission of the material to JCTLM. Yoshi replied that they first need to submit the paper to an international journal for publication and take several other steps. David Bunk commented that the next cycle for JCTLM nominations should end approximately May 2010 and then it would have to wait for another year. Greg commented that the material will likely not be available before 2011 due to these constraints.
- Glen Hortin asked if there is evidence that urinary albumin is de-lipidated or has normal fatty acid content. Yoshi responded that de-lipidated may be an option or use of re-combinant albumin may be better. These are important considerations and should be studied in the future.
- Mauro Panteghini commented that it would be good to include this material in the urine harmonization project; Greg agrees and will get in touch with Yoshi to add the Japanese candidate reference material to the study.
- David Bunk commented that this material will be useful as the reference material to use in the reference procedure being developed at Mayo; he is not concerned about commutability of the material for use as an IDMS calibrator and is more concerned about the value assignment. He will talk with Yoshi about this off-line.
5. Reference Material for Urine Creatinine and Urine Albumin, David Bunk, National Institute of Standards and Technology
- David Bunk commented that this project is trying to develop specifications for both urine albumin and creatinine reference materials. Advice is needed about the concentrations and collection protocols in order to move forward; it may be possible to augment creatinine levels by spiking and thus have a single material for both albumin and creatinine.
- Greg Miller commented that he thought the goal of the Japanese project is to develop a material that is for use by manufacturers so that commutability is an important characteristic. Yoshi Itoh commented that initially this was begun for standardization of immunoassay systems. Greg added that it would be good for Yoshi and David to collaborate on these two projects because we want to avoid two different reference materials coming out of these initiatives.
- A subgroup of volunteers was assigned to work with David Bunk on this project: Tony Killeen, John Lieske, and Glen Hortin. Once the project proposal is developed it will be circulated to the LWG by email for comments and suggestions.
6. Urine Albumin IDMS Candidate Reference Measurement Procedure, John Lieske, Mayo Clinic
John Lieske presented data on the validation of an LC-MS/MS procedure for quantification of albumin in a urine matrix. A 15N-labeled human serum albumin internal standard was developed and two separate peptides compared very well. Comparison of the LC-MS/MS to three clinical albumin assays showed good but not perfect agreement and demonstrated that there is variability between assays. The proposal for this project has 2 aims: Aim 1 is to validate the LC-MS/MS procedure as a reference method procedure to quantify intact albumin in urine; and Aim 2 is to compare clinical sensitivity and specificity of LC-MS/MS assay performed at Mayo and NIST, as well as with other immunoassays, using a population of patients with or without overt diabetic nephropathy. The sample requirement for this procedure is about 50 µL.
John Eckfeldt asked which of the two peptides was used for the comparison with the clinical assays because this may have an effect on some of the outliers. He speculated that the ratio of two different peptides may be different for the outliers and explain some of the lack of correlation. John Lieske replied that they suspect that some samples may have different fragments and this may be a reason to look for other fragments to include in the assay.
Part 3: Summary and Next Steps
- Max Robinowitz asked if there will be a recommendation for urine albumin to be assayed on a random urine and if creatinine should be assayed with urine albumin. Greg Miller commented that the paper from the NKDEP/IFCC conference published earlier this year stated that the first morning urine correlates best with 24 hour, random urine is not as good, and urine creatinine should be measured and the albumin / creatinine ratio (ACR) reported with all urine albumin measurements.
- Neil Greenberg expressed concern that many labs have not converted to IDMS-calibrated methods—that some manufacturers are phasing out the older non-standardized methods. He asked whether or not we need to reach out to labs to encourage adoption. Greg stated that he will show data from a survey of manufacturers (during the forum), which shows by the end of 2009: they will not sell anything that is not traceable to IDMS with two exceptions: Nova whole blood methods and Siemens Dimension/Vista Jaffe methods. Thus, in 2010 all labs will be using IDMS-traceable methods, and there is no need for NKDEP to address this further with labs.
- Mauro Panteghini asked about use of non-standardized methods used in other countries. Neil commented that they see isolated pockets of resistance that have no interest in IDMS traceability, but manufacturers still plan to discontinue non-IDMS methods. Graham Jones commented that the IFCC has set up a group to access the worldwide acceptance and provide assistance to areas that are resistant. Greg reviewed the survey of manufacturers, and confirmed that manufacturers plan to sell only IDMS-traceable creatinine methods in all markets.
- Graham Jones asked of the focus is on urine albumin or includes urine protein. Andy Narva commented that NKDEP is working on urine albumin exclusively. George Schwartz showed data suggesting that a significant number of children with kidney disease would be missed using urine protein/creatinine values but that they would be indentified using urine ACR.
- Glen Hortin commented that the outcome of an FDA-sponsored conference last year was they are closer to accepting urinary protein than urine albumin as a surrogate marker.
Summary of actions agreed upon during meeting:
- Greg Miller suggested that it is appropriate to develop some information about the new CKD-EPI equation. Lesley Stevens agreed to prepare a first draft to circulate to the LWG in the next couple of months.
- JSCC secondary reference material for urine albumin should be coordinated into the harmonization study.
- A Task Force has been established for developing parameters for the NIST calibrator for urine creatinine and albumin.
- Current information-gathering projects will continue.