Meeting Minutes – Manufacturers' Forum
July 28, 2010

National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)
National Kidney Disease Education Program (NKDEP) Manufacturers Forum
AACC Annual Meeting — Anaheim, CA

The International Federation of Clinical Chemistry and Laboratory Medicine's Working Group (WG) -Glomerular Filtration Rate Assessment (GFRA Chair, Neil Greenberg) and WG-Standardization of Albumin (SAU) in Urine (SAU Chair, Greg Miller) met in conjunction with the NKDEP Laboratory Working Group (LWG) as the membership and goals overlap, and many of the scientific activities are conducted as joint efforts.

Participants: Greg Miller, Lorin Bachmann, Diana Blanco, David Bruns, David Bunk, Glenn Carlisle, Joris Delanghe, John Eckfeldt, James Fleming, Mary Lou Gantzer, Paul Glavina, Neil Greenberg, Anders Grubb, Desire Hiraman, Glen Hortin, Yoshihisa Itoh, Chandra Jain, Graham Jones, Harvey Kaufman, Mary Kimberly, Anthony Kileen, David Koch, Hans-Joachim Kytzia, Katri Lahtinen, Yemi Lemma, Jack Levine, John Lieske, Andrei Malic, Gary Myers, Andrew Narva, Eileen Newman, Prasad Pamidi, Mauro Panteghini, Linda Parisi, Karen Phinney, James Ritchie, Max Robinowitz, Heinz Schimmel, David Seccombe, Jill Tate, Linda Thienpont, Dave Torrens, Clarke Xu, Dror Yahalom, Jack Zakowski, Ingrid Zegers

Meeting Minutes

1. Last year the group discussed an educational statement on drug dosing in the post-creatinine standardization environment. This provider resource has been posted on the NKDEP website.

2. At this time, all laboratories in the U.S. and other countries using methods from the major global manufacturers should be using creatinine methods calibrated to be traceable to isotope dilution mass spectrometry (IDMS). NKDEP will be removing estimated glomerular filtration rate (eGFR) equations for non-standardized creatinine methods from the website calculators (adult and pediatric). Equations on the website will be those appropriate for IDMS-traceable creatinine methods.

3. Informational Web Posting on CKD-EPI Equation, Andrew Narva (Director, NKDEP), NIDDK

An informational paragraph describing the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation preceded by this statement will be posted on the website: "The CKD-EPI equation is a new equation published in 2009 to estimate GFR from serum creatinine, age, sex, and race for adults age greater than 18 years. NKDEP has not made a recommendation on general implementation of this equation. The equation is still being validated and while offering some improvement for eGFR between 60 and 120 mL/min/1.73m², it is not clear that implementing CKD-EPI in place of the Modification of Diet in Renal Disease (MDRD) Study equation would alter clinical detection or management of patients with CKD."

4. Creatinine Specificity Project: Report and Recommendations, Neil Greenberg (Chair, IFCC WG-GFRA), Ortho-Clinical Diagnostics, Inc.

The object of the study is to evaluate selected groups of patient samples from known pathological groupings that may be targets for performing eGFR. Blood was collected from approximately 20 donors in 19 patient panels and 20 healthy control subjects; some normal serum was spiked with four levels of acetone, pyruvate, acetoacetate, and ascorbate. Four enzymatic and three Jaffe methods were evaluated and compared to a LC-MS/MS assay run in the United Kingdom. The methods evaluated were enzymatic and Jaffe methods from Siemens Dimension, Beckman-Coulter, and Roche, as well as enzymatic from Ortho Diagnostics. Results were summarized in a table with beige shaded boxes if the bias was +/- 3-5 samples and lavender if greater than 6 samples showed bias. The selection criteria for the patient panels were used to assess likely candidate interferences and patients for whom creatinine assays are used. It is important to note that we do not know the basis for a given bias, e.g., HbA1c probably is not the actual reason for bias in that group and it may be something else causing the bias. The clinical groups are intended to provide an overview and show a good picture of the common problems with representative populations. A key finding is that in the diabetic samples there were more problems with Jaffe than enzymatic methods. Another important observation is that there are no overall conclusions that can be made about any given type of technology (enzymatic vs. Jaffe) and within a technology the implementation details affect the influence of interferences. There were some surprising findings, e.g., a Jaffe method with no interference due to elevated bilirubin and enzymatic methods with interference in the elevated bilrubin group of samples. Future plans are to publish the findings and develop some general recommendations for laboratories.

5. Cystatin C Standardization, Anders Grubb (Chair, IFCC WG-SCC), Lund University, Sweden

It took six years to produce the international reference material traceable to SI units. The reference material is recombinant cystatin C that is identical to the human protein added to a serum pool from blood donors, and then lyophilized. The certified value and uncertainty is 5.48 +/-0.15 mg/L. A preliminary commutability between the reference material and patient samples for Siemens, Gentian, and Sentinel (Abbott) methods is good, showing the same dose-response curve for dilutions of the reference material and patient samples. The Roche method shows commutability for values below 3 mg/L, but not for higher values; and Roche has already started to improve their method to achieve full commutability. An ongoing commutability evaluation with 14 manufacturers and methods shows 11 to be commutable between the reference material and patient samples, and three which need additional development. The calibrator is available from IRMM in Belgium (http://irmm.jrc.ec.europa.eu/html/reference_materials_catalogue/index.htm) and the following publication describes a method for calibrating assays: Blirup-Jensen S, Johnson AM, Larsen M. Protein standardization V: Value transfer. A practical protocol for the assignment of serum protein values from a reference material to a target material. Clin Chem Lab Med 2008;46:1470-9.

In answer to a question about dilution, Anders Grubb replied that the dilution buffer was generally saline, but some manufacturers use their own priority solutions.

6. Urine Albumin Method Harmonization Study, Lori Bachmann, Virginia Commonwealth University

Project objective: Use native human urine to evaluate the agreement among manufacturers' albumin methods and also compare against the isotope dilution LC-MS/MS candidate reference procedure being developed at Mayo by John Lieske's group. These urine samples will also be used to compare analytical harmonization of total protein and creatinine methods. Another objective is to evaluate the commutability of the Japan Society of Clinical Chemistry candidate reference material developed by Yoshi Itoh's group in Japan. The study includes a comprehensive representation of manufacturers (N=5) and platforms (N=11). Participants who would like to be included were encouraged to contact Lori Bachman or NKDEP.

Study Design: Collect 340 fresh, residual native samples submitted for routine urine albumin measurement and represent a wide variety of disease pathologies. Samples will be collected over eight to ten weeks by three sites. Fresh aliquots of each sample will be shipped overnight to each manufacturer and frozen aliquots will be sent for LC-MS/MS. Shipments will include spiked matrix samples (salts, reducing substances, and drugs) to evaluate the influence of common matrix components on the methods.

Analysis: Imprecision will be evaluated by ANOVA, the bias for each method will be calculated versus LC-MS/MS method, and unusual or discrepant results will be evaluated for analyte-specific matrix influences. An aliquot will also be sent to the National Institute of Standards and Technology (NIST) where the candidate reference method procedure is being expanded to further characterize the measurand. Because these samples represent a wide variety of disease processes, the albumin molecular forms may contain disease-related post-translational modifications or truncations that will allow NIST to better define the measurand. Aliquots of the candidate reference material will be included in each manufacturer's assay run so the commutability of the reference material can be evaluated, as well as its ability to harmonize method results.

Confirmation of participation has been received from all manufacturers, and the target start date is the week of January 10, 2011.

A participant asked if the reference material for plasma protein will also be used in this study. Greg Miller replied that it is a good suggestion since that is the reference material to which most manufacturers' methods are now traceable. Dave Bruns asked what diluent will be used and Greg Miller answered that it should be the diluent currently being used by a manufacturer because that is the current status.

7. Urine Albumin Reference Measurement Procedure, John Lieske (Mayo Clinic)

Aims: Validate the LC-MS/MS as a reference measurement procedure to quantify intact albumin in urine, compare clinical sensitivity and specificity of the 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, and develop the LC-MS/MS methodology to characterize the urine albumin measurand in normal and disease states.

The method involves trypsin digestion of both the human sample and an internal standard that is a labeled, recombinant protein. There are three trypsin fragments that can be reliably detected; two near the beginning of the albumin protein and one at the end. Three quality control materials are used to monitor precision; the CVs range from 4-8% if one high intensity transition is used for each peptide and 3-7% if all transitions for each peptide are used. For each of the three peptides, intra-assay CVs range from 1-4% and inter-assay CVs range from 3-7%. The bias between fragments has been found to be 1-3% and assay of the several hundred human urine samples will confirm these biases between fragments. There may be some biology with pieces of the albumin being chewed off or other modifications that change the way it is detected.

In order to evaluate if there are native fragments in albumin that change the way that it can be quantified, 10 urine samples that had already been assayed were filtered through a 10 kDa filter; what was retained was re-suspended, measured, and compared to the pre-filtration result. Overall a slight 4% loss of albumin was found, but it correlated exactly with all three fragments so this was interpreted as simply loss caused by the filtration procedure. If native fragments were present, they should affect each of the trypsinized fragments differently, especially the ones at the beginning and end of albumin.

David Bunk, NIST

It is important to characterize albumin in urine and determine what types of structures are represented in normal and diseased states. In order to look at the protein structure, NIST will quantify many fragments, including the three fragments measured in the Mayo method. The isotope-labeled intact albumin that is used for an internal standard in the Mayo method can also act as an internal standard for the structural characterization because digestion of this intact protein yields a ratio of peptides that reflects the intactness of the protein. When compared to the ratio of peptides in the endogenous albumin in urine, you can determine if the albumin is intact or if there are post-translational modifications.

8. Urine Albumin Adsorption Study, Mary Kimberly (for Mary Robinson), Centers for Disease Control and Prevention

Urine albumin adsorption onto various types of containers is being studied. The main conclusions are:

• For all materials and urine samples tested, greater than 99% of albumin remained in solution after five hours at room temperature.
• Adsorption in sample cups was lower for polystyrene than polyethylene, and hydrophilic-coated sample vials adsorbed less albumin than polystyrene sample cups. The same conclusion was also determined in transfer experiments, but only when the urine pH was below 5.0.
• Albumin measurements using the immunoturbidimetric Roche Tina-Quant Assay are lower when urine sample pH values are lower than 5 even though the assay buffer is pH 8.0 (the effect of urine pH on other methods is unknown).
• Albumin concentration and pH are the most important factors affecting adsorption.

It will be necessary to determine whether pH adjustment is necessary for the samples collected for the Biologic/Pathologic Variability Study, which is the next phase of this project.

9. Urine Albumin Physiological Variability Study, David Koch, Emory University

The objective of the study is to determine which urine collection approach is optimal: 24 hour, first morning void, second morning void, or random collection with ratio to creatinine. Knowledge of physiologic variability is also needed to assist with measurement procedure performance requirements. Two institutions, Emory University and Grady Memorial Hospital, have submitted for IRB approval. Diabetic subjects with stable CKD and elevated urine albumin will be recruited. Samples will be collected based on a very difficult collection protocol and shipped to Mary Robinson at CDC for albumin and creatinine testing. The 12-week collection plan for each subject is as follows:

• Week 1: Collection over 24 hours with each void collected separately starting with the send morning void and ending the next day with the first morning void.
• Weeks 2-5: Random collections, first and second morning voids.
• Weeks 6-11: Repeat weeks 1-5.
• Week 12: Repeat week 1.