AACC Annual Meeting - San Diego, CA - August 1, 2017
NIDDK's National Kidney Disease Education Program (NKDEP) Laboratory Working Group (LWG) joint meeting with the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) Working Group for Standardization of Albumin in Urine (WG-SAU) final summary
Welcome and Introductions
W. Greg Miller, Virginia Commonwealth University (VCU)
Dr. Greg Miller, Chair of the National Kidney Disease Education Program (NKDEP) Laboratory Working Group (LWG), welcomed the participants to the meeting and summarized the group’s activities. The LWG was formed in 2003 to review and address the problems related to serum creatinine measurement used to determine estimated glomerular filtration rates (eGFRs). While engaged in these efforts, the group also began working to standardize urine albumin (UA) measurements and, in so doing, partnered with the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) Working Group for Standardization of Albumin in Urine (WG-SAU) in March 2007. The LWG completed its tasks related to establishing limits and recommendations for serum creatinine measurement in 2010 and subsequently has focused primarily on developing reference measurement procedures and reference materials for UA measurements. The reference system components will be submitted to the Joint Committee for Traceability in Laboratory Medicine (JCTLM) for listing and then will be available for use by manufacturers for calibration traceability. Planned activities for this meeting were to review the status of the reference materials, reference measurement procedures, and recommendations for UA measurement. The participants were invited to introduce themselves; a list of participants is provided as Attachment A.
National Institute of Standards and Technology (NIST) Reference Materials for Urine Albumin
Ashley Beasley-Green, NIST; Karen Phinney, NIST; Lorin Bachmann, VCU, Chair, IFCC WG-SAU
Standard Reference Material (SRM) 2925 Pure Albumin and SRM 3666 Albumin (and Creatinine) in Frozen Human Urine
Dr. Ashley Beasley-Green provided the NIST update for the status of candidate SRMs 2925 and 3666. She explained that NIST’s role is to develop a reference measurement system to link routine clinical UA measurements to higher-order standards via an unbroken chain of metrological traceability. The primary reference material or calibrator, SRM 2925, is highly purified recombinant human serum albumin (HSA) and will be used for calibration of the candidate reference measurement procedure (RMP) in the value-assignment of the secondary urine-based reference material, candidate SRM 3666. Thus far, the total protein concentration, density measurements, and peptide profile for SRM 2925 were uniform and homogenous across the lot. A qualitative assessment of the disulfide linkages of SRM 2925 was conducted under reducing and non-reducing conditions to confirm the tertiary structure of recombinant HSA. NIST is confident that the SRM 2925 material will be a suitable calibrator for the UA candidate RMP.
Dr. Beasley-Green noted that material procurement is in process for the secondary reference material, candidate SRM 3666. The intent is to generate a four-level reference material (i.e., pooled human urine samples) with UA concentrations within the following ranges: 5–10 mg/L, 20–50 mg/L, 60–150 mg/L, and 275–375 mg/L. The candidate SRM 3666 material will contain certified values for UA and creatinine using the NIST measurement procedures.
Freeze-Thaw Influence on Commutability Assessment for SRM 3666
Dr. Lorin Bachmann updated participants on the NKDEP/IFCC project to evaluate the effects of one freeze-thaw cycle on patient urine samples and how freeze-thaw cycles influence manufacturer commutability studies. As part of a large-scaled harmonization study conducted at VCU in collaboration with the University of Virginia and Hamilton Health Sciences, Ontario (Bachmann et al., Clinical Chemistry, 2014;60:471–80), evaluation of 17 commercially available UA measurement procedures on more than 300 unfrozen patient samples (i.e., samples stored between 2 and 8 °C) indicated bias as the dominant source of disagreement between the manufacturers. Evaluation of 44 pairs of frozen samples stored at 70°C for 45 minutes and unfrozen patient samples from the prior study using the prior 17 measurement procedures showed a difference in concentration of less than 1 percent between frozen and unfrozen samples. Noting the sample size of 44 and the freeze time of 45 minutes as limiting factors, additional evaluations were performed.
In a follow-up study, 66 samples were collected from residual patient samples submitted for routine care, ranging from 15 to 1,000 mg/L UA; these samples were centrifuged and divided into two aliquots. One aliquot was stored between 2 and 8 °C, and the second was stored at -70 °C for a minimum of 66 hours. The frozen samples were thawed, and the pairs were shipped overnight to five manufacturers representative of 10 commercially available methods. Blinded to the sample pairs’ treatment, manufacturers measured each sample in triplicate upon receipt. Preliminary results (unpublished data) showed that all manufacturers’ methods had biases of less than 1.2 percent for frozen versus non-frozen samples. Five of the 10 methods had outliers that were greater than 10 percent, but did not show a pattern of clustering in any one direction. The median coefficient of variation (CV) did not differ between frozen and non-frozen samples. Future steps include submitting the manuscript for publication and planning commutability assessments after the RMPs are finalized and the SRMs are made available.
- Dr. Miller asked about the projected availability date for SRM 2925. Dr. Beasley-Green explained that the plans are to repeat amino acid analysis of the SRM 2925 to achieve a lower standard deviation so that error is not propagated to the secondary standard. This will ensure that the uncertainties are suitable for use as calibrators for the clinical measurements and will involve adding more data points to the analyses. NIST anticipates releasing the SRM 2925 material in winter or spring of 2018.
- Dr. Miller asked how the SRM 2925 compared to existing commercially available HSA calibrators, such as Sigma-Aldrich products. Dr. Beasley-Green explained that the SRM 2925 was used in validation studies with the Mayo Clinic but direct comparisons with other commercially available HSA materials had not been conducted, but such experiments should be considered in the future.
- In response to a question from Dr. Graham Jones about the creatinine measurements, Dr. Beasley-Green clarified that the creatinine values will not be target values, but will be the “as found” values measured in the four levels of candidate SRM 3666.
Reference Measurement Procedures for Urine Albumin
Ashley Beasley-Green, NIST; John Lieske, Mayo Clinic; Jesse Seegmiller, University of Minnesota
NIST and Mayo Clinic
Dr. Beasley-Green detailed the NIST Candidate RMP for UA, which is based on an isotope dilution mass spectrometry (ID-MS) approach (Beasley-Green et al., J. Proteome Res., 2014; 13). The procedure consists of a 15N-labeled full-length HSA internal standard, which is added to a urine specimen prior to centrifugation to remove any interfering particulates and to account for pre-analytical variation in trypsin-digestion. The digested material is analyzed via liquid chromatography-tandem mass spectrometry (LC-MS/MS) using the Agilent 6460 platform and is assessed both quantitatively (UA concentration) and qualitatively (heterogeneity of HSA) based on 10 peptides that span the amino acid sequence of HSA.
Dr. Beasley-Green next described the status of the NIST–Mayo Clinic UA Method Comparison Study 3, a validation study comparing the UA measurements obtained via measurement procedures based on ID LC-MS/MS conducted by NIST to those conducted by the Mayo Clinic Renal Reference Laboratory. The goal of the study was to assess the dynamic range of both measurement procedures. Validation of the performance of the two candidate reference measurement procedures used single-donor urine samples ranging in concentration from 0.9 to 318 mg/L provided by the Mayo Clinic and calibrators (SRM 2925) and quality control materials (commercially available) ranging from 0.9 to 372 mg/L supplied by NIST. Each laboratory processed and analyzed the same materials and the UA concentrations based on three peptides were compared, the sequences of which were confirmed by Basic Local Alignment Search Tool to be unique to human albumin. The estimated UA content for the commercially available materials was supplied by the manufacturer, and the Mayo Clinic provided the estimated UA concentrations for the donor specimens based on the Roche immunoassay for UA. The NIST values were derived from the peak area ratios (unlabeled-to-labeled peak area ratios) from the standard calibration curves for each peptide. Initial observations showed clear differences in the immunoassay-estimated values and the NIST values, as well as nonlinearity in the standard curves across the broad dynamic range in the NIST method. Efforts to address these calibration issues are ongoing. Separate calibration curves in the concentration intervals of 0–100 mg/L and 100–400 mg/L produced UA measurements by the NIST method that were comparable to the immunoassay estimated values.
Dr. Beasley-Green pointed out that difference plots of the Roche immunoassay versus the NIST method using the three-peptide quantification system showed slight statistical differences and a slight decrease in measurement values for the NIST method compared to the immunoassay method, with an increase in the differences at higher concentrations. Difference plots using the 10-peptide quantification system showed lower biases between the two methods. Dr. Beasley-Green concluded that the NIST Candidate RMP for UA is fit-for-purpose for value-assignment of SRM 3666 as determined by a multi-level calibration system.
University of Minnesota
Dr. Jesse Seegmiller reported that the Advanced Research and Diagnostics Laboratory (ARDL) at the University of Minnesota is committed to supporting the UA standardization project. The ID-LC-MS/MS method has been established on the assay platform instrumentation. The ARDL is awaiting funding to support purchasing the isotopically labeled albumin internal standard and will partner with the Mayo Clinic for future comparison studies. Dr. Miller noted that the NKDEP LWG is supporting preparation of the labeled internal standard, which is not currently commercially available.
The calibration differences between the NIST and Mayo MS methods were discussed. The following points were made:
- Dr. Beasley-Green pointed out that the calibrator peak area ratio measurements should be consistent regardless of the platform used, suggesting an issue with the Mayo MS platform. The unlabeled peak areas were inconsistent, and the analytical performance of the method would need to be reevaluated before making any declarations. Historical data from the Mayo MS platform has not yet been reviewed.
- Dr. Bachmann remarked that the calibrators may not be commutable with the donor samples.
- Dr. Karen Phinney was hesitant to propose an explanation for the inconsistency of the signal intensity of the calibrators. She clarified that this update was meant to inform the LWG of the status of the project and recent challenges with the raw data, not to make any speculations.
- A participant noted the nonlinearity of the NIST full calibration curve and wondered whether there were plans to address this issue in the future. Dr. Phinney acknowledged the nonlinearity of the full calibrations and indicated that dual calibrations are better at quantifying donor specimens in the target ranges specified. Dr. Jones suggested that an additional curve-fitting process might improve the linearity of the NIST calibration.
- Dr. John Lieske noted that this comparison study was the first time that Mayo Clinic’s Renal Reference Laboratory had used its ID-LC-MS/MS method during this year. He advised the attendees not to consider the data definitive until further evaluations have been performed. Dr. Lieske added that the Mayo MS method has compared favorably with the Roche immunoassay method in prior studies.
- Dr. Miller explained that instead of a full calibration approach, a bracketing approach could be used for value-assignment of the NIST SRM 3666. Dr. Beasley-Green indicated that NIST will reevaluate reference standard calibration curves for the higher concentrations and it is a common practice to bracket the concentration of a SRM with suitable concentrations of the calibrator.
Issues associated with reporting high UA concentrations were discussed. The following points were made:
- Dr. Andrew Narva noted the challenge in standardizing albumin/creatinine ratios and highlighted the significance of being able to directly measure UA at the higher levels versus the use of dilutions for routine medical laboratory measurements. Patients often present to the clinic with high UA levels, and clinical laboratory reports for these patients often indicate levels “above assay limits,” rather than specific measurements. Dr. Jian Dai commented that manufacturer assays are validated over a specific measuring interval.
- Dr. Miller remarked that the usual practice for UA concentrations that exceed the validated measuring interval for medical laboratory methods is to dilute the urine and repeat the measurement to obtain a quantitative value. How to validating diluted samples that are above the assay measuring interval is a topic that can be addressed by the LWG. Dr. Bachmann pointed out that smaller community-based laboratories may choose to measure only within the manufacturer’s claimed measuring interval, while larger institutions may choose to internally validate assays at levels higher than the manufacturer’s measuring interval. Dr. Miller added that laboratories’ decisions likely are driven by increased regulatory requirements for accreditation and documented validations, rather than the measurement itself. He suggested generating an opinion paper addressing the significance of reporting UA above the current validated limits and recommending a generic validation design, with examples, for manufacturers.
Strategy for Implementing Calibration Traceability to a Reference System
W. Greg Miller, VCU
Dr. Miller reviewed the strategy for implementing calibration traceability to a reference system, which is based on the International Organization for Standardization 17511 requirements for the metrological traceability of values assigned to calibrators and control materials. Achieving traceability for UA measurement procedures that will be commercially available is the ultimate goal of the LWG. Two schemes for manufacturers to achieve standardization of UA are possible, and both depend on the availability of the SRM 2925 calibrator. The first standardization scheme involves the use of SRM 3666 albumin in urine as a commutable secondary reference material that manufacturers can use to calibrate their commercially available measurement procedures to achieve traceability to the NIST RMP and primary reference material SRM 2925. Manufacturers also can use the second standardization scheme, in which a reference measurement procedure calibrated with SRM 2925 is used to value assign a panel of patient samples that then can be used by manufacturers to calibrate their commercially available measurement procedures. Standardization Scheme 2 will depend on the availability of reference laboratories that can provide reference measurement services (e.g., Mayo Clinic or University of Minnesota) and human samples. Efforts are continuing to move closer to having the necessary reference system components available, but it may be several years before a suitable JCTLM listed reference system is available.
Dr. Miller introduced for discussion the idea of a NKDEP Certification Program for measurement procedures that achieve calibration traceability for UA similar to the Centers for Disease Control and Prevention’s (CDC’s) Lipids Standardization Program and the National Glycohemoglobin Standardization Program’s Hemoglobin A1c (HbA1c) Standardization. A steering committee, procedures, criteria, certificates, and reference laboratory resources would be needed, and manufacturer endorsement and the availability of resources would need to be addressed early.
Dr. Miller pointed out that another step in the UA standardization process that will be needed is surveillance of standardization based on proficiency testing (i.e., external quality assessment [EQA]), which requires commutable materials and reference laboratory target values. Such proficiency testing programs with commutable UA materials are available from providers in several countries. The NKDEP LWG could consider a program that coordinates such proficiency testing programs as an assessment tool to provide feedback to manufacturers on agreement among different methods. The next steps will be to complete standardizations with SRM 3666, RMPs, and reference laboratories; decide on the added value of a certification program; and collaborate with proficiency testing providers.
The concept of developing a NKDEP Certification Program for UA was discussed. The following points were made:
- Dr. Narva pointed out that the NIH is a research-based organization and could provide the scientific basis for such a program, but it cannot act in a regulatory capacity.
- Dr. Charbel Abou-Diwan and Dr. Dai noted the added cost associated with annual certification renewals for the various manufacturer platforms, which could be a challenge for smaller companies.
- Dr. Horst Klima commented that the availability of human samples that would be broadly distributed for certification could be a limiting factor. One solution would be for manufacturers to provide urine samples in the pre-established concentration range.
- Dr. Narva remarked that more than half of chronic kidney disease (CKD) diagnoses in the United States are made based on a single patient sample for UA.
- The sense of the group was that a certification program is likely too complex and expensive to pursue.
Contacting EQA providers to develop a formal coordinating service was discussed. The following points were made:
- Dr. Jones suggested providing quality standards or allowable performance limits for the reference laboratories.
- The sense of the group was that a coordination effort is likely not needed because manufacturers already participate in available proficiency testing programs and get suitable information for assessment purposes.
Standardization of Urine Albumin Measurements: Status and Performance Goals
W. Greg Miller, VCU
Dr. Miller provided an update on the performance goals for bias and imprecision discussed at the 2015 NKDEP LWG meeting, which have been submitted as an invited mini-review to the Journal of Applied Laboratory Medicine on behalf of the NKDEP LWG and the IFCC WG-SAU that is expected to be published in late 2017. The summary of how the recommendations were derived and the full recommendations can be found in the publication. The within-individual biological variation for urine albumin is difficult to determine, but reasonable estimates are 20 to 25 percent, and this within-individual variability dominates other sources of analytical variability. The report on performance of urine albumin measurement procedures in Clinical Chemistry (2014;60:471–80) supports that the imprecision and specimen-specific effects goals are met by almost all current measurement procedures and that the bias is the main challenge being addressed by the standardization program currently being developed. The desirable bias goal versus a reference measurement procedure is 13 percent or less, and the optimal bias goal is 7 percent or less.
- Dr. Narva commented that the opportunity exists to educate the public on the uncertainty of clinical laboratory results, which often are used to inform major decisions. He suggested including these types of data on the NKDEP website. Dr. Miller pointed out that preparing a comparable summary for the website is an option once the paper has been published.
Matrix for Low Serum Creatinine Concentration in NIST SRM 967
Johanna Camara, NIST
Dr. Johanna Camara reported on the status of the project to develop a low-concentration creatinine reference material in human serum and presented results from an interlaboratory study evaluating creatinine levels in synthetic serum. Six candidate low-level artificial serum creatinine materials derived from adding NIST materials (e.g., SRM 914a, pure creatinine; SRM 909c, frozen human serum) to commercially available materials (e.g., SeraFlx and SigMatrix Ultra from MilliporeSigma) were analyzed by the NIST isotope dilution LC-MS reference measurement procedure. Ten laboratories—five manufacturers and five routine laboratories—participated in the study and analyzed the candidate materials using a total of 11 Jaffee-based assays and 12 enzymatic assays. In addition, all participating laboratories analyzed the SRM 967a, creatinine in frozen human serum, at Level 1 (i.e., adult normal [0.847 mg/dL]) and Level 2 (i.e., high level [3.877 mg/dL]). Participants’ data were coded, and the manufacturers’ identities will not be disclosed unless permission is granted. Dr. Camara summarized the conclusions, noting that there were routine assay errors and quantification bias to the NIST reference measurement procedure. She suggested several options to improve results in the future. Procedural values could be minimized by diluting normal serum rather than spiking the blank material, and using artificial serum not containing phospholipids could reduce the solubility and interference issues. Future plans include determining whether any of the six candidates should be further developed as a new NIST material and performing stability testing at -20 °C and -80 °C.
- Dr. Miller commented that NIST Creatinine Candidate 4 looked promising and is worth pursuing further; if successful, this would help to address the lack of a low-creatinine reference material.
- Dr. Dai commented that glucose and uric acid that interfere with some Jaffe methods and are absent or diluted in these samples might account for the higher variability observed in this study.
- Dr. Klima suggested conducting a control study with the participant laboratories using a human serum sample measuring 0.4 mg/L creatinine. This would provide information on the accuracy and precision of the methods in this low range. Dr. Camara explained that single-donor samples with concentrations in this range are limited. NIST will rely on the LWG and their collaborators to assist with procuring samples.
- Dr. David Seccombe noted that the Canadian Laboratory Initiative on Pediatric Reference Intervals (CALIPER) maintains a biorepository and would be a resource to leverage to obtain serum samples with low concentration creatinine.
- Dr. James Fleming commented that his large commercial laboratory may be able to assist with obtaining serum samples with low creatinine.
- Dr. Narva added that the CKiD (Chronic Kidney Disease in Children) research program also has banked samples and may be able to assist with low creatinine levels.
- Dr. Seccombe also commented that he uses a dialysis approach to obtain low-creatinine samples in his external quality assessment program and has experienced good performance from those samples. He offered to discuss this further as a possible approach to prepare a low-concentration sample.
Creatinine Assay Standardization in Developing Countries
Graham Jones, St. Vincent’s Hospital Sydney
Dr. Jones discussed the status of creatinine assay standardization in developing countries. The in vitro diagnostic (IVD) manufacturers are crucial partners in standardizing creatinine measurements worldwide. The recommended NKDEP requirement for the United States and other countries is to calibrate serum creatinine methods to be traceable to an IDMS reference measurement procedure. This terminology is a useful shorthand for producing assays where the results are traceable to reference materials listed on the JCTLM database via JCTLM-listed reference methods (which are IDMS). In the United States and other developed countries, the instructions for use of commercial measurement procedures contain clear information indicating the traceability of the assay. Conversely, in much of Asia, Africa, South America, and the Middle East (i.e., developing countries), these instructions are not as clear. Dr. Jones’ 2015 assessment of instructions for use of creatinine measurement procedures reference materials that were available on manufacturer websites worldwide revealed that out of 88 assays, 15 contained clear statements that results were traceable to IDMS, and 17 calibrators were supplied with traceability data that could not be linked to IDMS. The majority of the assays either did not provide traceability data or supplied information that could not be verified. (Biljak et al., Biochemia Medica 2017;27(1):153–76). Furthermore a recent scientific poster from Sri Lanka indicated that six out of 14 manufacturers marketing creatinine reagents in that country did not provide information on calibration traceability. In this setting, not only are creatinine assays often producing results that are significantly biased, but laboratory workers are unable to determine the traceability from the instructions for use to consider a response. Dr. Jones outlined ways to improve the creatinine standardization issues in developing countries, suggesting that laboratories should become educated on assay traceability, manufacturers should produce and provide traceability information, and users should be aware of the problem and any governing factors.
Creatinine assay standardization in developing countries was discussed. The following points were made:
- Dr. Fleming suggested using CDC’s counterparts in other countries to disseminate information.
- Dr. Seccombe suggested that the NKDEP LWG consider engaging the IFCC Task Force on CKD, whose aim is to promote, support, and coordinate international activities related to laboratory testing around CKD.
- Dr. Narva suggested developing content to post on the NKDEP website regarding education on traceability. Dr. Miller also recommended posting to the NKDEP and IFCC websites information on obtaining quality reagents for clinical measurements in developing countries.
The meeting was adjourned at 11:22 a.m. EDT.
- A subgroup of LWG members will draft an opinion paper addressing the importance of reporting UA concentrations that are above the validated measuring interval of commercially available measurement procedures and the effect these concentrations have on patient health. The document will include criteria for validating dilution and recovery and will be circulated to the LWG members and manufacturers for input. Volunteers to write the report included Drs. Miller, Bachmann, Jones, Fleming, Lieske, and Narva.
- NIST and Mayo Clinic will continue to assess performance of their reference measurement procedures to resolve calibration and procedural details.
- Dr. Miller will prepare a graphical depiction or equivalent describing the uncertainty of UA measurements, which will be posted to the NKDEP website.
- LWG members will assist NIST in procuring low-level human serum samples for a control study.
Charbel Abou-Diwan, Ph.D.
Director, Medical and Scientific Affairs
Lorin Bachmann, Ph.D., DABCC
Associate Professor, Department of Pathology
Co-Director, Clinical Chemistry Laboratory
Virginia Commonwealth University
Ashley Beasley-Green, Ph.D.
Staff Scientist, Biomolecular Measurement Division
National Institute of Standards and Technology
Johanna Camara, Ph.D.
Research Chemist, Chemical Sciences Division
National Institute of Standards and Technology
Jian Dai, Ph.D., FACB, FCACB
Siemens Healthcare Diagnostics
James Fleming, Ph.D., FACB
Vice President and Director, Scientific Affairs
Laboratory Corporation of America
Paul Glavina, Ph.D.
Director, Research and Product Development
Lars-Olof Hansson, M.D., Ph.D.
Karolinska University Hospital
Graham Jones, M.D.
Associate Professor, Department of Chemical Pathology
St. Vincent’s Hospital Sydney
Horst Klima, Ph.D.
Manager, Research and Development
John Lieske, M.D.
Professor of Medicine
Director, O’Brien Urology Research Center
W. Greg Miller, Ph.D.
Working Group Chair
Professor, Department of Pathology
Co-Director, Clinical Chemistry Laboratory
Director, Pathology Information Systems
Virginia Commonwealth University
Andrew Narva, M.D., FACP
Director, National Kidney Disease Education Program
National Institute of Diabetes and Digestive and Kidney Diseases
National Institutes of Health
Tetsuya Nejime, Ph.D.
Serotec Co., Ltd.
Evan Ntrivalas, M.D., Ph.D.
Director, Medical and Scientific Affairs
Candice Oldenburg, M.B.A.
Senior Global Product Manager
Karen Phinney, Ph.D.
Research Chemist, Biomolecular Measurement Division
National Institute of Standards and Technology
David Seccombe, M.D., Ph.D., FRCPC
Canadian External Quality Assessment Laboratory
Jesse Seegmiller, Ph.D.
Assistant Professor, Department of Laboratory Medicine and Pathology
University of Minnesota
Nickolas Voskoboev, Ph.D.