AACC Annual Meeting – Atlanta, GA – July 29, 2015

NIDDK's National Kidney Disease Education Program (NKDEP) Laboratory Working Group (LWG) joint meeting with the International Federation of Clinical Chemistry (IFCC) Working Group for Standardization of Albumin in Urine (WG-SAU)


Greg Miller (chair), Dave Armbruster, Lori Bachmann, Ashley Beasley-Green, David Bunk, Johanna Camara, Jian Dai, Joris Delanghe, John Eckfeldt, James Fleming, Ben Godber, Neil Greenberg, Nebila Idris, Harvey Kaufman, Hans-Joachim Kytzia, John Lieske, Andrew Narva, Maria-Magdalena Patru, Karen Phinney, Heinz Schimmel, George Schwartz, David Seccombe, Dror Yahalom

Meeting Minutes

Summary of action items

  • IVD representatives will confirm with the researchers who will be performing the measurements that November 10 is an acceptable day for receiving samples for all of the manufacturers. (Note added after the meeting: the freeze-thaw study will be rescheduled.)
  • To ensure that fresh and frozen samples are diluted similarly in the freeze-thaw experiment, dilution ratios specific to each method will be recommended to the manufacturers for the samples that will need to be diluted.
  • For the freeze-thaw experiment, manufacturers that do not provide dilution instructions will be queried in a follow-up email as to instrument dilution factors for samples above their method’s AMR.
  • In the follow-up email to manufacturers regarding the freeze-thaw experiment, manufacturers will be asked whether their IFU specifies manual dilution.
  • Ten additional samples at 500 to 1,000 mg/L will be added to the freeze-thaw experiment protocol.
  • Manufacturers will be asked to comment on issues related to aspirating samples in sequence for triplicate measurements.
  • Feedback will be needed from the manufacturers on the size of sample groups to ensure that the nonfrozen and frozen samples will be assayed in a close time interval.
  • Participants with contacts in Seattle who would have access to patients with albuminuria should provide this information to NIST.
  • Dr. Patru will research international databases for UA allowable error.

Welcome and Introductions

Greg Miller, Virginia Commonwealth University, Richmond, VA

Dr. Greg Miller, Virginia Commonwealth University. Chair of the NKDEP LWG, welcomed the participants to the annual LWG meeting. The participants introduced themselves; a list of participants is provided as Attachment A. Dr. Miller indicated that the focus of the meeting would be on urine albumin (UA) standardization activities. The UA standardization project is a joint endeavor between the NKDEP LWG and the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) Working Group on Standardisation of Albumin Assay in Urine, chaired by Dr. Lori Bachman, Virginia Commonwealth University. Dr. Miller noted that this meeting would be different than previous meetings in that there would be no separate manufacturers’ forum. A number of projects are in different stages of development, and all of them depend on input from the manufacturers; therefore, it was logical to hold a joint meeting.

Summary of Workshop on Urine Albumin Standardization

Greg Miller, Virginia Commonwealth University

A workshop on UA standardization was held on February 5, 2015. All of the manufacturers of in vitro diagnostics (IVD) for UA measurement were invited to attend. Dr. Miller indicated that during the workshop, the participants discussed the need for kidney marker testing, reviewed reference systems components, and discussed the sequence of events needed to achieve standardization of UA (i.e., completion of a reference measurement procedure, albumin in human urine reference materials, a freeze-thaw study, and a commutability study). The participants had devised a task list and timeline. The timeline from the minutes of the February meeting is provided as Attachment B. The meeting began by obtaining status updates on each of the tasks, reviewing the timeline, and revising the schedule. The timeline needed to be revised so that it would be more realistic. The manufacturers will need sufficient lead time to schedule testing.

Review and Revise Timeline for Steps in Urine Albumin Standardization

Address Concentration-Dependent Biases and the Effect of Dilution of Urine on Bias in Routine Methods

The IVD manufacturers provided updates on efforts to address concentration- and dilution-dependent biases that were identified in the assessment of performance reported in Clinical Chemistry 2014;60:471–80.

  • Dr. Jian Dai, Siemens Healthcare Diagnostics, indicated that one of Siemens’ systems in particular was determined to have a low-end bias. There are two issues: the calibration anchoring point and the model to fit the calibration curve. His company is exploring different models to fit a curve to account for biases. The effort is ongoing, and he expected that it would be completed in 2 to 3 months. However, there could be regulatory ramifications.
  • Dr. Dror Yahalom, Abbott Laboratories, stated that his company also is working on the issue of bias on the low end. He did not have a timeline but indicated that the effort is ongoing.
  • Dr. Ben Godber, Beckman Coulter, Inc., indicated that some bias is associated with his company’s method at the low end. His company would like to verify its results by analyzing real patient samples using a mass spectrometry (MS) method. Dr. Godber estimated that the effort to address bias would be complete within 6 months.
  • Dr. Hans-Joachim Kytzia, Roche Diagnostics GmbH, stated that researchers at his company had reworked the way in which they establish their standard curves in terms of the concentrations measured and the mathematical calculations. They have made some progress and are considering what a change would mean relative to regulatory affairs. He indicated that for many countries, any change in traceability triggers a full regulatory review of the program.

Dr. Miller responded that addressing method biases by IVD manufacturers would be considered part of the UA standardization effort. The U.S. Food and Drug Administration (FDA) was informed about the plans of the IVD manufacturers to address bias and was supportive of a cooperative regulatory submission based on an agreed upon experimental design for the whole project. The FDA had attended LWG meetings in the past and had been informed early in the project about efforts to address biases by IVD manufacturers.

Dr. Bachman stated that some assays exhibited deviations from linearity in the lower UA concentration range, and some were nonlinear at concentrations of 50 to 100 mg/L. Dr. Miller noted that bias appeared to be a combination of calibration nonlinearity and dilution effects. Dr. Bachman stated that for follow-up with each company, she has contacts for oversite of the project and technical contacts who perform the assays.

Experimental Design for Freeze-Thaw Effects

Dr. Miller indicated that the experimental design to assess freeze-thaw effects was sent to the manufacturers for review in May 2015. The manufacturers have provided their responses. The protocol was to measure nonfrozen samples, as well as samples that were frozen at −70°C for 3 days and thawed, from the same patient in a single run in a defined sequence. The identity of the frozen and nonfrozen samples would be blinded to participants. Three batches of samples were to be analyzed over 3 weeks. All of the major platforms on the market—Abbott Architect, Beckman Coulter, Ortho Vitros, Roche Cobas C501, and Siemens—are participating.

A schedule was established with testing performed the day after the receipt of the sample. Manufacturers were asked about the days they could perform the tests; October 20 and 27 were acceptable to all of the manufacturers. Dr. Dai indicated that Siemens also could receive samples on November 10. IVD representatives will confirm with the researchers who will be performing the measurements that November 10 is an acceptable day for receiving samples for all of the manufacturers.

(Note added after the meeting: The freeze-thaw study will be rescheduled due to the inability to get the necessary contracts for funding the study in place in time to support the original dates.)

The protocol calls for testing on the same day of receipt of samples. Samples will be shipped by FedEx to deliver by 8:00 a.m. to allow sufficient time for measurements on the day of receipt.

The participants discussed centrifugation of samples. Dr. Miller indicated that centrifugation is not a mandatory part of the sample preparation protocol. Dr. Bachman noted that the majority of manufacturers centrifuge samples before analysis. She advised that manufacturers follow their instructions for use (IFUs). Dr. Miller added that the protocol called for centrifuging cloudy samples.

Dr. Miller reviewed the comments received on the protocol:

  • The samples should extend to lower and higher UA concentrations. Dr. Miller presented the target concentrations of UA in samples: 10 samples each at six concentration ranges, the lowest being 21 to 30 mg/L UA and the highest 151 to 300 mg/L. Dr. Miller observed that at the highest concentrations, some manufacturers must dilute the samples. In the discussion of the target concentrations, the following points were made:
    • The dilution of samples outside of a method’s analytical measurement range (AMR) was discussed. Potentially, dilution errors could be introduced in the fresh-frozen comparison if a fresh and frozen sample pair are at the borderline of needing to be diluted (e.g., initial measured values of 290 mg/L and 305 mg/L when the dilution cutoff is 300 mg/L). To ensure that fresh and frozen samples are diluted the same, as well as to ensure that there is sufficient sample for analysis, Dr. Miller agreed to provide recommended dilution ratios specific to each method to the manufacturers for the samples that will need to be diluted. Some manufacturers do not provide dilution instructions. Those manufacturers will be queried by email as to the dilution factors built into their instruments for samples above their methods’ AMRs.
    • At the recommendation of the participants, 10 additional samples at 500 to 1,000 mg/L UA, which often would be seen in patients, will be added to the protocol.
    • The question arose as to whether some systems require manual dilution of samples above the AMR. Manual dilution leads to a significant potential error. Manufacturers will be asked whether their IFUs specify manual dilution in the follow-up email.
  • Sample receipt and measurement dates should be clarified. Dr. Miller indicated that this question arose because the protocol text specifies that samples will be received and measured on Tuesdays. Logistics regarding arrival times of samples was discussed above.
  • Consider having manufacturers perform the thawing. Dr. Miller responded that this change in protocol would conflict with having the nonfrozen and frozen samples blinded, which is more important.
  • Will frozen and nonfrozen samples be measured in the same or different runs? Dr. Miller replied that they would be measured in the same runs.
  • If a technical problem occurs, should both the thawed and the nonfrozen samples be repeated? Dr. Miller answered that the samples will be blinded; therefore, this cannot be done.
  • Subtle evaporation may be a problem if the same “tray or rack” is repeated three times. Dr. Miller clarified that triplicates are to be run in sequence and that the protocol states that samples should not be uncapped for more than 20 minutes. The following points were made regarding errors from evaporation:
    • The participants discussed differences among platforms as to whether a single cup is sampled three times for triplicates, which many platforms do, or whether samples need to be poured into three separate cups. The sample volume of 1 mL per instrument will be sufficient even for platforms that use separate cups. Dead volume could be an issue for multiple cups. In the follow-up email, the manufacturers will be asked to comment on issues related to aspirating samples.
    • Dr. Miller noted that according to the protocol, all triplicates in different cups should be run together before proceeding to the next sample. Manufacturers will be asked to analyze samples in a specific order. Some sample designs randomize replicates. Dr. Miller clarified that measuring the triplicates in sequence was needed to support identifying random error components in the statistical model to be used.
    • It will be important to keep the nonfrozen and frozen sample pairs close together in the sequence. If manufacturers run samples in groups, the nonfrozen and frozen sample pairs should be run in the same group. Feedback will be needed from the manufacturers on the size of sample groups. The specification for assay order will ensure that nonfrozen and frozen samples are close together in the sequence.
    • The order of analyzing nonfrozen versus frozen samples in different pairs should be randomized.

After receiving feedback from the manufacturers, a clarified protocol will be issued.

Dr. Miller stated that the outcome of the freeze-thaw study will be important for the commutability study. It would be difficult to obtain the desired concentration ranges for the reference material without being able to freeze samples.

Desirable AMR (Measuring Interval) Recommendation

Dr. Miller suggested that the LWG recommend a desirable AMR of 2 to 400 mg/L UA. This recommendation was based on the recognition that values of 5 to 10 mg/L UA likely will be clinical risk points for albumin-creatinine ratios (ACRs) in the future and the desirability of not having dilutions within the 100 to 300 mg/L UA decision area. This AMR recommendation is intended to apply to new methods and modifications of old measurement procedures. In discussion of this recommendation, the participants made the following points:

  • There was support for having an AMR that covers the 5 to 10 mg/L UA range, which is important in the early-stage disease process, as well as the 100 to 300 mg/L UA range, which also is important for clinical decision making.
  • At higher levels, the principle use of UA by clinicians is monitoring a patient’s response to treatment.
  • It is preferable to have a single measurement procedure for the entire AMR rather than to run two procedures on every sample to achieve the desired AMR.
  • A 200-fold dynamic range might not be achievable for every procedure.
  • The lower end of the recommended AMR is a limit of quantitation, not a limit of detection, which is how some methods define the lower end of the AMR.
  • Given a choice, it will be more important to focus on the lower end of the AMR when developing a technology because the upper end can be achieved through sample dilution.

Performance Specification for Bias and Imprecision

Dr. Miller discussed bias and imprecision goals that should be met by UA measurement procedures. A comparison of routine measurement procedures for UA to isotope dilution tandem mass spectrometry (IDMS) identified that substantial biases were found between methods, with median differences between methods of approximately 40 percent; interquartile minimum to maximum differences between methods of approximately 60 percent. Analytical measurement imprecision, excluding sample-specific effects (CVa), was 6 percent or less for most methods.

The relative change value (RCV) for a given UA measurement arises from analytical (CVa) and within individual (CVi) variation. Dr. Miller showed that at a typical CVa of 6 percent, CVi dominates RCV and, similarly, CVi dominates a threshold categorization. In determining goals for performance specifications, the sum of the bias and the analytical imprecision must be small enough so as not to “excessively influence” the RCV or the threshold decision. The total allowable error (TEa) is a quantification of bias, analytical imprecision, and sample-specific imprecision. From a biological variability approach, a desirable TEa would be between 21 and 43 percent, assuming 20 or 40 percent CVi, respectively. Allowing for sample-specific effects of 7 percent, based on the assessment study,1 TEa would be between 15 and 36 percent, assuming 20 or 40 percent CVi, respectively. Given the relatively large amount of within-individual variability, a TEa of less than 25 percent is reasonable for the bias and CVa error components. Assuming a CVa of 6 percent or less, this TEa corresponds to a bias between methods of less than 13 percent.

Dr. Miller posed the question of whether it would be worthwhile to survey practitioners to ask how much change in UA would influence a care decision.

The participants discussed issues related to performance specification for bias and imprecision:

  • Within-individual variability might depend on the type of samples collected (i.e., 24-hour urine vs. first morning void vs. random samples). It is difficult to determine within-individual variation. The logistics of collecting 24-hour urine samples are challenging.
  • It is particularly important to determine within-individual variation in patients with kidney disease. The logistics of doing so are even more difficult than in the general population.
  • The determination of typical within-individual variability depends on how it is studied, for example, day-to-day or week-to-week. It also depends on the type of patients (e.g., whether they are changing their medications or are at steady-state). In an earlier report from the LWG, CVi values varied from 4 to 103 percent in 25 reports, likely due to differences in experimental design, patients examined, and approach to data analysis. The central tertile was 28 to 47 percent.
  • It is unclear whether variable intermittent proteinuria typically seen in children should be considered when determining within-individual variation of UA.
  • The Chronic Renal Insufficiency Cohort (CRIC) Study provides a cohort in which changes in UA were associated with outcomes with long-term follow-up. The UA results in the study are highly reproducible.
  • The participants agreed that surveying practitioners about the effects of UA variability on clinical decision making would be a large task that is unlikely to provide useful information.
  • Recognizing that there is a great deal of variability in the literature, Dr. Miller chose CVi values of 20 and 40 percent as being realistic for demonstration purposes. Dr. Maria-Magdalena Patru, Ortho Clinical Diagnostics, volunteered to research international databases (e.g., Clinical Laboratory Improvement Amendments [CLIA], Royal College of Pathologists of Australasia [RCPA], Ricos et al.) for UA allowable error.
  • Dr. Miller stated that the analytical precision of methods likely did not need improvement. A goal for bias of 10 to 12 percent agreement among different methods most likely would be sufficient.

Urine Albumin Reference Measurement Procedures

Ashley Beasley-Green, NIST
John Lieske, Mayo Clinic

Dr. John Lieske, Mayo Clinic, outlined the Mayo-NIST study design in which 15 clinical samples were collected, divided into two sets, frozen at −70°C, and sent to the Mayo Clinic and NIST. At each laboratory, the samples would be thawed, centrifuged, digested, and analyzed with the same calibrators. The laboratories would then swap digests and run the samples of the other laboratory. The initial goal of the study had been met for receiving three urine donations each in the target ranges of 5 to 10, 10 to 20, 20 to 40, 40 to 70, and 70 to 100 mg/L UA. The samples had been frozen and sent to the laboratories in May 2015. The Mayo Clinic also had received calibrants and quality control (QC) samples from NIST, including eight calibrants of 1.04 to 150.46 mg/L UA made from NIST SRM 2925 (SRM 2925) in a charcoal-stripped urine matrix; a charcoal-stripped urine matrix blank; and five QC samples, one made from SRM 2925 in a charcoal-stripped urine matrix and four commercially available QC samples. The Mayo-NIST study timeline was for the Mayo Clinic to collect clinical samples, freeze them, and ship one set to NIST by May 2015; NIST to prepare and ship calibrants by June 2015; the Mayo Clinic and NIST to run clinical samples with NIST calibrants by July 2015; the Mayo Clinic and NIST to swap digested samples and run them with NIST calibrants by August 2015; and the Mayo Clinic and NIST to complete data analysis by September 2015.

Dr. Beasley-Green described the NIST targeted multiplexed liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay for UA. To establish a UA reference measurement system, NIST used the primary reference calibrator, SRM 2925 (highly purified albumin), and is in the process to develop a secondary reference material, SRM 3666 (albumin in a urine matrix). For the NIST candidate reference measurement procedure (RMP), a sample is spiked with a 15N-labeled internal standard, a tryptic digestion of the sample is performed, and the sample is analyzed by LC-MS/MS in multiple reaction monitoring (MRM) mode. An advantage of monitoring multiple peptides is that the method is able to distinguish different domains of albumin. For each MRM transition, there is excellent linear agreement between the ratio of the unlabeled-to-labeled analyte peak area and unlabeled-to-labeled concentration.

Dr. Beasley-Green showed the results from QC samples, comparing the NIST-made samples containing SRM 2925 recombinant protein and commercially available QC samples made from protein in human urine. Commercially available samples showed much more variability among peptides than the NIST QC samples. The participants discussed the variability of the QC samples:

  • The coefficients of variation (CVs) of the QC samples reflect the heterogeneity of the analytes.
  • Dr. Beasley-Green stated that for quantitation in the future, one transition per peptide will be used because there was no effect on the quantitative value of the transition used.
  • When asked about peptide 6, which had deviant results in many samples, Dr. Beasley-Green answered that the peptide will be kept in the method but not used for quantitation; normally, one would not choose a peptide with a proline following an arginine residue in a tryptic digestion.
  • In response to a question about whether peptides are digested in an equimolar ratio, Dr. Lieske stated that in the Mayo Clinic assay, there are subtle differences of 1 to 2 percent in the molar ratios of the peptides. The participant suggested that although calibrants and samples are digested the same way, there might be matrix effects that would affect digestion of samples.
  • A participant asked whether the pH was adjusted in the urine before the trypsin digestion. The pH might affect the albumin conformation and therefore the digestion efficiency of the internal standard versus the samples. Dr. Beasley-Green responded that to reduce sample handling, she does not adjust the pH before the digest.

Urine Albumin Reference Materials

NIST SRM 2925 Pure Albumin and NIST SRM 3666 Albumin (and Creatinine) in Frozen Human Urine

Ashley Beasley-Green, NIST

Dr. Beasley-Green discussed two components of the UA reference measurement system: SRM 2925, human serum albumin solution, and SRM 3666, albumin/creatinine in frozen human urine. SRM 2925 is an aqueous solution of recombinant human serum albumin intended as a calibrator for secondary reference measurement procedures for UA. The standard’s concentration is assayed by amino acid analysis, it is assessed for purity and structural heterogeneity, and the homogeneity and stability of the solution is verified. The secondary SRM, SRM 3666, is intended as a matrix-based secondary reference material for UA manufacturers. Target UA levels are ≤ 10, 10 to 50, 20 to 200, and 200 to 400 mg/L UA with a minimum of 20 qualified donors per level. Currently, Solomon Park is experiencing difficulty recruiting qualified donors for the highest UA level.

Regarding prequalification of donors, Dr. Beasley-Green indicated that Solomon Park Research Laboratories, Inc., which was contracted to recruit urine donors, had experienced difficulty finding 20 donors with the higher range of UA for SRM 3666. Solomon Park has access only to donors, not patients. The following suggestions were made for contacts who might provide access to donors with high levels of UA:

  • Endocrinologists who are treating patients with diabetes.
  • Urban primary care facilities; approximately one-half of whose patients will have diabetes.
  • Diabetologists at the University of Washington.

Dr. Miller asked those participants with contacts in Seattle who would have access to patients with albuminuria to provide this information to NIST. Interested patients will be given the contact information for Solomon Park, which will collect the samples.

Experimental Design for Commutability Assessment for NIST Standard Reference Material 3666 (SRM 3666) UA in Frozen Human Urine (to Also Include NIST SRM 3667 [SRM 3667] Creatinine in Frozen Human Urine)

Dr. Beasley-Green described the commutability assessment for SRM 3666 and SRM 3667. The assessment is a three-phase study: Phase 1: material production/value assignment; Phase 2: material qualification assessment, a preliminary assessment of whether SRM 3666 is fit-for-purpose; and Phase 3: commutability study of SRM 3666 and SRM 3667. The experimental design of Phase 2 involves UA and creatinine as measurands, a diverse array of measurement platforms, a sample set that includes SRM 3666 and small number of single-donor urine specimens with concentrations close to SRM 3666 levels, and triplicate analysis of all samples.

The participants discussed Phase 2 of the commutability assessment:

  • Representative clinical samples that span the range of SRM 3666 levels will be chosen for the single-donor samples.
  • The single-donor samples can be 24-hour collections so that there will be sufficient volume.
  • The creatinine levels in the single-donor samples will not be known, but the primary objective of Phase 2 is UA measurement.

Plan for Urine Albumin Standardization Steps

Greg Miller, Virginia Commonwealth University

Dr. Miller updated the plan for the steps for UA standardization:

  • Address concentration-dependent and dilution biases in routine methods. Manufacturers are addressing bias issues.
  • Resolution of differences in IDMS procedure results. The experimental design submitted by NIST and the Mayo Clinic was reviewed at this meeting. The samples will be analyzed in October 2015.
  • Experimental design for freeze-thaw effects. The experimental design was reviewed at this meeting. Dates for analyzing samples are scheduled for October and November 2015. (Note added after the meeting: these dates will be rescheduled due to the inability to get the necessary contracts for funding the study in place in time to support the original dates.)
  • Prequalification of donors to SRM 3666. This was reviewed at this meeting. There have been difficulties in meeting recruitment goals, and contacts for physicians in the Seattle area who may be able to recruit donors will be provided to Solomon Park.
  • Experimental design for pilot commutability assessment for SRM 3666. The pilot commutability study actually is a material qualification study. The study cannot be scheduled until SRM 3666 has been produced. Once all of the donors are recruited, production will be rapid. It is hoped that SRM 3666 will be available by the end of 2015 and that the material qualification study will take place in early 2016.
  • Experimental design for full commutability assessment. The experimental design will be worked on offline by NIST, NKDEP, IVD manufacturers, and the FDA, incorporating suggestions from the IFCC Working Group on Commutability. Fall 2016 is the best estimate for the full commutability assessment to begin.
  • Education that quantitative values for UA and ACR should always be reported together. This goal has largely been achieved, with UA almost universally being reported with ACR. Further education efforts will follow the completion of other UA standardization steps. Dr. Miller indicated that clinicians will need to be educated about the recalibration. Manufacturers will need guidance on communicating with laboratories, and laboratories will need guidance on communicating with physician end-users.
  • Desirable AMR (measuring interval) recommendation. The goal to make a recommendation by July 2015 was met, as discussed at this meeting. There was general support for an AMR of 2 to 400 mg/L UA, with the understanding that if technical limitations were encountered, the lower limit was more important.
  • Desirable UA measurement procedure goals. Goals were discussed at this meeting, and final recommendations will require additional consideration. A reasonable TEa goal is 25 percent. Because most methods can achieve CVa less than 6 percent, an among-method bias goal of less than 13 percent is reasonable.
  • Experimental design for manufacturer recalibration of UA measurement procedures. Independent validation of recalibration is planned for mid-2017, after the commutability of SRM 3666 is confirmed and the reference measurement procedures have been validated and submitted to or listed by the Joint Committee for Traceability in Laboratory Medicine (JCTLM). An experimental design for recalibration is needed, including what experiments and data are needed by the FDA. This experimental design could be a primary topic for the 2016 NKDEP LWG meeting, which the FDA would be invited to attend.

The participants discussed the overall timeline for UA standardization. Product calibrators with revised values will need to be distributed into the field. For creatinine, the process took approximately 3 to 4 years. Dr. Miller anticipated that by 2020, UA measurement could be standardized by all of the IVD measurement procedures. Engaging the FDA with standardizing activities will be very helpful.

Dr. John Eckfeldt, University of Minnesota, offered some options for MS-based methods that might help address the problem of digestion efficiency. These included the use of the winged peptide internal standard approach; matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF) MS, which has been used for whole blood samples; and calibrated ion mobility analysis, which has been used to determine high-density lipoprotein particle concentration and might be able to quantify UA if a suitable internal standard can be found.

Plan for a Meeting on October 29, 2015, to Address Calibration Uniformity for Measured Glomerular Filtration Rate (MGFR) Based on Iohexol Clearance Measurement

John Eckfeldt, University of Minnesota

Dr. Eckfeldt introduced the topic of variability in GRF measurement procedures, which might be more significant than had been thought. The ideal agent for measuring GFR would meet the following criteria: freely filterable by the glomeruli; neither absorbed nor secreted by the renal tubules; neither synthesized nor metabolized by the kidney; not removed from the circulation by any organ other than the kidney; and accurately measurable, preferably in both serum and urine. GFR is defined as the volume of plasma totally cleared of the filtration agent by the kidney per unit time and is given by the clearance equation: urine concentration of the filtration agent multiplied by the urine volume generated per unit time divided by the plasma concentration of the filtration agent. Clearance of creatinine, an endogenous substance, is used to estimate GFR, but tubular secretion of creatinine is a problem, particularly at low levels of GFR. Various exogenous agents have been used to estimate GFR, including radioactive agents, the use of which many institutional review boards (IRBs) will no longer permit, as well as nonradioactive inulin, iothalamate, and iohexol (Omnipaque®). Inulin is the gold standard but is difficult to measure. Many variations exist in ways to estimate GFR using exogenous filtration agents, including single bolus injection versus continuous infusion; collection of timed urine, which raises the issue of emptying the bladder, and plasma; versus measuring only timed plasma disappearance and using a two-compartment model (i.e., distribution and elimination) for fitting plasma disappearance data. Radioactive agents are measured with gamma counters, nonradioactive iothalamate is usually measured by liquid chromatography-mass spectrometry (LC-MS), iohexol is usually measured by spectrophotometric high-performance liquid chromatography (HPLC) or LC-MS methods, and inulin can be measured by a variety of methods.

The tentative date for the meeting to address calibration uniformity for mGFR is October 29, 2015, and issues to be decided are who to invite and what topics to cover. Centers that are known to measure GFR using iohexol include the University of Minnesota, the University of Rochester, the Mayo Clinic, and (possibly) DaVita in Ft. Lauderdale. Possible topics include iohexol measurement procedure accuracy, GFR measurements more generally, the use of different filtration agents, the use of nonradioactive agents only, plasma disappearance versus methods using urine collection, and steady-state infusion versus single bolus.

The participants discussed who to invite:

  • Dr. Andrew Narva, NIDDK, suggested limiting the number of participants at the October meeting on iohexol measurement procedure accuracy.
  • Dr. Narva suggested that to address broader issues of GFR measurement, a large meeting could be proposed to the National Institutes of Health (NIH).
  • Dr. Eckfeldt will send a message via email to academic medical centers to determine whether they measure GFR and which agents they use. Dr. Narva pointed out that there was limited time to decide who to invite because the date of the meeting is only 13 weeks away.

The participants discussed the topic of the October meeting:

  • Dr. Narva pointed out that there are many large studies with iohexol-based GFR as a primary outcome. Iohexol measurement accuracy is an urgent issue for studies such as the Chronic Kidney Disease in Children (CKiD) prospective cohort study. The NIH has a strong interest in resolving issues of iohexol measurement procedure accuracy.
  • Dr. Narva stated that the use of iohexol is relatively benign, requiring a limited blood draw, and if validated, it has many future uses.
  • Dr. Narva noted that mGFR is used in chemotherapy as well.
  • Dr. Narva stated that the topic of accuracy is an important one because measured GFR is the gold standard on which all estimated GFR equations are based.
  • A participant indicated that radioactive agents are used most commonly in Europe. Pediatricians do not like to use radioactive agents, however, on their patients.
  • Dr. Eckfeldt noted that one study found 10 to 15 percent differences between mGFR by iohexol and iothalamate, possibly as a result of iothalamate secretion by renal tubules. Dr. Lieske suggested that protein binding is another possible explanation for the difference.
  • Dr. Lieske confirmed that iohexol has been measured successfully by LC-MS at the Mayo Clinic in Rochester, Minnesota.
  • Dr. George Schwartz, University of Rochester, indicated that studies in children and adults showed unusual changes in GFR, raising issues of calibration. Dr. Eckfeldt explained that there are different isomers of iohexol, an endo- and an exo-isoform. When Omnipaque® is used as a calibrant, change in the ratio of the two isomers is not an issue because it is heat stabilized and the ratio of isoforms stays constant. When commercially available iohexol in powdered form is used to produce an aqueous standard, the ratio of these isoforms in aqueous solution changes over time until an equilibrium is reached. Generally, the large peak is used, and the smaller peak is disregarded in spectrophotometric HPLC methods. Dr. Miller asked if using both peaks could address the calibration issue.
  • Dr. Eckfeldt stated that some drugs also interfere with the measurement of iohexol measurement by spectrophotometric HPLC methods.
  • Dr. Schwartz proposed initiating a small round-robin study of iohexol measurements that would include the Mayo Clinic. If the first samples were sent out by early October, the preliminary results of the analyses could be discussed at the October 29 meeting. A second, expanded round-robin of shared samples could be distributed after the meeting.
  • Dr. Miller indicated that standardization of iohexol measurement is a topic whose scope would be most appropriate for a 1-day meeting.