AACC Annual Meeting - Chicago, IL - July 31, 2014
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), Flavio Alcantara, David Armbruster, Lorin Bachmann, David Bunk (phone), Johanna Camara, Anchalee Chittamma, John Eckfeldt, James Fleming, Neil Greenberg, Lars-Olaf Hansson, Glen Hortin, Graham Jones, Hans-Joachim Kytzia, David Lacher, John Leiske (phone), Andrew Narva (Director NKDEP), Mauro Panteghini, Karen Phinney (phone), Heinz Schimmel, David Seccombe, Nickolas Voskoboev
Note: The urine albumin standardization activities are carried out as a joint project with the IFCC Working Group for Standardization of Albumin in Urine; chair, Lorin Bachmann.
Summary of action items
- Greg Miller and Lori Bachmann will schedule a meeting or conference call with manufacturers to discuss urine albumin method bias and non-linearity issues, approaches to address the issues, input on organizing the commutability assessment for new NIST SRMs, and the general approach for standardizing UA measurements.
- John Eckfeldt will circulate the report of the CAP urine albumin accuracy based survey.
- Greg Miller will circulate the protocol that was developed for the investigation of within person physiologic variability project.
- Greg Miller will appoint a small task force to make recommendations about modeling strategies to evaluate urine albumin measurement procedure performance requirements.
- Greg Miller will assign a chair to the task force for recommendations about creatinine performance requirements at low concentrations typically found in pediatric patients.
- Greg Miller and Andrew Narva will draft new language regarding eGFR equations for the web site and circulate to the LWG. If necessary, a conference call will be organized for discussion.
- Neil Greenberg will send the CAP educational publication about how to use the correct eGFR equation to Greg Miller and Andrew Narva.
- Once new eGFR equation language is on the web site, the LWG will ask CAP and AACC to issue an announcement in their e-news formats.
- Andrew Narva will circulate a pdf of the article “Chronic kidney disease controversy: how expanding definitions are unnecessarily labelling many people as diseased” published in BMJ 2013;347:f4298.
Standardization of Urine Albumin MeasurementsCommutability of current reference materials for Urine Albumin
Lori Bachmann, PhD, Virginia Commonwealth University
- The goals of this project are to assess the current state of agreement among routine urine albumin methods vs. an IDMS cRMP using native patient samples, to evaluate commutability characteristics of diluted IRMM ERM-DA470k/IFCC and JSCC candidate reference materials (cRMs) for urine albumin, and to assess the effectiveness of the cRMs for standardization of routine methods. 332 fresh native human urine samples from patients were analyzed using 16 routine quantitative methods.
- Overall, there was lack of agreement at all concentration ranges. Compared to the IDMS method, some methods showed non-linear bias, some showed constant bias and some showed proportional error. Imprecision was fairly good for most methods ranging from 5-8%. Five of 16 methods had total CV > 10%. Sample specific effects contributed to the total CV in three methods. However, for most methods sample specific effects did not contribute significantly to the lack of agreement. Non-constant bias was seen in 10 of the 16 methods and will be a challenge for commutability assessment. Bias was the dominant contribution to lack of agreement.
- Next steps are to assess the improvement in agreement among methods that is possible based on mathematical recalibration using the cRMs.
- Discussion –
- Glen Hortin suggested that the lack of linearity seen at lower concentrations could be the result of not enough calibration points at that level.
- Neil Greenberg felt that we need to know more from each manufacturer about the model used for the calibration and that more calibration points would help to create the model.
- Lori Bachmann commented that an attendee of the Wednesday afternoon symposium suggested that there are more proteins than just albumin in the IRMM materials and their value assignments were done using immunoassay based methods and the low concentration difference may be due to the reference materials themselves. The NIST candidate reference materials will be derived from undiluted native patient urine and should not have matrix problems.
- Greg Miller stated that the reason for the commutability study was to determine if the IRMM material could be diluted and used in urine assays. Manufacturers are using diluted IRMM material or purified albumin to prepare their own calibrators. The lack of standardized approaches and the fact that the IRMM material was not designed to be diluted and used for urine is contributing to the variability among results that we are seeing.
- Greg Miller noted that the long range goal is to use reference materials developed by NIST. However, understanding the characteristics of diluted IRMM material will be useful since it is frequently used as the basis for calibration traceability at the present time.
- Greg Miller suggested that we need to propose a meeting with manufacturers to review the data and discuss how to address the calibration issues identified.
- Glen Hortin commented that the data from the CAP accuracy based survey for urine albumin shows that all methods run lower than the Mayo candidate reference method. This survey gives a measure of the real world field performance.
- Heinz Schimmel cautioned about drawing conclusions on commutability based on the graphs and suggested there may be a method selectivity problem which is an assay problem and not necessarily a reference material commutability problem.
- Graham Jones suggested there would be a stronger case for recalibration with the manufacturers if it can be shown that there is a clinical effect, e.g. in his lab difference of 10% around the cut-off for urine albumin led to misclassification of about 2% of patients. He suggested that one of the big data bases like NHANES could be used to determine the effect of a bias on misclassification at the various cut points.
NIST Reference Materials Update
Karen Phinney, PhD and David Bunk, PhD, National Institute of Standards and Technology
- David Bunk presented an update of the status of the urine reference materials. The frozen urine creatinine reference material (SRM 3667) has been available for sale for about 1 year.
- The albumin reference material (SRM 3666) will be frozen human urine at four levels ranging from 5 mg/L to 300 mg/L and each level will be dispensed in 1 mL aliquots; the material will be prepared from a pool of at least 20 different patient samples.
- A plan for value assignment and commutability assessment will be developed; there will be a minimum of 80 samples available for assessment of commutability.
- Value assignment for the pure albumin primary standard (SRM 2925)has been done but the uncertainty was larger than they wanted. The collaborative study between NIST and Mayo to qualify the candidate reference measurement procedures will be used to re-value assign to reduce the measurement uncertainty.
- Discussion –
- John Eckfeldt felt a study like Lori Bachmann showed should be used to assess commutability; Lori stated that fresh urine was used for routine methods and frozen urine for the reference method. Commutability assessment should be performed using fresh patient samples across several methods but there are logistical considerations. Lori said they worked hard to have the most platforms at the fewest locations; the manufacturers needed to have their instruments ready to do testing within 1 day of sample receipt.
- Greg Miller suggested manufacturers be given plenty of advance notice about doing a commutability assessment; he suggested a two stage study in which stage one would use the individual frozen aliquots of the urine samples that made up the pools to see if the sum of their contribution to the pools is recovered in the pool. Stage two would be a smaller scale evaluation using fresh samples to be sure there is not an artifact of freezing.
- Greg Miller suggested that a small group be appointed to develop the experimental design and a time line. David Bunk estimates that if all goes well with the procurement process for the pool samples the materials could be available late this year or early in 2015. We need to work with manufacturers very soon if we want to do a study next year.
- John Eckfeldt suggested that the CAP accuracy based committee may be interested in helping with this project and suggested doing a small preliminary study using the CAP
- material. Greg Miller stated that the CAP material could be included in the commutability study. Glen Hortin suggested having some labs measure the samples fresh and frozen.
- Lori Bachmann commented that the accuracy assessment had fresh vs frozen data for the native urine sample but the samples were only frozen for about 45 min. at -70C so a new study needs to be done to really assess if there are artifacts from freezing.
Urine Albumin Reference Measurement Procedures Validation: NIST and Mayo Clinics
David Bunk, PhD, National Institute of Standards and Technology and John Leiske, MD, Mayo Clinic
- David Bunk reported that Mayo and NIST have completed a validation study using a common calibrant, NIST SRM 2925 but have not evaluated the data yet.
- Karen Phinney reported that a quick look at the data showed a few samples with more variability but there is a huge amount of data to evaluate and summarize. Timeline for data analysis is 2-3 mo.
- Greg Miller asked if it is possible to meet the JCTLM deadline for submission next year (May 2015) and David Bunk thinks a publication is needed prior to submission. However, Heinz Schimmel thinks that submission to JCTLM can be based on a manuscript that has been submitted and does not require publication.
- Neil Greenberg suggested that the manufacturers be encouraged to include methods that may be in the development process in the commutability study.
Establishing revised interpretive thresholds for Urine Albumin
- Urine albumin standardization has several parts: standardizing the assays, standardizing the nomenclature, standardizing the urine collection and then investigating interpretive thresholds/cutoffs for reported values.
- Consensus recommendations of NKDEP/IFCC:
- Use of the term “urine albumin” is recommended; “microalbumin” is discouraged.
- An albumin/creatinine ratio should be reported with all urine albumin measurements.
- International System of Units should be adopted. In the interim, uniform guidelines should be followed within a country or region.
- Albumin concentrations reported in milligrams per liter are difficult to interpret and concentrations in these units should not be the only value reported.
- NKDEP literature review looking at cutoffs and normal values:
- There is no specific quantitative evidence to justify a “normal” range. No evidence for race or ethnic-specific reference ranges.
- There is evidence of increased risk for CVD and kidney disease at urine albumin levels < 30 mg/g creatinine (e.g. PREVEND)
- There are well documented gender differences in creatinine excretion leading to over diagnosis in women.
- One suggested reference range for a “normal” ACR is 0-10mg/g in men and 0-15mg/g in women.
- Gender-specific cut-offs may eliminate gender disparity in albuminuria (NHANES III, Warram); ACR of 25mg/g women and 17 mg/g men correlate well with UAE of 30 ug/min
- Issues requiring more investigation:
- The ACR varies with age, sex, and ethnicity. Decision thresholds suitable for these subgroups need further investigation. A single decision threshold may not be adequately sensitive for each subgroup.
- The appropriate thresholds for risk for given populations (e.g., general population or high risk groups such as diabetes, hypertension or cardiovascular disease) need to be determined.
- Investigation of the usefulness of age- and sex-specific equations to convert ACR to an estimated AER for which a single reference limit may be appropriate.
- Recommendations to lower the ACR reference ranges cannot be implemented until standardization of UA assays is completed
- Based on literature review, research priorities are:
- Assess the reproducibility of the ACR values in the range that would now be considered abnormal.
- Determine whether proposed sex-specific criteria of 10mg/g in men and 15mg/g in women are appropriate based on prevalence.
- Special populations: African Americans, children, the elderly, transplant patients, and those with low lean body mass such as amputees given the influence of creatinine on ACR.
- Moving ahead:
- Standardize reporting format.
- Establish prevalence data base by performing standardized assays on reference population (e.g. NHANES, Kaiser Permanente).
- Establish new cutoff’s (+/- gender specific) or use existing 30mg/g and allow epidemiologic studies to assess risk using standardized assays.
- Educating professionals about changes.
- Collaborate with CRIC investigators to clarify research questions.
- Discussion –
- Dave Lacher commented that it is likely that the 2015 NHANES will collect 24 hour urines, 1st morning void and random samples for sodium measurement. He has lobbied for collecting enough urine to do other analytes, e.g. albumin.
- John Eckfeldt reported that the NHANES albumin assay is a fluorescent immunoassay with high sensitivity at low concentrations. He also suggests discussing with Joe Coresh the possibility of using ARIC Study samples because ARIC Study has long term follow-up on developed disease.
- Andrew Narva stated that the CRIC investigators would also be a good group for collaboration because it is a strong group of investigators working in the epidemiology and natural history of CKD.
- Lori Bachmann commented that current assay limits of quantification (LOQ) are around 4 and 12; if reference ranges are lowered, we need to let manufacturers know what the expectation for LOQ will be.
- John Eckfeldt commented that different manufacturers use different criteria for their claims.
- Mauro Panteghini feels that the LWG should define LOQ for urine albumin for the manufacturers.
- Greg Miller commented that after method standardization is adopted, and clinical investigations are published, LOQ at lower levels will become more important and we need to provide guidelines for method performance requirements at lower concentrations of UA.
- John Eckfeldt noted that data from some studies indicate that there is an increased risk below 10; methods that can measure accurately at the low end are needed to establish clinical risk.
- David Seccombe commented that the ratio includes urine creatinine so urine creatinine accuracy needs to be addressed too. Greg Miller agreed that efforts to develop recommendations for urine creatinine method performance requirements are needed.
- John Eckfeldt recalls that the CAP urine accuracy survey showed good agreement among urine creatinine methods; he will circulate the report to the Lab Working Group members.
Investigation of within person physiologic variability
- This project is not progressing. Greg Miller and Lori Bachmann have not been successful in finding the clinical collaborators and research coordinator for this project.
- A protocol has been developed and refined. Greg Miller will send the protocol to the LWG to consider if anyone is able to lead the project.
- Discussion –
- Mauro Panteghini cautioned that care be taken to look at the selectivity and specificity of the assay when choosing a method used for this study; Greg Miller commented that one of the better performing methods identified in the publication that used 320 freshly collected urine samples should be used.
- David Lacher commented that the NHANES study shows a lot of within-person variability; Greg Miller commented that the paper published by the LWG several years ago reported biological variability ranging from 4% to 110% in different literature reports and is a key reason that a careful study needs to be performed.
- Graham Jones commented that urine albumin may be reaching threshold in patients at various times throughout the day and the assumption that these are Gaussian may be false.
- Greg Miller suggested that using NHANES data to do modeling to see how much of a change in UA makes a difference in clinical decisions may be a better strategy. Greg will appoint a small task force to make recommendations about modeling study strategies to evaluate measurement procedure performance requirements.
Creatinine and Cystatin C Topics
Creatinine Performance Recommendations Outside the 1.0-1.5 mg/dL Interval (especially lower values seen in pediatric patients)
- Current web site says: “Ensure that comparable bias and imprecision extend throughout the analytical measurement range (measuring interval)”.
- Volunteers were identified for this task force last year but a chair needs to be identified.
Recommendations regarding MDRD vs. CKD-EPI equations for eGFR from creatinine; see attached CAP Survey information
- The web site needs to be revised regarding recommendations for eGFR equations.
- The suggested approach for web site content is to emphasize the importance of using an estimated GFR, use of a standardized creatinine assay and use of an IDMS traceable equation.
- CAP survey showed a substantial number (32%) of labs are using non-IDMS equation with an IDMS creatinine assay (assuming all creatinine assays used by participants are IDMS traceable).
- Neil Greenberg felt that the survey information may not be reliable because many labs do not really know which equation they are using.
- Greg Miller pointed out that 33% report results that are above 60 but only 11% are using the CKD-EPI equation meaning that 2/3 of those reporting results over 60 are using the wrong equation; the web site is very clear about which equation to use if reporting results above 60.
- John Eckfeldt suggested the web site be more neutral about using MDRD. Andrew Narva agrees that the web site language should be neutral.
- Lori Bachmann asked if CAP could facilitate getting labs to use the right equation.
- John Eckfeldt noted that virtually all creatinine assays are IDMS traceable. Perhaps CAP should be clearer about recommending in accordance with NKDEP web site language.
- Greg Miller suggested that the web site not make a strong recommendation for either equation but point out the proper use of each equation, e.g. if a lab wants to report results above 60, they should use CKD-EPI equation and if they choose to use MDRD equation, they should not report results above 60.
- John Eckfeldt commented that for older patients and women, the CKD-EPI equation is better, especially around 60.
- Neil Greenberg added that he and John Eckfeldt published a CAP educational piece about how to use the right equation; it was difficult to put out a clear message for the pathologists because they did not want to contradict the NKDEP web site.
- Greg Miller stated that he and Andrew will draft new language for the web site and circulate to the LWG and if necessary, organize a conference call for discussion.
- Neil Greenberg will send the CAP educational publication to Greg and Andrew to help them draft new language.
- Once new language is on the web site, we will ask CAP and AACC to issue an announcement in their e-news formats.
- Mauro Panteghini commented that in Europe KDIGO recommendations are becoming more popular; he recommends getting a uniform recommendation from both groups.
- Glen Hortin initiated a discussion of the reference range to use; Graham Jones stated that the KDIGO document says that values below 60 should be flagged. Greg Miller commented that a result below 60 does not mean that a patient has kidney disease because there are other factors that need to be considered, e.g. age, muscle mass, and for that reason a recommendation to flag results less than 60 may not be appropriate.
Standardization of Cystatin C Measurements and eGFR Equations Based on Cystatin C
Report on current implementation of standardization based on traceability to ERM DA471/IFCC
John Eckfeldt, MD, PhD, University of Minnesota
- Two regular “processed human plasma” (CYS-01 and CYS-02) and two “wildcard” off-the-clot, fresh frozen serum pools were prepared for the CAP cystatin C proficiency challenge. Results were reported by 132 laboratories.
- One wildcard was from normal volunteers (CYS-WC1) and one from CKD patients (CYS-WC2).
- International certified reference material ERM-DA471/IFCC was used to establish target values of the fresh frozen serum pools using a 2-point calibration curve: Normal pool target = 0.97 mg/L, CKD pool target = 2.37 mg/L
- There was a nearly two-fold variation in the concentration of cystatin C that the labs reported for each of the fresh frozen CAP samples.
- For the normal CYS-WC1 sample the central 95% of the results would yield an eGFR ranging from 116 to 65 mL/min/1.73 m2.
- For the CYS-WC2 CKD pool the central 95% of the results would yield an eGFR ranging from 35 to 19 mL/min/1.73 m2.
- Conclusion: Manufacturers of IVD’s for measurement of cystatin C concentrations and clinical laboratories using them need to improve the accuracy of their measurements if eGFR based on cystatin C is to be useful clinically.
- Discussion –
- Neil Greenberg commented that it would be interesting to see if DA471 dilutions that were used to establish the wild card pool values were run with each of the methods and then recalibrate the data. He wondered about the commutability of the DA471 material.
- John Eckfeldt reported that the IFCC cystatin group felt that the commutability study supported good commutability for the DA471 CRM.
- John Eckfeldt also commented that the DA471 dilutions were analyzed on two different methods in his lab, nephelometric (ProSpec) and turbidimetric (Gentian) assays and they gave values within 1%.
eGFR reporting based on creatinine and cystatin C equations for various categories of diseases
Lars-Olof Hansson, MD, PhD, Karolinska Hospital, Sweden
- Regression plots for cystatin C and creatinine measured for a series of pooled samples suggested that two linear segments could be identified that changed slope at about cystatin C of 2.0-2.5 mg/L. This data suggested that there may be a change in physiology of filtration that affects the two molecules differently.
- Data from eGFR calculated using a cystatin C based equation was compared to a creatinine based equation in several patient groups: primary care, cardiac clinic, neuro-ICU and oncology clinics.
- The relationship between eGFRcyst and eGFRcreat in a regression plot was different depending on the clinical group examined with a more abrupt change in the slope of the relationship occurring at about 50 mL/min/1.73m^2 observed for the neuro-ICU and oncology groups and less change in eGFRcreat than in the eGFRcyst.
- In Sweden, they will start to report eGFR based on both equations and there will be a bigger difference between estimates in sicker patients. The physician will have to interpret results based on the patient cohort.
- Discussion –
- Graham Jones suggested that the discrepancy seen in the ICU patients may be due to decreased muscle mass.
- A decision has not been made about publishing this data.
Screening of at-risk patients with appropriate biomarkers for CKD
- An evidence review on screening for early kidney disease was conducted by US Preventative Services Task Force and it was found that there are no trials evaluating screening or monitoring of kidney disease.
- The American College of Physicians (ACP) turned that evidence into a guideline on screening patients with CKD. Two of these recommendations relate to us:
- Recommendation 1: ACP recommends against screening for chronic kidney disease in asymptomatic adults without risk factors for chronic kidney disease.
- Recommendation 2: ACP recommends against testing for proteinuria in adults with or without diabetes who are currently taking an angiotensin-converting enzyme inhibitor or an angiotensin II receptor blocker.
- The American Society for Nephrology rejected the ACP recommendations.
- NKDEP recommends screening people at high risk and monitoring the urine albumin in patients under treatment. They also encourage physicians to manage blood pressure, glycemia, implement intervention to lower urine albumin and manage CVD.
- Joe Coresh developed an analysis that predicted a 60% lifetime risk of CKD and the National Kidney Foundation recommended screening everyone over 60. However, Paul Eggers, an epidemiologist from NIDDK did not feel a compelling case could be made for this recommendation
- An article worth reviewing is “Chronic kidney disease controversy: how expanding definitions are unnecessarily labelling many people as diseased” published in BMJ 2013;347:f4298 (Published 30 July 2013). Andrew will circulate a pdf of this paper.
Update from the IFCC-WASPaLM Task Force on CKD
Graham Jones, MBBS, DPhil, St Vincent’s Hospital, and University of New South Wales, Australia
- IFCC and WASPaLM (World Association of Pathology and Laboratory Medicine) have created a task force for CKD.
- Goals of the task force are to improve health in patients with CKD by supporting national adoption of CKD testing programs within countries, fostering collaboration between clinical and laboratory organizations and sharing experiences to get the best results.
- Reports from different countries:
- There is a wide variation in what is happening in this area world wide.
- In some countries such as Australia, UK, USA, Italy, Norway there are national programs with nephrologist and laboratory agreement.
- In some countries such as Croatia, India, South Africa, Uruguay a major issue is lack of creatinine standardization.
- There is limited contact with nephrologists in some countries.
- There is an issue of scale in other countries who reported >10,000 laboratories in their country.
- Major issues are:
- Creatinine measured with assays that are not IDMS calibrated. Even though the major manufacturers reported that their assays are IDMS calibrated, there are many assays in other parts of the world that are not.
- Liaison with nephrologists and between labs.
- Major manufacturers are IDMS aligned but an internet search produced 84 assays from 53 manufacturers across 4 or 5 continents. Only 12 use the term “IDMS traceable” in their package insert. Most assays were either not IDMS traceable or it was not possible to determine.
- Several agreements were made by the task force:
- Support 2012 KDIGO guidelines on Diagnosis and Management of CKD as template for National Plans.
- The task force provides a valuable resource for supporting members with activities in their countries. Members should feel free to seek assistance either from the whole group via e-mail, or any individual members as needed.
- The task force should prepare a document based on the KDIGO 2012 guidelines identifying the laboratory-related issues as a starting point for national discussions.
- Graham Jones was appointed to the W. Roman Travelling Lectureship which is awarded annually by the Australian Association of Clinical Biochemists to commemorate Dr. Wadim Roman who played a major role in the founding of the Association in 1961 and was President from 1964 to 1966. Graham’s topic for his traveling lectures is “Adventures with creatinine and eGFR” and he has been commending the work of the NKDEP and the Lab Working Group efforts to foster collaboration between labs and physicians and their achievement of making most creatinine assays from major international manufacturers IDMS traceable.