Summer 2010 Meetings
Clinical Advisory Committee | NGSP Steering Committee | Manufacturer Forum
2010 NGSP Manufacturer Forum
American Association for Clinical Chemistry Annual Exposition, July 2010
Presenters:
| Randie Little—NGSP Network Coordinator David Sacks—NGSP Steering Committee Chair Curt Rohlfing (for Curt Parvin)—NGSP |
Cas Weykamp—IFCC Network Coordinator Garry John—Chair, IFCC Integrated Project for HbA1c Carol Benson—U.S. Food and Drug Administration |
Present were members of the NGSP Steering Committee and representatives from various manufacturers, laboratories and agencies.
NGSP Progress Report: Randie Little
The NGSP has a laboratory network consisting of Primary and Secondary Reference Laboratories. The network is traceable to the IFCC network via sample comparisons performed twice a year. The number of certified methods and laboratories continues to increase. CAP results show much improvement in the comparability of HbA1c results since 1993. On the CAP 2010A survey a few methods still show significant bias and/or imprecision, but the biases for 23 of 30 methods were <0.3% HbA1c, and 93% of laboratories are using methods with CVs<5%, at all three HbA1c levels. Overall pass rates at the current CAP limits of +/-8% HbA1c were >95%. The NGSP certification criteria for the 95% CI of the differences between a method and the NGSP were tightened from +/- 0.85% HbA1c to +/- 0.75% in January 2010. The CAP began accuracy-based grading for HbA1c using NGSP-assigned target values in 2007 with an acceptable limit of +/-15%; this was subsequently reduced to +/- 12%, +/- 10% and +/- 8% in 2008, 2009 and 2010, respectively. It is difficult to directly compare the CAP and NGSP criteria since the former is based on a single sample while the latter is based on 40 samples. The most recent NGSP certification data for the 15 most-used methods on the CAP survey show that 14 methods had >95% of results within the current CAP +/-8% limit. Currently 4.2%, 21% and 6.5% of labs are using methods that show clinically significant interference (defined as ≥10% difference at 6 or 9% HbA1c) from HbAS and/or HbAC, HbAE and HbD, respectively. HbAE and HbAD interferences occur with ion-exchange HPLC methods and the interfering variant can generally be detected on the chromatogram prior to a result being reported.
CAP Grading, Future Plans: David Sacks
CAP will lower the acceptable limit for the GH-2 survey from the present +/- 8% to +/- 7% for 2011 and 2012. The original plan was to lower the limit to +/- 6% but CAP decided that this was not feasible given the current state-of-the-art. In 2012 data from the surveys will be evaluated to determine whether to lower the acceptable limit to +/-6% in 2013.
Uncertainty of CAP Value Assignments: Curt Rohlfing (for Curt Parvin)
Dr. Parvin calculated the uncertainty of both the NGSP target value assignments for the CAP survey (based on 2009 data) and IFCC value assignments for the secondary reference materials used by manufacturers. The combined uncertainty was CV=0.75%. This estimate was used to calculate the performance requirements for a manufacturer’s assay method to pass the current (+/-8%) and future (+/-7%) CAP criteria with 95 and 99% probability after accounting for value assignment uncertainty. For example, at a true HbA1c value of 7, a between-lab method CV of ~3.5% or less is required to pass the future +/-7% CAP criterion with 95% probability (99% probability requires a CV of ~2.5% or less) if there is no bias. If there is a bias of 0.1% HbA1c a CV of ~3% or less is required for 95% probability.
Discussion: In terms of a method meeting the CAP criteria value assignment uncertainty contributes less than 1% of the total +/-8%. The NGSP values are assigned by having all of the SRLs run the samples in triplicate on two different days then calculating the mean. With each survey CAP has been providing labs with information on how they would perform at the originally-proposed +/-6% limit. For HbA1c results in the upper range (~10%) accuracy is less critical, CAP is considering the possibility of setting a wider acceptance limit in this range.
IFCC Reference System Report: Cas Weykamp
The IFCC Working Group on HbA1c Standardization has been disbanded by the IFCC Scientific Division, having fulfilled its mission to develop a reference method and materials for HbA1c. The network will continue and will be overseen by the IFCC Committee on Traceability in Laboratory Medicine. Educational work will be intensified via a new IFCC Integrated Project chaired by Garry John. The IFCC network continues to re-approve network laboratories annually; there are currently two candidate laboratories. The mission of the IFCC network is to maintain the IFCC reference system and make HbA1c assays worldwide traceable to the IFCC reference method. There was an issue with the manufacturing of calibrators for the IFCC reference method, this has now been resolved. The master equations between the IFCC and NGSP as well as the IFCC and Swedish system have remained stable. There has been some deviation between the IFCC and JDS systems in the upper range of HbA1c values, this continues to be investigated. The IFCC network provides value-assigned calibrators and controls to manufacturers to establish and check traceability, as well as an ongoing monitoring program and value assignment of specimens for proficiency testing used by EQAS organizers.
HbA1c Standardisation-- the Current Situation: Garry John, Chair, IFCC Integrated Project for HbA1c
A 2010 update to the 2007 Worldwide Consensus Statement has been published. It reiterates that HbA1c results should be standardized worldwide and that the IFCC reference system represents the only valid anchor for worldwide HbA1c standardization. Estimated average glucose (eAG) is not mentioned, the number of decimal places for reporting results is specified (none for IFCC mmol/mol, one for NGSP%), and there is a recommendation that conversion tables listing both IFCC and NGSP units be made easily accessible to the diabetes community. Other recommendations are that editors of journals and other materials should require that manscripts report both units, and HbA1c should be the official term for the analyte although other abbreviations (e.g. A1C) can be used in guidelines and educational materials. The aim of the new Integrated Project for HbA1c is to establish an interface between the IFCC and clinical users of HbA1c to enable implementation of a scientifically sound reporting structure for HbA1c standardized to the IFCC Reference Method. It will consist of approximately five members; objectives will include advising the Reference Lab network and manufacturers, establishing links with scientific and clinical bodies, assist with implementation of the Consensus Statement, and develop quality targets for HbA1c measurement. Currently many countries in Europe, as well as New Zealand and Australia, are reporting IFCC and NGSP/DCCT units but not eAG. The U.S. is reporting DCCT/NGSP and eAG.
Global Implementation of HbA1c Testing: David Sacks
The president of the International Diabetes Federation, Jean Claude Mbanya, urges manufacturers to consider how they could provide accurate and reliable HbA1c assays to underdeveloped countries that currently lack of resources to make the test available. With the recent recommendation to use HbA1c for screening/diagnosis of diabetes this is more important than ever. Anyone who may have ideas in this regard should contact David Sacks.
HbA1c for Diagnostic Claim—Path to 510(K) Clearance: Carol Benson
The ADA has officially recommended that HbA1c be used for diagnosis, but currently no methods have 510(K) clearance for that use. Manufacturers will need to provide additional data to the FDA in order to make a claim that their method(s) can be used for diagnosis. The process includes the development of special control guidance to identify risks and mitigation of risks. The analytical evaluation will look at how the device performs at the recommended cutoff of 6.5% HbA1c. The method must be NGSP-certified; performance evaluation will include EP5-A2 precision evaluation at several HbA1c levels, interference studies and a robust method comparison with a NGSP SRL utilizing 360 samples over a range of HbA1c values. Analytical expections that are currently being considered include CV<5%, total error ≤6% for 99% of samples and little to no interference from common hemoglobin variants. Labeling requirements will include information on the ADA diagnostic cutoff and method performance as well as limitations. There is more work to do, the FDA welcomes input from manufacturers.
Discussion: The total error criterion of 6% for 99% of samples was developed with the idea that CAP would lower their acceptance limit to +/-6% for 2011. They have since decided on +/-7%, this criterion may need to be re-examined. There were concerns expressed about the large number of samples required for the comparison (360), this may not be practical for manufacturers or the SRLs. The criteria stated for hemoglobin variants involved spiking to obtain levels up to 20%, this may not be useful as these variants generally exist in the heterozygous state at levels of ~40%. The FDA welcomes input during this initial stage, and may invite experts in the field and manufacturers to an open workshop to advise them on the types of studies to be done. The FDA is not considering POC methods as the ADA recommendation specifically excludes them.