1. NGSP Progress Report—Randie Little , NGSP Network Coordinator

• Hemoglobin Variant Interference
  • Based on CAP survey data 11% of labs are currently using methods with clinically significant HbAS and/or HbAC interference (these are the two most common variants in the U.S.).
  • The Roche Cobas Integra is being phased out and replaced by the Roche Integra Gen. 2 which does not show interference from these variants, once this process is complete only 5% of labs will be using methods that show significant HbAS and/or HbAC interference.
  • Twenty-one percent of labs are using methods that show clinically significant interference from HbAE (the second most common variant worldwide) and 6% are using methods that show interference from HbAD (fourth most common).
  • Interference from hemoglobin variants is a concern but most labs are using methods that show no interference from common variants.
• Network monitoring
  • The NGSP PRLs and SRLs continue to demonstrate excellent comparability.
  • Mean between-lab CVs by month for the network were 2-3% over the past year with the exception of one month where the CV was barely above 3%.
• Certification
  • The number of certified methods has leveled off over the past several years but the number of certified labs continues to increase.
  • 70 laboratories (64 Level 1, 6 Level 2) were certified over the past year.
  • Most of the certified labs are from outside the US.
  • The NGSP is trying to get more large reference labs to certify as level 1 laboratories
• Reducing variability (CAP data)
  • The NGSP has resulted in dramatic improvement in the comparability of HbA1c results in the field (based on CAP survey data) since 1993.
  • Some methods still show significant bias and/or variability.  There are methods that show little overall bias but much variability; others show little variability but significant bias.
• Is HbA1c Measurement Adequate for Clinical Use?
  • Holmes et. al (Am J Clin Pathol 2008;129:540-7)
    • Compared on-site POC method (DCA2000) with clinical laboratory method (Variant II) over time
    • Compared 7 methods from CAP GH2 survey over time between 2004 and 2006.
    • Concluded that there is a gap between the analytical quality experts consider necessary for maximum clinical usefulness of the test and the current level of performance actually seen in the field.
  • Westgard: A War of Words in Laboratory Medicine, Part IV: A Sentinel Example (http://www.westgard.com)
    • Reviewed 2007A CAP GH2 survey data and evaluated accuracy, bias, total error and reference change values (RCVs)
    • RCV is the change in a patient’s serial test results that can be considered significant and not just due to analytical or biological variation.
    • Concluded that overall agreement in terms of bias is very good but some methods still have problems with accuracy and/or imprecision.
  • Although there has been considerable improvement in the comparability of HbA1c results among methods and laboratories since 1993, there has been virtually no improvement over the last several years; the between-lab CVs still remain at ~5%.
  • How to make RCV smaller?
    • Use lower confidence interval:  Is it OK to be 80 or 90% confident instead of 95%?
    • Reduce analytical CVs: At 7% HbA1c:
      1. At 95% probability, analytical CVs of 2 and 3% correspond to RCVs of 0.55 and 0.70% HbA1c, respectively.
      2. At 90% probability, RCVs at analytical CVs of 2 and 3% are 0.46 and 0.59% HbA1c, respectively.
  • Current efforts to decrease variability among methods
    • Beginning September 2007 the NGSP tightened the criteria for manufacturer certification (95% CI of the differences from +/-1% to +/-0.85% HbA1c).
    • Beginning with the 2007A GH2 survey, the CAP began using accuracy grading with NGSP targets as the only grading system.  A +/-15% limit was used in 2007; this has been tightened to +/-12% in 2008.
    • The acceptable limit will be reduced in future CAP surveys.

Discussion:   Methods showing little variability but significant bias tend to show these patterns repeatedly from one survey to the next, generally due to issues with assignment of calibrator values.  Fortunately the methods showing the worst CVs on the CAP surveys are used by a small number of labs. All-method CVs should be used to calculate RCVs.  Is it possible for the NGSP to assess how well the NGSP certification criteria correlate with CAP grading limits?

2. CAP Criteria Update—David Sacks, Chair, NGSP Steering Committee

• In the past the CAP used peer group grading for HbA1c
• In 2007 the CAP changed to accuracy grading based on DCCT target values set by the NGSP network
• The initial criteria were +/-15% which resulted in a 99% overall pass rate.
• In 2008 the criteria were reduced to +/- 12%, the CAP has now agreed to further reductions:
  • 2008B Survey—remains at +/-12%
  • 2009—reduced to +/-10%
  • 2010—reduced to +/-8%
  • 2011—reduced to +/-6%
• At that point the criteria are to be re-evaluated, further reductions are possible.
• Pass Rates (at the middle HbA1c level):
  • ~98% of labs would still have passed the current +/-12% criteria on the 2007B survey.
  • Data from the 2008A survey show that ~97% of labs would have passed at +/-10% and ~95% would pass at +/-8%.
• CAP will try to have the middle level at ~7% and the high level ~9.5-10% HbA1c in future surveys.

Discussion:  The high pass rates indicate that the relatively high overall imprecision seen on the CAP survey is due to a small # of methods that do not perform well.  CAP will notify end-users of the planned changes to the accuracy limits on the next survey.  Since HbA1c is not a “regulated” analyte laboratories that fail can still test but they must provide documentation of corrective action.  It was noted that, given the future 6% criteria, the NGSP needs to examine closely the uncertainty of the NGSP target value assignments and the performance of the NGSP network laboratories at a 6% criteria; the NGSP will do this.  Achieving the 6% criteria will be difficult for manufacturers but assays must improve to meet clinical requirements, especially now that HbA1c has been recommended for screening/diagnosis of diabetes.  It is likely that many POC methods will not be able to meet the more stringent criteria.  It would be very helpful to manufacturers if they could obtain more detailed information from the CAP surveys, the CAP is willing to provide what they have and consider additional questions for future CAP surveys.  Manufacturers wanting this information from CAP should contact Randie Little.

3. IFCC Manufacturer Meeting Report—Garry John, Chair, IFCC WG on HbA1c Standardization

• The IFCC held a meeting with manufacturer representatives in Milan, Italy in December 2007.  The implications of the ADA/EASD/IDF/IFCC Consensus Statement were discussed:
  • Reporting of average glucose is not a manufacturer issue, laboratory computer systems should be the method for reporting estimated average glucose (eAG).
  • Implies that three numbers (IFCC mmol/mol, NGSP % and eAG) will be reported.
  • Manufacturers should look toward reporting at least two numbers, IFCC and NGSP.
  • As of 1/1/2011 all instruments sold, will be required to report both IFCC and NGSP numbers.

Discussion: Requiring platforms developed prior to the deadline to report both numbers is problematic; many laboratories keep the older systems in service even after purchasing new systems.  It is likely that different countries will report numbers differently.  For example, the UK has decided to report IFCC and NGSP in the near future but not report eAG.

4. NGSP Clinical Advisory Committee Meeting Report—David Sacks, Chair, NGSP Steering Committee

• The CAC consists of representatives of the major clinical diabetes organizations; its purpose is to advise the NGSP on clinical aspects of HbA1c.
• RL gave an update on the NGSP and hemoglobin variant interference.
• There was a strong consensus that variability in HbA1c must be reduced.
• The move toward reporting HbA1c results as estimated average glucose (eAG):
  • The ADA has begun educational campaigns and are handing out flyers to physicians
  • The intent is to report results as HbA1c and eAG in the U.S.
  • Individual countries will likely decide what numbers they will report.
• The other issue discussed was the use of HbA1c for screening/diagnosis of diabetes.  MS is a member of the ADA committee that recently recommended HbA1c for screening/diagnosis, but specific HbA1c cutoffs and a timeframe for implementation were not decided.

 Discussion: The results of the ADAG study showed average glucose numbers that were lower than those reported from the DCCT data, which are the values physicians have been using for the past several years.  Concerns were expressed regarding the use of eAG; the relationship may not apply to some patients.  90% of patients in the ADAG study fell within the apriori criteria of eAG within +/-15% of the regression line but that means 10% of patients fell outside this window.  Also, reporting of three different numbers could potentially cause confusion, education of physicians will be extremely important.  The intent is that physicians will still treat based on the HbA1c number; eAG is simply a tool to explain results to patients.

The Expert Committee that recommended the use of HbA1c for screening/diagnosis of diabetes consisted of a very broad international group that included clinicians, epidemiologists and laboratory representatives.  In the end there was a quick consensus in favor of using HbA1c for diagnosis, although a specific cutoff was not set it was generally agreed that it should be somewhere around 6.5.  Only microvascular data were discussed, macrovascular data will probably come into play at some point.  Cardiologists will likely add HbA1c to the factors used to estimate CVD risk; this will make accurate results even within the normal range more important.   The recommendations will specify that HbA1c must be done in, or at least confirmed by, a central laboratory.