Hemoglobin A1c (HbA1c) measurements are used in the clinical management of diabetes to assess long-term efficacy of diabetic control. Given the medical implications that can result from the under- or overtreatment of a patient, it is therefore of paramount importance to be able to deliver high quality HbA1c test results.
Clinicians are becoming increasingly aware that the accuracy of HbA1c methods can be affected adversely by hemoglobin variants (also called hemoglobinopathies). Furthermore, with the inclusion of a sample containing HbS in the latest College of American Pathology (CAP) Glycohemoglobin (GH2) Proficiency Survey, it appears that there now is a renewed interest in hemoglobin variant interference and its effect on both the accuracy and precision of the HbA1c result.
CAP efforts toward improving HbA1c measurements
In recent years there has been an industry-wide effort to improve the measurement of HbA1c on commercially available methods. An example of this was CAP’s 2007 decision to transition from peer-based (precision-based) to accuracy-based grading.
CAP survey materials consist of three samples whose values are unknown and cover the clinical range. These are tested by participants on their laboratory method, and the results are sent back to CAP. Participants are informed whether or not they passed based on how closely the values they obtained match to a target National Glycohemoglobin Standardization Program (NGSP) Reference Value.
The NGSP Reference Value has been determined as the “true” value. It is determined from the mean of three replicate analyses per day for two days using only seven NGSP-certified secondary reference methods. Although not every commercially available method is represented, nor is every secondary reference method FDA-cleared, the main reason for using this system is so that the samples are tested blindly by the most accurate methods available.
The acceptability limit for passing was gradually lowered in each subsequent CAP survey in order to give manufacturers sufficient time to improve their methods (Table 1). Over time, methods that could not be improved would be replaced by superior ones so that laboratories could use only methods that provide the highest quality HbA1c results.
With accuracy-based grading, CAP is asking how confident a laboratory can be that it is getting the right result when it uses its chosen method. If a certain method is running falsely high or low in the clinical setting, this can translate into how confident a laboratory is that a patient is overtreated (falsely high) or undertreated (falsely low). However, if the result obtained using a laboratory method falls within the acceptable limit, the laboratory will pass at the current accuracy grade. It is worth mentioning that even though a method is capable of achieving the correct value, the method may not be able to do this consistently if its precision is low.
Impact of hemoglobin variants on CAP participants
Until recently, CAP survey samples were provided to evaluate the impact of reduced accuracy across the measuring range for normal samples. However, laboratorians should also be concerned with evaluating the performance of abnormal samples, so the most recent CAP survey included a hemoglobin variant sample. In part due to increased immigration to the United States by members of ethnic groups that are associated with a higher incidence of hemoglobin variants, it is becoming more important to be able to identify hemoglobin variants. Variants may interfere with an HbA1c result, depending upon the method being used.
Until the latest survey, overall coefficients of variation (all sample levels and all methods) for the CAP surveys since 2010 GH2-B had been decreasing to ≤4.0% heading toward the goal of ≤3.5%.1 However, in the last (2012 GH2-B) survey many methods were challenged when a hemoglobin variant sample was introduced. This HbAS sample resulted in overall CVs of 5.6% and a mean HbA1c as high as 6.8%, which is 20.35% above the target value.2 In fact, all the methods that showed interference were biased high and not low. The clinical significance of this fact is that patients with HbAS (sickle cell trait) will show a falsely elevated HbA1c result on those methods which show a positive bias. As a result, they could be overtreated based on the incorrect assumption that they are out of glycemic control.
The discrepancy in the HbA1c results is due to the fact that HbS interferes with the HbA1c on some methods. The NGSP provides a table on its website that laboratories can view in order to determine which variants affect all the different methods.3 The NGSP provides a resource which shows that for some methods “conflicting data in the literature” suggests interference despite manufacturer claims that specific hemoglobin variants should not interfere with the HbA1c result.4 Most often the discrepancy is the result of testing at more stringent accuracy limits than were required at the time the device was FDA-cleared.
The 2012 GH2-B survey results show that methods known to have HbS interference to the HbA1c result had a significantly elevated mean bias of >0.66% and CVs ≥3.4%. Laboratories using these methods should be aware of the interference.
If we examine methods that are not known to have HbS interference to the HbA1c result, one listed on the NGSP website as having “conflicting data in the literature” for HbS interference was similarly found in the 2012 GH2B survey to have some degree of HbS interference. The significantly elevated mean bias of 0.31% and CVs of 3.8% suggest variability from interference. Additionally, five other methods, not known previously to have any interference from HbS, had significantly elevated CVs of >6.5%. This was evident even though the calculated mean bias was within acceptable limits. This indicates that some laboratories using the method may have shown interference while others did not, which would keep the overall bias for the method within the allowable limits. The clinical significance of this phenomenon is that the laboratory is unaware of any potential interference from HbS—again, based on claims that were not established using current acceptability limits.
Furthermore, as there have been more than a thousand different hemoglobin variants identified, it is very difficult to definitively say which ones will interfere with the HbA1c result. Practically speaking, it makes sense to focus on the most common ones, such as HbS, HbC, HbD, HbE, HbF, and HbA2.
In theory, most methods should be able to produce correct HbA1c results in heterozygous variant samples (i.e., HbAS) given that the HbA1c derived from the adult hemoglobin (HbA) is sufficiently present. Similarly, HbA1c results from homozygous variant samples of any kind (i.e., HbSS) would be interfered by the large amount of hemoglobin variant present.
Certain methodologies allow the operator to determine if a hemoglobin variant is homozygous (i.e., HbSS) or heterozygous (i.e., HbAS). Only with heterozygous samples can we expect a reliable HbA1c result from the HbA fraction. With no HbA present in homozygous (i.e., HbSS) samples, the HbA1c cannot be detected. This is why the variant could be interfering with the calculation of the HbA1c. For example, use the ratios of 30% and 70% to describe relative minority and majority concentrations. In a normal HbAA sample you would see areas of >70% HbA0 and 0% HbS. For heterozygous HbAS sample you would see areas of >70% HbA0 and >30% HbS. For homozygous HbSS you would see areas of <30% HbA0 and >70% HbS.
Although sickle cell trait is the most common, HbF in Hereditary Persistent Fetal Hemoglobin (HPFH) patients usually exceeds 20%, and most available methods have known interference from elevated HbF. Only methods that allow the operator to see the quantity of HbF, and to determine if it exceeds the acceptable limit, will be able to provide a reliable HbA1c result for these patients.
“Acing” the HbA1c exam: key considerations
Since grading criteria is dropping each year, it will be more difficult for laboratories to pass their CAP survey if they are using a method which gives results that are outside of the acceptable grading limit and/or inconsistent. In order to “ace” the HbA1c exam, a laboratory method should be both accurate and precise, meaning it can achieve the “true” NGSP reference value and have the lowest CVs possible. Although it is useful to know that a method is accurate, unless the method can also be precise, the laboratory will not be able to achieve the “true” value consistently.
NGSP references demonstrate that each commercially available method today has some degree of interference. Laboratories need to be aware of which hemoglobin variants are present in their population and choose the method that best meets their needs. The survey results allow the laboratory to compare its results with other participants that share the same method (to evaluate the precision of the method across the install base) or even to a different method. If another available method performed consistently well over several surveys, this could be an opportunity for the laboratory to find a better method.
Today laboratories have the ability to select methodologies that show which hemoglobin variants may be present and to determine whether the sample is heterozygous or homozygous. There are significant differences in interference and precision among all the available methods. Therefore, it is worthwhile to review the NGSP website to see if manufacturer claims are still relevant at current accuracy limits.
- Sacks DB, Arnold M, Bakris GL, et al. Executive summary: guidelines and recommendations for laboratory analysis in the diagnosis and management of diabetes mellitus Clin Chem. 2011;57:793-798.
- GH2-B Hemoglobin A1c Participant Summary Discussion by David B. Sacks, MBChB.
- National Glycohemoglobin Standardization Program. HbA1c Assay Interferences HbA1c methods: Effects of hemoglobin variants (HbC, HbS, HbE and HbD traits) and elevated fetal hemoglobin (HbF). http://www.ngsp.org/interf.asp. Accessed March 27, 2013.
- National Glycohemoglobin Standardization Program. Factors that interfere with HbA1c test results. http://www.ngsp.org/factors.asp. Accessed March 27, 2013