Impact of heterozygous hemoglobin E on six commercial methods for hemoglobin A1c measurement

Introduction

Hemoglobin E (HbE) is a variant hemoglobin caused by a single point mutation, resulting in glutamic acid to lysine substitution at position 26 of the beta chain of the hemoglobin (β 26 Glu→Lys). The amino acid substitution shifts the overall molecular charge more basic, which can be detected by separative methods, such as electrophoresis and liquid chromatography. Globally, HbE is the second most common hemoglobin variant with a prevalence of up to 40% in certain populations in South and South-East Asia (Bachir & Galacteros, 2004).

Hemologbin A1c (HbA1c) measurement is recommended for monitoring of long-term glycemic control and treatment titration in patients with diabetes (American Diabetes Association, 2009). The measurement of HbA1c can be affected by the presence of hemoglobin variants, leading to spurious measurement that can adversely affect clinical decision making. This study examined the impact of heterozygous HbE on HbA1c measurements using six commonly used commercial methods and a modified isotope-dilution mass spectrometry (IDMS) reference method. The IDMS reference method has demonstrated comparability to the IFCC network (Liu et al., 2015).

Materials and Methods

Results

The Passing–Bablok regression between the results from the routine methods of six selected commercial methods and IDMS method are shown in Fig. 1. The regression equations are summarized in Table 2. Except for results from Cobas B 101 and the Variant II Turbo 2.0 methods, all the 95% confidence intervals of the regression lines overlapped, suggesting no statistically significant difference in results for the methods of the other four commercial methods. In the presence of heterozygous HbE, the Cobas B 101 method had positive bias at the range of concentrations examined (5.4–11.6%), while the Variant II Turbo 2.0 method had positive bias at concentrations up to approximately 9.5%.

The relative difference between heterozygous HbE and normal hemoglobin at HbA1c of 6% and 9% are summarized in Table 3. Results from the Cobas B 101 method (difference = +6.4% for HbA1c at 6%; difference = +7.4% for HbA1c at 9%) and the Variant II Turbo 2.0 method (difference = +9.9% for HbA1c at 6%) exceeded the a priori criteria for significance. The relative difference at 9% for the Variant II Turbo 2.0 method was +4.4%, which was within the a priori criteria for significance.

Discussion

The six commercial methods examined in this study included commonly used mainframe analyzer and point-of-care systems. Together, they represented the methods used by 36% of the College of American Pathologists proficiency testing survey participants (College of American Pathologists, 2020). This study examined the effect of HbE on HbA1c measurement across the clinically important measurement range by comparing against metrologically traceable reference values determined by a chemical metrology laboratory using IDMS measurements. Using this statistical approach, the effect of HbE on the methods can be examined relative to the normal hemoglobin across a range of concentration, which ameliorates the effects of inter-method calibrator bias.

The Cobas B 101 method showed significant positive bias across the measurement range investigated in the presence of heterozygous HbE. This finding extended the observation of significant positive bias in the Cobas B 101 method in a previous study that examined only a single patient sample (Lenters-Westra & Slingerland, 2014). This finding is contrary to prior published manufacturer’s claim, which indicated no significant analytical interference by heterozygous HbE (Food and Drug Administration, 2020). The discrepancy may be explained by the use of an acceptability criteria of ±10% relative difference (Food and Drug Administration, 2020), which is significantly wider than the clinical standards of ±6% required by the laboratory profession (College of American Pathologists, 2020). Laboratory practitioners should carefully communicate the significant positive interference in the Cobas B 101 method due to heterozygous HbE to the clinical user.

The significant positive bias observed in the Cobas B101 method is analytically unexpected. The Cobas B 101 method is a point-of-care assay that applies the principle of turbidimetric immunoinhibition. In general, immunoassays are thought to be more resilient against heterozygous HbE since the amino acid substitution is far from the N-terminus of the β chain where HbA1c glycation and antibody binding occur (Little et al., 2008). The cause of this discrepancy remains unclear at present.

On the other hand, the Variant II Turbo 2.0 method showed proportional positive bias in the presence of HbE. This finding is consistent with a previous report by Sthaneshwar et al. (2013). At lower HbA1c concentration, the positive relative bias is large (10%) and reduces at higher HbA1c concentrations. Interestingly, a previous publication reported no significant interference with HbE, despite having 1 out of 11 samples exceeding the acceptability criteria of ±10% relative difference in their internal evaluation (Food and Drug Administration, 2020). The positive interference from heterozygous HbE on the Variant II Turbo 2.0 method may be related to the incomplete cation-exchange chromatography separation. This potentially leads to overlapping HbE and HbA peaks that are sub-optimally resolved by the integration function of the algorithm.

These findings are of clinical concern for two reasons. Firstly, the acceptability criteria used by regulatory body is wider compared to the clinical requirements of routine laboratories. A 10% acceptability criteria implies that a relative difference of such magnitude is clinically acceptable. However, a 10% relative difference in HbA1c value can significantly alter the clinical interpretation of the results of individual patients. At a population level, the mean of HbA1c lies close to the clinical diagnostic threshold of 5.6% for pre-diabetes (Chai et al., 2017; American Diabetes Association, 2009). The significant bias will lead to over-diagnosis of diabetes, and the consequent unnecessary treatment and erroneous allocation of healthcare resources (Chai et al., 2017). As such, regulatory bodies should consider aligning their acceptability criteria with professional bodies to maintain the quality chain from manufacturer to laboratory and clinical end-users.

Secondly, the finding of significant bias meant that individual laboratory must exercise care when selecting a method. Additionally, in regions where a particular hemoglobin variant is prevalent, it may be prudent for the laboratory to verify the prior data, especially when its clinical requirements differ significantly from the manufacturer’s acceptability criteria. However, such verification exercise is often beyond the means of individual routine laboratories. This can be overcome by forming a collaborative network of laboratories to jointly evaluate the methods. When such exercise is performed, it is desirable to employ a reference laboratory method that is not influenced by the presence of hemoglobin variant in question.

Nevertheless, access to such reference laboratory method is limited. Increased reporting of hemoglobin variant interference in the literature will help laboratory practitioners make more informed decisions and assist manufacturers in improving their analytical methods. It may be necessary to periodically re-evaluate hemoglobin variant interference as a manufacturer may reformulate their reagents or modify their analytical methods, particularly for the CE-HPLC (Little & Rohlfing, 2017).

There is currently no consensus on how laboratory should manage hemoglobin variants that are incidentally detected during HbA1c measurements (Lewis et al., 2017). Some hemoglobin variants can interfere with HbA1c measurements leading to spurious results. Others may alter the lifespan of the red blood cells, leading to an altered glycation rate of the hemoglobin. An example of this is homozygous hemoglobin S that causes sickle cell disease. It is associated with shortened red blood cell lifespan, which causes lower HbA1c values relative to the ambience glucose. As such, it is advantageous to communicate the presence of hemoglobin variants to the clinicians to help them optimally interpret the HbA1c values. Additionally, such information also provides an opportunity for work up of significant hemoglobinopathy.

In general, it is desirable to avoid the use of methods that may be significantly interfered by the hemoglobin variant that is prevalent in the population served. This is particularly true for laboratory methods such as CE-HPLC, where the presence of hemoglobin variants may be inferred only by careful manual reading of the chromatograms by skilled technologists. It is desirable to avoid reporting an HbA1c result with known hemoglobin variant interference. When detected, alternative method not affected by the hemoglobin variant should be used to report the HbA1c instead. Alternate biomarker such as fructosamine may also be considered. However, such workflows require access to alternate instruments or esoteric biomarkers, which may not even be readily available to most laboratories.

The use of non-separative laboratory method such as the immunoassays for HbA1c measurement may mask an undetected interfering hemoglobin variant. At the same time, immunoassays are also liable to interfering antibodies (Rhea & Molinaro, 2014). Both of these may lead to erroneous reporting that can evade laboratory detection. The use of devices with such test principles in the point of care setting is considered an increased risk as laboratory supervision may be limited. As such clinicians should remain highly vigilant against clinically discordant HbA1c.

Conclusions

Four out of six commercial HbA1c laboratory methods examined in this study showed significant bias in measurement in the presence of heterozygous HbE when compared to an IDMS method. The impact of heterozygous HbE on HbA1c measurement may not be predicted from the assay principle, nor be adequately disclosed by the manufacturer. It is important for clinical laboratory to understand the impact of prevailing hemoglobin variant in their population during method selection and evaluation.

Supplemental Information