Quantitation of hemoglobins using Sebia Capillarys-2 capillary electrophoresis (CE) for A1c: Comparison to results using CE for hemoglobins

Abstract

Background

Measurement of A1c using the Sebia Capillarys-2 capillary electrophoresis (A1c CE) involves relative quantitative measurements of peaks for hemoglobins A1c, A, A2. We examined correlation of A1c CE results with results of CE analysis for hemoglobins (Hb CE) for homozygous A and S-trait patients. We specifically examined whether abnormalities in A2 or the A/S ratio by A1c CE alone would reasonably be the basis for recommendation of red cell indices for evaluation of possible thalassemia.

Methods

Selection of patients was from results for A1c CE, exhibiting either a normal pattern or a pattern consistent with S-trait. We then examined correlation of results of quantitation for A, S and A2 between A1c CE and Hb CE.

Results

%A2 by A1c CE (y) had high correlation with %A2 by Hb CE (x): y = 0.88 x; r = 0.948. %A2 in S-trait patients was right-shifted in comparison to normals by 0.5%. For S-trait patients, the A/S ratio by A1c CE (y) had high correlation with the A/S ratio by Hb CE (x): y = 1.02 x; r = 0.995.

Conclusions

Given high correlation of results between A1c CE and Hb CE, patent elevation of A2 by A1c CE for either normal or S-trait patients is a reasonable basis for recommendation of red cell indices for evaluation of possible beta thalassemia. For S-trait patients, patent abnormality in the A/S ratio by A1c CE is a reasonable basis for recommendation of red cell indices for evaluation of possible alpha or beta thalassemia.

Highlights

• •Compared quantitation of hemoglobins by A1c CE to those by hemoglobin CE.
• •A2 by A1c CE has negative bias relative to hemoglobin CE.
• •A2 is higher in S-trait than for normals in both A1 CE and hemoglobin CE.
• •In S-trait, both A and S by A1c CE have negative bias relative to hemoglobin CE.
• •In S-trait, A/S by A1c CE has nearly exact correspondence to hemoglobin CE.

1. Introduction

Our laboratory uses the Sebia Capillarys-2 capillary electrophoresis system for measurement of % hemoglobin A1c (A1c CE) [1]. The final report is of %A1c that is directly comparable to results produced in the Diabetes Control and Complications Trial (DCCT) [2], in accordance with recommendations of the National Glycohemoglobin Standardization Program (NGSP) [3] and the International Federation for Clinical Chemistry (IFCC) [4]. In addition to the final report of calibrated %A1c, the A1c CE procedure produces relative percentages of quantitative peak integrals for hemoglobins A1c, A, A2 and other hemoglobins. Information content of chromatography by CE for A1c thus in principle extends beyond identification of the presence of abnormal hemoglobins, such as the presence of patently elevated hemoglobin A2, or a patently abnormal A/S ratio in S-trait. These are conditions which are related to possible presence of a thalassemia [5,6]. In this context, we examined the extent to which information content of A1c CE correlates with results of Sebia Capillarys-2 capillary electrophoresis measurement of hemoglobins (Hb CE) for homozygous A and S-trait patients.

2. Methods

2.1. Data acquisition and calculations

In order to avoid preselection for thalassemia, primary selection of patients was from among samples submitted for routine measurement of A1c. Such samples were stored refrigerated before testing. After testing by A1c CE, individual samples were selected for the study based on having either a normal chromatographic pattern by A1c CE consistent with homozygous A, or a pattern consistent with S-trait. Sample selection was across numerous days, with restriction to days when Hb CE could be performed immediately after selection, on the same day that A1c CE was performed. All samples used in the study were de-identified.

Given results for chromatography by both A1c CE and Hb CE for samples selected for the study, we examined distributions and correlation of results of quantitation for A, S and A2 from A1c CE compared to Hb CE as the gold-standard measurement. Chromatograms (A1c CE, Hb CE) were examined for total of 120 random, de-identified subjects (60 normals, 60 S-trait). Quantitative results for all chromatographic peaks (% of total) as produced by the Sebia software were recorded. For A1c CE, the software does not report the quantitation for the A1c peak, so as to avoid confusion with the calibrated reportable %A1c result that appears on each individual patient report. The chromatographic %A1c was thus obtained by subtraction from 100% of the sum of the % of all other peaks identified in the A1c CE chromatogram. All calculations and data analyses were conducted using programming in R.

2.2. Ethics

This study was classified as “Exempt” by review of the Institutional Review Board of Jefferson University, pursuant to federal regulations governing exempted protocol declarations. Specifically, patient consent for use of de-identified specimens for this study was not required.

3. Results

3.1. Primary data

From examination of routine A1c reports for A1c CE, we selected a total of 120 samples from among patients exhibiting either a normal hemoglobin profile (profile consistent with presence of hemoglobins A, A1c, and A2, n = 60) or S-trait (profile consistent with presence of hemoglobins A, S, A1c and A2, n = 60). These samples were forwarded for measurement of hemoglobins by Hb CE. Fig. 1 shows examples for A1c CE and Hb CE chromatograms for normal and S-trait subjects. The A1c CE profile for normal subjects usually identifies and quantifies four peaks: A1c, other hemoglobin A, A, and A2. The A1c CE profile for S-trait subjects usually identifies and quantifies five peaks: A1c, other hemoglobins, A, S, and A2. “Other hemoglobins” can include forms of glycation of A besides A1c, glycated forms of hemoglobins other than A (including glycated S), carbamylated or acetylated hemoglobin, or hemoglobin degradation products [7]. For both normal and S-trait profiles, the presence of a minor peak in the position of A2 is a requirement for the A1c CE software to perform the calculation of reported %A1c. For Hb CE, the profile for adult normal subjects usually identifies and quantifies only two peaks: A and A2; the profile for most adult S-trait subjects usually identifies and quantifies only three peaks: A, S, and A2.

Fig. 1.

Example CE chromatograms. A. A1c CE for a normal adult subject. B. Hb CE for a normal adult subject. C. A1c CE for an adult S-trait subject. D. Hb CE for an adult S-trait subject.

3.2. General properties of A1c data

Fig. 2A shows the distribution of A1c within the primary dataset (n = 120) compared to that for all-comers data for a 3-month interval of A1c measurements (n = 9007). The primary dataset A1c distribution was representative of the at-large patient population distribution. Fig. 2B shows a comparison of distributions of A1c segregated for A and S-trait subjects in our primary dataset. There was a slight right-ward shift of median A1c results among S-trait subjects (6.0%) compared to normal subjects (5.7%).

Fig. 2.

General properties of %A1c data. A. Distribution of %A1c in the primary dataset (points, n = 120) compared to distribution for an all-comers %A1c dataset over a 3-month interval (line, n = 9007). B. Comparison of distributions of %A1c for A (n = 60) and S-trait (n = 60) in the primary dataset.

3.3. Comparisons of A, S and A/S in S-trait as measured by A1c CE vs. Hb CE

Fig. 3A shows correlation of %A from A1c CE with %A from Hb CE among S-trait subjects, for which %A results from A1c CE demonstrate a negative bias with respect to %A results from Hb CE (across samples, %bias = −7.3 ± 3.1). Negative bias is an expected result, given that the A peak in Hb CE is inclusive of A1c, whereas the A peak in A1c CE does not include A1c. Similarly, as shown in Fig. 3B, correlation of %S between A1c CE and Hb CE for S-trait subjects also demonstrates a negative bias (across samples, %bias = −8.9 ± 3.2). Negative bias is expected because S in A1c CE does not include S1c (which runs in the position of “Other hemoglobin”) [7]. Despite differences in the distributions of biases for both A and S in A1c CE relative to Hb CE, the ratio A/S as obtained from A1c CE had high numerical correspondence with (was roughly identical to) values obtained from Hb CE (Fig. 3C).

Fig. 3.

Correlations of %A, %S and A/S measurements between A1c CE and Hb CE among S-trait subjects. A. Correlation of %A. B. Correlation of %S. C. Correlation of A/S ratio.

3.4. Comparisons of A2 as measured by A1c CE vs. Hb CE

Fig. 4, Fig. 5 show properties of comparisons of A2 measurements between A1c CE and Hb CE. For correlation of %A2 as measured by A1c CE and Hb CE, %A2 by A1c CE is consistently an underestimate of %A2 by Hb CE (Fig. 4). Distributions for %A2 segregated according to normal and S-trait subjects are distinct (by t-test, p < 0.0001) both for A1c CE (Fig. 5A) and Hb CE (Fig. 5B), with higher values for %A2 in S-trait. For both A1c CE and Hb CE, separation of means for %A2 between normals and S-trait is approximately +0.5%. The data for %A2 in Hb CE are consistent with numerous previous observations regarding elevation of A2 in S-trait [[8], [9], [10], [11], [12]].

Fig. 4.

Correlation of %A2 between A1c CE and Hb CE.

Fig. 5.

Distributions for %A2. A. Distributions for %A2 among normal and S-trait subjects by A1c CE. C. Distributions for %A2 among normal and S-trait subjects by Hb CE.

4. Discussion

Our laboratory uses CE for A1c quantitation for reasons of ease of use and rapid throughput. Use of methods for A1c measurement such as CE that can additionally detect the presence of hemoglobin variants is arguably advantageous to clinicians [13,14]. It was of interest to us to delineate whether there was reasonable certainty in the information content of A1c CE with respect to indications of thalassemia, namely, detection of elevated A2 and, for S-trait patients, detection of an abnormal A/S ratio. Findings here indicate that for both A2 and A/S ratio, correlation of A1c CE results with those obtained from Hb CE are such that A1c CE data are in fact useful for identification of cases of potential thalassemias.

Consideration of beta thalassemia from A1c CE data is noted in the Sebia training manual for A1c CE, stating that A2 > 3.0% may be suspect for beta-thalassemia. Our results for normal subjects corroborate that recommendation for normal A subjects, with the added observation that the cutoff should be greater (A2 > 3.5%) for S-trait subjects. Our results suggest further that elevation in the A/S ratio as obtained by A1c CE might also warrant consideration for alpha-thalassemia.

In our study database, we found no instances of patently elevated A2 to suggest potential beta-thalassemia. Among S-trait patients, however, there were three instances (5% of results) in which the A/S ratio appeared to be significantly elevated (A/S > 2.0) relative to values exhibited by the remainder of subjects. A/S > 2.0 (S < 33%), when accompanied by a decreased MCV, is consistent with values that are observed in alpha thalassemia in S-trait [15]. For a given A/S ratio = α, S as a percentage of (A + S) can be determined by %S = 100/(α +1).

A significantly decreased A/S ratio (A/S < 1) and increased %S are associated with beta+ thalassemia in S-trait [16,17]. However, an elevation in A2, as a hallmark feature of beta-thalassemia [18], would be already evident from CE data, barring the presence of a delta chain variant [19]. CE provides an advantage over high-performance liquid chromatography (HPLC) in this regard, as the A2 peak in HPLC typically exhibits an interference from a glycated form of S when S is present [20,21].

The positive shift of A2 in S-trait subjects compared to normal subjects has been documented previously [[8], [9], [10], [11], [12]]. This phenomenon appears to be due to the decreased affinity of beta-S for alpha subunits, leading to a rearrangement of the overall distribution of alpha subunits among A, S and A2 hemoglobins [8,22,23]. A/S > 1 as the common finding in S-trait reflects this same phenomenon. Use of a separate reference range for A2 in S-trait does not appear to be common in hemoglobin variant analysis. The shift is likely to be well known, however, among pathologists who routinely interpret quantitative hemoglobin profiles.

Overall, results indicate that information content of A1c CE is sufficient to make grounded recommendations for follow-up for evaluation of beta thalassemia based on %A2, or, for S-trait subjects, recommendations for follow-up for evaluation of alpha thalassemia based on the A/S ratio. A1c CE is much more common among hospital laboratories than is CE for hemoglobin identification (Hb CE). In such laboratories, monitoring of A2 and A/S in A1c CE to recognize patently abnormal patterns would thus be a form of surveillance for thalassemias. It would be at the discretion of the attending pathologist as to whether results on a test order for A1c should be examined for suggestion of thalassemia, and whether a recommendation for follow-up is warranted. Appropriate results comments might be as follows: for elevated A2, “Chromatographic profile shows apparent elevation of A2. Recommend evaluation of red cell indices to assess for possible beta thalassemia.”; for elevated A/S: “Chromatographic profile shows apparent elevation of A/S ratio. Recommend evaluation of red cell indices to assess for possible alpha thalassemia."

From the constraints of our approval for use of deidentified specimens, we were unable to further evaluate any specific sample findings with respect to a diagnosis of thalassemia. It is thus a shortcoming of the study that no thalassemias were definitively discovered in our sample cohort. It would be of value to examine in a future study the rate at which A1c CE results would result in diagnosis of thalassemias among patients not otherwise considered for evaluation.

CRediT author statement

Catherine Tucker: Conceptualization, Investigation, Data Curation, Writing - Reviewing and Editing. Douglas Stickle: Conceptualization, Project Administration, Methodology, Investigation, Data Curation, Formal Analysis, Software, Writing, Visualization.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Catherine Tucker: No competing interests. Douglas Stickle: Recipient of consulting fees from Sebia, Inc.

Data availability

Data will be made available on request.