Analytical goals for the determination of HbA₂

Introduction

The measurement of HbA₂ is primarily used for the identification of β-thalassemia carriers with the aim to prevent the severe transfusion-dependent thalassemia disorders. Detection of carriers is easily performed in the classical phenotypes characterized by increased HbA₂ values (4.0%–6.0%) together with reduced mean cell volume (MCV) and mean cell hemoglobin (MCH). However, the existence of atypical β-thalassemia carriers with near normal hematological indices and borderline HbA₂ values (3.3%–3.8%) can make the diagnosis more difficult [1, 2]. Borderline HbA₂ levels are generally due to the presence of mild β⁺ mutations or may be the consequence of a number of genetic factors, such as the interaction with other molecular defects (α- and δ-allele), β-promoter mutations, triplication of the α-globin gene (ααα/αα) and KLF1 gene mutations [3, 4]. In addition, some acquired conditions (iron deficiency, megaloblastic anemia, hyperthyroidism) can affect the HbA₂ levels, contributing to the risk of misinterpretation of HbA₂ results.

Since an incorrect diagnosis may translate into major clinical implications for affected individuals, it is very important that a reliable HbA₂ measurement with a high degree of both reproducibility and accuracy is provided to clinicians [5]. To this regard, data obtained in a large external quality assessment scheme [UK NEQAS (H)] and results from a recent inter-laboratory study showed a significant bias between different routine methods for HbA₂ measurement [6, 7]. However, it is rather difficult to evaluate the actual analytical quality of HbA₂ testing since at present there is no reference measurement procedure approved for this analyte and no defined specifications for allowable analytical bias and imprecision have been defined.

With regard to the analytic performance goals, there are at least six different ways for establishing them (regulations and EQAS, biologic variation, survey of clinicians, effects on medical decisions, patterns for follow-up tests, formal decision models), as clearly reviewed [8, 9].

The aim of this work is to define some goals for the determination of HbA₂ in human blood, and to relate them to the metrological reference system for HbA₂ under development by the IFCC Working Group on HbA₂ standardization.

Approach I: goals based on biologic variation

Experimental work has been performed on 17 apparently healthy subjects (9 men and 8 women, aged 26–52 years) by analyzing their HbA₂ values every 2 weeks, for 2 months, as previously described [10]. The subjects were enrolled among the laboratory staff and gave their written consent to be tested. The blood samples were collected after overnight fasting and without any morning exercise. Venous blood was collected by the same experienced phlebotomist between 9:00 and 9:30 with subjects in the sitting position for 1–5 min, with minimal stasis using a 20 G straight needle directly into 3 mL siliconized vacuum tubes containing K₂ EDTA (Terumo Europe NV, Belgium). The blood specimens were immediately aliquoted and stored at –80°C until analyzed all in the same run, in order to minimize the analytical variation. On each day of blood collection a whole blood cell count was performed on an aliquot of the EDTA-blood specimen by means of an automated Advia 2120 analyzer (Siemens Healthcare Diagnostics).

The determination of HbA₂ was performed by an automated HPLC system (Bio-Rad Variant II, dual kit, Bio-Rad Laboratories, Segrate, Italy) using the calibrators provided by the manufacturer. Each blood specimen was analyzed in duplicate in order to evaluate the analytical imprecision. The whole set of blood samples was analyzed over four different runs, taking care that all the specimens derived from a particular subject were analyzed within the same run. Finally, the results obtained by the HPLC analysis were adjusted in order to express each result with two decimal places in order to achieve the minimum number of significant numbers. With regard to the statistical treatment of the data, an ANOVA analysis was performed, in order to extrapolate, from the overall variance (σ²_tot), the biologic intra-individual component (σ²_I) and the biologic between-individuals component (σ²_G), according to the formula:

σ²_tot = σ²_anal + σ²_I + σ²_G

as already described with regards to a similar study performed to obtain the analytical goals for glycated hemoglobin [11]. The analytical variance (σ²_anal) was obtained, as reported above, from the analysis of the differences in the duplicates.

All the data obtained by this investigation were included in the evaluation, since no abnormalities in the other red cell parameters (total hemoglobin, MCV, MCH and hematocrit) were found. The results of such evaluations are reported in Figure 1, where the mean HbA₂ values for each subject is reported, together with the minimum to maximum range. Just by looking at the graph, it is clearly evident how much the variability in the measured HbA₂ in each subject was smaller than that measured between subjects. Moreover, no statistical significant differences in the HbA₂ values were found between different genders (women, mean±SD: 2.63±0.26%; men: 2.74±0.16%; t-test p=0.265). From the data obtained so far, the analytical goals for imprecision, bias and total error were calculated (Table 1).

Figure 1

Individual mean and absolute range of HbA₂ values measured in 17 healthy subjects during a 2 month period.

Women and men are represented as open circles and closed squares, respectively. Reprinted from reference 9 with permission of the Editor.

Approach II: goals based on clinical needs and opinion of experts

A survey has been performed on a selected group of 15 clinicians dealing mostly with patients affected by thalassemic syndromes and various hemoglobin disorders, mostly sickle cell anemia (team A) and by an international team of more than 100 laboratory professionals (team B). The survey was performed during two recent international meetings (Sebia advisory board meeting, 25 May 2012, Sebia, Lisses (France), and VI international symposium on hemoglobins [12]), and the results were collected by email (team A) and by televoting during the conference (team B). They were asked to answer the questions related to three clinical cases, as illustrated in Table 2.

The first case history reports on an HbA₂ test repeated over a short time (typically 1 week) from the preceding result, in order to confirm a result of difficult interpretation, such for the so-called borderline values. The second case describes a situation where another repeated test is performed, but over a longer period of time during which clinical action was undertaken (i.e., iron therapy). The third case describes a somewhat less likely situation, where the test is performed four times over 3 months without any apparent change either in the pre-analytical than in the analytical phase. None of the cases were related to any real physical patient.

Responses were collected from 10 clinicians from team A and from 83 laboratory professionals from team B. A summary of the responses is shown in Table 3, where, together with the answers, we have tabulated the mean values and the absolute HbA₂ changes about whom we were asking opinions. The corresponding allowable total error, expressed from the hypothetical mean HbA₂ values and their absolute changes, are displayed in the fourth column of Table 3.

As can be seen, there is a quite good agreement between the laboratory professionals and the clinicians, although a rigorous comparison cannot be done, because the number of answers collected so far was very unbalanced, with team A in a much smaller quantity in respect to team B.

Finally, another approach based on clinical needs has been already reported by our group [1] and is just briefly outlined here. Considering a subject with a true HbA₂ value of 3.6%, the measurement error should not exceed, in relative units, of more than 7.0% in order to exclude the possibility of misclassifying him as a β-thalassemia carrier (HbA₂ ≥3.8%) or as a non-β-thalassemia subject (HbA₂ <3.3%).

Conclusions

The quantitation of HbA₂ is primarily used to diagnose, or exclude the diagnosis of β-thalassemia trait in pregnant women and other patients, or to assist with the elucidation of microcytic, hypochromic anemia [5]. Historically, when only a small number of subjects were tested there appeared to be a clear difference in HbA₂ levels between carriers and non-carriers, but recent work has shown that interpretation of results is more complicated than initially thought. The interpretation of the numerical analytical results is made more complicated for two main reasons: there are more than 200 mutations causing β-thalassemia and the HbA₂ is raised by different amounts according to the mutation [13], and second that some medical conditions affect the percentage of HbA₂ found. An HbA₂ level of 3.5% is often given as a cut-off point to categorize carriers from non-carriers [14].

To our knowledge, up until now only a few figures on serial measurements have been published to show how much biologic variation there is [9], and as it happens this is very small, as would be expected for any biochemical parameter mainly regulated by genetic mechanisms. Also, given the fact that nobody would make a diagnosis only on the basis of the HbA₂ value, without knowing other important information (clinical picture, family history, whole blood cell count, iron metabolism etc.), we felt it was important to determine values considered appropriate for cut-off points and variability of results which will demonstrate improvements necessary in the analytical goals for manufacturers of automated analyzers. To this regard, we have summarized our findings in Table 4, where a minimal goal for the total error could be set at 6.9%, a kind of a compromise between the information gained from the study on biological variation, clinical needs and opinion of experts.

Of course, the opinion of experts and clinical need may change over time, as a possible result of the improvements in the analytical aspects. Up until now, there are only few methods able to achieve an overall imprecision of <1%, but the scenario is continuously evolving, so it would be advisable to revise such goals in the future.

Moreover, we hope that in a couple of years the complete IFCC reference system for HbA₂ is realised, so that all manufacturers are able to calibrate by means of a common commutable certified reference material with an assigned value derived from the primary reference measurement procedure. A layout of the system this Working Group is developing is shown in Figure 2. This is a new version of that already presented some years ago [3], and the main new issue concerns the reference measurement procedure and the primary reference materials. Both these issues are under development, following the successful approach already developed within the sphere of glycated hemoglobin [15]. The data on the analytical goals for HbA₂, derived by means of what presented in this document, could be regarded as an additional outcome of the aims the WG is trying to achieve.

Figure 2

Reference system and traceability chain for the determination of HbA₂.