Clinical Chemistry 45:2

237–243 (1999) Hematology

Evaluation of a Capillary Electrophoresis Method

for Routine Determination of Hemoglobins

Downloaded from https://academic.oup.com/clinchem/article-abstract/45/2/237/5642837 by guest on 03 August 2020

A2 and F
Frédéric Cotton,1* Changying Lin,1 Bernard Fontaine,1 Béatrice Gulbis,1
Jacques Janssens,2 and Françoise Vertongen1

Hemoglobin A2 (Hb A2) and hemoglobin F (Hb F) are compared with those obtained with well-accepted chro-
important analytes in the diagnosis and follow up of Hb matographic techniques.
diseases. We evaluated a new capillary zone electro-
phoresis (CZE) kit for Hb A2 and Hb F measurements. Materials and Methods
The imprecision ranged from 3% to 6% for Hb A2 and specimens
Hb F at physiological and pathological concentrations. Samples were routine blood samples (EDTA anticoagu-
The method compared well with cation-exchange HPLC lated) sent to the laboratory for Hb analysis. Reference
for Hb A2 and Hb F and with anion-exchange chroma- values were established by the analysis of samples from
tography in microcolumns (MAEC), for Hb A2. Never- 107 healthy blood donors. Samples containing known Hb
theless, higher results were obtained [Hb A2 CZE (%) 5 variants (S, C, D-Punjab, E, O-Arab, and G-Philadelphia)
1.233 Hb A2 HPLC 2 0.2; Hb A2 CZE (%) 5 1.190 Hb detected by conventional electrophoretic and chromato-
A2 MAEC 1 0.1; Hb FCZE (%) 5 1.118 Hb FHPLC 1 0.4], and graphic methods and identified by peptide or DNA
new reference values had to be determined (Hb A2 analysis were also used. The study was in accordance
2.7–3.8%; Hb F <1.2%). Quantification of Hb A2 was not with the current revision of the Helsinki Declaration of
influenced by Hb S. Measurement of Hb F was accurate 1975.
and precise except at low concentrations in Hb AS
patients. This new CZE kit is rapid, precise, and reliable, analytical methods
and seems appropriate for use in clinical laboratories. Capillary electrophoresis. We tested the “Analis kit for Hb
A2 quantification and Hb Variants screening with the
Hemoglobin A2 (Hb A2)3 and hemoglobin F (Hb F) are Beckman P/ACETM” (Analis, Namur, Belgium). The re-
minor Hb components that are clinically important in the agents for this assay are patented and proprietary (8).
diagnosis and management of Hb disorders (1, 2). Com- Assays were performed following the manufacturer’s
monly used quantitative methods include microchro- instructions on a Beckman P/ACE 5500 System (Beckman
matographic techniques and automated HPLC for Hb A2 Inc.) equipped with a 25 mm (i.d.) 3 24 cm (effective
(3, 4), and alkaline denaturation, radial immunodiffusion, length 17 cm) fused-silica capillary thermostated at 26 °C.
and automated HPLC for Hb F (3). Capillary electro- The capillary was first rinsed for 0.5 min under a pressure
phoresis is an emerging technique in routine clinical of 138 kPa with the “initiator” solution (solution of a
chemistry, and analyses of Hb fractions and variants polycation in a malic acid buffer, pH 4.7). This process
based on this technique have arisen (5–7). We evaluated a coats the capillary wall with the polycation because a
new capillary zone electrophoresis (CZE) kit for Hb A2 large number of anchor points form between the negative
measurement and for its ability to estimate Hb F concen- charges of the silanol groups of the capillary glass and the
trations. The results obtained with this method were positive charges of the polycation. This rinse was fol-
lowed by a 1.0-min rinse with the “buffer” solution
(solution of a sulfated polyanion in arginine buffer, pH
1
Department of Clinical Chemistry, Hôpital Erasme, Université Libre de
Bruxelles, 808 route de Lennik, B1070 Brussels, Belgium.
2
Analis SA, 14 rue Dewez, B5000 Namur, Belgium.
3
*Author for correspondence. Fax 32 2 555 6655; e-mail fcotton@ulb.ac.be. Nonstandard abbreviations: Hb, hemoglobin; CZE, capillary zone elec-
Received September 8, 1998; revision accepted November 20, 1998. trophoresis; and MAEC, microcolumn anion-exchange chromatography.

237
238 Cotton et al.: Hemoglobins A2 and F by Capillary Electrophoresis

8.7), which adds a second layer of coating because a large 1.0%) to obtain various Hb A2 and Hb F concentrations
number of anchor points form between the polycation between these limits. Seven mixtures were prepared in
and the polyanion. This double layer of polymer provides duplicate and analyzed. Because accurate expected con-
a large number of negative sulfated charges facing the centrations of Hb A2 and Hb F were unknown (they are
inside of the capillary, creating a zeta potential higher related to hematocrit and the mean corpuscular Hb con-
than the original charge from the silanol groups. This centration of each sample), the results obtained for Hb F
double layer produces a strong, reproducible electroos- were plotted against those obtained for Hb A2 in each
motic flow and a decrease in protein adhesion to the mixture. Linear regression was applied, and results were
capillary wall. graphically interpreted.
After hemolysis with a provided reagent [10 mL of
whole blood 1 50 mL of hemolyzer (arginine buffer, pH Limit of quantification for Hb F. Packed red cells from a

Downloaded from https://academic.oup.com/clinchem/article-abstract/45/2/237/5642837 by guest on 03 August 2020

8.7)], the sample was injected for 3 s by pressure (3.5 kPa). blood displaying high Hb F concentrations (7.4% with
The buffer solution was then injected twice (1 s and 10 s CZE) were serially diluted with packed red cells from a
under pressure) to clean the outside of the capillary and to blood having undetectable Hb F concentrations. Six mix-
push the sample inside. Separation of Hbs according to tures displaying theoretical Hb F concentrations ranging
their mass-to-charge ratio was performed at 10.8 kV for from 0.2% to 7.4% were analyzed in triplicate by CZE. The
6.0 min in the same buffer solution. After separation, the limit of quantification was defined as the least amount of
coating was removed by first rinsing the capillary with Hb F that could be quantified reproducibly (CV ,10%).
the “conditioner” solution (0.2 mol/L NaOH) for 0.75 min
under pressure (138 kPa) and for another 0.75 min under Interference. The influence of Hb S on Hb A2 and Hb F
pressure and with current (200 mA) and then rinsing with quantification was tested by analyzing samples contain-
water for 0.5 min. Detection was made at 415 nm using an ing Hb S, using all three methods.
on-line ultraviolet-visible light detector. Electrophero-
grams were analyzed with the Beckman P/ACE Station comparison of methods
software, Ver. 1.0. Results were expressed as the corrected Fifty-seven samples containing no abnormal Hbs were
area percentage for each Hb fraction. Expected values selected and analyzed by all three methods.
provided by the manufacturer, obtained from a popula-
tion of 120 randomly selected ambulatory and hospital- reference values
ized subjects, were 2.1–3.3% for Hb A2 and 0.1–1.1% for The reference range for Hb A2 was determined by calcu-
Hb F. lating the 2.5–97.5 percentile interval obtained in the 107
blood donors. The 97.5 percentile for Hb F in the same
Automated HPLC. Automated cation-exchange HPLC population was considered as the high reference value.
(VariantTM Hemoglobin System Beta-Thalassemia Short
Program; Bio-Rad Laboratories) was used as the compar- statistical analysis
ison method for Hb A2 and Hb F, following the manufac- Comparisons of methods were made with the Passing and
turer’s instructions (3, 9). The expected values provided Bablok method using the Analyse-It for Microsoft Excel
by the manufacturer were 2.1–3.0% for Hb A2 and ,1.0% (Analyse-It Software Ltd.). Linear regression was used in
for Hb F. the linearity studies.

Anion-exchange chromatography. Microcolumn anion-ex- Results

change chromatography (MAEC; Quik-SepTM Hemoglo- The different patterns obtained with the CZE method for
bin A2 Test System; Isolab) was used as the comparison different patients are shown in Fig. 1. The separation of
method for Hb A2 only. The method was applied follow- Hb fractions was achieved within 6 min. The hemolyzer
ing the manufacturer’s instructions (4). The expected marker, Hb A2, Hb S, Hb F, and Hb A displayed migra-
values provided by the manufacturer were 1.5–3.0% for tion times of 4.59, 4.92, 4.99, 5.12, and 5.27 min, respec-
Hb A2. tively. These correspond to relative migration times (cal-
culated as the Hb fraction migration time divided by the
analytical performance hemolyzer marker migration time) of 1.00, 1.07, 1.09, 1.12,
Imprecision. Total imprecision for migration times and and 1.15, respectively.
concentrations of Hb fractions were determined by ana-
lyzing three samples once a day for 10 days and by imprecision
calculating the CVs. The total CVs for the migration times are summarized in
Table 1, and the total CVs for quantification of the Hb
Linearity. The linearity of Hb A2 and Hb F quantification fractions are summarized in Table 2. The imprecision was
was evaluated by assaying mixtures of different volumes always ,2% for migration times and ,1% for relative
of a normal cord blood (Hb A2 undetectable; Hb F, 84.8%) migration times. The CV for Hb F quantification was 5%
and of a blood from a healthy adult (Hb A2, 2.9%; Hb F, for concentrations ranging from 1.8% to 18.9%; the CV for
 Clinical Chemistry 45, No. 2, 1999 239

Downloaded from https://academic.oup.com/clinchem/article-abstract/45/2/237/5642837 by guest on 03 August 2020

Fig. 1. Typical electropherograms obtained by CZE.

(A), healthy patient; (B), patient with b-thalassemia trait; (C), S heterozygote; (D), S homozygote.
240 Cotton et al.: Hemoglobins A2 and F by Capillary Electrophoresis

Table 1. Imprecision for migration times.

Relative migration
Migration time time

Fraction Mean, min CV, % Mean CV, % n

Hemolyzer 4.59 1.3 1.00 10
Hb A2 4.92 1.4 1.07 0.1 10
Hb F 5.12 1.6 1.12 0.7 10
Hb A 5.27 1.5 1.15 0.2 10

Hb A2 was 3– 6% for concentrations ranging from 2.0% to

Downloaded from https://academic.oup.com/clinchem/article-abstract/45/2/237/5642837 by guest on 03 August 2020

5.6%.

linearity
The measured concentrations of Hb A2 and Hb F obtained
in mixtures prepared as explained in Materials and Meth-
ods are shown in Fig. 2 . The relationship between Hb A2
and Hb F seemed linear (r2 5 0.9934; Fig. 2A). The same
procedure applied to HPLC gave also good results (r2 5
0.9839; Fig. 2B).

limit of quantification for Hb F

The Hb F results in serial dilutions (given as the mean
concentration and the CV) were as follows: mean, 7.4%,
CV, 2%; mean, 3.4%, CV, 4%; mean, 1.7%, CV, 6%; mean,
0.9%, CV, 3%; undetectable; and undetectable. The value
of 0.9% was thus considered as the quantification limit for
Hb F.

comparison of methods
The comparisons for Hb A2 quantification are shown in
Fig. 3. All three methods gave similar results but signifi-
cant proportional biases were observed between CZE and
other methods, and a significant constant bias was ob-
served between CZE and MAEC as well as between
HPLC and MAEC (Table 3). The comparison of Hb F
concentrations obtained with HPLC and CZE is shown in
Fig. 4. Only samples with Hb F concentrations above the
CZE quantification limit were included (n 5 33). Results Fig. 2. Linearity of Hb A2 and Hb F quantification by CZE (A) and
of both methods were comparable despite constant and HPLC (B).
proportional biases (Table 3).

reference values interference

The distributions of Hb F and Hb A2 are shown in Fig. 5. We tested the effect of Hb S on Hb A2 and Hb F
The calculated reference values were 2.7–3.8% for Hb A2
quantification, according to the method of Suh et al. (10),
and ,1.2% for Hb F.
by assaying samples that contained [Hb AS (n 5 8) and
Hb SS (n 5 7)] or did not contain Hb S [Hb AA (n 5 57)]
Table 2. Imprecision for quantification. by all three methods. The biases in Hb A2 values obtained
Hb fraction Mean, % CV, % n between HPLC or CZE and MAEC were compared in
Hb A2 2.0 6 10 samples with or without Hb S (Table 4). Using the Student
2.4 3 10 t-test, we determined that the mean bias in Hb A2 values
3.1 6 10 between MAEC and CZE was not significantly higher in
5.6 6 10 Hb S than in Hb AA samples. On the other hand, the bias
Hb F 1.8 5 10 between MAEC and HPLC was significantly higher (P
2.3 5 10
,0.000001) in Hb S than in Hb AA samples. The bias in
18.9 5 10
Hb F values between CZE and HPLC was not statistically
 Clinical Chemistry 45, No. 2, 1999 241

Downloaded from https://academic.oup.com/clinchem/article-abstract/45/2/237/5642837 by guest on 03 August 2020

Fig. 4. Comparison of Hb F values obtained with HPLC and CZE.

Hb variants
Analysis of samples containing common Hb variants gave
the following results: Hb S was separated from Hb A and
Hb A2, but not from Hb D-Punjab and Hb G-Philadelphia;
Hb C and Hb O-Arab were separated from Hb A, but not
completely from Hb A2; and Hb E coeluted with Hb A2
(not shown, see Discussion).

Discussion
Measurement of Hb A2 and Hb F concentrations is of
great importance in the field of hemoglobinopathies (1, 2).
An increase in Hb A2 is one of the best biological markers
for the b-thalassemia trait, whereas low Hb A2 concentra-
tions are seen in the a-thalassemia trait and in iron
deficiency. Increases in Hb F can be observed in condi-
tions such as hereditary persistence of fetal Hb, b-thalas-
semia intermedia, b-thalassemia major, and some drug
Fig. 3. Comparison of Hb A2 quantification.
treatments. Hb F inhibits polymerization of Hb S and thus
(A), CZE and MAEC results (n 5 57); (B), CZE and HPLC results (n 5 57).
prevents painful crises in sickle cell disease. Hydroxyurea
is used to increase Hb F concentrations in sickle cell
patients. The monitoring of Hb F concentrations during
different in samples with Hb S and samples without Hb S. the follow up of these patients is thus mandatory.
Nevertheless, in S heterozygotes, measurement of low Hb Ion-exchange chromatography is considered as a ref-
F concentrations (#3%) was not possible, because of an erence method for Hb A2 and Hb F measurements (3, 4).
incomplete return to the baseline of the electropherogram The latter is also measured by alkaline denaturation and
between Hb A and Hb S (Fig. 1C). immunological methods (3). We compared a CZE method

Table 3. Comparison of Hb A2 and Hb F values.

Passing and Bablok comparison 95% CIa for slope 95% CI for intercept n
Hb A2 CZE (%) 5 1.233 Hb A2 HPLC 2 0.2 (1.156–1.386) (20.6 to 0.0) 57
Hb A2 CZE (%) 5 1.190 Hb A2 MAEC 1 0.1 (1.113–1.260) (20.1 to 0.3) 57
Hb A2 HPLC (%) 5 1.000 Hb A2 MAEC 1 0.2 (0.963–1.000) (0.2–0.3) 57
Hb FCZE (%) 5 1.118 Hb FHPLC 1 0.4 (1.045–1.210) (0.3–0.6) 33
a
CI, confidence interval.
242 Cotton et al.: Hemoglobins A2 and F by Capillary Electrophoresis

The results obtained with samples containing Hb vari-

ants showed that this kit might be suitable only for a
limited screening. For further characterization of the vari-
ant, this method should be combined with another (11).
Moreover, in the presence of some of these variants (e.g.,
Hb C, Hb E, and Hb O-Arab), quantification of Hb A2
becomes impossible because of incomplete resolution.
This drawback is seen in almost all methods (3, 7, 9) but
seems clinically insignificant.
The last 5 years have seen the appearance of an
increasing number of capillary electrophoresis applica-

Downloaded from https://academic.oup.com/clinchem/article-abstract/45/2/237/5642837 by guest on 03 August 2020

tions in clinical chemistry. In the field of Hb study, several
methods have been developed, mainly based on capillary
isoelectric focusing (5–7). Although highly effective, this
technique is less suitable for routine use than older
chromatographic methods because of unacceptable im-
precision at low concentrations of minor fractions, time
consumption, and higher costs (7). A recent study de-
scribed a homemade CZE method for Hb A2 quantifica-
Fig. 5. Distribution of Hb F (M) and Hb A2 (u) values (%) in 107 healthy
tion (6). Unfortunately, the CVs are quite high, and the
blood donors. method does not allow Hb F separation and measure-
ment. At the same time, a highly effective commercial
CZE method for Hb A1c determination has been released
developed for Hb A2 and Hb F quantification to com-
(5). It uses the same original procedure of dynamic
monly used ion-exchange chromatographic methods. This
coating, which maintains constant electroosmotic flow
new method compares favorably with the traditional
from run to run and from capillary to capillary and
methods. Nevertheless, higher values are obtained, which
decreases protein adhesion to the capillary wall. Hence,
are also reflected by the higher reference values provided
variations in electroosmotic flow are the main source of
by the manufacturer. The reference values that we deter-
imprecision encountered in CZE.
mined for Hb A2 are somewhat higher than those pro-
This new capillary electrophoresis method is precise,
vided with the kit. This can be explained by the fact that
quick, and comparable to well-accepted chromatographic
the manufacturer used blood from patients, potentially
methods for Hb A2 measurement without interference
including patients with iron deficiency.
from Hb S; thus, it is suitable for routine use. Moreover, it
The CZE method displays excellent CVs for both
allows precise estimation of Hb F. This method could be a
migration times and quantification, similar to those pub-
worthwhile tool in the clinical laboratory studying Hb
lished for chromatography (3, 4, 6, 7, 9).
diseases.
The presence of Hb S was shown to influence Hb A2
quantification with the HPLC method used in this study
(10). Spuriously increased concentrations are obtained,
We thank Analis for supplying the CZE reagents for this
probably because of Hb S adducts coeluting with Hb A2 in
study.
this system. Hb A2 quantification is not influenced by Hb
S in the CZE method, which provides accurate Hb A2
References
values. This allows the detection of the b-thalassemia trait
1. Weatherall DJ. The thalassemias. In: Beutler E, Lichtman MA,
and the differentiation of S-b°-thalassemia patients from S Coller BS, Kipps TJ, eds. Williams hematology. New York: McGraw-
homozygotes, for example. To the same extent, Hb F Hill, 1995:581– 615.
quantification is accurate in Hb SS patients, allowing the 2. Beutler E. The sickle cell disease and related disorders. In: Beutler
induction of Hb F by hydroxyurea in sickle cell patients to E, Lichtman MA, Coller BS, Kipps TJ, eds. Williams hematology.
be followed. New York: McGraw-Hill, 1995:616 –50.
3. Tan GB, Aw TC, Dunstan RA, Lee SH. Evaluation of high perfor-
mance liquid chromatography for routine estimation of haemoglo-
Table 4. Influence of Hb S on Hb A2 and Hb F bins A2 and F. J Clin Pathol 1993;46:852– 6.
4. Huisman THJ, Schroeder WA, Brodie AN, Mayson SM, Jakway J.
quantification.
Microchromatography of hemoglobins. III. A simplified procedure
Bias
for the determinations of hemoglobin A2. J Lab Clin Med 1975;
No Hb S (n 5 57) Hb S (n 5 15) P 86:700 –2.
Hb A2 HPLC 2 Hb A2 MAEC 0.17% 1.05% ,0.000001 5. Doelman CJA, Siebelder CWM, Nijhof WA, Weykamp CW, Janssens
J, Penders TJ. Capillary electrophoresis system for Hb A1c deter-
Hb A2 CZE 2 Hb A2 MAEC 0.62% 0.49% 0.11
minations evaluated. Clin Chem 1997;43:644 – 8.
Hb FCZE 2 Hb FHPLC 0.84% 2.1% 0.09
6. Jenkins MA, Hendy J, Smith IL. Evaluation of hemoglobin A2
 Clinical Chemistry 45, No. 2, 1999 243

quantification assay and hemoglobin variant screening by capillary 9. Papadea C, Cate JC. Identification and quantification of hemoglo-
electrophoresis. J Capillary Electrophor 1997;4:137– 43. bins A, F, S, and C by automated chromatography. Clin Chem
7. Mario N, Baudin B, Aussel C, Giboudeau J. Capillary isoelectric 1996;42:57– 63.
focusing and high-performance cation-exchange chromatography 10. Suh DD, Krauss JS, Bures K. Influence of hemoglobin S adducts
compared for qualitative and quantitative analysis of hemoglobin on hemoglobin A2 quantification by HPLC. Clin Chem 1996;42:
variants. Clin Chem 1997;43:2137– 42. 113– 4.
8. Janssens J, Chevigné R, inventors. Capillary electrophoresis 11. Lin C, Cotton F, Fontaine B, Gulbis B, Janssens J, Vertongen F.
method using initialized capillary and polyanion-containing buffer Capillary zone electrophoresis: an additional technique for identi-
and chemical kit therefore. US patent 5,611,903, 1997. fication of hemoglobin variants. Hemoglobin 1999;23:in press.

Downloaded from https://academic.oup.com/clinchem/article-abstract/45/2/237/5642837 by guest on 03 August 2020