Clinical Chemistry 43:1
34 39 (1997)
Cation-exchange HPLC evaluated for presumptive
identification of hemoglobin variants
Jean Riou, Christian Godart, Didier Hurtrel, Mireille Mathis, Catherine Bimet,
Josiane Bardakdjian-Michau, Claude Prehu, Henri Wajcman,* and
Frederic Galacteros
A battery of relatively simple tests allows the presump-
tive identification of hemoglobin (Hb) variants, making
unnecessary structural analysis by protein chemistry
methods or DNA sequencing. The primary step in this
strategy involves the use of a matrix of electrophoretic
mobilities obtained under various experimental condi-
tions. This leads to an unambiguous result in ;90% of
the cases. Additional tests are required to characterize
with more confidence the remaining 10%. We describe
here the use of cation-exchange HPLC on the Bio-Rad
Variant automated analyzer with the b Thalassemia
Short program. By comparing the elution time of 125
human Hb mutants, we found that some variants with
almost identical pI values or produced by the same type
of amino acid substitution displayed different elution
times. We present several examples in which use of the
HPLC profile helped establish the diagnosis.
INDEXING TERMS: sickle cell disease thalassemia electro-
phoresis isoelectric focusing globin chains
The use of cation-exchange high-performance liquid chro-
matography (CE-HPLC) to separate and quantify various
normal and abnormal hemoglobin (Hb) fractions has been
increasing [1 4].1 This method has also been proposed for
screening Hbs of clinical significance [3, 5, 6]. The Bio-
Rad Variant Hemoglobin Testing System (Bio-Rad Labs.,
Hercules, CA), a totally automated CE-HPLC instrument,
has been used in our laboratory for routine quantification
of HbA2 and HbF for .2 years. In a recent evaluation of
the Bio-Rad Variant b Thalassemia Short program for
Department of Biochemistry and INSERM U91, Hopital Henri Mondor,
94010 Creteil, France.
*Author for correspondence. Fax (33-1) 49 81 28 95; e-mail wajcman@
im3.inserm.fr.
1
Nonstandard abbreviations: Hb, hemoglobin; CE-HPLC, cation-exchange
HPLC; and IEF, isoelectric focusing.
Received May 20, 1996; revised and accepted August 27, 1996.
34
Molecular Pathology
and Genetics
quantification of Hbs A, S, C, and F, Papadea and Cate [7]
obtained results that were comparable with or exceeded
those of traditional methods. We think CE-HPLC is also a
valuable additional tool for presumptive identification of
Hb variants.
The strategy used in our laboratory to characterize rare
mutant Hbs primarily involves multiple electrophoreses
in several experimental conditions [8, 9]. This leads to an
unambiguous result in .90% of the cases. Nevertheless, a
few variants may behave similarly during electrophoresis
and share several common biochemical or functional
properties; for those cases, any additional test that would
allow discrimination will be of interest. Here we describe
the use of the Bio-Rad Variant analyzer with the b
Thalassemia Short program to discriminate Hb variants.
Working under the experimental conditions specified by
the manufacturer, with a few technical modifications, we
obtained highly reproducible retention times for repeated
assays with the same column and reagent batch. There-
fore, we evaluated the possibility of using a normalized
elution time as an additional variable in our approach to
the presumptive identification of Hb variants.
Materials and Methods
blood samples
Samples (Hb concentrations ;2 g/L) were obtained by
hemolyzing 20 mL of blood in 1 mL of a buffer of
potassium hydrogen phthalate (5 g/L) and potassium
cyanide (5 g/L). We applied 20 mL of hemolysate onto the
column for analysis.
The blood samples were obtained either from patients
who required an investigation of their Hb because of the
presence of an abnormal Hb component or from our
reference collection of rare Hb variants, stored in liquid
nitrogen. In all cases, the structural abnormality was
checked by protein structure analysis [10]. A flow dia-
gram showing the techniques used in our laboratory to
identify Hb variants is shown in Fig. 1.
 Clinical Chemistry 43, No. 1, 1997
Fig. 1. Flow diagram showing the techniques used to evaluate a Hb
variant.
When the electrophoretic data match for several Hbs, additional tests (e.g.,
reversed-phase HPLC, measurement of functional properties) may be performed
before a structural determination is made. IEF agarose, CE-HPLC, citrate agar
electrophoresis (CAE), and solubility tests are systematically performed on each
sample in our laboratorya strategy that aids detection of unusual variants,
such as a common Hb with more than one substitution within a given chain.
hplc analysis
The Bio-Rad Variant, a fully automated HPLC system,
uses double-wavelength detection (415 and 690 nm).
Several elution methods, including specific columns, buff-
ers, and softwares, are available from the manufacturer.
The b Thalassemia Short program, the most widely used
Variant program, has been designed to separate and
determine in 5 6 min the area percentages for HbA2 and
HbF and to provide qualitative determinations of abnor-
mal Hbs. Windows for retention times have been estab-
lished for presumptive identification of the most com-
monly occurring Hb variants. In the b Thalassemia Short
program, a 3 3 0.46 cm nonporous cation-exchange
column is eluted at a flow rate of 2 mL/min by a gradient
of two phosphate buffers that differ in pH and ionic
strength. The various Hb components give slight differ-
ences in elution time from one column to another and
from one reagent batch to another. The elution time for a
Hb component also varies slightly according to its con-
centration in the sample. We found that, for a given
column, a more accurate calibration than that proposed
by the manufacturer could be obtained by using HbA2 as
a reference. This Hb is present only between narrow
concentration limits, which prevents any significant mod-
ification of its elution time.
35
electrophoretic studies
The methods for electrophoretic analysis of Hbs included
electrophoresis on cellulose acetate at alkaline pH, citrate
agar electrophoresis, isoelectric focusing (IEF) of Hb, and
electrophoresis of globin chains in 6 mol/L urea in
Tris-EDTA buffer at pH 6.0 and 9.0 or in the presence of
Triton X-100 [7, 8]. In our laboratory, the electrophoretic
mobilities of .350 Hb variants are available in a data
bank under a format convenient for comparison. In this
approach, IEF is expressed in mm from HbA, and the
mobilities of the other electrophoretic systems are scaled
by using several slow and fast-moving Hbs as references.
Migration of the globin chains in urea is estimated from a
scale in which the values for normal a- and b-chain are
110 and 120, respectively.
Results
When the blood samples were prepared according to the
manufacturers instructions, we found that some Hb
components that were present in low amounts (;1% of
total Hb) were eluted together with HbF in the HbF
retention time frame. These additional fractions, most
probably reversibly glycated Hbs [11], disappeared when
we hemolyzed the samples with a potassium hydrogen
phthalate buffer instead of the Bio-Rad Hemolysis reagent
(Fig. 2). This procedure for sample preparation is the same
as that used for HPLC determination of HbA1c. Under
these experimental conditions, the elution profile of the
HbF window in the b Thalassemia Short program was
improved. It allowed an excellent agreement between
chromatographic measurement of HbF, starting at ,1%,
and resistance of the variant to alkaline denaturation.
Linear regression of a study involving 140 samples gave a
slope of 1.07 (r 5 0.963). Aged Hb specimens display some
degraded products, which are eluted in the P2 and P3
windows.
The most commonly occurring variants are HbS
(b6Glu3 Val), HbC (b6Glu3 Lys), HbE (b26Glu3 Lys),
and HbD-Punjab (b121Glu3 Gln). Presumptive identifi-
Fig. 2. Elution profile of a hemolysate prepared (A) according to the
manufacturers instructions and (B) with potassium hydrogen phtha-
latepotassium cyanide buffer to improve resolution of the HbF
window.
The small artifactual peak eluted at the position of HbF in A has disappeared in
B. This method for sample preparation allows analysis of samples that contain
a low amount of HbF or a weakly expressed mutant Hb eluting in the HbF window.
36 Riou et al.: Hb variants identified by cation-exchange HPLC

Table 1. Elution times and windows for normal Hb and some of the more frequently occurring variants.a
Elution time, min Elution time, min

Hb n Mean SD CV, % Min. Max. Windowb DRTc

HbF 200 1.109 0.039 3.5 1.02 1.23 0.21 22.35
(1.15) (1.00) (1.30) (0.30) (22.68)
P2d (1.45) (1.30) (1.60) (0.30)
P3d (1.75)
HbA 200 2.43 0.037 1.5 2.3 2.51 0.21 21.02
(2.60) (2.20) (3.30) (0.90) (21.23)
HbA2 200 3.456 0.019 0.54 3.4 3.5 0.10 0
(3.83) (3.68) (3.98) (0.30)
HbE 37 3.439 0.028 0.8 3.38 3.49 0.11 0
D Window (4.05) (3.98) (4.12) (0.14)
HbS 200 4.359 0.033 0.7 4.2 4.42 0.22 0.90
S Window (4.27) (4.12) (4.42) (0.30) (0.44)
HbC 100 5.017 0.011 0.2 4.98 5.04 0.06 1.56
C Window (5.03) (4.88) (5.18) (0.30) (1.20)
a
Values listed in parentheses are those given as examples by the manufacturer.
b
Width of elution window (difference between minimum and maximum elution times).
c
Mean difference in retention time as compared with HbA2.
d
Minor peaks associated with HbA, probably HbA1c and HbA1d.

cation of these abnormal Hbs is made by using the
retention time windows S-Window, D-Window, A2-
Window,and C-Window specified by the manufac-
turer. A study of the retention times of these common
variants, determined in .200 samples, is shown in Table
1. The actual elution time of these variants, if present in
the hemolysate at a similar range of concentration, was
highly reproducible in relation to that of HbA2 as a
calibrator. Moreover, these Hbs were found to be eluted
within a window of smaller size than that given as an
example by the manufacturer. Nevertheless, the elution
time of a given Hb component differs slightly according

Table 2. Difference in the retention time, relative to HbA2, of variants with pI near that of Hb S.
IEF,a mm, DRT,b min, relative to
Hb Substitution relative to HbA HbA2 6 0.05
a-Chain variants
Moabit a86Leu3Arg 28.5 10.76
Stanleyville a78Asn3Lys 28.5 10.39
Kokura a47Asp3Gly 28.5 10.63
G-Norfolk a85Asp3Asn 28.5 10.42
Inkster a85Asp3Val 28.7 10.26
Fort de France a45His3Arg 29.0 20.20
Hasharon a47Asp3His 29.0 11.02
Arya a47Asp3Asn 29.0 10.74
Savaria a49Ser3Arg 29.2 10.95
b-Chain variants
S b6Glu3Val 28.5 10.90
G-Makassar b6Glu3Ala 28.5 10.72
Ocho-Rios b52Asp3Ala 28.7 0.00
Osu-Christiansborg b52Asp3Asn 29.0 0.00
Cocody b21Asp3Asn 29.1 0.00
Karlskoga b21Asp3His 29.1 10.37
Kenitra b69Gly3Arg 19.2 0.00
G-Copenhagen b47Asp3Asn 29.2 10.14
G-Ferrara b57Asn3Lys 29.4 0.00
Poissy b56Gly3Arg, b86Ala3Pro 29.4 0.00
S-Antilles b6Glu3Val, b23Val3Ile 29.5 10.92
Maputo b47Asp3Tyr 29.5 10.24
a
Experimental values obtained with a column from which HbA2 elutes at 3.47 min.
b
Difference in retention time.
 Clinical Chemistry 43, No. 1, 1997 37

Fig. 3. Elution pattern of hemolysates from patients carrying

HbD-Punjab or Hb Korle Bu: (A) adult with HbA and HbD-
Punjab; (B) adult with HbA and Hb Korle Bu; (C) newborn
baby, composite heterozygote for HbS and Hb Korle Bu; and
(D) patient, composite heterozygote for HbS and HbD-Punjab.

to its concentration in the hemolysate. We have observed that varied from one case to another, as exemplified in
this for adult Hbs in neonatal screening, or when the Table 3 for those carrying a Glu3 Lys substitution. The
variant is unstable or associated with a thalassemic trait. most extreme situation is that of HbE and HbC, which
As shown in Table 2, some of the frequently encoun- elute near HbA2 and at 5.0 min, respectively. As Fig. 4
tered a- or b-chain variants, for which the pI is identical or shows, combining the IEF mobilities and the HPLC reten-
very close to that of HbS, may be easily recognized by tion times unambiguously identifies all the abnormal Hbs
CE-HPLC. This is important for differentiating the diag- belonging to this group.
nosis, already suggested by electrophoresis, between The elution times of several a- and b-chain Hb vari-
HbD-Punjab and other HbD-like variants, which are ants, in comparison with those of HbA, HbF, and HbA2,
frequently found in association with HbS but have oppo- and the position of the different windows are shown in
site clinical consequences. Hb Korle Bu, the most common Fig. 5. To show in the same diagram the elution times
of these, elutes at the same position as HbA2, whereas observed with different columns or reagent batches, one
HbD-Punjab elutes just after (Fig. 3). Hb variants with an must normalize the experimental data. A convenient way
identical amino acid substitution displayed elution times to do this is to convert the observed rough data into a
scale wherein the reference value is the mean elution time
of HbA2 given by the manufacturer.
Table 3. Difference in retention time, relative to HbA2, of
variants carrying a Glu3 Lys substitution.*
IEF,a mm, DRT,b min,
Hb Substitution relative to HbA relative to HbA2 6 0.05
a-Chain variants
Chad a23Glu3Lys 211.7 11.26
O-Padova a30Glu3Lys 211.7 11.19
O-Indonesia a116Glu3Lys 215.0 11.32
b-Chain variants
C b6Glu3Lys 216.0 11.56
Siriraj b7Glu3Lys 213.0 11.48
E-Saskatoon b22Glu3Lus 216.2 10.60
E b26Glu3Lys 215.1 0
Agenogi b90Glu3Lys 213.0 11.40
O-Arab b121Glu3Lys 215.0 11.37
a
Experimental values obtained with a column from which HbA2 elutes at 3.47
min.
b
Fig. 4. CE-HPLC and IEF combined analysis of the Hb variants carrying
Difference in retention time.
the Glu3 Lys amino acid substitution.
38 Riou et al.: Hb variants identified by cation-exchange HPLC

Fig. 5. Order of elution of several

normal and abnormal Hbs.
To eliminate between-column and between-
(reagent)batch variabilities, we normalize
the elution times in terms of the mean
elution value for HbA2 given by the manu-
facturer (3.8 min).

Some Hbs that are difficult to distinguish from HbA
electrophoretically, such as Hb Rainier, Alzette, or Putte-
lange, are clearly separated by HPLC. HbH, which it is
important to recognize in a-thalassemic patients, elutes
before the start of the standard integration program. Hb
Constant Spring, an elongated thalassemic a-chain vari-
ant seen frequently in Southeast Asian populations, elutes
in the C-window. The presence of an abnormal HbA2
eluting after the main abnormal peak is a convenient way
to distinguish between a- or b-chain variants that have
identical elution times. The retention times of .125 dif-
ferent Hb variants are now available in our data bank.
The sensitivity of the method is as efficient as that of
polyacrylamide gel IEF and allows recognition of an
abnormal adult Hb in a blood sample obtained in a
neonatal screening program for the main hemoglobinop-
athies. However, the concentration effect, which might
slightly modify the retention time, has to be taken into
account. Calibration of the elution time with a lesser
amount of abnormal Hb (,10%) is required.
Discussion
As of 1995, ;700 Hb mutations have been structurally
characterized [12]. Each year, our laboratory, which
 Clinical Chemistry 43, No. 1, 1997
works as a reference center for abnormal Hbs, character-
izes several hundreds of samples thought to contain a rare
Hb variant. To avoid performing a complete structural
determination by protein chemistry methods or by DNA
sequencing for each unusual samplewhich usually con-
firms a rare but already described variantwe have
developed a multicriteria approach to reach a presump-
tive diagnosis. This strategy is based primarily on electro-
phoretic characterization of the variant in various exper-
imental conditions by comparison with a data bank that
contains references for the .350 variants we have ob-
served. Including additional fast, reliable, and relatively
simple biochemical methods may improve the electro-
phoresis approach. Recently, we proposed the use of
reversed-phase perfusion chromatography for chain anal-
ysis as a complementary tool for the characterization of
Hb variants [13]. Additional considerations to take into
account include the patients ethnic origin [14] and clinical
presentation and the functional properties of the erythro-
cytes or blood lysate. Measuring the molecular mass of
the variant by electrospray mass spectrometry may indi-
cate which kind of amino acid exchange has occurred [15].
Taken together, all of these informative results lead to
a presumptive diagnosis that, even in the more exotic
cases, is almost always found to be accurate. This was
indeed the case for .100 cases, where we also checked the
defect at the molecular level. Including another simple
method such as the well-standardized CE-HPLC de-
scribed here will certainly decrease the number of incom-
plete matches between the variant under investigation
and a reference from the data bank and hence will
minimize the need to work up an identification at the
molecular level.
We acknowledge the support of the Bio-Rad Co. in this
work.
References
1. Wilson JB, Headlee ME, Huisman THJ. A new high-performance
liquid chromatographic procedure for the separation and quanti-
tation of various hemoglobin variants in adults and newborn
babies. J Lab Clin Med 1983;102:174 85.
39
2. Kutlar A, Kutlar F, Wilson JB, Headlee ME, Huisman THJ. Quanti-
tation of hemoglobin components by high-performance cation-
exchange liquid chromatography: its use in diagnosis and in the
assessment of cellular distribution of hemoglobin variants. Am J
Hematol 1984;17:39 53.
3. Rogers BB, Wessels RA, Ou CN, Buffone GJ. High-performance
liquid chromatography in the diagnosis of hemoglobinopathies and
thalassemias. Am J Clin Pathol 1985;84:671 4.
4. Samperi P, Mancuso GR, Dibenedetto SP, Di Cataldo A, Ragusa R,
Schiliro G. High performance liquid chromatography (HPLC): a
simple method to quantify HbC, O-Arab, Agenogi, and F. Clin Lab
Haematol 1990;13:169 75.
5. Shapira E, Miller VL, Miller JB, Qu Y. Sickle cell screening using a
rapid automated HPLC system. Clin Chim Acta 1989;182:301 8.
6. Ou CN, Rognerud CL. Rapid analysis of hemoglobin variants by
cation-exchange HPLC. Clin Chem 1993;39:820 4.
7. Papadea C, Cate JC. Identification and quantification of hemoglo-
bins A, F, S, and C by automated chromatography. Clin Chem
1996;42:57 63.
8. Schneider RG, Barwick RC. Electrophoretic mobilities of mutant
hemoglobins and mutant globin chains. In: Schmidt RM, Fairbanks
VF, eds. CRC handbook series in clinical laboratory science.
Section I: Hematology, Vol. 4. Boca Raton, FL: CRC Press,
1986:12539.
9. Lacombe C, Riou J, Godard C, Rosa J, Galacteros F. Characteriza-
tion approach of silent beta-chain hemoglobin variants. Acta
Haematol 1986;78:119 22.
10. Wajcman H, Bardakdjian J, Ducrocq R. Structural characterization
of abnormal hemoglobins from dried blood specimens in a neo-
natal screening program. Ann Biol Clin 1993;50:86770.
11. Tan GB, Aw TC, Dunstan RA, Lee SH. Evaluation of high-perfor-
mance liquid chromatography for routine estimation of haemoglo-
bins A2 and F. J Clin Pathol 1993;46:852 6.
12. International Hemoglobin Information Center. Variant list. Hemo-
globin 1995;19:39 124.
13. Wajcman H, Ducrocq R, Riou J, Mathis M, Godart C, Prehu C,
Galacteros F. Perfusion chromatography on reversed-phase col-
umn allows fast analysis of human globin chains. Anal Biochem
1996;237:80 7.
14. Chami B, Braconnier F, Riou J, Bardakdjian-Michau J, Prehu C,
Blouquit Y, et al. Geographic distribution of 119 alleles of the a
and b globin genes detected in 432 French Caucasian carriers of
haemoglobin variant. Ann Genet 1995;38:206 16.
15. Lacombe C, Prome D, Blouquit Y, Bardakdjian J, Arous N, Bost M,
et al. New results of hemoglobin variant structure determinations
by fast atom bombardment mass spectrometry. Hemoglobin
1990;14:529 48.