Review Article

Detection and quantitation of normal and variant haemoglobins:

an analytical review
Barbara J Wild1 and Barbara J Bain2

Abstract
Addresses As well as six ‘normal’ haemoglobins that occur at various stages of development,
1
Department of Haematological Medicine more than 800 abnormal or variant haemoglobins have been described. Many of
King’s College Hospital
these variant haemoglobins have no significant clinical consequences apart from
London SE5 9RS, UK
2
Department of Haematology causing confusion to clinicians and in laboratories; however, some of the variant
St Mary’s Hospital Campus of Imperial haemoglobins result in major morbidity or mortality. The laboratory challenge is to
College School of Medicine detect these clinically significant haemoglobins and to identify them with sufficient
London W2 1NY, UK accuracy for clinical purposes, as well as to quantitate both these and the ‘normal’
haemoglobins. The techniques used to detect and quantitate these haemoglobins in
Correspondence routine service laboratories are discussed in detail. Methods used by referral
Dr Barbara Wild
laboratories, such as mass spectrometry and DNA analysis, are briefly discussed.
E-mail: Barbara.wild@kcl.ac.uk
Haemoglobin analysis is most often undertaken as part of neonatal, antenatal or
This article was prepared at the invitation of pre-anaesthetic screening; these programmes are reviewed, together with possible
the Analytical Investigations Standing changes to neonatal screening and antenatal screening that may occur as part of the
Committee of the Association of Clinical NHS National Plan.
Biochemists.
Ann Clin Biochem 2004; 41: 355–369

Introduction screening since the precise identity will not usually

have been determined at this stage. For this reason, it
Six normal and more than 800 abnormal haemoglo- is incumbent upon the laboratory not only to design
bins, now usually called variant haemoglobins, have screening programmes that detect the appropriate
been described to date.1,2 The ‘normal’ haemoglobins variants reliably, but also to ascertain which patients
are de¢ned as those that usually occur during need to proceed to further investigation.
embryonic, fetal, neonatal and adult life and the
structure of the ‘abnormal’or variant haemoglobins is Normal and variant haemoglobins
compared to them. Mutations resulting in abnormal The six normal haemoglobins are two adult (HbA and
haemoglobin structure may a¡ect the function of the HbA2), one fetal (HbF) and three embryonic haemo-
haemoglobin molecule; if this occurs, there may be globins (Gower1, Gower 2 and Portland). Most clinical
signi¢cant clinical implications for an individual who problems relate to structural abnormalities of the
has such a haemoglobin. Investigations for the major adult haemoglobin, HbA. Occasionally, clinical
presence of a haemoglobin variant are undertaken for conditions may result from disturbance of HbF
one of two reasons: either as part of a screening production, but in most situations these conditions are
programme, where persons are tested because they benign. The minor adult haemoglobin, HbA2 is present
may be at risk of having a particular clinically impor- in such small amounts that physiologically it is
tant haemoglobin variant; or they may be tested to immaterial whether it is present or not, but its quanti-
elucidate the cause of speci¢c clinical symptoms or tation is essential for the diagnosis of b-thalassaemia
haematological abnormality. The analytical proce- trait, in which condition it is characteristically
dures used for haemoglobinopathy screening will elevated.
detect, but not identify, both the common clinically Normal adult haemoglobins are composed of tetra-
important haemoglobin variants and many variants mers of four globin polypeptide chains each with a
that have no clinical signi¢cance. It is therefore very haem group attached. In normal adults, the principal
di⁄cult for clinicians to give de¢nitive clinical or haemoglobin, HbA (comprising approximately 97%),
genetic advice about a variant that is found during is composed of two a and two b chains, each enclosing

& 2004 The Association of Clinical Biochemists 355

356 Wild and Bain

a haem group. The remainder consists of approxi- variant haemoglobin. If its presence is not already
mately 2-3% HbA2, which has two a and two d chains, known the patient should be referred to a haematolo-
and less than 1% HbF, which has two a and two g gist so that its clinical signi¢cance can be determined
chains. In the neonate, the major haemoglobin is HbF (using the approach given in this review).
(approximately 80-90%), with a variable amount of
HbA (approximately 10-20%). Embryonic haemoglo-
bins with a-like and b-like chains (called z and e, Clinical significance of haemoglobin variants
respectively) are present in embryonic and fetal life but Many of the haemoglobin variants described are of no
will not be covered in this review. The switch from g clinical signi¢cance and it is important to be able to
chain synthesis (which predominates in the fetus and determine which haemoglobins are likely to have
neonate) to b chain synthesis (which predominates in clinical or genetic implications and which are not.
children and adults), together with the production of d Clinical signi¢cance arising from the variant haemo-
chains, is usually virtually complete by 12 months globins is attributable to either an abnormality of
after birth. Hence, at this time, the infant will have solubility or stability or capacity to deliver or carry
developed mature levels of HbA2 and HbF. oxygen. These manifest as either the sickle cell
a and z chains are encoded by a cluster of genes on disorders,5 as a b-thalassaemia phenotype resulting
chromosome 16, whereas b, g, d and e chains are from the presence of Hb Lepore (a hybrid haemoglobin
encoded by a cluster of genes on chromosome 11.3 of b and d globin chains) or from the compound
Variant haemoglobins usually result from a single heterozygote of HbE with b-thalassaemia,3 as mild or
mutation in either an a, b, g or d gene, leading to the severe haemolysis resulting from HbC or unstable
substitution of a single amino acid; occasionally, haemoglobins,6,7 as cyanosis due to the presence of
however, they result from more than one mutation in HbMs 3,8 or as erythrocytosis or anaemia as a result of
the same globin chain and occasionally from a cross- an alteration in a⁄nity for oxygen.9 Of these con-
over leading to a hybrid of two globin chains. The ditions, the sickling disorders are by far the most
globin chains present in the normal and some of the common, followed by the compound heterozygote
common variant haemoglobins are shown in Table 1. condition HbE b-thalassaemia, which is extremely
Normal and variant haemoglobins can undergo post- common in South-East Asia, whereas the remaining
synthetic modi¢cations such as glycation. For groups are relatively rare. People who are carriers of
example, column chromatography can separate HbA the sickle gene are asymptomatic, whereas homo-
into a major peak (HbA0), and two smaller peaks zygotes, and some compound heterozygotes such as
(glycated HbAI, and HbAIII, in which glutathione is HbSC, su¡er signi¢cantly in terms of morbidity and
bound onto the cysteine at b93 as an ageing phenom- mortality.5 The collective term ‘sickle cell disease’ is
enon). HbAI can be further subdivided into HbAIa-e. used to encompass all those genotypes in which clini-
Glycated haemoglobin analysis has been reviewed cally important sickling may occur. Compound
previously 4 and will be referred to in this review only heterozygosity of Hb Lepore with b-thalassaemia is
when it impinges on the detection or quantitation of uncommon and homozygosity for Hb Lepore is even
variant haemoglobins. If a variant haemoglobin is rarer, but both of these genotypes can lead to a
detected as part of glycated haemoglobin assay the phenotype indistinguishable from b-thalassaemia
clinician should be informed of the presence of the major. HbE b-thalassaemia has a variable phenotype
but usually leads to a thalassaemia intermedia
phenotype which may have considerable morbidity.
Table 1. Globin chain constitution of normal and some variant
Heterozygotes for some of the mildly unstable haemo-
haemoglobins
globins are asymptomatic, but other unstable variants
Haemoglobin Globin chain constitution cause severe haemolytic anaemia, which may neces-
sitate treatment by blood transfusions and/or sple-
A (adult) a2 b2 nectomy even in the heterozygous state. It is thought
A2 (minor adult) a2 d2 that some unstable haemoglobins are so severe that
F (fetal) a2 g 2 the homozygous state is incompatible with life;
S (sickle) a2 bS2 indeed, reports of homozygosity for unstable haemo-
globins are very few. Some hyper-unstable variants
C a2 bC2 are so unstable that no variant is detectable in the
E a2 bE2 peripheral blood until after splenectomy, and these
GPhiladelphia aG2 b2 can present phenotypically like thalassaemia trait due
OArab a2 bO2 to reduced availability of one of the globin chains.
HbMs (haemoglobinopathies) are a rare group of
DPunjab a2 bD2
disorders in which the variant haemoglobin will

Ann Clin Biochem 2004; 41: 355–369

 Detection and quantitation of normal and variant haemoglobins 357

readily oxidize to methaemoglobin; the carriers are either selective (also called targeted), where selection
consequently cyanotic but are otherwise quite well. is based on the subject’s ethnic origin, or universal,
where all individuals are tested. It has been recom-
Detection of haemoglobin variants: mended that, in areas where the ethnic minority
haemoglobinopathy screening programmes component exceeds 15%, universal screening
There has been a considerable increase in immigration programmes should be practised.10 However, universal
into the UK over the last 50 years and the numbers of neonatal screening is undertaken in most of the USA
subjects carrying genes for some of the clinically and is likely to be recommended for introduction
signi¢cant haemoglobin variants such as sickle throughout the UK by 2004/05.
haemoglobin have increased enormously. Interest-
ingly, there has also been an increase in the number of Laboratory investigation of haemoglobin variants
di¡erent variants detected and subsequently described suspected from clinical symptoms or the presence of
as a consequence of the increase in HbA1c monitoring haematological abnormality
using automated high-performance liquid chromato- In addition to haemoglobinopathy screening
graphy (HPLC). Due to the increase in the population programmes, it is important for the laboratory to be
at risk of having a haemoglobinopathy, haemoglobi- able to diagnose (or refer for diagnosis) patients who
nopathy screening programmes have been widely are su¡ering from a haematological abnormality and/
introduced. These are broadly divided into three or clinical symptoms that may be the result of a
categories: pre-anaesthetic, antenatal and neonatal haemoglobin variant. As previously described, these
screening. The objectives of the di¡erent categories of include the sickling disorders, the unstable haemo-
screening are quite di¡erent. Pre-anaesthetic globins, the altered a⁄nity haemoglobins and the
screening is undertaken to detect the presence of HbMs. In attempting to diagnose variants other than
sickle haemoglobin and, if present, to determine HbS it is essential to be aware that the abnormality
whether the subject is a heterozygote, homozygote or may be electrophoretically or chromatographically
compound heterozygote in order that appropriate silent and thus the application of the common, simple
clinical management may be delivered during and screening procedures may be inadequate. In these
after general anaesthesia. Antenatal screening is cases it is necessary to apply speci¢c techniques that
undertaken in order to predict any genetic risk to the will identify defective functional properties of the
fetus of sickle cell disease, b-thalassaemia major or the abnormal haemoglobin rather than a di¡erence in
severe form of a-thalassaemia (Hb Barts hydrops charge or solubility of the molecule. These situations
fetalis) and also to ensure appropriate clinical require tests such as those for haemoglobin instability,
management of women with sickle cell disease during investigation of oxygen a⁄nity or electrophoresis
pregnancy. This necessitates screening for the under di¡erent conditions. However, if precise identi-
common clinically important variants HbS, HbC, ¢cation is required, then either detailed protein
HbDPunjab, HbE, Hb Lepore and HbOArab; for the analysis, by mass spectrometry, or DNA sequencing is
thalassaemia syndromes b-thalassaemia trait and a0 - needed. The clinical indications for testing for a
thalassaemia trait; and for the rarer conditions db- variant haemoglobin are shown in Box 1.
thalassaemia trait and hereditary persistence of fetal
haemoglobin (HPFH). If the fetus is at risk of a major Box 1. Clinical indications for testing for a variant haemoglobin
haemoglobinopathy then the parents are usually
o¡ered prenatal diagnosis which can be undertaken at
. Pre-anaesthetic screening (for sickle cell haemoglobin)
11 weeks of gestation, or at any time after that. . Antenatal screening to permit genetic counselling and
Antenatal screening, including partner testing and adequate management of pregnant women
appropriate counselling, should be completed before . Neonatal screening to permit early diagnosis and
11 weeks if at all possible. Neonatal haemoglobino- appropriate management of sickle cell disease and
pathy screening is undertaken for the detection of b-thalassaemia major
sickle cell disease and b-thalassaemia major since . Diagnosis of patients presenting with clinical features
awareness of these conditions at an early stage suggestive of sickle cell disease or b-thalassaemia
permits institution of treatment programmes. It major
. Diagnosis of patients with otherwise unexplained
has been clearly demonstrated that the administration haemolytic anaemia, polycythaemia, anaemia or cya-
of prophylactic penicillin from early infancy in nosis
subjects with sickle cell disease signi¢cantly reduces . Confirmation that a variant haemoglobin is responsible
infant mortality.5 Haemoglobinopathy screening for an atypical elution pattern when the total HbA1 or
programmes are becoming widely implemented in HbA1c is being quantitated by high-performance liquid
areas where the percentage of the ethnic minority chromatography
population is high. The screening programmes are

Ann Clin Biochem 2004; 41: 355–369

358 Wild and Bain

Methodology the sensitivity and accuracy of recent batches of

reagents have been questioned and some reagent
Overview of analytical procedures used batches have been withdrawn.When unusual variants
Haemoglobinopathy diagnosis involves not only the are found, it is both analytically and clinically
accurate performance of laboratory tests but also the useful to determine whether the variant is a or b
interpretation of the resultant data together with basic in origin, and to this end globin chain electro-
haematological data, knowledge of the patient’s phoresis is employed. In the future, other immuno-
ethnicity and clinical history of the patient and family. logical methods for the detection of speci¢c
Thus, although this review will deal principally with haemoglobin variants may become available, but these
the analytical aspects of the detection and quantita- would need to be carefully assessed for accuracy and
tion of normal haemoglobins and haemoglobin speci¢city.
variants, it is necessary to place these in an appro- Quantitation of HbA2 involves separating the HbA2
priate clinical context. It should be noted that, in clin- from any other haemoglobin(s) present and then
ical practice, the detection of haemoglobin variants determining the proportion of HbA2 present. This may
cannot be dissociated from the diagnosis of thalas- be undertaken using one of three procedures: haemo-
saemic conditions. This review encompasses the globin electrophoresis with elution of the resultant
detection and quanti¢cation of HbA2 and HbF but not haemoglobin bands, micro-column chromatography
the other aspects of thalassaemia diagnosis. or automated HPLC. HPLC has the advantage that it
Historically, the variant haemoglobins were allows simultaneous quantitation of HbA2 and HbF
detected by the use of haemoglobin electrophoresis, together with any haemoglobin variants found, but
using paper or starch as the support medium.11 For the the disadvantage that it may give erroneous results for
last 30 years, haemoglobin electrophoresis has been HbA2 and HbF in the presence of a haemoglobin
undertaken using cellulose acetate membrane (CAM) variant. Quantitation of haemoglobin variants may
as the support medium, resulting in superior separa- also be achieved using haemoglobin electrophoresis
tion and resolution of the variants.12 Haemoglobin with elution or with automated scanning densito-
electrophoresis is undertaken at both alkaline pH metry. However, automated scanning densitometry is
using CAM and at acid pH using citrate agar;13 more not accurate enough for HbA2 quantitation since this
recently, alkaline and/or acid agarose gels have been requires approximately 10 times greater precision
used as support media. Acid agarose gels provide a when used in the diagnosis of b-thalassaemia trait
similar, but not identical, separation of haemoglobins than the quantitation of a variant. HbF is the major
as citrate agar but have the advantage in that prepared haemoglobin present in neonates (comprising
gels are commercially available whereas only one approximately 80% of the total haemoglobin). It is
company supplies citrate agar gels ready for use. largely replaced by HbA during the ¢rst year of life so
Automated HPLC was introduced as a tool for haemo- that it has usually reached adult proportions (less
globinopathy screening in the early 1990s and has than 1%) by the end of the ¢rst year of life. The HbF
become popular in laboratories where large numbers level may remain raised due to this change being
of samples are tested. Isoelectric focusing (IEF) was incomplete or not occurring at all, as occurs in HPFH,
originally described by Righetti et al.,14 and is a very of which there are several types,3 or due to the inheri-
reliable technique for the detection of haemoglobin tance of db-thalassaemia.3 The most common type of
variants; however, it is usually only appropriate in HPFH in the UK usually occurs in people of African
laboratories with a high throughput or in reference descent in whom the heterozygote will have approxi-
laboratories. It is usual for laboratories to employ mately 30% HbF evenly distributed in the red cells
either haemoglobin electrophoresis at alkaline pH or (pancellular) and the homozygote will only have HbF.
automated HPLC or the sickle test as the primary Other types of HPFH will have HbF levels of the order
analytical procedure for the detection of variants in of 5-10% which may be evenly distributed in the red
general or HbS speci¢cally. On detecting a haemo- cells (pancellular) or variably distributed (hetero-
globin variant, it is essential to utilize a second and/or cellular). HbF may also be raised in pregnancy (up to
third procedure in order to increase the accuracy 5%) and in association with various inherited condi-
of the presumptive identi¢cation. Where the initial tions (such as sickle cell disease) and acquired condi-
results suggest the presence of sickle haemoglobin, the tions (such as myelodysplasia). Historically, HbF was
sickle solubility test is indicated. Where the variant measured by utilizing its resistance to alkali in order
detected is clearly not sickle haemoglobin, haemo- to separate it from other haemoglobins. The Singer
globin electrophoresis at alkaline and acid pH, or 1-min alkali denaturation test15 was described in 1951
IEF, is indicated. Immunoassay has been used for and can give reproducible results with an accuracy
con¢rmation of the presence of HbS, HbC or HbE, but suitable for clinical use when the level is above 5%, but
this method has proved to be unreliable since both it is less useful at levels below this. The 2-min alkali

Ann Clin Biochem 2004; 41: 355–369

 Detection and quantitation of normal and variant haemoglobins 359

denaturation test described by Betke et al.16 in 1959 is Box 2. Analytical procedures employed in the various clinical
much better in the range usually found in clinical situations
situations; better still is the modi¢cation published in
1974 by Pembrey et al.17 which can give precise and Analytical procedures used for the detection of
accurate results well down into the normal range of haemoglobin variants
0.2-0.8% and also up to levels of 50%, but the method . Haemoglobin electrophoresis at alkaline pH (cellulose
must be followed very carefully. Above 50% HbF, the acetate membrane or alkaline agarose)
method may underestimate the HbF but this is rarely . Haemoglobin electrophoresis at acid pH (citrate agar
of any clinical signi¢cance and such specimens are or acid agarose)
rare after the neonatal period. This 2-min alkali dena- . Automated high-performance liquid chromatography
turation test was recommended in 1979 as the inter- . Isoelectric focusing
national reference method.18
Specific test for the determination of sickle haemoglobin
HbF can also be measured by HPLC, giving precise . Sickle solubility testing
results down into the normal range, but the peak . Sickling test (useful post-transfusion)
detectors are not always able to recognize the peak in
the lower half of the normal range. If this happens, it is Detection of the unstable haemoglobins
clinically satisfactory to report the result as ‘less than . Heinz body preparation
1%’ although this can cause problems with the statis- . Heat-stability test
tical analysis of some quality assessment schemes that . Isopropanol-stability test
can not cope with a result that is ‘less than’. It is
Detection of the altered affinity haemoglobins
essential to examine all chromatograms very carefully . Oxygen dissociation curve
before authorizing HPLC results, and especially when
HbF is concerned. The reason for this is that auto- Detection of HbMs (haemoglobinopathies)
mated peak detection and integration systems may . Quantitation of methaemoglobin
falsely label a glycated haemoglobin A peak as HbF; . Haemoglobin electrophoresis at pH 7
thus, whenever the HbF level is above 10%, a di¡erent . Spectrophotometry of haemolysate
qualitative technique (such as CAM at alkaline pH)
should be used to check that the peak is really HbFand Analytical procedures used for the direct or indirect
identification of haemoglobin variants
not an early eluting variant haemoglobin from the
. Mass spectrometry of whole blood
HbN or HbJ group. If these precautions are taken, . DNA analysis
automated HPLC can be a very satisfactory method for
quantitating Hbf.
Radial immunodi¡usion can also be used for
measuring HbF but is less precise than the 2-min The results of these analyses are interpreted in
alkali denaturation technique within and close to the conjunction with results of the full blood count and
normal range. This is of little clinical importance and blood ¢lm, clinical details and the patient’s ethnic
so it can be a useful technique if the laboratory origin. These analytical procedures will detect the
throughput does not warrant HPLC. majority of the common clinically important haemo-
For the rarer haemoglobin variants, such as the globin variants and enable a presumptive identi¢ca-
unstable haemoglobins, tests for haemoglobin tion of the variant to be made, which is often su⁄cient
instability using either the heat-stability or isopro- for clinical purposes. In order to identify the variant
panol-stability tests are indicated.19 Investigation for unequivocally it is necessary to analyse either the
altered a⁄nity haemoglobins requires the speci¢c protein structure by the use of mass spectrometry,21
measurement of the haemoglobin a⁄nity for oxygen which has largely replaced ¢ngerprinting with amino
by plotting the haemoglobin-oxygen dissociation acid analysis, or to perform DNA analysis using gene
curve. HbMs require haemoglobin electrophoresis at sequencing or speci¢c restriction enzyme techni-
pH 7 and spectrometric analysis by wavelength scan- ques.22,23 A summary of the analytical procedures
ning of haemolysate.8,11 The investigation of a patient employed in the various clinical situations is given in
for the presence of an HbM should include quantita- Box 2.
tion of the methaemoglobin level. However, it is
important to remember that if using the traditional Full blood count and film
method of Evelyn and Malloy 20 the result may be Most haemoglobin variants are not associated with
inaccurate in the presence of HbM due to the di¡er- any abnormality in the blood count, but some are
ence in the peak absorbance of the induced ferrihae- associated with microcytic, hypochromic red cells and
moglobin of an HbM molecule compared with that of these changes will be identi¢ed when assessing the
HbA. full blood count and blood ¢lm. The blood ¢lm can also

Ann Clin Biochem 2004; 41: 355–369

360 Wild and Bain

be very useful in assessing the signi¢cance of a posi- cellulose acetate is very convenient to use and is often
tive sickle test. In practice, most blood count analysers referred to as a cellulose acetate ‘plate’. Haemoglobin
now produce a comprehensive set of variables, solutions (haemolysates) are prepared by lysing
including all aspects of red cell, white cell and platelet packed cells in Triton-X 100 or EDTA, and these are
measurements. In analysing samples for the purpose applied to the cellulose acetate by an applicator that
of haemoglobinopathy diagnosis it is important to gives a discrete line. Tris-EDTA-borate bu¡er pH 8.4
measure the haemoglobin level, red cell count, mean is used and the electrophoresis run at 350 V for
cell haemoglobin, mean cell volume and red cell 25 min.26 The sample plates are then ¢xed and stained
distribution width as these will indicate if anaemia or simultaneously using Ponceau S stain in trichloro-
erythrocytosis is present and distinguish between acetic acid solution. Excess stain is removed by
normochromic, normocytic and hypochromic, micro- washing in acetic acid solution and the plates are left
cytic conditions. For a full review of the blood count to dry; the samples appear as discrete red bands on a
and ¢lm examination the reader is referred to Dacie white background. The method reliably detects the
and Lewis’ Practical Haematology.24 clinically signi¢cant haemoglobin variants HbS, HbC,
HbDPunjab, HbE, HbOArab and Hb Lepore, and will also
Haemoglobin electrophoresis detect many of the other haemoglobin variants
In 1949, Pauling used zone electrophoresis to demon- described. It is a sensitive method, detecting abnormal
strate that the electrophoretic mobility of haemo- fractions comprising as little as 1% of the total
globin from people with sickle cell anaemia di¡ered haemoglobin component. It is possible to render the
from that of normal people. Haemoglobin electro- cellulose acetate transparent by using a clearing
phoresis was ¢rst undertaken using paper as the solution but this does not o¡er any de¢nite advantage
supporting medium and borate discontinuous bu¡er other than permitting the use of automated densito-
systems. As the interest in the abnormal haemoglo- metry. Examples of this technique are shown in Figs 1
bins increased in the1960s, especially by investigators and 2.
such as Lehmann and Schneider, the repertoire of
electrophoretic techniques developed enormously to
encompass separations at alkaline and acid pH using a Haemoglobin electrophoresis at alkaline pH using agarose
variety of support media. gels
Alkaline agarose gels can be used as an alternative to
Haemoglobin electrophoresis at alkaline pH using CAM CAM. A variety of gels are available commercially for
This was ¢rst introduced and optimized for the this purpose but the quality of separation is often
separation of haemoglobins by Marengo-Rowe in the inferior to that of the CAM method described. Whilst
1960s 25 and was further developed for the qualitative the gels are capable of separating the common
separation of haemoglobin variants. The various types haemoglobin variants, some are less satisfactory than
of CAM di¡er in pore size, length of cellulose chain and the CAM technique both at separating HbF from HbA
degree of acetylation of the cellulose. Plastic-backed and at detecting certain uncommon haemoglobin

Figure 1. Cellulose acetate electrophoresis at

alkaline pH, adult samples (small bands are
listed in parentheses). 1: A, S, (A2). 2: (A), S,
(Sb+thal). 3: S, C. 4: A, C. 5: A, N, (A2). 6: A, S,
G. 7: A. 8: A, (A2).

Ann Clin Biochem 2004; 41: 355–369

 Detection and quantitation of normal and variant haemoglobins 361

Figure 2. Cellulose acetate electrophoresis at alka-

line pH, neonatal samples. 1: A, F, S, C control. 2: A, F
(normal neonate). 3: F, S. 4: A, F, S. 5: A, F. S, 6: F, S,
C. 7: A, F, C. 8: A, F, S, C control.

variants. Some are also often less able to detect small Sickle solubility testing
quantities of the variants. The characteristic reduced solubility of HbS was put
to use in the development of an analytical procedure to
detect sickle haemoglobin. Sickle solubility testing is
undertaken by introducing 10 mL of packed red cells
Haemoglobin electrophoresis at acid pH using into a phosphate bu¡er that contains a reducing agent.
citrate agar or acid agarose gels Red cells containing sickle haemoglobin will cause
Haemoglobin electrophoresis at acid pH using citrate refraction of the light and the solution will appear
agar gel as the support medium was ¢rst described in turbid. Red cells that do not contain sickle haemo-
1957 by Robinson et al.13 This method was introduced globin will lyse and the sample remains clear. The
to supplement the information obtained from separa- sickle solubility test has the disadvantage that it does
tions performed at alkaline pH since the resultant not di¡erentiate between sickle cell trait and sickle cell
mobilities of di¡erent haemoglobins give character- disease: any sample containing more than 15% HbS
istic patterns partly due to the acidic pH and partly due will give a positive result. It is therefore always neces-
to the solubility of the di¡erent haemoglobins in agar. sary to establish the haemoglobin phenotype using
With the information obtained from the two techni- either haemoglobin electrophoresis at alkaline pH or
ques, it became possible to di¡erentiate between HbS HPLC with quantitation of the amount of HbS present.
and HbD, and also between HbC, HbE and HbOArab, False negative solubility tests may occur if the sample
which co-migrate on alkaline electrophoresis. has too low a haemoglobin concentration; such false
The staining system used for citrate agar gel elec- negatives should be avoided by the use of packed red
trophoresis is usually haem-speci¢c and as such di¡ers cells. False positive results can occur if there is exces-
from the general protein stain commonly used for sive lipid in the patient’s sample, as in some patients
cellulose acetate and agarose separations. Citrate agar undergoing parenteral nutrition, or if a paraprotein is
gel electrophoresis was widely used for many years as present. Again, the use of packed red cells will mini-
the second-line test to con¢rm the presence of mize such errors. Since b globin development is not
haemoglobin variants but now has been largely fully achieved until 6-12 months of age, use of the
superseded by the use of acid agarose gels since the solubility test is generally inappropriate in neonates
latter are more convenient to use. Unfortunately, and infants under 6 months.
migration patterns obtained from acid agarose gel
separations are not always identical to those tradi- Automated HPLC
tionally produced by citrate agar and therefore HPLC has been used for haemoglobin analysis since
reference to electrophoretic mobility in texts that the late 1970s, when it was used for the separation of
relate to the latter technique may give misleading peptides during the identi¢cation of abnormal
information. An example of citrate agar electrophor- haemoglobins.27,28 In the early 1990s, automated
esis is shown in Fig. 3 and an example of acid agarose HPLC was developed to meet the need created by
electrophoresis is shown in Fig. 4. expanding screening programmes. The objectives of

Ann Clin Biochem 2004; 41: 355–369

362 Wild and Bain

Figure 3. Citrate agar electrophoresis at acid pH,

adult samples (small bands are listed in
parentheses). 1: A, S. 2: A, C. 3: F, (A) control.
4 & 5: homozygous OArab. 6: A, S. 7: A, C. 8: F, (A)
control.

Figure 4. Acid agarose gel electrophoresis,

adult samples (small bands are listed in
parentheses). 1: A, (F). 2: A, (F) (normal
adult). 3: A, hereditary persistence of fetal
haemoglobin. 4: S, A. 5: S, S. 6: SC. 7: A, (F).
8: F, A, S, C control.

the technology were to detect and quantitate both systems unless adequate safeguards are both available
normal and abnormal haemoglobin fractions. Several and activated, and to ensure that all the chromato-
manufacturers now o¡er speci¢c programmes for grams are inspected by an experienced analyst prior to
application within haemoglobinopathy screening. All authorization since minor changes in retention times
of these employ calibrated bu¡ers and a column can result in peaks being wrongly identi¢ed. The fact
accompanied by a pre-programmed gradient on an that glycated and other post-translationally modi¢ed
integrated circuit card or CD-ROM. variant haemoglobins often elute with the same
Automated HPLC will detect the common clinically retention time as HbA0 or HbA2 can cause problems
important haemoglobin variants and many of the less with automatic transfer of data. Similarly, the fact that
common variants and will reliably quantitate fractions HbE often elutes with HbA2 can result in absurd
as small as1%. However, it has the disadvantage that it values for HbA2 being reported. For all these reasons,
separates haemoglobins that have undergone post- the initial identi¢cation and/or quantitation of peaks
translational modi¢cation, such as HbAI, HbSI, HbAIII may be incorrect and if not detected can lead to
and HbSIII, not only from each other but also from the diagnostic errors.
unmodi¢ed forms and this makes interpretation more
di⁄cult. As with haemoglobin electrophoresis, Isoelectric focusing14
di¡erent haemoglobins often co-elute, demonstrating IEF separates haemoglobins on the basis of their
that speci¢c retention times are not unique for a isoelectric points. For this reason, it leads to the
speci¢c haemoglobin. As with most column chroma- formation of more discrete sharp bands and often
tography procedures, the systems are highly sensitive separates haemoglobins that do not separate on CAM
to temperature; di¡erent temperatures produce at alkaline pH. IEF has been used in large-scale
di¡erent retention times and therefore a haemoglobin screening programmes, notably neonatal screening,
variant cannot be even provisionally identi¢ed on the when it is convenient to batch and analyse samples in
basis of retention time alone. It is important to be wary this way. Most IEF gels will accommodate up to 80
of on-line transfer of HPLC data to pathology computer samples. However, IEF has the disadvantage that it

Ann Clin Biochem 2004; 41: 355–369

 Detection and quantitation of normal and variant haemoglobins 363

Figure 5. Isoelectric focusing (small bands

are listed in parentheses). 1: A, F, S, E control.
2, A, F, D, C control. 3: A, (A2) (normal adult).
4: A, S, (A2). 5: (F), S. 6: S, C. 7: S, hereditary
persistence of fetal haemoglobin. 8: A, S. 9: A,
J. 10: F, A, S (infant 3 months). 11: F, S (infant
3 months). 12: A, F, S, E control. 13: A, F, D, C
control.

separates post-translationally modi¢ed forms of the Stability testing

haemoglobins, such as glycated and oxidized haemo- The stability of haemoglobin was ¢rst exploited as a
globins, and this makes interpretation more di⁄cult. characteristic of potential diagnostic use by Grimes 30
An example of IEF is shown in Fig. 5. in 1962. Normal adult haemoglobin will withstand
temperatures of up to 508C for 1h without signi¢cant
deterioration, whereas haemoglobin variants that
Globin chain electrophoresis have substitutions at or near the haem pocket will fail
Electrophoresis of globin chains is undertaken to to withstand the temperature and will precipitate from
speci¢cally identify which of the globin chains carries solution. In 1972, Carrell and Kay 31 described a modi-
a mutation. Globin chain electrophoresis is performed ¢cation to the heat-stability test which utilized
at alkaline and/or acid pH in the presence of 6 mol/L isopropyl alcohol in the reaction mixture. This meant
urea and demonstrates the mobility of variant globin that the test could be undertaken at 378C as opposed to
chains in comparison to the normal globin.29 It is the higher temperatures used by Grimes and Meisler
essential that the sample is also analysed using in the heat-denaturation test. It also enabled positive
haemoglobin electrophoresis at alkaline pH, usually and negative controls to be prepared more readily.
on CAM, in order that correct interpretation of the
mobility patterns can be undertaken. It is important to Oxygen affinity measurement
know which chain is a¡ected by the substitution since Substitutions at the a1b1 interface can have an adverse
the clinical signi¢cance will vary. In general, haemo- e¡ect on the quaternary structure of the haemoglobin
globin variants with substitutions on the b globin molecule, thus inhibiting or exaggerating the ability of
chain are potentially more severe than those on the a the molecule to take up and release oxygen. Such
chain. Variants on the g chain only cause problems in haemoglobin variants are termed the altered a⁄nity
very early life; they are, of course, impossible to detect haemoglobins. They are uncommon, but they are
in adults but may cause confusion during neonatal signi¢cant in that they can cause marked erythrocy-
screening. Variants in d chain structure are very tosis or anaemia.9 It is clinically important to detect
common, but in reality are rarely clinically signi¢cant; such haemoglobin variants because if they are missed
however, they can cause problems in the diagnosis of the patient may be misdiagnosed as having polycy-
b-thalassaemia trait by interfering with the quantita- thaemia rubra vera, which may lead to inappropriate
tion of HbA2. An example of globin electrophoresis at treatment with 32P with consequent risk of post-
alkaline pH is shown in Fig. 6. treatment malignancy. Haemoglobin electrophoresis

Figure 6. Globin electrophoresis at alkaline

pH, adult samples. 1: A (a+b). 2: A, S
(a+b+bS). 3: A, C (a+b+bC). 4: A, GPhiladelphia
(a+aG+b). 5: A, GPhiladelphia (a+aG+b). 6: A
(a+b). 7: A, S (a+b+bS). 8: A, C (a+b+bC).

Ann Clin Biochem 2004; 41: 355–369

364 Wild and Bain

of the altered a⁄nity haemoglobins is often unin- in more than 95% of cases, which is similar to that
formative since the variants are frequently electro- obtained with other techniques. The method involves
phoretically silent. Evidence for the clinical e¡ect of electrospray ionization of whole blood in order to
the suspected haemoglobin variant will be obtained assess mass di¡erence and determine which globin
from the measurement of the oxygen dissociation chain is a¡ected so as to identify possible single amino
curve. Surprisingly, case reports of people with high- acid substitutions. Tandem mass spectrometry of
a⁄nity haemoglobins indicate that many of them also whole blood identi¢es which section of the relevant
have a raised white cell count but the reason for this chain contains the mutation. Electrospray ionization
has not been established.9 Low-a⁄nity haemoglobins of a tryptic digest of the blood identi¢es which peptide
cause a reduced haemoglobin concentration, since contains the variant amino acid, and tandem mass
normal oxygen delivery to the tissues is usually main- spectrometric analysis of the suspect peptide will
tained. This is of no clinical signi¢cance but may cause pinpoint the precise location. It is possible to identify
diagnostic confusion with other possible causes of a abnormalities present in very small proportions (5%)
reduced haemoglobin concentration. of the total haemoglobin. Using a combination of
phenotypic and mass spectrometry procedures, it is
HbMs8 easily possible to identify variants with only 1 Da mass
Methaemoglobinaemia may be due to three di¡erent di¡erence, such as HbC, HbE and HbDPunjab from HbA,
mechanisms: (1) exposure to toxic oxidizing agents since these haemoglobins have a di¡erence in charge
that cannot be adequately neutralized by red cell redu- and therefore separate on electrophoresis and/or
cing enzymes (mainly NADH diaphorase); (2) reduced chromatography.
or absent activity of red cell NADH diaphorase; or (3)
the presence of one of the haemoglobin variants DNA analysis
known as the HbMs.2 DNA analysis has a valuable role in that it permits
The usual method of measuring the amount of identi¢cation of a precise mutation in an individual
methaemoglobin present involves measuring the but, as with protein analysis, it is essential to ask the
di¡erence in absorption before and after the addition right question because there is no single analytical
of cyanide to convert any methaemoglobin present to DNA procedure that can provide all the clinical
cyanmethaemoglobin, which has a di¡erent absorp- answers needed. Like protein analysis, it is also easier
tion spectrum to methaemoglobin.20 This method is to con¢rm, or exclude, the presence of a speci¢c
very satisfactory in the ¢rst two situations as long as mutation than to detect or exclude the presence of an
fresh reagents are used, but can give misleading unknown mutation in a particular sample.
results in the presence of one of the HbMs. The HbMs DNA analysis is essential for ¢rst trimester prenatal
result from a mutation in the haem pocket of one of the diagnosis since no other analytical approach can
globin chains which usually involves the histidine predict what protein will, or will not, be synthesized
residues that lie close to the haem iron. At the present later in life. DNA analysis is also very useful in
time we know of two a chain and three b chain muta- con¢rming, or excluding, the presence of HbDPunjab in
tions that result in HbMs. Since these mutations only a blood sample from a patient in whom other tests have
a¡ect the histidine residues, these haemoglobins do shown the presence of a D-like haemoglobin (there are
not usually separate from HbA on electrophoresis at about 20 di¡erent D-like haemoglobins that have
pH 8.4 and are therefore said to be electrophoretically already been identi¢ed). Prenatal diagnosis carries a
silent. However, since histidine ionizes at neutral pH, risk of inducing a miscarriage (approximately 1%
these haemoglobin variants can be detected by above background) and it is therefore important not to
undertaking electrophoresis at pH 7 in a phosphate undertake this procedure unless there is good
bu¡er.11 evidence that it is clinically necessary. The results of
such DNA analysis are sometimes required quickly,
Mass spectrometry such as in antenatal screening when a decision has to
Mass spectrometry has been used for the character- be made as to whether or not prenatal diagnosis is
ization of haemoglobin variants by various labora- appropriate as it may produce information that may
tories since the 1980s but its use is currently limited to lead to a decision to terminate a pregnancy.
specialized centres. The development of the technique
to give a rapid procedure that can be undertaken on Practical aspects of providing
very small samples of blood has been described
recently.21 This procedure has been undertaken in
haemoglobinopathy service (Box 3)
conjunction with a range of electrophoretic and chro- Sample requirements
matographic separations and has shown that positive EDTA-anticoagulated blood samples are usually used
identi¢cation of the abnormal haemoglobin is possible for the analysis of variant haemoglobins. Samples are

Ann Clin Biochem 2004; 41: 355–369

 Detection and quantitation of normal and variant haemoglobins 365

Box 3. Factors to be considered when selecting methods for a haemoglobins or their post-synthetic adjuncts can
haemoglobinopathy service cause misleading results for HbA2 measurement by
HPLC.
. Number of samples per day, week, year
. Nature of samples (small samples and/or dried blood Quality assurance
spots) Controls should be used with each electrophoretic
. Ethnic origin of patient population run, and should contain the common haemoglobin
. Turnaround time required (is antenatal screening to be variants as well as the normal haemoglobins. A
performed? will quantification of HbS be required control mixture of HbF, HbA, HbS and HbC is widely
urgently in patients needing exchange transfusion?) used. UK NEQAS (UK National External Quality
. Resource implications (funding for staff, equipment and Assessment Scheme) circulates samples for the iden-
reagents) ti¢cation of abnormal haemoglobins and for the
. Availability of a consistent supply of reliable reagents
measurement of HbA2, HbF and HbS. Participation in
or kits
. Availability of support for maintenance of equipment these external quality assessment programmes is a
requirement for Clinical Pathology Accreditation of
haemoglobinopathy laboratories. Commercial
controls are available from several manufacturers for
ideally tested fresh, but liquid samples up to 1 month
both HbA2 and HbF quantitation and as electro-
old can be used for haemoglobin electrophoresis, IEF,
phoretic/chromatographic markers for haemoglobins
HPLC and mass spectrometry. Red cells can even be
such as HbA, HbF, HbA2, HbS, HbC, HbD and HbE.
recovered from blood clots and analysed. In addition,
Reference materials certi¢ed by the World Health
samples that have been frozen can be used for
Organization are available from the National Institute
some of the procedures. Samples for stability
of Biological Standards and Controls to assist in the
testing must be tested as soon as possible after collec-
calibration of methods and instruments used to
tion to avoid false positive results resulting from the
quantitate HbA2 and HbF.
presence of methaemoglobin. Oxygen dissociation
measurement requires the use of heparinized samples Storage of electrophoretic strips, gels and HPLC
which also should be tested within a few hours of chromatograms
venesection. Apart from neonatal screening cards which should be
Haemoglobin electrophoresis and HPLC use only a kept for 20 years, there are no speci¢c guidelines for
very small quantity of sample, usually 10 mL will the retention of laboratory data results used for genetic
su⁄ce. Haemoglobin stability tests and the quantita- purposes in the National Health Service, although
tion of HbA2, HbF and haemoglobin variants by de¢nite guidelines are given for personal health
manual techniques require a larger sample (up to records.32 However, it is prudent to retain data at least
5 mL). DNA techniques require a sample with nucle- until the outcome of any related pregnancy is known
ated cells; usually white blood cells from a peripheral and probably for very much longer, as the unexpected
blood sample (the bu¡y coat) are used. birth of a baby with a major haemoglobinopathy may
lead to medicolegal issues that can continue for several
Inaccuracy
years.
Carbonic anhydrase and carbonic anhydrase variants
can cause confusion when electrophoretic prepara- Safety
tions are stained with protein stains rather than with Electrical safety is important, as with any laboratory
haem-speci¢c stains; if there is any suspicion, the procedure.
electrophoresis should be repeated using a haem-
speci¢c stain such as o-dianisidine. Since the lifespan Throughput
of a red cell is 120 days, inaccuracies can result if For electrophoresis it is usual to process 8-12 samples
samples are analysed within 4 months of the last blood per plate, and each electrophoresis tank will accom-
transfusion. Since there are more than 800 variant modate three sample plates. An experienced worker is
haemoglobins, they can easily be confused with each likely to analyse up to 80 samples per day. Most auto-
other when using the methods available to routine mated HPLC analysers can process 100 samples per
service laboratories. batch.

Imprecision
Poor sample application, di¡ering protein concentra-
Haemoglobinopathy screening strategies
tions in the samples, inadequately blotted CAM, delay Pre-anaesthetic screening33,34
in applying the current and delay in ¢xation after The goal of such screening is to detect, or exclude, the
electrophoresis will all cause poor results. Variant presence of sickle haemoglobin and if present, to

Ann Clin Biochem 2004; 41: 355–369

366 Wild and Bain

Figure 7. Flow diagram for pre-anaesthetic screening. HPLC, high-performance liquid chromatography; RBC, red blood cell.

determine whether the a¡ected individual is a age. A £ow diagram of a recommended approach is
heterozygote (sickle cell trait), a homozygote (sickle given in Fig. 9.
cell anaemia) or a compound heterozygote with HbC,
HbDPunjab, HbOArab or b-thalassaemia. A £ow diagram Prenatal diagnosis35,36
to indicate a recommended approach is given in Prenatal genetic diagnosis is undertaken in order to
Fig. 7. detect whether or not a fetus has inherited a condition
that will cause medical problems during gestation or
Preconceptional and antenatal screening34,35 later in life. In the haemoglobinopathy ¢eld, it is
The goal in such screening is to detect the presence or undertaken to detect homozygosity for the most severe
absence of sickle haemoglobin or b-thalassaemia trait form of a-thalassaemia (Hb Barts hydrops fetalis)
or of one of the haemoglobin variants that interact which results in obstetric problems for the mother and,
with them. HbC, HbDPunjab, HbOArab, Hb Lepore and usually, intrauterine death of the fetus at around 30
b-thalassaemia all interact with sickle haemoglobin to weeks of gestation. Prenatal diagnosis is also under-
produce the clinical signs and symptoms of sickle cell taken to detect sickle cell disease and b-thalassaemia
disease. HbE, HbOArab and Hb Lepore all interact with major. Neither of these conditions causes problems
b-thalassaemia to produce the clinical signs and during gestation but both can cause severe, life-
symptoms of b-thalassaemia intermedia. Screening is threatening problems at any time after the ¢rst 3
also undertaken to detect db thalassaemia, HPFH and months of post-natal life, so potential parents may
a0 thalassaemia. If one of these haemoglobinopathies wish to consider termination of the pregnancy.
is present it is important to test the partner in order to For both social and obstetric reasons, prenatal diag-
be able to assess whether or not the baby will be at risk nosis is best undertaken during the ¢rst trimester; the
of developing a major haemoglobinopathy. It is also current recommendations are that it should be
important to know whether the mother has one of the undertaken at 11 weeks of gestation or as soon as
genotypes associated with sickle cell disease in order possible after that time. At this stage, a sample of DNA
to manage her pregnancy appropriately. A £ow is obtained by chorionic villus sampling; later in
diagram of a recommended approach is given in Fig. 8. gestation a similar procedure is called placental
biopsy. To minimize diagnostic error it is important to
Neonatal screening26,34 obtain samples from both parents at the same time for
The goal of neonatal screening is to detect infants with analysis in parallel with the fetal sample. If the results
sickle cell disease or b-thalassaemia major before of parental DNA analysis are available prior to the fetal
clinical symptoms develop so that appropriate clinical analysis, the results can usually be available in 3
management can be implemented before 3 months of working days.

Ann Clin Biochem 2004; 41: 355–369

 Detection and quantitation of normal and variant haemoglobins 367

Figure 8. Flow diagram for antenatal screening. HPLC, high-performance liquid chromatography; GCS, globin chain synthesis; MCV,
mean corpuscular volume; MCH, mean corpuscular haemoglobin; SCD, sickle cell disease. *If HPLC is not available, HbA2 and HbF
should be quantitated if the MCH is 527 pg.

Figure 9. Flow diagram for neonatal screening. HPLC, high-performance liquid chromatography. *For emergency anaesthesia, a
sickle solubility test should be performed but results should subsequently be confirmed by standard procedures.

Ann Clin Biochem 2004; 41: 355–369

368 Wild and Bain

An alternative approach is to obtain fetal blood by References

cordocentesis, but this cannot generally be under- 1 Huisman T, Carver M, Efremov G. Human Hemoglobin Variants,
taken until 19-20 weeks of gestation. Although 2nd edn. Augusta, GA: Sickle Cell Anemia Foundation, 1998
undertaken much later in pregnancy, this analysis 2 Hb Variant Database. http://globin.cse.psu.edu/
does have the advantage that it examines the haemo- 3 Weatherall DJ, Clegg JB. The Thalassaemia Syndromes, 4th edn.
globin that is being synthesized rather than having to Oxford: Blackwell Scientific, 2001
predict what haemoglobin will be synthesized by 4 John WG. Glycated haemoglobin analysis. Ann Clin Biochem 1997;
analysis of the DNA. Analysis of the fetal blood sample 34: 17-31
requires incubation for 1h with radiolabelled leucine 5 Serjeant GR, Serjeant BE. Sickle Cell Disease, 3rd edn. Oxford:
to allow the leucine to be incorporated into newly Oxford University Press, 2001
synthesized globin. This is followed by preparative 6 White J. The unstable haemoglobin disorders. Clin Haematol 1974;
column chromatography and scintillation counting of 3: 333
the eluted fractions to determine which globin chains 7 Dacie J. The Haemolytic Anaemias. Vol. 2: The Hereditary
are, or are not, being synthesized. This procedure also Haemolytic Anaemias, 3rd edn. Edinburgh: Churchill Livingstone,
takes approximately 3 days to produce an answer and, 1988
unlike DNA analysis, does not require samples from 8 Kiese M. Methemoglobinemia: A Complete Treatise. Cleveland,
the parents. Single cell analysis is currently being OH: CRC Press, 1974
9 Stephens A. Polycythaemia and high affinity haemoglobins. Br J
developed for pre-implantation genetic diagnosis for
Haematol 1977; 36: 153-159
the haemoglobinopathies.
10 Standing Medical Advisory Committee on Sickle Cell, Thalassaemia
and other Haemoglobinopathies. Report of the Standing Committee
Future developments on Sickle Cell, Thalassaemia and other Haemoglobinopathies.
London: Department of Health, 1993
In terms of screening, it is possible that capillary 11 Lehmann H, Huntsman RG. Man’s Haemoglobins, 2nd edn.
electrophoresis will be developed for haemoglobin Amsterdam: North-Holland, 1974
separation. Semi-automated kits are being introduced 12 Schneider RG. Identification of hemoglobin by electrophoresis. In:
for the common DNA mutations for thalassaemia and Schmidt RM, Huisman THJ, Lehmann H, eds. The Detection of
some of the haemoglobin variants. Development of Hemoglobinopathies. Cleveland, OH: CRC Press, 1974: 11
automated mass spectrometry could lead to the appli- 13 Robinson AR, Robson M, Harrison AP, Zuelzer WW. A new
cation of this procedure for screening for speci¢c technique for differentiation of hemoglobins. J Lab Clin Med 1957;
haemoglobin variants. 50: 745-52
In terms of service delivery, there is currently a 14 Righetti PG, Gianazza E, Bianchi-Bosisio A, et al. The
review of both antenatal and neonatal haemo- hemoglobinopathies: conventional isoelectric focusing and immobi-
lized pH gradients for hemoglobin separation and identification. In:
globinopathy screening by the National Screening
Methods in Hematology 1986; 15: 47–71
Committee and it is intended that the end of 2005,
15 Singer K, Chernoff AI, Singer L. Studies on abnormal
a national programme will be implemented in
haemoglobins: 1. Their demonstration in sickle cell anaemia and
England.37 other haematological disorders by means of alkali denaturation.
Blood 1951; 6: 413-28
Conclusions 16 Betke K, Marti HR, Schlicht L. Estimation of small percentage of
fetal haemoglobin. Nature 1959; 184: 1877
The recommended analytical procedures for 17 Pembrey ME, McWade P, Weatherall DJ. Reliable routine
screening programmes are shown in the relevant £ow estimation of small amounts of foetal haemoglobin by alkali
charts. Using screening programmes, it is possible to denaturation. J Clin Pathol 1972; 25: 738-40
detect speci¢ed conditions with an acceptable success 18 International Committee For Standardization in Haematology.
rate. As with all population screening programmes, Recommendations for fetal haemoglobin reference preparations
there will be failures as some very rare conditions are and fetal haemoglobin determination by the alkali denaturation
not detectable by simple automated techniques. method. Br J Haematol 1979; 42: 133-6
Awareness and acceptability of the error and failure 19 Carrell RW. The hemoglobinopathies: methods of determining
hemoglobin instability (unstable hemoglobins). In: Huisman THJ,
rates are important for laboratory and clinical sta¡
ed. Methods in Hematology. Edinburgh: Churchill Livingstone, 1986;
and for the patient.
15: 109–24
Investigation of patients for the rarer haemo-
20 Evelyn KA, Malloy HT. Microdetermination of oxyhemoglobin,
globinopathies, which can cause clinical symptoms of methemoglobin and sulphhemoglobin in a single sample of blood. J
cyanosis, erythrocytosis or even haemolytic anaemia, Biol Chem 1938; 126: 655-62
requires detailed investigation using a variety of tech- 21 Wild BJ, Green BN, Cooper EK, Lalloz MRA, Erten S, Stephens
niques, which ultimately may require protein analysis AD, et al. Rapid identification of haemoglobin variants by
using mass spectrometry, or DNA techniques to electrospray ionisation mass spectrometry. Blood Cells Mol Dis
characterize the variants involved. 2001; 27: 691-704

Ann Clin Biochem 2004; 41: 355–369

 Detection and quantitation of normal and variant haemoglobins 369

22 Old JM, Higgs DR. Gene analysis. In: Weatherall DJ, ed. The 30 Grimes AJ, Meisler A. Possible cause of Heinz bodies in congenital
Haemoglobinopathies. London: Baillière Tindall, 1983: 74-102 Heinz body anaemia. Nature 1962; 194: 190
23 Cao A, Rosatelli MC. Screening and prenatal diagnosis of 31 Carrell RW, Kay R. A simple method for the detection of unstable
the haemoglobinopathies. In: Higgs DR, Weatherall DJ, eds. haemoglobins. Br J Haematol 1972; 23: 15
The Haemoglobinopathies. London: Baillière Tindall, 1993: 32 Royal College of Pathologists, Science IoB. The Retention and
263-86 Storage of Pathological Records and Archives. London: The Royal
24 Lewis SM, Bain BJ, Bates I, eds. Dacie and Lewis’ Practical College of Pathologists, 1999
Haematology, 9th edn. London: Churchill Livingstone, 2001 33 British Committee for Standardisation in Haematology. Guidelines
25 Marengo-Rowe AJ. Rapid electrophoresis and quantitation of for haemoglobinopathy screening. Clin Lab Haematol 1988; 10: 335
hemoglobins on cellulose acetate. J Clin Pathol 1965; 18: 790 34 British Committee For Standards in Haematology Working Party of
the General Haematology Task Force. The laboratory diagnosis of
26 International Committee for Standardization in Hematology.
haemoglobinopathies. Br J Haematol 1998; 101: 783-92
Recommendations for neonatal screening for haemoglobinopathies.
35 British Committee For Standardization in Haematology. Guidelines
J Clin Lab Haematol 1988; 10: 335
for the investigation of the a and b thalassaemia traits. J Clin Pathol
27 Hanash SM, Shapiro DN. Separation of human hemoglobins by ion-
1994; 47: 289
exchange high performance liquid chromatography. Hemoglobin
36 British Committee for Standardization in Haematology. Guidelines
1981; 5: 165-75
for the fetal diagnosis of globin gene disorders. J Clin Pathol 1994;
28 Schroeder WA, Shelton JB, Shelton JR, Powars D. Hemoglobin 47: 199-204
Sunshine Seth. Hemoglobin 1979; 3: 145-59 37 NHS Sickle Cell and Thalassaemia Screening Programme. National
29 Schneider R. Differentiation of electrophoretically similar hemoglo- Plan. http://www.kcl-phs.org.uk
bins such as S, D, G and P or A2, C, E and O by electrophoresis of
the globin chains. Clin Chem 1974; 20: 1111 Accepted for publication 10 March 2004

Ann Clin Biochem 2004; 41: 355–369