Large-scale blood group genotyping – clinical implications


  • Neil D. Avent

    1. Centre for Research in Biomedicine and UWE, Bristol Genomics Research Institute, Faculty of Health and Life Sciences, University of the West of England, Bristol, UK
    Search for more papers by this author

Neil D. Avent, Centre for Research in Biomedicine and UWE, Bristol Genomics Research Institute, Faculty of Health and Life Sciences, University of the West of England, Coldharbour lane, Frenchay, Bristol BS16 1QY, UK. E-mail:


The molecular background of blood group antigen expression of the major clinically significant blood group antigens has been largely accomplished. Despite this large body of work, blood group phenotype prediction by genotyping has a marginal supporting role in the routine blood bank. It has however had a major impact in the prenatal determination of fetal blood group status in the management of haemolytic disease of the fetus and newborn. In the past few years several high throughput systems have been in development that have the potential capacity to perform genotyping on a mass scale. Such systems have been designed for use on donor- and patient-derived DNA and provide much more comprehensive information regarding an individuals blood group than is possible by using serological methods alone. DNA-based typing methodology is easier to standardize than serology and has the potential to replace it as a front line diagnostic in blood banks. This review overviews the current situation in this area and attempts to predict how blood group genotyping will evolve in the future.

For over a century blood groups have been defined by the simple, yet highly effective, agglutination test between red cells and blood group specific antisera. This testing strategy has evolved into a cheap, highly efficient means of defining blood group status of both blood donors and patients. Despite this, there are circumstances where blood group serology is surpassed by molecular genetic techniques, most notably during prenatal diagnosis and determination of blood group of multi-transfused patients. During the past two decades, a substantial body of information has been accrued regarding the molecular background of blood group antigens. Whilst some are caused by simple single nucleotide polymorphisms, a large number of variant Rh genotypes cause phenotypic differences in RH antigenicity, including weakened and partial D expression. A similar large number of genotypes account for variant ABO expression, and over fifty O alleles are described. Both of these situations make genotyping for ABO and Rh status challenging, but for Rh in particular, assessment of which RhD allele (partial or variant D) is impossible or very difficult to define serologically. Rh genotyping is thus more comprehensive than serology. Mass scale genotyping, if applied to routine blood donors, would significantly change the management of blood provision. Better matching of donor blood to patient is the most significant benefit. This is primarily because a large numbers of low frequency antigens (or absence of high frequency antigen) are not routinely tested for, and donor-patient mismatches are only detected by serological cross-matching (only if an antibody has been generated) immediately prior to transfusion. This review summarises the progress to date to develop suitable platform technologies that have the potential to deliver blood group genotyping into routine use in blood banks. Throughout the review, terminology used for human blood groups is that of the International Society for Blood Transfusion (ISBT) (Daniels et al, 2007a).

Blood groups and the molecular era

Although blood groups have been known for over 100 years, understanding their molecular basis at the level of the gene is a comparatively recent development. It is beyond the scope of this review to provide a detailed description of the century-plus worth of serological and chemical characterisation of blood groups. Readers with interest in this are referred to references contained within the following section: Tomita and Marchesi (1975) were the first to describe the primary amino acid sequence of a human membrane protein, glycophorin A. This was shortly followed in 1976/77 by the description of the amino acid exchanges necessary for M and N blood group antigenicity, carried by this protein (Wasniowska et al, 1976; Dahr et al, 1977). Protein sequencing alone was not able to provide the information necessary regarding the molecular basis of blood group antigens due to technical limitations. In the 1980s, molecular biology was able to first provide the identities of the proteins involved in blood group antigen expression followed by their cDNA cloning and their biosynthetic enzymes, in the case of carbohydrate-dependent antigens. The first description of a cDNA encoding a blood group active polypeptide species was made in 1986, (glycophorin C, which carries the minor Gerbich blood group antigens). However it wasn’t until 1990 that the molecular basis underlying the two major blood group systems, ABO (Yamamoto et al, 1990) and RH (Avent et al, 1990; Cherif-Zahar et al, 1990), came into focus with the cloning of the AB and O transferase cDNAs and a mRNA encoding an Rh polypeptide.

During the 1990s a huge body of information was uncovered concerning the molecular basis of blood group antigenicity, and a diversity of genetic events was shown to be responsible for the expression of blood group polymorphisms. Some illustrative examples are described here, but the reader is referred to more comprehensive reviews on the molecular biology of blood group antigen expression for details in greater depth (Avent, 2003; Reid & Mohandas, 2004; Daniels, 2005). The majority of blood group polymorphisms are caused by simple single nucleotide polymorphisms (SNPs) in either genes encoding the protein involved in blood group antigenicity or glycosyltransferase, which is an enzyme that catalyses the addition of monosaccharide onto a nascent blood group oligosaccharide moiety (for example the ABO system). An example of a protein-dependent blood group antigen is the K/k (K1/K2) polymorphism, where a SNP in exon 6 of the gene encoding the Kell blood group (KEL) alters codon 193 producing a Met (K) to Thr (k) amino acid exchange (Lee et al, 1995). The major blood group antigens Rh C/c and E/e; Fya/Fyb, Jka/Jkb, S/s are also caused by SNPs in their structural genes, as are a significant number of the known 300+ blood group polymorphisms (Fig 1). Small insertions or deletions of nucleotides causing frame-shift mutations have been shown to cause significant effects on blood group antigen expression, the most notable being the 1060/1061G frameshift nucleotide deletion within the ABO transferase gene (ABO) causing a multitude of O alleles (see Complexities with ABO blood group genetics that lead to variation in the strength of A and B antigen expression have been reviewed extensively (Chester & Olsson, 2001; Yamamoto, 2004). Within the RH system a wide variety of mutations cause altered antigen expression, most notably that of the D antigen. Variation in D expression is due to the fact that RHD is commonly mutated producing the following variations – missense mutations (Weak D, D-elute and partial D phenotypes), nonsense mutations (D-negative phenotypes), splice-site mutations (D-elute phenotype), small deletions, hybrid genes (partial, D-elute and D-negative phenotypes) and a pseudogene (RHDΨ). Rh antigen variability and its molecular biology have been reviewed extensively in the literature (Avent & Reid, 2000; Westhoff, 2005; Avent et al, 2006; Flegel, 2006a; Avent, 2007). This wide degree of D antigen variability has confounded some serological approaches-for example it is extremely difficult to characterise partial D and D-elute phenotypes by serology alone, as a very large panel of monoclonal anti-D representing each of the different anti-epD (D-epitope) specificities would be required, over 50 of which have been described (Scott, 2002). It is known that certain D epitopes are more clinically relevant than others – for example those missing on DVI variant erythrocytes are known to elicit powerful immune responses. Furthermore, determination of the D-elute phenotype requires specialist adsorption and elution tests. RHD genotyping has significant advantages over serology, and therefore represents the best potential application of genotyping to assist red cell typing in the blood bank. There are 49 Rh antigens described to date, but in contrast there are over 200 presently described RH allelic variants defined at the molecular level with new alleles still being discovered (see Fig 1).

Figure 1.

 Numbers of described blood group antigens and alleles of ABO and RH systems. The left hand pie-chart shows the numbers of described blood group antigens for each of the 29 blood group systems as described by the International Society of Blood Transfusion (ISBT) (Daniels et al, 2007a). RH and MNS systems are the most complex as defined serologically, but ABO is relatively simple with just four antigens described. However, the study of ABO and RH genetics reveals a significant level of complexity with a growing list of alleles described in the literature. Information regarding the numbers of alleles of both systems are taken from dbRBC ( and Rhesus base

Blood group serology

For routine applications of typing of patients and donors, serology has been at the forefront of all testing technologies of choice. This is for a host of reasons including reliability, cost and familiarity. For some applications this is not always the case. Some blood group antigens do not have a readily accessible antibody (e.g. is not commercially available), or a poorly standardised polyclonal human serum is often only available, and testing is possible via only specialist laboratories that have limited stocks of it. One of the first clinical applications following the advent of monoclonal antibody (MAb) technology was the production of blood-group specific MAbs (Voak et al, 1980, 1982). Consequently a large number of MAbs with blood group specificity have been generated and are commercially available. These MAbs have been well characterised by a series of workshops organised by the International Society of Blood Transfusion in 1987, 1990, 1996 and 2001 (Scott et al, 1996; Scott, 2002). Almost all routine blood grouping is performed using automated platforms provided by several different commercial suppliers, and using reagents derived from a wide variety of sources. Genotyping platforms must therefore be of similar throughput and greater accuracy to be potential replacements to routine blood group phenotyping.

Blood group genotyping-application and potential applications

Prenatal diagnosis

The initial impetus to translate the recently acquired data surrounding the molecular basis of blood group antigen expression into a clinically applicable diagnostic assay was in the clinical management of Haemolytic disease of the Fetus and Newborn (HDFN). Early determination of feto-maternal blood group incompatibility enabled more focussed obstetric care to at risk antigen-positive pregnancies, and avoided the unnecessary and risky exposure of antigen-negative fetuses to cordocentesis. Initially, DNA extracted from amniocytes, which were normally disposed of during the now obsolete Liley curve investigations, was the first source of material to be utilised, but more recently, analysis based on free fetal DNA in maternal plasma has circumvented the requirement for risky amniocentesis to acquire fetal DNA. As maternal anti-D is the major cause of HDFN, the discovery of RHD and the demonstration that it is absent in the vast majority of D-negative Caucasians immediately led to the development of prenatal RHD typing assays (Bennett et al, 1993; Wolter et al, 1993; Simsek et al, 1994, 1995). Other blood group antigens can cause HDFN, notably Rh c and K (K1) [for review see (Avent et al, 2000)]. The development of maternal plasma-based testing for RhD (Faas et al, 1998; Lo et al, 1998), led to the introduction of clinically applicable tests for RhD using non-invasively sampled fetal material from maternal blood (Finning et al, 2002, 2004; Rouillac-Le Sciellour et al, 2004). Presently, most fetal blood group genotyping is directed toward fetuses that are at risk, namely the mother has become alloimmunised, and assays are performed predominantly during the second trimester.

Feasibility studies have proven the effectiveness of applying mass-scale maternal plasma fetal RHD genotyping to large cohorts of D-negative pregnant women. Much debate has also ensued, arguing that antenatally administered prophylactic anti-D could be conserved, and not be given to mothers carrying D-negative fetuses (van der Schoot et al, 2003; Ait Soussan et al, 2004; Rouillac-Le Sciellour et al, 2004; Bianchi et al, 2005; Van der Schoot et al, 2006; Legler et al, 2007; Urbaniak, 2008). The approach to RHD genotype such fetuses has been to examine at least two RHD exons to minimise the chance of obtaining false-negative results. Whilst one assay is able to type RHDΨ alleles correctly as RhD negative (Finning et al, 2002), others are unable to do this and will incorrectly type the fetus as RhD positive. In this circumstance, the mother will be given prophylactic anti-D unnecessarily, which, in the absence of prenatal RHD genotyping, is the current situation. For greater accuracy some new generation RHD assays may be applicable to testing maternal plasma samples (depending on their availability, the maternal plasma testing market is rigorously protected intellectual property), and the detection of certain maternal Rh variants during pregnancy may be an advantage. It is possible to envisage that mothers that have detectable anti-D may benefit from genotyping to exclude the possibility that they have an Rh variant phenotype. There have been incidences of severe alloimmunisation events resulting in fetal death in partial D phenotype mothers, for example the DVI phenotype (Lacey et al, 1983), which has led to a policy of deliberately serotyping such individuals as D-negative, so that they receive prophylactic anti-D during the pregnancy. RHD genotyping the mother may lead to a further decrease in the incidence of anti-D alloimmunisation by other partial D phenotypes during pregnancy. The frequency of the occurrence of such alloimmunisations has not really been established, but several events have been reported by one group who had monitored this situation (Mayne et al, 1991). The frequency of DVI in European populations is estimated to be at least 1:4000 (Avent et al, 1997). Maternal genotyping may therefore identify partial D phenotype individuals who have the potential to produce anti-D, and may require the administration of prophylactic anti-D during the pregnancy. Furthermore, maternal genotyping may help in the identification of partial D phenotypes that may require some adjustments to the dose of given antenatal prophylactic anti-D. The maternal partial D red cells may adsorb the anti-D, depending on the concentration and epD profile of these cells.

Multi-transfused patients

For those individuals whose quality of life is dependent on regular blood transfusions, for example those suffering from sickle cell disease, alloimmunisation and the resultant problems of matching blood can be a serious problem. Very often these patients are very difficult to serotype as their peripheral blood contains transfused cells, and often they were not fully typed before the first transfusion took place. Several studies have shown the application of polymerase chain reaction using sequence-specific primer (PCR-SSP)-based assays to be effective in the management of such patients (Wenk & Chiafari, 1997; Legler et al, 1999; Reid et al, 2000; Rozman et al, 2000). Often, these patients are of African descent (for example sickle cell patients) and will sometimes make unusual antibodies, for example of the MNS system, the Henshaw antigen (He), is found at a frequency of 3–7% in such individuals (Tippett et al, 1992). Sometimes they will also lack high frequency antigens (e.g. Hr0, Hr), and thus will pose difficulties for matching if antibodies have been made to such. Thus, for effective application of genotyping to such vulnerable individuals it is important that suitably genotyped donors are available. This may be achieved by genotyping a large cohort of regular blood donors at a blood bank, sufficient to provide compatible blood for these patients.

Delayed haemolytic transfusion reactions (DHTR) are sometimes fatal, and are caused by blood group-specific antibodies that are difficult to detect by serological cross matching, and are thus a major issue in multi-transfused patients. Despite the clinical course of DHTR and known ‘culprit’ antibodies, there have been recent descriptions of its occurrence among multi-transfused patients, but of course secondary immunisation can occur after just one blood transfusion – as primary alloimmunisation could have been caused by pregnancy. The most common causes of DHTR include antibodies against antigens in RH, JK (Kidd), FY (Duffy), KEL (Kell) and MNS systems but there have been descriptions of DHTR caused by antibodies to LU (Lutheran), DI (Diego), CO (Colton) and DO (Dombrock) system antigens (Klein & Anstee, 2005).

Elimination of risky units (i.e. those that may potentially cause alloimmunisation) from bloodbanks and identification of at risk patients

There has been some debate regarding the screening of all D-negative blood in blood banks for the presence of rare RHD variants. These units (conventionally typed by serological means as D-negative), if transfused into D-negative recipients, may induce an immune response to D in this vulnerable patient group (for example D-negative women of child bearing age). Whilst there is some doubt as to whether there are sufficient D antigen site numbers on weak D red cells to induce an immune response (Kumpel, 2006), it is worth bearing in mind that the site densities of some partial D red cells are similar to normal D-positive red cells [for example DVI type III (Wagner et al, 1998); DIII and DV (Jones et al, 1996)] and, as such, may induce immune responses. Furthermore, many partial D phenotype red cells express low frequency antigens. Whilst most routinely used monoclonal anti-D for D typing detects most D variants, some notably do not, and may lead to a limited number of partial D units entering the blood banks wrongly typed as D-negative. Certain weak D individuals have made anti-D when transfused with D-positive blood, including weak D types 4·2, 11 and 15, and therefore these individuals should ideally receive D-negative blood. However the common weak Ds (type 1, 2, 3 and 4·1) do not appear to be prone to D immunisation (Flegel, 2006b). In fact, the Rh D protein weak D type 4·2 is identical in predicted primary amino acid sequence to the partial D DAR, so strictly should not be considered a weak D phenotype. Because of such a fuzzy distinction between partial and weak D phenotypes, it has been argued that the term D variant be used instead, with each phenotype considered individually on the merits of its clinical ability to cause alloimmunisation (if transfused) or be alloimmunised by normal D positive blood (if a patient) (Daniels et al, 2007b).

In one Canadian-German study that took place over an 18-month period, 33 864 donors were examined by DNA-based methods for the presence of variant RHD status (Denomme et al, 2005). Fifty-five of these samples were discrepant to serologically recorded RhD status, and 54 of them had variant RHD alleles. Of those that were identified (20) the following phenotypes were recorded: – DV family (7) DAR (8) and four novel alleles, related to DAU (Wagner et al, 2002; Flegel et al, 2008). Confirmed partial D phenotype individuals should only safely receive D-negative blood and should be given prophylactic anti-D after giving birth to a D-positive baby.

Different serological testing combinations have been adopted depending on whether the red cells being tested are derived from a donor or a patient to determine transfusion policy (for example DVI-phenotype mothers, mentioned previously). Genotyping would provide informed choices of how better to manage potentially risky units and identify vulnerable patients, rather than depend on the current approach of detecting antibodies after a possibly preventable primary alloimmunisation event.

Reagent red cells (namely those used for crossmatching and antibody identification tests). Within other blood group systems there are known issues with the use of panel red cells for the detection of blood group specific antibodies, particularly with respect to Fyb antigen expression (Hult et al, 2005). If presumed FybFyb homozygous red cells are used to detect anti-Fyb, then two normal copies of the FY*B allele need to be expressed by the donors that have provided the cells. If the FY*Bweak allele is expressed, then reduced Fyb antigen expression will occur, and may compromise the detection of weak-anti-Fyb antibodies in patient sera. FY*Bweak arises through missense mutations of a normal FY*B allele, which results in weakened Fyb antigen expression (Olsson et al, 1998; Parasol et al, 1998; Tournamille et al, 1998). Other weakening (e.g. weak D) or silencing (null alleles) mutations need to be excluded from presumed homozygous cell panels used for antibody screening too.

Solid organ and stem cell transplantation

It has been known for decades that blood group antigens are expressed on a variety of other tissues, and that ABO matching is critical during transplantation as hyperacute rejection episodes will occur due to the existence of naturally occurring antibodies, anti-A and -B. The role of other blood group antigens as minor histocompatibility antigens has recently been explored (Lerut et al, 2007). This study revealed that mismatched FY and JK renal transplants, whilst although they had identical 10-year survival rates, had some initial signs of rejection. The FY mismatched transplants in particular had an increased rate of chronic lesions, which was also found to a lesser extent in the JK mismatched transplants. This study is significant in that for the first time, it indicates that red cell antigens (although FY and JK are expressed on other tissues) may act as minor histocompatibility antigens. FY matching at least, would be prudent if at all possible during solid organ transplantation.

Blood group compatibility can cause complications during stem cell therapy, and again matching for at the very least ABO and RH will be prudent, as clinical complications have been described in the literature. The generation of anti-D in a D-positive patient given a stem cell transplant from a D-negative donor has been reported (Mijovic, 2002) and anti-A production and other anti-Rh, and have contributed significantly to post-transplantation haemolysis (Sokol et al, 2002; Worel et al, 2002). Naturally, individuals that have had previous stem cell replacement therapy will have a haemopoietic genotype different to that if DNA was isolated from elsewhere in the body, and should be considered carefully if further red cell transfusions are required.

High throughput genotyping systems

In order to meet the demand of routine blood group genotyping of hundreds, if not thousands, of donors or patients per day, genotyping technologies need to be high throughput and, above all, automated, accurate and cost effective. Cost effectiveness must not however be judged simplistically on the raw cost per test. The potential benefits of having a comprehensive genotype of a donor or patient may minimise transfusion complications as alloimmunisation may be reduced. The full economic cost of providing complex serological investigations for such individuals should also be considered. For genotyping, the procedures required will be automated DNA extraction, PCR setup and running (if required by the downstream platform) and all manipulation of post-PCR and processing of the array or similar analytical system. Subsequent to these processes, robust bioinformatics (and thus software) is required to interpret the results of the genotyping, in particular to score combinations of SNPs that are diagnostic of hybrid genes found in the ABO, RH and MNS systems.

Why mass-scale genotyping for blood groups in the future?

DNA-based diagnostics are applied routinely in a large number of clinical- and forensics-based diagnostics. This includes the diagnosis of cancer, pharmacogenomics, detection and subclassification of microorganisms, diagnosis of disease, prenatal diagnosis. Most of these technologies involve the detection of SNPs and also gene copy number analysis, and DNA arrays have moved into a prime position as the technology platform of choice for the detection of these mutations (Beaudet & Belmont, 2007). With such widespread application there is almost universal acceptance that the defined genotype in the diagnostic assay always correlates with phenotype, be it for disease (for example factor V Leiden, Cystic fibrosis, Down syndrome, mutations in cancer) or the detection of the presence of a microorganism, although the problems associated with false positive and false negative results are well known. In particular, the latter assays have found widespread use in transfusion medicine for the nucleic acid testing (NAT)-based strategies for detection human immunodeficiency virus, hepatitis B virus and hepatitis C virus in plasma minipools (Ratcliff et al, 2007). NAT-based technology used in blood banks are high throughput genotyping on a mass scale, proving that these technical challenges are not an issue to the rapid provision of blood within 24 h or less after donation.

Genotyping for blood groups should therefore be a simple adaptation for blood banks to make. However, a century plus of blood group serology does not make this transition seamless, and there is some perceived resistance to the widespread application of DNA-based approaches. In my view this is a great shame, and widespread adoption of blood group genotyping would lead to safer transfusions. A common criticism of blood group genotyping, and widely stated in reviews and at conferences within the transfusion medicine community is that ‘genotype does not always reflect phenotype’. This is a misconception; in circumstances where defined genotype does not correlate with serological phenotype it is commonly due to the application of an incorrect genotyping strategy or, frequently, by the discovery of an unknown allele. In certain circumstances the reasons for the lack of expression of a particular blood group antigen are complex, for example in the very rare Rhnull phenotype, no Rh antigens are expressed but RHCE and RHD are normal. Rhnull mutations are caused by alterations in the RHAG structural gene [for review see (Cartron, 1999)], which acts as a chaperone in assembling the Rh core complex trimer that is composed of two subunits of RhAG and one of either RhD or RhCE (Conroy et al, 2005; Callebaut et al, 2006). Null phenotypes occur in other blood group systems, most commonly in ABO where over 50 null genotypes have been described.

Technology available for blood group genotyping

This review considers high throughput approaches to blood grouping, namely those that are array-based. This has arisen from over a decade of work developing single-sequence primer based approaches for blood grouping that are now in widespread use. Some of these are commercially available, for example produced by BAGene (Prager, 2007). The Beckman-Coulter genomelab SNPStream system has also been adapted for blood group and platelet antigen genotyping (Denomme & Van Oene, 2005). However, the focus in the remainder of the review will be on mass scale blood group genotyping systems, of which two that are currently commercially available are described here. Both of these systems are described as ‘high throughput’, but both systems will require amendments including automation (robotic PCR set up and genomic DNA extraction) before they could be utilised into routine genotyping in the blood bank.

BLOODchip – the Bloodgen project

From 2003–2006, framework V of the European commission funded a demonstration project, the objective of which was to illustrate the applicability of glass array based genotyping approaches to the determination of a large number of blood group polymorphisms. The Bloodgen consortium (Avent et al, 2007) comprises academic laboratories, blood banks and Progenika Biopharma SA, a Spanish biotech company that specialises in genotyping and personalised medicine. The project developed the commercially available product, BLOODchip, which is an oligonucleotide array with multiple probes corresponding to allelic pairs of blood-specific SNPs. PCR products corresponding to DNA sequences flanking each SNP are amplified by a multiplex amplifiable probe hybridisation (MAPH)-based multiplex reaction (Beiboer et al, 2005). The PCR products are fragmented, fluorescently labelled, and then hybridised to probes that are arranged on a glass array. Differential hybridisation to allelic probe pairs for each SNP is then detected using a conventional scanner, and then genotyping is completed by software devised to detect hetero or homozygosity for each SNP (Fig 2). Bloodchip includes ABO (33 haplotypes) RHD (91 haplotypes) RHCE (9 alleles), KEL (8 alleles), JK (4 alleles, including 2 JKnull) FY (4 alleles), MNS (9 haplotypes), DI, DO and CO. The major content of the array are the various RHD alleles that cause D-negative, partial, weak D and D-elute phenotypes. BLOODchip has recently been CE-marked for diagnostic purposes in the European Union (currently for RHCE, KEL, FY, JK, CO, MNS, DI and DO) with RHD in progress, and this clinical validation proved the superiority of this platform over conventional serology. Thousand samples, which included patient samples, newborns, weak D and A or B positive samples were analysed by BLOODchip and 116 different SNPs were analysed for each sample and phenotypes scored by the detection of each SNP and the inheritance of a combination of SNPs (for example in hybrid gene type partial D phenotypes). Only two errors in genotype compared to phenotype were observed, one RhC scoring and one Kpa/Kpb scoring. The mis-typed RhC sample was later resolved by an improved fragmentation protocol, although the reasons for the incorrect Kpa/Kpb typing are unclear. Two unknown combinations of SNPs were found in the ABO system, and six in the RH system and are currently being investigated by DNA sequence analysis. BLOODchip demonstrated effectively that a high throughput genotyping technology has superior accuracy over serology, for Rh CcEe typing 999/1000 were scored correctly by genotyping, whereas 995/1000 were concordant by serology. For the following blood groups there was 100% concordance between BLOODchip defined genotype and serological phenotype, (except 1/358 Kpa/Kpb typings). In contrast serological testing recorded the following concordances K/k (1000/1000); Kpa/Kpb (358/358); Jsa/Jsb (122/123); Jka/Jkb (596/597); FY serotypes (498/506), MN (425/445), Ss (479/483), CO (169/170). DI and DO types were confirmed by DNA sequence analysis due to the unavailability of serological reagents. At present the source of serological errors, which could either be technical or clerical, is unclear (see Table I).

Figure 2.

 Technical approach using one commercially available blood group genotyping platform – BLOODchip. Genomic DNA extracted from peripheral blood is (1) subjected to multiplex amplifiable probe hybridisation (MAPH) multiplex (MPX) polymerase chain reaction (PCR) where the PCR contains a mix of gene-specific primers (3 here shown for clarity), and primers which are tagged with MAPH sequences. The MAPH tagged primer sets are at a higher concentration in the MPX and ensures uniform amplification of all PCR products due to their common sequence. (2) The MPX PCR products are fragmented, labelled and made single stranded by denaturation. Fluorescent dyes are incorporated into the DNA. (3) The labelled DNA is hybridised to a BLOODchip array using an automated hybridisation station (for example Ventana discovery), and arrays are rinsed and scanned using a conventional scanner (4) Bespoke software analyses the data output (there are multiple probes per SNP, and positive and negative controls) and scores the spot as ‘good’ hybridisation (perfect match) or ‘poor’ (imperfect match). Zygosity for each SNP is devised using a patented algorithm. The software then interprets combinations of SNPs and converts this into a phenotype where these combinations are known. This is simple for KEL/DARC/GYPA(/B/E)/SLC4A1/ART$/AQP1 genotyping but complex for ABO and RH genotypes.

Table I.   BLOODchip CE marking (KEL, FY, JK, DI, DO, CO, MNS) results – genotyping versus serological testing.
Blood group antigen (ISBT System)Serological testingBLOODchipComments
  1. Figures in bold indicate full concordance.

Rh C/c (RH) 998/1000999/1000 
Rh E/e (RH) 997/10001000/1000Includes one typographical error
CW+ (RH)18*28*Serological confirmation of genotype
CX+ (RH)0*2*Serological confirmation of genotype
VS+ (RH)9*15*Serological confirmation of genotype
K/k (KEL)1000/10001000/1000 
Kpa/Kpb (KEL)358/358 357/358 
Jsa/Jsb (KEL) 122/123123/123 
Jka/Jkb (JK) 596/597 597/597 
Fya/Fyb/Fy(a-b-) (FY) 498/506 506/506 
MN (MNS) 425/455 455/455 
Ss (MNS) 479/483 483/483 
Dia/Dib (DI)  120/120Confirmed by DNA sequencing
Doa/Dob/(DO)  120/120Confirmed by DNA sequencing
Coa/Cob (CO) 169/170 170/170 

Human erythrocyte antigen testing platform

Bioarray Solutions Ltd. (Warren, NJ, USA) has produced a human erythrocyte antigen (HEA) typing platform that is based around functionalised colour-coded beads, and have attached oligonucleotide probes. PCR products are amplified for DNA flanking the regions surrounding each SNP. This is then denatured and then annealed to the derivatized beads. Then a short extension reaction incorporates a labelled dNTP onto the 3′ end of the probe, and the labelled bead is then placed on a beadchip array. After reading, the genotype is determined. The HEA testing system currently includes RHCE, KEL, FY, DO, LW, CO, SC, LU, DI, JK, MNS and HbS (Hashmi et al, 2005), but does not include ABO or RHD testing at present, which are highly clinically significant. Using this platform, a cohort of 2355 donors whom had been phenotyped serologically, were genotyped for the minor blood group antigens MNS, LU, KEL, FY, JK, DO and CO. Twenty-four discordant results were obtained, the bulk of which were assumed to be typographical errors, and eight further discordant samples were found to be due to GYPB (Ss) silencing mutations by DNA sequencing. This led to modifications of the system to include two further GYPB silencing mutations being added (Hashmi et al, 2007).

Future application

There is considerable ongoing speculation that blood group genotyping will replace serology in its entirety over the next decade or so. In reality, this may be a long time coming, but advances in genotyping technology over the past decade have been astonishing, and will not cease. In contrast, serological methods reached their peak several decades ago, and a possible way forward would be antibody array technology, which would permit ‘multiplexing’ of serological reactions for detection of blood group specific antibodies in patient sera (Campbell et al, 2006; Petrik, 2006; Robb et al, 2006). There is no doubt however that, for the foreseeable future, serological methods will remain a normal procedure among transfusion medicine specialist, as the primary method for detecting blood group-specific antibodies in patient serum. Recombinant blood group antigens may offer opportunities to replace reagent (panel) red cells and methods demonstrating the effectiveness of this approach for FY (Sheffield et al, 2006; Ridgwell et al, 2007), Gerbich (Jaskiewicz et al, 2002; Schawalder et al, 2004) and Lutheran and Kell (Ridgwell et al, 2007) have been published. However, recombinant Rh antigens pose the biggest challenge for such antibody detection technology as anti-Rh are very clinically significant, and any array-based system must include this on the platform to have any useful application in transfusion medicine.

Mass genotyping to diagnose genetically inherited diseases is now a reality, particularly in newborn screening. In several US states it is mandatory for such diseases as cystic fibrosis, phenylketonuria, and medium chain acyl-CoA dehydrogenase deficiency (Green et al, 2006). For diseases where there are therapeutic prevention regimens available, this approach is crucial, but where there is no effective treatment and the inheritance of a genotype is merely an indicator of a high risk of disease then the potential benefits are less obvious.

For blood group and, for that matter, human leucocyte antigen (HLA) status determination, the benefits to the individual are clear if that information is present on their medical records from birth. There is a good case for concerted HLA and blood group determination in the management of organ and stem cell transplants as mentioned earlier. In that scenario, individuals that become patients would be able to receive exactly or closely matched blood, and their comprehensive blood group genotype to be already defined would also greatly benefit the blood donating process if they chose to become donors. For blood grouping and HLA-typing to be included in a high-throughput genotyping system that includes screening for genetically inherited diseases, then significant changes in donor and patient testing can be envisaged; indeed it may not be necessary for blood banks or hospitals to determine blood groups at all. If organ donation moves to a situation whereby consent is presumed, then comprehensive HLA typing of the population is extremely desirable. Comprehensive genotyping thus would be a driver to implement rapid ‘electronic’ cross matching in instances of trauma, and potential elimination of any form of laboratory test prior to an emergency blood transfusion.

Concluding remarks

It is now clear that DNA-based determination of blood group status is a highly useful addition to the armoury of transfusion medicine diagnostics to provide safe and effective blood transfusion. It remains to be determined whether genotyping will emerge as a contender to replace serology in its entirely, or will remain as a secondary supportive role. Alternatively, blood group genotyping may completely replace serology for all blood groups except perhaps ABO typing, which is simple and cheap to define serologically. Furthermore, there is reluctance among blood bankers to switch immediately to ABO genotyping because of the known complexity of this system at the genetic level, and the relative simplicity of routine serology. However, it must be stressed that DNA-based technologies have massively advanced in a relatively short space of time, whilst there have been only modest advances in blood group serology based technologies. With the emergence of rapid and accurate genome resequencing techniques, the possibility that rapid sequencing of an individuals complete genome emerges as a distinct possibility. With this in mind, I have no doubt that we are witnessing the beginning of the end for red cell serology as the frontline method for defining blood group antigen expression.


NDA is a member of the Scientific Advisory board for Progenika Biopharma SA.