Bristol Institute for Transfusion Sciences, NHS Blood and Transplant, Bristol, UK
A blood group may be defined as an inherited character of the red cell surface detected by a specific alloantibody. This definition would not receive universal acceptance as cell surface antigens on platelets and leucocytes might also be considered blood groups, as might uninherited characters on red cells defined by autoantibodies or xenoantibodies. The definition is suitable, however, for the purposes of this chapter.
Most blood groups are organized into blood group systems. Each system represents a single gene or a cluster of two or more closely linked homologous genes. Of the 339 blood group specificities recognized by the International Society for Blood Transfusion, 297 belong to one of 33 systems (Table 3.1). All these systems represent a single gene, apart from Rh, Xg and Chido/Rodgers, which have two closely linked homologous genes, and MNS with three genes [1,2].
Table 3.1 Human blood group systems.
Most blood group antigens are proteins or glycoproteins, with the blood group specificity determined primarily by the amino acid sequence, and most of the blood group polymorphisms result from single amino acid substitutions, though there are many exceptions. Some of these proteins cross the membrane once, with either the N-terminal or C-terminal outside the membrane, some cross the membrane several times and some are outside the membrane to which they are attached by a glycosylphosphatidylinositol anchor. Some blood group antigens, including those of the ABO, P1PK, Lewis, H and I systems, are carbohydrate structures on glycoproteins and glycolipids. These antigens are not produced directly by the genes controlling their polymorphisms, but by genes encoding transferase enzymes that catalyse the final biosynthetic stage of an oligosaccharide chain.
The two most important blood group systems from the clinical point of view are ABO and Rh. They also provide good models for contrasting carbohydrate- and protein-based blood group systems.
The ABO system
ABO is often referred to as a histo-blood group system because, in addition to being expressed on red cells, ABO antigens are present on most tissues and in soluble form in secretions. At its most basic level, the ABO system consists of two antigens, A and B, indirectly encoded by two alleles, A and B, of the ABO gene. A third allele, O, produces neither A nor B. These three alleles combine to effect four phenotypes: A, B, AB and O (Table 3.2).
Table 3.2 The ABO system.
Two key factors make ABO the most important blood group system in transfusion medicine. First, almost without exception, the blood of adults contains antibodies to those ABO antigens lacking from their red cells (Table 3.2). In addition to anti-A and anti-B, group O individuals have anti-A,B, an antibody to a determinant common to A and B. Second, ABO antibodies are IgM, though they may also have an IgG component, have thermal activity at 37°C, activate complement and cause immediate intravascular red cell destruction, which can give rise to severe and often fatal haemolytic transfusion reactions (HTRs) (see Chapter 7). Major ABO incompatibility (i.e. donor red cells with an ABO antigen not possessed by the recipient) must be avoided in transfusion and, ideally, ABO-matched blood (i.e. of the same ABO group) would be provided.
ABO antibodies seldom cause haemolytic disease of the fetus and newborn (HDFN) but when they do it is usually mild. The prime reasons for this are (1) that IgM antibodies do not cross the placenta, (2) IgG ABO antibodies are often IgG2, which do not activate complement or facilitate phagocytosis and (3) ABO antigens are present on many fetal tissues and in body fluids, so the haemolytic potential of the antibody is greatly reduced.
A and B subgroups
The A (and AB) phenotype can be subdivided into A1 and A2 (and A1B and A2B). In a European population, about 80% of group A individuals are A1 and 20% A2 (Table 3.2). A1 and A2 differ quantitatively and qualitatively. A1 red cells react more strongly with anti-A than A2 cells. In addition, A2 red cells lack a component of the A antigen present on A1 cells and some individuals with the A2 or A2B phenotype produce anti-A1, an antibody that agglutinates A1and A1B cells, but not A2 or A2B cells. Anti-A1 is seldom reactive at 37°C and is generally considered clinically insignificant.
There are numerous other ABO variants, involving weakened expression of A or B antigens, but all are rare. They are often subdivided into categories (A3, Ax, Am, Ael, B3, Bx, Bm and Bel) based on serological characteristics, but molecular analyses have shown that most of the subgroups are genetically heterogeneous.
Biosynthesis and molecular genetics
Red cell A and B antigens are expressed predominantly on oligosaccharide structures on integral membrane glycoproteins, mainly the anion transporter band 3 and the glucose transporter GLUT1, but are also on glycosphingolipids embedded in the membrane. The tetrasaccharides that represent the predominant form of A and B antigens on red cells are shown in Figure 3.1, together with their biosynthetic precursor, the H antigen, which is abundant on group O red cells. The product of the A allele is a glycosyltransferase that catalyses the transfer of N-acetylgalactosamine (GalNAc) from a nucleotide donor substrate, UDP-GalNAc, to the fucosylated galactose (Gal) residue of the H antigen, the acceptor substrate. The product of the B allele catalyses the transfer of Gal from UDP-Gal to the fucosylated Gal residue of the H antigen. GalNAc and Gal are the immunodominant sugars of A and B antigens, respectively. The Oallele produces no transferase, so the H antigen remains unmodified.
Fig 3.1 Diagram of the oligosaccharides representing H, A, B, Lea and Leb antigens, and the biosynthetic precursor of H and Lea. R, remainder of molecule.
The ABO gene on chromosome 9 consists of seven exons. The A1 and B alleles differ by seven nucleotides in exons 6 and 7, four of which, in exon 7, encode amino acid substitutions in their glycosyltransferase products: Arg176, Gly235, Leu266 and Gly268 in A; Gly176, Ser235, Met266 and Ala268 in B . It is primarily the amino acids at positions 266 and 268 that determine whether the gene product is a GalNAc-transferase (A) or Gal-transferase (B). The most common O allele (O1) has an identical sequence to A1, apart from a single nucleotide deletion in exon 6, which shifts the reading frame and introduces a translation stop codon before the region of the catalytic site, so that any protein produced would be truncated and have no enzyme activity. Another common O allele, called O1v, differs from O1 by at least nine nucleotides, but has the same single nucleotide deletion as that in O1 and so cannot produce any functional enzyme. O2, which represents about 3% of O alleles in a European population, does not have the nucleotide deletion characteristic of most O alleles and encodes a complete protein product, but with a charged arginine residue instead of a neutral glycine (A) or alanine (B) at position 268, which completely blocks the donor GalNAc-binding site and prevents any enzyme activity. The A2 allele has a sequence almost identical to A1, but has a single nucleotide deletion immediately before the translation stop codon. The resultant frame shift abolishes the stop codon, so the protein product has 21 extra amino acids at its C-terminal, which reduces the efficiency of its GalNAc-transferase activity and might alter its acceptor substrate specificity .
Biochemically related blood group systems – H, Lewis and I
H antigen is the biochemical precursor of A and B (Figure 3.1). It is synthesized by an α1,2-fucosyltransferase, which catalyses the transfer of fucose from its donor substrate to the terminal Gal residue of its acceptor substrate. Without this fucosylation neither A nor B antigens can be made. Two genes, active in different tissues, produce α1,2-fucosyltransferases: FUT1, active in mesodermally derived tissues and responsible for H on red cells, and FUT2, active in endodermally derived tissues and responsible for H in many other tissues and in secretions. Homozygosity for inactivating mutations in FUT1 leads to an absence of H from red cells and, therefore, an absence of red cell A or B, regardless of ABO genotype. Such mutations are rare, as are red cell H-deficient phenotypes. In contrast, inactivating mutations in FUT2 are relatively common and about 20% of White Europeans (nonsecretors) lack H, A and B from body secretions despite expressing those antigens on their red cells. Very rare individuals who have H-deficient red cells and are also H nonsecretors (Bombay phenotype) produce anti-H together with anti-A and -B and can cause a severe transfusion problem.
Antigens of the Lewis system are not produced by erythroid cells, but become incorporated into the red cell membrane from the plasma. Their corresponding antibodies are not usually active at 37°C and are not generally considered clinically significant. Lea and Leb are not the products of alleles. The Lewis gene (FUT3) product is an α1,3/4-fucosyltransferase that transfers fucose to the GlcNAc residue of the secreted precursor of H in nonsecretors to produce Leaand to secreted H in secretors to produce Leb (Figure 3.1). Consequently, H secretors are Le(a–b+) or Le(a+b+), H nonsecretors are Le(a+b–) and individuals homozygous for FUT3 inactivating mutations (secretors or nonsecretors) are Le(a–b–).
I antigen represents branched N-acetyllactosamine (Galβ1–4GlcNAc) structures in the complex carbohydrates that also express H, A and B antigens. The I gene (GCNT2) encodes a branching enzyme, which only becomes active during the first months of life. Consequently, red cells of neonates are I-negative. Rare individuals are homozygous for inactivating mutations in GCNT2 and never form I on their red cells . This phenotype, called adult i, is associated with production of anti-I, which is usually only active below 37°C, but may occasionally be haemolytic at body temperature.
The Rh system
Rh is the most complex of the blood group systems, with 54 specificities. The most important of these is D (RH1).
Rh genes and proteins
The antigens of the Rh system are encoded by two genes, RHD and RHCE, which produce D and CcEe antigens, respectively. The genes are highly homologous, each consisting of 10 exons. They are closely linked, but in opposite orientations, on chromosome 1  (Figure 3.2). Each gene encodes a 417 amino acid polypeptide that differs by only 31–35 amino acids, according to Rh genotype. The Rh proteins are palmitoylated, but not glycosylated, and span the red cell membrane 12 times, with both termini inside the cytosol and with six external loops, the potential sites of antigenic activity (Figure 3.2).
Fig 3.2 Diagrammatic representation of the Rh genes, RHD and RHCE, shown in opposite orientations as they appear on the chromosome, and of the two Rh proteins in their probable membrane conformation, with 12 membrane-spanning domains and 6 extracellular loops expressing D, C/c and E/e antigens.
The most significant Rh antigen from the clinical point of view is D. About 85% of White people are D+ (Rh-positive) and 15% are D– (Rh-negative). In Africans, only about 3–5% are D– and in the East Asia D– is rare.
The D– phenotype is usually associated with the absence of the whole D protein from the red cell membrane. This explains why D is so immunogenic, as the D antigen comprises numerous epitopes on the external domains of the D protein. In White people, the D– phenotype almost always results from homozygosity for a complete deletion of RHD. D-positives are either homozygous or heterozygous for the presence of RHD. In Africans, in addition to the deletion of RHD, D– often results from an inactive RHD (called RHDΨ) containing translation stop codons within the reading frame. Other genes containing inactivating mutations are also found in D– Africans and Asians.
Numerous variants of D are known, though most are rare . They are often split into two types, partial D and weak D, though this dichotomy is not adequately defined and of little value for making clinical decisions. Partial D antigens lack some or most of the D epitopes. If an individual with a partial D phenotype is immunized by red cells with a complete D antigen, they might make antibodies to those epitopes they lack. The D epitopes comprising partial D may be expressed weakly or may be of normal or even enhanced strength. Weak D antigens appear to express all epitopes of D, but at a lower site density than normal D. D variants result from amino acid substitutions in the D protein occurring either as a result of one or more mis-sense mutations in RHD or from one or more exons of RHD being exchanged for the equivalent exons of RHCE in a process called gene conversion.
Anti-D is almost never produced in D– individuals without immunization by D+ red cells. However, D is highly immunogenic and at least 30% of D– recipients of transfused D+ red cells make anti-D. Anti-D can cause severe immediate or delayed HTRs and D+ blood must never be transfused to a patient with anti-D.
Anti-D is the most common cause of severe HDFN. The effects of HDFN caused by anti-D are, at its most severe, fetal death at about the seventeenth week of pregnancy. If the infant is born alive, the disease can result in hydrops and jaundice. If the jaundice leads to kernicterus, this usually results in infant death or permanent cerebral damage. The prevalence of HDFN resulting from anti-D has been substantially reduced by anti-D immunoglobulin prophylaxis. In 1970, at the beginning of the anti-D prophylaxis programme, there were 1.2 deaths per thousand births in England and Wales due to HDFN caused by anti-D; by 1989, this figure had been reduced to 0.02.
Prediction of fetal Rh genotype by molecular methods
Knowledge of the molecular bases for D– phenotypes has made it possible to devise tests for predicting fetal D type from fetal DNA. This is a valuable tool in assessing whether the fetus of a woman with anti-D is at risk from HDFN . Most methods involve PCR tests that detect the presence or absence of RHD. It is important that the tests are devised so that RHDΨ and other variant RHD genes do not give false phenotype predictions. Initially, the usual source of fetal DNA was amniocytes obtained by amniocentesis, which has an inherent risk of fetal loss and of feto-maternal haemorrhage. A far superior source of fetal DNA, avoiding invasive procedures, is the small quantity of free fetal DNA present in maternal plasma. This noninvasive form of fetal D typing is now provided as a reference service in some countries for alloimmunized D– women. In addition, in a few European countries noninvasive fetal RHDgenotyping is offered to all D– pregnant women, so that only those with a D+ fetus receive routine antenatal anti-D prophylaxis (see Chapter 32).
C and c, E and e
C/c and E/e are two pairs of antigens representing alleles of RHCE. The fundamental difference between C and c is a serine–proline substitution at position 103 in the second external loop of the CcEe protein (Figure 3.2), though the situation is more complex than that . E and e represent a proline–alanine substitution at position 226 in the fourth external loop. Taking into account the presence and absence of D and of the C/c and E/e polymorphisms, eight different haplotypes can be recognized. The frequencies of these haplotypes and the shorthand symbols often used to describe them are shown in Table 3.3.
Table 3.3 Rh phenotypes and the genotypes that produce them (presented in DCE and shorthand terminology).
Anti-c is clinically the most important Rh antibody after anti-D and may cause severe HDFN. On the other hand, anti-C, -E and -e rarely cause HDFN and when they do the disease is generally mild, though all have the potential to cause severe disease.
Other Rh antigens
Of the 54 Rh antigens, 20 are polymorphic, i.e. have a frequency between 1 and 99% in at least one major ethnic group, 22 are rare antigens and 12 are very common antigens. Antibodies to many of these antigens have proved to be clinically important and it is prudent to treat all Rh antibodies as being potentially clinically significant .
Other blood group systems
Of the remaining blood group systems (Table 3.1), the most important clinically are Kell, Duffy, Kidd, Diego and MNS, and these are described below.
The Kell system
The original Kell antigen, K (KEL1) (Met193), has a frequency of about 9% in Caucasians, but is rare in other ethnic groups. Its antithetical (allelic) antigen, k (KEL2) (Thr193), is common in all populations. The remainder of the Kell system consists of one triplet and five pairs of allelic antigens – Kpa, Kpb and Kpc; Jsa and Jsb; K11 and K17; K14 and K24; VLAN and VONG; KYO and KYOR – plus 17 high frequency and 3 low frequency antigens. Almost all represent single amino acid substitutions in the Kell glycoprotein.
Anti-K can cause severe HTRs and HDFN. About 10% of K– patients who are given one unit of K+ blood produce anti-K, making K the next most immunogenic antigen after D. In most cases of HDFN caused by anti-K, the mother will have had previous blood transfusions. HDFN caused by anti-K differs from Rh HDFN in that anti-K appears to cause fetal anaemia by suppression of erythropoiesis, rather than immune destruction of mature fetal erythrocytes. Anti-k is a very rare antibody. It is always immune and has been incriminated in some cases of mild HDFN . Most other Kell system antibodies are rare and best detected by an antiglobulin test.
The Kell antigens are located on a large glycoprotein, which crosses the cell membrane once and has a glycosylated, C-terminal extracellular domain, maintained in a folded conformation by multiple disulfide bonds. The Kell glycoprotein belongs to a family of endopeptidases, which process biologically important peptides, and is able to cleave the biologically inactive peptide big endothelin-3 to produce endothelin-3, an active vasoconstrictor.
The Duffy system
Fya and Fyb represent a single amino acid substitution (Gly42Asp) in the extracellular N-terminal domain of the Duffy glycoprotein. Their incidence in Caucasians is Fya 68%, Fyb 80%. About 70% of African Americans and close to 100% of West Africans are Fy(a–b–) (Table 3.4). They are homozygous for an FY*B allele containing a mutation in a binding site for the erythroid-specific GATA-1 transcription factor, which means that Duffy glycoprotein is not expressed in red cells, though it is present in other tissues  (Table 3.5). The Duffy glycoprotein is the receptor exploited by Plasmodium vivax merozoites for penetration of erythroid cells. Consequently, the Fy(a–b–) phenotype confers resistance to P. vivax malaria. The Duffy glycoprotein (also called Duffy antigen chemokine receptor, DACR) is a red cell receptor for a variety of chemokines, including interleukin-8.
Table 3.4 The Duffy system: phenotypes and genotypes.
Table 3.5 Nucleotide polymorphisms in the promoter region and in exon 2 of the three common alleles of the Duffy gene.
Anti-Fya is not infrequent and is found in previously transfused patients who have usually made other antibodies. Anti-Fyb is very rare. Both may cause acute or delayed HTRs and HDFN varying from mild to severe .
The Kidd system
Kidd has two common alleles, JK*A and JK*B, which represent a single amino acid change (Asp280Asn) in the Kidd glycoprotein. Both Jka and Jkb antigens have frequencies of about 75% in Caucasian populations. A Kidd-null phenotype, Jk(a–b–), results from homozygosity for inactivating mutations in the Kidd gene, SLC14A1. It is very rare in most populations, but reaches an incidence of greater than 1% in Polynesians. The Kidd glycoprotein is a urea transporter in red cells and in renal endothelial cells.
Anti-Jka is uncommon and anti-Jkb is very rare, but they both cause severe transfusion reactions and, to a lesser extent, HDFN . Kidd antibodies have often been implicated in delayed HTRs. They are often difficult to detect serologically and tend to disappear rapidly after stimulation.
The Diego system
Diego is a large system of 22 antigens: three pairs of allelic antigens – Dia and Dib (Leu854Pro), Wra and Wrb (Lys658Glu), Wu and DISK (Gly565Ala) – plus 16 antigens of very low frequency. All represent single amino acid substitutions in band 3, the red cell anion exchanger. The original Diego antigen, Dia, is very rare in Caucasians and Black people, but relatively common in Mongoloid people, with frequencies varying between 1% in Japanese and 50% in some native South Americans. Anti-Dia and -Dib are immune and rare, and can cause severe HDFN . Wra has a frequency of about 0.1%. Its high frequency allelic antigen, Wrb, is dependent on an interaction of band 3 with glycophorin A for expression. Naturally occurring anti-Wra is present in approximately 1% of blood donors and can cause severe HTRs. Very rarely, anti-Wra causes HDFN. Autoanti-Wrb is a relatively common autoantibody and may be implicated in autoimmune haemolytic anaemia.
The MNS system
MNS, with a total of 46 antigens, is second only to Rh in complexity. These antigens are present on one or both of two red cell membrane glycoproteins, glycophorin A (GPA) and glycophorin B (GPB). They are encoded by two homologous genes, GYPA and GYPB, on chromosome 4.
The M and N antigens, both with frequencies of about 75%, differ by amino acids at positions 1 and 5 of the external N-terminus of GPA (Ser1Leu, Gly5Glu). S and s have frequencies of about 55% and 90%, respectively, in a Caucasian population and represent an amino acid substitution in GPB (Met29Thr). About 2% of Black West Africans and 1.5% of African Americans are S– s–, a phenotype virtually unknown in other ethnic groups, and most of these lack the U antigen, which is present when either S or s is expressed. The numerous MNS variants mostly result from amino acid substitutions in GPA or GPB and from hybrid GPA–GPB molecules, formed by intergenic recombination between GYPA and GYPB. The phenotypes resulting from these hybrid proteins are rare in Europeans and Africans, but the GP.Mur (previously Mi.III) phenotype occurs in up to 10% of East Asians. GPA and GPB are exploited as receptors by the malaria parasite Plasmodium falciparum.
Anti-M and -N are not generally clinically significant, though anti-M is occasionally haemolytic . Anti-S, the rarer anti-s and anti-U can cause HDFN and have been implicated in HTRs. Although rare elsewhere, anti-Mur, which detects red cells of the GP.Mur phenotype, is common in East Asia and Oceanic regions and causes severe HTRs and HDFN.
The biological significance of blood group antigens
The functions of several red cell membrane protein structures bearing blood group antigenic determinants are known, or can be deduced from their structure. Some are membrane transporters, facilitating the transport of biologically important molecules through the lipid bilayer: band 3 membrane glycoprotein, the Diego antigen, provides an anion exchange channel for HCO3− and Cl− ions; the Kidd glycoprotein is a urea transporter; the Colton glycoprotein is aquaporin 1, a water channel; the GIL antigen is aquaporin 3, a glycerol transporter; the Junior and Lan antigens are possibly porphyrin transporters. The Lutheran, LW and Indian (CD44) glycoproteins are adhesion molecules, possibly serving their primary functions during erythropoiesis. The MER2 antigen is located on the tetraspanin CD151, which associates with integrins within basement membranes, but its function on red cells is not known. The Duffy glycoprotein is a chemokine receptor and could function as a ‘sink’ or scavenger for unwanted chemokines. The Cromer and Knops antigens are markers for the decay accelerating factor (CD55) and complement receptor 1 (CD35), respectively, which protect the cells from destruction by autologous complement. Some blood group glycoproteins appear to be enzymes, though their functions on red cells are not known: the Yt antigen is acetylcholinesterase, the Kell antigen is an endopeptidase and the sequence of the Dombrock glycoprotein suggests that it belongs to a family of adenosine diphosphate (ADP)-ribosyltransferases. The C-terminal domains of the Gerbich antigens, GPC and GPD, and the N-terminal domain of the Diego glycoprotein, band 3, are attached to components of the cytoskeleton and function to anchor it to the external membrane. The carbohydrate moieties of the membrane glycoproteins and glycolipids, especially those of the most abundant glycoproteins, band 3 and GPA, constitute the glycocalyx, an extracellular coat that protects the cell from mechanical damage and microbial attack [12,13].
The Rh proteins are associated as heterotrimers with the glycoprotein RhAG in the red cell membrane, and these trimers are part of a macrocomplex of red cell surface proteins that include tetramers of band 3 plus LW, GPA, GPB and CD47, and are linked to the red cell cytoskeleton through protein 4.2 and ankyrin. There is probably another protein complex containing Rh proteins and dimers of band 3, plus Kell, Kx and Duffy blood group proteins, and is linked to the cytoskeleton through glycophorin C (Gerbich blood group), MMP1, and protein 4.1R. It is likely that RhAG forms a carbon dioxide and, possibly, oxygen channel, and could function as an ammonia/ammonium transporter [13,14]. The function of the Rh proteins is not known, but they may play a role in facilitating the assembly of the band 3 macrocomplex.
The structural differences between antithetical red cell antigens (e.g. A and B, K and k, Fya and Fyb) are small, often being just one monosaccharide or one amino acid. The biological importance of these differences is unknown and there is little evidence to suggest that the product of one allele confers any significant advantage over the other. Some blood group antigens are exploited by pathological microorganisms as receptors for attaching and entering cells, so in some cases absence or changes in these antigens could be beneficial. It is likely that interaction between cell surface molecules and pathological microorganisms has been a major factor in the evolution of blood group polymorphism.
1. The International Society of Blood Transfusion recognizes 339 blood group specificities, 297 of which belong to one of 33 blood group systems.
2. The most important blood group systems clinically are ABO, Rh, Kell, Duffy, Kidd and MNS.
3. ABO antibodies are almost always present in adults lacking the corresponding antigens and can cause fatal intravascular HTRs.
4. ABO antigens are carbohydrate structures on glycoproteins and glycosphingolipids.
5. Anti-RhD is the most common cause of HDFN.
6. Red cell surface proteins serve a variety of functions, though many of their functions are still not known.
1. Red Cell Immunogenetics and Blood Group Terminology Working Party of the International Society of Blood Transfusion. http://www.isbtweb.org/working-parties/red-cell-immunogenetics-and-blood-group-terminology.
2. Storry JR, Castilho L, Daniels G, et al. International Society of Blood Transfusion Working Party on red cell immunogenetics and blood group terminology: Berlin Report. Vox Sanguinis 2011; 101: 77–82.
3. Yamamoto F, Clausen H, White T, Marken J & Hakomori S. Molecular genetic basis of the histo-blood group ABO system. Nature 1990; 345: 229–233.
4. Chester AM & Olsson ML. The ABO blood group gene: a locus of considerable genetic diversity. Transfus Med Rev 2001; 15: 177–200.
5. Yu L-C, Twu Y-C, Chou M-L, et al. The molecular genetics of the human I locus and molecular background explaining the partial association of the adult I phenotype with congenital cataracts. Blood 2003; 101: 2081–2088.
6. Wagner FF & Flegel WA. RHD gene deletion occurred in the Rhesus box. Blood 2000; 95: 3662–3668.
7. Rhesus base website. http://www.uni-ulm.de/∼fwagner/RH/RB.
8. Daniels G, Finning K, Martin P & Massey E. Non-invasive prenatal diagnosis of fetal blood group phenotypes: current practice and future prospects. Prenat Diagn 2009; 29: 101–107.
9. Mouro I, Colin Y, Chérif-Zahar B, Cartron J-P & Le Van Kim C. Molecular genetic basis of the human Rhesus blood group system. Nature Genet 1993; 5: 62–65.
10. Poole J & Daniels G. Blood group antibodies and their significance in transfusion medicine. Transfus Med Rev 2007; 21: 58–71.
11. Tournamille C, Colin Y, Cartron JP & Le Van Kim C. Disruption of a GATA motif in the Duffy gene promoter abolishes erythroid gene expression in Duffy-negative individuals. Nature Genet 1995; 10: 224–228.
12. Daniels G. Functions of red cell surface proteins. Vox Sanguinis 2007; 93: 331–340.
13. Anstee DJ. The functional importance of blood group-active molecules in human red blood cells. Vox Sanguinis 2011; 100: 140–149.
14. Mohandas N & Gallagher PG. Red cell membrane: past, present, and future. Blood 2008; 112: 3939–3948.
Anstee DJ. Red cell genotyping and the future of pretransfusion testing. Blood 2009; 114: 248–256.
Anstee DJ. The relationship between blood groups and disease. Blood 2010; 115: 4635–4643.
Avent ND & Reid ME. The Rh blood group system: a review. Blood 2000; 95: 375–387.
Daniels G. Human Blood Groups, 3rd edn. Oxford: Wiley-Blackwell 2013.
Daniels G. The molecular genetics of blood group polymorphism. Hum Genet 2009; 126: 729–742.
Daniels G & Bromilow I. Essential Guide to Blood Groups, 2nd edn. Oxford: Blackwell Publishing; 2010.
Daniels G & Reid ME. Blood groups: the past 50 years. Transfusion 2010; 50: 3–11.
Reid ME & Lomas-Francis C. The Blood Group Antigen Facts Book, 2nd edn. New York: Academic Press; 2004.
Roback JD, Combs MR, Grossman BJ & Hillyer CD. AABB Technical Manual, 17th edn. Bethesda, MD: AABB; 2011.
Watkins WM. (ed.) Commemoration of the centenary of the discovery of the ABO blood group system. Transfus Med 2001; 11: 239–351.