Treatment of HDFN

If alloimmunization is detected in time, the disease can be prevented in most cases. Treatment of HDFN depends on the severity of fetal anemia. In the past, bilirubin levels in amniotic fluid were used to assess the anemia. Today, advances in Doppler ultrasonography have made noninvasive detection of fetal anemia possible. Mari et al. showed that ultrasonic Doppler determination of the middle cerebral artery peak systolic velocity (MCA-PSV) can be used as a surrogate measurement of fetal anemia . In an international multicenter study it was demonstrated that MCA-PSV is superior to amniocentesis, with a sensitivity of 88% and a specificity of 82% . In case of severe fetal anemia, the anemia can be corrected by intrauterine blood transfusions (IUT) and/or preterm labor might be induced followed by neonatal treatment. The first IUT is on average given around 26 weeks, range 16–35 weeks, the earlier transfusions are especially given in pregnancies complicated with anti-K antibodies. The perinatal survival rate of cases treated with IUT is around 95% with a normal neurodevelopmental outcome in > 95% . The development of kernicterus after birth can be prevented by exchange transfusion after birth. In mild cases of HDFN phototherapy is sufficient. For timely treatment, all developed countries have screening programs in place to identify alloimmunized pregnancies at risk. Without a screening program, HDFN during pregnancy can only be suspected by decreased fetal movements or sudden fetal death, or by (early) neonatal jaundice after birth.

Anti-D Immunoprophylaxis to Prevent Immunization

Before the introduction of postnatal anti-D immunoprophylaxis HDFN was one of the major causes of perinatal death. In the Netherlands postnatal immunoprophylaxis was introduced in 1969 and resulted in a drastic decrease of the prevalence of anti-D immunization from 3.5% to 0.5% in the 1990s . Postnatal prophylaxis has to be given within 72 h after delivery and traditionally on the basis of postnatal RhD typing of cord blood from the newborn . Already in 1978, Bowman and colleagues demonstrated that the use of antenatal anti-D prophylaxis can further reduce immunization incidents . The effects of combined antenatal and postnatal prophylaxis have been substantiated by more recent meta-analyses . Combined prophylaxis reduces the immunization risk by half as compared to postnatal prophylaxis only, with a parallel reduction by half in severe HDFN cases . Thus in several countries, antenatal prophylaxis is currently combined with postnatal prophylaxis to further minimize immunization risk. Antenatal prophylaxis is routinely given either as a single dose of 250–300 μg of immunoglobulin at gestational weeks 28–31, or as two 150-μg doses, respectively, at gestational weeks 29 and 34 . In addition, it is given after invasive procedures during pregnancy, abdominal trauma, late miscarriage, or termination of pregnancy .

We recently obtained strong evidence that anti-D immunoprophylaxis is also preventing immunization toward other RBC antigens. Only 1 out of 99 pregnant women who became alloimmunized against non-Rh antigens during their first pregnancy had received anti-D immunoprophylaxis in this first pregnancy, which was significantly less often than the 10 women expected based on calculations for the general population . Also ABO incompatibility can protect against immunization against other blood group antigens . Theoretically, also immunoprophylaxis against Rhc or K, or by giving antibodies against a universal fetal antigen could have the same “wider” preventive effect, but so far, only anti-D immunoprophylaxis is available.

HDFN Mediated by RhD and Non-RhD Alloantibodies

Despite adequate antenatal and postnatal anti-D immunoprophylaxis still 1–3 in 1000 D-negative women develop anti-RhD antibodies, and in about 30% of these cases severe HDFN develops . The immunization against RBC antigens other than RhD became relatively more important when the incidence of anti-D immunization decreased. The prevalence of non-D immunization as determined in large-scale (> 70,000 women) studies of unselected pregnant women is calculated to be about 3 in 1000 women ( Table 1 ). The specificities of non-D alloantibodies detected in a predominantly European population of pregnant women are, in order of frequency, anti-E, anti-K, anti-c, anti-C ^w , anti-Fy ^a , anti-S, anti-Jk ^a , and anti-e . Occasionally antibodies against Kp ^a , anti-f, anti-Jk ^b , and anti-s are detected. Antibodies against other blood group antigens are extremely rare. Anti-M antibodies are frequently found, but almost always of the IgM subclass and therefore harmless.

Any antibody that can pass the placenta and react with an antigen expressed by fetal RBCs can theoretically result in HDFN. However, only in a minority of immunized pregnancies, the presence of the antibody results in clinical disease. In 2008 we performed a Dutch nationwide prospective index-cohort study on 305,000 consecutive pregnancies screened in the 12th week of pregnancy, and determined the risk that the presence of non-RhD antibodies will lead to clinical signs of HDFN . In Table 2 the results of this study are summarized. In total non-RhD antibodies were detected in 1279 pregnancies and in 403 cases the fetus was positive for the antigen. Severe HDFN, defined as need for intrauterine treatment and/or blood transfusion after birth, was almost exclusively seen in pregnancies with anti-c and anti-K antibodies, and rarely in pregnancies with other (non-D/c) Rh-antibodies. The risk of severe HDFN in case of anti-K antibodies has probably been underestimated in Table 2 , because at present we observe that 50% of cases with anti-K antibodies lead to severe HDFN. Others have reported similar data . All other antibodies (observed in this single center study) did not cause severe HDFN. In a cohort of 426 IUT-treated fetuses from the reference center for fetal treatment in the Netherlands after the introduction of antenatal prophylaxis, the vast majority of IUTs were still given because of alloimmunization against D (78%), compared to K in 15%, c in 3.5% and occasionally, antibodies against other antigens. Similar observations were done in Denmark and Germany.

In literature case reports of severe HDFN caused by a wide variety of other specificities have been described, but in population studies these specificities are seldom found as cause of severe HDFN. Anti-Duffy, anti-Kidd, and anti-MNS antibodies are relatively frequently found, but rarely result in severe HDFN , although they may lead to slightly decreased hemoglobin levels at birth and need for phototherapy. For antibodies against I, Le, P1, Lu, and Yt there is no risk for developing disease, because of very low expression of these antigens by fetal cells. Albeit the high rate of ABO incompatibility between the mother and her fetus, clinically significant hemolysis in neonates due to anti-A or anti-B is relatively low. And, if it occurs, it is usually mild . For these reasons in Western countries pregnant women are not screened for the presence of anti-A or anti-B of IgG class. Because placental tissue expresses A and B antigens the maternal alloantibodies may be absorbed to some extent by the placenta. Furthermore, fetal RBCs show weak expression of A and B blood group antigens. There is a striking, and unexplained, difference in the incidence of ABO-mediated HDFN between populations. The incidence is around 0.3%–0.8% in the Caucasian population, vs 3%–5% in Black or Asian populations, also with a more severe clinical course . Other alloantibodies detected in East Asian populations from China and Taiwan are antibodies specific for hybrid glycophorins of the MNS system, especially antibodies reactive with RBC expressing GP.Mur. This phenotype is virtually absent in European populations, whereas it occurs in 6%–8% of Chinese and Thai individuals, and 0.13% of pregnant Chinese women have anti-Mur antibodies . The majority of anti-MUR antibodies are IgM, but IgG anti-MUR antibodies can result in very severe HDFN and might cause, in the absence of antibody screening, fatal hydrops fetalis .

In conclusion, in the Western world especially pregnancies complicated with anti-D, anti-c, anti-K, and possibly also anti-E should be carefully monitored. If fetal blood group typing demonstrates that the fetus is antigen positive, the titer of the antibody has to be investigated at regular intervals to check whether a critical titer is exceeded . In that case the woman can be referred to a special care center for further clinical examinations like Doppler ultrasonography to determine the severity of anemia, and to start IUT in time or to induce labor, if needed.

Rh System

The Rh blood group system consists of two genes RHD and RHCE on chromosome 1 (1p36.11), positioned in opposite directions and separated by 31.8 kb, in which the TMEM50A gene (previously SMP1) is located . The RHD gene arose as a duplication of the RHCE gene in the common ancestors of humans, chimpanzees, and gorillas . Both genes have 10 exons and share an overall 93.8% gene sequence identity and 96.4% exon sequence identity . The RHD gene is flanked by two 9 kb regions of 98.6% homology, the so-called Rh boxes . The function of the RhD and RhCE polypeptides is unknown. They belong to the Rh family, but in contrast to three other members of this family (RhAG, RhBG, and RhCG) they most likely do not transport ammonia . It has been suggested that they might play a role in CO ₂ transport in RBC but are functionally redundant. The RHD gene ( NG_007494 ) encodes the RhD protein (CD240D), carrying the D antigen (RH1). The RHCE gene ( NG_009208 ) encodes the RhCE protein (CD240CE), carrying the C (RH2) or c (RH4) and E (RH3) or e (RH5) antigens. Both proteins are red cell specific, consist of 417 amino acids, are not glycosylated, and have 12 membrane spanning domains. As shown in Fig. 1 A , 37 nucleotides are specific for the coding sequence of RHD and not present in any of the RHCE alleles. Five nucleotides (in exon 2) are specific for RHc , of which the 307C encoding 103Pro is best correlated with c expression . Exon 2 of RHC is identical to exon 2 of RHD , therefore only 48C (in exon 1) (16Cys) is specific for RHC . However, 16Cys is not correlated to C-expression and 74% of African blacks have 48C in a ce-allele with normal expression of Rhc . Therefore RHC genotyping assays have to be based on a 109-bp insert in intron 2, that is, only present in RHC . The E/e polymorphism is caused by a single nucleotide variation 676C > G (Pro226Ala) in exon 5 of RHCE . The 676G is also present in the RHD allele, but is in RHD surrounded by seven D specific nucleotides . The frequency of the different alleles in the different populations is indicated in Fig. 1 A.

Schematic representation of the RH genes and the most prevalent variant RHD alleles and their frequency in the different populations. The exons (1–10) are represented as yellow boxes, on scale; the introns are black for RHD and red for RHCE and are on scale with each other. The RHD⁎01 gene, encoding the RhD polypeptide, is given as the reference sequence. The positions of the nucleotides differing between RHD and RHCE are listed above the top row, with the RHD nucleotide at that position, within red boxes the c and E/e determining nucleotides. Nucleotide substitutions are shown as red (T), black (G), green (A), or blue (C) lines, and if not given in the top row also the position is indicated. The ISBT allele name and the encoded phenotype are given at the left, and the frequencies in different populations at the right. (A) Five major alleles encoding the D and CcEe polypeptides. (B) Three most frequently occurring D-negative RHD alleles. (C) Most frequently occurring variant D-positive RHD alleles. In the bottom row three different weak D alleles, each characterized by a single nucleotide substitution. The blue diamond ( ) represents an intronic insert and the blue triangle ( ) represents a 37-bp duplication.

In the RBC membrane the Rh proteins are complexed with RhAG, another member of the Rh family. The most likely in vivo subunit configuration of the Rh core complex is RhD-(RhAG) ₂ and RhCcEe-(RhAG) ₂ . RhAG is essential for the assembly of the Rh complex in the erythrocyte membrane . Mutations in the RHAG gene can result in the complete absence (regulator Rh _null ) or severe reduction (Rh _mod ) of both RhD and RhCE antigens . The Rh complex is directly anchored to the cytoskeleton via Ankyrin R, and also mutations in ANK1 can result in weakened expression of RhD and RhCE antigens . The RH locus is highly polymorphic, and many hybrid RHD-RHCE gene variants have been described. The Rh antigens and RH alleles are numbered by the Working Party on Red Cell Immunogenetics and Blood Group Terminology of the International Society of Blood Transfusion and published on their website ( http://www.isbtweb.org/working-parties/red-cell-immunogenetics-and-blood-group-terminology/ ). A complete database for RHD variants has been established by Wagner and Flegel and is continuously updated ( http://www.rhesusbase.info /) . At present (December 2017), 54 antigens, 378 RHD alleles, and 116 RHCE alleles have been recognized in the Rh blood group system. The main mechanism responsible for the generation of hybrid RH genes is thought to be gene conversion, explained by the opposite orientation of the highly homologous RHD and RHCE genes. Multiple exons can be converted, but often microconversion events lead to single amino acid changes. In addition, gene conversion events can be associated with untemplated mutations, in which the mutated nucleotides are not derived from either gene .

RhD Negativity

The frequency of the D negative phenotype is 15%–17% in populations of European descent, 5% in Africa, and very rare in Asian populations. Because D negativity will reduce reproductive fitness due to HDFN, it is surprising that D negativity has become so frequent in Europe, suggesting a positive selection for an as-of-yet unknown fitness benefit of the RHD deletion. Perry et al. found, however, no evidence that positive natural selection affected the frequency of the RHD deletion . Thus the emergence of the RHD deletion in European populations may simply be explained by genetic drift/founder effect. Mourant indeed proposed already in 1954 a mixing of two populations, one essentially D − (Paleolithic peoples from the Basque region, which survived the last ice age) and the other D + (Neolithic migrants) as cause for the high frequency of D negativity in Europeans . Genetic analysis using mtDNA and Y chromosome markers has now provided strong evidence for this hypothesis . In European and Asian people the D negative phenotype is usually caused by deletion of RHD , which has occurred between the Rh boxes, flanking the RHD gene . Although many different nonsense mutations or splice site mutations have been described, each of them is rare. In contrast, in the African population deletion of RHD gene is not the only mechanism responsible for D negativity. Approximately 66% carry the RHDΨ gene and 15% carry a RHD-CE-Ds (r’s) hybrid allele ( Fig. 1 B) . The RHDΨ gene (RHD⁎08N.01) contains a 37 base pair duplication causing a frameshift, consisting of the last nucleotides of intron 3 and the first nucleotides of exon 4, and a nonsense mutation in exon 6 (807T > G, Tyr269stop) . The African RHD-CE-Ds allele consists of exon 1, 2, and 3 of RHD⁎DIIIa ; exons 4–7 of RHCE ; and exons 8–10 of RHD (type 1 = RHD⁎03N.01) , but occasionally it is a hybrid RHD (exon1–2)- RHCE (exon3–7)- RHD (exon8–10) allele (type 2 = RHD⁎01N.06) . These two genes produce no D, but do produce an abnormal C antigen. Because the RHD-CE-Ds alleles miss intron 2 on which RHC genotyping assays are based, reliable RhC prediction is also complicated in Africans. The observation that in Africa different mechanisms resulting in D negativity have emerged might indicate that there has been an ancient (unknown) selective pressure in Africa .

In Asian populations (China, Korea, Japan), 15%–30% of serologically typed D negative individuals carry the RHD⁎01EL.01 ( RHD⁎1227A ) gene, harboring a single mutation (rs549616139) causing a splice site defect resulting in the DEL phenotype . In DEL individuals D expression on RBCs is extremely weak and can only be recognized by sensitive serological techniques such as the adsorption-elution test. These individuals are not at risk for RhD immunization . Between 3% and 8% of truly D negative Asian individuals carry the D-negative hybrid RHD⁎D-CE(2–9)-D allele (RHD⁎01N.03) ( Fig. 1 B). In addition, the silent RHD⁎711delC (RHD⁎01N.16) allele is relatively frequent (> 1%) in Chinese D negative individuals .

The presence of silent RHD genes in D negative individuals, shown in Fig. 1 B, might hamper noninvasive fetal RHD genotyping assays. If the mother carries a silent RHD allele that is detected in the fetal genotyping assay, the genotype of the fetus cannot be reliably predicted. If the fetus inherited a silent RHD gene from the father, this will result in false-positive results. For that reason, it is important to take these variant RHD genes into account when designing fetal RHD genotyping assays, or at least in the interpretation of the assay results. This might also be true for the Asian silent alleles. In Asia, routine fetal RHD genotyping for D negative pregnant women seems to be not that relevant, because < 5% of fathers are hemizygous for RHD . However, in a multiethnic society fetal genotyping assays should also be reliable for Asian D negative pregnant women, thus assays based on RHD exon 10 should be preferably avoided.

RHD Variant Genes With D Expression

As described previously the RH locus is highly polymorphic and numerous RHD variants have been described (see http://www.isbtweb.org/working-parties/red-cell-immunogenetics-and-blood-group-terminology/ ). RHD variants were originally subdivided into so-called DEL, weak D, and partial D. The most frequently occurring variant alleles are shown in Fig. 1 C.

The Del phenotype is characterized by a very weak expression of the complete RhD protein, which expression is only detectable with the very sensitive absorption-elution technique . The Del phenotype is most often caused by mutations that disturb splice sites or mutations in the C-terminal region of the RhD protein. DEL types are common in Asia (see earlier) and less common in Europe. The DEL frequency in the European population is 1:350 to 1:2000 .

Individuals with weak D expression express the complete RhD protein, however, in low quantities . 0.2%–1% of populations of European descent express weak D antigens. Weak D expression is caused mostly by single mutations in RHD that cause amino acid change in the transmembrane or intracellular parts of the protein . At present (December 2017), 139 different weak D antigens have been described. The most frequent weak D types are weak D type 1 (rs121912763; allele frequency 0.3% in ExaC database), weak D type 2 (rs71652374; allele frequency 0.1%), and weak D type 3 (rs144969459; allele frequency 0.3%).

RBCs of individuals carrying a partial D variant express an RhD protein that lacks 1 or several of the 30 D-epitopes and/or express new RhD epitopes, due to amino acid changes in (an) extracellular loop(s) of the RhD protein. Most partial variants arise from hybrid alleles in which parts of RHD are exchanged with the very homologous RHCE gene. In many of the partial D variants the expression is also weakened compared to normal RhD expression .

Fetal RBC expressing any of the gene products of these RHD variant genes can theoretically induce alloimmunization in the mother. DEL individuals cannot make anti-D, since they express the complete protein, so pregnant women do not need immunoprophylaxis. Furthermore, it was supposed that also weak D individuals cannot make anti-D, in contrast to individuals carrying partial D variants that miss some D epitopes. However, since the serological recognition of partial D variants can be difficult and it became clear that some weak D individuals, for example, with weak D type 4.2 (DAR), type 11, 15, 21, or 57 (reviewed in ) and also DEL individuals carrying the RHD⁎DEL8 allele or the RHD⁎DEL5 can become immunized against D, the DEL/weak D/partial D nomenclature is misleading and nowadays the term “D variant” is applied for all RHD alleles that result in qualitatively and/or quantitatively aberrant expression . The general consensus is that pregnant women or transfusion recipients with weak D type 1, 2, and 3 and Asian type DEL should be considered as RhD positive and not at risk for alloimmunization . Pregnant women carrying any of the other RHD variants should be categorized as D negative and included in prophylaxis programs . As outlined previously, the presence of these variant RHD genes will mask the fetal RHD , thereby hindering a straightforward prediction of the fetal D type. Furthermore, fetuses carrying a D-expressing- RHD variant might be missed by fetal genotyping and result in false-negative results.

Kell Blood Group System

The Kell antigens are located on a single-pass type II transmembrane glycoprotein (CD238) of 93 kD (732 amino acids), which functions as a metalloendopeptidase that processes endothelin-3 . On RBCs the Kell glycoprotein is covalently linked to the XK protein. The absence of XK protein leads to the McLeod syndrome that is characterized by mild hemolysis, late onset forms of muscular and neurological defects, red cell acanthocytosis, and a greatly reduced amount of the Kell protein . The KEL gene, consisting of 19 exons distributed over 21.5 kb, is located at chromosome 7q33. Kell appears early in erythropoiesis but may also be expressed on myeloid progenitors and weak expression is found in other tissues including brain and muscles. Kell antigens are already expressed in the fetus from 6 to 7 weeks of gestation onward .

The Kell system consists of 34 antigens. The clinically most relevant antigen is K (KEL1). As described previously anti-K antibodies are second to anti-D antibodies the most frequently encountered antibodies in severe HDFN. A single nucleotide variation (rs8176058) 578T > C in exon 6 resulting in Met193Thr is responsible for the Kk (KEL1, KEL2) polymorphism . Ninety-one percent of Europeans, 98% of Africans and South Asians, and almost all East Asians are K negative (kk or homozygous KEL2), and thus at risk for alloimmunization if carrying a K + fetus. But also antibodies against other antigens of the KEL blood group system (anti-k (KEL2), anti-Kpa (KEL3, rs8176059), and anti-Jsb (KEL7, rs8176038)) may result in severe HDFN . Recently, a single fatal case has been described against a novel, low frequency Kel alloantigen (KEAL, rs557358978) . Js ^a is a low frequency antigen in Caucasians (< 0.01%), whereas 20% of Africans are positive. Consequently, anti-Js ^b is more frequently found in Africans .

In the KEL system 29 silent alleles, so-called K _o alleles, have been identified and numbered by the ISBT ( http://www.isbtweb.org/working-parties/red-cell-immunogenetics-and-blood-group-terminology/ ). These alleles are caused by many different nonsense mutations, single nucleotide insertions or deletions, splice site mutations, and missense mutations. In addition, 12 so-called K _mod alleles resulting in weak Kell expression have been annotated. The majority of the presently known mutations (29 K _o and 11 K _mod ) occur in a k allele. Although these mutations are rare, they are relatively frequent in individuals with RBCs that are reactive only with anti-K and not with anti-k. In Europeans only 2 out of 1000 individuals have this phenotype, but 3.5%–7% of them were shown to be genotypically KEL⁎1/KEL⁎2, in which the KEL⁎2 allele was a K _o or K _mod allele . As a fetus carrying K _o alleles and K _mod alleles does not express K and hence is not at risk for HDFN, fetal K genotyping is relevant in all anti-K alloimmunized kk women, even if the assured father is phenotypically homozygous K. Their fetus will not be at risk if it inherited a paternal K _o or K _mod allele, and this aberrant k allele will not be detected in the K PCR.

Noninvasive Fetal RHD Typing to Guide Prophylaxis

In countries where antenatal prophylaxis is part of routine care, it is common practice that all D negative pregnant women are offered antenatal prophylaxis because the fetal RhD type remains unknown until birth. In individuals of European descent, approximately 40% of D negative pregnant women carry a D negative fetus. These women are not at risk for D immunization. With the advent of noninvasive analysis of cell-free DNA in maternal plasma (cfDNA), it has become possible to target antenatal RhD prophylaxis to those nonimmunized D negative women carrying an RHD positive fetus . Such targeting prevents unnecessary exposure of RhD negative pregnant women to anti-D immunoglobulin, in Western countries this is > 6% of all pregnant women. The prophylaxis product, that is, anti-D immunoglobulin, is a human-derived blood product that carries a theoretical risk of contaminants or unknown infective agents, such as prions, that may be detrimental to the woman or the fetus. Furthermore, a certain risk is also posed to the volunteer donors who allow to be hyperimmunized with RBCs to produce anti-D. Thus it has become an ethical necessity to offer pregnant women fetal RHD genotyping to avoid unnecessary immunoprophylaxis and the waste of a gift of donors . In addition, access to the prophylaxis product is limited in some countries, and these countries would thus benefit from restricting the product only to pregnant women with a demonstrated indication for its use. Nevertheless, the argumentation is quite opposite from a US perspective. The consensus in the United States is that the current regimen with untargeted and thus unnecessary use of antenatal prophylaxis should be left unchanged. The arguments are that the supply of anti-D immunoglobulin is not limited in the United States, the antenatal prophylaxis is perfectly safe, and that introducing targeted antenatal prophylaxis will cause more immunizations unless the sensitivity of the cfDNA testing is 100% .

In Europe, the implementation of noninvasive fetal RHD typing to guide targeted antenatal and postnatal prophylaxis is in progress. In Table 3 an overview of large-scale trials is given, the studies demonstrate high sensitivities, above 99% as early as GA 10–11 weeks and 99.9% at a GA of 25–28 weeks. At present, all the studies are based on multiplex real-time PCR assays and do not include positive controls for the presence of fetal DNA. It is important to note that similar high sensitivities have been attained in the three countries (Denmark, the Netherlands, and Finland) where fetal RHD typing and targeted prophylaxis were implemented in a nationwide program . The high sensitivity justified the abolition of cord blood serology for postnatal prophylaxis. In the studies evaluating the performance of implemented screening programs, only 21 false-negative cases in almost 50,000 (0.04%) screening tests were observed. The false-negative (FN) results are typically caused by low levels of cfDNA, failed DNA extraction, or human error . None of the FN results were caused by RHD variants in the fetus. It seems feasible to perform antenatal RHD screening early in pregnancy , although the sensitivity decreases in very early pregnancy. Implementation of early testing enables the targeted use of prophylaxis for potential sensitizing events, such as amniocentesis and chorionic villus sampling.

In the Dutch study, nine screening results were false negative and in 225 cases fetal RHD results were positive and cord blood serology was negative. Because in a screening program especially FN have to be prevented, the scoring algorithm is less stringent toward false-positive (FP) results. Thus, firstly, low levels of nonspecific amplification, or from DNA derived from a vanishing twin, will result in an RHD positive typing. Therefore in all studies listed in Table 3 the specificity is lower than the sensitivity. A second cause of FP or inconclusive results are variant genes in the mother or the child. This was the case in 100 out of 225 cases in the Dutch nationwide screening program evaluation. And thirdly, discrepancies between fetal RHD typing and cord blood serology can be caused by FN cord blood serology results . In 10 of the 225 presumed FP cases analysis of backup plasma showed that the “cord blood serology” (generating RhD negative results) had mistakenly been performed on maternal blood samples or cord blood mixed with maternal blood. In an additional 22 FP cases with negative cord blood serology, this serology result appeared to be false as the newborn carried an RHD variant gene with weak D expression . Thus the net effect of implementation of fetal RHD genotyping might even be fewer immunizations.

Noninvasive Fetal Blood Group Typing in Alloimmunized Women

As described previously in most countries pregnant women are serologically screened for the detection of alloantibodies against RBC antigens. If an anti-RBC antibody is detected, the specificity of the alloantibody(ies) is determined to assess the risk of HDFN. As listed in Table 2 especially anti-Rh antibodies and anti-K are frequently found and clinically relevant. Before fetal blood group genotyping was possible, the supposed father was serologically typed for the presence of the involved antigen and the antithetical allele. If the father was either homozygous positive or negative for the involved alloantigen, the fetal phenotype could be predicted. In particular in case of anti-K, the father is in the far majority of cases K-negative, and the mother has been immunized by a previous incompatible blood transfusion . In the past, homozygosity or hemizygosity for RHD was predicted based on CcEe phenotype, because of the linkage disequilibrium between the different RHD and RHCE alleles. However, these predictions can be false and nowadays genotyping assays are used to determine RHD homo- or hemizygosity in the father. These assays can be based on the presence of a hybrid Rh-box, which is only present in an Rh locus from which the RHD is deleted . Owing to the high frequency of mutated Rhesus boxes in African individuals this assay is not accurate in Africans , whereas it is in China . The RHD gene copy number can be determined by quantitative PCR , double Amplification Refractory Mutation System , ddPCR, MLPA, or Malditof assays .

However, direct fetal blood group typing is to be preferred. Before the discovery of cell-free DNA, fetal blood group genotyping was performed with fetal DNA from amniotic fluid . Since the introduction of cell-free DNA-based fetal blood group typing, invasive procedures for blood group typing have become obsolete for this indication. They have a small risk of fetal loss and a risk of FMH which might boost the maternal immune response. Noninvasive fetal blood group typing is offered by specialized diagnostic laboratories in many countries. For noninvasive fetal RHD genotyping in D immunized women, the accuracy was reported to be 98.5% in a meta-analysis of 41 publications. However, this meta-analysis included also publications on screening as well as studies from the early development of the method. Thirty of the 41 publications reported 100% accuracy .

Several of the laboratories have reported on the accuracy of their diagnostic tests for the other clinically relevant antibodies , and as shown in Table 4 in most studies the accuracy is 100% for RHC, RHc, and RHE, only for KEL1 detection some false results are obtained. Only in the Polish and Dutch studies the presence of fetal DNA was controlled , see for a discussion on the necessity of adding a positive control for blood group genotyping the review of Scheffer et al. .

Fetal RHD Genotyping

So far, all laboratories involved in fetal RHD screening apply real-time PCR with Taqman probes, mostly a duplex PCR. RHD exons 4, 5, 7, and 10 are used as targets ( Table 3 ). Detecting at least two RHD exons is highly recommended to increase the chance of a positive detection among most variants. The Danish developed an elegant method to further increase the sensitivity of the assay by combining the two signals for exon 7 and exon 10 into one signal by applying the same reporting dye for each assay .

The most widely applied assay is the duplex exon 5/exon 7 PCR that has been validated in many European laboratories in the EU granted SAFE project . This assay examines the RHD sequence that encodes the portion that harbors most exofacial D epitopes. Nevertheless, also this duplex assay will give false-negative results with some rare hybrid RHD variants: DBT-1 (RHD-Ce(5–7)-D), DBT-2(RHD-Ce(5–9)-D) , and DBU (RHD-cE(5–7;226P)-D) . The relatively frequent African alleles DAR, DAR-E, and Weak D type 29 have single mutations both in exon 5 (667T > G) and exon 7(957G > A), and this might theoretically hamper their detection. In the SAFE exon 5/exon 7 PCR these variants appeared to be amplified however .

Exon 7 and the untranslated 3′-end of exon 10 harbor multiple D specific nucleotides, which makes it relatively easy to develop specific assays. The theoretical disadvantage of exon 10 as compared to exon 7 is that it does not encode for D epitopes, which is exemplified by its presence in D negative RHD variants with gross deletions such as the African (C)ce ^s and the Asian RHD⁎01N.03 genes ( Fig. 1 B).

PCR assays in exon 4 or exon 5 can be designed in such a way that they will not amplify the most common RHD variant genes in Africans and Europeans, RHD⁎Ψ and RHD⁎DVI , respectively ( Fig. 1 B and C). However, in duplex assays with exon 7 or exon 10, the recognition of an RhD positive fetus in women carrying the RHD⁎Ψ and RHD⁎DVI will depend on the amplification of only a single exon 4 or 5. This makes the assay less sensitive and all fetal RHD variants with mutations in, respectively, exon 4 or 5 might be missed. Furthermore, the concomitant amplification of another exon derived from the maternal variant gene might interfere with the amplification of the fetal exon 4 or 5. At present, the reliability of fetal RHD genotyping for screening in mixed ethnic populations has not been established yet.

Fetal RHc Genotyping

RHc genotyping is rather straightforward and does not suffer from nonspecific amplification as is the case for Kell genotyping (see later) because of the five RHc specific nucleotides in exon 2 ( Fig. 1 A). The 307A nucleotide encoding 203Pro is strictly correlated with Rhc expression . Although RHC and RHc alleles also differ in exon 1 (48C in RHC and 48G in RHc , Fig. 1 A), this nucleotide should not be used, because in most of the African population the RHCE⁎ce allele has a 48C at this position but Rhc and not RHC expression . All laboratories reporting results of diagnostic assays for fetal RHc genotyping in alloimmunized pregnant women listed in Table 4 applied real-time PCR. Three obtained results with a similar assay, originally developed in Bristol, UK . In this assay both the forward and reverse primers are specific for the RHc allele at their 3′-end (201A, 203A, and 307C, respectively). The German group applied an approach in which only the reverse primer is specific (307C) and in this case the probe harbor the c-specific nucleotides 201A and 201A . Theoretically the assay developed in the UK might be superior, also because of the lower amplicon size (106 vs 175 bp).

Fetal K Genotyping

Expression of the K antigen is determined by only a single nucleotide substitution (c.578C > T, rs8176058), and the development of a specific real-time PCR assay is, among others for that reason, more challenging. The amplification of the maternal k-allele by mispriming of the K-specific primer has been tackled in several ways. Finning et al. designed a K-specific primer with an extra mismatch and two locked nuclei (LNA) acid bases, to increase specificity, however, at the expense of lower sensitivity . Scheffer et al. included a k allele-specific peptide nucleic acid (PNA) probe in the PCR reaction. Clamping of this PNA probe to the maternal k-allele diminishes the nonspecific amplification of the maternal allele. In this study 24 Kell positive and 32 Kell negative newborns were identified by antenatal fetal K genotyping. In 4 out of 60 cases results were inconclusive because of weak amplification signals or because presence of fetal DNA could not be confirmed. In one case with low amplification signals, an intrauterine demise was diagnosed (and potentially already present at the time of the first blood draw) before a second sample could be taken, and a Kell positive child was born . Several labs have developed alternative approaches. Rieneck et al. showed proof of principle using a targeted NGS approach. Li et al. applied Maldi-TOF for K-genotyping but missed 2 out of 13 K-positive fetuses . Bohmova successfully applied a minisequencing approach, but reported remarkably low relative levels of fetal DNA (< 1%) and only 4 out of 128 tested women carried a K positive fetus . In conclusion, from all noninvasive blood group genotyping PCRs the K-assay is the most cumbersome and awaits technical improvement, for example, by digital droplet PCR.

In addition to these challenges with the assay, it is worth mentioning that extremely rare variants can lead to false-negative results with deleterious clinical consequences. Poole et al. described a variant K gene in which another SNP at codon 193 (c.577A > T, rs61729031) results in K expression, and which will be missed by most K genotyping assays . However, the frequency of this SNP is below 0.01% (ExaC database).