Significance of ABO Blood Group Incompatibility between Mother and Child Regarding Incidence and Severity of Fetal and Neonatal Alloimmune Thrombocytopenia Inaugural Dissertation submitted to the Faculty of Medicine in partial fulfillment of the requirements for the PhD-Degree of the Faculties of Veterinary Medicine and Medicine of the Justus Liebig University Gießen by Miserre-Roth, Laila Gießen 2023 Faculty of Medicine of the Justus Liebig University Gießen, Center for Transfusion Medicine and Hemotherapy Supervisor: Prof. Dr. med. Gregor Bein Supervisor: Prof. Dr. med. Matthias Wolff Date of Doctoral Defense: October 9th, 2023 Table of Contents 1 Introduction ............................................................................................................ 1 1.1 FNAIT .............................................................................................................................................. 1 1.2 Pathogenesis ..................................................................................................................................... 2 1.3 Diagnosis .......................................................................................................................................... 4 1.4 Therapy ............................................................................................................................................. 5 1.5 Predictors of Incidence and Severity ................................................................................................ 7 1.6 General Screening in Pregnancy ..................................................................................................... 10 1.7 Prophylaxis of Maternal Immunization in HDFN and FNAIT ....................................................... 11 1.8 ABO Incompatibility ...................................................................................................................... 13 1.9 ABO Blood Group System ............................................................................................................. 13 1.10 ABO Antigens on Platelets ............................................................................................................. 16 1.11 Hypothesis and Aim of the Study ................................................................................................... 18 2 Material and Methods ......................................................................................... 20 2.1 Subjects ........................................................................................................................................... 20 2.2 Materials and Manufacturer ............................................................................................................ 22 2.2.1 Reagents and buffer .................................................................................................................. 22 2.2.2 Consumables ............................................................................................................................. 24 2.2.3 Hardware ................................................................................................................................... 25 2.2.4 Software .................................................................................................................................... 26 2.3 MAIPA ........................................................................................................................................... 27 2.4 DNA Extraction .............................................................................................................................. 29 2.5 Measurement of DNA Concentration ............................................................................................. 30 2.6 ABO Genotyping ............................................................................................................................ 30 2.7 Principle TaqMan® Real Time PCR .............................................................................................. 35 2.8 Statistical Analysis of Data ............................................................................................................. 35 3 Results ................................................................................................................... 38 3.1 Retrospective Case-Control Study .................................................................................................. 38 3.1.1 ABO phenotype frequencies do not differ between FNAIT cases and controls ........................ 38 3.1.2 Fetomaternal ABO incompatibility does not protect against immunization to fetal platelet antigens by pregnancy ............................................................................................................... 40 3.2 Retrospective Cohort Study ............................................................................................................ 41 3.2.1 Fetomaternal ABO incompatibility and FNAIT severity .......................................................... 41 3.2.1.1 Fetomaternal ABO incompatibility is not associated with fetal/neonatal ICH ........... 41 3.2.1.2 Fetomaternal ABO incompatibility is not associated with neonatal platelet count nadir ............................................................................................................................ 42 3.2.1.3 Fetomaternal ABO incompatibility is not associated with neonatal birth weight ...... 43 3.2.1.4 Fetomaternal incompatibility for blood group A1 is not associated with FNAIT severity ....................................................................................................................... 44 3.2.2 Maternal ABO phenotypes and FNAIT severity....................................................................... 45 3.2.2.1 Maternal ABO phenotypes are not associated with neonatal ICH .............................. 45 3.2.2.2 Maternal ABO phenotypes are not associated with neonatal platelet count nadir ...... 46 3.2.2.3 Maternal phenotype A is associated with neonatal birth weight ................................ 47 3.2.3 Maternal ABO gene dose and FNAIT severity ......................................................................... 48 3.2.3.1 Maternal ABO gene dose is not associated with neonatal ICH .................................. 49 3.2.3.2 Maternal ABO gene dose is not associated with neonatal platelet count nadir .......... 51 3.2.3.3 Maternal ABO gene dose is not associated with neonatal birth weight ...................... 53 3.2.4 Neonatal phenotype O is associated with ICH .......................................................................... 55 3.2.5 Neonatal ABO gene dose and FNAIT Severity ........................................................................ 56 3.2.5.1 Neonatal ABO gene dose is not associated with ICH ................................................. 56 3.2.5.2 Neonatal ABO gene dose is not associated with neonatal platelet count nadir .......... 57 3.2.5.3 Neonatal ABO gene dose is not associated with birth weight .................................... 58 4 Discussion ............................................................................................................. 59 4.1 ABO Phenotype Distribution .......................................................................................................... 59 4.2 Fetomaternal ABO Incompatibility and FNAIT ............................................................................. 59 4.3 Maternal ABO Phenotypes and FNAIT Severity ............................................................................ 62 4.4 Maternal ABO Gene Dose and FNAIT Severity ............................................................................ 64 4.5 Neonatal ABO Blood Groups and FNAIT Severity ....................................................................... 65 4.6 Strength and Weaknesses ................................................................................................................ 66 5 Synopsis (English and German) ......................................................................... 69 6 List of Abbreviations ........................................................................................... 71 7 List of Figures ....................................................................................................... 73 8 List of Tables ........................................................................................................ 74 9 References ............................................................................................................. 75 10 List of Publications, Congressional Contributions ........................................... 93 11 Declaration of Academic Honesty ...................................................................... 94 12 Acknowledgement ................................................................................................ 95 Introduction 1 1 Introduction 1.1 FNAIT Fetal and neonatal alloimmune thrombocytopenia (FNAIT) involves the mother generating alloantibodies against fetal platelet antigens inherited from the father (Pearson et al. 1964). After transplacental transfer of IgG antibodies, this results in the opsonization of fetal platelets and subsequent destruction of platelets by different mechanisms; thrombocytopenia and signs of bleeding are present in the newborn or remain unapparent. Most commonly, bleeding signs such as petechiae (90%), hematomas (66%) and melena (30%) occur. Other less frequent bleeding signs are hemoptysis (8%), retinal bleeding (7%) and hematuria (3%) (Mueller-Eckhardt et al. 1989). The incidence of FNAIT in Caucasians is 1 in 1000 pregnancies and the incidence of severe bleeding complications due to FNAIT is 1 in 10,000 pregnancies (Kamphuis et al. 2014). Approximately 30% of FNAIT cases result in severe disease progression (Kamphuis et al. 2010). Intracranial hemorrhage (ICH) occurs in 10% of severely thrombocytopenic cases (Kamphuis et al. 2010) and leads to perinatal mortality in almost 50% of these cases, with severe neurodevelopmental impairment occurring in 60% of surviving children (Winkelhorst et al. 2019). Alloimmunization occurs when the father is positive for a target antigen which the mother does not inherit. In FNAIT, target antigens are human platelet antigens (HPA), expressed as molecular variants of glycoproteins (GP), located primarily on the surface of thrombocytes and megakaryocytes. Several different human platelet antigen-systems are known so far, and each system comprises different alleles, of which the frequent allele is labelled with the letter a and the less common allele with the letter b (Sachs 2013). In Caucasians, the most frequent alloantibody responsible for approximately 75% of all cases of FNAIT is directed against HPA-1a (Ghevaert et al. 2007b), which is expressed on the integrin β3-chain (GP IIIa) and present in various complexes on different cells, for example on thrombocytes (see also chapter Pathogenesis). 75% of all Caucasians are homozygous HPA-1aa, 23% heterozygous HPA-1ab and 2% are HPA-1bb homozygous (Ahlen et al. 2012; Décary 1982). The risk of HPA-1a alloimmunization in HPA-1bb women after delivery of an HPA-1a- positive child is strongly associated with the HLA class II allele HLA-DRB3*01:01 Introduction 2 (Wienzek-Lischka et al. 2017). 12.7% of HPA-1bb mothers who are HLA-DRB3*01:01 positive undergo alloimmunization during pregnancy if they encounter the fetal HPA-1a peptide, as opposed to only 0.5% who lack this allele (Kjeldsen-Kragh and Olsen 2019). HPA-1a and HPA-1b differ in a single leucine/proline substitution at position 33 (leu33/pro33) of the integrin β3 chain. In HPA-1a, leu33 acts as anchor residue for the peptide binding groove of HLA-DRB3*01:01. Its hydrophobic side chain is predicted to fit to the hydrophobic P9 pocket of DRA/DRB3*01:01 (Wu et al. 1997) and is buried in the peptide binding groove, which enables the stable MHC binding necessary for T-cell activation. In contrast, the pro33 in HPA-1b associates poorly to the hydrophobic P9 pocket of DRA/DRB3*01:01; only non-physiologically high concentrations of autologous HPA-1b peptides enable HPA-1a-specific T-cell activation (Ahlen et al. 2016). Another frequently targeted antigen (approximately 15% of FNAIT cases) in Caucasians is HPA-5b on GPIIb, whereas in Asian regions the most frequent antibody is directed against HPA-4b (Mueller-Eckhardt et al. 1989; Shibata et al. 1986). 1.2 Pathogenesis Regarding hemolytic disease of the fetus and newborn (HDFN), the red blood cell (RBC) counterpart of FNAIT, the alloimmunization of pregnant women against RBC antigens primarily occurs during parturition, when fetomaternal hemorrhage is at its strongest (Woodrow and Finn 1966). Consequently, HDFN does not regularly occur in the first pregnancy of mothers not immunized before by pregnancy or blood transfusion. In contrast, immunization against HPA-1a does already occur early during the first pregnancy (Skogen et al. 2009). However, it is doubtful whether the amount of fetal platelets present in antenatal FMH could sufficiently cause immunization against HPA- 1a, taking into account that platelets seldomly stimulate anti-HPA-1a immunization after allogeneic transfusions (Kiefel et al. 2001). Therefore, Kumpel et al. propose that particles of the syncytiotrophoblast may trigger alloimmunization in early pregnancy. These particles expressing the β3 integrin (carrier of the HPA-1a/1b polymorphism) are shed into the maternal bloodstream (Kumpel et al. 2008) as early as 4-6 weeks p.c. and are phagocytized by maternal dendritic cells, whereupon HPA-peptides are presented on Introduction 3 MHC molecules and matching maternal T-cells are activated (Kumpel and Manoussaka 2012). However, this route of maternal immunization could account for alloimmunization against HPA located on the β3 integrin only (Curtis 2015). Once fetal antigens enter the maternal circulation, the alloantigens are processed and presented by professional antigen presenting cells, e.g., dendritic cells, to T-cells. Ahlen et al. were able to demonstrate the existence of HPA-1a– specific HLA-DRB3*01:01– restricted CD4+ T cells in alloimmunized women (Ahlen et al. 2009). Activated, antigen-specific CD4+ T helper cells in turn activate antigen-specific B cells in undergoing a class switch from IgM to IgG and secreting anti-HPA-1a antibodies of IgG class. The level of antibody synthesis depends on the specificity of cognate T-B cell interaction and the strength of the co-stimulatory signal (Kumpel and Manoussaka 2012). The maternal IgG-antibodies are transferred into the fetal circulation by fetal Fragment crystallizable Receptor neonatal (FcRn) receptors on the syncytiotrophoblast (Chen et al. 2010) and opsonize fetal platelets, which are afterwards degraded in the reticuloendothelial system of the fetus. Already in week 16 of gestational age, fetal platelets express HPAs (Gruel et al. 1986). Since low platelet counts were already detected in some cases prior to gestational week 20, the transplacental IgG transfer must be active before gestational week 20 (Bussel et al. 1997). The β3 integrin is expressed as a heterodimer with αIIb (GP IIbIIIa, fibrinogen receptor), mainly on platelets, or as a heterodimer with αv (αvβ3 integrin, vitronectin receptor), mainly on endothelial cells (Bennett 1996; Bennett et al. 1997; Bennett et al. 1999; Paul et al. 2003). The expression of αvβ3 integrin on endothelial cells and interaction with extracellular matrix plays a major role in the formation of new vessels, e.g., angiogenesis (Bennett et al. 1999; Brooks et al. 1994). Santoso et al. demonstrated that three different types of anti-HPA-1a antibodies exist: anti-αIIbβ3, anti-β3 and anti- αvβ3.Whereas anti-αIIbβ3 is only reactive with the fibrinogen receptor (GP IIbIIIa) on platelets, anti-β3 and anti-αvβ3 (anti-αvβ3 predominantly) react with integrin αvβ3 (receptor for vitronectin) on endothelial cells (as well as smooth muscle cells and platelets). Only anti-αvβ3 impairs angiogenesis and induces apoptosis by blocking the attachment to vitronectin and increasing the synthesis of reactive oxygen species (ROS) (Santoso et al. 2016). In support of these findings, van Gils et al. described the Introduction 4 interference of HPA-1a antibodies with the formation of a stable endothelial monolayer by reallocation of junctional proteins (van Gils et al. 2009). Regarding the pathogenesis of ICH, Yougbaré et al. showed that mouse fetuses without platelets survived in utero without developing ICH, while in a mouse FNAIT model of isoimmunization in β3 -/- mice, ICH only occurred in the presence of anti-β3 integrin- antibodies (Yougbaré et al. 2015). Santoso et al. detected anti-αvβ3 antibodies in sera from mothers with children suffering from ICH (Santoso et al. 2016). Therefore, it can be assumed that ICH is more likely caused by an impairment of angiogenesis by anti- αvβ3 antibodies rather than because of low platelet counts. Since the αvβ3-integrin is expressed on infiltrating trophoblasts, anti-β3 antibodies could possibly be responsible for causing intrauterine growth retardation (IUGR) and miscarriage in FNAIT cases (Eksteen et al. 2017; Yougbaré et al. 2015). Additionally, FNAIT is associated with chronic placental inflammation mostly affecting the fetomaternal interface, e.g., chronic chorioamnionitis and villitis (Althaus et al. 2005; Dubruc et al. 2016). Killer cells naturally occurring in the uterus contribute to this inflammatory environment and tend to impair trophoblast function in the presence of anti- β3 antibodies and therefore impair placental development (Yougbaré et al. 2017). 1.3 Diagnosis FNAIT is usually discovered when the first obviously affected child is born and other causes for thrombocytopenia (e.g., septicemia) have been ruled out. Early onset thrombocytopenia and very low platelet counts in otherwise healthy newborns are found in FNAIT cases; other maternal clinical features (e.g. immune thrombocytopenia, ITP) and thrombocytopenia among sick, preterm children usually do not cause a neonatal platelet count this low and this early (Burrows and Kelton 1993). Laboratory diagnosis of FNAIT is based on a positive crossmatch between maternal serum and paternal platelets, discrepancy between maternal and paternal or maternal and neonatal HPA- genotypes, and the detection of maternal platelet-specific alloantibodies. Commonly, the detection of anti-HPA is performed serologically via monoclonal antibody immobilization of platelet antigens (MAIPA). MAIPA was also used in this Introduction 5 study to detect maternal antibodies and is therefore described in detail in chapter 2, Material and Methods (Kiefel et al. 1987). Another possibility for detecting anti-HPA antibodies is the platelet immunofluorescence test (PIFT), where test platelets are incubated with maternal serum. If anti-HPA antibodies exist, they are made visible by adding a fluorescence labelled “secondary antibody” and measurement of fluorescent signals with flow cytometry afterwards (Borne et al. 1978). However, since pregnant mothers develop anti-HLA antibodies in approximately 50% of pregnancies at term, it is not possible to distinguish between HPA and HLA antibodies using platelets as targets (Kiefel et al. 1987). Thus, glycoprotein-specific capture tests are the gold-standard for the detection of HPA antibodies. If initial platelet serology results are negative, but FNAIT is nonetheless strongly suspected, serology should be repeated some weeks later (Socher et al. 2009). Anti- HPA antibodies are not always detectable directly after delivery (Killie et al. 2008) and detection of low-avidity antibodies, which can demonstrably induce platelet destruction in a NOD-SCID mouse model, may be missed by MAIPA due to its steps involving washing (Bakchoul et al. 2011). After affinity maturation, these antibodies can be detected some weeks after delivery. 1.4 Therapy Neonatal treatment depends on clinical presentation and platelet count. Neonates without bleeding signs and no family history of ICH should receive platelet (PLT) transfusions if the PLT count is < 25 x 109/l, in the case of bleeding signs and family history if the PLT count is < 50 x 109/l, and in the case of ICH if the PLT count is < 100 x 109/l (Lieberman et al. 2019; New et al. 2016). If antigen-negative platelets or washed platelets of the mother are not available, random buffy coat platelets can be used, although their time of survival might be shortened (Kiefel et al. 2006). The subsequent pregnancy is designated as risk pregnancy and antenatal fetal bleeding prophylaxis depends on the outcome of the previously affected child. Standard care remains the administration of intravenous immunoglobulin (IVIG) for the expectant mother (Winkelhorst et al. 2017). IVIG might block the neonatal Fc receptor on the Introduction 6 syncytiotrophoblast and on fetal macrophages, inhibiting the transplacental transfer and the binding of anti-HPA on opsonized platelets to fetal macrophages (Ueda et al. 2015). Mechanisms of the downregulation of maternal antibodies independent of the FcRn receptor pathway and how IVIG acts against ICH cannot yet be explained. Direct effects on endothelial cells or systemic platelet-mediated cytotoxicity are conceivable (Yougbaré et al. 2015), and this assumption is supported by the prevention of chronic villitis through IVIG (Althaus et al. 2005). IVIG administration can reduce the recurrence risk of ICH from 79% to 11% (Tiller et al. 2013). In order to do so, the onset of administration should be tailored to the approximate gestational age in which ICH occurred in the previous sibling (Bussel et al. 2010), mostly starting before 20 weeks of gestation because ICH mainly occurs by the end of the second trimester (Tiller et al. 2013), (Giovangrandi et al. 1990). In the 1990s, fetal blood sampling (FBS) was offered to all pregnant women with HPA- 1a antibodies, to assess the fetal platelet count (Brojer et al. 2016). FBS comes with a high risk of transplacental hemorrhage (Denomme and Fernandes 2007), boosting of alloimmunization, 1.6% fetal loss and 2.4% other complications per procedure. FBS was combined with intrauterine platelet transfusion (IUT) to avoid prolonged bleeding from the umbilical vein. Nowadays IUTs are mostly used as a rescue strategy only, since IUT alone is associated with premature birth or abortion in 1-2% per intervention and non-invasive approaches i.e. empiric IVIG gave adequate results in several studies (Winkelhorst et al. 2017). Nevertheless, IVIG administration contains disadvantages. Its isolation from a large donor pool involves the (albeit small) risk of transmission of blood associated diseases. The standard dose of 1g/kg maternal weight weekly until delivery is based on a recommendation for ITP and has not been investigated in randomized studies. Dose- related maternal side effects occur: light discomfort and headaches are common, and aseptic meningitis, renal and cardiovascular dysfunctions are rare (Paridaans et al. 2015; Winkelhorst et al. 2017). IVIG still contains isohemagglutinins, which can cause intravascular hemolysis, reticuloendothelial degradation of erythrocytes and complement-activation. Blood group A women therefore have a higher risk for anemia than blood group O women, especially reported for treatment with 2g IVIG per kg maternal body weight weekly. A reduction of the IVIG dose to 0.5 g/kg body weight Introduction 7 was reported in a case series with pregnant women with standard risk (no ICH in history) (Lakkaraja et al. 2016). Data on the value of adding steroids to IVIG therapy is insufficient (Winkelhorst et al. 2017). A prospective screening study from Norway has recommended performing a Caesarean section two to four weeks before term (Kjeldsen-Kragh et al. 2007). Practice guidelines advise induction of labor in week 37 and vaginal delivery in cases without history of ICH and a history of vaginal delivery in the previous pregnancy (Lieberman et al. 2019; van den Akker et al. 2006). In general, potentially traumatic delivery assistance which increase the neonatal bleeding risk, such as forceps delivery or vacuum delivery and scalp electrodes, should be avoided (Lieberman et al. 2019). However, trials assessing the safest delivery mode for pregnancies affected by FNAIT are needed. New experimental treatment methods are investigated, mostly focusing on administration of non-destructive antibodies which block pathogenic maternal antibodies and are unable to activate complement, phagocytosis or antibody-dependent cytotoxicity (Bakchoul et al. 2013; Ghevaert et al. 2013; Mathiesen et al. 2013). Further approaches relate to blocking FcRn to prevent the transplacental IgG-transfer (Chen et al. 2010), targeting uterine natural killer cells and/or their receptors (Yougbaré et al. 2017), targeting T-cell response in order to inhibit alloimmunization (Ahlen et al. 2009) and oral administration of peptides in a tolerogenic formulation to induce tolerance and prevent anti-HPA-1a formation (Ahlen et al. 2016). However, these new treatment methods are not yet established and cannot replace IVIG, making further research necessary. 1.5 Predictors of Incidence and Severity Many predictors for the incidence and severity of FNAIT have been proposed, but only a few appear reliable and can be used to decide on a therapy regime. One predictor for the appearance of HPA-1a alloimmunization is the HLA- DRB3*01:01 status of the mother (Delbos et al. 2016; Wienzek-Lischka et al. 2017). 90% of the immunized HPA-1bb women are carriers of HLA-DRB3*01:01, whereas only 27% of the general population carry HLA-DRB3*01:01 (Ahlen et al. 2009). Recently, a dose-dependent effect of maternal HLA-DRB3*01:01 status (negative, Introduction 8 hemi- or heterozygous) on the neonatal platelet count was described, depicting an inversely proportional relationship between maternal HLA-DRB3*01:01 allele dose and neonatal platelet counts (Kjeldsen-Kragh et al. 2019). Another positive correlation with the incidence of FNAIT might be the HLA- DQB1*02:01 status (Ahlen et al. 2009). Whether HLA-DRB4*01:01 might play a role in FNAIT is still discussed. While Delbos et al. stated that therapy response in the presence of HLA-DRB4*01:01 was significantly better and no ICH in mothers with low-avidity antibodies occurred (Delbos et al. 2016), Loewenthal et al. observed an aggravation of FNAIT progression and impaired therapy response in presence of a combination of HLA-DRB3*01:01 and HLA-DRB4*01:01. In contrast to both, Wienzek-Lischka et al. in turn found no association between FNAIT and HLA- DRB4*01:01 (Wienzek-Lischka et al. 2017). Study results on maternal antibody levels as severity predictors are inconsistent. Killie et al. demonstrated an association between antibody-levels in gestational weeks 22 and 34 and FNAIT severity for both primiparous and multiparous women (Killie et al. 2008). Tiller et al. also confirmed an association between maternal antibody levels and neonatal platelet counts (Tiller et al. 2016). In contrast, Bertrand et al. came to the conclusion that the antibody level did not correlate with the severity in index cases, but in subsequent pregnancies (Bertrand et al. 2011), while Ghevaert et al. observed that neither the potency nor the bioactivity of the maternal antibody permitted conclusions about the severity of FNAIT (Ghevaert et al. 2007a). The commonly accepted assumption that the severity of FNAIT increases with subsequent pregnancies, for example illustrated by the data of Kamphuis et al. (Kamphuis et al. 2010) and Delbos et al. (Delbos et al. 2016), has recently been challenged by Tiller et al., whose study on the natural course of FNAIT in subsequent pregnancies revealed unchanged or higher platelet counts in about 66% of subsequent siblings at the time of delivery (Tiller et al. 2016). Concerning severity, the only reliable predictor thus far seems to be the outcome of the previous child regarding ICH (Bussel et al. 1997; Kamphuis et al. 2010; Kjeldsen- Kragh et al. 2007; Porcelijn et al. 2008). If ICH has once occurred, the reoccurrence rate in subsequent pregnancies is very high (Birchall et al. 2003; Radder et al. 2003), Introduction 9 ranging between 72% and 79% including fetal death. If the index child did not present with ICH, the risk of ICH for the subsequent child without IVIG prophylaxis is estimated at about 7% (Radder et al. 2003). The mode of delivery has no influence on the risk of ICH, not even in fetuses with platelet counts below 50 x 109/l (van den Akker et al. 2006). Anti-HLA class I antibodies cannot be excluded as possible predictors of incidence or severity of FNAIT so far, since their detection in association with FNAIT can be pure coincidence and independent from the pathogenesis of FNAIT. Since anti-HLA- antibodies are detectable in 30-50% of all pregnant woman at term and their frequency rises with the number of pregnancies, Marin et al. and Delbos et al. hypothesized that anti-HLA class I antibodies work synergistically with anti-HPA and may aggravate the severity of FNAIT (Delbos et al. 2016; Marin et al. 2005). Marin speculates that anti- HLA class I antibodies, which enter the fetal circulation, are mainly neutralized by thrombocytes and thus cause fetal thrombocytopenia (Marin et al. 2005). Fetal HLA- antigens -C, -E, -G are expressed on extravillous trophoblasts and HLA-antigens -A, -B, -C can also become accessible during fetomaternal hemorrhaging, when fetal thrombocytes enter maternal circulation. There are several FNAIT case reports in which anti-HLA are the only detectable antibodies (Thude et al. 2006), though admittedly in the majority of cases the sera were not tested for antibodies against rare HPA. However, in all of these cases other particular circumstances applied, e.g. infections, asphyxia, maternal ITP (Refsum et al. 2017). A recent large study in 817 cases of suspected FNAIT demonstrated no association of maternal anti-HLA class I antibodies with FNAIT incidence and severity (Sachs et al. 2020). Recently, a correlation between maternal ABO blood groups and the severity of FNAIT has been reported. Ahlen et al. reported that women with non-O-blood groups had a higher risk of having a child with severe FNAIT than blood group O women (Ahlen et al. 2012), while alloimmunization occurred independent from ABO blood groups. Introduction 10 1.6 General Screening in Pregnancy The majority of ICH occurs in utero (Ghevaert et al. 2007b; Jin et al. 2019; Kamphuis et al. 2014; Winkelhorst et al. 2019) and often the platelet count of affected fetuses is below 20 x 109/l (Ghevaert et al. 2007b). An early general screening in pregnancy and prophylactic interventions in women where anti-HPA-1a antibodies have been detected could possibly prevent ICH. Although costs would be high, the treatment costs of FNAIT sequelae are higher (Kamphuis et al. 2010). A prophylactic screening could proceed as follows (Kamphuis et al. 2010): 1. Detection of HPA-1bb women in the first trimester through HPA genotyping or antigen detection by ELISA. 2. Fetal HPA genotyping from cell-free DNA in maternal plasma in HPA-1bb pregnant women. This replaces paternal genotyping and excludes HPA-1bb children from a heterozygous father, where no alloimmunization will occur. 3. The group of HPA-1bb women could be reduced by testing for HLA-DRB3*01:01. Note that in 10% of cases alloimmunization occurs independent from HLA- DRB3*01:01. 4. Screening HPA-1bb pregnant women at risk for immunization for anti-HPA-1a antibodies. Reconsidering from a preclinical standpoint, it seems wiser to narrow the group of women at risk for FNAIT as much as possible before the application of fetal HPA genotyping and thus change the proposed order and reverse step 3 before step 2. Currently, a general screening program is being evaluated in the Netherlands (Kjeldsen- Kragh et al. 2007; Tiller et al. 2017). In Germany, however, no screening measures have yet been implemented. Introduction 11 1.7 Prophylaxis of Maternal Immunization in HDFN and FNAIT The concept of prophylaxis of maternal immunization in FNAIT has been proposed in analogy to the efficacy of anti-RhD prophylaxis in RhD-negative women to prevent hemolytic disease of the fetus and newborn (HDFN). In HDFN, anti-RhD and anti-ABO antibodies are frequently involved in prenatal and postnatal disease in the fetus or newborn, respectively. Albeit ABO incompatibility between mother and fetus is present in between 14% to 20% of pregnancies (depending on ethnicities) (Akanmu et al. 2015; Cariani et al. 1995; Clarke 1973), only approximately 1% of children is affected by ABO hemolytic disease (ABO-HD) (Peevy and Wiseman 1978; Voak and Bowley 1969). The course of the disease is mainly benign and includes jaundice, likely because most of the transferred antibodies are neutralized by ABO antigens on other cells tissues and plasma proteins (Ottenberg 1911). The quantitative expression of ABO antigens varies widely among infants’ RBCs (Grundbacher 1980) and thus, red blood cells of the fetus with low expression of ABO antigens survive in the presence of maternal ABO antibodies and disease is almost always observed only postnatally. In HDFN due to anti-RhD antibodies, severe manifestations with fetal anemia and hydrops, severe hyperbilirubinemia and kernicterus in the newborn used to occur more often. Since immunization of the mother occurs mainly during delivery by fetomaternal (micro-) hemorrhage, the subsequent child is at risk (Costumbrado and Ghassemzadeh 2019). By introduction of RhD phenotyping for all primiparous women and anti-D- immunoprophylaxis in RhD-negative pregnant women more than 50 years ago, the incidence of HDFN due to anti-RhD decreased from 16% (Bowman 1997) to less than 0.1% (Hendrickson and Delaney 2016). By default, possibly after non-invasive detection of fetal RhD, RhD-negative women in Germany receive 300 μg anti-D- immunoglobulin between 28th and 30th week of gestation as well as within 72 hours after birth in case the newborn is RhD-positive (and in case of invasive interventions or fetomaternal hemorrhage during pregnancy) (Legler 2018). Interestingly, anti-D- immunoprophylaxis not only reduces the risk for anti-D-alloimmunizaton, but also for non-D-alloimmunization against other antigens located on RBCs (Zwiers et al. 2018). One possibility for explaining this non-antigen specific immunosuppression is the Introduction 12 “rapid clearance”, i.e., the destruction of the RBC by splenic macrophages before the maternal immune system is triggered for alloimmunization (Woodrow et al. 1975). Although the rate of FNAIT in newborns of primiparous women is already high (at least 25%), alloimmunization may occur in association with delivery in a major proportion of anti-HPA-1 immunized women (Killie et al. 2008; Kjeldsen-Kragh et al. 2007; Kjeldsen-Kragh et al. 2012). Therefore, FNAIT may resemble anti-RhD HDFN more than previously assumed (Skogen et al. 2009). Profnait, a project set up by eleven European project partners and promoted by the European Union since 2012, works to establish a safe and effective FNAIT immunoprophylaxis (Geisen 2013; Kjeldsen-Kragh et al. 2012). In HPA-1bb women who gave birth to an HPA-1a positive child without the occurrence of alloimmunization, it is planned to administer anti-HPA-1a immunoprophylaxis immediately after delivery (similar to anti-D-immunoprophylaxis), in order to degrade fetal HPA-1a positive platelets before they are able to trigger the maternal immune system (Kamphuis et al. 2010). It is still uncertain whether a postpartal prophylaxis for FNAIT would be as successful as anti-D-immunoprophylaxis, if antigens on particles of the syncytiotrophoblast alternative to fetal platelets trigger for already early immunization in FNAIT. However, in cases of severe FNAIT complicated by intracranial hemorrhage, maternal immunization nonetheless occurs mostly before delivery (Jin et al. 2019). To prevent all severe FNAIT cases, antenatal prophylaxis is required in addition to postnatal prophylaxis (Kjær et al. 2020). Assuming that fetal platelet antigen expression is initiated by 12 weeks of gestation and since the transplacental IgG transfer must be active before gestational week 20 (in order to cause intracranial hemorrhage before gestational week 18), available prophylaxis for pregnant women positively screened for FNAIT would need to be administered early in pregnancy (Bussel et al. 2010; Jin et al. 2019). Introduction 13 1.8 ABO Incompatibility In ABO-incompatible pregnancies, isohemagglutinins are directed against the fetal blood group antigens. As early as 1943, Levine noticed that the prevalence of anti-RhD alloimmunization was higher in ABO-compatible than in ABO-incompatible pregnancies (Levine 1943, 1959) pregnancies. This observation was confirmed by several authors and led to the development of anti-RhD immunoprophylaxis, trying to mimic the described protection through ABO incompatibility with a suitable antibody. In 1997 Bowman depicted the risk of anti-RhD alloimmunization to be 16% in ABO- compatible pregnancies, but only 2% in ABO-incompatible pregnancies (Bowman 1997). Recently, Zwiers confirmed that ABO incompatibility has a preventive effect on anti-D and non-D alloimmunization (Zwiers et al. 2018). Furthermore, Zizka et al. hypothesize that ABO incompatibility significantly reduces the risk of severe fetomaternal hemorrhage (FMH). They suspect that in case of fetomaternal blood- contact, agglutination almost immediately closes the pathological leaks of placental vessels and the fetal blood loss is smaller (Zizka et al. 2008). Regarding FNAIT, in 1977 Gratwohl and Shulman described 25 cases of “isoimmune neonatal thrombopenia” (FNAIT), all resulting from ABO-compatible pregnancies and assumed that not only fetomaternal incompatibility regarding platelet antigens but also fetomaternal compatibility regarding ABO blood groups are requirements for the development of FNAIT. Consequently, they speculated ABO incompatibility might prevent alloimmunization, probably because fewer fetal platelets enter maternal circulation or their time of survival is shortened (Gratwohl and Shulman 1977). 1.9 ABO Blood Group System In 1900 Karl Landsteiner described three blood groups: A, B and C (later called O), after observing different agglutinating effects in mixing sera and RBC of different individuals together (Schwarz and Dorner 2003). Based on Landsteiner’s discoveries, the ABO blood group antigens on erythrocytes, small carbohydrate determinants on glycoproteins and glycolipids, were identified. Products of the encoding genes are glycosyltransferase enzymes, which transfer the immunodominant terminal monosaccharide to the membranous acceptor substrates. Introduction 14 The ABO system is now structured in four major groups A, B, O, AB (Yamamoto et al. 1990b) and various subgroups and variants. Since ABO antigens are not only expressed on erythrocytes but on other cells and tissues as well, e.g., leukocytes and epithelial cells, they are often referred to as “histo-blood group antigens.” Several genes collaborate for their synthesis. The glycosyltransferases’ genes are located on chromosome 9 q34.1-q34.2 and include 7 exons which span over 18kb. 77% of the coding sequences and 91% of the catalytically active soluble transferase proteins are comprised in exon 6 and 7 (Bennett et al. 1995; Ferguson-Smith et al. 1976). All ABO phenotypes are closely related to DNA sequence variants of the glycosyltransferases. A sequence encoding for the frequent phenotype A1, a 1062 bp long coding region (Yamamoto et al. 1990a) is used as reference to describe the differences between ABO alleles, which are found mainly in exons 6 and 7, where substrate specificity and activity of the glycosyltransferase is coded (Bennett et al. 1995). Two deletions, seven sense mutations and ten missense mutations recurring in various alleles depict the evolutionary history of the ABO gene (Blumenfeld and Patnaik 2004). Blood group A individuals express the α13 N-acetylgalactosaminyltransferase (GTA), which transfers N-acetylgalactosamine (GalNAc), while B individuals express α13 galactosyltransferase (GTB), which transfers galactose (Gal) to the acceptor substrate called H substance. Uridine diphosphate-GalNAc (UDP-GalNAc) operates as donor substrate for GTA and UDP-Gal for GTB. The DNA sequence of A- and B- glycosyltransferases differs in 7 base substitutions, which results in the exchange of 4 amino acids of altogether 353 amino acids (Yamamoto et al. 1990b). The amino acids in position 266 and 268 communicate with the donor- and acceptor-substrate and therefore distinguish the glycosyltransferase enzymes substrate specificity and activity (Yamamoto and McNeill 1996; Yip 2002). Leu/Met266 is most important for the selection of donor carbohydrates; the smaller Leu266 in GTA interacts with the larger acetamido group of UDP-GalNAc, while the larger Met266 in GTB interacts with the smaller UDP-Gal (Yamamoto and McNeill 1996). However, GTA and GTB share a small degree of overlap in their substrate specificity, i.e. GTA can transfer Gal if GalNAc is absent or barely there and vice versa (Yates and Watkins 1982). AB individuals express both the A- and B-glycosyltransferases. Introduction 15 Alleles ABO*A1.01 and ABO*A2.01 differ from each other in a base substitution (nucleotide position 297, exon 6) and a deletion (nucleotide position 1059, exon7), which leads to the loss of a stop codon and consecutive shift of the reading frame and results in an ABO*A2.01 glycosyltransferase prolonged by 21 amino acids at the C- terminus (Yamamoto et al. 1992). Its enzymatic activity is weakened 5 to 10 times, thus the antigenic density of ABO*A2.01 RBC is only 20% of the ABO*A1.01 antigenic density on RBC (Schachter and Michaels 1971). The ABO*O.01.02 allele (= O1v) arises through a deletion at position 261 in the ABO*A1.01 DNA sequence, resulting in the generation of a premature stop codon. The ABO*O.01.01 allele (= O1) is formed subsequently through interallelic exchange between ABO*A1.01 and ABO*O.01.02 (Roubinet et al. 2004). These alleles encode truncated proteins with no catalytic domain. In contrast to the nonsense mutations generating ABO*O.01.02 and ABO*O.01.01, a single missense mutation (base substitution in position 802, exon 7) generating allele ABO*O.02 (= O2) has an inactivating effect on the ABO*A1.01 glycosyltransferase albeit lacking 261delG (Yamamoto et al. 1993). In ABO*O.02, arginine in position 268 blocks the donor-GalNAc binding site of the A-glycosyltransferase (Lee et al. 2005). The eventuality that non-deletional O alleles can likely produce small amounts of the A antigen cannot yet be ruled out (Seltsam et al. 2005; Yazer et al. 2008). If ABO*O.02 is not considered in genotyping, one might mistake genotype ABO*O.02/O.01 for ABO*A1.01/O.01. The H acceptor substance is synthetized by a α12 fucosyltransferase encoded through the FUT1-gene on chromosome 19. The dominant H allele is necessary for the expression of the α12 fucosyltransferase, however, in rare cases the FUT1-gene is homozygous h and no fucosyltransferase, and accordingly, no acceptor substrate for A- and B- glycosyltransferases will be synthesized, which results in the Bombay phenotype without any H-substance and hence no ABO antigens (Morgan and Watkins 1969). Approximately 80% of the Caucasian population also has ABO antigens in soluble form in plasma and other body fluids (Kelly et al. 1995). Indispensable presupposition is the secretion of H-substance in body fluids, which is controlled through FUT2, a gene Introduction 16 closely related to FUT1. Individuals who inherit the dominant allele FUT2 Se are called secretors; Se is dominant over non-secretor Se (Morgan and Watkins 1969). The ABO system meets the criteria for definition as a blood group because natural antibodies, isohemagglutinins, are built as a result of immunization against A-and B- substances on bacteria (Ahlstedt et al. 1977; Springer et al. 1959) during the settlement and construction of intestinal microbiota. Therefore, neonatal IgM isohemagglutinins are first detected at the age of three months and reach adult titers within the age of 5 to 10 years (Maur et al. 1993). This immunization occurs only against the ABO phenotypes that the neonate itself lacks. AB individuals therefore lack isohemagglutinins in their sera completely, while O individuals express antibodies against RBC phenotypes A and B, and individuals with Bombay phenotype even generate antibodies against the H-substance, reacting with all red blood cells regardless of ABO group. 1.10 ABO Antigens on Platelets On platelets, the blood group A and B determinants are expressed on various glycoproteins (e.g. GPIb, GPIIa, GPIIb, GPIIIa, GPIV, GPV, CD109), PECAM and glycosphingolipids (maintained from the time as megakaryocyte, newly synthetized, passively adsorbed from plasma to a minor extent). The ABO*A1.01 antigen expression on adult platelets varies widely, while ABO*A2.01 and B determinants almost always demonstrate a minimal level of expression. (Cooling et al. 2005; Curtis et al. 2000; Farias et al. 2016; Hou et al. 1996; Kelton et al. 1998; Moureau and Andre 1954; Santoso et al. 1991; Skogen et al. 1988; Stockelberg et al. 1996). In most cases, the platelet ABO*A1.01 antigenic density is so low that ABO- incompatible platelet transfusions are thought to not have much impact on the duration of platelet survival (Curtis et al. 2000). However, approximately 7% of adults with the ABO*A1.01 phenotype and 4% of adults with the B phenotype are "high expressers" (HXP) for their ABO determinants on platelets (Curtis et al. 2000; Ogasawara et al. 1993), defined through an increase of the antigenic density beyond two standard deviations. Respectively, approximately 5% of A and B phenotype adults are “low expressors” (O'Donghaile et al. 2020). Introduction 17 In the case of HXP type 1, the antigenic density is three times higher than average, and for HPX type 2, seven times higher than average. In HXP, ABO*A1.01 antigens on platelets are mainly expressed on GPIIb and PECAM; as a consequence, incompatible transfusions can also affect the duration of platelet survival (Curtis et al. 2000). The ABO*A1.01 antigenic density on HXP RBC increases the H antigen expression drops to the same effect, but on platelets, the H antigen seems much more strongly expressed than the A antigen. Cooling et al. speculated that the ABO antigen expression on platelets depends more on the activity of the fucosyltransferase FUT1, proposing that the H antigen on platelets is expressed on sterically hidden, complex carbohydrates. The ABO glycosyltransferase may not modify these residues to ABO*A1.01 antigens (Cooling et al. 2005). The H-to-A antigen expression ratio on platelets appears to be directly related to the genotype of the individual, with ABO*A1.01/A1.01 individuals showing a lower H-to-A ratio than for example ABO*A1.01/O.01 individuals (DeLelys et al. 2013; Xu et al. 2019). O’Donghaile et al. concluded that the ABO*A1.01/A1.01 genotype must be one of the major determinants of ABO high-expresser trait (O'Donghaile et al. 2020), while factors such as epigenetic effects, ABO gene transcription rates and messenger RNA stability may also contribute to the expression traits and explain the variety of antigen expression among individuals of the same genotypes (DeLelys et al. 2013). The HXP phenotype, among others, is of importance, with Curtis et al. reporting a case of FNAIT with an RBC phenotype O mother and a Type 2 HPX child with RBC phenotype B. No other antibodies than anti-B were detectable. They may have induced FNAIT despite the competition of antibody binding through erythrocytes and other tissues (Curtis et al. 2008). Likewise, Kato et al. presented a case where FNAIT was possibly induced by maternal high-titer anti-A antibodies. Anti-HPA or anti-HLA were excluded serologically. Unfortunately, these authors were not able to examine the A antigen expression on the infant´s platelets because of its death due to severe ICH, but maternal anti-A reacted strongly with paternal platelets. (Kato et al. 2013). The ISBT (International Society of Blood Transfusion) terminology (Table 1) is used throughout this paper. Since further discrimination between ABO*O.01.01 and ABO*O.01.02 was not implemented, these alleles were summarized as ABO*O.01. Introduction 18 Table 1: ABO terminology Phenotype ISBT 001 (version v1.1 170123) 1 dbRBC/ BGMUT2 Traditional designation A1 ABO*A1.01 A101 A1 A2 ABO*A2.01 A201 A2 B ABO*B.01 B101 B O ABO*O.01.01 O01 O1 ABO*O.01.02 O02 O1v ABO*O.02.01 O03 O2 1.11 Hypothesis and Aim of the Study FNAIT is a disorder which, in severe cases, can proceed to fetal/neonatal death or neurologic sequelae. Non-invasive monitoring of the fetal platelet count is not possible. Prophylaxis of fetal bleeding during pregnancy itself is not harmless either and cannot be applied to all pregnant women at risk. Thus far, a predictor for FNAIT severity, which may be relied on in order to find the necessary intensity of therapy with the least significant side-effects, does not yet exist, especially not for antenatal management in pregnant women with a history of FNAIT in a previous pregnancy. An FNAIT screening program in the general population of pregnant women would also require reliable prediction of the fetal bleeding risk to avoid overtreatment. In hemolytic disease of the fetus and newborn (HDFN), ABO incompatibility between mother and fetus has a profound effect on maternal immunization to fetal red blood cell antigens (Bowman 1997; Levine 1959; Zwiers et al. 2018). It is hypothesized that the ABO group of the mother and ABO incompatibility between pregnant mother and fetus may be associated with the incidence of anti-HPA-1a immunization and/or may represent a disease-modifying factor (Ahlen et al. 2012; Gratwohl and Shulman 1977). The aim of this study is to verify whether there is a genetic association between ABO incompatibility between mother and fetus, and the incidence and severity of FNAIT. To this purpose, maternal and fetal/neonatal ABO blood groups are determined. 1 International Society of Blood Transfusion 2 Blood Group Antigen Gene Mutation Database, Patnaik et al. (2012). Introduction 19 In the course of blood group examination, the proposed correlation between maternal ABO blood groups and FNAIT severity (Ahlen et al. 2012) is investigated and a possible association between fetal/neonatal ABO blood groups and FNAIT severity is tested. The results of this study may enable refined prediction of maternal anti-HPA-1a immunization risk as well as prediction of fetal bleeding risk in already immunized mothers. Furthermore, results may be informative for the development of prepartal and/or postpartal anti-HPA-1a immunoprophylaxis strategies in HPA-1bb pregnant women. Material and Methods 20 2 Material and Methods 2.1 Subjects 165 blood samples from mother-child pairs with a history of FNAIT living in Germany were collected by the Center for Transfusion Medicine and Hemotherapy of the Justus Liebig University Giessen between 2000 and 2015. The FNAIT diagnosis was based on maternal anti-HPA antibodies or HPA-1a antigen incompatibility in association with typical clinical data entries, including the neonatal platelet count (nadir) and the presence or absence of intracranial hemorrhage. Thrombocytopenia was defined as a platelet count ≤ 150 x 109/l and severe thrombocytopenia less than 50 x 109/l. The common denominator was the circumstance that neither the mother nor the unborn child received prenatal therapy in any form. In average, maternal age upon childbirth was 30½ years and the affected child arose from the second pregnancy as well as second delivery. Of the 165 affected children, 113 were male, 51 female and one intersex (at a later record classified as male). 15 children suffered from intracranial hemorrhage; data presented in table 2. Material and Methods 21 Table 2: Data of children suffering from ICH. Maternal antibody Maternal ABO genotype Allele 1 Allele 2 Neonatal ABO genotype Allele 1 Allele 2 ABO- incompatible Neonatal PLT count nadir (/μl) Birth weight (g) Anti- HPA-1a ABO*O.01 ABO*O.01 ABO*O.01 ABO*O.01 no 8000 2810 Anti- HPA-1a ABO*O.01 ABO*O.01 ABO*O.01 ABO*O.01 no 3000 unknown Anti- HPA-1a ABO*O.01 ABO*O.01 ABO*A1.01 ABO*O.01 yes 21000 2275 Anti- HPA-1a ABO*O.01 ABO*O.01 ABO*O.01 ABO*O.01 no 13000 unknown Anti- HPA-1a ABO*O.01 ABO*O.01 ABO*A1.01 ABO*O.01 yes 13000 unknown Anti- HPA-1a ABO*O.01 ABO*O.01 ABO*O.01 ABO*O.01 no 30000 2275 Anti- HPA-1a ABO*O.01 ABO*O.01 ABO*O.01 ABO*O.01 no 4000 3205 Anti- HPA-1a ABO*A1.01 ABO*O.01 ABO*O.01 ABO*O.01 no 5000 3680 Anti- HPA-1a ABO*A1.01 ABO*O.01 ABO*O.01 ABO*O.01 no 45000 3000 Anti- HPA-1a ABO*A1.01 ABO*O.01 ABO*O.01 ABO*O.01 no 11000 3270 Anti- HPA-1a ABO*A1.01 ABO*O.01 ABO*O.01 ABO*O.01 no 16000 3320 Anti- HPA-1a ABO*A1.01 ABO*B.01 ABO*B.01 ABO*O.01 no 6000 2230 Anti- HPA-1a ABO*B.01 ABO*O.01 ABO*B.01 ABO*O.01 no 9000 2720 Anti- HPA-1a ABO*B.01 ABO*O.01 ABO*A1.01 ABO*B.01 yes 26000 unknown Anti- HPA-1a ABO*B.01 ABO*O.02 ABO*O.01 ABO*O.02 no 5000 3420 Material and Methods 22 All ICH-kids arose from HPA-1a incompatible pregnancies (the most common cause) and had severe thrombocytopenia. Most pregnancies were ABO-compatible. In 125 cases, FNAIT was caused by maternal anti-HPA-1a (anti-PLA1), including 4 cases caused by a combination with anti-HPA-5b (anti-Bra) and one case in combination with anti-HPA-2b (anti-Koa), as well as one case combined with anti-A. 24 cases were caused by anti-HPA-5b (anti-Bra) alone, 4 cases by anti-HPA-3a (anti-Baka) and 2 cases by anti-HPA-15a (anti-Govb). Anti-HPA-15b (anti-Gova), anti-HPA-2b (anti-Koa) and anti-HPA-8bw (anti-Sra) were responsible for one case each. The anti-HPA antibodies were detected using MAIPA (monoclonal antibody immobilization of platelet antigen assay). In 7 cases, FNAIT was clinically diagnosed but no antibodies were detectable. In these cases, however, fetomaternal HPA-1a incompatibility was confirmed through HPA- genotyping of mother and child. The Ethics Committee of the Medical Faculty at the Justus Liebig University in Giessen, Germany, officially approved the use of all human material (vote 21.07.2009, docket file nr. 82/09). 2.2 Materials and Manufacturer 2.2.1 Reagents and buffer Table 3: Reagents, in alphabetical order. Affinity Pure Goat Anti-Mouse IgG, Fc8 Fragment Specific Jackson ImmunoResearch Laboratories, INK, Pennsylvania, USA (ordered via: DIANOVA GmbH, Ham- burg, GER) Aqua destillata Baxter Deutschland GmbH, Unter- Schleißheim, GER BSA (Bovine Serum Albumine 22%) Ortho Clinical Diagnostics GmbH, Neckargemünd, GER C6H8O7 x H2O (citric acid monohydrate) Merck KGaA, Darmstadt, GER Material and Methods 23 CaCl2 (calcium chloride dihydrate) Sigma-Aldrich Chemie GmbH, München, GER EZ1® DNA Blood 350 µl Kit Qiagen GmbH, Hilden, GER, REF H2O2 (hydrogen peroxide) Merck KGaA, Darmstadt, GER H2SO4 (sulphuric acid (2.5 N)) Merck KGaA, Darmstadt, GER Isotonic saline solution 0.9% Braun injection solution B. Braun Melsungen AG, Melsungen, GER NABO*A2.01CO3 (sodium carbonate) Merck KGaA, Darmstadt, GER NABO*A2.01HPO4 x 12 H2O (sodium hydrogen phosphate) Merck KGaA, Darmstadt, GER NaHCO3 (sodium hydrogen carbonate) Merck KGaA, Darmstadt, GER NaN3 (sodium acid) Merck KGaA, Darmstadt, GER Nuclease-free water Promega, REF: P119E, USA OPD-tablets (2 mg Orthophenylenediamine) Kem-En-Tec Diagnostics A/S, Taastrup, DK PBS (Phosphate Buffered Saline (1x)) PAA Laboratories GmbH, Cölbe, GER Peroxidase-conjugates Affinity Pure Goat Anti-Human IgG Fc8 Fragment Specific Jackson ImmunoResearch Laboratories, INK, Pennsylvania, USA (ordered via: DIANOVA GmbH, Hamburg, GER) Probes for ABO genotyping Applied Biosystems by Thermo Fisher Scientific, LSG Strategic Oligo Solutions, California, USA TaqMan® Universal 2× PCR Master Mix Applied Biosystems, Life Technologies LTD, Warrington, UK TRIS PUFFERAN® ≥99.3%, Buffer Grade (Tris-(hydroxymethyl)- aminomethan) Carl Roth GmbH + Co. KG, Karlsruhe, GER Triton™ X-100 Sigma-Aldrich Chemie GmbH, München, GER Tween20 ® Sigma-Aldrich Chemie GmbH, München, GER Material and Methods 24 Table 4: Buffer, in alphabetical order. Buffer Components Coating buffer (durability: 14 d) 1.59 g NABO*A2.01CO3 2.93 g NaHCO3 0.2 g NaN3 dissolve in 1000 ml distilled water (target pH 9.6) PBS/BSA 2% 10 ml Phosphate Buffered Saline (PBS) 1 ml 22% Bovine Serum Albumine (BSA) Substrate buffer peroxidase (durability: 3 months) 3.65 g C6H8O7 x H2O 11.94 g NABO*A2.01HPO4 x 12 H2O bring volume up to 500 ml with distilled water (target pH 5.0), take 7.5 ml of this solution and add 3 OPD tablets and 7.5 µl H2O2 Trisbuffer / washing buffer (durability: 3 months) 3.63 g Tris, dissolve in 3 l isotonic saline solution (target pH 7.4) 15 ml Triton x 100 15 ml Tween20 1.5 ml 1mCaCl2 2.2.2 Consumables Table 5: Consumables, in alphabetical order. MicroAmp® Fast 96-Well Reaction Plate Applied Biosystems, Life Technologies, Beijing, China MicroAmpTM Optical Adhesive Film Life Technologies Corporation, Carlsbad, California, USA Microtiter tray (high binding capacity, F-form) Greiner Bio One International GmbH, Kremsmünster, GER Pipettes tips: 10 μl, achromatic 200 μl, yellow 1000 μl, blue Biozym tips, achromatic, 10 μl Sarstedt, Nürnbrecht, GER Sarstedt, Nürnbrecht, GER Sarstedt, Nürnbrecht, GER Biozym Scientific GmbH, Material and Methods 25 Filter Mikro reach low binding, 10 μl Filter Tip XL Low Binding, 200 μl Filter Tip, 1250 μl Hess. Oldendorf, GER Biozym Scientific GmbH, Hess. Oldendorf, GER Biozym Scientific GmbH, Hess. Oldendorf, GER Biozym Scientific GmbH, Hess. Oldendorf, GER Reaction tubes: 0.5 ml 1.5 ml 1.5 ml DNA LoBind Tubes Sarstedt, Nürnbrecht, GER Sarstedt, Nürnbrecht, GER Eppendorf AG, Hamburg, GER 2.2.3 Hardware Table 6: Hardware, in alphabetical order. Centrifuges: Biofuge pico Rotina 380 Heraeus Instruments GmbH, Hanau, GER Hettich Zentrifugen, Mühlheim a.d. Ruhr, GER DNA extraction: Bio Robot EZ1 Advanced XL Qiagen Instruments AG, Hombrechtikon, CH pH measurement: inoLab pH Level 1 Wissenschaftlich-Technische Werkstätten GmbH, Weilheim, GER Photometer: NanoDrop® ND-1000 Tecan Sunrise NanoDrop Technologies, Wilmington, USA Tecan Austria GmbH, Grödig, Austria Pipettes: Proline® Plus 0.5-10 µl Eppendorf Reference (0.5-10 µl) Eppendorf Reference (10-100 µl) Sartorius AG, Göttingen, GER Eppendorf AG, Hamburg, GER Eppendorf AG, Hamburg, GER Material and Methods 26 Eppendorf Research (0.5-10 µl) Eppendorf Research (10-100 µl) Eppendorf Research (100-1000 µl) Starlab MicroOne® Eppendorf Multipette Eppendorf Titerman 4908 Eppendorf AG, Hamburg, GER Eppendorf AG, Hamburg, GER Eppendorf AG, Hamburg, GER Starlab GmbH, Ahrensburg, GER Eppendorf AG, Hamburg, GER Eppendorf AG, Hamburg, GER StepOnePlus™ Real-Time PCR System Applied Biosystems, Life Technologies GmbH, Darmstadt, GER Vortexer: MS2 Minishaker IKA Reax 2000 Sea Star Vortexer All-in-one Heidolph Instruments GmbH & Co.KG, Schwabach, GER Biozym Scientific GmbH, Hessisch Oldendorf, GER 2.2.4 Software Table 7: Software, in alphabetical order. IBM SPSS Statistics Version 25 for Windows IBM Deutschland GmbH, Ehningen, GER MAGELLAN V7.2 TRA. 2PC PAC SUNRISE Tecan Austria GmbH, Grödig, Austria Microsoft® Excel 2010 Microsoft Corporation, Redmond, USA Microsoft® Office 365 Word Microsoft Corporation, Redmond, USA NanoDrop-1000 V. 3.8.1 NanoDrop Technologies, Wilmington, USA Prism 8, GraphPad Inc. GraphPad Software, La Jolla, California, USA StepOnePlus™Software v2.3 Thermo Fisher Scientific, Life Technologies GmbH®2012, Darmstadt, GER Material and Methods 27 2.3 MAIPA The verification of anti-HPA was performed serologically via indirect monoclonal antibody immobilization of platelet antigens (MAIPA); after incubation of test platelets with maternal serum, the relevant glycoproteins were able to be immobilized by adding monoclonal murine antibodies. Maternal and murine antibodies and HPA form a stable trimolecular complex, which was liberated during platelet solubilization and afterwards fixated via a murine antibody on a microtiter tray coated with polyspecific anti-murine- antibodies. By adding enzyme-labelled anti-human-IgG, the antibody complex with the maternal antibody was visualized through substrate reaction. This cross-match procedure allowed the detection of fetal-maternal incompatibilities beyond the frequent HPA-1 and -5 and of HLA antibodies. Prerequisite was the existence of monoclonal antibodies for all the relevant glycoproteins (Table 8). Table 8: Monoclonal murine antibodies used for MAIPA. Specificity Clone Isotype Form Sales GP IIb IIIa Ü Gi 5 Liquid Dr. Santoso GP Ia IIa Ü Gi 9 Liquid Dr. Santoso GP Ib IX CD42a FMC25 IgG1 Liquid Serotec GP V CD 42d SW16 IgG1 Liquid CLB HLA Cl. I ABO* B.01G6 IgG2a Purified Immunotech The MAIPA was conducted by employees of the laboratory from the Center for Transfusion Medicine and Hemotherapy. Test platelets were isolated from EDTA-coagulated blood through cell-fractioning, washed three times and stored at 4 °C for at least 12 hours in isotonic saline solution. Samples, not stored on ice, were examined within 2 days after blood drawing. Examination of hemolytic samples was not possible. A microtiter tray was coated with 100 μl buffer per well (consisting of coatingbuffer and anti-Mouse-IgG in a ratio of 1:500) and incubated for at least 120 minutes at 37 °C or allowed to rest for 24 hours at room temperature. Material and Methods 28 20 million test platelets (in one reaction tube) were centrifuged (1 min, 10000 rpm) and resuspended in 30 μl PBS 2% BSA immediately after the supernatant was discarded. Serum/platelet-free plasma was also centrifuged (1 min, 10000 rpm), then 20 – 50 μl were added to the platelets and the mixture was incubated for 30 minutes at 37 °C. After a wash step with 100 μl isotonic saline solution and centrifugation (1min, 10000 rpm), the supernatant was again discarded and the platelets were again resuspended in 30 μl PBS 2% BSA. After adding 10 μl of monoclonal mouse antibodies, the mixture was incubated for 30 minutes at 37 °C and then washed three times in 100 μl isotonic saline solution, then centrifugation was repeated (1 min, 10000 rpm) and the supernatant discarded. For platelet lysis, the remaining platelet pellet was resuspended in 100 μl solubilisation buffer (containing Tris-buffered saline and Triton x 100 in the same ratio as the washing buffer), incubated for 30 minutes at 4 °C and then centrifuged for 30 minutes at 4 °C (13000 rpm). Afterwards, 50 μl supernatant was carefully taken (without stirring up insoluble platelet residues) and diluted with 200 μl washing buffer. During solubilization time, the microtiter tray was washed four times (with 200 μl washing buffer / well) and blocked for 15 minutes at 4 °C (with 100 μl washing buffer / well). Subsequently, the microtiter tray was beaten and 100 μl of the diluted supernatant was added in each well, then the tray was incubated at 4 °C for 90 minutes. After this, the tray was washed again four times with 200 μl washing buffer / well. 100 μl peroxidase labelled Affinity Pure Goat Anti-Human IgG was added per well. After incubation for 120 minutes at 4 °C the tray was washed again in the usual manner (4x, 200 μl washing buffer / well) to remove excess antibodies. Then 100 μl substrate puffer per well was added and the tray was incubated in the dark for 10 minutes at room temperature. Finally, the color reaction was stopped with 50 μl 2.5 N H2SO4. Using a Photometer (Sunrise™, Tecan), the extinction was measured at 450 and 620 nm. Each extinction and a blank value (distilled water) was measured twice. The single extinction values should not differ by more than 20%. The patients mean value of extinction was subtracted from the mean blank value, and the resulting value of optical density (OD-value) was interpreted according to table 9. A positive control for each monoclonal murine antibody was also included. Material and Methods 29 Table 9: Interpretation of OD-Values. HPA-genotyping was performed by employees of the laboratory at the Center for Transfusion Medicine and Hemotherapy using TaqMan® real time PCR. 2.4 DNA Extraction DNA was isolated from whole blood and buffy coats using the “Bio Robot EZ1 Advanced XL” and its proven EZ1® DNA Blood 350 μl Kit. The DNA extraction was carried out as specified by the manufacturer, using the preprogramed protocol on EZ1 Advanced XL Cards. After arranging the disposable filter-tips and tip-holders, elution tubes and the sample tubes (filled with 350 μl EDTA-coagulated blood) and loading the Bio Robot with the prefilled cartridges (Geno Prep Cartridge B 350, Qiagen) containing all required reagents, the instrument prepared the DNA extraction automatically. After the cell lysis and in the presence of a chaotropic salt, which keeps the proteins denatured and soluble, the DNA bound to the silica surface of magnetic particles. Using a magnet, these particles were separated from the lysates and the DNA was washed and eluted in a 200 μl elution volume in the elution tubes. With this set-up, 14 samples can be processed in approximately 20 minutes (EZ1 DNA Blood Handbook 04/2010). Assessment of OD-Values Altitude OD-Value Negative < 0.15 Borderline (+) 0.15-0.20 Weakly positive + 0.21-0.40 Moderately positive ++ 0.41-0.80 Explicitly positive +++ 0.81-1.20 Strongly positive ++++ > 1.21 Material and Methods 30 2.5 Measurement of DNA Concentration The DNA concentration was measured using the NanoDrop®ND-1000 Spectrophotometer. After 2 μl samples were placed between two fiber optic cable ends, a ray of light from a pulsed xenon flash lamp was analyzed by the spectrometer after passing through the sample. The sample absorbance at a given wavelength was calculated relative to the initial absorbance of distilled water set as a baseline value. If the ratio of sample absorbance at 260 and 280 nm was lower than 1.8, the DNA sample was considered pure. The calculated absorbance was then correlated with the concentration using a modified Beer-Lambert equation with an extinction coefficient for double stranded DNA of 50 ng/ml. The ND-1000 software V.3.8.1 run from a PC was used to control the instrument. A template volume of 2 μl contained enough DNA for a TaqMan® real time PCR. 2.6 ABO Genotyping Maternal and fetal ABO genotyping was performed with a TaqMan® real time PCR assay to detect the major ABO alleles ABO*A1.01, ABO*A2.01, ABO*B.01, ABO*O.01 (O1/O1v) and ABO*O.02 (O2). Further discrimination between ABO*O.01.01 and ABO*O.01.02 was not implemented. To identify the mentioned ABO alleles, 4 SNP sites were selected (Table 10) and detected by customized primers and probes. Material and Methods 31 Table 10: The selected SNP sites in the mentioned ABO alleles. Allele ABO*A1.01 served as consensus sequence. It was only detected by exclusion of the other alleles. Probes were stored at -15 to -25 °C. Samples were stored at -25 to -80 °C. The TaqMan® probes were labelled with a fluorescent reporter dye at the 5′ end as well as a fluorescent quencher dye at the 3′ end (Table 11). Until the DNA-polymerase separated the 5′ reporter dye from the physical proximity to the quencher dye due to its 5′- 3′ exonuclease activity, the fluorescence of the reporter dye would not be emitted. Depending on which probe hybridized (the wildtype or the variation specific probe), different fluorescence signals were able to be detected and monitored in real time. SNP site ABO*O.01.01, ABO*O.01.02 ABO*B.01 ABO*O.02 ABO*A2.01 rs-number 77641731 8176743 41302905 56392308 Exon 6 7 7 7 Chr:bp 9:133257521 9:133256028 9:133255929 9:133255670 Codon Val87/Thr88=fs 235 268 354 Nucleotide position c.260/262insG 703 802 1061 ABO*A1.01 (consensus) G G G C ABO*A2.01 G G G deletion ABO*B.01 G A G C ABO*O.01.01, ABO*O.01.02 deletion G G C ABO*O.02 G G A C Material and Methods 32 Table 11: Sequences, quencher and reporter molecules of TaqMan® probes used for the detection of the selected SNPs. The target in each allele is underlined. Deletions are indicated with an asterisk. For alleles ABO*O.01, ABO*O.02 and ABO*A2.01 probe 1 always detected the allele- specific target and probe 2 detected ABO*A1.01 consensus sequence and vice versa for allele ABO*B.01. For each probe, a reaction mix was set up in 1,5 ml DNA LoBind Tubes. This reaction mix consisted of 10 μl TaqMan® Universal 2× PCR Master Mix, containing AmpliTaq Gold® DNA polymerase, dNTPs with dUTP, Uracil-DNA Glycosylase, buffer components and a passive internal reference ROX™ dye, as well as 1μl assay mix (made up of primers and probes) and 7 μl nuclease-free water. After preparation, the reaction mix was vortexed for a few seconds. The 2 μl DNA templates were pipetted in the according wells of a 96-well-plate following the plate layout in the StepOneTM software v2.3. By adding the reaction mix, the final reaction volume was brought up to 20 μl per well. Negative (NTC = no template control) and positive controls for each target were prepared in the same manner. The 96-well-plate was then covered with a MicroAmpTM Optical Adhesive Film. SNP Sequence of probes (P), quencher and reporter molecules ABO*O.01.01, ABO*O.01.02 P1: 5’-VIC-GCCTCGTGGTGCCCCTTGG-3’-MGB P2: 5’-FAM-GCCTCGTGGTACCCCTTGG-3’-MGB ABO*B.01 P1: 5’-VIC-CCGTAGAAGCTGGGGTGCAGG-3’-MGB P2: 5’-FAM-CCGTAGAAGCCGGGGTGCAGG-3’-MGB ABO*O.02 P1: 5’-VIC-CCGAAGAACCCCCCCAGGT-3’-MGB P2: 5’-FAM-CCGAAGAACCTCCCCAGGT-3’-MGB ABO*A2.01 P1: 5’-VIC-AGCCGCTCACGGGTTCCGGAC-3’-TAMRA P2: 5’-FAM-AGCCGCTCAC*GGTTCCGGAC-3’-TAMRA Material and Methods 33 Finally, thermal cycling was undertaken in a StepOne PlusTM Real-time PCR System using the following cycling procedure: 60 °C for 30 seconds, initial denaturation at 95 °C for 10 minutes, followed by 40 cycles of 95 °C for 15 seconds, 60 °C annealing and elongation for 1 minute, concluding with 60 °C for 30 seconds. Analysis of the results was again undertaken with the StepOneTM software v2.3 by examining the allelic discrimination plots (Figure 1) and following the evaluation scheme (Table 4). Figure 1: Allelic discrimination plot. Exemplary portrayal of an allelic discrimination plot with three clusters (homozygous allele 1 = C/C, heterozygous allele 1/allele 2 = C/T, homozygous allele 2 = T/T), here illustrated for SNP ABO*O.02 (rs41302905). Material and Methods 34 Table 12: Evaluation scheme of ABO genotyping. ABO genotypes Allele-specific target ABO* A1.01/A1.01 G , G C , C C , C G , G ABO*O.01/O.01 A , A C , C C , C G , G ABO*B.01/B.01 G , G T , T C , C G , G ABO*O.02/O.02 G , G C , C T , T G , G ABO* A2.01/A2.01 G , G C , C C , C * , * ABO*A1.01/O.01 G , A C , C C , C G , G ABO*B.01/O.01 G , A C , T C , C G , G ABO*A2.01/O.01 G, A C , C C , C G , * ABO*A1.01/B.01 G , G C , T C , C G , G ABO*A2.01/B.01 G , G C , T C , C G , * ABO*A1.01/O.02 G , G C , C C , T G , G ABO*O.01/O.02 G , A C , C C , T G , G ABO*B.01/O.02 G , G C , T C , T G , G ABO*A2.01/O.02 G , G C , C C , T G , * ABO*A1.01/A2.01 G , G C , C C , C G , * The evaluation scheme depicts all 15 genotype patterns possibly resulting from the 5 detected alleles. Both maternal and paternal nucleotides are given for each SNP. Variations of the consensus sequence (gray background) defining the different genotypes are displayed with a light gray background. Deletions are again indicated with an asterisk. In the case of discrepancies between serological ABO phenotypes and the determined genotypes, blood group genotyping was confirmed using a PCR with sequence-specific primers (PCR-SSP) or repeated using TaqMan-PCR, but with a 5 μl DNA sample instead of 2 μl and a reduced water fraction (from 7 μl to 4 μl) in each reaction mix. Material and Methods 35 2.7 Principle TaqMan® Real Time PCR The intensity of the fluorescent signal was proportional to the generation of amplicons, thus, the more the signal increases, the farther away from the zero point the clusters will be depicted in the dot plot analysis. Therefore, a qualitative and quantitative detection at the same time was possible. The data was collected in the exponential phase of the PCR run, where the fresh products doubled precisely, assuming 100% reaction efficiency. In the following linear phase, the reaction slowed down due to consumption of reagents and, lastly, ended in the plateau phase, where no more amplicons were produced. The integration of uracil in the amplicons guaranteed the end of the reaction (ThermoFisher Scientific). Traditional PCR takes measurements in this plateau phase (“endpoint detection”), but due to different reaction kinetics in each sample, the same initial concentrations would produce different amounts of amplicons. Furthermore, traditional PCR requires the detection and comparison of the amplified band to known standards in gel electrophoresis, which only gives off 'semi-quantitative' results. TaqMan PCR enabled the dropping of these time-consuming post-PCR steps and provides more accurate quantification (ThermoFisher Scientific). Calculations reveal that the minimal number of probes needed to distinguish all 5 alleles –and thus all 15 resulting genotypes– is at least 8 (MinProb: Analysis and Optimization of Sets of Oligonucleotideprobes by Carlheinz Mueller, University of Ulm, Germany). 2.8 Statistical Analysis of Data The collected results were documented together with other pseudonymized patient information in a Microsoft Excel database. Mr. Johannes Herrmann provided professional support and performed the first statistical analyses with IBM SPSS Statistics Version 25 for Windows. A p-value less than 0.05 was considered significant. The graphical illustration was undertaken with GraphPad Prism 8. Material and Methods 36 Several case definitions were set up to avoid confounding effects of more than one antibody in mothers and to assess possible differences between different antibody specificities causing FNAIT:  Case definition 1: FNAIT proven or possible (including cases of HPA-1bb mothers and HPA-1ab newborns without detectable anti-HPA-1a antibodies), n = 165.  Case definition 2: Cases with any anti-HPA antibody specificity or more than one antibody and newborns positive for the corresponding antigen, n = 158.  Case definition 3: All women who were HPA-1a-negative (HPA-1bb), had an HPA-1a-positive offspring and had anti-HPA-1a antibodies (n=118) or had no anti-HPA-1a antibodies (n=7).  Case definition 4: All women who were HPA-1a-negative (HPA-1bb), had an HPA-1a-positive offspring and had anti-HPA-1a antibodies (n=118).  Case definition 5: All women who were HPA-5b-negative (HPA-5aa), had an HPA-5b-positive offspring and had anti-HPA-5b antibodies (n=24). Regarding influence of fetomaternal ABO incompatibility and ABO blood groups, focus was directed towards case definition 3 and 4, since HPA-1a antigen incompatibility is primarily responsible for FNAIT in Caucasians and ICH as marker for severity only occurred within case definition 4. Therefore, results are reported for case definition 1 to give an overview and for case definition 3 or 4. For case definition 5, most statistics were not able to be computed due to the small number of individuals. Therefore, only descriptive analysis was reported. The ABO phenotype distribution among the immunized mothers and their children affected by FNAIT was compared to 45295 mainly German individuals within the local bone marrow registry by using a Chi-square (χ²) test. The proportion of ABO-incompatible and compatible pregnancies among cases was compared to the data of 522 mother-child-pairs with former suspicion of FNAIT. Fetomaternal ABO incompatibility, maternal and neonatal phenotypes and allele features were correlated to the severity of FNAIT, defined by platelet count nadir, incidence of intracranial hemorrhage and birth weight of the affected children. Since Material and Methods 37 some genotypes occurred only rarely, statistics concerning the influence of the underlying genotypes were not performed. Regarding the incidence of ICH, Fisher’s exact tests and χ² tests (with continuity correction) were used to assess the effects of fetomaternal ABO incompatibility, maternal and neonatal ABO phenotypes and hetero- or homozygosity for ABO alleles. Regarding gene doses, only ABO*A1.01 and ABO*O.01 could be taken into consideration because for ABO*O.02 and ABO*B.01, the number of homozygous individuals was not high enough for valid statistics. Concerning the magnitude of neonatal thrombocytopenia, Mann-Whitney-U tests were implemented for the assessment of fetomaternal ABO incompatibility and hetero- or homozygosity for allele ABO*A1.01 and ABO*O.01. A Kruskal-Wallis test was conducted to evaluate differences among ABO phenotypes. Considering the infant’s birth weight, data was only available for 96 (58%) of the 165 neonates. A t-test was implemented to assess the effects of fetomaternal ABO incompatibility and ABO*O.01 allele dose and a Mann-Whitney-U test was performed for assessment of the ABO*A1.01 allele dose. A Welch-ANOVA was conducted to assess the effect of ABO phenotypes followed by a Games-Howell post-hoc test to see which phenotypes differed significantly. Results 38 3 Results 3.1 Retrospective Case-Control Study 3.1.1 ABO phenotype frequencies do not differ between FNAIT cases and controls The ABO phenotypes of mothers and children affected by FNAIT were deduced from ABO genotypes that are shown in table 13. Table 13: Distribution of ABO blood group genotypes in mother-child-pairs with a history of FNAIT. Genotype ABO*A2.01/*A2.01 was not present in any subject. ABO phenotype ABO genotype Allele 1 Allele 2 Mothers (n = 165) Children (n = 165) O ABO*O.01 ABO*O.01 54 62 ABO*O.01 ABO*O.02 6 4 ABO*O.02 ABO*O.02 1 - A1 ABO*A1.01 ABO*O.01 48 44 ABO*A1.01 ABO*A1.01 8 9 ABO*A1.01 ABO*A2.01 2 6 ABO*A1.01 ABO*O.02 2 2 A2 ABO*A2.01 ABO*O.01 13 7 ABO*A2.01 ABO*O.02 1 1 B ABO*B.01 ABO*O.01 18 15 ABO*B.01 ABO*O.02 3 2 ABO*B.01 ABO*B.01 2 1 AB ABO*A1.01 ABO*B.01 6 10 ABO*A2.01 ABO*B.01 1 2 Results 39 ABO phenotype frequencies of all mothers and neonates were compared to the ABO phenotype frequencies among the control group (Table 14). The differences did not prove to be statistically significant. Table 14: ABO phenotype distribution pictured for case definition 1. Phenotype distribution mothers to controls: χ² test, p = 0.507 Phenotype distribution children to controls: χ² test, p = 0.585 * The subgroups A1 and A2 were aggregated to phenotype A. The same holds true for the subgroup with case definition 3 (FNAIT due to anti-HPA- 1a antibodies). ABO phenotype frequencies did not differ between cases and controls (Table 15). Table 15: ABO phenotype frequencies for case definition 3. Phenotype distribution mothers to controls: χ² test, p = 0.670 Phenotype distribution children to controls: χ² test, p = 0.720 * The subgroups A1 and A2 were aggregated to phenotype A. ABO phenotype Mothers (n = 165) Children (n = 165) Population (n = 45295) O 36% 38% 41% A* 46% 44% 42% B 14% 11% 12% AB 4% 7% 5% ABO phenotype Mothers (n = 124) Children (n = 124) Population (n = 45295) O 35% 39% 41% A* 47% 44% 42% B 13% 10% 12% AB 5% 7% 5% Results 40 3.1.2 Fetomaternal ABO incompatibility does not protect against immunization to fetal platelet antigens by pregnancy ABO-incompatible pregnancies are defined as pregnancies in which the mother has isohemagglutinins directed against fetal ABO blood group antigens (Table 16). Table 16: Compatible and incompatible fetomaternal ABO blood group combinations. The proportion of ABO-incompatible pregnancies in cases was similar to our control population composed of 522 mother-child-pairs with FNAIT excluded (Figure 2): Figure 2: Distribution of ABO-compatible and ABO-incompatible pregnancies Distribution of ABO-compatible and ABO-incompatible pregnancies for case definitions 1 (all FNAIT cases; illustration A) and 3 (FNAIT cases due to anti-HPA-1a antibodies, illustration B) among cases and controls (χ² test, p > 0.05). Maternal blood group Fetal blood group compatible incompatible O O A, B, AB A A, O B, AB B B, O A, AB AB O, A, B, AB - Results 41 3.2 Retrospective Cohort Study 3.2.1 Fetomaternal ABO incompatibility and FNAIT severity 3.2.1.1 Fetomaternal ABO incompatibility is not associated with fetal/neonatal ICH The incidence of ICH in neonates suffering from FNAIT was compared between ABO- compatible and ABO-incompatible pregnancies. No association between ABO incompatibility and ICH was observed (Figure 3). No ICH was observed in case definition 5 (suspected FNAIT, detection of anti-HPA-5b antibodies). Figure 3: Fetomaternal ABO incompatibility and neonatal ICH incidence. Comparison of the incidence of ICH in neonates suffering from FNAIT among ABO- compatible and ABO-incompatible pregnancies for case definitions 1 – 4 (illustrations A – D). A) n = 165, two-sided Fisher’s exact test, p = 1.000 B) n = 157, two-sided Fisher’s exact test, p = 1.000 C) n = 124, two-sided Fisher’s exact test, p = 1.000 D) n = 118, two-sided Fisher’s exact test, p = 1.000 Results 42 3.2.1.2 Fetomaternal ABO incompatibility is not associated with neonatal platelet count nadir There was no significant difference between ABO-compatible or incompatible pregnancies regarding the magnitude of neonatal thrombocytopenia (Figure 4). Case definition 5 not illustrated due to low number of individuals (n = 21, Mann-Whitney-U test, p = 0.052). Figure 4: Fetomaternal ABO incompatibility and neonatal platelet count nadir. Comparison of platelet counts nadir in neonates suffering from FNAIT among ABO- compatible and ABO-incompatible pregnancies for each case definition (illustrations A – D). Dotted line threshold of severe FNAIT (PLT count <50000/μl). Interquartile range and median displayed. A) n = 156, Mann-Whitney-U test, p = 0.113 B) n = 150, Mann-Whitney-U test, p = 0.212 C) n = 120, Mann-Whitney-U test, p = 0.190 D) n = 115, Mann-Whitney-U test, p = 0.401 Results 43 3.2.1.3 Fetomaternal ABO incompatibility is not associated with neonatal birth weight No significant differences between ABO-compatible or incompatible pregnancies regarding the neonatal birth weight were found (Figure 5). Case definition 5 not illustrated due to low number of individuals (n = 12, t-test, p = 0.809). Figure 5: Fetomaternal ABO incompatibility and neonatal birth weight. Comparison of birth weight in neonates suffering from FNAIT among ABO-compatible and ABO-incompatible pregnancies for each case definition (illustrations A – D). Interquartile range and median displayed. A) n = 95, t-test, p = 0.836 B) n = 91, t-test, p = 0.799 C) n = 72, t-test, p = 0.866 D) n = 68, t-test, p = 0.654 Results 44 3.2.1.4 Fetomaternal incompatibility for blood group A1 is not associated with FNAIT severity The A antigen is only barely detectable on platelets from A2 donors (Curtis et al. 2000). Thus, we analyzed the subgroup of mothers with anti-A antibodies and neonates with phenotype A1. According to this definition of ABO incompatibility, FNAIT severity did not differ between ABO-incompatible and compatible pregnancies (Figure 6). Figure 6: Association of platelet-adjusted ABO incompatibility and FNAIT severity. Pictures A – C provide an overview on the influence of platelet-adjusted ABO incompatibility on neonatal ICH, platelet count nadir and birth weight for case definition 3. Interquartile range and median are displayed. A) n = 124, two-sided Fisher’s exact test, p = 1.000 B) n = 120, Mann-Whitney-U test, p = 0.223 C) n = 72, t-test, p = 0.539 Results 45 3.2.2 Maternal ABO phenotypes and FNAIT severity 3.2.2.1 Maternal ABO phenotypes are not associated with neonatal ICH The comparison of maternal ABO phenotypes and the occurrence of neonatal ICH disclosed no significant associations. The cases with maternal blood group AB were excluded due to the low number of individuals (Figure 7). Figure 7: Maternal ABO phenotypes and neonatal ICH incidence. Distribution of maternal ABO phenotypes and the occurrence of ICH in neonates suffering from FNAIT for case definitions 1 - 4 (illustrations A – D). A) n = 158, two-sided Fisher’s exact test, p = 0.530 B) n = 151, two-sided Fisher’s exact test, p = 0.506 C) n = 118, two-sided Fisher’s exact test, p = 0.405 D) n = 113, two-sided Fisher’s exact test, p = 0.355 Results 46 3.2.2.2 Maternal ABO phenotypes are not associated with neonatal platelet count nadir The maternal ABO phenotypes were compared to the platelet count nadir in their neonates. There were no significant associations between the magnitude of neonatal thrombocytopenia and the maternal ABO phenotype (Figure 8). The cases with maternal blood group AB were excluded due to the small number of individuals. Figure 8: Maternal ABO phenotypes and neonatal platelet count nadir. Distribution of maternal ABO phenotypes and platelet count nadir in neonates suffering from FNAIT for case definitions 1 - 4 (illustrations A – D). Dotted line threshold of severe FNAIT (PLT count <50000/μl). Logarithmic box and whisker plot; vertical line inside the box equals the median. A) n = 149, Kruskal-Wallis test, p = 0.856 B) n = 144, Kruskal-Wallis test, p = 0.914 C) n = 114, Kruskal-Wallis test, p = 0.874 D) n = 110, Kruskal-Wallis test, p = 0.851 Results 47 In case definition 5 (FNAIT suspected, anti-HPA-5b antibodies detected), the median platelet count nadir for O, A1, A2 and B was 39000, 64500, 77500, 70000/μl. 1 out of 8 (12.5%) phenotype A mothers and 4 out of 7 (57%) phenotype O mothers gave birth to children with severe instead of moderate thrombocytopenia (OR 0.1071, 95% CI 0.0082 – 1.4071). 3.2.2.3 Maternal phenotype A is associated with neonatal birth weight The conducted Welch-ANOVA depicted a statistically significant difference in mean birth weight levels for the different maternal ABO phenotypes for case definitions 1 and 2. The subsequent Games-Howell post-hoc test revealed a significant difference between maternal phenotypes A1 and A2. On average, children of A2 mothers were 679 g for case def. 1 (95% CI: 140.48 – 1217.00, p = 0.010) and 642 g for case def. 2 (95% CI: 106.05 – 1178.81, p = 0.015) heavier than children from A1 mothers. The children of A2 mothers had a mean birth weight of 3551 g (for case def. 1 and 2) and of A1 mothers 2872 g (case def. 1) and 2908 g (case def. 2) respectively (Figure 9). Case definition 5 not illustrated due to low number of individuals. Results 48 Figure 9: Maternal ABO phenotypes and neonatal birth weight. Distribution of maternal ABO phenotypes and birth weight in neonates suffering from FNAIT for case definitions 1 - 5 (illustrations A – D). Logarithmic box and whisker plot; vertical line inside the box represents the median. A) n = 89, Welch-ANOVA, p = 0.012 B) n = 86, Welch-ANOVA, p = 0.016 C) n = 67, Welch-ANOVA, p = 0.130 D) n = 64, Welch-ANOVA, p = 0.066 3.2.3 Maternal ABO gene dose and FNAIT severity According to a study of Ahlen et al. (Ahlen et al. 2012), the ABO genotype was associated with FNAIT severity. Among mothers with blood group A, the frequency of newborns with severe NAIT was lower in pregnancies where the mother carried only one A allele, and higher where mothers carried two A alleles. To analyze the possible association between maternal ABO genotype and neonatal outcomes, mothers were stratified according to zygosity for A and O alleles (ABO*A1.01 and ABO*O.01 alleles). For alleles ABO*O.02 and ABO*B.01, the number of homozygous mothers was too small for valid statistics. Results 49 3.2.3.1 Maternal ABO gene dose is not associated with neonatal ICH There was no significant difference of the incidence of ICH in neonates suffering from FNAIT born to mothers that were hetero- or homozygous for ABO*A1.01 or ABO*O.01 (Figures 10 and 11). Figure 10: Maternal hetero- or homozygosity for allele ABO*A1.01 and neonatal ICH incidence. Comparison of maternal ABO*A1.01 hetero- or homozygosity and the occurrence of ICH in neonates suffering from FNAIT for case definitions 1 – 4 (illustrations A – D). Heterozygous mothers carry one ABO*A1.01 allele on one haplotype and one ABO*O.01 or ABO*A2.01 allele on the other haplotype. A) n = 66, two-sided Fisher’s exact test, p = 1.000 B) n = 62, two-sided Fisher’s exact test, p = 1.000 C) n = 53, two-sided Fisher’s exact test, p = 1.000 D) n = 51, two-sided Fisher’s exact test, p = 1.000 Results 50 Figure 11: Maternal hetero- or homozygosity for allele ABO*O.01 and neonatal ICH incidence. Comparison of maternal ABO*O.01 hetero- or homozygosity and the occurrence of ICH in neonates suffering from FNAIT for case definitions 1 – 4 (illustrations A – D). Heterozygous mothers carry one ABO*O.01.01 or one ABO*O.01.02 allele, whereas homozygous mothers carry two (ABO*O.01.01 and ABO*O.01.02 not discriminated). A) n = 139, χ² test with continuity correction, p = 0.386 B) n = 133, χ² test with continuity correction, p = 0.396 C) n = 104, two-sided Fisher’s exact test, p = 0.220 D) n = 100, two-sided Fisher’s exact test, p = 0.222 Results 51 3.2.3.2 Maternal ABO gene dose is not associated with neonatal platelet count nadir There was no significant difference in the platelet count nadir in neonates suffering from FNAIT born to mothers that were hetero- or homozygous for ABO*A1.01 or ABO*O.01 (Figure 12 and 13). Figure 12: Maternal hetero- or homozygosity for allele ABO*A1.01 and neonatal platelet count nadir. Comparison of maternal ABO*A1.01 hetero- or homozygosity and platelet count nadir in neonates suffering from FNAIT for case definitions 1 – 4 (illustrations A – D). Heterozygous mothers carry one ABO*A1.01 allele on one haplotype and one ABO*O.01 or ABO*A2.01 allele on the other haplotype. Dotted line threshold of severe FNAIT (PLT count <50000/μl). Interquartile range and median displayed. A) n = 62, Mann-Whitney-U test, p = 0.748 B) n = 59, Mann-Whitney-U test, p = 0.985 C) n = 51, Mann-Whitney-U test, p = 0.892 D) n = 49, Mann-Whitney-U test, p = 0.536 Results 52 Figure 13: Maternal hetero- or homozygosity for allele ABO*O.01 and neonatal platelet count nadir. Comparison of maternal ABO*O1.01 hetero- or homozygosity and platelet count nadir in neonates suffering from FNAIT for case definitions 1 – 4 (illustrations A – D). Heterozygous mothers carry one ABO*O.01.01 or one ABO*O.01.02 allele, whereas homozygous mothers carry two (ABO*O.01.01 and ABO*O.01.02 not discriminated. Dotted line threshold of severe FNAIT (PLT count <50000/μl). Interquartile range and median displayed. A) n = 133, Mann-Whitney-U test, p = 0.895 B) n = 129, Mann-Whitney-U test, p = 0.983 C) n = 101, Mann-Whitney-U test, p = 0.778 D) n = 98, Mann-Whitney-U test, p = 0.897 Results 53 3.2.3.3 Maternal ABO gene dose is not associated with neonatal birth weight There was no significant difference in the birth weight of neonates suffering from FNAIT born to mothers that were hetero- or homozygous for ABO*A1.01 or ABO*O.01 (Figure 14 and 15). Figure 14: Maternal hetero- or homozygosity for allele ABO*A1.01 and neonatal birth weight. Comparison of maternal ABO*A1.01 hetero- or homozygosity and birth weight of neonates suffering from FNAIT for case definitions 1 – 4 (illustrations A – D). Heterozygous mothers carry one ABO*A1.01 allele on one haplotype and one ABO*O.01 or ABO*A2.01 allele on the other haplotype. Interquartile range and median displayed. A) n = 37, Mann-Whitney-U test, p = 0.184 B) n = 35, Mann-Whitney-U test, p = 0.418 C) n = 30, Mann-Whitney-U test, p = 0.083 D) n = 28, Mann-Whitney-U test, p = 0.231 Results 54 Figure 15: Maternal hetero- or homozygosity for allele ABO*O.01 and neonatal birth weight. Comparison of maternal ABO*O1.01 hetero- or homozygosity and birth weight of neonates suffering from FNAIT for case definitions 1 – 4 (illustrations A – D). Heterozygous mothers carry one ABO*O.01.01 or one ABO*O.01.02 allele, whereas homozygous mothers carry two (ABO*O.01.01 and ABO*O.01.02 not discriminated). Interquartile range and median displayed. A) n = 78, t-test, p = 0.469 B) n = 76, t-test, p = 0.490 C) n = 59, t-test, p = 0.946 D) n = 57, t-test, p = 0.961 Results 55 3.2.4 Neonatal phenotype O is associated with ICH We analyzed the possible association between neonatal ABO phenotype and neonatal outcomes. Results showed a significant difference of the incidence of ICH in newborns stratified according to ABO phenotype (Figure 16). Further tests revealed that ICH occurred significantly more often in neonates with phenotype O compared to those with phenotype A (χ² (1) = 5.336, p = 0.021 with continuity correction, φ = 0.264). Newborns grouped according to ABO phenotype did not show significant differences in platelet count nadir and birth weight. N um be r of p at ie nt s (n ) Figure 16: Neonatal ABO phenotypes and ICH or platelet count nadir. Pictures A and B provide an overview on the influence of the neonatal ABO blood group on occurrence of ICH and platelet count nadir for case definition 4. In picture B interquartile range and median displayed. A) n = 110, χ² test, p = 0.035 B) n = 107, Kruskal-Wallis test, p = 0.067 Results 56 3.2.5 Neonatal ABO gene dose and FNAIT Severity 3.2.5.1 Neonatal ABO gene dose is not associated with ICH There was no significant difference of the incidence of ICH in neonates hetero- or homozygous for alleles ABO*A1.01 or ABO*O.01 (Figure 17). Figure 17: Neonatal hetero- or homozygosity and ICH. Incidence of ICH in FNAIT neonates hetero- or homozygous for allele ABO*A1.01 or ABO*O.01 for case definitions 4 (illustrations A and B). A) Heterozygous newborns carry one ABO*A1.01 allele on one haplotype and one ABO*O.01 allele on the other haplotype. B) Heterozygous newborns carry one ABO*O.01.01 and one ABO*O.01.02 allele. A) n = 52, two-sided Fisher’s exact test, p = 1.000 B) n = 96, χ² test with continuity correction, p = 0.262 Results 57 3.2.5.2 Neonatal ABO gene dose is not associated with neonatal platelet count nadir There was no significant difference in platelet counts in neonates hetero- or homozygous for alleles ABO*A1.01 or ABO*O.01 (Figure 18). Figure 18: Neonatal hetero- or homozygosity and platelet count nadir. Platelet count of FNAIT neonates hetero- or homozygous for allele ABO* A1.01 and ABO*O.01 for case definitions 4 (illustrations A and B). A) Heterozygous newborns carry one ABO*A1.01 allele on one haplotype and one O allele on the other haplotype. B) Heterozygous newborns carry one ABO*O.01.01 and one ABO*O.01.02 allele. Dotted line threshold of severe FNAIT (PLT count <50000/μl). Interquartile range and median displayed. A) n = 50, Mann-Whitney-U test, p = 0.870 B) n = 95, Mann-Whitney-U test, p = 0.072 Results 58 3.2.5.3 Neonatal ABO gene dose is not associated with birth weight There was no significant difference in birth weight in neonates hetero- or homozygous for alleles ABO*O.01 (Figure 19). Birth weight data was not available for neonates homozygous for allele ABO*A1.01, therefore statisti