NEW METHODS AND APPROACHES Noninvasive fetal genotyping of human platelet antigen-1a using targeted massively parallel sequencing €hner,1 Holger Hackstein,1 Sandra Wienzek-Lischka,1 Annika Krautwurst,1 Vanessa Fro €hner,2 Andreas Bra € uninger,2 Roland Axt-Fliedner,3 Jan Degenhardt,3 Stefan Gattenlo Christina Deisting,3 Sentot Santoso,1 Ulrich J. Sachs,1 and Gregor Bein1
BACKGROUND: Fetal human platelet antigen (HPA) genotyping is required to determine whether the fetus is at risk and whether prenatal interventions to prevent fetal bleeding are required in pregnant women with a history of fetal and neonatal alloimmune thrombocytopenia (FNAIT). Methods for noninvasive genotyping of HPA alleles with the use of maternal plasma cell-free DNA were published recently but do lack internal controls to exclude false-negative results. STUDY DESIGN AND METHODS: Cell-free DNA was isolated from plasma of four pregnant women with a history of FNAIT caused by anti-HPA-1a and controls. A primer panel was designed to target sequences flanking single-nucleotide polymorphisms (SNPs)/exonic regions of ITGB3 (HPA-1), ITGA2B (HPA-3), ITGA2 (HPA-5), CD109 (HPA-15), RHD, RHCE, KEL, DARC, SLC14A1, GYPA, GYPB, and SRY. These regions and eight anonymous SNPs were massively parallel sequenced by semiconductor technology. RESULTS: The mean (6SD) number of reads for targeted SNPs was 5255 (62838). Fetal DNA was detected at a median of 4.5 (range, 2-8) polymorphic loci. The mean fractional fetal DNA concentration in cell-free maternal plasma was 8.36% (range, 4.79%-15.9%). For HPA-1, nonmaternal ITGB3 sequences (c.176T, HPA-1a) were detected in all HPA-1ab fetuses. One HPA-1bb fetus was unequivocally identified, showing the pregnancy was not at risk of FNAIT. CONCLUSION: We have successfully established massively parallel sequencing as a novel reliable method for noninvasive genotyping of fetal HPA-1a alleles. This technique may also allow the safe detection of other fetal blood group polymorphisms frequently involved in FNAIT and hemolytic disease of the newborn.
F
etal and neonatal alloimmune thrombocytopenia (FNAIT) is caused by maternal alloantibodies against paternally inherited human platelet antigens (HPAs) on fetal platelets (PLTs). After transplacental transport from maternal to fetal circulation, maternal HPA alloantibodies can result in fetal thrombocytopenia. Most cases of FNAIT in Caucasians are caused by alloantibodies against the HPA-1a alloantigen on glycoprotein IIb/IIIa.1-3 This epitope is characterized by a single-amino-acid substitution at Position 33 (L33P) of the integrin b 3 chain caused by a T>C mutation at Position c.176 of the ITGB3 gene.4 Newborns with FNAIT present with thrombocytopenia, petechiae, and hematomas.5 The most devastating consequence of FNAIT is intracranial hemorrhage (ICH), which leads to death or severe neurologic disability in almost 90% of cases.6 Based on calculations of a recent meta-analysis of unbiased postnatal studies on neonatal thrombocytopenia, the incidence of severe FNAIT (neonatal PLT count < 50 3 109/L) in the Caucasian population
ABBREVIATIONS: FNAIT 5 fetal and neonatal alloimmune thrombocytopenia; ICH 5 intracranial hemorrhage; SNP(s) 5 single-nucleotide polymorphism(s). From the 1Institute for Clinical Immunology and Transfusion Medicine; the 2Institute of Pathology; and the 3Division of Prenatal Medicine, Department of Obstetrics and Gynecology, Justus-Liebig-University, Giessen, Germany. Address reprint requests to: Gregor Bein, MD, Institute for Clinical Immunology and Transfusion Medicine, JustusLiebig-University Giessen, Langhansstrasse 7, 35392 Giessen, Germany; e-mail: [email protected]. Received for publication July 14, 2014; revision received February 23, 2015; and accepted March 1, 2015. doi:10.1111/trf.13102 C 2015 AABB V
TRANSFUSION 2015;55;1538–1544 1538 TRANSFUSION Volume 55, June 2015
NONINVASIVE FETAL HPA-1A GENOTYPING
is estimated to be 0.15% and the incidence of ICH is estimated to be 10 in 100,000 pregnancies.7 A recent large retrospective study, which included 43 cases of ICH due to FNAIT, assessed the gestational age at the onset of bleeding.6 Contrary to previous assumptions, that study showed that most instances of ICH occur in utero and the majority (54%) of cases occur before 28 weeks of gestation. No cases of ICH occurred intrapartum in the aforementioned study, and only two of the 43 hemorrhagic events occurred after delivery. Thus, prospective screening programs proposed to reduce the risk of ICH8 will require the identification of at-risk pregnancies before the 20th week of gestation, defined as an anti-HPA-1a– immunized woman carrying an HPA-1a–positive fetus.6 Without routine screening of general populations of pregnant women, interventions to reduce ICH due to FNAIT are restricted to the management of subsequent pregnancies. Antenatal preventive measures for at-risk pregnancies include intrauterine PLT transfusion and weekly infusion of intravenous immunoglobulin (IVIG) with or without additional steroids to the mother.9,10 Previously, monitoring fetal PLT count by fetal blood sampling and intrauterine PLT transfusion of compatible PLTs was performed for this purpose; however, they are associated with premature birth and abortion in 1% to 2% of procedures (5%-8% per pregnancy).11,12 Maternal IVIG therapy was introduced in 198813 and has been demonstrated to be efficient in prevention of fetal ICH in the majority of studies. Observational studies also support the efficacy of noninvasive management, that is, IVIG therapy in at-risk pregnancies without the requirement of diagnostic fetal blood sampling.14 Prophylactic interventions in pregnant immunized women or those with a history of FNAIT are only indicated if fetal PLTs carry the cognate HPA antigen. In approximately 30% of cases, the father is heterozygous for the HPA-1a allele (genotype HPA-1ab), and the likelihood that the fetus will inherit the HPA-1b allele is 50%. In this case, there is no risk of FNAIT due to maternal antibody in the current pregnancy, and there is no indication for prophylactic IVIG therapy. The fetal HPA type can be determined by amniocentesis or fetal blood sampling, which carries the risk of miscarriage and/or a boosted humoral immune response in the mother. The discovery of cell-free fetal DNA in maternal plasma in 1997 by Lo and colleagues15 led to the development of noninvasive prenatal testing for fetal RHD status,16,17 fetal aneuploidy,18 and monogenic diseases.19 Noninvasive genotyping of fetal HPA-1a alleles has been described using real-time polymerase chain reaction (PCR) with20 and without21 prior digestion of maternally derived HPA-1b alleles with the MspI restriction enzyme. These methods do not have internal positive controls for the presence of fetal DNA and may lead to false-negative results in cases where the fraction of fetal circulating DNA
in maternal plasma is below the detection limit. In this study, we employed targeted massively parallel sequencing of ITGB3 alleles in combination with additional polymorphic regions coding for common blood group antigens and anonymous single-nucleotide polymorphisms (SNPs) from plasma samples of pregnant antiHPA-1a–positive women and controls.
MATERIALS AND METHODS Subjects Four pregnant women with a history of FNAIT due to anti-HPA-1a were included. Gestational ages at the time of blood sampling were 14, 21, 26, and 31 weeks, respectively. Three husbands were genotyped HPA-1aa. Thus, the pregnant women presumably carried fetuses that inherited one paternal HPA-1a allele (fetal genotype HPA1ab). One husband was genotyped HPA-1ab. In this case, 50% of the fetuses would inherit the genotype HPA-1ab and 50% would inherit the genotype HPA-1bb. All fetal HPA-1 genotypes were confirmed after birth in three cases. In one case (Case 4), the fetal HPA-1 genotype was confirmed at 28 gestational weeks by analyzing DNA from amniotic fluid. Additional blood samples were obtained from four healthy blood donors. HPA-1a–immunized women with HPA-1a–positive fetuses (n 5 3) received IVIG (1 g/kg/week) starting at approximately 20 weeks of gestation. The birth PLT counts of the neonates were 70 3 109, 183 3 109, and 283 3 109/L. Bleeding complications were not observed. This study protocol was approved by the ethics committee of the Medical Faculty of the Justus-LiebigUniversity (Giessen, Germany; file numbers 178/2013 and 05/2000). Written informed consent was obtained from all patients and blood donors.
Genotyping of cases and controls Genotyping for blood group polymorphisms of cases and controls was performed using commercially available PCR assays (BAG Health Care, Lich, Germany; Inno-train, Kronberg, Germany) and in-house TaqMan PCR, respectively. Protocols are available on request.
Cell-free plasma sample preparation Blood samples (2 3 10 mL) were collected into blood collection tubes (Cell-Free DNA BCT, Streck, Omaha, NE). After centrifugation at 1,200 3 g for 20 minutes at room temperature, the plasma was carefully removed and centrifuged a second time at 2,400 3 g for 10 minutes. The plasma samples were stored at 220 C until analysis. Cellfree DNA was extracted from 2 mL of maternal plasma using a vacuum manifold method (QIAamp circulating nucleic acid kit, QIAGEN, Hilden, Germany), lowering the final extraction volume to 60 mL. Volume 55, June 2015 TRANSFUSION 1539
WIENZEK-LISCHKA ET AL.
TABLE 1. Targeted SNPs and exonic regions encoding common blood group phenotypes of PLTs and RBCs System/target
Gene
SNP ID
Phenotype
HPA-1 HPA-3 HPA-5 HPA-15 RHCE Exon 2 RHCE Exon 5 FY MNS MNS KEL JK FY RHD Exon 4 RHD Exon 7
ITGB3 ITGA2B ITGA2 CD109 RHCE RHCE DARC GYPA GYPB KEL SLC14A1 DARC RHD RHD
rs5918 rs5911 rs10471371 rs10455097 rs676785 rs609320 rs12075 rs7682260 rs7683365 rs8176058 rs1058396 rs2814778
HPA-1a/b HPA-3a/b HPA-5a/b HPA-15a/b RhC/c RhE/e Fya/b M/N S/s K/k Jka/b Fy(a–b–) RhD RhD
Targeted massively parallel sequencing Library preparation was performed according to the manufacturer’s instruction (Ion AmpliSeq Library Kit 2.0, Life Technologies/Thermo Fisher Scientific, Carlsbad, CA) using a custom-made primer panel (Ion AmpliSeq Designer). The targeted sequences were chosen to flank SNPs/exonic regions encoding common blood group systems of PLTs and red blood cells (RBCs) that are frequently involved in FNAIT or hemolytic disease of the newborn: ITGB3 (HPA-1), ITGA2B (HPA-3), ITGA2 (HPA5), CD109 (HPA-15), RHD, RHCE, KEL, DARC, SLC14A1, GYPA, and GYPB (Table 1). The SRY sequences and eight anonymous SNPs (Ion AmpliSeq Sample ID-Panel) provided an additional internal positive control for the presence of fetal DNA to avoid false-negative results. The Ion AmpliSeq Sample ID Panel includes eight primer pairs targeting eight SNPs that are unlinked, nonexonic markers that show consistently high minor allele frequency across a diverse group of human populations.22 After library preparation and removal of the target-specific primers, the adaptors and barcode adapters (Ion Xpress barcode adapter kit) used for multiplexing two samples into a single run were ligated. Subsequently, enrichment amplification and further purification were performed. The quality of the DNA libraries was controlled using a bioanalyzer (Agilent Technologies, Palo Alto, CA) instrument. The template was prepared with an automated workflow system (Ion ONE Touch system, Life Technologies/Thermo Fisher Scientific). Massively parallel sequencing was performed on semiconductor chips (Ion Torrent 314, Life Technologies/Thermo Fisher Scientific), using a personal genome machine (Ion Torrent PGM platform, Life Technologies/ Thermo Fisher Scientific; analysis flow, 500, 125 cycles).
Statistical analysis Data were initially processed with the Ion Torrent platform–specific software Torrent Suite (Version 4.0.1, Life 1540 TRANSFUSION Volume 55, June 2015
Technologies/Thermo Fisher Scientific) to generate sequence reads. The Torrent Variant Caller and Integrative Genomics Viewer23 mapping programs were used to align the sequence data to the hg19 human reference genome.
Definitions and a priori defined cutoff values Total reads for a given locus were defined as the number of all sequence reads that crossed the targeted SNP. Unsuccessful amplification of a targeted SNP was defined as total reads less than 1,000 at a particular locus in more than one subject. Unexpected base calls were defined as all base calls at a given SNP position that did not belong to the known biallelic system. Sequences were assumed to be of fetal origin if the pregnant woman was homozygous for a given SNP and the antithetical allele at that locus was detected with a frequency of at least 2% of the total reads. In cases where the targeted sequence was not present in the maternal genomes (SRY, RHD Exon 4, and RHD Exon 7), sequences were assumed to be of fetal origin if the number of sequence reads was more than 100. The fractional fetal DNA concentration in maternal plasma, f, was calculated from total reads as per the equation f ¼
2p ; p1q
where p is the number of sequence reads of the paternally inherited fetal allele and q is the number of sequence reads of the antithetical allele of the mother, which is shared by the maternal and fetal genomes.24
RESULTS Quality of the sequence data Raw data from the sequencing reaction were filtered and polyclonal sequences, low-quality reads, and adapter dimers were excluded. The aligned bases showed a mean of more than Q20 quality in 90.98%. Per individual sequenced, we obtained mean (6SD) of 246,904 (647,562) raw reads, representing 51.55% of the raw data after filtering. Mean (6SD) read length was 104 (68.04) nucleotides. The mean (6SD) total number of assigned amplicon reads was 81,124 (613,339) resulting in 2,897 (6476) reads per amplicon (28 amplicons in total; several of the target sequences were covered by more than one amplicon).
Coverage and total sequence reads We targeted 12 SNPs, coding for biallelic blood group polymorphisms, RHD Exon 4, RHD Exon 7, eight anonymous SNPs, and SRY. Of these 23 targets, 21 were successfully amplified and sequenced from cell-free plasma DNA of four pregnant women and four controls. The targeted sequences of GYPA (rs7682260, associated with blood
NONINVASIVE FETAL HPA-1A GENOTYPING
all subjects). Thus, the a priori defined cutoff value of 2% for paternally inherited fetal alleles was clearly separated from the mean frequency of unexpected base calls plus sixfold SD (Fig. 1).
Positive control for the presence of fetal DNA in cell-free maternal plasma Sequences of fetal origin were detected at a median of 4.5 (range, 2-8) loci including RHD in one RHD-negative woman carrying a confirmed RHD-positive fetus and SRY in one woman carrying a confirmed male fetus (Table 2). The mean frequency of paternally inherited fetal alleles was 4.18% of the total reads (range, 2.40%-7.95%; n 5 15, excluding RHD and SRY). Thus, the mean fractional fetal DNA concentration in cell-free maternal plasma was 8.36% (range, 4.79%-15.9%). The number of sequence reads for RHD Exon 4 was 545 and the number of sequence reads for RHD Exon 7 was 249 in the RHD-negative pregnant woman carrying a confirmed RHD-positive fetus. The number of sequence reads for SRY was 172 in the woman carrying a confirmed male fetus. Thus, the presence of circulating fetal DNA in cell-free maternal plasma above the a priori defined threshold was confirmed in all cases.
Fig. 1. Sequence reads (% of total reads) for unexpected base calls (18 targets in eight plasma samples; n 5 288 unexpected calls) and paternally inherited alleles (n 5 15 in plasma samples of four pregnant women). Dashed line 5 predefined cutoff for the presence of paternally inherited alleles.
group M/N polymorphism) and DARC (rs2814778, associated with Duffy null phenotype) were excluded from further analysis in seven of eight cases as the number of sequence reads in these cases was below the a priori defined threshold. We obtained a mean (6SD) of 5,255 (62,838) sequence reads (range, 355-13,902; n 5 144 [18 targets in eight subjects]) per target for the remaining targets (RHD and SRY were excluded from this calculation as they were not present in the genome of all subjects). We obtained a mean (6SD) of 6,009 (62,415) sequence reads for RHD Exon 4 and 5,183 (63,013) sequence reads for RHD Exon 7 in seven RHD-positive subjects. A total of 1,667 and 2,989 sequence reads for SRY were detected in cell-free plasma DNA of two male control subjects (data not shown).
Noninvasive genotyping of the fetal HPA-1a antigen HPA-1bb phenotyped mothers are homozygous for C at Position c.176 of the ITGB3 gene. The mean (6SD) number of total reads for sequences at Position c.176 of the ITGB3 gene was 4,176 (61,619; n 5 8). We sequenced cellfree plasma DNA of control subjects representing the genotypes HPA-1aa, HPA-1ab, and HPA-1bb. The number of false-positive base calls for the antithetical allele in homozygous subjects was 3, 11, and 2 (Table 3). The paternally inherited fetal T allele was detected in all three pregnant women with a history of FNAIT due to antiHPA-1a carrying HPA-1ab fetuses. The mean frequency of fetal T alleles was 3.68% (range, 2.40%-4.99%) of total reads (Table 2). Thus, the fetal HPA-1a antigen could be detected in all cases by our approach. One HPA-1bb fetus was unequivocally identified at 14 gestational weeks in a woman with a history of FNAIT due to anti-HPA-1a where the husband was heterozygous HPA-1ab. In this case, there was one false-positive base call (of 2,930 reads) for the T allele at Position c.176 of the ITGB3 gene (Table 2).
DISCUSSION Unexpected base calls Unexpected base calls were detected with a mean (6SD) frequency of 0.13% (60.21%) of total reads (sixfold SD 6 1.27%; range, 0%-1.20%, n 5 288; RHD and SRY were excluded from this calculation as they were not present in
Here we report a novel method for the noninvasive genotyping of fetal HPA-1a alleles using targeted massively parallel sequencing that reliably prevents false-negative results due to the inclusion of additional polymorphic loci. Fetal DNA constitutes approximately 10% of cell-free Volume 55, June 2015 TRANSFUSION 1541
WIENZEK-LISCHKA ET AL.
TABLE 2. Detection of fetal alleles in cell-free maternal plasma* Case 1. HPA-1ab (ch) HPA-1bb (mo) 3 HPA-1aa (fa) 2. HPA-1ab (ch) HPA-1bb (mo) 3 HPA-1aa (fa)
3. HPA-1ab (ch) HPA-1bb (mo) 3 HPA-1aa (fa)
4. HPA-1bb (ch) HPA-1bb (mo) 3 HPA-1ab (fa)
System
A reads
C reads
G reads
T reads
% fetal allele
HPA-1 SNP #7 HPA-1 HPA-5 HPA-15 RHCE E/e FY SNP #3 RHD Exon 4 RHD Exon 7 HPA-1 MNS S/s SNP #2 SNP #3 SNP #5 SRY HPA-1 HPA-15 MNS M/N
0 412† 0 4,027 4,118 1 6,182 264† 0 0 0 281† 1 408† 306† 0 0 49† 1,881
4,369 0 7,133 0 109† 10,439 0 0 0 0 3,698 2 7 1 1 0 2,913 1,137 0
22 8,012 37 193 0 280† 221† 9,940 0 249† 30 4,940 308† 4,722 4,998 172† 16 0 72†
166† 3 176† 0 0 2 2 0 545† 0 196† 14 7,165 0 0 0 1** 0 1
166/4,557 (3.64%‡) 412/8,436 (4.88%) 176/7,346 (2.40%‡) 193/4,220 (4.57%) 109/4,227 (2.58%) 280/10,722 (2.61%) 221/6,405 (3.45%) 264/10,205 (2.59%) 545/545 (100.00%) 249/249 (100.00%) 196/3,924 (4.99%‡) 281/5,237 (5.37%) 308/7,481 (4.12%) 408/5,132 (7.95%) 306/5,306 (5.77%) 172/172 (100.00%) 1/2,930 (0.034%**) 49/1,186 (4.13%) 72/1,954 (3.68%)
* The genotypes of mothers, fathers, and fetuses are given: mo 5 mother; fa 5 father; ch 5 child. ** False positive base call of the HPA-1a allele. † Alleles of fetal origin. ‡ Percentage of fetal HPA-1a allele.
DNA in the plasma of pregnant women.25 As the fetus inherits one allele from its mother, the paternal allele constitutes approximately 5% of the circulating DNA molecules. Thus, it is technically challenging to detect the paternally inherited allele against the background of 95% maternal alleles. Scheffer and colleagues20 proposed pre-PCR digestion of the maternal HPA-1b allele employing the MspI restriction enzyme to avoid nonspecific signals due to the high background of maternal sequences. Le Toriellec et al.21 demonstrated noninvasive fetal HPA-1a genotyping by real-time allele-specific PCR and melting curve analysis of PCR amplicons. However, these methods do not have an internal control for the presence of fetal DNA, which may lead to falsenegative results. In these methods, the HPA-1a–negative phenotype of the fetus is inferred merely by the absence of a positive signal. However, a negative result could also arise because of the absence of fetal DNA or a fraction of fetal DNA being below the detection limit of the assay. False-negative results may lead to omission of prophylactic measures against ICH, leading to potentially severe consequences. Thus, a negative result can be confirmed only if fetal DNA is detected with an internal control.
External positive control systems for fetal DNA include the detection of the male SRY gene, genetic variations that differ between the mother and the fetus, and epigenetic variation, for example, hypermethylation of the placental RASSF1A gene.26 Rieneck and coworkers27 used massively parallel sequencing of the KEL1 allele of the Kell system from cell-free fetal DNA in maternal plasma. However, internal controls for the presence of fetal DNA were not included in this study as well. Targeted massively parallel sequencing of DNA fragments spanning SNPs, either anonymous or coding for common blood groups, and of SRY, in addition to the allele in question, provides an internal control for the presence of fetal DNA. In biallelic systems with an allele frequency of 50%, there is a probability of 0.25 that the system is informative for the presence of fetal DNA; that is, the mother is homozygous for one or the other allele and the fetus is heterozygous for the antithetical allele. Thus, the inclusion of 20 biallelic systems increases the likelihood of at least one of the polymorphic systems being informative in those cases where the SRY gene of a male fetus is not detected to a p value of 0.99. Lo and coworkers24 showed that complete fetal and maternal genomes are represented in maternal plasma at
TABLE 3. Targeted massively parallel sequencing of cell-free plasma DNA from nonpregnant controls: reads at Position c.176 of ITGB3 Controls
A reads
C reads (HPA-1b)
G reads
T reads (HPA-1a)
% T reads
HPA-1aa HPA-1aa HPA-1ab HPA-1bb
0 1 1 0
3 11 2,331 4,023
5 4 16 18
1,661 4,640 1,935 2
1,661/1,669 (99.52%) 4,640/4,656 (99.66%) 1,935/4,283 (45.18%) 2/4,043 (0.05%)
1542 TRANSFUSION Volume 55, June 2015
NONINVASIVE FETAL HPA-1A GENOTYPING
a constant relative proportion. Circulating fetal DNA fragments show a fragmentation pattern of approximately 140 bp.24 Consequently, the targeted massively parallel sequencing of cell-free maternal plasma DNA detects fetal and maternal alleles without bias, if the amplicons used for the library preparation are short enough to cover the fragmented DNA. Unbiased amplification of different alleles was demonstrated experimentally in this study by dilution of two different cell-free plasma samples in a ratio of 1:10. The allelic read counts were in the expected range (data not shown). A limitation of noninvasive prenatal diagnosis by analysis of fetal DNA in maternal plasma is the low concentration of fetal DNA in early pregancy.28 However, the massively parallel sequencing of polymorphic targets enables the counting of paternally inherited alleles and thus the precise determination of the fetal DNA concentration. In cases where the fetal DNA concentration is below predefined thresholds; for example, below 4% fractional fetal DNA concentration, a follow-up sample should be analyzed. In conclusion, we described the noninvasive detection of fetal HPA-1a alleles by targeted massively parallel sequencing in HPA-1bb mothers who developed antiHPA-1a antibodies. This approach offers a reliable and safe method to determine whether the fetus is at risk for FNAIT and whether interventions to prevent ICH, that is, weekly infusion of IVIG, are indicated. Several groups have proposed population-based screening programs for HPA-1a alloimmunization to prevent ICH in the first pregnancy.29 Noninvasive detection of fetal HPA-1a alleles in anti-HPA-1a-positive mothers would allow detection of at-risk pregnancies without paternal zygosity testing. We did not validate noninvasive fetal blood grouping of common blood groups that were included as polymorphic controls for the presence of fetal DNA in this study, as this is the subject of another ongoing study. However, the preliminary results of this study suggest that targeted massively parallel sequencing may be the method of choice for noninvasive risk determination for FNAIT due to HPA antibodies or hemolytic disease of the newborn due to common blood group antibodies.
2. Ghevaert C, Campbell K, Walton J, et al. Management and outcome of 200 cases of fetomaternal alloimmune thrombocytopenia. Transfusion 2007;47:901-10. 3. Davoren A, Curtis B, Aster R, et al. Human platelet antigenspecific alloantibodies implicated in 1162 cases of neonatal alloimmune thrombocytopenia. Transfusion 2004;44:1220-5. 4. Newman PJ, Derbes RS, Aster RH. The human platelet alloantigens, PlA1 and PlA2, are associated with a leucine33/proline33 amino acid polymorphism in membrane glycoprotein IIIa, and are distinguishable by DNA typing. J Clin Invest 1989;83:1778-81. 5. Mueller-Eckhardt C, Kiefel V, Grubert A, et al. 348 cases of suspected neonatal alloimmune thrombocytopenia. Lancet 1989;1:363-6. 6. Tiller H, Kamphuis MM, Flodmark O, et al. Fetal intracranial haemorrhages caused by fetal and neonatal alloimmune thrombocytopenia: an observational cohort study of 43 cases from an international multicentre registry. BMJ Open 2013;3: e002490. 7. Kamphuis MM, Paridaans NP, Porcelijn L, et al. Incidence and consequences of neonatal alloimmune thrombocytopenia: a systematic review. Pediatrics 2014;133:1-3. 8. Kjeldsen-Kragh J, Killie MK, Tomter G, et al. A screening and intervention program aimed to reduce mortality and serious morbidity associated with severe neonatal alloimmune thrombocytopenia. Blood 2007;110:833-9. 9. Bussel JB. Diagnosis and management of the fetus and neonate with alloimmune thrombocytopenia. J Thromb Haemost 2009;7:253-7. 10. Kaplan C. Neonatal alloimmune thrombocytopenia: a 50year story. Immunohematology 2007;23:9-13. 11. Birchall J, Murphy M, Kaplan C, et al. European collaborative study of the antenatal management of feto-maternal alloimmune thrombocytopenia. Br J Haematol 2003;122:275-88. 12. Rayment R, Brunskill SJ, Soothill PW, et al. Antenatal interventions for fetomaternal alloimmune thrombocytopenia. Cochrane Database Syst Rev 2011;(5):CD004226. 13. Bussel JB, Berkowitz RL, McFarland JG, et al. Antenatal treatment of neonatal alloimmune thrombocytopenia. N Engl J Med 1988;319:1374-8. 14. van den Akker ES, Oepkes D, Lopriore E, et al. Noninvasive antenatal management of fetal and neonatal alloimmune thrombocytopenia: safe and effective. BJOG 2007;114:469-
CONFLICT OF INTEREST The authors have disclosed no conflicts of interest.
73. 15. Lo YM, Corbetta N, Chamberlain PF, et al. Presence of fetal DNA in maternal plasma and serum. Lancet 1997;350:485-7. 16. Lo YM, Hjelm NM, Fidler C, et al. Prenatal diagnosis of fetal
REFERENCES 1. Kroll H, Yates J, Santoso S. Immunization against a lowfrequency human platelet alloantigen in fetal alloimmune thrombocytopenia is not a single event: characterization by the combined use of reference DNA and novel allele-specific cell lines expressing recombinant antigens. Transfusion 2005;45:353-8.
RhD status by molecular analysis of maternal plasma. N Engl J Med 1998;339:1734-8. 17. Faas BH, Beuling EA, Christiaens GC, et al. Detection of fetal RHD-specific sequences in maternal plasma. Lancet 1998; 352:1196. 18. Bianchi DW, Parker RL, Wentworth J, et al. DNA sequencing versus standard prenatal aneuploidy screening. N Engl J Med 2014;370:799-808. Volume 55, June 2015 TRANSFUSION 1543
WIENZEK-LISCHKA ET AL.
19. Lo YM. Non-invasive prenatal testing using massively parallel sequencing of maternal plasma DNA: from molecular karyotyping to fetal whole-genome sequencing. Reprod Biomed Online 2013;27:593-8.
25. Tsui NB, Lo YM. Recent advances in the analysis of fetal nucleic acids in maternal plasma. Curr Opin Hematol 2012;19:462-8. 26. Chan KC, Allen Ding C, Gerovassili A, et al. Hypermethylated RASSF1A in maternal plasma: a universal fetal DNA marker
20. Scheffer PG, Ait Soussan A, Verhagen O, et al. Noninvasive
that improves the reliability of noninvasive prenatal diagno-
fetal genotyping of human platelet antigen-1a. BJOG 2011; 118:1392-5.
sis. Clin Chem 2006;52:2211-8. 27. Rieneck K, Bak M, Jønson L, et al. Next-generation sequenc-
21. Le Toriellec E, Chenet C, Kaplan C. Safe fetal platelet geno-
ing: proof of concept for antenatal prediction of the fetal Kell
typing: new developments. Transfusion 2013;53:1755-62. 22. Pakstis AJ, Speed WC, Fang R, et al. SNPs for a universal indi-
blood group phenotype from cell-free fetal DNA in maternal plasma. Transfusion 2013;53(11Suppl2):2892-8.
vidual identification panel. Hum Genet 2010;127:315-24. ttir H, Winckler W, et al. Integrative 23. Robinson JT, Thorvaldsdo genomics viewer. Nat Biotechnol 2011;29:24-6. 24. Lo YM, Chan KC, Sun H, et al. Maternal plasma DNA sequencing reveals the genome-wide genetic and mutational profile of the fetus. Sci Transl Med 2010;2:61ra91.
1544 TRANSFUSION Volume 55, June 2015
28. Lo YM, Tein MS, Lau TK, et al. Quantitative analysis of fetal DNA in maternal plasma and serum: implications for noninvasive prenatal diagnosis. Am J Hum Genet 1998;62:768-75. 29. Kamphuis MM, Paridaans N, Porcelijn L, et al. Screening in pregnancy for fetal or neonatal alloimmune thrombocytopenia: systematic review. BJOG 2010;117:1335-43.
BACKGROUND: Fetal human platelet antigen (HPA) genotyping is required to determine whether the fetus is at risk and whether prenatal interventions to prevent fetal bleeding are required in pregnant women with a history of fetal and neonatal alloimmune thrombocytopenia (FNAIT). Methods for noninvasive genotyping of HPA alleles with the use of maternal plasma cell-free DNA were published recently but do lack internal controls to exclude false-negative results. STUDY DESIGN AND METHODS: Cell-free DNA was isolated from plasma of four pregnant women with a history of FNAIT caused by anti-HPA-1a and controls. A primer panel was designed to target sequences flanking single-nucleotide polymorphisms (SNPs)/exonic regions of ITGB3 (HPA-1), ITGA2B (HPA-3), ITGA2 (HPA-5), CD109 (HPA-15), RHD, RHCE, KEL, DARC, SLC14A1, GYPA, GYPB, and SRY. These regions and eight anonymous SNPs were massively parallel sequenced by semiconductor technology. RESULTS: The mean (6SD) number of reads for targeted SNPs was 5255 (62838). Fetal DNA was detected at a median of 4.5 (range, 2-8) polymorphic loci. The mean fractional fetal DNA concentration in cell-free maternal plasma was 8.36% (range, 4.79%-15.9%). For HPA-1, nonmaternal ITGB3 sequences (c.176T, HPA-1a) were detected in all HPA-1ab fetuses. One HPA-1bb fetus was unequivocally identified, showing the pregnancy was not at risk of FNAIT. CONCLUSION: We have successfully established massively parallel sequencing as a novel reliable method for noninvasive genotyping of fetal HPA-1a alleles. This technique may also allow the safe detection of other fetal blood group polymorphisms frequently involved in FNAIT and hemolytic disease of the newborn.
F
etal and neonatal alloimmune thrombocytopenia (FNAIT) is caused by maternal alloantibodies against paternally inherited human platelet antigens (HPAs) on fetal platelets (PLTs). After transplacental transport from maternal to fetal circulation, maternal HPA alloantibodies can result in fetal thrombocytopenia. Most cases of FNAIT in Caucasians are caused by alloantibodies against the HPA-1a alloantigen on glycoprotein IIb/IIIa.1-3 This epitope is characterized by a single-amino-acid substitution at Position 33 (L33P) of the integrin b 3 chain caused by a T>C mutation at Position c.176 of the ITGB3 gene.4 Newborns with FNAIT present with thrombocytopenia, petechiae, and hematomas.5 The most devastating consequence of FNAIT is intracranial hemorrhage (ICH), which leads to death or severe neurologic disability in almost 90% of cases.6 Based on calculations of a recent meta-analysis of unbiased postnatal studies on neonatal thrombocytopenia, the incidence of severe FNAIT (neonatal PLT count < 50 3 109/L) in the Caucasian population
ABBREVIATIONS: FNAIT 5 fetal and neonatal alloimmune thrombocytopenia; ICH 5 intracranial hemorrhage; SNP(s) 5 single-nucleotide polymorphism(s). From the 1Institute for Clinical Immunology and Transfusion Medicine; the 2Institute of Pathology; and the 3Division of Prenatal Medicine, Department of Obstetrics and Gynecology, Justus-Liebig-University, Giessen, Germany. Address reprint requests to: Gregor Bein, MD, Institute for Clinical Immunology and Transfusion Medicine, JustusLiebig-University Giessen, Langhansstrasse 7, 35392 Giessen, Germany; e-mail: [email protected]. Received for publication July 14, 2014; revision received February 23, 2015; and accepted March 1, 2015. doi:10.1111/trf.13102 C 2015 AABB V
TRANSFUSION 2015;55;1538–1544 1538 TRANSFUSION Volume 55, June 2015
NONINVASIVE FETAL HPA-1A GENOTYPING
is estimated to be 0.15% and the incidence of ICH is estimated to be 10 in 100,000 pregnancies.7 A recent large retrospective study, which included 43 cases of ICH due to FNAIT, assessed the gestational age at the onset of bleeding.6 Contrary to previous assumptions, that study showed that most instances of ICH occur in utero and the majority (54%) of cases occur before 28 weeks of gestation. No cases of ICH occurred intrapartum in the aforementioned study, and only two of the 43 hemorrhagic events occurred after delivery. Thus, prospective screening programs proposed to reduce the risk of ICH8 will require the identification of at-risk pregnancies before the 20th week of gestation, defined as an anti-HPA-1a– immunized woman carrying an HPA-1a–positive fetus.6 Without routine screening of general populations of pregnant women, interventions to reduce ICH due to FNAIT are restricted to the management of subsequent pregnancies. Antenatal preventive measures for at-risk pregnancies include intrauterine PLT transfusion and weekly infusion of intravenous immunoglobulin (IVIG) with or without additional steroids to the mother.9,10 Previously, monitoring fetal PLT count by fetal blood sampling and intrauterine PLT transfusion of compatible PLTs was performed for this purpose; however, they are associated with premature birth and abortion in 1% to 2% of procedures (5%-8% per pregnancy).11,12 Maternal IVIG therapy was introduced in 198813 and has been demonstrated to be efficient in prevention of fetal ICH in the majority of studies. Observational studies also support the efficacy of noninvasive management, that is, IVIG therapy in at-risk pregnancies without the requirement of diagnostic fetal blood sampling.14 Prophylactic interventions in pregnant immunized women or those with a history of FNAIT are only indicated if fetal PLTs carry the cognate HPA antigen. In approximately 30% of cases, the father is heterozygous for the HPA-1a allele (genotype HPA-1ab), and the likelihood that the fetus will inherit the HPA-1b allele is 50%. In this case, there is no risk of FNAIT due to maternal antibody in the current pregnancy, and there is no indication for prophylactic IVIG therapy. The fetal HPA type can be determined by amniocentesis or fetal blood sampling, which carries the risk of miscarriage and/or a boosted humoral immune response in the mother. The discovery of cell-free fetal DNA in maternal plasma in 1997 by Lo and colleagues15 led to the development of noninvasive prenatal testing for fetal RHD status,16,17 fetal aneuploidy,18 and monogenic diseases.19 Noninvasive genotyping of fetal HPA-1a alleles has been described using real-time polymerase chain reaction (PCR) with20 and without21 prior digestion of maternally derived HPA-1b alleles with the MspI restriction enzyme. These methods do not have internal positive controls for the presence of fetal DNA and may lead to false-negative results in cases where the fraction of fetal circulating DNA
in maternal plasma is below the detection limit. In this study, we employed targeted massively parallel sequencing of ITGB3 alleles in combination with additional polymorphic regions coding for common blood group antigens and anonymous single-nucleotide polymorphisms (SNPs) from plasma samples of pregnant antiHPA-1a–positive women and controls.
MATERIALS AND METHODS Subjects Four pregnant women with a history of FNAIT due to anti-HPA-1a were included. Gestational ages at the time of blood sampling were 14, 21, 26, and 31 weeks, respectively. Three husbands were genotyped HPA-1aa. Thus, the pregnant women presumably carried fetuses that inherited one paternal HPA-1a allele (fetal genotype HPA1ab). One husband was genotyped HPA-1ab. In this case, 50% of the fetuses would inherit the genotype HPA-1ab and 50% would inherit the genotype HPA-1bb. All fetal HPA-1 genotypes were confirmed after birth in three cases. In one case (Case 4), the fetal HPA-1 genotype was confirmed at 28 gestational weeks by analyzing DNA from amniotic fluid. Additional blood samples were obtained from four healthy blood donors. HPA-1a–immunized women with HPA-1a–positive fetuses (n 5 3) received IVIG (1 g/kg/week) starting at approximately 20 weeks of gestation. The birth PLT counts of the neonates were 70 3 109, 183 3 109, and 283 3 109/L. Bleeding complications were not observed. This study protocol was approved by the ethics committee of the Medical Faculty of the Justus-LiebigUniversity (Giessen, Germany; file numbers 178/2013 and 05/2000). Written informed consent was obtained from all patients and blood donors.
Genotyping of cases and controls Genotyping for blood group polymorphisms of cases and controls was performed using commercially available PCR assays (BAG Health Care, Lich, Germany; Inno-train, Kronberg, Germany) and in-house TaqMan PCR, respectively. Protocols are available on request.
Cell-free plasma sample preparation Blood samples (2 3 10 mL) were collected into blood collection tubes (Cell-Free DNA BCT, Streck, Omaha, NE). After centrifugation at 1,200 3 g for 20 minutes at room temperature, the plasma was carefully removed and centrifuged a second time at 2,400 3 g for 10 minutes. The plasma samples were stored at 220 C until analysis. Cellfree DNA was extracted from 2 mL of maternal plasma using a vacuum manifold method (QIAamp circulating nucleic acid kit, QIAGEN, Hilden, Germany), lowering the final extraction volume to 60 mL. Volume 55, June 2015 TRANSFUSION 1539
WIENZEK-LISCHKA ET AL.
TABLE 1. Targeted SNPs and exonic regions encoding common blood group phenotypes of PLTs and RBCs System/target
Gene
SNP ID
Phenotype
HPA-1 HPA-3 HPA-5 HPA-15 RHCE Exon 2 RHCE Exon 5 FY MNS MNS KEL JK FY RHD Exon 4 RHD Exon 7
ITGB3 ITGA2B ITGA2 CD109 RHCE RHCE DARC GYPA GYPB KEL SLC14A1 DARC RHD RHD
rs5918 rs5911 rs10471371 rs10455097 rs676785 rs609320 rs12075 rs7682260 rs7683365 rs8176058 rs1058396 rs2814778
HPA-1a/b HPA-3a/b HPA-5a/b HPA-15a/b RhC/c RhE/e Fya/b M/N S/s K/k Jka/b Fy(a–b–) RhD RhD
Targeted massively parallel sequencing Library preparation was performed according to the manufacturer’s instruction (Ion AmpliSeq Library Kit 2.0, Life Technologies/Thermo Fisher Scientific, Carlsbad, CA) using a custom-made primer panel (Ion AmpliSeq Designer). The targeted sequences were chosen to flank SNPs/exonic regions encoding common blood group systems of PLTs and red blood cells (RBCs) that are frequently involved in FNAIT or hemolytic disease of the newborn: ITGB3 (HPA-1), ITGA2B (HPA-3), ITGA2 (HPA5), CD109 (HPA-15), RHD, RHCE, KEL, DARC, SLC14A1, GYPA, and GYPB (Table 1). The SRY sequences and eight anonymous SNPs (Ion AmpliSeq Sample ID-Panel) provided an additional internal positive control for the presence of fetal DNA to avoid false-negative results. The Ion AmpliSeq Sample ID Panel includes eight primer pairs targeting eight SNPs that are unlinked, nonexonic markers that show consistently high minor allele frequency across a diverse group of human populations.22 After library preparation and removal of the target-specific primers, the adaptors and barcode adapters (Ion Xpress barcode adapter kit) used for multiplexing two samples into a single run were ligated. Subsequently, enrichment amplification and further purification were performed. The quality of the DNA libraries was controlled using a bioanalyzer (Agilent Technologies, Palo Alto, CA) instrument. The template was prepared with an automated workflow system (Ion ONE Touch system, Life Technologies/Thermo Fisher Scientific). Massively parallel sequencing was performed on semiconductor chips (Ion Torrent 314, Life Technologies/Thermo Fisher Scientific), using a personal genome machine (Ion Torrent PGM platform, Life Technologies/ Thermo Fisher Scientific; analysis flow, 500, 125 cycles).
Statistical analysis Data were initially processed with the Ion Torrent platform–specific software Torrent Suite (Version 4.0.1, Life 1540 TRANSFUSION Volume 55, June 2015
Technologies/Thermo Fisher Scientific) to generate sequence reads. The Torrent Variant Caller and Integrative Genomics Viewer23 mapping programs were used to align the sequence data to the hg19 human reference genome.
Definitions and a priori defined cutoff values Total reads for a given locus were defined as the number of all sequence reads that crossed the targeted SNP. Unsuccessful amplification of a targeted SNP was defined as total reads less than 1,000 at a particular locus in more than one subject. Unexpected base calls were defined as all base calls at a given SNP position that did not belong to the known biallelic system. Sequences were assumed to be of fetal origin if the pregnant woman was homozygous for a given SNP and the antithetical allele at that locus was detected with a frequency of at least 2% of the total reads. In cases where the targeted sequence was not present in the maternal genomes (SRY, RHD Exon 4, and RHD Exon 7), sequences were assumed to be of fetal origin if the number of sequence reads was more than 100. The fractional fetal DNA concentration in maternal plasma, f, was calculated from total reads as per the equation f ¼
2p ; p1q
where p is the number of sequence reads of the paternally inherited fetal allele and q is the number of sequence reads of the antithetical allele of the mother, which is shared by the maternal and fetal genomes.24
RESULTS Quality of the sequence data Raw data from the sequencing reaction were filtered and polyclonal sequences, low-quality reads, and adapter dimers were excluded. The aligned bases showed a mean of more than Q20 quality in 90.98%. Per individual sequenced, we obtained mean (6SD) of 246,904 (647,562) raw reads, representing 51.55% of the raw data after filtering. Mean (6SD) read length was 104 (68.04) nucleotides. The mean (6SD) total number of assigned amplicon reads was 81,124 (613,339) resulting in 2,897 (6476) reads per amplicon (28 amplicons in total; several of the target sequences were covered by more than one amplicon).
Coverage and total sequence reads We targeted 12 SNPs, coding for biallelic blood group polymorphisms, RHD Exon 4, RHD Exon 7, eight anonymous SNPs, and SRY. Of these 23 targets, 21 were successfully amplified and sequenced from cell-free plasma DNA of four pregnant women and four controls. The targeted sequences of GYPA (rs7682260, associated with blood
NONINVASIVE FETAL HPA-1A GENOTYPING
all subjects). Thus, the a priori defined cutoff value of 2% for paternally inherited fetal alleles was clearly separated from the mean frequency of unexpected base calls plus sixfold SD (Fig. 1).
Positive control for the presence of fetal DNA in cell-free maternal plasma Sequences of fetal origin were detected at a median of 4.5 (range, 2-8) loci including RHD in one RHD-negative woman carrying a confirmed RHD-positive fetus and SRY in one woman carrying a confirmed male fetus (Table 2). The mean frequency of paternally inherited fetal alleles was 4.18% of the total reads (range, 2.40%-7.95%; n 5 15, excluding RHD and SRY). Thus, the mean fractional fetal DNA concentration in cell-free maternal plasma was 8.36% (range, 4.79%-15.9%). The number of sequence reads for RHD Exon 4 was 545 and the number of sequence reads for RHD Exon 7 was 249 in the RHD-negative pregnant woman carrying a confirmed RHD-positive fetus. The number of sequence reads for SRY was 172 in the woman carrying a confirmed male fetus. Thus, the presence of circulating fetal DNA in cell-free maternal plasma above the a priori defined threshold was confirmed in all cases.
Fig. 1. Sequence reads (% of total reads) for unexpected base calls (18 targets in eight plasma samples; n 5 288 unexpected calls) and paternally inherited alleles (n 5 15 in plasma samples of four pregnant women). Dashed line 5 predefined cutoff for the presence of paternally inherited alleles.
group M/N polymorphism) and DARC (rs2814778, associated with Duffy null phenotype) were excluded from further analysis in seven of eight cases as the number of sequence reads in these cases was below the a priori defined threshold. We obtained a mean (6SD) of 5,255 (62,838) sequence reads (range, 355-13,902; n 5 144 [18 targets in eight subjects]) per target for the remaining targets (RHD and SRY were excluded from this calculation as they were not present in the genome of all subjects). We obtained a mean (6SD) of 6,009 (62,415) sequence reads for RHD Exon 4 and 5,183 (63,013) sequence reads for RHD Exon 7 in seven RHD-positive subjects. A total of 1,667 and 2,989 sequence reads for SRY were detected in cell-free plasma DNA of two male control subjects (data not shown).
Noninvasive genotyping of the fetal HPA-1a antigen HPA-1bb phenotyped mothers are homozygous for C at Position c.176 of the ITGB3 gene. The mean (6SD) number of total reads for sequences at Position c.176 of the ITGB3 gene was 4,176 (61,619; n 5 8). We sequenced cellfree plasma DNA of control subjects representing the genotypes HPA-1aa, HPA-1ab, and HPA-1bb. The number of false-positive base calls for the antithetical allele in homozygous subjects was 3, 11, and 2 (Table 3). The paternally inherited fetal T allele was detected in all three pregnant women with a history of FNAIT due to antiHPA-1a carrying HPA-1ab fetuses. The mean frequency of fetal T alleles was 3.68% (range, 2.40%-4.99%) of total reads (Table 2). Thus, the fetal HPA-1a antigen could be detected in all cases by our approach. One HPA-1bb fetus was unequivocally identified at 14 gestational weeks in a woman with a history of FNAIT due to anti-HPA-1a where the husband was heterozygous HPA-1ab. In this case, there was one false-positive base call (of 2,930 reads) for the T allele at Position c.176 of the ITGB3 gene (Table 2).
DISCUSSION Unexpected base calls Unexpected base calls were detected with a mean (6SD) frequency of 0.13% (60.21%) of total reads (sixfold SD 6 1.27%; range, 0%-1.20%, n 5 288; RHD and SRY were excluded from this calculation as they were not present in
Here we report a novel method for the noninvasive genotyping of fetal HPA-1a alleles using targeted massively parallel sequencing that reliably prevents false-negative results due to the inclusion of additional polymorphic loci. Fetal DNA constitutes approximately 10% of cell-free Volume 55, June 2015 TRANSFUSION 1541
WIENZEK-LISCHKA ET AL.
TABLE 2. Detection of fetal alleles in cell-free maternal plasma* Case 1. HPA-1ab (ch) HPA-1bb (mo) 3 HPA-1aa (fa) 2. HPA-1ab (ch) HPA-1bb (mo) 3 HPA-1aa (fa)
3. HPA-1ab (ch) HPA-1bb (mo) 3 HPA-1aa (fa)
4. HPA-1bb (ch) HPA-1bb (mo) 3 HPA-1ab (fa)
System
A reads
C reads
G reads
T reads
% fetal allele
HPA-1 SNP #7 HPA-1 HPA-5 HPA-15 RHCE E/e FY SNP #3 RHD Exon 4 RHD Exon 7 HPA-1 MNS S/s SNP #2 SNP #3 SNP #5 SRY HPA-1 HPA-15 MNS M/N
0 412† 0 4,027 4,118 1 6,182 264† 0 0 0 281† 1 408† 306† 0 0 49† 1,881
4,369 0 7,133 0 109† 10,439 0 0 0 0 3,698 2 7 1 1 0 2,913 1,137 0
22 8,012 37 193 0 280† 221† 9,940 0 249† 30 4,940 308† 4,722 4,998 172† 16 0 72†
166† 3 176† 0 0 2 2 0 545† 0 196† 14 7,165 0 0 0 1** 0 1
166/4,557 (3.64%‡) 412/8,436 (4.88%) 176/7,346 (2.40%‡) 193/4,220 (4.57%) 109/4,227 (2.58%) 280/10,722 (2.61%) 221/6,405 (3.45%) 264/10,205 (2.59%) 545/545 (100.00%) 249/249 (100.00%) 196/3,924 (4.99%‡) 281/5,237 (5.37%) 308/7,481 (4.12%) 408/5,132 (7.95%) 306/5,306 (5.77%) 172/172 (100.00%) 1/2,930 (0.034%**) 49/1,186 (4.13%) 72/1,954 (3.68%)
* The genotypes of mothers, fathers, and fetuses are given: mo 5 mother; fa 5 father; ch 5 child. ** False positive base call of the HPA-1a allele. † Alleles of fetal origin. ‡ Percentage of fetal HPA-1a allele.
DNA in the plasma of pregnant women.25 As the fetus inherits one allele from its mother, the paternal allele constitutes approximately 5% of the circulating DNA molecules. Thus, it is technically challenging to detect the paternally inherited allele against the background of 95% maternal alleles. Scheffer and colleagues20 proposed pre-PCR digestion of the maternal HPA-1b allele employing the MspI restriction enzyme to avoid nonspecific signals due to the high background of maternal sequences. Le Toriellec et al.21 demonstrated noninvasive fetal HPA-1a genotyping by real-time allele-specific PCR and melting curve analysis of PCR amplicons. However, these methods do not have an internal control for the presence of fetal DNA, which may lead to falsenegative results. In these methods, the HPA-1a–negative phenotype of the fetus is inferred merely by the absence of a positive signal. However, a negative result could also arise because of the absence of fetal DNA or a fraction of fetal DNA being below the detection limit of the assay. False-negative results may lead to omission of prophylactic measures against ICH, leading to potentially severe consequences. Thus, a negative result can be confirmed only if fetal DNA is detected with an internal control.
External positive control systems for fetal DNA include the detection of the male SRY gene, genetic variations that differ between the mother and the fetus, and epigenetic variation, for example, hypermethylation of the placental RASSF1A gene.26 Rieneck and coworkers27 used massively parallel sequencing of the KEL1 allele of the Kell system from cell-free fetal DNA in maternal plasma. However, internal controls for the presence of fetal DNA were not included in this study as well. Targeted massively parallel sequencing of DNA fragments spanning SNPs, either anonymous or coding for common blood groups, and of SRY, in addition to the allele in question, provides an internal control for the presence of fetal DNA. In biallelic systems with an allele frequency of 50%, there is a probability of 0.25 that the system is informative for the presence of fetal DNA; that is, the mother is homozygous for one or the other allele and the fetus is heterozygous for the antithetical allele. Thus, the inclusion of 20 biallelic systems increases the likelihood of at least one of the polymorphic systems being informative in those cases where the SRY gene of a male fetus is not detected to a p value of 0.99. Lo and coworkers24 showed that complete fetal and maternal genomes are represented in maternal plasma at
TABLE 3. Targeted massively parallel sequencing of cell-free plasma DNA from nonpregnant controls: reads at Position c.176 of ITGB3 Controls
A reads
C reads (HPA-1b)
G reads
T reads (HPA-1a)
% T reads
HPA-1aa HPA-1aa HPA-1ab HPA-1bb
0 1 1 0
3 11 2,331 4,023
5 4 16 18
1,661 4,640 1,935 2
1,661/1,669 (99.52%) 4,640/4,656 (99.66%) 1,935/4,283 (45.18%) 2/4,043 (0.05%)
1542 TRANSFUSION Volume 55, June 2015
NONINVASIVE FETAL HPA-1A GENOTYPING
a constant relative proportion. Circulating fetal DNA fragments show a fragmentation pattern of approximately 140 bp.24 Consequently, the targeted massively parallel sequencing of cell-free maternal plasma DNA detects fetal and maternal alleles without bias, if the amplicons used for the library preparation are short enough to cover the fragmented DNA. Unbiased amplification of different alleles was demonstrated experimentally in this study by dilution of two different cell-free plasma samples in a ratio of 1:10. The allelic read counts were in the expected range (data not shown). A limitation of noninvasive prenatal diagnosis by analysis of fetal DNA in maternal plasma is the low concentration of fetal DNA in early pregancy.28 However, the massively parallel sequencing of polymorphic targets enables the counting of paternally inherited alleles and thus the precise determination of the fetal DNA concentration. In cases where the fetal DNA concentration is below predefined thresholds; for example, below 4% fractional fetal DNA concentration, a follow-up sample should be analyzed. In conclusion, we described the noninvasive detection of fetal HPA-1a alleles by targeted massively parallel sequencing in HPA-1bb mothers who developed antiHPA-1a antibodies. This approach offers a reliable and safe method to determine whether the fetus is at risk for FNAIT and whether interventions to prevent ICH, that is, weekly infusion of IVIG, are indicated. Several groups have proposed population-based screening programs for HPA-1a alloimmunization to prevent ICH in the first pregnancy.29 Noninvasive detection of fetal HPA-1a alleles in anti-HPA-1a-positive mothers would allow detection of at-risk pregnancies without paternal zygosity testing. We did not validate noninvasive fetal blood grouping of common blood groups that were included as polymorphic controls for the presence of fetal DNA in this study, as this is the subject of another ongoing study. However, the preliminary results of this study suggest that targeted massively parallel sequencing may be the method of choice for noninvasive risk determination for FNAIT due to HPA antibodies or hemolytic disease of the newborn due to common blood group antibodies.
2. Ghevaert C, Campbell K, Walton J, et al. Management and outcome of 200 cases of fetomaternal alloimmune thrombocytopenia. Transfusion 2007;47:901-10. 3. Davoren A, Curtis B, Aster R, et al. Human platelet antigenspecific alloantibodies implicated in 1162 cases of neonatal alloimmune thrombocytopenia. Transfusion 2004;44:1220-5. 4. Newman PJ, Derbes RS, Aster RH. The human platelet alloantigens, PlA1 and PlA2, are associated with a leucine33/proline33 amino acid polymorphism in membrane glycoprotein IIIa, and are distinguishable by DNA typing. J Clin Invest 1989;83:1778-81. 5. Mueller-Eckhardt C, Kiefel V, Grubert A, et al. 348 cases of suspected neonatal alloimmune thrombocytopenia. Lancet 1989;1:363-6. 6. Tiller H, Kamphuis MM, Flodmark O, et al. Fetal intracranial haemorrhages caused by fetal and neonatal alloimmune thrombocytopenia: an observational cohort study of 43 cases from an international multicentre registry. BMJ Open 2013;3: e002490. 7. Kamphuis MM, Paridaans NP, Porcelijn L, et al. Incidence and consequences of neonatal alloimmune thrombocytopenia: a systematic review. Pediatrics 2014;133:1-3. 8. Kjeldsen-Kragh J, Killie MK, Tomter G, et al. A screening and intervention program aimed to reduce mortality and serious morbidity associated with severe neonatal alloimmune thrombocytopenia. Blood 2007;110:833-9. 9. Bussel JB. Diagnosis and management of the fetus and neonate with alloimmune thrombocytopenia. J Thromb Haemost 2009;7:253-7. 10. Kaplan C. Neonatal alloimmune thrombocytopenia: a 50year story. Immunohematology 2007;23:9-13. 11. Birchall J, Murphy M, Kaplan C, et al. European collaborative study of the antenatal management of feto-maternal alloimmune thrombocytopenia. Br J Haematol 2003;122:275-88. 12. Rayment R, Brunskill SJ, Soothill PW, et al. Antenatal interventions for fetomaternal alloimmune thrombocytopenia. Cochrane Database Syst Rev 2011;(5):CD004226. 13. Bussel JB, Berkowitz RL, McFarland JG, et al. Antenatal treatment of neonatal alloimmune thrombocytopenia. N Engl J Med 1988;319:1374-8. 14. van den Akker ES, Oepkes D, Lopriore E, et al. Noninvasive antenatal management of fetal and neonatal alloimmune thrombocytopenia: safe and effective. BJOG 2007;114:469-
CONFLICT OF INTEREST The authors have disclosed no conflicts of interest.
73. 15. Lo YM, Corbetta N, Chamberlain PF, et al. Presence of fetal DNA in maternal plasma and serum. Lancet 1997;350:485-7. 16. Lo YM, Hjelm NM, Fidler C, et al. Prenatal diagnosis of fetal
REFERENCES 1. Kroll H, Yates J, Santoso S. Immunization against a lowfrequency human platelet alloantigen in fetal alloimmune thrombocytopenia is not a single event: characterization by the combined use of reference DNA and novel allele-specific cell lines expressing recombinant antigens. Transfusion 2005;45:353-8.
RhD status by molecular analysis of maternal plasma. N Engl J Med 1998;339:1734-8. 17. Faas BH, Beuling EA, Christiaens GC, et al. Detection of fetal RHD-specific sequences in maternal plasma. Lancet 1998; 352:1196. 18. Bianchi DW, Parker RL, Wentworth J, et al. DNA sequencing versus standard prenatal aneuploidy screening. N Engl J Med 2014;370:799-808. Volume 55, June 2015 TRANSFUSION 1543
WIENZEK-LISCHKA ET AL.
19. Lo YM. Non-invasive prenatal testing using massively parallel sequencing of maternal plasma DNA: from molecular karyotyping to fetal whole-genome sequencing. Reprod Biomed Online 2013;27:593-8.
25. Tsui NB, Lo YM. Recent advances in the analysis of fetal nucleic acids in maternal plasma. Curr Opin Hematol 2012;19:462-8. 26. Chan KC, Allen Ding C, Gerovassili A, et al. Hypermethylated RASSF1A in maternal plasma: a universal fetal DNA marker
20. Scheffer PG, Ait Soussan A, Verhagen O, et al. Noninvasive
that improves the reliability of noninvasive prenatal diagno-
fetal genotyping of human platelet antigen-1a. BJOG 2011; 118:1392-5.
sis. Clin Chem 2006;52:2211-8. 27. Rieneck K, Bak M, Jønson L, et al. Next-generation sequenc-
21. Le Toriellec E, Chenet C, Kaplan C. Safe fetal platelet geno-
ing: proof of concept for antenatal prediction of the fetal Kell
typing: new developments. Transfusion 2013;53:1755-62. 22. Pakstis AJ, Speed WC, Fang R, et al. SNPs for a universal indi-
blood group phenotype from cell-free fetal DNA in maternal plasma. Transfusion 2013;53(11Suppl2):2892-8.
vidual identification panel. Hum Genet 2010;127:315-24. ttir H, Winckler W, et al. Integrative 23. Robinson JT, Thorvaldsdo genomics viewer. Nat Biotechnol 2011;29:24-6. 24. Lo YM, Chan KC, Sun H, et al. Maternal plasma DNA sequencing reveals the genome-wide genetic and mutational profile of the fetus. Sci Transl Med 2010;2:61ra91.
1544 TRANSFUSION Volume 55, June 2015
28. Lo YM, Tein MS, Lau TK, et al. Quantitative analysis of fetal DNA in maternal plasma and serum: implications for noninvasive prenatal diagnosis. Am J Hum Genet 1998;62:768-75. 29. Kamphuis MM, Paridaans N, Porcelijn L, et al. Screening in pregnancy for fetal or neonatal alloimmune thrombocytopenia: systematic review. BJOG 2010;117:1335-43.
