Non‐invasive prenatal testing (NIPT) for fetal Kell, Duffy and Rh blood group antigen prediction in alloimmunised pregnant women: power of droplet digital PCR

Abstract

Pregnant women who present with an antibody to a blood group antigen may require intensive monitoring throughout pregnancy to manage early warning signs of fetal anemia. The most clinically significant antibodies are anti-D (in the Rh blood group system) and anti-K (Kell system), followed by anti-c, anti-E, anti-C (Rh system), and anti-Fya or anti-Fyb (Duffy system) (Royal College of Obstetricians and Gynaecologists, 2014). For anti-D, anti-K, and anti-c, there is a high (>50%) risk for mild to severe haemolytic disease of the fetus or newborn (HDFN) developing should the fetus inherit the target paternal red cell antigen (de Haas et al., 2015). Non-invasive prenatal testing (NIPT) of cell-free fetal (cff) DNA in maternal plasma determines whether the target paternal blood group allele has been inherited, thus predicting whether the fetus is at risk for HDFN. NIPT has proven reliable for fetal RHD genotyping for D-negative pregnant women (Finning et al., 2002; Mackie et al., 2016; Hyland et al., 2017). Conventional real-time PCR allows ready detection of fetal RHD sequences because the RHD gene is deleted on the D-negative haplotype in the majority of cases (Colin et al., 1991). There are therefore no maternal background DNA sequences to compete for detection with fetal RHD sequences. In contrast to RhD, the majority of blood group antigens, including K, c, E, C and Fya/Fyb, arise from single nucleotide variant(s) (SNVs). For this reason, there are limitations for NIPT using real-time PCR assays due to competition from background maternal cell-free (mcf) allelic DNA sequences. Adverse pre-analytic sample conditions which result in maternal white cell lysis further increase background mcfDNA and reduce the fraction of cffDNA, which itself is in low abundance (Lun et al., 2008). This places restrictions on the use of real-time PCR, specifically in the case of K (Finning et al., 2007; Scheffer et al., 2011). To manage the challenge of detecting low copy number fetal alleles in complex samples, we evaluated a droplet digital PCR (ddPCR) approach for a suite of non-RhD NIPT blood group assays. The ddPCR partitions each reaction mixture into tens of thousands of droplets or ‘nano-reactions’ prior to amplification. Each droplet is expected to contain zero or one copy of either maternal or fetal cell-free (cf) DNA, removing the competition effects caused by mcfDNA sequences. The NIPT suite comprised duplex assays for KEL*01.01/KEL*02 (K/k) and FY*A/FY*B (Fya/Fyb), and published singleplex assays for RHCE*c (c), RHCE*E (E) (Finning et al., 2007) and RHCE*C (C) (Legler et al., 2002). For the duplex assays, the maternal allele serves both as a ‘housekeeper’ DNA control marker and allows measurement of the fetal fraction of the cfDNA component. For the singleplex assays, a separate housekeeper DNA control, CCR5, was included, measuring total plasma (maternal and fetal) cfDNA levels (Finning et al., 2002). All assays performed optimally under the same thermal cycling conditions permitting a simplified laboratory protocol, Table 1. The system was evaluated within the clinical environment, under human research ethics approval, for 81 non-RhD alloimmunised cases, of median gestation age (GA) 19+1 weeks (range 10–37), enrolled across 31 geographically diverse antenatal clinics. The majority of cases ‒ 46/81 ‒ exhibited maternal anti-K antibodies. Six cases had alloantibodies to two antigens, making a total of 87 requests for NIPT fetal blood group assessment, Table 2. Maternal blood samples were collected into EDTA and/or cell-free DNA tubes ‒ the latter designed to minimise maternal white cell lysis ‒ and transported to the laboratory for processing: median time to processing was 47·4 h (range 1 h to 6 days). NIPT ddPCR reactions were prepared in five replicates using plasma-derived cfDNA (Hyland et al., 2009). Droplets were prepared, amplified, read and analysed using the Bio-Rad QX200 ddPCR system (Bio-Rad Laboratories Inc, CA, USA). For the 87 fetal blood group investigations, fetal-specific signals were detected for 37 (42·5%; gestation 10–36+1 weeks), predicting inheritance of the antigen targeted by the maternal antibody, as shown in Table 2. The earliest detection was for a 10 weeks GA case with anti-E antibodies, with fetal RHCE*E signals detected at 0·5 copies/µl, Fig 1A. This case was notable because the total cfDNA, measured by CCR5, was elevated at 188·5 copies/µl, due to maternal white cell lysis over five days of transport. A further case with maternal anti-K antibodies at 13 weeks exhibited 0·34 copies/µl, Fig 1B. This sample was collected in Cell-Free DNA BCT® (Streck, Omaha, Nebraska, USA), providing assessment of the biological fetal fraction (16·8%), and not confounded by pre-analytical factors such as delayed transport time increasing background maternal cfDNA. For duplex assays, samples collected in EDTA showed lower fetal fraction when the background mcfDNA was elevated; however, this did not impact on detection of fetal-specific signals. Fifty of the 87 investigations (57·5%) showed no signals for the target blood group, predicting that the fetus had not inherited the paternal allele and was not at risk for HDFN from the associated antibody, as shown in Table 2. Where no fetal blood group-specific signals were detected, further testing using SRY and hypermethylated RASSF1A (Chan et al., 2006; Hyland et al., 2009) was performed to check for the presence of cffDNA markers. Infant cord or buccal swab outcomes were available for 43 of the 81 cases (53·1%): NIPT genotyping predictions were concordant with infant blood group in all cases. Accuracy is 100% (CI 92·0–100·0). The challenge of confirming cord or infant blood group information (other than RhD) has been noted elsewhere (Scheffer et al., 2011; Jenkins et al., 2018). Where available, infant gender predicted by SRY correlated for all cases. For the epigenetic marker, where fetal RASSF1A is hypermethylated and maternal is presumed hypomethylated, we observed higher hypermethylated levels from samples with high maternal cfDNA levels. This requires further investigation and suggests tight pre-analytical conditions are required when testing for this epigenetic marker. We conclude that the partitioning power of droplet digital PCR separates maternal and fetal targets, permitting accurate detection of fetal-specific SNVs for the target blood groups from late first trimester. Early knowledge of the fetal blood group provides guidance for clinical management, permitting reduction in monitoring where it is predicted that the fetus is not at risk for HDFN. The authors are grateful to the mothers who participated in this study, and to all clinicians involved in recruitment and providing infant outcomes. Australian governments fund Australian Red Cross Lifeblood to provide blood, blood products and services to the Australian Community. GG, CH, HO, and RF designed the research study. HO and ES developed and optimised the assays and performed the laboratory work. HO, ES, CH and GG analysed the data. GG and JD provided clinical input and guidance, recruited participants and obtained relevant infant outcome data from the participating sites. CH, HO, GG and ES wrote the paper and RF and JD provided critical revision of the manuscript. The authors have disclosed no conflicts of interest.

Topics & Concepts