Two Reliable Methodical Approaches for Non-Invasive RHD Genotyping of a Fetus from Maternal Plasma

Noninvasive fetal RHD genotyping is an important tool for predicting RhD incompatibility between a pregnant woman and a fetus. This study aimed to assess a methodological approach other than the commonly used one for noninvasive fetal RHD genotyping on a representative set of RhD-negative pregnant women. The methodology must be accurate, reliable, and broadly available for implementation into routine clinical practice. A total of 337 RhD-negative pregnant women from the Czech Republic region were tested in this study. The fetal RHD genotype was assessed using two methods: real-time PCR and endpoint quantitative fluorescent (QF) PCR. We used exon-7-specific primers from the RHD gene, along with internal controls. Plasma samples were analyzed and measured in four/two parallel reactions to determine the accuracy of the RHD genotyping. The RHD genotype was verified using DNA analysis from a newborn buccal swab. Both methods showed an excellent ability to predict the RHD genotype. Real-time PCR achieved its greatest accuracy of 98.6% (97.1% sensitivity and 100% specificity (95% CI)) if all four PCRs were positive/negative. The QF PCR method also achieved its greatest accuracy of 99.4% (100% sensitivity and 98.6% specificity (95% CI)) if all the measurements were positive/negative. Both real-time PCR and QF PCR were reliable methods for precisely assessing the fetal RHD allele from the plasma of RhD-negative pregnant women.


Introduction
Alloimmunization of RhD-negative pregnant women by the highly immunogenic fetal D antigen leads to a hemolytic transfusion reaction and hemolytic disease of the fetus and newborn. For this reason, anti-D immune prophylaxis is preventively administered to RhD-negative pregnant women [1][2][3]. By conducting a molecular analysis of cell-free fetal DNA (cffDNA) circulating in the peripheral blood of a pregnant woman, it is possible to determine the fetal RHD genotype at an early stage of pregnancy.
Peripheral blood samples of RhD-negative pregnant women, confirmation samples of newborn buccal swabs, and control plasma samples for calibration and optimization were collected in collaboration with the Department of Medical Genetics, the Department of Obstetrics and Gynecology, and the Department of Transfusion Medicine of the University Hospital Olomouc. All of the women enrolled in the study signed an informed consent form approved by the Ethics Committee of the University Hospital Olomouc (approval code: 150/10; approved on 20 September 2010).
The total number of analyzed samples was 337 triplets when it was possible to examine a pregnant woman, a fetus, and a newborn as a control of the fetal RHD genotype together. The mothers had already been phenotypically (serologically) and genotypically tested so newborn phenotyping was not performed in this study. The real-time PCR and QF PCR analyses were performed in randomly selected plasma samples taken from RhD-negative pregnant women >18 years old with a singleton pregnancy. The characterizations of the tested RhD-negative pregnant women are in Table 1. Determination of the fetal RHD genotype was evaluated in parallel using two methods: TaqMan real-time PCR and endpoint QF PCR.  15 15.5 1 RhD-negative phenotype, n-total sum of analyzed samples, BMI-body mass index.

Sample Preparation and DNA Isolation
All 337 blood samples from RhD-negative pregnant women were collected into two parallel 9 mL tubes ("A" and "B") containing ethylenediaminetetraacetic acid (EDTA). Anticoagulated blood was placed on ice immediately after collection and was processed within 4 h after sampling. Plasma was separated from the cellular fraction of blood using double centrifugation (2700× g for 10 min and 3500× g for 20 min). The plasma samples were frozen until further processing at −28 • C. Plasma-cell-free (cf) DNA was isolated in each of the two parallel tubes ("A" and "B"). The DNA isolation of 1 mL of plasma was performed using the QIAamp DNA Mini Kit (Qiagen, Venlo, The Netherlands). The incubation step for the isolation took place at 56 • C, with an elution volume of 65 µL. The isolation of maternal DNA from peripheral blood leukocytes was performed with a Qiacube automated isolator (Qiagen) using the QIAamp DNA mini kit (Qiagen), according to the manufacturer's instructions. The isolation of the control DNA from newborn buccal swabs was performed using the QIAamp DNA Mini kit (Qiagen), according to the manufacturer's instructions.

Determination of Fetal RHD Genotype by TaqMan Real-Time PCR Using Internal Amplification Control
Plasma DNA ("A" and "B") were analyzed from each sample in two parallel reactions (each cffDNA sample was measured four times). The maternal RHD genotype was determined from the DNA sample from peripheral blood leukocytes and the fetal genotype was confirmed from a neonatal buccal swab. Specific primers for exon 7 and for the internal control were used to amplify and quantify the multiplex using the TaqMan real-time PCR system, where their sequences were: 5 -GGGTGTTGTAACCGAGTGCTG-3 , forward and 5 -CCGGCTCCGACGGTATC-3 , reverse. The sequence and labeling of the TaqMan probe for RHD exon 7 was 5 -FAM-CCCACAGCTCCATCATGGGCTACAA-BHQ1-3 . The primer and probe sequences from the β-globin gene for the internal total plasma DNA amplification control were GTGCACCTGACTCCTG AGGAGA, forward, CCTTGATACCAACCTGCCCAG, reverse, and 5 -JOE-AAGGTGAACGTGGATGAAGTTGGTGG -BHQ1-3 , TaqMan probe.
DNA samples isolated from RHD-positive blood and RHD-positive fetal DNA isolated from plasma were used as amplification controls. PCR water was used to control the contamination of the PCR premixes. Amplification for all samples was performed in a real-time PCR system Mx3005P (Stratagene, Santa Clara, CA, USA) under the following conditions: 95 • C 15 min, (95 • C 15 s, 60 • C 60 s) 55×. The software Prox-Mx3005P v3.00 Build 311 (Stratagene) was used for the evaluation. The threshold cycle (Ct) values (the number of cycles at which the fluorescence exceeds the threshold value) were determined for each group of samples. Before reading, the measured fluorescence was logarithmized.
Within four parallel measurements, four criteria for evaluating positivity and negativity were set: • Criterion 1-Positive if all four plasmas were positive, negative if all four plasmas were negative. • Criterion 2-Positive if three or more plasmas were positive, negative if two or more plasmas were negative. • Criterion 3-Positive if three or more plasmas are positive, negative if three or more plasmas were negative. • Criterion 4-Positive if two or more plasmas were positive, negative if three or more plasmas were negative.

Determination of Fetal RHD Genotype with Endpoint QF PCR Using Internal Amplification Control with Capillary Electrophoresis
Plasma DNA ("A" and "B") was analyzed in two parallel reactions. Specific primers for exon 7 of the RHD gene and for AMELX/Y sequences were used to amplify and quantify the multiplex using endpoint QF PCR. The sequence and labeling of the primers for RHD exon 7 were 5 -HEX-CCCTGGGCTCTGTAAAG-3 , forward, and 5 -CCGGCTCCGACGGTATC-3 , reverse. The primers for the AMELX/Y gonosomal sequences were used as internal amplification controls, where their sequences were: 5 -6FAM-CCCTGGGCTCTGTAAAG-3 , forward, and 5 -ATCAGAGCTTAAACTGGGAAGCT-3 , reverse. The PCR reactions for DNA isolated from the plasma of the RhD-negative pregnant women were amplified in a final 20 µL volume.
DNA samples isolated from the RHD-positive blood and RHD-positive fetal DNA isolated from plasma were used as the amplification controls. PCR water was used to control for the contamination of PCR premixes. PCR conditions were 95 • C 10 min, (94 • C 30 s, 59 • C 60 s, 72 • C 60 s) 35-40×, 72 • C 10 min, 60 • C 30 min. PCR amplification was performed for all samples in a Thermocycler C 1000 (Bio-Rad, Hercules, CA, USA).
The fluorescence intensity and size of the PCR products were determined using capillary electrophoresis on an ABI PRISM 3130 Genetic Analyzer (Applied Biosystems, Waltham, MA, USA) using polymer POP-4 (Applied Biosystems) and a 47 cm capillary with a diameter of 50 µm. One microliter of PCR product was mixed with 8.5 µL Hi-Di formamide (Applied Biosystems) and the GeneScan 500 TAMRA Size Standard (Applied Biosystems). Each sample of plasma cffDNA was assessed in two capillary electrophoresis conditions. The first injection was for 4 s at 4 kV with electrophoresis for 18 min at 15 kV, and the second injection was for 5 s at 10 kV with electrophoresis for 18 min at 15 kV. The capillary conditions for peripheral blood samples and buccal swab samples were injected for 5 s at 10 kV with electrophoresis for 18 min at 15 kV. The data were analyzed using 310 GeneScan 3.1.2 software (Applied Biosystems). The RFU (relative fluorescence unit) parameter expressed as a peak height was used for quantitative analyses. QF PCR was evaluated with two replicates as we had a limited amount of total plasma DNA.
Within two parallel measurements, two criteria for evaluating positivity and negativity were set: • Criterion 1-Positive if two plasmas were positive, negative if two plasmas were negative. • Criterion 2-Positive if one plasma was positive, negative if two plasmas were negative.

Data Collection
Data from plasma DNA were analyzed, evaluated, and collected independently from the newborn DNA ones.

Study Limitation
The data from the evaluation of pregnancy and clinical characteristics associated with plasma or cffDNA concentrations were not available. We did not carry out the molecular analysis of weak RHD variants.

Statistical Evaluation
Standard descriptive statistics were applied in the analysis: arithmetic mean with standard deviation (SD) and median with a 5th to 95th percentile range were adopted for the continuous variables and absolute and relative frequencies were adopted for categorical variables.
Fisher's exact test was applied for the computation of the statistical significance of relations between categorical variables (association analyses).
The predictive power of variables was quantified on the basis of a standard set of statistics: AUC and its statistical significance derived from ROC analysis, specificity, sensitivity, NPV, PPV, and overall accuracy; ROC analysis was applied for the identification of optimal cut-offs of continuous variables as predictors.

TaqMan Real-Time PCR
Determination of the fetal RHD genotype was possible in a total of 333 out of the 337 triplets using real-time PCR. The analysis failed in two plasma samples. The signal intensity of the PCR products from the RHD gene and from the internal control β-globin gene was not detectable in these samples. The analysis was not possible, probably due to the low concentration of cffDNA in the plasma, maybe due to DNA degradation. Determination of the fetal RHD genotype was not possible in two samples due to repeated RHD-positive findings in the maternal DNA. Examples of the output of RHD-positive and RHD-negative fetuses using the probe for RHD exon 7 from real-time PCR are shown in Figure 1.  The predictive power of the diagnostic test and ROC analysis with cut-off Ct values are available in Tables 2 and 3. It was not possible to evaluate all of the data for any of the four assessed criteria (Tables 2 and 3). In the real-time PCR method, the greatest power to predict the RHD genotype was with criterion 1 ( Table 2). ROC analysis and predictive values using the indentified cut-offs showed the best accuracy if the mean of all four parameters with a cut-off (Ct RHD-Ct globin) of 16.314 was calculated (Table 3).

End-Point QF PCR
Determination of the fetal RHD genotype was possible in a total of 335 out of 337 triplets using endpoint QF PCR. Determination of the fetal RHD genotype was not possible in two samples due to repeated RHD-positive findings in the maternal DNA. Examples of the electrophoretograms of RHD-positive and RHD-negative fetuses using the probes for RHD exon 7 and for AMELX/Y are shown in Figure 2. dpoint QF PCR. Determination of the fetal RHD genotype was not possible in two samples due t eated RHD-positive findings in the maternal DNA. Examples of the electrophoretograms of RHD sitive and RHD-negative fetuses using the probes for RHD exon 7 and for AMELX/Y are shown i ure 2.
The predictive power of the diagnostic test and ROC analysis with a cut-off of Ct values ar own in Tables 4 and 5. It was not possible to evaluate all of the data for any of the two assesse teria (Tables 4 and 5). In endpoint QF PCR, the greatest power to predict RHD genotype was wit terion 1 ( Table 4). The ROC analysis and predictive value using the identified cut-offs showed th st accuracy if the mean of all two parameters with a cut-off (RFU RHD/AMELX) of 0.023 wa lculated (Table 5).  The predictive power of the diagnostic test and ROC analysis with a cut-off of Ct values are shown in Tables 4 and 5. It was not possible to evaluate all of the data for any of the two assessed criteria (Tables 4 and 5). In endpoint QF PCR, the greatest power to predict RHD genotype was with criterion 1 ( Table 4). The ROC analysis and predictive value using the identified cut-offs showed the best accuracy if the mean of all two parameters with a cut-off (RFU RHD/AMELX) of 0.023 was calculated (Table 5).

Discussion
Our study focused on comparing the commonly used TaqMan real-time PCR methodology with the less commonly used endpoint QF PCR with capillary electrophoresis. The sensitivity threshold using the TaqMan real-time PCR system was determined in our previous study. RHD calibration was performed using a dilution series of an artificial mixture of RHD genotypes. TaqMan real-time PCR was able to capture a 0.22% admixture of an RHD-positive heterozygote in an RHD-negative homozygote. The sensitivity of QF PCR was assessed using simulations of artificial mosaics of gonosomal sequences in the studies. Even a 0.5% mosaic was reliably captured using this method [23,24]. Both methodologies are therefore suitable for detecting fetal DNA fractions due to the highly sensitive detection limit (below 1%). However, the sensitivity of the methodologies may also be affected by other factors, such as the overall concentration and quality of the fragmented cfDNA. The amount of total cfDNA was also affected by the time lag between the collection of peripheral blood into EDTA tubes and the separation of plasma from plasma cellular components. The amount of total cfDNA increased over time due to the lysis of peripheral leukocytes [25]. The lysis of maternal leukocytes leads to a relative decrease in the fetal fraction and an increase in the maternal fraction, which disrupts the subsequent quantitative analyses.
If it is not possible to process the blood within 6 h after the collection, it is advised to use specialized cfDNA preservation tubes (e.g., Cell-Free DNA BCT (Streck, La Vista, NE, USA), CellSave (Cell-Search, Huntingdon Valley, PA, USA), and PAXgene Blood DNA (Qiagen)), which prolong the stability of peripheral leukocytes and allow for storage of whole blood at room temperature for several days [31][32][33]. Barrett et al. reported that the concentration of short cffDNA fragments did not change at room temperature for 72 h when collected in Cell-Free DNA BCT tubes (Streck). In EDTA tubes, the number of long (maternal) DNA fragments increased after 72 h [34]. In our study, blood samples were separated within 4 h after collection, which is consistent with numerous studies that confirm that EDTA collection tubes provide cfDNA stability for at least 6 h, and in some studies up to 24 h [33,[35][36][37][38].
Plasma separation conditions can also affect the concentration and quality of cfDNA. The optimized centrifugation method allows for better separation of not only cellular components of blood, but also longer cfDNA fragments of maternal origin from shorter fetal DNA fragments. Commonly used methods are one-step centrifugation for 10 min at 2000-3000× g [39][40][41] or two-step centrifugation, where the first step is 1600-3000× g for 10 min and the second centrifugation is 10,000-16,000× g for 10 min [42][43][44][45][46]. The yield of cffDNA also depends on the chosen methodology for isolating cfDNA. The QIAamp Circulating Nucleic Acid Kit (Qiagen) is currently the most commonly used method for cffDNA isolation.
Jain et al. compared the use of kits for isolating viral DNA and circulating DNA using the QIAamp DSP Virus Kit and the QIAamp Circulating Nucleic Acid Kit. A statistically higher yield of cffDNA was achieved using the QIAamp Circulating Nucleic Acid Kit [47].
In our work, we used the QIAamp DNA Mini Kit (Qiagen, USA), the yield of which was sufficient due to the relatively easy evaluation of the detection of the presence of the RHD gene and due to the sensitivity of our methods. Our methodology for RHD genotyping using TaqMan real-time PCR and endpoint QF PCR was based on the detection of a unique exon 7 sequence that does not contain the highly homologous RHCE gene [7]. Functional (antigenic) variants of the Rh system are caused by insertions/deletions, single nucleotide polymorphisms, or gene conversion between RHD and RHCE genes [48,49].
In Whites, the RhD-negative genotype is in most cases the result of a homozygous deletion of the entire RHD gene [50,51], which was confirmed by our study. The deletion of the RHD gene results from uneven recombination between two homologous Rh boxes of paired chromosomes 1, leading to the formation of a hybrid Rh box [52].
It is reported that 12% to 18% of the White population has an RhD-negative genotype [53]. The expression of the D allele in the D/d heterozygote is due to the dominant type of inheritance [51]. In the Black population, RHD is 5% negative [53]. There are three common variants in the Black population that do not produce the D antigen. The most common cause of the D negative phenotype (66%) is the presence of an RHD pseudogene (RHDΨ), which contains a 37 bp nucleotide duplication, resulting in the formation of a premature stop codon in exon 6 [54]. The second variant is the RHD-CE-D hybrid gene, which contains nucleotide sequences from the RHCE gene, does not produce antigen D, and antigen C is formed abnormally. RHD-CE-D hybrid genes probably originated due to the pairing between RHD and RHCE genes on the same chromosome during meiosis and subsequent gene conversion [55,56]. The third variant is a complete deletion of the RHD gene, as in the White population [54].
The detection of the RHD gene deletion using exon 7 was shown to be sufficient for the Czech population in this study; for example, in contrast to the Black population, only two (0.6%) discrepancies were found between the mother genotype and the phenotype in our group. We did not test other variants associated with RhD negatives. Our test is diagnostically suitable and usable for such populations in which the vast majority of RhD negativity is caused by the deletion of the entire gene, which is very accurately detectable by exon 7.
For the populations in which several variants are responsible for RhD-negative status, other methods should be considered, for instance, the NGS (Next Generation Sequencing) method [57].
Currently, TaqMan real-time PCR is, due to its high sensitivity and specificity, the most common method for detecting the RHD allele of the fetus from free fetal DNA circulating in the peripheral blood of a pregnant woman. A general overview of currently used methods is given in Table 6. In the Czech Republic, real-time PCR [16,17,58] and droplet digital (dd) PCR [16] are used for non-invasive prenatal testing of the fetal RHD allele. Our study showed that the endpoint QF PCR method can fully replace real-time PCR. In the case of RHD-positive or male-specific fetuses, the amount of fetal fraction can be also relatively easily estimated. n-number of evaluated participants, a-blood samples from RhD-negative pregnant women collected in EDTA tubes, b-blood samples from RhD-negative pregnant women collected in BCTs tubes, *-when is provided, NGS-Next Generation Sequencing.
Accuracy characteristics were similar for both methods. The real-time PCR does not need a capillary electrophoresis separation step and so it takes about 2 h less.
On the other hand, the advantage of QFPCR lies in its direct confirmation of PCR specificity using the length of PCR fragments. In addition, an AMELY-specific probe can be used in male fetuses as a cffDNA control and quantificator.
The only currently published studies in the Czech Republic were on small groups of patients. Hromadníkova et al. analyzed a group of 45 pregnant women using real-time PCR [58]. Svobodová et al. compared dd PCR and real-time PCR on a group of 35 pregnant women [17].
Our results are comparable to those published (see Table 6).

Conclusions
The main goal of this study was to asses and compare possible clinical utilization of two methodological approaches of noninvasive fetal RHD genotyping in RhD-negative pregnant women. The study proved there was a minimal discrepancy between the RhD phenotype and the RHD genotype for the Czech population and since both the methods showed excellent power to predict the fetal RHD genotype from maternal plasma, it is possible to introduce them into clinical practice.