Prenatal non-invasive foetal RHD genotyping: diagnostic accuracy of a test as a guide for appropriate administration of antenatal anti-D immunoprophylaxis

Abstract

Background

Foetal RHD genotyping can be predicted by real-time polymerase chain reaction (qPCR) using cell-free foetal DNA extracted from maternal plasma. The object of this study was to determine the diagnostic accuracy and feasibility of non-invasive RHD foetal genotyping, using a commercial multiple-exon assay, as a guide to appropriate administration of targeted antenatal immunoprophylaxis.

Material and methods

Cell-free foetal DNA was extracted from plasma of RhD-negative women between 11–30 weeks of pregnancy. The foetal RHD genotype was determined non-invasively by qPCR amplification of exons 5, 7 and 10 of the RHD gene using the Free DNA Fetal Kit^® RhD. Results were compared with serological RhD cord blood typing at birth. The analysis of diagnostic accuracy was restricted to the period (24–28⁺⁶ weeks) during which foetal genotyping is usually performed for targeted antenatal immunoprophylaxis.

Results

RHD foetal genotyping was performed on 367 plasma samples (24–28⁺⁶ weeks). Neonatal RhD phenotype results were available for 284 pregnancies. Foetal RHD status was inconclusive in 9/284 (3.2%) samples, including four cases with RhD maternal variants. Two false-positive results were registered. The sensitivity was 100% and the specificity was 97.5% (95% CI: 94.0–100). The diagnostic accuracy was 99.3% (95% CI: 98.3–100), decreasing to 96.1% (95% CI: 93.9–98.4) when the inconclusive results were included. The negative and positive predictive values were 100% (95% CI: 100–100) and 99.0% (95% CI: 97.6–100), respectively. There was one false-negative result in a sample collected at 18 weeks. After inclusion of samples at early gestational age (<23⁺⁶ week), sensitivity and accuracy were 99.6% (95% CI: 98.7–100) and 95.5% (95% CI: 93.3–97.8), respectively.

Discussion

This study demonstrates that foetal RHD detection on maternal plasma using a commercial multiple-exon assay is a reliable and accurate tool to predict foetal RhD phenotype. It can be a safe guide for the appropriate administration of targeted prenatal immunoprophylaxis.

Introduction

Haemolytic disease of the foetus and the newborn (HDFN) has been the main cause of neonatal and perinatal morbidity and mortality for many decades^1,2. The impact of this disease in economically advanced countries has been greatly reduced by the existence of surveillance and prevention programmes. Until the 1960s HDFN affected about 7,000 neonates per year with a mortality of 1.5/1,000 births. The introduction of postnatal anti-D immunoglobulin (RhIg) (late 1960s) drastically decreased the risk of anti-D alloimmunisation, such that the current incidence of RhD HDFN is 0.01–0.03% and the mortality rate is lower than 2/10,000 births^3–6.

National guidelines from scientific societies and health institutions^7–13 strongly recommend that all RhD-negative women are routinely offered RhIg post-partum, during antenatal care (at 28 weeks) and following any potentially sensitising event in which foeto-maternal haemorrhage may have occurred. Current policy and legal practice is that women should be given appropriate information about RhIg - its benefits to foetal health (in current and future pregnancies) and potentially adverse events - in order to give a conscious consent.

The rate of RhD maternal-foetal incompatibility depends on the prevalence of RhD-negative and RhD-positive phenotypes, which is linked to RHD haplotype frequencies. In a predominantly white population about 40% of RhD-negative women carry a RhD-negative foetus^14–16. Therefore, during a pregnancy, 40% of RhD-negative women receive unnecessary administration of one or more RhIg, prepared from pooled human plasma and, even if current preparations are safe, they are exposed to a risk of infection from viral or prion contamination^17–21. Furthermore, there are ethical concerns about the source of hyperimmune plasma, its world-wide shortage and the wastage of an expensive product. Haemovigilance reports registered incidents involving neglected, inappropriate and/or unnecessary administration of RhIg: in 2016 SHOT reported 333 adverse events out of 409 reports (81.4%) related to omission or late administration of RhIg (2 anti-D immunisation) and 69/409 inappropriate administrations²².

The likely future direction of prevention of RhD HDFN lies in defining the RHD fetal genotype from cell-free foetal DNA (cffDNA) in maternal plasma⁵. The discovery of circulating foetal DNA in maternal plasma allowed invasive procedures, associated with the risk of miscarriage, transplacental haemorrhage and alloimmmunisation stimulus, to be abandoned²³. Large-scale studies demonstrated the feasibility of real-time polymerase chain reaction (PCR)-based screening for foetal RHD to guide targeted antenatal immunoprophylaxis^24–29, restricting the administration of this immunoprophylaxis to RhD-negative women who carry a RhD-positive foetus. In 2010 and 2011 Denmark and the Netherlands, respectively, implemented nationwide antenatal screening for foetal RHD genotyping; regional availability of prenatal RHD screening is also reported in Sweden, Belgium, United Kingdom, Czech Republic, France and Germany^30,31.

At present, in Italy, a routine protocol for non-invasive foetal RHD genotyping does not exist: this test is performed by a limited number of specialised Transfusion Service laboratories for anti-D alloimmunised patients only or by private laboratories, which are not to be trusted. In 2015 the Regional Blood Centre of the Region of Emilia-Romagna (Italy) supported a project including two phases: the first had the aim of determining the diagnostic accuracy and feasibility of non-invasive foetal genotyping at different gestational ages, comparing results with serological RhD typing on cord blood; the second phase planned to introduce RHD foetal genotyping into the antenatal screening programme in Emilia-Romagna. We report here the results of the first phase of the study.

Materials and methods

Five Regional Immunohaematology and Transfusion Services participated in the first phase of the project: Bologna (S.Orsola-Malpighi Polyclinic and Maggiore Hospital), Ferrara, Imola and Rimini.

RhD-negative pregnant women, with RhD-positive partners or partners of unknown RhD phenotype, presenting in hospitals at different gestational ages for antenatal immunohaematological tests (first trimester screening, third trimester - around 28 weeks) or invasive diagnostic procedures, were asked to take part in the study. Local physicians obtained written informed consent for genetic tests and for personal information/biological samples, and completed a report with pregnancy data. Approval was obtained from the Ethical Committee of the coordinating centre (S. Orsola-Malpighi Polyclinic - Bologna).

EDTA-anticoagulated blood (6–9 mL) was drawn and shipped, at room temperature within 48 hours of collection, to the Advanced Immunohaematology Laboratory in S. Orsola-Malpighi Polyclinic in Bologna. Samples for which informed consent was absent or incomplete, haemolysed samples and samples received more than 48 hours after collection were not accepted.

The compliant samples were double-centrifuged, aliquoted and stored at −30 °C until further processing. Two aliquots of maternal blood were frozen at −30 °C, as a source of maternal DNA for analysing potential maternal RhD variants.

Foetal DNA was treated twice for DNA extraction, using manual extraction by microcolumn: the QIAamp^® DNA DSP Blood Mini Kit (Qiagen, Hilden, Germany) was used initially and then replaced by the QIAamp^® DSP DNA Virus Kit (Qiagen), which is more specific for small DNA fragments³². Given the high percentage of inconclusive results and low threshold cycle (Ct) in controls, we finally performed extraction with the QIAamp^® Circulating Nucleic Acid Kit (Qiagen). Foetal DNA was extracted from 1 mL plasma samples containing 5 μL of diluted maize DNA (as an extraction/amplification control), according to the manufacturer’s instructions.

Using the Free DNA Fetal Kit^® RhD (Institut de Biotechnologies Jacques-Boy, Reims, France) and LightCycler^® 480 Probes Master (Roche, Rotkreuz, Switzerland), we carried out, in duplicate, real-time PCR analysis for RHD exons 5, 7 and 10 on cffDNA isolated from maternal plasma, generating six test results for each sample. Three controls were added for each series of extractions: RhD-negative and RhD-positive plasma controls provided with the kit, a blank control with extraction water instead of plasma, and nuclease-free water used for PCR mix (single amplification).

All PCR tests were performed with the Dx Real-time System (Biorad, Hercules, CA, USA) applying the amplification conditions indicated by the manufacturer. Traceability of all phases of the analysis protocol (plasma storage, DNA extraction, PCR amplification) with manual transfer of plasma/DNA was guaranteed by the supervision and signatures of two operators on dedicated check-lists.

The manufacturer’s instructions suggested the following Ct values as references for the acceptance of batch and sample results: Ct values between 35 and 41 for each target in samples, Ct values <39 for the positive control, Ct values ≤ 37 for the maize DNA IVR2 exon for both samples and controls. These values were applied during the validation of the method. Subsequently, in accordance with scientific consultant suggestions, Ct value ranges of each target (RHD exons 5, 7 and 10 in samples and in positive control, maize DNA) were determined for each lot of probes (Table I). Ct values <40 for exon 5 and <41 for exons 7 and 10 were interpreted as a positive signal.

Table I.

Cut-off Ct values applied in the typing algorithm.

	Target	Range
Positive samples	Exon 5	34.33–39.33
	Exon 7	35.84–40.80
	Exon 10	35.54–40.78
Positive controls	Exon 5	34.24–36.18
	Exon 7	35.95–37.81
	Exon 10	35.34–37.96
Maize DNA		34.42–35.64

Ct: Cycle threshold

Results were validated only if: (i) no amplification curve was observed for each target of the RhD-negative control, blank control and PCR water (to exclude contamination); (ii) the Ct values for exons 5, 7 and 10 of the RhD-positive control were, respectively, 35.21±0.97, 36.88±0.93 and 36.65±1.31; and (iii) the maize DNA IVR2 exon was amplified with a Ct range of 35.03±0.61, confirming the efficacy of DNA extraction and the absence of PCR inhibition for each sample/control.

The genotype was reported as RHD negative when all RHD PCR reactions were negative (6/6) and the RHD genotype was considered positive when at least 5/6 PCR reactions were positive. RHD PCR-positive reactions ≤4/6 were reported as inconclusive. Discrepant or inconclusive results were repeated on a new plasma aliquot to identify a technical error, a different sensibility of probes with low cffDNA concentrations or a coding/not coding RhD variant.

A RHD PCR-positive reaction with Ct values below 35 or lower than the Ct of the positive control for each target, in one or more RHD exons, suggested the presence of a silent RHD gene in the maternal genome³³ and invalidated the foetal RHD genotype, which was considered as RHD positive. Afterwards serological RhD phenotyping, antiglobulin RhD test and PCR-single strand polymorphism (SSP) molecular typing were performed on the maternal sample to identify or exclude a RhD variant.

Predicted RhD phenotype from the non-invasively determined foetal RHD genotype was confirmed by serological RhD cord blood typing at birth, performed locally following validated procedures.

Results

From February 2016 to January 2018, 455 RhD-negative pregnant women were recruited; 31/455 collected samples were not tested (Figure 1). One of the aims of the study was to determine the diagnostic accuracy of non-invasive foetal genotyping at different gestational ages, however, we collected a small number of samples from women in the first trimester of pregnancy. Therefore, we restricted the analysis of diagnostic accuracy to the period of 24–28⁺⁶ weeks, during which foetal genotyping is usually performed for targeted antenatal RhIg administration. Out of 424 genotyped samples, 31 collected before 23⁺⁶ weeks and 26 after 29 weeks were excluded; one sample was excluded because no cord blood phenotype was obtained owing to a stillbirth (Figure 1). The foetal RHD genotype and the RhD serological cord blood phenotype were available for 284 pregnancies; 82 pregnancies are still ongoing.

Figure 1.

Flow chart summarising the results of non-invasive foetal RHD genotyping.

With regards to ethnicity, the majority of the women were Caucasian (273 samples, 96.1%), nine (3.1%) were African, one (0.3%) was Asian (India), and one (0.3%) came from Central America. It should be noted that these rates do not represent the proportion of racial mixture in the region (31%)³³. In fact, because of language difficulties, participation in the project by foreign women was low (Table II). The antibody screening performed at the beginning of third trimester of pregnancy was negative in 260/284 pregnant women and positive in 24/284 owing to passive anti-D from previous RhIg, administered after chorionic villous sampling in the first trimester. Post-natal screening, performed 6 months after delivery, was negative in all tested women (248/284).

Table II.

Population characteristics.

Maternal age (years) mean (range)	33.8±4.8
Ethnicity
Caucasian	273
African	9
Asian	1 (India)
Central America	1 (Cuba)
Parity
Nulliparous	118
Multiparous	166
Twin pregnancies	6
Prenatal antibody screening
Negative	260/284
Positive*	24/284
Postnatal antibody screening**
Negative	248/284
Positive	0/284

^*Anti-D from previous immunoprophylaxis, administered in first trimester;

^**Performed 6 months after delivery; 36/284 women have not yet been screened, because they gave birth less than 6 months ago.

As a result of a validation process, which overcame the problems that had surfaced with the QIAamp^® DNA DSP Blood Mini Kit and QIAamp^® DSP DNA Virus Kit (Qiagen), the foetal DNA isolation is now performed with QIAamp^® Circulating Nucleic Acid Kit (Qiagen), more specific than the previous ones for extraction of small DNA fragments of human origin³⁴. Table III shows details about the efficiency of DNA recovery of the extraction kits.

Table III.

Typing results.

Extraction kit	Typed samples	Conclusive results with one measurement	Conclusive results with two measurements	Conclusive results with three or more measurements	Inconclusive results	False-positive	False-negative
QIAamp DNA DSP Blood Mini Kit	46 N	38	4	2	2	0	0
QIAamp DSP DNA Virus Kit	209 N+2 I	170	19	2	20	1	1
QIAamp Circulating Nucleic Acid Kit	96 N+16 I	104	4	0	4	1	0

N: new samples; I: sample with inconclusive results obtained with other extraction kits.

In 198/284 samples a RHD-positive test result (69.7%) and in 77/284 samples (27.1%) a RHD-negative test result were obtained. In two cases the foetuses have been reported as RHD positive, whereas the RhD cord blood serology showed the neonates to be RhD negative. In the first sample, the suspicious of a mistake during one of the manual steps of the typing process prompted us to introduce the check-lists; the second sample was probably mistaken by contamination. No pseudogenes or non-coding variant genes were identified. There was one false-negative result (RhD-positive foetus with a RHD-negative maternal cffDNA result) in a sample taken at 18 weeks of gestation.

Genotyping results were inconclusive in 9/284 (3.2%) samples; in five cases the reaction pattern was variable in subsequent repetitions and not interpretable according to our criteria: RhD phenotyping at birth confirmed one as RhD negative and four as RhD positive. In 4/11 samples we could not identify the RHD fetal genotype because of maternal RhD variants (Table IV). In two samples 6/6 positive signals with low Ct values were obtained, while RhD serological typing was repeatedly negative. PCR-SSP analysis on maternal DNA predicted the presence of RhD variants with very low antigen expression (RHD*01W.29 and RHD*11). In two pregnancies the reaction pattern was a single amplification signal only with the exon 10 probe and low Ct values: the antiglobulin RhD typing was negative and the PCR-SSP analysis revealed the presence of hybrid genes, associated with a RhD-negative phenotype, in which RHD exons 5 and 7 are substituted by the corresponding RHCE.

Table IV.

Maternal RhD variants.

ID	Week of gestation	Rh phenotype	RhD AG test	Ct exon 5	Ct exon 7	Ct exon 10	Maternal RHD genotype	Predicted maternal RhD phenotype
227/2016	28	Ccee	Positive	28.07	29.42	29.03	RHD*01W.29 (RHD*weak D type 29)	CcDuee
230/2016	28	Ccee	Positive	32.47	34.15	33.60	RHD*11 (RHD*weak D type 11)	CcDuee
245/2016	29	Ccee	Negative	N.A.	N.A.	33.42	RHD-CE(2–9)-D	Ccdee
248/2016	28	*Ccee	Negative	N.A.	N.A.	31.76	RHD-CE(3–7)-D	Ccdee

N.A.: not amplified; AG: anti-globulin;

^*partial C antigen associated with d(C)ces aplotype.

The sensitivity of the non-invasive foetal RHD genotyping was 100% and the specificity was 97.5% (95% CI: 94.0–100). The diagnostic accuracy was 99.3% (95% CI: 98.3–100), decreasing to 96.1% (95% CI 93.9–98.4) when the inconclusive results were included. The negative and positive predictive values were 100% (95% CI: 100–100) and 99.0% (95% CI: 97.6–100) respectively.

Discussion

The determination of foetal genotyping on circulating cffDNA in maternal plasma transformed prenatal cares in all respects³⁵. About 20 years ago, Lo and co-workers³⁶ demonstrated the presence of male foetal DNA sequences in maternal plasma, where it can be detected as early as at 5 weeks of gestation; however, the paucity of cffDNA and its coexistence with maternal DNA were limitations to its diagnostic use, especially at early gestational ages. The diagnostic application of non-invasive RHD foetal genotyping, previously defined at 16 weeks by Lo et al.³⁶, was subsequently antedated to 11–14 weeks³⁷.

cffDNA from maternal plasma is usually isolated manually with commercially available kits, but automation of the process allows its application on a large scale^5,38 with good recovery of foetal DNA. Finning et al.¹⁹ demonstrated that a high-throughput technique using robotic isolation of DNA from maternal plasma could reduce the false-negative rate to 0.2% in the third trimester, although it remained 3.5% at 11–13 weeks³⁷. Recently Moise et al.³⁹ reported a 0.32% false-negative rate in first trimester samples analysed with a mass spectrometry platform, opening new perspectives in testing methods. Different, specific and highly sensitive methodologies for the detection of cffDNA were successfully applied for diagnostic purposes from 11–12 weeks of gestation^37,40–49, exploiting the presence/absence of the RHD gene amplification and assuming a negative RHD maternal genotype (the RHD gene is deleted in Caucasians and 19% of Afro-Americans) corresponding to the RhD-negative phenotype. A recent meta-analysis by Zhu et al.⁵⁰, performed on 41 publications and including ~11,000 samples, reported an overall diagnostic accuracy of 98.5% (99% in the first trimester) and a negative predictive value of 98%.

The clinical application was initially restricted to anti-D immunised women at high risk of HDFN^5,51,52 and was then extended to non-immunised RhD-negative women for the appropriate administration of targeted prenatal RhIg. Since the first large-scale feasibility studies^24,25, many authors have demonstrated the accuracy of foetal RHD genotyping, also at early gestational age^37,47,48, and have evaluated strategies to implement it (Table V). Besides nations in which, nationally or regionally, non-invasive foetal RHD genotyping is part of routine antenatal screening, it is under evaluation in other European countries.

Table V.

Published studies on non-invasive RHD foetal genotyping performed with real-time polymerase chain reaction.

Reference (First Author, year, ref)	Country	Samples (n)	Gestational week	RHD exons	Controls	False-positive results	False-negative results	Inconclusive results	Sensitivity (%)	Specificity (%)	Accuracy (%)
Rouillac-Le Sciellour C, 2004, 53	France	893	7–40	7, 10	None						99.5
van der Schoot CE, 2006, 24	The Netherlands	1,257	30*	7	None	0	3	0	99.6
Rouillac-Le Sciellour C, 2007, 54	France	300	10–34	7, 10	Maize DNA (T)	2			100	99.9	99.3
Finning K, 2008, 19	UK	1,869	8–38	5, 7	CCR5 (T) SRY (F)	14 (0.8%)	3 (0.2%)	56 (3%)	99.7§	94§	99.7§
Minon JM, 2008, 55	Belgium	545	10–38	4, 5, 10	CCR5 (T) SRY (F)	1			100	99.8
Muller SP, 2008, 25	Germany	1,113	6–32	5, 7	β-globin (T)	NR	NR	NR	99.7/99.8#	99.2/98.1#	99.5/99.2#
Hyland CA, 2009, 56	Australia	140	12–40	4, 5, 10	CCR5 (T) SRY RASSF1A (F)		0	100			96.4
Wang XD, 2009, 57	China	78	14–40		β-globin (T) SRY, STR’s (F)	3	0	3	100		94.9
Morano, D 2010, 58	Italy	11	14–22	4, 5, 10	GAPD (T) SRY (F)	1	1	3			54
Mohammed NB, 2010, 59	Pakistan	21	20–39	5	β-globin (T)	3	1		62.3	62.5
Chinen PA, 2010, 60	Brazil	102	7–36	7, 10	β-globin (T)	2	0		100
Scheffer PG, 2011, 34	The Netherlands	212	7–38	5, 7	Albumin (T) SRY, biallelic polymorphisms (F)	0	0		100
Akolekar R, 2011, 37	UK	586	11–14	5, 7	CCR5 (T)		6 (3.5%)	84 (14.3%)	98.2	100	98.8
Clausen FB, 2012, 61	Denmark	2,312	25*	5, 7/5, 10/5, 10 **	CCR5 SOD GAPDH (T)	0.26% (6)	0.087% (2)	3.2% (74)	99.9	99.3	99.6
Daniels G, 2012, 26	UK	4,876	7–24	5, 7	NR	17	3		>99.2&
De Haas M, 2012, 62	The Netherlands	6,941	NR	5, 7	None	0.2%	56 (1.1%)	12 (0.2%)	>99.6@
Grande M, 2013, 63	Spain	284	24–26	5, 7 (+6, 10)	β-actin (T) SRY (F)	1			100	99	99.6
Macher HC, 2012, 64	Spain	1,012	10–28	5, 7	SRY (F)	3			100	98.6	99.3
Oliveira J, 2011, 31	Portugal	19	22–37	7, 10	SRY (F) GLO (T)	1 (5.9%)		80	100	67	94
Sbarsi I, 2012, 65	Italy	31	NR	4, 5, 10	CCR5 (T) SRY (F)				100	100	100
Wikman AT, 2012, 28	Sweden	4,118	8–40	4	GAPDH (T)	14 (1.1%)	23 (1.1)	14	98.9$
Dovč-Drnovšek T, 2013, 66	Slovenia	153	7–38	5, 7, 10 (+ int 4)	ALB (T) SRY (F),				100	100	100
Tiblad E, 2013, 67	Sweden	8,374	29*	4	GAPDH (T)				98.9	98.9
Chitty LS, 2014, 48	UK	2,288	11–23	5, 7	CCR5 (T) SRY (F)	18	19	393 (8%)	99.34	94.91	91.24
Schmidt LC, 2014, 68	Brasil	55	12–29	5, 7	SRY (F)		1	5 (9.1%)	100	100	100
De Haas M, 2016, 29	The Netherlands	25,789	27*	5, 7	None	225 (0.57%)	9 (0.03%)	0.21%	99.94	97.74	99.09
Vivanti A, 2016, 49	France	416	10–14	10	Mouse GALT gene (T)	7	0	9 (2.2%)	100	95.2	98.3
Trucco Boggione C, 2017, 69	Argentina	298	19–28	4, 5, 7, 10	SRY (F)	1		8	100	98.8
Geifman-Holtzaman O, 2006, 70	Meta-analysis	3,261 (37 articles)	8–42						95.4	98.6	91.4***
Zhu Y, 2014, 50	Meta-analysis	11,129 (41 studies)							99	98	98.5 (99% in first trimester)

^§The foetus was categorised as RhD positive if RHD predicted phenotype was RhD positive, RHD variant or inconclusive;

^#extraction by spin-column method/magnetic tips;

^$the sensitivity was 99.3% at 10 weeks and 100% after 22 weeks;

^&specificity was >99.8 at 11 weeks (3 false negatives out 4,011 samples);

^@Estimated from reported false-negative results <0.25%;

^*mean;

^**multicentre study;

^***(90.8 % in the first trimester, 85.0% in the second trimester, 85.3% in the third trimester); CCR5: C-C chemokine receptor type 5; GAPDH: glycerldehyde 3-phosphate dehydrogenase; SOD: superoxidase dismutase; (T): total cell-free DNA control; (F): foetal cell-free DNA control; NR: not reported.

In Italy the first experience⁵⁸ of foetal RHD genotyping was conducted in 2010, when cffDNA was tested by real-time PCR and RhD foetal status was successfully determined. This was followed by the Pavia experience⁶⁵ in 2012, validating the local protocol and demonstrating feasibility and 100% accuracy.

The project supported by the Regional Blood Centre of Emilia-Romagna gave us the opportunity to evaluate the diagnostic accuracy of a commercial test for non-invasive foetal RHD genotyping and to introduce it into the routine antenatal protocol for screening RhD-negative pregnant women.

The validation of the process was complex because of the difficulties encountered with foetal DNA extraction, which is a critical step²⁵. In fact, the quantity of foetal nucleic acids is minimal in comparison with the background of maternal DNA and, despite the high sensitivity and specificity of testing methods, they are prone to false-negative results. False-negative results could cause the lack of anti-D administration and the risk of alloimmunisation. In our study we had one false negative result in an 18-week sample: RhD-positive typing of the neonate was confirmed on a new sample from the newly born baby. After exclusion of an operating mistake, the retrospective analysis showed that the typing session (in which extraction was performed with the QIAamp^® DSP DNA Virus Kit) had RhD positive control Ct values at the higher limits, although in the defined range. The genotyping on a second maternal plasma aliquot, extracted with the QIAamp^® Circulating Nucleic Acid Kit, confirmed the positive serological result. The QIAamp^® DSP DNA Virus Kit extraction kit is one of three tested during this study and, although the best in an International Workshop³², it showed a low specificity in foetal DNA recovery, as well as a failure to meet defined Ct values for amplification and extraction controls. These performances forced us to repeat the test even more than once (Table III), but the results were often only clarified when using the QIAamp^® Circulating Nucleic Acid Kit, which revealed greater efficiency in foetal DNA recovery and showed a higher conclusive result rate at first analysis. Therefore, in these samples, two of the main causes of false-negative results are concurrent: the efficiency of foetal DNA extraction⁶³ and the early gestational age, when the quantity of foetal DNA is minimal^37,48.

The maize DNA provided in the Free DNA Fetal Kit^®RhD is considered an adequate control for both DNA extraction and PCR amplification⁷¹ and the lack of a positive control for cffDNA is considered acceptable if foetal genotyping is used for screening. Relying on our experience and with the aim of avoiding false-negative results, we will implement a second analysis at least 4 weeks later, as recommended^54,72, to confirm negative results obtained at an early gestational age, as an alternative to a cffDNA control^37,54.

The inconclusive result rate was 3.2%, as expected with amplification of more targeted RHD exons. In fact, this strategy could increase indeterminate results, conceivably because of low foetal DNA recovery (unrelated to gestational age) in RhD-positive samples and different primer affinity, and because of contamination or non-specific amplification in RhD-negative samples, but at the same time it reduces false results, compared to analysis of single exons. Even if there are experiences demonstrating the reliability of single-exon assays (with a low false-negative rate)^{24,28,49,67,70} and their ease of application in routine analysis, we chose the opportunity, offered by a commercial CE-marked kit, to explore more RHD exons simultaneously and to identify maternal and/or foetal RHD variants that may not be recognisable by the conventional serological techniques or may cause maternal alloimmunisation, respectively. In a complex and polymorphic gene such as RHD, the amplification/not amplification of a single exon (5, 7, 10 or others) is informative only for the single exon analysed and it could deceive, giving only a partial view of the gene status. In fact, the disadvantage of amplifying only one RHD exon is the risk of false-positive results, if the target is a non-coding exon (i.e. exon 7 in RHDψ), or false-negative results, if the target is substituted by a RHCE coding sequence (i.e. exon 5 in RHD*DVI).

On the other hand, the availability of primers that amplify three pivotal exons of RHD (exons 5, 7 and 10) allows the gene to be studied from a different point of view and with more accuracy. After the first decade of experience, the Special Non-invasive Advances in Fetal and Neonatal Evaluation Network⁷³ recommended the use of specific primers for exons 5 and 7 to overcome the complexity of RHD. Exon 5 is involved in many RHD variants and hybrid genes: for example, it does not amplify in RHDψ (coding for the RhD-negative phenotype in 67% of black Africans) and RHD*6(RHD*DVI) (the most common RHD variant in Caucasians)^29,74 alleles. The analysis of exon 5 alone gives a correct negative result in the first case, but a false negative result in the second one: in the latter case it would mean no RhIg administration in a RhD-incompatible pregnancy and the risk of anti-D alloimmunisation. Exon 7 has many D-specific nucleotides and is absent only in a minority of RHD variants (such as RHD-CE-D^s)^54,74, while it is amplified in the RHDψ allele giving a false-positive result, which would mean inappropriate RhIg administration. However, if exons 5 and 7 are analysed together, a discrepant amplification result between them suggests a variant, induces a more detailed study and guides appropriate management of pregnancy. Finally, as in many hybrid RHD-CE-D genes (some of which associated with weak or partial antigen expression) exons 5 and 7 are both replaced by the corresponding RHCE exons, the inclusion of exon 10 confirms or rules out the presence of the RHD gene, depending on positive or negative results; in fact, exon 10 is one of the most preserved RHD exons (although it is a non-coding exon and it could give false-positive results if tested alone)⁷⁵ and warrants an high specificity of the test.

This compound of primers is useful in RHD foetal genotyping of pregnant African women, in whom the frequency of RhD variants is higher (up to 6–7%) than in Caucasians (0.2–1%)¹⁴. In fact one of the RHD maternal variants (RHD*CE(3–7)-D) was identified in an African woman associated with a d(C)ce^s allele and partial C; the other hybrid gene (RHD-CE(2–9)-D), associated with a RhD-negative phenotype and RHCE*Ce or RHCE*cE alleles in Europeans, was found in a Caucasian woman.

About one decade ago, Rouillac-Le Sciellour et al.⁵⁴ performed preliminary studies with the Free DNA Fetal Kit^® RhD (with primers for RHD exons 7 and 10) and reported high sensitivity and specificity rates (100% and >99% respectively); however, in this article the authors affirmed that “the first generation kit is not suitable for correct genotyping of foetuses from women carrying a RHDψ pseudogene in their genome”. In fact, the kit could not discriminate the presence of RHDψ in a foetus from a normal coding RHD gene because exons 7 and 10 were amplified in both cases: if RHDψ was present, this would mean inappropriate immunoprophylaxis or wrong evaluation of HDFN risk in the case of an alloimmunised pregnant woman. The subsequent implementation of exon 5 primers is helpful for the identification of RhD variants.

The low number of samples collected at an early gestational age is a weakness of our investigation and we, therefore, excluded such samples from the statistical analysis of the study. Nevertheless, after the inclusion of samples taken at an early gestational age (<23⁺⁶ week), the sensitivity and accuracy of the test were 99.6% (95% CI: 98.7–100) and 95.5% (95% CI: 93.3–97.8) respectively. These results encourage the performance of further studies to confirm the reliability of this protocol also in the first trimester.

The high sensitivity (100%) and diagnostic accuracy (99.3%) rates, obtained in the period during which foetal genotyping is usually performed for targeted antenatal RhIg administration (24–28⁺⁶ weeks), show that fetal RHD genotyping is a reliable test for managing RhD-negative pregnancies and a powerful diagnostic tool in prenatal care, if appropriate strategies are applied. Beyond economic and ethical advantages, we believe that it is a way to prevent anti-D alloimmunisation and HDFN with an integrated and multidisciplinary management of pregnancy: it deflects attention from RhIg to pregnant women and their babies and improves safety, without increasing the risk of immunisation or foetal disease.

Conclusions

The findings of this study demonstrate that foetal RHD detection on maternal plasma using a commercial multiple-exon assay is a reliable and accurate tool to predict foetal RhD phenotype, with acceptable rates of false-negative and false-positive results. It can be a safe guide for the appropriate administration of targeted prenatal RhIg, avoiding unnecessary exposure to immunoprophylaxis. The presence of three RHD exon primers increases the sensibility and provides the opportunity to identify RhD variants in mixed ethnic populations. The validation of the process allows its introduction into routine clinical practice in the Region.