SEARCH

SEARCH BY CITATION

Keywords:

  • Fetal cells;
  • fetal DNA;
  • fetal RhCE;
  • prenatal diagnosis

Abstract

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References

Background  The successful prevention of RhD disease has brought attention to other red blood cells’ antigens causing alloimmunisation including RhC/c and RhE/e. Prenatal diagnosis of fetal Rh genotype from maternal blood is in clinical use in Europe but not in the USA.

Objective  To estimate the collective reported diagnostic accuracy of fetal RhCE genotyping from peripheral maternal blood and compare the results of genotyping when fetal cells and free fetal DNA (FfDNA) are used.

Search strategy  English-written literature describing fetal RhCE determination from maternal blood using fetal cells or FfDNA was performed using medical subject headings and text words. The sources included Pubmed (1966–2007), Ovid (1966–2007), CINAHL, The Cochrane Library, ACP Journal Club and OCLC. Key words were prenatal diagnosis, fetal RhCE, fetal DNA in maternal blood and alloimmunisation.

Selection criteria  A study was considered eligible if it described fetal RhCE type determination using maternal peripheral blood reported in the English literature. Abstracts were excluded.

Data collection and analysis  From each study, we determined the number of samples tested, fetal RhCE genotype, the source of the fetal DNA, gestational age, presence of alloimmunisation and confirmation of fetal RhCE type. Exclusions and inclusions were noted. We calculated composite estimates of accuracy using a weighted random effects model. We assessed the papers against an international quality, STARD checklist which is standards for reporting studies of diagnostic accuracy.

Main results  We identified 20 protocols in six English-written publications reporting fetal RhC/c (seven protocols) and/or E/e (13 protocols) genotyping using DNA obtained from maternal blood for a total of 369 samples. For RhC/c, 176 samples were tested and for RhE/e, 193 samples were tested. Accuracy was determined for each study and for all studies. The combined accuracy of fetal genotype was 96.3% for RhC/c and 98.2% for RhE/e. Only a few samples of the sorted cells were found to be a source for accurate diagnosis, but plasma was consistently the best source of fetal RhCE genotyping in 147/147 (100%) for RhC/c and 168/168 (100%) for RhE/e.

Conclusions  The combined accuracy of noninvasive fetal RhC/c or RhE/e determination using maternal peripheral blood is 96.3% and 98.2%, respectively. FfDNA in maternal plasma is a better source for genotyping reported to be 100% correct for both RHCE genotypes. Further studies and reports of accuracy from laboratories performing the tests are required before prenatal determination of fetal RhC/c or RhE/e genotypes from maternal blood can safely replace the current methods used in the management of the RhC/c or RhE alloimmunised pregnancies.


Background

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References

Cloning of the Rh genes and the ability to determine fetal Rh genotype prenatally using fetal DNA has brought the management of alloimmunised pregnancies into a new era. Using polymerase chain reaction (PCR), fetal RhD, C/c, E/e and Kell types have been determined from fetal DNA extracted from amniocytes obtained during amniocentesis in alloimmunised pregnancies.1,2 This test has been incorporated into the current management of the alloimmunised pregnancy and if the fetus is negative for the specific antigen causing alloimmunisation, no further surveillance is required. Since amniocentesis is an invasive test with a risk of up to 0.5% for pregnancy loss, the research in the field is aimed at developing a noninvasive, risk-free, maternal blood test for fetal Rh determination.

Fetal RhD genotyping from maternal blood was first reported using free fetal DNA (FfDNA) in the plasma or using DNA extracted from fetal nucleated red cells isolated from maternal blood.3,4 These initial studies were followed by multiple reports of successful prenatal diagnosis of fetal RhD type and other fetal conditions by using fetal DNA obtained from peripheral maternal blood.5–7 Fetal DNA enters the maternal circulation either as free DNA form or as fetal-derived cells. The maternal immune system may facilitate the release of FfDNA by causing destruction of the villous trophoblast or the fetal cells entering the maternal blood. Apoptosis of villous trophoblast in situ or apoptosis of nucleated fetal cells circulating in maternal blood may also lead to release of fetal DNA into the maternal circulation.8,9

The successful prevention of RhD disease has brought attention to other red blood cells’ antigens including RhC/c and RhE/e. The presence of RhE/e and RhC/c antibodies causing alloimmunisation was shown to be 14% and 10.7%, respectively, in an upstate New York female population with RhD still being the leading antigen at 21%.10 Being the most common, the majority of research in the field has focused on fetal RhD genotyping from maternal blood. In 2005, editorials in the obstetric literature11–13 have encouraged bringing fetal RhD genotyping from peripheral maternal blood to clinical use in the USA. These are based on work performed in laboratories in Europe that are already using this test routinely as part of their prenatal testing of the RhD-negative pregnant women.14 Previously, we reported meta-analysis of articles describing fetal RhD genotyping from maternal blood with diagnostic accuracy of 94.8% for noninvasive RhD typing.15 Considering the high risk of severe haemolytic disease in pregnancy alloimmunised to Rhc and occasionally to RhE and RhC, and because of increased interest in noninvasive testing from maternal blood, we searched for the information that is available on fetal RhCE genotyping using maternal blood. We compared the methods and results in the English literature describing fetal RhE/e and/or RhC/c genotyping from maternal blood. Our goal was to estimate, based on the existing studies, the collective reported diagnostic accuracy and the current limitations of fetal RhE/e and RhC/c genotyping from peripheral maternal blood.

Search strategy

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References

An independent literature search was performed by the authors (O.G.H. and C.A.G.) that combined medical subject headings and text words with no restrictions on language, publication type or publication date. The sources included Pubmed (1966–2007), Ovid (1966–2007), CINAHL, The Cochrane Library, ACP Journal Club, OCLC, abstracts from scientific forums and bibliographies of published articles, as previously described.15

Key words for the search used separately and in combination were prenatal diagnosis, fetal Rh, fetal RhC/c and/or fetal RhE/e, fetal DNA, fetal DNA in maternal blood, maternal plasma and maternal serum and alloimmunisation. The reference lists of the articles that were identified in the search were reviewed for relevant articles. Authors who had published more than one study were contacted to clarify case duplication in their reported studies, as previously described.15

Selection criteria

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References

A study was considered eligible for further review and analysis if it described fetal RhC/c or fetal RhE/e type determination in maternal blood, plasma or serum, and if confirmation of the fetus/newborn RhC/c or E/e type was provided. The pregnant women who participated in the studies could be alloimmunised or not. Abstracts were not included in the data analysis.

Data collection and analysis

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References

Standardised data collection forms were developed prior to data extraction in a Microsoft Excel format. Uniform headings for all articles were generated that included primary author, year of the study, journal title, number of women and tests, and inclusions and exclusions were recorded, gestational age, correct and incorrect fetal RhC/c and RhE/e determination, source of fetal DNA, number of women, neonate/fetal RhC/E type and the presence of alloimmunisation. Articles were reviewed against standards for reporting studies of diagnostic accuracy (STARD) checklist and guidelines.16

The rate of correct identification was calculated for each study based on the total correct tests divided by the total tests. Weighted averages based on sample size of each study were calculated. To model the between-study heterogeneity and synthesise the results appropriate for meta-analysis, we used a weighted random effects linear model (SAS PROC GLIMMIX). By the nature of studies from different laboratories in different parts of the world and using different protocols, we used the random effect model, assuming that heterogeneity is present between the studies. The random effects composite was calculated together with the 95% CI. The random effect model assumes that the individual studies vary randomly. We weighted the analysis based on the number of women in each study since we analysed multiple tests per women from the various studies. The overall diagnostic accuracy for fetal RhC/c and RhE/e determination from maternal blood was calculated separately for RhC/c from RhE/e. Diagnostic accuracies based on trimester tested were determined from studies where gestational age was given. Diagnostic accuracies were also calculated based on the source of the fetal DNA. Comparisons between various rates for each trimester or source were made using 95% CI for differences in Poisson rates.

Results

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References

We identified six English-written peer-reviewed publications that present 20 protocols for the fetal RhC/c and RhE/e determination in maternal blood7,17–21 (Tables 1–4). A protocol was defined as a method that is described in each article to determine fetal Rh genotyping. For RhC/c, there were a total of 176 samples from 125 women who were tested for fetal RhC/c genotype in peripheral maternal blood, and for RhE/e there were a total of 193 samples from 123 women who were tested for RhE/e genotype in peripheral maternal blood. For fetal RhC/c genotype from maternal blood, one study included three protocols17 and each of the other four studies18–21included one protocol each, with a total of seven protocols (Table 2). For fetal RhE/e genotype from maternal blood, one study included nine protocols7 and each of the other four18–21 studies included one protocol each for a total of 13 protocols (Table 4).

Table 1.  Weighted random effects estimate for RhCc of accurate diagnosis: 96.3% (CI: 80–100%) of fetal RhCc type from maternal blood
First author (year)Women (n)Samples (n)Correct diagnosis (n)Correct (%)
Geifman-Holtzmam17 (1998)11291758.6
Legler18 (2002)162424100
Hromadnikova19 (2005)416161100
Hromadnikova20 (2005)355100
Finning21 (2007)545757100
All reports12517616493.2
Table 2.  Fetal RhCc determination from maternal blood by protocol
First author (year)Women (n)Maternal samples tested (n)Correct (n)Correct (%)
  • *

    No enrichment.

  • **

    Mononuclear cell.

  • ***

    Sorted cell.

  • ****

    FfDNA from maternal blood reported in four different articles.

Geifman-Holtzman17 (1998)*1111327.3
Geifman-Holtzman17 (1998)**99666.7
Geifman-Holtzman17 (1998)***89888.9
Legler18 (2002)****162424100.0
Hromadnikova19 (2005)****416161100.0
Hromadnikova20 (2005)****355100.0
Finning21 (2007)****545757100.0
All reports12517616493.2
Table 3.  Weighted random effects estimate for RhE of accurate diagnosis: 98.2% (CI: 91.4–100%) of fetal RhE type from maternal blood
First author (year)Women (n)Samples (n)Correct diagnosis (n)Correct (%)
Geifman-Holtzman7 (2000)15403485
Legler18 (2002)163535100
Hromadnikova19 (2005)456767100
Hromadnikova20 (2005)355100
Finning21 (2007)444646100
All reports12319318796.9
Table 4.  Fetal RhE determination from maternal blood by protocol
First author (year)Women (n)Maternal samples tested (n)Correct (n)Correct (%)
  • *

    Cell sort CD36+/GPA+.

  • **

    Cell sort CD71+/GPA+.

  • ***

    Cell sort CD36−/GPA−.

  • ****

    Cell sort CD71−/GPA−.

  • *****

    Cell sort CD45−/CD71+/GPA+.

  • ******

    Cell sort CD45−/CD71−/GPA−.

  • *******

    Cell sort CD45−/CD36+/GPA+.

  • ********

    Cell sort CD45−/CD36−/GPA−.

  • *********

    Cell sort CD45+.

  • **********

    FfDNA from maternal blood reported in four different articles.

Geifman-Holtzman7 (2000)*77685.7
Geifman-Holtzman7 (2000)**33266.7
Geifman-Holtzman7 (2000)***33266.7
Geifman-Holtzman7 (2000)****333100.0
Geifman-Holtzman7 (2000)*****444100.0
Geifman-Holtzman7 (2000)******44375.0
Geifman-Holtzman7 (2000)*******344100.0
Geifman-Holtzman7 (2000)********344100.0
Geifman-Holtzman7 (2000)*********78675.0
Legler18 (2002)**********163535100.0
Hromadnikova19 (2005)**********456767100.0
Hromadnikova20 (2005)**********355100.0
Finning21 (2007)**********444646100.0
All reports12319318796.9

This analysis demonstrated that the combined accuracy for RhC/c was 96.3% (CI: 80–100%) (Table 1) with 164/176 samples correctly diagnosed. The fetal RhC/c genotype was correctly identified when the peripheral maternal blood sample was processed as follows: 3/11 (27.3%) samples using maternal whole blood without any separation, 6/9 (66.7%) samples using mononuclear cell layer and 8/9 (88.9%) samples using cell sorting. The overall results for correct fetal RhC/c genotyping were correctly determined in 17/29 (58.6%) samples when fetal cells were used and was correctly identified in 147/147 (100%) samples using FfDNA that was obtained from maternal plasma and amplified for the RhC/c gene as reported in a four different articles18–21 (Tables 2 and 5).

Table 5.  Diagnostic accuracy of fetal RhCc and fetal RhEe determination by fetal DNA source
Rh typeSourceSamples (N)Correct (N)Correct (%)
  1. NS, difference in Poisson rates.

  2. For fetal RhCc: n= 176 samples (125 women), difference = 41.4 (0.03, 0.80) P < 0.05, difference in Poisson rates. For fetal RhEe: n= 193 samples (123 women), difference 15 (−19, 50) P > 0.05.

CcMaternal blood, fetal cells (DNA)291758.6
CcMaternal plasma, FfDNA147147100.0
EeMaternal blood, fetal cells (DNA)403485
EeMaternal plasma, FfDNA153153100

The combined accuracy for RhE/e was 98.2% (CI: 91.4–100) (Table 3) with 187/193 samples correctly diagnosed. The fetal RhE/e genotype was correctly determined when the peripheral maternal blood sample was processed as follows: 6/7 (85.7%) samples of CD36+/glycophorin A (GPA)+ sorted cells, 2/3 (66.7%) samples of CD71+/GPA+ sorted cells, 2/3 (66.7%) samples of CD36−/GPA− sorted cells, 3/3 (100%) samples of CD71−/GPA− sorted cells, 4/4 (100%) samples of CD45−/CD71+/GPA+ sorted cells, 3/4 (75%) samples of CD45−/CD71−/GPA− sorted cells, 4/4 (100%) samples of CD45−/CD36+/GPA+ sorted cells, 4/4 (100%) samples of CD45−/CD36−/GPA− sorted cells and 6/8 (75%) samples of CD45+ sorted cells. The overall results for fetal RhE/e genotyping were correctly determined in 34/40 (85%) when fetal cells were used and were correctly identified in 168/168 (100%) samples when free DNA in maternal plasma was used as reported in four different articles.18–21 (Tables 4 and 5)

The studied samples were obtained from women at different trimesters of pregnancy and specific gestational ages were not always provided. The accuracies were demonstrated to be different for gestational age when fetal cells were used but no difference was noted when FfDNA was used. For RhC/c fetal typing, correct diagnosis was 75% in the first trimester when sorted fetal cells from maternal blood were used and accuracies of 56.5 and 50% were noted for the fetal RhC/c typing when sorted fetal cells were used in second and third trimester, respectively (Table 6). For both RhC/c and RhE/e, correct diagnosis of 100% was documented in first, second and third trimester when FfDNA was used (Table 6). In one study, 100% accuracy was documented when FfDNA in maternal plasma was used between 10 and 36 weeks of gestation but the exact trimesters for the samples were not reported.21 In protocols using fetal cells, the highest accuracies (100%) were found when the sorting was performed after depletion of the sample from maternal leucocytes with CD45 antibodies that was followed by sorting with CD36+ and GPA+, antibodies specific for erythroid cells. Therefore, with depletion of the sample using CD45 antibodies, followed by positive fetal cell sorting from maternal blood using red cells’ antibodies the results were comparable with the use of FfDNA from maternal plasma and both demonstrated similar high accuracies of 100% for RhC/c and RhE/e genotyping (Tables 2 and 4).

Table 6.  Accuracy of fetal RhCc determination reported on 95 samples by gestational age and fetal DNA source (n= 176 samples [125 women]) and accuracy of fetal RhEe determination reported on 72 samples by gestational age and fetal DNA source (n= 193 samples [123 women])
Rh typeFetal DNA source/gestational ageSamples (n)Correct (n)Correct (%)
  • NS, difference in Poisson rates.

  • *

    Difference = 18.4 (−63, 100).

  • **

    Difference = 25 (−100, 100).

  • ***

    Difference = 18.4 (−63, 100) P > 0.05.

Maternal blood, fetal cells
CcFirst trimester4375.0*
CcSecond trimester231356.5NS**
CcThird trimester2150.0 NS***
Maternal plasma, free DNA
CcFirst trimester44100.0
CcSecond trimester2121100.0
CcThird trimester4141100.0
Maternal blood, fetal cells
EeFirst trimesterNot reportedNot reportedn/a
EeSecond trimesterNot reportedNot reportedn/a
EeThird trimesterNot reportedNot reportedn/a
Maternal plasma, free DNA
EeFirst trimester44100.0
EeSecond trimester2424100.0
EeThird trimester4444100.0

Assessment of the case series articles that were selected against the STARD checklist found that the authors did not always report their results following the STARD checklist. Some of the articles were published before this checklist was available. Furthermore, no standard reference testing exists for noninvasive fetal genotyping to which the fetal RhCE genotyping from maternal blood could be compared. And the only other pertinent existing reference standard involves fetal Rh genotyping using amniotic fluid or Rh typing using serology after delivery. This lack of comparable procedures and studies limited the use of the STARD guidelines for the selected articles in this meta-analysis.

Discussion

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References

In this study, fetal RhC/c and fetal RhE/e genotype determination from peripheral maternal blood is demonstrated to be correct at 96.3% and 98.2%, respectively. This accuracy is compared with the reported accuracy of fetal RhD (94.8%) genotyping from maternal blood.15 However, this study demonstrates that fetal RhCE determination using FfDNA obtained from maternal plasma was 100% correct.

These accuracies are determined from published articles that used various methods and different sources of fetal DNA for fetal Rh genotyping. Therefore, we examined the variation in accuracies and their improvement when using FfDNA compared with fetal cells.

Since the molecular basis of human blood groups polymorphism was first described, fetal RhD, c and E, K, Fya, Jka genotype determination using fetal DNA extracted from amniotic fluid or chorionic villi has been used in clinical practice and has been part of modern management of the alloimmunised women.22 The availability of accurate noninvasive testing, for determination of fetal genotype could provide early first trimester testing for pregnant women with alloimmunisation that would eliminate the risk of the amniocentesis procedure-related pregnancy loss (up to 0.5%). Women whose fetus is determined to be negative for the specific antigen will continue with routine care. Women whose fetus is determined to be positive for the specific antigen will continue to be monitored closely with ultrasound and other testing as indicated.

The Rh antigens are the leading cause of alloimmunisation, and are also the most complex.10 The Rh blood group antigens of D, C/c and E/e are polypeptides present in the red cell membranes encoded by two highly related genes, RHD and RHCE. The RHD and RHCE genes have a high level of homology with 93.8% identity, indicating that the two Rh genes are most likely derived by duplication of an ancestral gene. Both genes contain ten coding exons in a tail-to-tail configuration (5′-RhD-3′-3′-RhCE-5′) on chromosome 1-p36.2-p34.23 Determination of RhD C/c or E/e antigen type requires understanding of these blood group polymorphisms and the existing variants in different populations. A change of a single nucleotide as well as gene deletion, pseudogene, and gene conversions could result in a wide variety of Rh phenotypes and impact the test results.24 The C/c polymorphism and activity are defined by a 307C>T single nucleotide polymorphism in exon 2 of RHCE encoding a Ser103Pro substitution in the second extracellular loop that determines the RhC or Rhc protein. Molecular testing for the C gene is difficult because exon 2 of RHD and of the C allele of the RHCE have identical sequences but testing for c, which is the antigen causing severe disease in alloimmunised pregnancies, is more straightforward as C307 is unique to the c allele of RHCE gene. The main way that European laboratories genotype for RhC is to detect a 109-bp insert in exon 2 of the RHCE gene found only in individuals with C expression.

The E/e polymorphism results from a 676G>C nucleotide polymorphism in exon 5 of RHCE gene encoding a Pro226Ala substitution in the fourth extracellular loop of the antigen. Detection of C676 loci consists with E genotype. The c and E tests employed restriction enzymes or allele-specific primers in exons 2 and 5 of RHCE, respectively, and thus the c and E genotypes and phenotypes can be determined.14

The Rh epitopes are conformational and a single amino acid substitution may create a new antigen and effect the expression of existing antigens and interfere with an accurate phenotype prediction. The RHCE gene encodes the C/c and E/e antigens and many others such as Cw (RH8), Cx (RH9) and VS (RH20).25 A significant number of exon 5 mutations are known that generate E and e variants in individuals of African descent too. Variants of RHCE gene coincide with antigen variations and effect accurate genotyping and phenotyping. These variants could be explained by complete or partial expression defect of the Cc/Ee antigens resulting from reduced transcriptional activity of the RHCE gene that stems from rearrangement of the Rh locus.26

In attempting fetal genotyping from maternal blood especially for clinical use, the different phenotypes should be considered and should not be missed to assure higher accuracy of the test. As demonstrated in the publications that were reviewed in our study, accurate prediction of the complete Rh phenotype is accomplished when specific primers for single nucleotide polymorphisms or for multiple polymorphic sites are used. Additionally, the use of quadruplicate testing with real-time PCR partially compensated for lack of sensitivity with excellent specificity to predict RHD and the C/c, and E/e allele as was reported.18,19,21

The limitations of RhC/c and RhE/e genotyping in blood from larger general patient population (sensitivity of 98.3–96.9%)27 compared with standard reference of serology (100%) were addressed previously in numerous publications but with more experience in the field, with cloning and sequencing of Rh genes, improving the molecular techniques and accumulating knowledge on the gene structure, all have increased accuracy to predict Rh phenotype.26

If fetal genotyping from maternal blood is attempted in alloimmunised pregnancies at risk, the knowledge of different genotypes in diverse population is of major importance to avoid false-positive or false-negative fetal genotyping that may result in poor outcome of affected pregnancies. Our study includes small number of reported studies with few alloimmunised pregnancies. The samples probably do not include all possible gene/antigen rearrangements and therefore should be interpreted cautiously. The accuracy of determination of fetal RhCE from maternal blood is high and close to 100% when the source is fetal DNA from maternal plasma. In reviewing the methods used in the selected studies, we found that using sorted cells with antibodies specific to red blood cells’ antigens as the source of fetal DNA, rather than using cells that were not sorted, led to similar accuracies of FfDNA use. However, the laboratory method required to sort cells is more complicated, time-consuming and very costly compared with the relative simple use of FfDNA in maternal plasma. The laboratory methods using plasma can be automated, accurate, reproducible and are less time-consuming.4 This test can be brought to clinical practice and applied at the same time to many blood samples with results available in less than 24 hours. An additional advantage of using FfDNA is its very short postpartum half-life, therefore avoiding the risk of obtaining fetal DNA from previous pregnancies.

The majority of the inaccuracies of the tests that were reported were inconclusive or resulted from failure to amplify/find DNA in the sample. The accuracy increased with the use of methods to ‘purify’ the maternal peripheral sample for the fetal DNA. In the selected studies, we noticed an improvement in accuracy from whole blood, to mononuclear layer, to sorted cells, to FfDNA. There was a significant increase in accuracy from 27% in whole blood to 100% for FfDNA. In addition, in one study, it was demonstrated that accuracy improved with gestational age. When blood samples for fetal genotyping that gave inconclusive results early in the second trimester were repeated later in pregnancy, the results were conclusive and predicted correct fetal phenotype.21

The successful amplification of the fetal DNA in the overwhelming background of maternal DNA in the plasma sample and confirmation that the source of the amplified DNA is the fetus of the current pregnancy remains most difficult. Cell-free DNA isolated from the plasma of pregnant women contains 94–97% maternal DNA and 3–6% fetal DNA.28 A technique that will enable depletion of maternal DNA from the studied material will improve detection of fetal DNA. When fetal cells were used, there were negative and positive selections of maternal cells with attempts to separate them from the fetal cells. By doing this, high accuracy was achieved similar to the accuracy of FfDNA from maternal plasma. Additionally, when FfDNA is used, a carefully selected internal positive control to confirm the presence of fetal DNA must be used, either SRY gene in male fetuses or other polymorphic markers in female fetuses, such as HLA gene29 for when HLA differences exist between both parents. Alternatively, using fetal/maternal methylation differences such as the hypermethylated fetal RASSF1A gene or the hypomethylated Maspin gene that serves as control for the presence of FfDNA as was previously reported.30

The possibility of finding fetal cells from a prior pregnancy is not resolved yet. Designing a test that includes maternal, paternal and sibling samples as an internal control for the different polymorphic sites will optimise the test and decrease the risks of erroneous results due to the presence of fetal DNA circulating in the maternal blood from previous pregnancies. The accuracy of fetal genotyping could also be improved if the maternal and paternal genotypes and phenotypes are known and are predetermined, and if specific primers are designed for the RhDCcEe genes of the tested couple or to fit the common genes in the studied population.

In the studies that used FfDNA from maternal plasma, fetal RhE/e or RhC/c genotyping was demonstrated with 100% accuracy as early as the first trimester and the same accuracy was maintained in the second and third trimester. From these results, gestational age was not shown to be a limiting factor and accurate diagnosis can be achieved whenever the test is clinically indicated.

Our study is limited by the small numbers of women and samples that have been reported in only a few case series articles in the medical literature describing fetal RhC/c and RhE/e determination from maternal blood. Another limitation of the reports is that fetal RhCE determination was performed by using multiple and repeat tests for the same woman. In statistical terms, it is limiting the analysis as the results of multiple samples of the same women are considered as a ‘non-independent observation’ and the accuracies must be reported as percentage of the correct diagnosis, unable to report the sensitivity and specificity of the test. The same limitation of small numbers of women and tests that were performed on multiple samples from the same woman (e.g. multiple events per person) led us to use Poisson calculations of the percentage of correct testing and the ‘random effects’ model assuming heterogeneity among the selected studies.

The limitation of the studies reviewed for this meta-analysis is recognised when assessing the articles against STARD guidelines. First, there is no reference standard testing for noninvasive fetal RhCE genotyping, In addition, the articles did not always report inconclusive results, the reports include repeat tests using the same sample, did not report the characteristics of the population studied and did not report estimates of statistical uncertainty. The published reports may have described their accurate results, biased towards the potential of fetal RhCE genotyping. Risk of bias may also be present in most studies reviewed as tests were performed on selective population. The ethnic origin of the woman could play an important role in understanding polymorphism of the Rh test’s results and its interpretation. It is, therefore, important to know the prevalence of the population from where the samples are drawn. For example, C has a frequency of 70% and c a frequency of 80% in Europeans, whereas in Africans the frequency of c is much higher (99%) and the frequency of C much lower (17%).25

However, some of the articles were reported before the publication of the STARD checklist and all of them reported on an innovative test that needs special evaluation as reflected in a suggested specific proforma (checklist) to evaluate the articles describing noninvasive fetal Rh genotyping (K. Freeman et al., submitted). In addition, a larger study is needed for this test to become qualified as a diagnostic standard test.

Despite the above limitations, and while recognising the ethnic diversity, the polymorphisms complexity of the Rh group and the risks resulting from false-positive or false-negative results, we suggest using the promising results of this article as the basis for a large-scale study on fetal Rh genotyping from maternal blood. Only perfection of the method in different ethnic groups and in the presence alloimmunisation will enable a safe incorporation of this test into routine prenatal care.

Acknowledgement

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References

We thank Traci Hutchins and Janet Ober for their help in preparation of this manuscript and continued support of our research.

References

  1. Top of page
  2. Abstract
  3. Background
  4. Search strategy
  5. Selection criteria
  6. Data collection and analysis
  7. Results
  8. Discussion
  9. Acknowledgement
  10. References