Non-invasive prenatal paternity testing, using autosomal STR or SNP markers.

(1)

Non-invasive prenatal paternity

testing, using autosomal STR or SNP

markers.

Ingemarie van Gilse (11388439) Master Forensic Science

Supervisor: prof. dr. A.D. (Ate) Kloosterman Co-assessor: mw. dr. P.J. (Pernette) Verschure Date: 2-9-2018

(2)

Abstract

Each year, sexual assaults result in 100.000 victims in the Netherlands. After raping, 7% of the women get pregnant. Sometimes, it is not clear who the father is and a paternity test is wanted. Paternity testing is currently mostly performed with invasive techniques like AC and CVS. This literature thesis intends to investigate whether non-invasive prenatal paternity testing is possible. The following research question is studied: “Is non-invasive prenatal paternity testing possible with the use of autosomal STR or SNP markers?”.

Invasive techniques like AC and CVS are currently mostly used for prenatal paternity testing. Even though the procedural risk of the techniques is between 0,5-1%, a lot of

physical and emotional stress of the pregnant women can be reduced when a NIPAT would be available. Since 1997 are studies about cffDNA published, which led to developing NIPT for medical usage, such as aneuploidy screening. NIPAT also would make use of cffDNA. STRs and SNPs are often used as genetic markers and are also used for NIPAT. STR-based NIPAT seemed promising, but a lot of challenges are faced like low average fetal fraction, high fragmentation of cffDNA and spontaneous mutations. Some studies

recommend to replace or combine STRs with SNP markers. Since cffDNA is fragmented and SNPs have a shorter variation, this marker is preferred over STR markers. Also the mutation rate of SNPs is lower than for STR markers.

NIPAT is a rather new technique and more research should be done, especially since a lot of articles were proof-of-concept studies. For instance, first trimester sampling,

accuracy in clinical trials, more SNPs and various populations should still be studied. Nevertheless, NIPAT with autosomal STR and/or SNP markers seem very promising and hopefully become available soon for medical use and forensics.

Keywords: non-invasive prenatal paternity testing, paternity testing, single nucleotide polymorphism, short tandem repeats, chorionic villus sampling, amniocentesis, NIPT

(3)

1. Introduction

In the Netherlands, 8100 woman were sexually assaulted in 2016 [1]. These are assaults registered by the “Centraal Bureau voor Statistiek” (Central Bureau for Statistics), but a lot of assaults are not reported. A fact sheet from ‘Centrum Seksueel Geweld’ (Centre Sexual Violence) in 2013 shows much higher numbers, namely 100.000 victims of sexual assaults each year, of which 90% of the victims are women. Also, it is stated that 1 out of 8 women have been raped in their life. After being raped, 7% of these women get pregnant [2].

Unfortunately, it is not always clear who impregnated the woman, because it is possible that the woman was trying to get pregnant from her partner or had intercourse with other men. This can lead to lots of stress for the mother, because she has to decide if she wants to keep the baby. Knowing whether the child is from her partner or from the rapist will have a great influence on this decision. A paternity test could be the answer to this problem. Most frequently, paternity testing is done by chorionic villus sampling (CVS) or amniocentesis (AC) after the 13th week of pregnancy [3, 4]. The problem with these tests are, that they are invasive, could lead to pregnancy loss and can only be performed after 13 weeks of

pregnancy [3, 4]. Postnatal paternity tests are also available, but these miss the main purpose of knowing who the father of the child is before birth. Nevertheless, some elements from postnatal paternity testing can be used for prenatal paternity tests, such as which markers should be targeted.

A real improvement would be a non-invasive prenatal paternity test (NIPAT).

Currently there are non-invasive techniques, like the Non-Invasive Prenatal Test (NIPT), that can be performed before 13 weeks of pregnancy for aneuploidy screening (e.g. Down

syndrome) in cell-free fetal DNA (cffDNA) in maternal blood [5]. NIPTs can determine the gender of the fetus and can amplify the chromosomal STR alleles of male fetuses [4]. Y-chromosomal STRs are routinely used for DNA analysis and invasive paternity testing, as well as autosomal STRs, so it is possible that these STRs also can be used for NIPAT [4]. However, Y-STRs are only applicable to male fetuses. Therefore, NIPAT should make use of autosomal markers (autosomal STRs and SNPs) to reveal the genotypes of both male and female fetuses.

This literature thesis will elaborate on the currently used techniques and new techniques for paternity testing. First the postnatal and invasive techniques will be explained. Thereafter, newer techniques and studies into NIPAT will be discussed. The main aim of this literature thesis is to investigate the following research question: “Is non-invasive prenatal paternity

(5)

2. Current paternity testing

As mentioned in chapter 1, the currently used prenatal paternity tests are invasive, like CVS and AC. Also, postnatal paternity tests are available. The markers used in postnatal paternity tests and invasive tests, can possibly be adopted in a non-invasive test. This chapter will discuss how current paternity testing is performed and what the risks are of these invasive techniques.

2.1 Postnatal techniques

Postnatal paternity testing is already available. For these tests, common STRs are used to determine the paternity. Because STRs are short repeated regions that vary a lot between individuals, these markers are effectively used for identification purposes [6]. The more markers (STRs or SNPs) are compared, the higher the paternity index (PI) will be when the markers match with each other. The paternity index is calculated by dividing the probability that the alleged father is the biological father by the probability that the genetic results are found when a random man is the biological father [7,8]. The product of PI is the combined paternity index (CPI) [7]. The probabilities depends on the allele frequencies [9].

Short tandem repeat analysis can handle forensic samples that are of low quantity and of poor quality [9]. There are various types of STR systems: di-, tri-, tetra-, penta- and hexanucleotide repeats. Tetra repeats are the most popular, because they form only little stutter products compared to the other nucleotide repeats [9]. Since NIPAT deals with low quantity samples of the fetus with a high risk of stutter, STRs (preferably tetra repeats) would be a good target for NIPAT.

In a study by El-Alfy (2012), the paternity was determined post-natal. Blood samples were taken from the mother, child and alleged father. They used a multiplex system of 15 STR loci and applied these for paternity testing. The PI is calculated with the allele frequencies and measures the strength of a genetic match between child and the alleged father. When a low frequency allele is shared between parent and child, this will contribute to a high PI. Also the probability of paternity was calculated. To confirm if someone is the biological parent of the child, the probability of parentage should be at least 99%. There is legal proof of paternity when the probability of a biological relationship is 99 % or higher [9].

Also in forensic cases, such as sexual assaults, postnatal paternity tests are used to determine the paternity of the rapist. DNA is extracted from the aborted fetus or born child and the STR profiles are compared between the mother and suspect(s). If the fetus was a boy, also Y-STR markers can be analyzed [10]. Since the victim has to decide if she wants to keep the baby, while she does not know who the father is, it is important that a NIPAT becomes available. This can reduce the emotional and physical stress on the woman. Also, suppose that the aborted child was not from the rapist, this will be another painful moment for the victim.

2.2 Invasive techniques

Two invasive techniques that are commonly used for paternity testing are CVS and AC [10]. These invasive tests are usually taken by 5- 10% of the pregnant women [11]. When women have a higher risk for trisomy 21, there is an exponential increase of the invasive testing rate. Less than 1% of the women take an invasive test when the risk of trisomy 21 is 1 in 10,000, while 95% of the women take the test when the risk is 1 in 50 [12]. How often invasive tests are used for paternity tests, was not found in the literature.

In the Netherlands, prenatal screening tests are done for aneuploidy, such as Down, Edwards and Patau syndrome. There are non-invasive prenatal techniques, such as blood tests and nuchal scans, and since 2017, the NIPT test is available. When an abnormal result is found, the pregnant women can still choose to do an invasive prenatal test. Since 2011, the participation to these non-invasive tests is slightly increased, from 24% in 2011 to 33,7% in 2015 [13]. NIPT has a higher accuracy than CVS or AC and might therefore also be a more

(6)

accurate test for paternity testing [13]. Also, the usage of NIPT for paternity testing will possibly increase, because it is an non-invasive technique and women are more likely to use this test. In this section is the use of CVS and AC in the medical field further explained, because there is more knowledge about their medical use than applications for paternity testing. However, this medical knowledge is applicable to paternity testing with AC or CVS, because it is performed in the same way. Only the information gathered from the application is used differently.

In the 1970s, AC was established and in the 1980s CVS followed. The main purpose of these techniques was to observe aneuploidy, like Down syndrome, in cases where the family history contains specific gene mutations that might cause health problems. They can also be used for DNA analysis such as paternity testing [11,14]. AC makes use of the amniotic fluid to obtain fetal cells and CVS extract cells from

the placenta, as seen in figure 1. CVS is subdivided into transcervical (TC) and transabdominal (TA) CVS. TC-CVS is done with inserting a catheter through the cervix into the placenta. For TA-CVS, a needle is inserted through the abdomen and uterus into the placenta. Both techniques are shown in figure 2. It depends on the location of the placenta what technique is used [14]. CVS makes mostly use of polymerase chain reaction (PCR) and STR markers[15]. Also, fetal cells collected from CVS can earlier be analyzed than the cells from AC.

Figure 1: The difference in collecting fetal cells between AC and CVS is shown. AC extracts amniotic fluid, while CVS extracts tissue from the chorionic villi of the placenta. [16]

(7)

The risk of pregnancy loss by CVS and AC was always suggested to be 1%. However, different studies have shown that there are inconsistencies in the data. For example, a systematic review by Mujezinovic and Alfirevic (2007) [11] shows that there are inconsistencies in the risk data for pregnancy losses and complication rates. These inconsistencies are found in both AC and CVS procedures. Measurements were taken 14 days, 30 days and 24 weeks after the procedure. Only homogeneity was found for AC after 14 days with a pregnancy loss rate of 0.6%, with a 95% confidence interval of 0.5-0.7%. The other measurements showed heterogeneity for CVS and AC. The only correlation that was found was the positive correlation between the rate of fetal loss and period between procedure and follow-up. Within 14 days of the procedure this rate is 0.6%, while for total pregnancy loss this rate is 1.9%. The background risk will be higher with CVS than AC, since CVS is performed earlier and the risk of spontaneous pregnancy loss is higher [11].

Another study by Niederstrasser et al. (2017) [14] states that invasive prenatal testing results in a higher risk of pregnancy loss than the absolute background risk for spontaneous miscarriage. In a large Danish national register-based cohort study, the fetal loss for AC was determined to be 1.4% and 1.9% for CVS. But, AC and CVS have a different timing of the procedures. When AC is performed prior to 15 weeks, this is associated with higher fetal loss rates. Therefore, AC is recommended after 15 weeks of pregnancy. CVS,

however, can be performed after 10 weeks of pregnancy. The advantage of CVS is therefore that it can be performed earlier in the pregnancy, so decisions about the continuation of the pregnancy can be made earlier [14]. The total fetal loss rates of TA-CVS, TC-CVS and AC were analyzed. It was concluded that the rates of TA-CVS and AC were comparable and that TC-CVS is related to higher pregnancy loss risks. The total fetal loss rate, so procedure and background risk together, measured in this cohort study was 1.73% for TA-CVS, 2.01% for TC-CVS and 1.18% for AC [14]. Taking into account the earlier mentioned background risks for fetal loss (1.4% AC and 1.9% CVS), the results from this study assume that the procedure-related risks for AC and CVS are lower than the accepted 0.5-1% in the literature. The conclusion of this study is that TA-CVS, TC-CVS and AC are all supported as equal methods for invasive prenatal testing [14].

In a study by Kollmann et al. (2013), higher risks of pregnancy loss were shown due to procedure related complications of AC and CVS. The study determined fetal loss for AC at 0.75%, TA-CVS at 2% and TC-CVS at 3.13% [18].

Lastly, a systematic review of Akolekar et al. (2015) [19] estimated the procedure-related risks of miscarriage following AC and CVS. This study also states that there is a lot of inconsistency in the complication rates and that they are probably lower than currently thought. It is suggested that AC has a fetal loss of 0.1% and CVS of 0.2% with data combined from several studies [19].

What should be taken into account when the miscarriage rate is interpreted, is the background risk. It is hard to determine an exact risk for procedure-related miscarriage, because the tested cohort can have a higher background risk than the general population [14]. This means that the women from the cohort could already have a higher risk to miscarriage, before the procedure had taken place, than the general population. Women who undergo CVS will have

Figure 2: The difference in collecting fetal cells between transabdominal and transcervical CVS is shown. TA-CVS is shown in the top picture and TC-CVS is shown in the lower picture. [17]

(8)

a higher background risk, because it is performed in an earlier stage than AC [11]. As said before, AC can be performed after 15 weeks of pregnancy, while CVS can be performed after 10 weeks [14]. Because AC is performed in a later stage of the pregnancy, when there is a lower risk of spontaneous miscarriage, these women will have a lower background risk. The higher miscarriage rates noticed in women that undergo CVS is probably influenced by this higher background risk. The CVS studies are mostly done with TC-CVS; less data is found for TA-CVS pregnancy losses. The miscarriage risks of CVS after 10 weeks are comparable with AC after 15 weeks [11].

Between the different articles, a wide range of pregnancy loss rates is observed. Despite the fact that there is no homogenous result, it can be concluded that most articles state that invasive prenatal testing has a higher risk of fetal loss than the background risk for spontaneous pregnancy loss. This is also applicable to invasive prenatal paternity testing, because it is the same invasive procedure. Next to the higher risk, the invasive procedure itself will also cause more emotional and physical stress for the mother. Therefore, it is important that a non-invasive prenatal paternity test becomes available.

2.3 Overview invasive techniques

In table 1 are the different techniques found and their risk of pregnancy loss. It differed between the articles if they mentioned the background, procedure or combined risk of the invasive technique. Overall, the procedure risks seem not higher than the accepted 0.5-1% mentioned in the literature [14].

Table 1: Article overview of pregnancy loss risk. There is mentioned which technique was used and what risk of pregnancy loss was measured.

Article Technique Background/Procedure/Combined Risk (%)

Mujezinovic (2007) [11]

CVS & AC After 14 days: 0.6% (CVS), 0.6% (AC)

Niederstrasser et al. (2017) [14] TA-CVS, TC-CVS & AC Background: 1.9% (CVS), 1.4% (AC) Combined: 1.73% (TA-CVS) 2.01% (TC-CVS), 1.18% (AC) Kollmann et al. (2013)[18] TA-CVS, TC-CVS & AC 2% (TA-CVS), 3.13% (TC-CVS), 0.75% (AC) Akolekar et al. (2015) [19]

(9)

3. Non-invasive prenatal paternity testing

Currently, prenatal tests are performed on fetuses to screen for health problems and risks for certain genetic diseases. An upcoming test is NIPT, which probably will reduce the practice of invasive prenatal testing, such as AC and CVS [19]. Because this test is performed non-invasively, there is no higher risk for miscarriage. Since NIPT is already possible, it seems that NIPAT would be possible as well. Autosomal, mitochondrial or Y-chromosomal DNA markers are used for DNA typing postnatally [9]. These markers combined with NIPT seem promising components for NIPAT. As mentioned in the introduction, this thesis is focused on NIPAT with autosomal STRs and SNPs as markers and will be further discussed in this chapter.

3.1 Start of non-invasive testing

As discussed in chapter 2, even though the risks for mother and fetus are not that high, CVS and AC are associated with pregnancy loss[20]. The fact that there is even a small risk, is a problem for many women, as well as the invasive procedure that the women should undergo. NIPAT can be a solution to avoid the added risk.

Since 1997, reports are published of cffDNA analysis in the maternal blood as a technique for non-invasive prenatal genetic diagnosis[4,7,21]. NIPT is a technique that makes use of cffDNA as shown in figure 3. CffDNA has a higher concentration in the maternal plasma compared to fetal cells. Also, cffDNA can be isolated with easier methods and be observed after the 6th_{week of}

pregnancy [21]. Several studies have validated this technique in prenatal diagnostics, such as for determining the fetal sex, aneuploidies and rhesus typing, but only a few studies mentioned NIPAT developments[4,20,21]. Down syndrome and other chromosomal anomalies can be detected in fetuses by amplifying and sequencing cffDNA fragments. Studies have compared if the detection rate of Down syndrome is different between cffDNA testing and CVS or AC. Pregnant women with a high risk for chromosomal abnormalities were used for these studies. The detection rate for Down syndrome using NIPT was 100% and the false positive rate was less than 1% [20]. Since 2014, the NIPT test is available in the Netherlands for pregnant women with a higher risk for chromosomal abnormalities. Although NIPT is highly reliable, invasive techniques are still used to provide extra certainty [13].

Although NIPT is a good advancement in determining the gender of the fetus and aneuploidies, the main question remains whether it can be used as a NIPAT [4]. A start for NIPAT was made in a study by Wagner et al. (2009) [4]. They used the same methods as for NIPT, namely extracting DNA from the maternal plasma. Multiplex PCR was used and adjusted for amplification of low amounts of DNA. Two types of amplification were used. The first amplification showed that the maternal DNA suppresses the autosomal fetal alleles, which is a common complication in the examination of low-copy DNA samples. Only the locus

Figure 3: In the mother’s bloodstream is cfDNA found from the placenta. NIPT makes use of maternal blood to obtain cffDNA to analyze for e.g. aneuploidies. [22]

(10)

amelogin that reveals the fetal gender is not suppressed. The second amplification obtained between 6 and 16 fetal loci, but where only from male fetuses. These loci matched the alleles of their presumed father’s. The tested paternity was herewith affirmed [4].

Thus, a first start for NIPAT was made by Wagner et al. (2009) [4]. The drawback of this study was that the fetal alleles were suppressed by the maternal alleles. If fetal alleles could be identified and less stutter would be ensured, cffDNA could be a good option for NIPAT. The challenge is to find a technique that is suitable and reliable for both male and female fetuses [4]. However, this study was only performed on male fetuses. The next sections will discuss the new studies and discoveries of non-invasive prenatal paternity testing.

3.2 Short tandem repeats

Invasive techniques (CVS and AC) use STRs for paternity testing [23]. Since STRs can deal with low quantity samples, it is a logical step to use STRs as markers for NIPAT. Screening for fetal aneuploidies with analysis of cffDNA has recently become available with NIPT. NIPT is performed by analyzing the cffDNA with massive parallel sequencing (MPS) and can be performed at the 9th week of pregnancy [24]. MPS can be used for SNP and Y-STR typing for NIPAT[24,25]. This chapter will elaborate on NIPAT using Y-STR markers. Several studies have been performed on NIPAT using STRs. Different techniques are used and two recent studies will be discussed. A first example is a study by Gysi et al. (2015) [25] where the researchers used a microarray assay to analyze the STRs [25]. This article set up a pilot study using a next-generation sequencing (NGS) assay that can analyze cffDNA from maternal blood. Blood samples from two pregnant women were collected to analyze the cell-free DNA (cfDNA) in the blood plasma. The cfDNA profile should be interpreted as a mother-child mixture, where the mother is the major contributor (20-50 fold excess) [25]. The mother’s profile masks every maternal allele of the fetus, so the minor contributor shows the paternal alleles of the fetus. The paternal alleles are detected when they are not masked by stutter or maternal alleles. Also mini-STR typing with NGS is possible, that has as the advantage of detecting more fetal alleles. The paternity probabilities were estimated for the two fetuses, which gave a probability of >99.9997% with 9 markers for one fetus and 99.9999997% for the other with 12 markers[25]. Some markers from the father were masked, if they were included, the paternity probabilities would be higher. This pilot study detected a high number of paternal alleles and a low number of dropouts, which illustrates the potential of this technique for paternity testing in forensics [25]. However, this study by Gysi et al. (2015) [25] is preliminary and more studies should be done to acquire more accuracy by comparing prenatal and postnatal STR typing for the true fetal genotypes [25]. Also, future studies are needed to validate the technique, measure drop-out probabilities, stutter ratios and signal-to-noise ratios [25].

A second study is done by Whittle et al.(2017) [24], who based their work on a previous pilot study to evaluate STRs for NIPAT [24]. The participants in this study were informed that the NIPAT was not reliable when there existed consanguinity between the parents or when the true father was the brother of the alleged father. After the 9th week of pregnancy, 20 milliliter venous blood was obtained from women with a singleton pregnancy. When the article was

(11)

Normally, absence of expected paternal alleles is associated with allele drop-out. But it can also strengthen the exclusion of paternity [24].

The study suggested a pre-measurement of the fetal fraction in the blood plasma by measuring fetal-specific markers. This pre-measurement could be performed before massive parallel sequencing (MPS) in cases where small amounts of maternal blood are used, because MPS is expensive and labor-intensive. In this study, it was not necessary, since 20 milliliter maternal blood was required [24]. The fetal fraction of male fetuses can be evaluated by determining the Y-chromosomal markers in the plasma [24].

Whittle et al. (2017) studied STR-based and showed some promising results. However, there are some reservations about the use of STRs for paternity testing, because stutter results in information loss. Moreover, using this technology, STRs are insufficiently sequenced when compared to non-repeat DNA. The researchers suggest that analyzing single nucleotide polymorphisms (SNPs) with MPS could improve the current STR testing [24]. Although STRs seem to be a useful marker, several researches have also reported that SNPs have great potential for NIPAT. But to use SNPs for non-invasive prenatal paternity testing, a lot of loci need to be analyzed [24]. SNP markers for NIPAT are further discussed in chapter 3.4.

3.3 Complications STRs

STR loci or microsatellites have a high power of discrimination and are therefore often used as markers in forensic casework [26]. For paternity testing with male offspring, STR loci from the Y chromosome are useful, since fathers pass these on to their sons[26]. For female offspring, Y-STR testing is not possible and autosomal STRs should be used. But as mentioned in the previous section, many complications are faced with NIPAT using STRs. For example, low average fetal fraction and high fragmentation of cffDNA, but also spontaneous mutations can lead to false paternity exclusions[24,25]. In this chapter those complications are further discussed.

The first complication is the average fetal fraction of all circulating DNA. Around the 9th week of the pregnancy, this fraction is 10%. Since the circulating DNA is a mixture from the mother and fetus, there is a large amount of maternal cfDNA[24]. Combined with a high amount of stutter, it is hard to purify the fetal cfDNA and to align the STR loci with MPS[24]. Some loci will not be revealed due to high stutter amounts, because the fragments are too small (between 145 and 166 base pairs) and cannot be differentiated from the stutter[24]. Furthermore, the maternally inherited alleles of the fetus can usually not be assigned because the alleles in the plasma are a mixture from mother and fetus. Therefore, the mother’s profile will mask all maternal inherited alleles of the fetus [25].

Another challenge is formed by spontaneous mutations. In paternity testing, the similarities or differences in genetic marker loci between the offspring and the presumed father are analyzed. If there are differences in loci, this is associated with exclusion of biological paternity. Nevertheless, it is possible that spontaneous mutations lead to a false exclusion. The mutation rate of a locus is correlated with the degree of polymorphism. If the mutation rate is high, there is more variability in the locus. These loci are interesting for forensic purposes, because of the high discrimination power[26]. Mutations ensure that polymorphic STR loci evolve and, therefore, knowledge about the mutation rates at each locus is required to interpret the genetic profiles. These different mutation rates for autosomal STR loci rely on sequence-specific parameters [26].

In a study by Kayser and Sajantila (2001)[26], a first effort was made to study the mutation rate of some Y-STR loci used in forensics. One of the conclusions was that, on average, an Y-STR mutation occurs for up to three out of every thousand father/son pairs [26]. The mutation rates of different autosomal STRs are studied in humans, fruit flies and yeast. Mutation rates seem to be dependent on the length of the repetitive array, the longer the array, the higher the mutation rate [26]. This correlation is not found for the Y-STRs, probably assigned to the comparably small number of mutations [26]. In data of human STRs, it is

(12)

demonstrated that when sustained homogeneous repeats are lengthened, the mutation rate also increases. Because every locus has specific molecular features, this results in different mutation rates. It is also observed that there could be allele-specific mutation rates, but this should be further investigated to determine the forensic implications of this phenomenon [26].

Non-paternity is currently determined when more than two out of 6-15 STR loci differ between the alleged father and child. When there are only one or two different loci, this is assigned to mutation. In a study by Kayser and Sajantila (2001)[26], 415 biologically related father and son pairs were analyzed and the relevance of mutations in Y-STR loci was investigated. It was calculated that two mutations within one transmission of germlines are not statistically unexpected and this was also found in one father and son pair. In this study, the average mutation rate for Y-chromosomal and autosomal STRs was 2.88x10-3_{[26]. This rate}

affects the determination between mutations and exclusions in paternity testing. Due to the mutation rate and these observed mutations in a father and son pair, the exclusion criteria in paternity testing should be changed to a minimum of three Y-STR loci with a sufficient number of Y-STR loci during analysis. The study by Kayser and Sajantila (2001) recommended at least nine Y-STRs to be analyzed. This also applies to autosomal STR paternity testing, since the mutation rates and features of autosomal and Y-STRs are comparable. Nevertheless, the specific characteristics of alleles and loci should be studied before they are used for paternity testing [26].

When this study was published, there was no clear protocol by the DNA commission of the International Society for Forensic Genetics (ISFG) on mutation consideration for paternity and forensic examinations using STRs. Nevertheless, the exclusion criteria with a minimum of three loci was also recommended by the German Association of Expert Witnesses for Paternity Testing [26].

Another optimization point for STR paternity testing is the way of testing. Because NGS is becoming more commercially available, certain challenges of the fragmented and low amount of DNA can be avoided. The study of Gysi et al. (2015) already used NGS for paternity testing with STRs. The use of mini-STRs is also enabled, since size separation of STRs is no longer necessary and fetal peaks that are masked by stutter or maternal alleles can be revealed [25]. Since STRs have many complications, researchers have looked for other markers that can be used for paternity testing, such as SNP markers. SNPs have a lower mutation rate than STRs. Some studies already recommended to replace or combined SNPs with STR loci. It is too early to use Y-SNP data, because it is currently collected from numerous populations and has a high population specificity [26]. The next section will elaborate on SNP-based NIPAT.

3.4 Single Nucleotide Polymorphisms

The development of molecular analysis of cffDNA has made it possible to determine the sex of the fetus, the rhesus D blood group, aneuploidy identification and mutations derived from the parents[23]. In forensics, STRs are already well established and have large databases. Therefore, STRs are preferred as markers over SNPs[25]. Nevertheless, it is hard to find enough useful STR loci, because the cffDNA only has a size of <150 bp. Mini-STR analysis was suggested as an alternative. However, the analysis of SNPs has a shorter variation and

(13)

Moreover, a mathematical model was developed for the SNP-based NIPAT test. The algorithm, based on Bayes’ theorem, calculated a PI based on the effective SNPs that resulted from the sequence data of the maternal plasma[7,21]. Effective SNPs are homozygous SNPs in the mother, which means you can determine which allele is inherited from the father. These alleles are the minor contributor to the maternal plasma. The minor allele frequency (MAF) was evaluated for its influence on the NIPAT. The binomial distribution probability model that was used, gave a positive correlation between MAF, the total SNP number and effective SNP number. High frequency (HF) SNPs (MAF >0.3) are suitable when the sequencing depth is not deep enough, in contrast to low frequency (LF) SNPs (MAF <0.3). Previous studies demonstrate the same results. Therefore, NIPAT uses HF SNPs[7].

The paternity of 17 cases were determined by calculating the combined paternity index (CPI). This was done by combining bioinformatics and deep targeted sequencing of selective SNPs. Elements that could affect CPI calculated by NIPAT were evaluated, such as the effect of the fetal fraction. A significant positive correlation was shown between the fetal fraction and calculated CPI. When the fetal fraction is changed from 1% to 10%, it demonstrates a significant effect on the CPI, while above 10% only a slight effect was seen. Secondly the effect of sequencing depth was evaluated with a fetal fraction of 10%. Changing the sequencing depth from 10X to 75X shows a considerable improvement of the CPI. When the sequencing depth was further increased, the CPI improved only little. Increasing the sequencing depth over 200X showed no improvement at all of the calculated CPI. It was observed that, when the sequencing depth was below 30X, paternity determination was incorrect and the calculated CPI was lower than zero, regardless of the number of effective SNPs. All effects result in the following conditions for SNP-based NIPAT: the number of effective SNPs should be between 1*103_~2*103_{, MAF of the SNPs should be larger than 0.3,}

sequencing depth of 75x and the fetal fraction should be over 3%. When the fetal fraction is below 3%, the recommended sequencing depth is >125X or the number of effective SNPs should be increased to >1*105_[7].

Next to NIPAT, all seventeen women also underwent amniocentesis around the 20th

week. A bioinformatics method was used to verify the results of paternity determination in NIPAT and compared these results with invasive paternity tests. The calculated CPI for the biological father was 5.51*103_{with an error rate of 2.29‰ and the calculated CPI for unrelated}

males was -6.31*104 _{with an error rate of 41.55‰. The number of effective SNPs were,}

surprisingly, the same in the SNP-based NIPAT method as with AC, respectively 1.756*104

SNPs and 1.97*104_SNPs[7].

A limitations of this study was, for example, that all samples were taken in the second trimester (12-20th week), where the fetal fraction is relatively high (5.69%~29.74%). There is recommended that first trimester sampling with lower fetal fraction should be studied as well in the future. Although it is hard with a low fetal fraction (<3.5%) to determine paternity, NIPAT usage in the first trimester is strived for. Moreover, other populations should be studied, because this study was limited to a Chinese population. Lastly, this method should be tested in a large-scale study with clinical samples to assess the accuracy[7].

The second study of a SNP-based method that is discussed is a study by Yang et al.(2018). They studied massively parallel sequencing (MPS) as a method to analyze cffDNA in the maternal plasma. With this technique, fetal and maternal DNA can be distinguished from the maternal plasma. Evaluation of the procedure showed validation of the reliability and therefore opportunities for non-invasive prenatal paternity testing (NIPAT)[23]. The cfDNA samples of 20 women were examined for paternally inherited alleles, including 11 samples from first-trimester pregnancies (9 to 12 weeks)[23].

This study mentioned a large SNP analysis by Ryan et al. (2012) with impressive results. Yet, the cffDNA data interpretation was complicated by the low signal-to-noise values and the restricted number of probes of the microarray [27]. In a previous study from the same research group, inconclusive results were found in a large sample analysis. This demonstrates that maternal or fetal loci do not methylate completely or restriction enzymes do not fully digest the unmethylated DNA pieces[23].

(14)

Ion PGM technology was used as NGS technology to sequence the paternally inherited alleles in the maternal plasma. When this technology is combined with technical advancements and optimized data interpretation of the MPS results, this could be applied in non-invasive fetal genotyping. As a result of the short amplicons, this amplicon-based strategy has an advantage in evaluating fragmented cfDNA samples compared to assays of STRs. Moreover, adjustable SNPs can be used for investigation, only the number of SNPs is established. Also, the costs are lower compared to STR assays, causing a higher coverage [23].

A first effort has been made by Yang et al. (2018) to investigate first trimester sampling. Nonetheless, this amplicon-based NGS strategy for SNP-based NIPAT can profit from further developments. For example, the SNPs with a low coverage, heterozygote imbalance or high background noise need an optimized protocol. Also, to obtain more reliable results, the sequencing technique should be improved to limit errors and obtain sufficient cffDNA. Future studies should be expanded, such as higher sample sizes of different gestations during pregnancies, to gain better knowledge about the sequencing noise and alleles inherited from the father. Further improvements should consider stochastic effects and define a success rate prediction model to create an algorithm for the MPS results and paternity determination performance [23].

A third study by Qu et al. (2018) studied 1795 unlinked polymorphic SNPs (with MAF≥0.30 in the East Asian population), which were chosen from the previous study by Jiang et al. (2016) [7,28]. The prenatal paternity test was performed with MPS in 11 first-trimester and 23 second-trimester pregnancy cases. The mathematical model from Jiang et al. (2016) was used for paternity determination [7]. This model was in accordance with the Bayesian model, which is recommended most highly, where the PI is an LR. For the PI calculation, the SNP mutation rates were not considered, since these are very small for SNPs (namely 10-8_{) [28]. The CPI}

was determined by multiplying all separate PI values of the loci, since they were considered to be unlinked. In this study, a significant separation is found between the CPI values of the biological father and unrelated males, even in cases with low fetal fraction. A minimum mean sequencing depth of 75x was set and all samples met this requirement[28]. Two methods were used to interpret the sequencing data for NIPAT, namely identification of paternal alleles with the counting method and the mathematical algorithms for PI calculation[28].

MPS-based technology was used for NIPAT, mainly due to the efficient handling in cfDNA samples. MPS is increasingly used in forensic science and the cost are also decreasing. Therefore, MPS seems a promising technique for practical NIPAT development. In the 34 cases, 30,82% of the actual paternally inherited alleles were below the allele fraction threshold cutoff of 2%, which led to an insufficient number of effective SNPs with the counting method. However, the maternally inherited homozygous SNPs were included in the PI calculation model as effective loci and led to sufficient loci for paternal identification. Drop-ins were found in 12 cases. With MPS, it is likely to generate some incorrect alleles during sequencing, which are considered to be stochastic effects[28]. Therefore, differentiation between paternal alleles and sequencing errors is difficult in maternal plasma with low fetal fractions.

Guidelines for interpretation of the data for paternity testing, suggests a CPI value of 10.000 or larger to support paternity. Exclusion of paternity is suggested when the CPI value

(15)

3.5 Overview non-invasive techniques

In table 2 is an overview shown of the non-invasive techniques discussed in this literature thesis.

Table 2: Overview of non-invasive techniques with STR and SNP markers. The advantages and complications are mentioned, as well as some recommendations for future research.

Article STR/ SNP

Advantages Disadvantages Future study

Gysi et al. (2015) [25]

STR High number of paternal alleles, low number of dropouts

Preliminary, low fetal contribution

Validate technique, dropout probabilities, stutter ratios, signal noise Whittle et al. (2017) [24] STR Sufficient informative loci identified Maternally inherited alleles cannot be assigned, stutter peaks smaller amounts of maternal blood, SNP analysis with MPS Jiang et al. (2016) [7]

SNP Short DNA fragments, vast number of SNPs across whole genome

Only second trimester and Chinese population used

First trimester (low fetal fraction), other

ethnicities, accuracy

Yang et al. (2018) [23]

SNP Short amplicons, high coverage, first & second trimester pregnancies

Low signal-to-noise Optimize protocol, larger sample size, success rate prediction model

Qu et al. (2018) [28]

SNP First & second

trimester pregnancies, Bayesian mathematical model Drop-ins, Chinese population Other ethnicities, improve testing efficiency

(16)

3.6 Costs NIPAT

Yang et al. (2018) [23] made an estimation of the running costs for NIPAT. They stated that the costs for the primer for 720 amplicons per sample were around $11. If a study has a design with 3000 amplicons, the costs for the primer per sample will be around $48. The manufacture time is approximately 15 working days. Per cfDNA sample, there are also costs for library preparation, PCR and MPS, which will be approximately $250. The samples can be sequenced simultaneously in 2 to 5 hours using for example the Ion PGM[23].

Oligonucleotide probes were used in previous studies to detect the paternally inherited alleles in maternal plasma. These probes were required for preparation of the library and costed more than $200 per sample. The manufacture time was between 30 and 60 working days. The time that is needed to manufacture and sequence the samples should also be taken into account when choosing a method[23].

Since NIPAT is not yet widely available, the market price is not found in the literature. Nevertheless, an estimation can be made when comparing with the NIPT test, since the method is based on the same principle and makes use of cffDNA. Therefore it will be likely that the costs for NIPAT will be around the same price range. In the US, a NIPT test costed around $800 to $2000 in 2013. Outside the US, the costs range from $500 to $1500. Another study by Song et al. (2013) states that the cost of a NIPT test are around $695 and $995[29]. Currently, the NIPAT test is not included in health insurances[30]. Also, other paternity tests are currently also not reimbursed by the health insurances, so it is not very likely that NIPAT will be reimbursed in the future. Heredity studies are only reimbursed when there is a disease or congenital abnormality in the family[31].

3.6 Statistics

Very little is mentioned in the articles about the statistical certainty of the NIPAT test. Mostly the number of the correctly identified biological fathers or correctly exclusion of paternity is mentioned. Also, paternity identification between the different techniques was compared. For example, in the study by Jiang et al. (2016) was mentioned that 17 out of the 17 cases were consistent between NIPAT and amniocentesis data[7]. Another study by Ryan et al. (2012) [27] reports that 99.95% of the SNP-based NIPAT correctly excluded paternity and 0.05% was uncertain[21]. Paternity determination in 21 pregnancies were performed of which 20 pregnancies were successful with a p-value <0.0001 and one undetermined [7].

Another study by Jiang et al. [7] showed that the probability of paternity exclusion can be used as a bio statistical standard for NIPAT, by calculating the CPI. Standards have been set up to assure accuracy of the NIPAT test. The paternity was correctly confirmed in all of the 17 tested cases. The alleged father could be determined as the biological father when the logarithm of CPI was larger than 4 and a p-value smaller than 10-4_{. There should be noted that}

only a Chinese population and second trimester pregnancies were tested. Also, only a small number of cases is used as a population for accuracy confirmation of this methods [21]. However, invasive techniques have a accuracy of 97,7% for CVS and 99,8% for AC. The lower accuracy of CVS is probably caused by the timing of the procedure, since CVS can be

(17)

4. Conclusion and recommendations

Chapter 1 discusses why non-invasive paternity testing can be a useful tool for determining who the biological father is. This is especially important in forensic cases, where a woman is raped and got pregnant. If the paternity test is non-invasive and there is no risks for miscarriage, no ethical issues would be involved. Furthermore, NIPAT can be used for medical applications such as detecting aneuploidies and other genetic anomalies. Thus, there is a great importance for research into non-invasive paternity testing.

4.1 Conclusions

The intention of this thesis is to investigate whether non-invasive prenatal paternity testing is possible. A literature overview was given of the current invasive paternity testing and how non-invasive techniques have been developed. The most recent studies were reviewed with two types of markers, namely STRs and SNPs.

Invasive techniques like AC and CVS are currently mostly used for prenatal paternity testing. Inconclusive results are found for the procedural risk of the techniques. Nevertheless, there can be concluded that AC and CVS have a slightly higher risk of miscarriage due to the procedure, adding 0.5-1% to the background risk. Even though the added risk is not that high, it still will add emotional and physical stress to the pregnant women, as well as the invasive procedure itself. Since studies about cffDNA were published, also techniques for non-invasive prenatal testing were developed, like the NIPT test. NIPAT makes use of the same principle and the first studies used STRs as markers for the test, since these were also used for the invasive techniques. However, researches of these studies recommended that STR markers were replaced or combined with SNPs in future studies, since STR-based NIPAT faced a lot of challenges, such as high fragmentation of cffDNA and spontaneous mutations. SNP markers can deal with these challenges, considering the short variation of SNPs and lower mutation rate than STR markers. Nevertheless, most SNP-based NIPAT studies are performed in a Chinese population, mostly second trimester pregnancies are used and the test should be validated in a large-scale study. In my opinion, these non-invasive techniques seem very promising, but further development is needed.

Considering the main findings of this literature review, the research question that was mentioned in chapter 1 is revisited: “Is non-invasive prenatal paternity testing possible

with the use of autosomal STR or SNP markers?”. Given the studies mentioned, especially

in chapter 3, there can be concluded that STR- or SNP-based NIPAT is possible. In my opinion, the preference lies with SNP markers, since these are shorter, widely available over the whole genome and have a lower mutation rate. Nevertheless, this field is rather new and, in my opinion, more research should be done before it can be applied in forensics. Possibly in a few years, NIPAT based on autosomal STR and/or SNP markers is validated and performed in a more various sample group, so it can be applied in forensic (and medical) practice.

4.2 Recommendations

The previous NIPAT studies provided some recommendations that I have applied to a forensic setting. This includes a larger sample size with various populations, more SNPs, first-trimester sampling and evaluation of the accuracy in clinical models. Moreover, recommendations on which NGS technique and which marker are preferred for NIPAT are mentioned below.

First, SNP-based NIPAT is recommended over STR-based NIPAT. Paternity testing with cffDNA is difficult when using STRs because cffDNA is fragmented and sufficient effective loci are hard to obtain. Since SNP- based NIPAT is suitable for short DNA fragments, it could give more reliable results. Also, SNPs are found through the whole genome, which ensures that sufficient loci can be obtained. Some difficulties for SNP-based NIPAT are the amount of maternal blood needed and a shortage of SNP databases. Therefore, expansion of SNP databases is recommended.

(18)

Most studies performed on NIPAT used Chinese populations and second-trimester samples. Allele frequencies are different in various populations, which can influence the results. Second trimester samples have a relatively high fetal fraction compared to first trimester samples. Therefore, future NIPAT studies should include different populations and first trimester samples. This can also reduce physical and emotional stress on the women, since she can earlier decide if she wants to terminate the pregnancy.

To evaluate the accuracy of NIPAT, a large-scale study should be performed in clinical trials. Also some influencing factors have not been evaluated yet, such as the allele frequency, minimum of SNPs/STRs and fetal fraction threshold.

Lastly, in order to improve the efficiency of NIPAT, it is recommended that the number of effective SNPs should be increased or that deeper sequencing is performed. An alternative method would be the analysis of multi-allelic SNPs, since their discrimination power is more robust than regular bi-allelic SNPs.

(19)

5. References:

[1] Centraal Bureau voor Statistiek. Geregistreerde criminaliteit; soort misdrijf, regio

http://statline.cbs.nl/Statweb/publication/?DM=SLNL&PA=83648ned&D1=0-4&D2=0,25,33,35,40-41,46,53,58,73-75&D3=0&D4=4-6&VW=T (accessed 22th of November 2017)

[2] september 2015 - Op basis van de meest recente wetenschappelijke onderzoeken heeft 'Centrum Seksueel Geweld' de cijfers in kaart gebracht over seksueel geweld in Nederland:

http://www.centrumseksueelgeweld.nl/wp-content/uploads/2013/09/FactsheetCSG_HR.pdf (accessed 22th of November 2017)

[3] Shulman, L. P., MURAM, D., & SPECK, P. M. (1992). Counseling sexual assault victims who become pregnant after the assault: benefits and limitations of first-trimester paternity determination. Journal of Interpersonal Violence, 7(2), 205-210.

[4] Wagner, J., Džijan, S., Marjanović, D., & Lauc, G. (2009). Non-invasive prenatal paternity testing from maternal blood. International journal of legal medicine, 123(1), 75-79.

[5] Yaron, Y. (2016). The implications of non‐invasive prenatal testing failures: a review of an under‐ discussed phenomenon. Prenatal diagnosis, 36(5), 391-396.

[6] Chen, D. P., Tseng, C. P., Tsai, S. H., Wang, M. C., Lu, S. C., Wu, T. L., ... & Sun, C. F. (2009). Use of X-linked short tandem repeats loci to confirm mutations in parentage caseworks. Clinica

Chimica Acta, 408(1-2), 29-33.

[7] Jiang, H., Xie, Y., Li, X., Ge, H., Deng, Y., Mu, H., ... & He, N. (2016). Noninvasive prenatal paternity testing (NIPAT) through maternal plasma DNA sequencing: a pilot study. PloS one, 11(9), e0159385.

[8] Drábek, J., & Cereda, G. (2014). Interpreting noninvasive prenatal paternity tests. GenetIcs in

medIcIne, 16(10), 793.

[9] El-Alfy, S. H., & El-Hafez, A. F. A. (2012). Paternity testing and forensic DNA typing by multiplex STR analysis using ABI PRISM 310 Genetic Analyzer. Journal of Genetic Engineering and

Biotechnology, 10(1), 101-112.

[10] Csete, K., Beer, Z., & Varga, T. (2005). Prenatal and newborn paternity testing with DNA analysis. Forensic science international, 147, S57-S60.

[11] Mujezinovic, F., & Alfirevic, Z. (2007). Procedure-related complications of amniocentesis and chorionic villous sampling: a systematic review. Obstetrics & Gynecology, 110(3), 687-694.

[12] Nicolaides, K. H., Chervenak, F. A., McCullough, L. B., Avgidou, K., & Papageorghiou, A. (2005). Evidence-based obstetric ethics and informed decision-making by pregnant women about invasive diagnosis after first-trimester assessment of risk for trisomy 21. American Journal of Obstetrics &

Gynecology, 193(2), 322-326.

[13] Volksgezondheid en zorg. Zorg rond de geboorte; cijfers & context; screeningen.

https://www.volksgezondheidenzorg.info/onderwerp/zorg-rond-de-geboorte/cijfers-context/screeningen#node-resultaten-prenatale-screening-down-edwards-en-patausyndroom (accessed 12-03-2018)

[14] Niederstrasser, S. L., Hammer, K., Möllers, M., Falkenberg, M. K., Schmidt, R., Steinhard, J., ... & Schmitz, R. (2017). Fetal loss following invasive prenatal testing: a comparison of transabdominal chorionic villus sampling, transcervical chorionic villus sampling and amniocentesis. Journal of

perinatal medicine, 45(2), 193-198.

[15] Jorge, P., Mota-Freitas, M. M., Santos, R., Silva, M. L., Soares, G., & Fortuna, A. M. (2014). A 26-Year Experience in Chorionic Villus Sampling Prenatal Genetic Diagnosis. Journal of clinical

(20)

[16] Qdos Ultrasound. CVS and Amniocentesis. https://www.qdosultrasound.com.au/pregnancy/cvs-amniocentesis-perth/ (accessed 19-08-2018)

[17] Summit medical group. Chrorionic villus sampling (genetic test) during pregnancy.

https://www.summitmedicalgroup.com/library/adult_health/obg_chorionic_villus_sampling/ (accessed 29-8-2018)

[18] Kollmann, M., Haeusler, M., Haas, J., Csapo, B., Lang, U., & Klaritsch, P. (2013). Procedure-related complications after genetic amniocentesis and chorionic villus sampling. Ultraschall in der

Medizin-European Journal of Ultrasound, 34(04), 345-348.

[19] Akolekar, R., Beta, J., Picciarelli, G., Ogilvie, C., & D'antonio, F. (2015). Procedure‐related risk of miscarriage following amniocentesis and chorionic villus sampling: a systematic review and meta‐ analysis. Ultrasound in Obstetrics & Gynecology, 45(1), 16-26.

[20] Langlois, S., Brock, J. A., Wilson, R. D., Audibert, F., Carroll, J., Cartier, L., ... & Okun, N. (2013). Current status in non-invasive prenatal detection of Down syndrome, trisomy 18, and trisomy 13 using cell-free DNA in maternal plasma. Journal of Obstetrics and Gynaecology Canada, 35(2), 177-181. [21] Zhang, S., Han, S., Zhang, M., & Wang, Y. (2018). Non-invasive prenatal paternity testing using cell-free fetal DNA from maternal plasma: DNA isolation and genetic marker studies. Legal Medicine. [22] Amy, E. and Thompson, MD. Noninvasive Prenatal Testing.

https://jamanetwork.com/journals/jama/fullarticle/2396480 (accessed 20-08-2018)

[23]Yang, D., Liang, H., Lin, S., Li, Q., Ma, X., Gao, J., ... & Ou, X. (2018). An SNP panel for the analysis of paternally inherited alleles in maternal plasma using ion Torrent PGM. International journal

of legal medicine, 132(2), 343-352.

[24] Whittle, M. R., Francischini, C. W., & Sumita, D. R. (2017). Routine implementation of noninvasive prenatal paternity testing with STRs. Forensic Science International: Genetics

Supplement Series.

[25] Gysi, M., Arora, N., Sulzer, A., Voegeli, P., & Kratzer, A. (2015). Non-invasive prenatal paternity testing with STRs: A pilot study. Forensic Science International: Genetics Supplement Series, 5, e291-e292.

[26] Kayser, M., & Sajantila, A. (2001). Mutations at Y-STR loci: implications for paternity testing and forensic analysis. Forensic Science International, 118(2-3), 116-121.

[27] Ryan, A., Baner, J., Demko, Z., Hill, M., Sigurjonsson, S., Baird, M. L., & Rabinowitz, M. (2012). Informatics-based, highly accurate, noninvasive prenatal paternity testing. Genetics in Medicine,

15(6), 473.

[28] Qu, N., Xie, Y., Li, H., Liang, H., Lin, S., Huang, E., ... & Ou, X. (2018). Noninvasive prenatal paternity testing using targeted massively parallel sequencing. Transfusion.

[29] Song, K., Musci, T. J., & Caughey, A. B. (2013). Clinical utility and cost of non-invasive prenatal testing with cfDNA analysis in high-risk women based on a US population. The journal of

(21)

7. Search strategy (appendix)

To collect the relevant literature for this study, several keywords and search engines have been used. Pubmed and Google were used, however Google Scholar was used the most. In table 3 is shown which keywords were used, according to the search engine. The literature was selected by reading the title and abstract. Also the reference list of the selected articles were searched for relevant articles. When a title seemed promising for relevant information, the abstract of the article was evaluated in order to determine whether the article could be used for the study.

Anther search strategy was using the “cited by” button on Google Scholar for certain articles. Then, the articles were assessed on title and abstract for relevant information. The following articles were used for this search: Yang et al. (2018)[17], Wagner et al. (2009)[4], Whittle et al. (2017)[18]

Table 3: Search strategy with keywords for each search engine that was used.

Search engine Keywords

Google Scholar Non invasive prenatal paternity testing NIPAT

Prenatal paternity testing (Y-)STR Prenatal paternity testing SNP Prenatal paternity testing Amniocentesis

(Transabdominal/ transcervical) Chorionic villus sampling Paternity testing

Invasive paternity testing NIPT test

NIPT prenatal testing

Google Cijfers seksueel misbruik

Verkrachtingen statistieken Nederland Prenatale screening cijfers

Non-invasive prenatal paternity testing, using autosomal STR or SNP markers.