Noninvasive prenatal testing by maternal plasma DNA analysis: Current practice and future applications

Abstract

Prenatal screening of fetal chromosomal aneuploidies and some common genetic diseases is an integral part of antenatal care. Definitive prenatal diagnosis is conventionally achieved by the sampling of fetal genetic material by amniocentesis or chorionic villus sampling. Due to the invasiveness of those procedures, they are associated with a 1 in 200 chance of fetal miscarriage. Hence, researchers have been exploring noninvasive ways to sample fetal genetic material. The presence of cell-free DNA released by the fetus into the circulation of its mother was demonstrated in 1997. Circulating fetal DNA is therefore obtainable through the collection of a blood sample from the pregnant woman without posing any physical harm to the fetus. By analyzing this source of fetal genetic material, researchers have succeeded in developing DNA-based noninvasive tests for the assessment of Down syndrome and single gene diseases. Since the end of 2011, tests for the noninvasive assessment of chromosomal aneuploidies have become commercially available in parts of the world. Recommendations from professional groups have since been made regarding how these tests could be incorporated into the framework of existing prenatal screening programs. More recently, cell-free circulating fetal DNA analysis have been shown to be applicable to the deciphering of the fetal molecular karyotype, genome and methylome. It is envisioned that an increasing number of the noninvasive prenatal tests will become clinically available. The ethical, social and legal implications of the introduction of some of these tests would need to be discussed in the context of different cultures, societal values and the legal framework.

Key Words::

• Noninvasive prenatal diagnosis
• maternal plasma DNA
• Down syndrome

Background

The definitive diagnosis of fetal chromosomal aneuploidies and genetic diseases are conducted through the analysis of fetal genetic material. To obtain fetal genetic material during pregnancy, conventionally, one would need to rely on invasive means, such as amniocentesis or chorionic villus sampling. However, these procedures are associated with a 1 in 200 chance of fetal miscarriage. To obviate such a risk, researchers explored sources of fetal genetic material that could be sampled noninvasively without posing harm to the fetus. In 1997, Lo et al. reported the presence of fetal DNA in plasma of pregnant women [1]. This finding offered the possibility to obtain fetal genetic material via a maternal blood sample. Researchers have since reported that the fetal DNA is cell-free in nature, exists as short fragments of < 200 bp long, contributes around 10–15 % of DNA in plasma collected from the first and second trimesters of pregnancy, and has an apparent half-life of < 1 h [2]. Noninvasive prenatal assessment of fetal Down syndrome and other chromosomal aneuploidies have since been achieved through the analysis of maternal plasma DNA. The test sensitivity and specificity have been shown to surpass 99 % which led to its commercial launch. The scientific principles for achieving noninvasive prenatal assessment of other conditions have been achieved. Hence, we are currently witnessing a paradigm shift in how prenatal genetic diagnosis is conducted.

Noninvasive prenatal Down syndrome screening

Down syndrome, a condition that is associated with mental retardation and physical morbidities such as congenital heart disease, is the commonest chromosomal aneuploidy in humans [3]. On average, Down syndrome occurs in 1 in 750 fetuses [3]. Because the risk of fetal miscarriage from the invasive prenatal diagnostic procedures is even higher than the average risk of fetal Down syndrome, couples are faced with a dilemma when they consider prenatal diagnosis. Therefore, it has been customary to use screening tests that are safe and noninvasive to stratify women according to their fetuses’ risks for Down syndrome before recommendations for invasive prenatal testing are made. Conventionally, screening tests have been based on the measurement of proteins and hormones in maternal serum, fetal ultrasonographic assessment, maternal age and family history. However, those screening modalities are associated with 3–5 % false-positive rates [4]. Therefore, a disproportionate number of pregnancies are faced with the dilemma of opting for invasive prenatal diagnosis or not. Hence, there has been a practical need to refine the conventional prenatal screening schemes to reduce the false-positive identification of high risk pregnancies.

Down syndrome is typically caused by the presence of an additional copy of chromosome 21, amounting to a total of three copies (i.e. trisomy 21), in the genomes of affected individuals. Researchers in the field of noninvasive prenatal diagnosis therefore aimed to develop approaches to detect the increase in chromosome 21 DNA content among the cffDNA molecules in maternal plasma. Massively parallel sequencing (MPS), also referred to as next-generation sequencing, is a technology that allows the determination of the nucleotide sequences of millions to billions of short DNA fragments in each analysis run. Once the nucleotide sequence of a DNA fragment is available, one could determine its genomic identity by comparing with the reference human genome sequence. MPS was applied to the analysis of cell-free DNA in maternal plasma for Down syndrome detection [5,6]. The rationale was that DNA fragments in maternal plasma (including the DNA molecules from the fetus and mother) are sequenced by MPS. The sequenced DNA reads are then compared with the human reference genome sequence to determine if the DNA fragment originated from chromosome 21 or another chromosome. Because millions of DNA fragments are sequenced from each maternal plasma sample, the amount of chromosome 21 DNA fragments sequenced as a proportion of other non-chromosome 21 DNA fragments sequenced from the sample could be determined. If the fetus has trisomy 21, there should be disproportionately more chromosome 21 DNA fragments sequenced from the sample in comparison to samples of normal pregnancies. Since the test directly detects the core pathology of Down syndrome, namely the overrepresentation in chromosome 21 DNA content among the fetal DNA sequences, its accuracy should theoretically be better than the screening tests based on the assessment of phenotypic features. In addition, by detecting the core pathology of Down syndrome, the MPS-based noninvasive prenatal test (MPS-NIPT) is applicable across different gestational ages.

Performance of MPS-NIPT test for fetal chromosome aneuploidy detection

Soon after the initial development of the MPS-based approaches for noninvasive Down syndrome detection, researchers investigated the diagnostic performance of the MPS-NIPT for trisomy 21 detection. Several studies focused on the prospective recruitment of women pregnant with a single fetus and were clinically indicated for invasive diagnostic testing with full karyotyping results available [7–10]. The indications for invasive testing included pregnancies deemed to be at high risk for fetal aneuploidies based on conventional screening results, maternal age, fetal ultrasound findings or family history. In other words, those studies were primarily designed to investigate the performance of MPS-NIPT among high risk pregnancies.

Palomaki et al. [8] conducted a nested case- control study where MPS analysis was performed on 1484 euploid pregnancies and 212 trisomy 21 pregnancies. The mean maternal age was 37 years. The mean gestational age was 15 weeks (interval: 8.1 to 21.5). Result reporting was not available for 13 eligible samples (0.8 % of 1696 samples) due to quality control failures. The test correctly identified 209 of 212 trisomy 21 fetuses giving a detection rate of 98.6 % (95 % CI, 95.9–99.7). Three of the reported 1417 euploid pregnancies were reported as trisomy 21, giving a false-positive rate of 0.2 % (95 % CI, < 0.1–0.6).

The study by Bianchi et al. [9] was a nested case-control study with 90 trisomy 21 pregnancies and 426 non-trisomy 21 pregnancies that were eligible for analysis [9]. The mean maternal age was 35 years and mean gestational age was 15 weeks (interval: 10–23). Test results were not available for 23 samples (inclusive of one trisomy 21 sample) due to failure to meet criteria for reporting. This resulted in a non-reportable rate of 4.5 %. Of the remaining 89 trisomy 21 pregnancies, all were reported as trisomy 21 that gave a sensitivity of 100 % (95 % CI, 95.9–100). All 404 of the non-trisomy 21 pregnancies were reported as not trisomy 21 (95 % CI, 99.1–100).

Norton et al. [10] conducted a cohort study with MPS analysis performed on 3,228 samples, inclusive of 84 cases of trisomy 21 pregnancies and 2,888 euploid pregnancies. Results were not issued for 148 samples (4.6 % of the cohort), three of which had trisomy 21, due to quality control failures. The mean maternal age was 34.3 years and mean gestational age was 16.9 weeks (range: 10–38.7). A total of 80 of the 81 trisomy 21 cases were identified as high risk for trisomy 21 by the test while 2,887 of 2,888 euploid cases were identified as low-risk. These results gave a detection rate of 100 % (95 % CI, 95.5–100) and a false-positive rate of 0.03 % (95 % CI, 0.002–0.2).

In summary, the data from studies suggest that MPS-NIPT offers > 99 % detection, at < 0.5 % false-positive rate, for trisomy 21 among high-risk pregnancies. However, the rate of non-reportable results varied between protocols and ranged from some < 1–4.6 %.

Besides trisomy 21, trisomy 18, trisomy 13 and sex chromosome aneuploidies are among the other common fetal chromosomal aneuploidies that are targeted in the conventional prenatal screening programs. In theory, the principles for using MPS on maternal plasma DNA to noninvasively detect trisomy 18 and trisomy 13 should be similar to that for trisomy 21. However, studies reported that the extreme cytosine and guanine nucleotide contents of chromosomes 18 and 13 resulted in biased estimations by MPS, termed GC bias, of the contents of chromosome 18 and chromosome 13 DNA in maternal plasma, resulting in insensitive detection of trisomies 18 and 13 [11,12]. Thus, MPS-NIPT for trisomy 18 and trisomy 13 required the incorporation of steps (e.g. computation) to minimize, or correct the GC bias.

With these optimized protocols in place, a number of studies addressed the performance of trisomy 18 and trisomy 13 detection by MPS-NIPT. In general, the detection rates were > 95 % and the false-positive rates were ∼1 %. More ambitiously, researchers have explored the development of MPS protocols to determine the fetal molecular karyotype noninvasively [13,14]. Such protocols allow the noninvasive screening of microdeletions and microduplications.

Adoption of MPS-NIPT of fetal chromosomal aneuploidies for clinical use

The National Coalition for Health Professional Education in Genetics (NCHPEG) and National Society of Genetic Counselors (NSGC) in the US have jointly developed a factsheet [15], published in August 2012, addressing NIPT for fetal chromosomal aneuploidy screening. The factual information of the findings of the published studies was presented. In the factsheet, it was stated that the ‘NSGC supports NIPT as an option for patients whose pregnancies are considered to be increased risk for certain chromosome abnormalities’ [15]. The NSGC also recommended NIPT only be offered with informed consent and counseling by qualified providers. The factsheet indicated that MPS-NIPT was sensitive and specific for trisomies 21 and 18. However, due to the chance of false-positivity, confirmatory testing by CVS or amniocentesis was recommended for cases tested positive by MPS-NIPT. It was stressed that a negative result did not absolutely rule-out trisomy 21 or trisomy 18. Fetal anatomic ultrasound to look for features of aneuploidies added value to interpreting the MPS-NIPT results.

Subsequently, recommendations and position statements from a number of professional groups were published with largely similar conclusions. Such professional groups included the American College of Obstetricians and Gynecologists [16], International Society for Prenatal Diagnosis [17], Society of Obstetricians and Gynaecologists of Canada [18]. High risk pregnant women generally refers to those screened positive by conventional aneuploidy screening protocols, women with advanced maternal age, ultrasound findings suggestive of trisomies 21, 18 or 13, and family or personal history of the birth of trisomies 21, 18 or 13. Some of the published statements further highlighted the need to develop guidelines for laboratory standards, quality controls and proficiency testing.

The use of MPS-NIPT for fetal chromosomal aneuploidy screening among high risk pregnancies have since become commercially available and adopted for clinical use in many parts of the world. Researchers began to ponder if the test would also be applicable to non-high risk pregnancies and whether the test performance would be as sensitive and specific. The fractional content of DNA in a maternal plasma sample that is contributed by the fetus, termed fetal fraction, is one of the key parameters that affect the sensitivity of MPS-NIPT [7]. Samples with low fetal DNA fractions have higher chances of false-negative detection. Hudecova et al. [19] recruited women receiving Down syndrome screening by the first trimester combined test. MPS was used to assess the Y chromosomal content in plasma of women conceived with male fetuses. The study found that the range and distribution of fetal fractions among the general pregnant population during early pregnancy did not show statistically significant differences from data previously reported for high-risk pregnancies. These data suggested that the performance of MPS-NIPT for chromosomal aneuploidy detection should theoretically be equivalent between high and average risk pregnancies.

Bianchi et al. [20] directly compared the screening of chromosomal aneuploidies by MPS-NIPT and conventional approaches among 1914 average risk women. MPS-NIPT detected all eight aneuploidy cases with a 0.3 % false-positive rate. These data were similar to those previously reported for high-risk pregnancies. The conventional screening approach also identified the eight aneuploid cases as high-risk. However, 3.6 % of the non-aneuploidy cases were also identified as high risk. The authors concluded that MPS-NIPT provided a positive predictive value of 45.5 % for trisomy 21 but only 4.2 % by conventional screening. The authors reasoned that MPS-NIPT should be seriously considered as a primary screening tool.

Other developments on the horizon

Besides fetal chromosomal aneuploidies, it would be ideal to achieve noninvasive prenatal detection of single gene diseases. Because fetal DNA circulates in maternal plasma among a background of maternal DNA, researchers have initially focused on assessing the presence or absence of paternally-transmitted mutations in maternal plasma for the assessment of paternally-transmitted autosomal dominant traits [2]. For autosomal recessive diseases, one would need to determine which maternal allele the fetus has inherited. To achieve this goal despite the interference by the maternal background DNA, quantitative approaches could be adopted [21]. The approaches aim to compare the relative amounts of cell-free DNA molecules carrying the mutation against the corresponding normal allele. If a mother being a carrier for the said mutation is pregnant with a fetus heterozygous for the mutation, the relative amounts of the mutant and normal alleles are equal. If the fetus is homozygous for the mutation, there would be more mutant DNA molecules in the plasma sample than the normal DNA molecules and vice versa. Fetal DNA is the minor population in maternal plasma, hence precise quantitative methods are required to achieve reliable detection of the fetal genotype with respect to the mutation. A digital polymerase chain reaction approach has been developed and was termed relative mutation dosage assessment [22]. The noninvasive assessments of beta-thalassemia [22], hemophilia [23] and sickle cell anemia [24] have been demonstrated using relative mutation dosage. MPS-based methods with or without target capture of the genetic loci of interest have been developed and was termed relative haplotype dosage analysis [25]. These approaches have been demonstrated for the prenatal assessments of beta-thalassemia [26] and congenital adrenal hyperplasia due to 21 hydroxylase deficiency [27].

When the relative haplotype analysis approach is applied to the entire genome, one could determine the fetal genotype at multiple polymorphic sites and disease loci, and thereby noninvasively decoding the fetal genome [25,28]. In addition to the fetal genome, when bisulfite treatment is applied to maternal plasma DNA before MPS analysis, one could determine the placental methylome noninvasively [29]. It is now recognized that the placenta is the main source of cffDNA in maternal plasma while the maternal blood cells are the main source of maternal DNA. Bisulfite is a substance that converts unmethylated cytosines into thymines and methylated cytosines remain unchanged. This results in sequence changes enabling one to distinguish methylated and unmethylated DNA molecules. By knowing the methylation status of the background maternal DNA and the fetal fraction, one could noninvasively deduce the placental DNA methylation status by direct analysis of maternal plasma [29].

The future is now

DNA-based noninvasive prenatal detection of fetal chromosomal aneuploidies and single gene diseases was once an elusive goal. The presence of cffDNA in maternal plasma provided a source of fetal genetic material that could be sampled safely without causing physical harm to the fetus. Researchers have subsequently succeeded in developing approaches for the noninvasive detection of the genetic basis of fetal chromosomal aneuploidies and single gene diseases. Approaches have subsequently been extended to determining the fetal molecular karyotype, genome and methylome noninvasively. We are witnessing a start in a paradigm shift regarding how prenatal screening of chromosomal aneuploidy is performed clinically. It can be foreseen that more applications of NIPT will be adopted for clinical use. However, as with many new technologies, ethical, social and legal issues have emerged [30]. For example, equality of test access, adequacy of the informed consent process, fetal sex selection and eugenics are among issues that have been raised. To ensure the safe and effective implementation of NIPT, research, discussions and debates on such issues should go hand-in-hand with the scientific developments in the field. Nonetheless, the future of prenatal diagnosis has begun.

Questions and answers

Q (Lazzarotto): What is the turnaround time for the test?

A (Chiu): For screening of Down syndrome with the latest models of sequencer, up to 144 samples can be processed per run. The turnaround time varies between laboratories but is between 5 days and 2 weeks. For the fetal genome it is longer because we analyze fewer samples per run. The time of the run is the same but fewer samples could be processed per run.

Q (Lazzarotto): Is it possible to perform different tests in one run?

A (Chiu): That depends on the design of the sequencer. Some sequencers allow two different flow cells so two different protocols can be performed in the same run, provided you are using the same sequencing run time.

Q (Anderson): Are there issues with confined placental mosaicism?

A (Chiu): Confined placental mosaicism may lead to false positive results. Also, unknown maternal mosaicism is another cause of false positive results. The data show that false positives are typically less than 1 % of results. Confirmation by invasive procedures is recommended.

Q (Ferrari): There are many private companies offering this test. What do you think is the possibility of transferring this service into the public sector, in the light of economic considerations?

A (Chiu): The service is now offered in the private sector because of the cost of the sequencing, but the cost will come down and public services will offer these tests in the future. The demand for these tests will increase dramatically and the cost of service provision will be cheaper.

Comment (Spencer): I fully agree with Dr Chiu. In 3–5 years these tests will be routine in the public service and for example, my laboratory will do 40–50,000 tests per annum for screening by sequencing. It becomes economically viable at that level. There are likely to be about four such centers in the UK. The other scenario is the utilization of bench top sequencers for smaller workloads. I don't know if the cost of bioinformatics will drop as quickly as the cost of the sequencing.

Q (Young): May I ask you to comment on the bioinformatics issue? In particular, could you comment on the possibilities of automation and the time it takes to interpret the data being generated.

A (Chiu): I will not cover cancer because it is more heterogeneous and the analysis is more complicated. In prenatal diagnosis we have to detect a specific target, for example in Down syndrome and the cost of analysis is not high. The cost is higher for data storage. In the future utilizing a cloud approach may be the answer and may be less costly.

Q (Spencer): We have not even begun to scratch the surface of ethics arguments around using these techniques. How will we quality assure laboratories conducting these programs?

A (Chiu): We evaluate quality in an identical way to a clinical biochemistry laboratory. In my laboratory we check all steps in the process and there is a QC program. Also, EQA schemes have been set up and my laboratory participates in the UKNEQAS scheme. Quality considerations are more highly developed than might be thought.

Q: Is there interference from maternal DNA? What is the process of fetal DNA clearance and how long does it take? Is there a possibility of interference with screening during the next pregnancy?

A (Chiu): In 1999, we demonstrated with real time PCR that all male fetal DNA is gone within 2 hours of pregnancy but we recently used a sensitive sequencing technology which showed that it has all disappeared in about 1 day, so the half life is short and will definitely not cause problems in screening in any subsequent pregnancy

Q: How important is the evaluation of the fetal DNA fraction.

A (Chiu): It is the single most important factor. The protocol we use is based on 4 % fetal DNA. Another important factor is maternal weight. The higher the maternal weight, the lower the fetal DNA.

Acknowledgements

Supported by the University Grants Committee of the Government of the Hong Kong Special Administrative Region, China, under the Areas of Excellence Scheme [AoE/M-04/06], an Innovation and Technology Fund [ITS095/14FP] and a sponsored research agreement from Sequenom, Inc.

Declaration of interest: The author reports no conflicts of interest. The author alone is responsible for the content and writing of the paper.