Screening of Fetal Chromosomal Aneuploidy by Noninvasive Prenatal Testing: From Innovation to Setting Public Health Agendas to Potential Impact on Other Fields

My group first reported the presence of cell-free fetal DNA in the plasma of pregnant women in 1997 (1). At that time, I had just joined The Chinese University of Hong Kong after having spent 13 years abroad in the UK. My work in those early years was focused on uncovering the basic biology of cell-free fetal DNA and then working towards translating such knowledge into clinical applications (2). In the early days of research in the field of cell-free fetal DNA, many workers expressed skepticism that a cell-free method would allow fetal chromosomal aneuploidies to be detected robustly. Many investigators thought intuitively that cell-based approaches based on the isolation of fetal nucleated cells in maternal blood would be the preferred approach for such an application. Hence, it was an exciting development when my group and Stephen Quake's group at Stanford reported in 2007 that a single molecular counting approach such as digital PCR could be used to detect the presence of an increased dosage of chromosome 21 in a cell-free system such as maternal plasma (3, 4). In the subsequent year, both groups reported the application of random massively parallel sequencing to sample plasma DNA molecules from chromosome 21 and other chromosomes in maternal plasma and demonstrated an extremely high concordance rate with fetal chromosomal aneuploidy status (5, 6). After these initial demonstrations of the feasibility of noninvasive prenatal testing (NIPT) for detection of fetal chromosomal aneuploidies, my group went on to illustrate the robustness of this technique in a clinical trial that was published in January 2011 (7). We worked with Sequenom to commercialize the assay. The Quake team worked with Verinata, which was subsequently acquired by Illumina, to commercialize their assay. Such efforts were joined by many other companies around the world, including the BGI Group, Berry Genomics, Natera, Ariosa, and others. Hence, within a relatively short period of time, NIPT became available in many countries around the world. It is estimated that over 6 million NIPTs are now done annually worldwide. Many experts have commented that NIPT may be one of the most rapidly adopted genomic tests globally. It has been gratifying to see the initiation of pioneering national efforts in NIPT such as the programs in the Netherlands (8) and Belgium (9). In addition, in Hong Kong, a city where I am based, NIPT will become available under the public healthcare system probably by the end of 2019 or early 2020. Such developments will help to address the equity of access issue, because NIPT previously has been available primarily under private healthcare systems in many places around the world. In the first commercial use of NIPT, the technology was used for the screening of Down syndrome. The list of conditions detectable by NIPT then gradually increased to include trisomy 18 and trisomy 13 (10). Several vendors added the sex chromosomal aneuploidies and, recently, various subchromosomal aberrations (11, 12). Because of the very low prevalence of many of the detectable subchromosomal aberrations, the positive predictive values of their detection can be very low. Hence, some workers in the field have questioned the clinical utility of expanding NIPT to include such aberrations (13). In the future this likely will be an area of ongoing debate, accompanied by technological improvements. One area of technological development is in the area of the so-called cell-based NIPT (14, 15). Cell-free fetal DNA in maternal plasma is derived from the placenta. As the placental karyotype can be complicated by the presence of confined placental mosaicism, even if the detection technology is perfect, one can never be sure that the karyotype as predicted by cell-free NIPT will always be concordant with that of the fetus. Accordingly, in the context of aneuploidy detection, it can be argued that cell-free NIPT will always be a screening test and will require confirmation using a diagnostic test. Several researchers have argued that cell-based NIPT has potential to be developed into a noninvasive prenatal diagnostic test. However, it is important to bear in mind that because several cell-based NIPT prototype systems are based on the isolation of trophoblast cells, it is likely that they also may be affected by confined placental mosaicism. Although preliminary work has been reported in this area on cases with placental mosaicism (15), much more information will be needed to assess the relative strengths and weaknesses of cell-free vs cell-based NIPT. Apart from placental mosaicism, another reason the results of cell-free NIPT may not necessarily reflect the karyotype of the fetus is the presence of other maternal conditions, especially malignancy. My group has shown that, using a similar technology to NIPT, one can detect a variety of cancers using maternal plasma DNA sequencing (16). Several groups have published data demonstrating the occurrence of abnormal NIPT results from maternal malignancies (17, 18). Maternal malignancies are therefore one class of conditions that should be included when consenting pregnant women for NIPT. Another maternal condition that has been found to occasionally lead to unexpected NIPT results is organ or bone marrow transplantation previously received by the pregnant mother (19). This phenomenon traces back to an observation first made in 1998 that following transplantation, donor-derived DNA sequences could be detected in the plasma of transplantation recipients (20). This discovery was based on the same approach used for the first detection of fetal DNA in maternal plasma (1), i.e., via the detection of Y chromosomal sequences (from a male transplantation donor or from a male fetus) in the plasma DNA. In addition to being a confounding factor in NIPT for pregnant women who have previously received transplantation, the measurement of donor-derived DNA in the plasma of transplantation recipients has also been developed into a tool for detection of rejection (21, 22). Beyond research to directly advance the practice of NIPT, development of a fundamental understanding of the biology of circulating nucleic acids, in my opinion, will ultimately guide the future development of the field. Two areas are particularly interesting. The first is the fragmentation of plasma DNA, the study of which can be referred to as “fragmentomics.” Circulating tumor DNA (23) and fetal DNA (24) are short fragments of DNA. Furthermore, circulating fetal (24) and tumor DNA (25) molecules are slightly shorter than the background plasma DNA. High-resolution plasma DNA sequencing studies have shown that circulating DNA of various sizes could be mapped back to various portions of the DNA backbone of a nucleosome (26). One question that puzzled me for many years was the nature of the “scissors” that cleaved the DNA to form the fragments of DNA in plasma. It seemed likely that the process involved in the cleaving was catalyzed by an enzyme. Because DNASE1 was thought to be a major nuclease that is present in human plasma, early studies in my laboratory focused on the plasma DNA of mice in which the deoxyribonuclease 1 (Dnase1)3 gene had been knocked out. Disappointingly, no apparent differences in plasma DNA size profile could be observed in the knocked-out vs the wild-type mice (27). We then moved onto another nuclease, DNASE1L3. This time around, mice in which the Dnase1l3 gene had been knocked out exhibited interesting aberrations in plasma DNA size, with increased prominence of the plasma DNA fragments of dinucleosomal and higher order nucleosomal lengths (28). DNASE1L3 might be only one of several enzymes involved in the generation of the fragmented nature of plasma DNA. It is noteworthy that mice in which the Dnase1 and/or Dnase1l3 genes have been knocked out develop autoimmune disorders. It would be interesting to further explore the relationship between the length of circulating DNA molecules and their potential immunogenicity. The second area that I have found particularly interesting concerns the tissue of origin of circulating DNA. The first approach that my group used to investigate this area was transplantation, in which the genetic differences between the donor and recipient would allow us to detect and measure DNA released from the donated organ (19, 29). This line of work demonstrated that the main bulk of the circulating DNA pool was derived from the hematopoietic system (30). Another approach was based on the fact that genomic DNA from different tissues within the body would exhibit different DNA methylation profiles. For plasma-based detection, we first demonstrated that differential methylation patterns between the fetus and the pregnant mother allowed us to differentiate fetal and maternal DNA in maternal plasma (31). Using DNA methylation patterns closely associated with the placenta, we further demonstrated the placental origin of the “fetal” DNA in maternal plasma (32–34). Further developments have allowed us to perform methylome-wide profiling in plasma of pregnant women (35) and in cancer patients (36). Such technologies have stimulated the recent intense interest in mapping the tissue of origin of plasma DNA, thus opening up many new applications of circulating nucleic acids, from elucidating the origin of an aberrant signal in NIPT (37) to detecting myocardial damage (38) and liver damage (39) to investigating the causes of anemia (40) and tracing the tissue of origin of cancer (37). Circulating nucleic acids have provided a treasure trove of molecular diagnostics and research. I have been fortunate to be involved in this exciting field for the past 2 decades. Reflecting on the developments of this field over the past 20 years, I have learned a number of important lessons. First, it is often insightful to step outside of one's immediate field and to look at principles and patterns across apparently divergent fields. Hence, by seeing the similarities between circulating fetal and cancer DNA, one could pool the fruits of the efforts from workers in these 2 fields and accelerate developments. Second, it is useful to work on both the basic biology and the technological aspects simultaneously, because data generated from the biological and technological sides would synergize with one another. For example, on the one hand, if I had not known the basic biological parameters of circulating fetal DNA, such as gestational variation, concentrations, and clearance kinetics, then I would have had no way to logically work out the analytical requirement of an, NIPT system. On the other hand, if I had not been aware of technological developments such as digital PCR and massively parallel sequencing, then I would not have the means to use them to further elucidate the details of fetomaternal trafficking. Having seen the rapid adoption of NIPT into clinical practice that has resulted in a paradigm shift in prenatal practice worldwide, I am hopeful that further developments in this area will also impact fields such as oncology, transplantation, and other branches of medicine. deoxyribonuclease 1 deoxyribonuclease 1 like 3.