Digital Fetal Aneuploidy Diagnosis by Next-Generation Sequencing

Since the identification of fetal lymphocytes in maternal blood in 1969, investigators have endeavored to develop genetics-based noninvasive prenatal diagnostics (NIPD)1 (1). A robust noninvasive approach would augment or potentially supplant amniocentesis and chorionic villus sampling, which, although gold standards, carry a risk of fetal loss. Despite considerable efforts, the use of fetal cells for NIPD has never reached clinical implementation because of their paucity (on the order of a few cells per milliliter of maternal blood) and concerns of fetal cell persistence in the maternal circulation between pregnancies (2). A new avenue was opened in 1997 by the discovery of circulating cell-free fetal DNA in maternal blood (3). Cell-free fetal DNA constitutes between 5% and 10% of the total DNA in maternal plasma and increases during gestation. It rapidly clears from the circulation postpartum, a feature that enhances its attractiveness as a per-pregnancy–specific analyte. Several clinical applications based on the analysis of cell-free fetal DNA have been developed. These assays include determining fetal Rh D status in Rh D-negative women (4), sex in sex-linked disorders (3)(5), and detection of paternally inherited autosomal recessive and dominant mutations (6). In the context of these successes, however, there remains the outstanding challenge of the use of cell-free fetal DNA for the detection of chromosomal aneuploidy, in particular trisomies 21, 18, and 13. A cell-free fetal DNA–based approach for chromosomal aneuploidy must overcome several technical hurdles. By virtue of being cell free, fetal DNA in the maternal plasma cannot be easily analyzed by traditional visualization methods such as fluorescence in situ hybridization. The high “background” of maternal DNA (cell free or from residual cells) further complicates analysis by diluting the fetal genetic information. To address these issues, investigators have worked to identify fetal-specific molecular markers. The origin of circulating cell-free fetal DNA is primarily the placenta, whereas maternal cell-free DNA is derived from maternal leukocytes (7). By studying differences in genomic DNA methylation between the placenta and paired maternal leukocytes, investigators have characterized placenta-specific epigenetic markers (8). The finding of circulating cell-free placenta-derived mRNA allowed the identification of placenta-specific mRNA production (9). These findings have been the basis for the development of techniques for determining fetal chromosomal dosage that quantify the allelic ratio of single-nucleotide polymorphisms present in circulating fetal and maternal RNA or DNA. Although promising, these approaches depend on allelic heterozygosity between the fetus and mother for the analyzed single-nucleotide polymorphisms. Multiple markers are therefore needed for genetically diverse populations. To overcome this limitation, investigators have pursued polymorphism-independent methods, the first of which was digital PCR (10)(11). In digital PCR, individual fetal and maternal circulating cell-free DNA fragments are amplified in wells or microfluidics compartments under limiting-dilution conditions. By amplifying a target locus, on chromosome 21 for example, and quantitatively comparing its amplification with a locus on a reference chromosome, one should be able to measure an imbalance in chromosome 21 dosage. Specifically, the total number of chromosome 21 amplifications (representing maternal plus fetal contributions) divided by the number of reference chromosome amplifications should yield a ratio indicating an over- or underrepresentation of chromosome 21. Although the digital PCR approach is conceptually solid, the low percentage of cell-free fetal DNA in the maternal plasma sample requires the performance of thousands of PCRs to generate a ratio with statistical confidence. Ongoing investigations to enhance this method include the use of multiple target amplifications and enrichment of cell-free fetal DNA. In the setting of the above accomplishments, next-generation sequencing (NGS) has entered the picture. NGS comprises several recently developed technologies that share the fundamental approach in which clonally amplified DNA templates (or, most recently, single DNA molecules) are sequenced in a massively parallel fashion within a flow cell (12)(13). The chemistries include pyrosequencing, sequencing by synthesis with reversible dye terminators, and sequencing by oligonucleotide probe ligation. In addition to high-throughput sequence information, NGS provides digital quantitative information, in that each sequence read is a countable “sequence tag” representing an individual clonal DNA template or a single DNA molecule. This quantification allows NGS to expand the digital PCR concept of counting cell-free DNA molecules, and in 2008, Fan et al. and Chiu et al. described noninvasive detection of trisomy 21 by NGS (14)(15). Fan et al. also provided preliminary evidence for the detection of trisomies 13 and 18. Both groups extracted cell-free DNA from maternal plasma samples from both euploid and trisomy pregnancies. DNA from each sample was sequenced on the Illumina Genome Analyzer, and each sequence read was aligned to the reference human genome. Both groups used different data-analysis strategies to quantify the number of sequence reads aligning to each chromosome. The increased number of reads aligning to chromosome 21 in trisomy 21 cases was statistically separable from diploid chromosomes within the sample and disomy 21 cases. Impressively, these results were derived from sequencing <1 genome equivalent per sample. Two important caveats emerged from these studies. First, the investigators observed a nonuniform distribution of aligned sequence reads along chromosomes. The nonuniformity was related to GC content, potentially due to the PCR and bridge-amplification methods used in the Illumina technology. Second, the results obtained for the small number of trisomy 13 and 18 cases examined by Fan et al. were quantitatively less precise than those for trisomy 21. In this issue of Clinical Chemistry, Chiu et al. build on their earlier work with the Illumina Genome Analyzer and demonstrate noninvasive NGS-based trisomy 21 detection with the sequencing-by-ligation approach on the Life Technologies SOLiD platform (16). In the current study, cell-free DNA was isolated from the plasma of 15 pregnant women, 5 of whom carried trisomy 21 fetuses. The plasma samples were collected in the first trimester, between 11 and 14 weeks gestational age. Cell-free DNA was extracted from each sample, clonally amplified by emulsion PCR, and sequenced in 1 chamber of an 8-chamber SOLiD slide. This process yielded a median of 59 × 106 50-base reads per sample. A median of 12 × 106 reads (or 21%) were each aligned uniquely to one location of the reference human genome (with masking of repeat regions), for a coverage of approximately 20% of the haploid human genome. With their previously reported data-analysis strategy, these investigators compared the number of unique reads aligning to each chromosome with the total number of unique reads aligning to all chromosomes to derive a percent genomic representation for each chromosome. They then generated a “z score,” which is the difference between the percent genomic representation of the chromosome and the mean percent genomic representation for the same chromosome in the euploid control group, divided by the SD. For each trisomy 21 case, the chromosome 21 z score value indicated a 99% chance of a statistically significant difference from the chromosome 21 z scores for the controls. As reported earlier with the Illumina Genome Analyzer, a nonuniform distribution of aligned sequence reads was observed with the SOLiD data. Whereas Fan et al., using the Illumina Genome Analyzer, reported a positive bias for read alignment to chromosomes with higher GC content, the SOLiD data showed a negative bias. The existence of GC content bias with different platform trends suggests analytical influences, as opposed to biological factors. Combining the current SOLiD data with the previously reported data indicates that 28 trisomy 21 pregnancies and 30 euploid pregnancies have been studied and correctly identified with 2 different NGS platforms and statistical methods. These encouraging results set the stage for larger prospective clinical studies that use NGS. Several outstanding issues remain to be addressed. To date, only 2 trisomy 18 cases and 1 trisomy 13 case have been reported from studies with the Illumina Genome Analyzer. Although these cases were distinguishable from controls, further studies are needed to determine the ability of NGS to identify these trisomies. The current turnaround time for sample processing, sequencing, and data interpretation in experienced hands is 5 to 8 days for the Genome Analyzer and SOLiD platforms, respectively. The material costs for sequencing are approximately $700–$1000 per sample. Going forward, one can expect streamlining and automation of technical processes and data analysis, coupled with reduced sequencing costs. As sequencing throughput increases, it will be leveraged to increase genomic coverage, which should improve the precision of measurement of aligned sequence reads. Theoretically, NGS could detect other chromosomal aneuploidies, such as chromosome 16, that, although not found in live births, are responsible for spontaneous miscarriages (17). Structural aberrations across the genome may also be detected. An example of the trisomies is the partial chromosome duplications of the Down syndrome critical region (18). Combining sequence and copy number information could facilitate accurate determination of duplication boundaries to produce both a greater understanding of the clinical features associated with chromosomal regions and the prospect of predicting clinical severity. The nonuniformity of read alignment and GC content bias may be mitigated by new PCR-independent sequencing technologies for single DNA molecules, such as those developed by Helicos and Pacific Biosciences (19)(20). How might NGS fit into the current algorithm of prenatal diagnosis, in which screening results for maternal serum and ultrasound measurements of nuchal thickness guide the use of amniocentesis and chorionic villus sampling? Assuming noninvasive NGS-based diagnostics reach the clinic, one can envision initial implementation when the results of maternal serum screening are positive and NGS is elected over invasive procedures. Ultimately, reduced sequencing costs and turnaround times could pave the way for NGS-based NIPD to be considered as an alternative to serum biomarker screening, which, while cost-effective, remains prone to false positives. Forty years after the discovery of circulating fetal cells, the vision of NIPD appears clearer and closer. Nonstandard abbreviations: NIPD, noninvasive prenatal diagnostics; NGS, next-generation sequencing. Author Contributions:All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article. Authors’ Disclosures of Potential Conflicts of Interest:No authors declared any potential conflicts of interest. Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.