Noninvasive fetal whole-genome sequencing from maternal plasma: feasibility studies and future directions.

The discovery of cell-free fetal DNA in maternal plasma in 1997 has catalyzed the development of noninvasive prenatal diagnosis (1 ). Fetal DNA represents on average approximately 15% of the total circulating DNA in maternal plasma. The advent of massively parallel sequencing has made possible the analysis of circulating fetal DNA with unprecedented precision. In 2010, Lo et al. demonstrated that a fetal genomewide genetic map could be assembled by using a combination of DNA-sequencing data from maternal plasma and genetic maps of the parents (2 ). The latter data include a genotype map of the father and a haplotype-resolved genotype map of the mother. In this approach, the resolution of the deduced fetal genomic map is governed by the resolutions of the parental genetic maps. This approach was used to create a genomewide map of single-nucleotide polymorphism (SNP) alleles that the fetus had inherited from the father but that were absent in the mother’s genome (Fig. 1A). The focus was then shifted to the maternally inherited half of the fetal genome and the SNP alleles that were heterozygous in the mother and homozygous in the father (Fig. 1B). These heterozygous SNPs would form 2 haplotypes within the mother’s genome and on a pair of homologous chromosomes. Analyzing DNAsequencing data of the maternal plasma allowed comparisons of the relative dosages of these haplotypes in maternal plasma. One could then deduce which haplotypes had been passed on to the fetus, an approach termed “relative haplotype dosage analysis.” The maternal inheritance of the fetus was resolved as a series of 3000 haplotype blocks across the entire genome, and the approach was demonstrated to be useful for the noninvasive prenatal diagnosis of monogenic diseases such as -thalassemia (2 ). In this proof-of-concept study, Lo et al. opted to construct the maternal haplotype map by using fetal genotype information obtained from a chorionic villus sample, but the maternal haplotype could also have been deduced with samples from other members of the mother’s family. Similar end points would have been obtained aside from the dozens of haplotype blocks of the total of 3000 in which meiotic recombination had taken place in the passage of the genetic information from the mother to the fetus. A more recent report by Kitzman et al. (3 ) confirmed this approach of Lo et al. (2 ). Kitzman et al. demonstrated the ability to construct a fetal genomic map by combining maternal plasma DNA-sequencing data, paternal genotype information, and maternal haplotype information. In addition, they described several additional developments. First, they sequenced the maternal plasma DNA to a depth of 78-fold haploid genome coverage (3 ), compared with the 65-fold haploid coverage previously reported (2 ). Second, the number of parental heterozygous SNP sites analyzed was increased by 10-fold (3 ) compared with the earlier study (2 ). Third, the maternal haplotype was deduced with a direct sequencing– based approach. Fourth, they explored the analysis of maternal plasma DNAsequencing data for detecting fetal de novo mutations and identified 2.5 10 candidate de novo mutations. Compared with the 44 actual fetal de novo mutations identified by sequencing cord blood of the baby at delivery, only 39 of the candidate mutations deduced from the maternal plasma data were proved true. The sensitivity of their approach was thus 88.6%; however, the positive predictive value of the method was very low because of the extremely large number of falsepositive results. An overall analysis of the points described above and the data reported by Kitzman et al. demonstrates that the fetal-genome sequencing approach proposed by Lo et al. is scalable and robust, even when it is pushed to a greater sequencing depth and higher resolution. The concept of detecting fetal de novo mutations is interesting, although the specificity and sensitivity would require substantial improvement before the clinical applications could be investigated. Further advances in sequencing technology to allow greater 1 Li Ka Shing Institute of Health Sciences and 2 Department of Chemical Pathology, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong SAR, China. * Address correspondence to the author at: Department of Chemical Pathology, Prince of Wales Hospital, 30-32 Ngan Shing St., Shatin, New Territories, Hong Kong SAR, China. Fax 852-26365090; e-mail [email protected]. Received January 18, 2013; accepted January 31, 2013. Previously published online at DOI: 10.1373/clinchem.2012.191098 Clinical Chemistry 59:4 000 – 000 (2013) Perspectives