Researchers from Stephen Quake's laboratory at Stanford University have combined a haplotyping technique with whole-genome sequencing, demonstrating in a proof-of-principle study that an entire fetal genome can be determined from maternal plasma without genetic information from the father.
The proof-of-principle study was published last week in Nature and demonstrates a noninvasive method for determining clinically relevant alleles in a fetal genome from maternal plasma.
While the method is not yet practical for a clinical setting, co-lead author Wei Gu told Clinical Sequencing News that a possible application for the approach would be to do carrier screening on the mother for a small number of diseases, and then to follow that up with targeted sequencing of maternal plasma.
The Nature study builds on the Quake team's previous work in developing a method for whole-genome haplotyping, which was published in Nature Biotechnology (IS 12/21/2010).
That technique uses microfluidics to phase chromosomes. When applying the method to maternal plasma, the researchers can determine the two different maternal haplotypes. Then, whole-genome sequencing of maternal plasma plus "allele counting" enables the team to determine which maternal haplotype was passed along to the fetus. Piecing together the inherited paternal haplotype requires both the sequence data plus imputation. The fetal genome is then constructed from the inherited maternal and paternal haplotypes.
Quake's team verified the approach on two different pregnant women, one of whom was carrying a normal fetus and another who was carrying a fetus with a heterozygous deletion on chromosome 22, which is associated with DiGeorge syndrome, a heterogeneous disorder that can cause congenital heart disease, defects in the palate, neuromuscular problems, and learning disabilities, among other problems.
They used their microfluidic phasing technique on three or four cells from each sample and also genotyped cord blood that was collected at delivery to serve as a reference for the fetal genotypes.
All samples were sequenced on the Illumina platform. DNA from the healthy fetus during the first trimester was sequenced to 53-fold haploid coverage and again during the second trimester to 21-fold coverage. The affected fetus was sequenced during the third trimester to 1.3-fold coverage.
Paternally inherited haplotypes were reconstructed through a combination of detecting paternal-specific alleles from the sequence data and also through imputation. The researchers were able to detect between 66 percent and 70 percent of the paternally inherited alleles, and were also able to deduce about 70 percent of paternally inherited haplotypes using imputation.
The method was able to identify the deletion on chromosome 22 from the mother carrying the fetus affected with DiGeorge syndrome.
Next, they wanted to see if they could determine the clinically relevant portions of the fetal genome by applying the counting method to the exome. They tested the method from the sample with the healthy fetus from all three trimesters, sequencing the fetal exome to over 100-fold coverage in each sample.
The greater sequencing depth increased their ability to assign fetal genotypes and detect paternal specific alleles. "In all three trimesters, fetal genotypes could be assigned robustly at loci where the mother is homozygous," the authors wrote.
Additionally, paternal-specific alleles were detected with a sensitivity of 96 percent to 99.8 percent.
The team also found that the percentage of fetal DNA increased throughout pregnancy. In the first trimester, fetal DNA made up just 6.6 percent of total DNA, but that percentage increased to 20.1 percent and 26.3 percent in the second and third trimesters, respectively.
The greater the fetal DNA percentage, the more sensitive the assay was at distinguishing between heterozygous and homozygous SNPs. For instance, maternal heterozygous SNPs could only be detected with 16 percent sensitivity and 90 percent specificity from the first trimester sample, but sensitivity was increased to 93 percent from the third trimester sample.
Gu said that the method is similar to two other published fetal sequencing techniques: one by Dennis Lo's group from the Chinese University of Hong Kong that was published in Science Translational Medicine in 2010 (IS 12/14/2010), and another by Jay Shendure's team from the University of Washington that was published in Science Translational Medicine last month.
One main difference between those methods and this one, said Gu, is that both other methods rely on having paternal DNA to determine the fetal genome. "We wanted to do this without the dad's DNA," said Gu.
Nevertheless, Gu said that whole-genome sequencing of fetal DNA is not yet practical in a clinical setting. The current study was a "proof-of concept to show that you can cover the whole genome. But, there's a question of, 'do you need the whole thing?'" he said.
A more likely first step for clinical implementation is to first do carrier screening of the mother, potentially also the father, and then adapt this proof of principal to targeted sequencing of maternal plasma, he said.
Then, instead of scouring the entire genome for disease-causing mutations, only the areas where there was a risk for a disease allele being inherited would need to be analyzed.
That would help reduce the complexity of analyzing the data as well as the sequencing costs, Gu said. Additionally, it would help focus the analysis on clinically useful portions of the genome where known disease variants reside. Currently, for most of the genome, there is not "well-validated data for many mutations," Gu explained. For the whole genome to truly be useful, a "better and integrated clinical database that covers all ethnicities is needed."
A targeted sequencing test would also have a quicker turnaround time and a lower cost than whole-genome sequencing.
The study's authors also note that the method could be useful for testing for conditions "that are not survivable or lead to medical complications." Additionally, it will enable "diagnosis of conditions that would benefit from treatment immediately after delivery," including metabolic and immunologic disorders.
However, they also acknowledge that the study did not address the fact that the ability to determine an entire fetal genome raises ethical issues, such as how that knowledge could affect decisions about abortion.