SAN FRANCISCO (GenomeWeb) – A research team from the National University of Singapore has developed a targeted sequencing method to calculate fetal fraction in cell-free DNA from maternal plasma, which it plans to use in research studies to develop noninvasive prenatal screens for single-gene disorders like beta thalassemia.
The method, which was described this week in PLOS One, involves analyzing a panel of 35 indel polymorphisms and comparing allele frequencies to determine what proportion of cell-free DNA in a plasma sample originated from the fetus.
Knowing fetal fraction is critical for identifying disease-causing mutations in single-gene disorders, said Angela Barrett, lead author of the study and senior research fellow at the National University of Singapore. Beta thalassemia is an autosomal recessive disease, so a fetus would have to inherit two disease alleles to be affected. Finding this out relies on accurately determining the proportion of mutated alleles compared to wildtype. "The algorithm we use to do that relies on knowing the fetal fraction," Barrett said.
Fetal fraction can be calculated by a variety of methods. For whole-genome sequencing-based NIPT, researchers have developed methods that rely on statistical modeling of read count across the genome. Researchers from Sequenom, for instance, developed an algorithm that they trained to estimate fetal DNA fraction based on read counts.
In addition, Dennis Lo's lab at the Chinese University of Hong Kong has demonstrated size differences between maternal and fetal DNA fragments that can be analyzed to determine fetal fraction.
In addition, a team from VU University Medical Center in the Netherlands developed a similar method that exploits differences in how fetal and maternal DNA is digested by analyzing nucleosomes.
Barrett said that those methods are primarily geared toward whole-genome sequencing approaches and her lab wanted a method that would be compatible with targeted sequencing. Natera, for instance, uses a targeted SNP-based approach, looking at more than 13,000 SNPs across five chromosomes.
Barrett said that the team had tried to use methods that look at methylation differences, since both Sequenom and Lo's lab had previously demonstrated that there are methylation differences between maternal and fetal cell-free DNA.
However, she said that the lab has not had success with methylation methods when they used them to look at specific genes, rather than a genome-wide approach, likely due to methylation fluctuating during pregnancy.
Instead, the team decided to focus on analyzing polymorphisms, specifically indel polymorphisms, because the Illumina MiSeq has a slightly lower error rate for indels than SNVs, Barrett said.
The researchers identified a set of indels that have global allele frequencies greater than 0.25 or an average heterozygosity greater than 0.3. Barrett noted that these requirements result in a "greater chance of both parents having a different allele at the same locus." They then designed amplicons that covered the indels and flanking sequences.
First, the team tested samples from six women in order to compare three different library prep kits: Illumina's TruSeq PCR-free and TruSeq Nano DNA kits, as well as Rubicon Genomics' ThruPlex-FD kit.
The three kits yielded comparable results, but the TruSeq PCR-free kit resulted in a significantly higher number of average reads, so the researchers used that one for their remaining analyses.
To test the panel, the team first analyzed samples with known fetal fraction — looking at maternal blood cells and amniotic fluid from an individual carry a male fetus. They also analyzed mixtures, diluting genomic maternal DNA with fetal DNA at concentrations from 1.6 percent to 50 percent . They found that the assay performed as expected and was quantitative even at 1.6 percent fetal DNA.
In addition, they analyzed samples from 61 patients carrying a male fetus and compared the indel panel to shotgun sequencing and an analysis of the Y chromosome reads, which is considered to be the most accurate way to measure fetal fraction for male fetuses. The two methods were correlated at a ratio of 0.69. The researchers also looked at a targeted analysis of the X and Y chromosome, and found that the indel panel performed better than the targeted Y chromosome analysis.
Barrett added that methods for determining fetal fraction vary a lot "depending on the algorithm used to analyze the data." Different methods for estimating fetal fraction can yield different results for the same sample, she added.
After validating the assay, they ran it on an addition 90 samples that included women with both male and female fetuses. The researchers determined that in order to estimate fetal fraction, a minimum of three indels on the panel should be informative. Informative alleles are those in which the mother is homozygous and the fetus inherits the alternate allele from the father, Barrett explained. On average, 12 indels per sample yielded informative information about fetal fraction, and all samples contained enough informative alleles to estimate fetal fraction. Mean fetal fraction across all 157 samples was 14.4 percent.
Erik Sistermans, head of genome diagnostics at the VU University Medical Center, said that the method seemed similar to SNP-based approaches in other targeted sequencing NIPT methods, except that it used indels.
He anticipated that it would be useful for the targeted prenatal testing applications the researchers are pursuing but would have limited utility for whole-genome-based NIPT. When doing whole-genome sequencing, "it's better to directly calculate the fetal fraction from the sequencing data" and it wouldn't make sense to have a separate test to calculate fetal fraction, he said. But for targeted prenatal tests, such as for beta thalassemia, "you cannot derive the fetal fraction from the data itself, so you need an add-on," he said.
Kimberly Martin, Natera's senior global media director of women's health, said that the approach was novel, but was concerned about its robustness. The correlation of 0.69 was not great, she said. Natera's method, which looks at more than 13,000 SNPs to measure fetal fraction, has a correlation of more than 0.96, she said. She said the group would have to demonstrate that their fetal fraction method combined with single-gene prenatal testing gave reliable calls.
Barrett said that the team has since analyzed around 400 samples and that the panel has been working well, although she said that more samples would be needed to validate that the chosen indels would work for women from different populations.
She said the next step is to incorporate the fetal fraction panel with the beta thalassemia test. To do that, she said the lab would multiplex in the primers for the indels with the amplicons for the beta thalassemia gene, in order to minimize extra cost. The panel could then be run on the MiSeq.
She said that the team has also submitted a manuscript for noninvasive beta thalassemia testing in cases where the parents are known carriers. The paper is a proof of concept, she said, and the next step would be to do a clinical trial with a greater number of samples to demonstrate the analytical and clinical validity of the assay.
The indel method for fretal fraction determination could also be combined with other targeted tests for single-gene disorders, like cystic fibrosis or sickle cell anemia, she said.