NEW YORK – Using a recently developed long-read methylation sequencing technique, investigators from the Chinese University of Hong Kong have found that atypically long cell-free DNA molecules in maternal plasma could be a source for new methods to diagnose fetal monogenic disorders and maternal pregnancy complications.
Reporting their findings in the Proceedings of the National Academy of Sciences, investigators described the detection of cfDNA molecules up to 23,635 bp in length in maternal plasma and their development of an approach using the abundance of CpG sites on these molecules to deduce their tissue of origin of based on single-molecule methylation analysis.
In an email, senior author Dennis Lo said that the current data represents just a first step – proof of concept that will have to be further validated in larger cohorts of samples, particularly for prenatal testing of monogenic disorders.
His team at CUHK has plans to begin these studies which, if successful, would provide the necessary supporting evidence for Take2, Lo and his colleagues' commercial spinout, to move forward. Although he said that monogenic NIPD is a target for the company, he declined to comment on specific plans.
Although now ubiquitous, NIPT has largely remained limited to broadly applicable testing for detection of chromosomal disorders, and to disease-specific tests for families with known monogenic carrier status. However, the possibility of diagnosing de novo pathogenic mutations in fetuses has long been a hope as cell-free DNA technologies have advanced.
"We have been looking for approaches that are potentially more scalable [and we] hope that in time, long cfDNA in maternal plasma would offer us an additional approach to get there," he added.
One of the first commercial offerings to employ standard short-read sequencing is Baylor Genetics' PreSeek, also marketed by Natera under the name Vistara, a single-gene screening test for dominant disorders. Baylor and Natera researchers recently reported on a cohort of 422 pregnancies in which the test identified all positives and provided no false-positive or false-negative results for those cases where follow-up information was available.
But Lo said that he and his team believe that there is still an unmet need to further expand the use of noninvasive prenatal testing for autosomal recessive disorders, which they hope long cfDNA sequencing could help address.
In the PNAS study this week, he and his team used a method they had developed to improve the detection of DNA cytosine methylation via PacBio SMRT sequencing, which they call the holistic kinetic (HK) model.
Studying samples from seven pregnant women, the team found that cfDNA molecules over 500 bp were prevalent in maternal plasma at fractions between about 15 percent and 32 percent depending on pregnancy trimester.
According to the authors, one advantage of being able to analyze long cell-free DNA molecules in the plasma at a single-molecule level is the possibility to use the greater ubiquity of CpG sites to obtain more genetic and epigenetic information.
Using this advantage, the team developed an approach that analyzes methylation patterns of the series of CpG sites along a long DNA molecule to determine its tissue origin — either fetal or maternal — showing that it had relatively high accuracy (an AUC of 0.88) in differentiating between fetal and maternal plasma DNA molecules.
The utility for prenatal disease diagnosis in this differentiation lies in the ability to directly compare single nucleotide polymorphisms in corresponding maternal and fetal DNA sequences.
Paternal haplotype information wasn't available in this initial study, so the team determined maternal inheritance based on a quantitative analysis of the imbalance between the two maternal haplotypes. With dual parental information, an easier qualitative approach would be possible, the authors wrote.
According to the team, one limitation of the study is that their tissue-of-origin analysis did not attain an accuracy of 100 percent. However, the authors wrote that the protocol could potentially be improved using bead-based or other enrichment techniques for long DNA molecules.
"The longer the DNA molecule, the more CpG sites the molecule will likely carry," the authors wrote.
Another limitation cited was the current relatively low throughput of Pacific Biosciences SMRT sequencing, which the authors wrote is a cost challenge for immediate clinical implementation.
"Nonetheless," they said, "we believe that our current work demonstrating the presence of a large population of previously unexplored long cell-free DNA molecules would create a motivation to overcome such technical challenges."
The study also explored the possibility of diagnosing pregnancy complications via long cfDNA molecules, with investigators able to demonstrate that plasma DNA size and end-motif patterns were different in pregnant women with and without preeclampsia.
The authors wrote that the small sample sizes in the study mean that this would require further validation to confirm. Future studies exploring whether these differences in fragment size and end-motif prevalence might predict the development of preeclampsia before the onset of clinical symptoms would also be warranted, they added.
Although the CUHK team employed their own HK SMRT sequencing method, Lo said the methodology for mining this long cfDNA trove may not matter as long as the techniques used are "conducive to preserving and analyzing [these] molecules."
Lo and his colleagues haven't limited their research to cell-free DNA length, having published previous findings regarding patterns of fragment "jaggedness," and circular orientation with the potential to identify fetal origin.
In his email, Lo said that it makes sense that various cfDNA-based biomarkers, including fragmentomic patterns, long cfDNA molecules, and other factors, would "carry additive information that together would allow us to describe a physiological or pathological condition better." As such, the team continues to be "open" to potentially employing multiple targets in commercial clinical products.