NEW YORK (GenomeWeb) – In a new study published in Cell today, researchers have shown for the first time that by analyzing deeply sequenced cell-free DNA and the spacing of nucleosomes, it is possible to determine the bodily origin of individual cfDNA fragments.
Despite growing adoption of technologies that can detect mutations in circulating cfDNA, the range of information that can be gleaned from these nucleic acids has thus far been limited to distinguishing somatic mutations from background DNA in the context of diagnosing or monitoring cancer.
In the new study, researchers from the University of Washington piloted an approach to infer nucleosome patterns from sequenced cfDNA and compare them to what might be expected in DNA released from cells dying in different tissues of the body.
The method opens up for the first time the possibility of tying bits of circulating DNA to a tissue of origin or particular disease process, and explored further, the results could both expand the utility of liquid biopsy in cancer and also potentially open up new uses of cfDNA to inform about other health conditions in which cells die and release DNA into the blood.
In the study, investigators led by Matthew Snyder and Martin Kircher, a graduate student and a postdoctoral fellow respectively in the laboratory of Jay Shendure at UW, set out to examine whether they could mark circulating DNA fragments as having a particular origin by deeply sequencing them to determine how nucleosomes were arranged when they were part of a contiguous strand of DNA.
According to the authors, each cell type in the body can theoretically be linked to a unique pattern of nucleosomes, which are key players in packaging DNA tightly into the cell nucleus. The group hoped to be able to pick up a blueprint for the bodily origin of circulating DNA molecules in their patterns of fragmentation, as a surrogate for the initial pattern of nucleosomes.
To examine this possibility, Snyder, Kircher, and colleagues first performed initial deep sequencing of cfDNA, in order to develop a method to infer the original positions of nucleosomes from the fragmentation seen in cfDNA molecules.
Then the team looked at the fragmentation of cfDNA from healthy individuals, and the associated nucleosome patterns, which they found correlated most strongly with epigenetic features of lymphoid and myeloid cells. This is consistent with the hematopoietic cell death one would expect to be the primary source of cfDNA absent a cancer or other disease in the body.
Finally, the researchers looked at blood samples from five cancer patients, and observed, as they had hoped, that different fragmentation/nucleosome fingerprints in the cell-free DNA could indeed be traced to different types of cancer. Moreover, for some of the cancers, the researchers could also identify the source of the tumor in a particular part of the body.
The five cancer patients in this proof-of-concept were chosen specifically because they had a high burden of circulating tumor DNA in their blood. This somewhat "stacked the odds" in their favor, the authors wrote. However, the team's goal for the technique is not necessarily to outperform mutation-based cfDNA analysis, but rather to offer a complementary approach that could answer different clinical questions, such as what the tissue of origin is in a case of cancer with an unknown primary site.
"This could be particularly relevant in the five percent of metastatic cancers whose original source is unknown," Shendure said in a statement, adding that the test "could aid in diagnosing what kind of cancer it is and to help guide treatment."
More remotely, the results also hint at the approach's potential to inform on a much wider variety of conditions in which cells die and release DNA into circulation, including heart attacks, strokes, and autoimmune diseases.
"Contributions from these tissues to cfDNA cannot be readily detected under the current paradigm of analyzing genotypic differences, which are effectively non-existent in these conditions," the authors wrote. "By contrast, the approach presented here should generalize to detecting contributions to cfDNA from any non-hematopoietic cell lineage.
Overall, the results were limited by small sample sizes and reference data, the authors wrote. Analyzing more samples representing more physiological states and diseases will be necessary to fully evaluate the potential and limitations of their approach, they said.