NEW YORK (GenomeWeb) – A difference in size between circulating cancer DNA and normal cell-free DNA may improve the ability of liquid biopsies to pick up tumor DNA in patients' blood.
Liquid biopsies — in which blood samples are analyzed to detect and diagnose cancer as well as monitor its recurrence or response to treatment — have been hampered by sensitivity issues. But as researchers from the University of Utah and the University of Washington reported in PLOS Genetics today, DNA fragments originating from tumors tend to be shorter than fragments from normal cells, a feature that could be taken advantage of to improve liquid biopsies.
"This development has the potential to enable earlier detection of solid tumors through a simple blood draw by substantially improving our ability to detect very low quantities of circulating DNA derived from tumor cells," first author Hunter Underhill from Utah said in a statement. Underhill began the work while in Jay Shendure's lab at UW.
Underhill and his colleagues first noticed a difference in circulating tumor and normal DNA in an animal model of glioblastoma multiforme. A human GBM stem-like cell line had been implanted into rat brains and lesions formed. Human circulating tumor DNA (ctDNA) was found in the rats, but the researchers noted a size difference between the human tumor and normal rat DNA fragments: the human ctDNA was commonly 134 basepairs in length, while the rat cell-free DNA (cfDNA) was 167 basepairs in length. The human ctDNA also exhibited a 10-basepair periodicity that the rat cfDNA did not.
This pattern, they noted, was present in all rats in which human ctDNA could be detected and held in a follow-up study of another GBM line implanted into rats. The researchers likewise detected a difference in DNA fragment size between circulating human ctDNA from rats implanted with human hepatocellular carcinoma and rat cfDNA. This indicated to the researchers that the pattern wasn't restricted to glioblastoma and could be a general feature of ctDNA.
To determine whether it was a species-specific finding and limited to animal models, Underhill and his colleagues examined the length of circulating DNA from melanoma patients and compared it to that of healthy controls. Via densitometry they found that melanoma patients had globally shorter DNA fragment lengths.
The researchers also sequenced samples from a melanoma patient with an elevated level of cfDNA and compared the data to a pooled sample of cfDNA from healthy controls. Fragments from the melanoma patient were, again, typically shorter — about 20 basepairs — than controls. There was also some indication of fragment length periodicity in the melanoma patient cfDNA. Further, the researchers noted that among the cfDNA from the melanoma patient, the BRAF V600E allele was more frequently present among the shorter fragments than the wild-type allele.
In a cohort of 15 lung cancer patients and nine healthy controls, the researchers likewise found that cfDNA from lung cancer patients was typically shorter than that from controls, through both densitometry and sequencing. In a subset of cases and controls, the researchers also used a gene capture approach combined with sequencing, and found that shorter fragments were more likely to house mutant alleles associated with lung cancer such as the EGFR T790M mutation.
To test whether enriching for smaller fragments beefs up the detection of tumor DNA, Underhill and his colleagues put together cell-free DNA sequencing libraries from four lung cancer patients with EGFR T790M mutations and a healthy control. Three of the four cancer patient samples exhibited an increase in mutant allele frequency when the researchers isolated DNA fragments that were some 20 basepairs to 50 basepairs shorter than the overall average. There was a 2.5-fold to 9.1-fold increase in mutant allele frequency, the researchers added, noting that it was particularly improved for the two patients who had low levels of short DNA in circulation.
This suggests that selecting for shorter cfDNA might improve the portion of mutant alleles detected and improve the detection of low burden tumors, the investigators said.
"It's possible that [a] jump in sensitivity could make the difference between being able to detect a cancer, and not," Underhill added. However, he also noted that the mechanism behind this pattern is not yet clear.