Editors' ChoiceCancer

Following bloody footprints: Cancer diagnosis from cell-free DNA

See allHide authors and affiliations

Science Translational Medicine  30 Oct 2019:
Vol. 11, Issue 516, eaaz3724
DOI: 10.1126/scitranslmed.aaz3724


Nucleosome footprinting in circulating tumor DNA can identify active tumor-specific transcription factors.

Although tissue biopsies continue to be the gold standard for confirming a cancer diagnosis, less invasive liquid biopsies from plasma are emerging as increasingly reliable screening and staging tools. The basis of liquid biopsy performance is the shedding of DNA that accompanies the normal turnover of cells. In patients with cancer, the volume of cell-free DNA (cfDNA) tends to increase with tumor burden, as DNA of cancer origin represents the population of cells most actively dying and dividing. When sampled from plasma, detection of circulating tumor DNA (ctDNA) is largely based on next-generation sequencing of cfDNA and identifying somatic mutations that arose from tumor cells.

Most ctDNA detection platforms use hybrid capture enrichment to rapidly and preferentially sequence coding regions of oncogenes or tumor suppressors that can guide precision therapies, especially in the absence of a tissue biopsy. In this study, Ulz et al. used low-coverage whole-genome sequencing of plasma cfDNA from healthy volunteers and patients with cancer to identify the patterns of sequencing that accompany transcription factor (TF) binding, or footprinting. The occupancy of a TF on chromatin prevents nucleases from cleaving DNA at that site. Although the vast majority of cfDNA molecules are heavily fragmented, the authors found this fragmentation was not random. First, the authors devised a method to determine the binding accessibilities of 504 TFs and used sequence read depth patterns to assign scores of normal or abnormal binding site accessibility for each TF. Then, using a series of plasma samples from patients with prostate cancer, the authors demonstrated increased accessibility of binding sites mapping to AR, HOXB13, and FOXA1, transcription factors known to function in prostate cancer development. They showed similar results with samples from breast and colon cancer patients.

The authors showed in four patients with prostate cancer that the pattern of TF binding changed as tumors became resistant to treatment. In particular, the switch from hormone-dependent adenocarcinoma to hormone-independent neuroendocrine prostate cancer was accompanied by a decrease in the cfDNA accessibility of AR and HOXB13, which are down-regulated in treatment-resistant prostate cancers. Last, Ulz et al. developed a classifier to predict stage I or II colon cancer in 592 patients based on the accessibility of TF binding sites in cfDNA. As these methods are still quite new, future validation studies will determine whether treatment decisions based on earlier detection of cancer from cfDNA improves long-term survival of patients.

Highlighted Article

View Abstract

Navigate This Article