Identify Tissue-Of-Origin In Cancer Cfdna By Whole Genome Sequencing

Cell free DNA (cfDNA) has been shown to be an emerging non-invasive biomarker to monitor tumor progression in cancer patients. Elevated cfDNA has been found not only from tumors, but also from normal tissues. Thus, the identification of cfDNA’s tissue-of-origin is critical to understand the mechanism of cfDNA release and tumor progression. Recent efforts to identify cfDNA’s tissue-of-origin begin to utilize cfDNA’s epigenomic status, such as DNA methylation and nucleosome spacing. However, both of these methods have limitations: (1) For nucleosome positioning, lack of reference nucleosome maps in different tumor and normal tissues has limited its application to tissue-of-origin deconvolution; (2) For DNA methylation, large DNA degradation during whole genome bisulfite sequencing (WGBS) library preparation, even with current low-input DNA technology, is still the major hurdle for its clinical application, although extensive DNA methylation studies by WGBS in tumor and normal tissues during the last decade have provided many reference maps. Very recently, a pioneer study showed significant differences between DNA fragment lengths of methylated and unmethylated cfDNA. Taking advantage of this experimental observation, we developed a machine learning approach to infer the base pair resolution DNA methylation level from fragment size information in whole genome sequencing (WGS). The predicted DNA methylation, from not only high coverage but also dozens of ultra-low-pass WGS (ULP-WGS), showed high concordance with the ground truth DNA methylation level from WGBS in the same cancer patients. Furthermore, by using hundreds of WGBS datasets from different tumor and normal tissues/cells as the reference map, we deconvoluted cfDNA’s tissue-of-origin status by inferred DNA methylation level at ULP-WGS from thousands of breast/prostate cancer samples and healthy individuals. The cfDNA’s tissue-of-origin status in cancer patients showed high concordance with confirmed metastasis tissues from physicians. Interestingly, some clinical information, such as cancer grades/stages, seemed to be correlated with cfDNA’s tissue-of-origin status. Overall, our methods here pave the road for cfDNA’s application in clinical diagnosis and monitoring. Citation Format: Yaping Liu, Sarah Reed, Atish D. Choudhury, Heather A. Parsons, Daniel G. Stover, Gavin Ha, Gregory Gydush, Justin Rhoades, Denisse Rotem, Samuel Freeman, Viktor Adalsteinsson, Manolis Kellis. Identify tissue-of-origin in cancer cfDNA by whole genome sequencing [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 5689. doi:10.1158/1538-7445.AM2017-5689