INVESTIGATION OF BLOOD-BASED BIOMARKERS AND PATTERNS OF DNA METHYLATION IN TUMORS
Embargo until
Date
2020-03-20
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Johns Hopkins University
Abstract
One epigenetic hallmark of many cancer types is differential DNA methylation at multiple loci compared to normal tissue. Detection of cancer specific DNA methylation, particularly in the context of a liquid biopsy, could provide an effective and minimally invasive cancer diagnostic metric. We have recently identified the CpG island promoter region of ZNF154 as a hypermethylated locus in 15 solid epithelial tumor types from 13 different organs, making it a promising multi-cancer biomarker for use in a blood-based cancer detection assay. We used methylation array data from 9817 solid tissue samples, followed by a combination of bisulfite sequencing, high resolution DNA melt, and droplet digital polymerase chain reaction (ddPCR) assays on 304 plasma samples in total representing 16 different tissue types and noncancer disease states to assess the diagnostic performance of ZNF154 methylation. We found ZNF154 hypermethylation relative to controls in early stages both at the tissue level and in patient plasma. Notably, we achieved 100%/80% sensitivity/specificity using plasma from patients with early stage pancreatic adenocarcinoma. However, we observed heterogeneous methylation at the ZNF154 locus in both cases and controls, suggesting further work is needed to establish reliable cutoffs to account for background methylation and improve the use of methylated biomarkers in liquid biopsy-based assays in general. We developed a methylation density binary classification method called EpiClass that can be used to predict and optimize the performance of methylation biomarkers, particularly in challenging, heterogeneous samples such as liquid biopsies. Finally, we sought to identify molecular commonalities between a subset of tumors characterized by a CpG Island Methylator Phenotype (CIMP) across different tissues. Identification of a common etiology of these tumors may define new mechanistic targets in cancer treatment and help reduce the heterogeneity seen in collections of tumor samples. We found that CIMP-positive colorectal and uterine tumors are significantly associated with hypoxic gene expression profiles, suggesting the tumor microenvironment may predispose tumors toward CIMP. We also explored drug sensitivity and metabolic profiles of the extensively characterized NCI-60 cell line dataset to identify compounds specifically sensitive to CIMP tumors. We identified several compounds with significant CIMP sensitivities for additional in-vivo testing.
Keywords:
ZNF154; multi-cancer marker; pan-cancer screening; liquid biopsy; DNA methylation; CpG island methylator phenotype; cell-free DNA; ovarian cancer; pancreatic cancer; cancer diagnostics; intramolecular variation
Description
Keywords
ZNF154, multi-cancer marker, pan-cancer screening, liquid biopsy, DNA methylation, CpG island methylator phenotype, cell-free DNA, ovarian cancer, pancreatic cancer, cancer diagnostics, intramolecular variation