Conference Proceeding

Characterizing a biomarker of pan-cancer importance for use in biofluid diagnostics

Dr. Laura Elnitski,

Dr. Laura Elnitski received her Ph.D. from The Pennsylvania State University in the area of Biochemistry and Molecular Biology. She later carried out her post-doctoral research at The Pennsylvania State University, in the area of Computational Biology. Currently she serves as a Tenured Principal Investigator at the National Human Genome Research Institute at the National Institutes of Health, Bethesda, MD, USA. She has many honors and awards, and serves on many scientific review committees. Her research involves the study of tumor epigenomes, with a focus on computational methods to identify tumor subclasses by their DNA methylation landscapes. Her research accomplishments include the identification of the first candidate pan-cancer biomarker for use in cancer diagnostics, using a blood-based or liquid biopsy.

Introduction: Cancer diagnostics is moving into an area of noninvasive screening and rapid detection through the emerging technologies of blood-based assays and DNA sequencing. Examples include isolation of circulating tumor cells and circulating tumor DNA. The challenges in moving these methodologies from the laboratory into the clinic include the diversity of somatic mutations found in different types of cancer, the cost of implementing sequencing approaches for individual patient samples, and the limited amounts of starting material available from a blood draw. In light of these considerations, alternatives to mutation based screening were considered. We previously identified a putative pan-cancer marker in 15 tumor types from 13 different organs and we now examine the feasibility of use in a diagnostic application.
Materials and Method: DNA methylation was assessed from five types of tumors including184 tumors, plus normal tissues for each type. Bisulfite conversation, PCR amplification and squencing were used to detect DNA methylation in a 300 base pair region of the genome. Compuational analyses were applied to determine the best classification method to distinguish tumors from normal samples.
Results: We measured the magnitude and pattern of differential methylation of the ZNF154 CpG islandin colon, lung, breast, stomach, and endometrial tumor samples and found that all tumor types and subtypes are hypermethylated at this locus compared with normal tissue. To evaluate this site as a possible pan-cancer marker, we developed several sequence analysis methods to determine whether we could distinguish the five tumor types from normal tissue samples. The classification performance for the strongest method, measured by the area under the receiver operating characteristic curve (AUC), is 0.96, close to a perfect value of 1. We then performed a computational simulation of circulating tumor DNA, in which we mixed the methylation signals from individual tumor samples and normal samples in ratios up to 1% tumor DNA in 99% normal DNA. Analysis of these data addressed whether we could detect limited amounts of tumor DNA diluted with normal DNA, and yielded classification performance AUCs of up to 0.79, which is deemed as highly promising.
Conclusion: The amount of tumor DNA circulating in the blood varies from less than 1% to more than 50%. Therefore, our findings suggest that in a blood-based assay, we can reliably expect to distinguish tumor samples through hypermethylation of the ZNF154 CpG island compared to normal samples. As such, this biomarker has potential utility for blood-based cancer screening, in the attempt to catch many types of cancer earlier than with conventional methods.

Published: 11 May 2017