19026 – Minimalist Approaches to Cancer Tissue-of-Origin Classification by DNA Methylation

The technology developed by lead PI Dr. Daniel Xia at UHN’s Princess Margaret Cancer Centre and Dr. Alberto Jose Leon Arroyo at the Ontario Institute for Cancer Research presents a so-called minimalist approach to cancer classification by DNA methylation. The technology uses a small and focused set of methylation biomarkers to scale down methylation testing and adapt it to more popular platforms, including PCR and targeted NGS-based sequencing. The test is expected to be less expensive than array-based methods, without significantly compromising accuracy– making it a suitable test for all laboratories, including smaller ones within the community setting. In comparison to standard immunohistochemistry used in pathology, the methylation biomarkers identified are sensitive and precise. The inventors have developed an extensive series of classifiers, which are able to accurately classify cancers using small numbers of CpG sites. Of particular significance, they have devised a pan-cancer classifier that uses information from just 53 CpG sites and was able to achieve an accuracy of 94% across 28 different cancer types in a validation dataset from The Cancer Genome Atlas (TCGA). One of the primary applications envisioned for this pan-cancer classifier is the development of targeted NGS-based testing for the classification of cancers of unknown primaries (CUPs), which are cancers that have already metastasized, but where the primary tumour site cannot be identified by routine laboratory pathology tests. This pan-cancer classifier presents an accurate and inexpensive method to determine the primary tumour site of CUPs and therefore allow oncologists to develop much more informed treatment plans for these patients. Another classifier developed used information from only 5 CpG sites to classify 3 types of cancer, with an accuracy of 93%. Currently available methods for DNA methylation array profiling, while medically and diagnostically informative for cancer tissue-of-origin testing, are quite costly and technologically challenging to implement on a large scale as a new platform. This technology proposes that cancer tissue-of-origin testing by DNA methylation can be substantially simplified in an inexpensive and accurate manner, through the use of sets of only the most informative CpG markers and widely available platforms such as targeted NGS and quantitative PCR. Andrew Roberts andrew.roberts@uhnresearch.ca 4166348033