Most comprehensive map of human epigenomes helps identify cancers of unknown origin

19 Feb 2015

by ecancer reporter Janet Fricker

A collection of 24 papers from the Roadmap Epigenomics Program (REP), published online in Nature and six other journals, provides the most comprehensive catalogue of epigenomic data from primary human cells and tissues to date.

Of particular note for oncologists was the study providing a new approach to identify the primary site of metastatic cancer of unknown origin.

Launched in 2008, the REP is a public resource of human epigenomic data to catalyse basic biology and disease orientated research.

Mapping for the 10 year $240 million US government research program took place in four Roadmap Epigenomics Mapping Centers (REMC).

Just as the Human Genome Project provided a map of the genes of the human genome, REP aims to provide information on how genes are interpreted in different ways in different cells and tissues.

“Although all cells have a copy of the same book they’re reading different chapters, and booking different pages and highlighting different paragraphs and words. The human epigenome is the collections of marks placed in each cell type,” explained Manolis Kellis, at a Nature press briefing to publicise the papers.

Such chemical epigenetic ‘marks’ could be the addition of methyl groups in or near genes, or modifications to histones which can determine whether genes are available to be transcribed and translated into proteins.

Although epigenetic marks are stable, they are reversible, and can be altered by environmental factors, such as diet and exposure to toxins.

Such changes influence gene expression and can lead to disease.

In one paper Kellis and colleagues, from MIT, detailed 111 distinct human epigenetic patterns across different tissues and cell types.

The data, wrote the authors, provides the "the most comprehensive map of the human epigenomic landscape so far".

In a related paper, Shamil Sunyaev and colleagues from Brigham & Women’s Hospital, analysed a total of 173 cancer genomes from cancer types and compared them to over 400 epigenetic features measured by the consortium.

Cancers examined included melanoma, multiple myeloma, lung adenocarcinoma, liver cancer, colorectal cancer, glioblastoma, oesophageal adenocarcinoma and lung squamous cell carcinoma.

“We showed that mutational patterns found in cancers closely matched epigenetic patterns of the normal cells that gave rise to the cancer,” said John Stamatoyannopoulos, one of the authors.

The finding, he added, has important clinical implications.

“Every year there are thousands of patients who show up in oncology clinics with metastatic cancers where we do not know the primary site. This work opens the door for sequencing the patient’s tumour to identify the cell type of the primary cancer.”

The investigators found they could predict with nearly 90 per cent accuracy where metastatic cancer originated, which is unknown in 2 to 5 per cent of patients.

Directions for future work, the press conference heard, include analysing epigenomic variations between individuals associated with disease, and modifying them to elevate symptoms.

Furthermore, understanding how the epigenome varies with disease state could identify biomarkers to characterise disease in different individuals and identify targeted therapies.

“Instead of the epigenome map being the end I very much see it was beginning a decade of epigenomics,” said Kellis. 


P Polak, R Karlic, A Koren, et al. Cell-of-origin chromatin organization shapes the mutational landscape of cancer. Nature. 

Epigenomics Consortium. Integrative analysis of 111 reference human epigenomes. Nature.