NEW YORK (GenomeWeb News) – A flurry of new studies is outlining findings from the first phase of an analysis that brings together Cancer Genome Atlas data on up to a dozen cancer types.
Members of the Pan-Cancer Initiative scoured sequence data for thousands of samples assessed by TCGA to look for genomic features, oncogenic signatures, and potential treatment targets shared between multiple or many cancer types.
Indeed, results of their studies suggest that while there are some mutations that tend to characterize cancers arising from a given site in the body, many of the genetic alterations contributing to cancer are shared between tumors in ways that are independent of that tissue-of-origin.
"For years we've been looking at one tumor type at a time, but there are patterns you can only spot by making connections across different tissues and tumor types," University of California at Santa Cruz researcher Joshua Stuart said in a statement.
"Finding these similarities across tissues can have important implications for treatment," said Stuart, a co-organizer of the TCGA's Pan-Cancer effort and corresponding author on a commentary in Nature Genetics that describes the project.
By combining information from a large tumor set, Pan-Cancer researchers were also able to pick up cancer-related hiccups in the genome that were missed previously, including recurrent amplifications and deletions that appear to interfere with the function of chromatin-related pathways.
Investigators involved in the ongoing effort outlined their approaches — together with some early findings from it — in 18 studies and commentary articles appearing in Nature Genetics and elsewhere.
In one of the studies, for example, researchers from Memorial Sloan-Kettering Cancer Center and elsewhere searched for oncogenic signatures in almost 3,300 tumor samples from 12 cancer types mapped by TCGA members prior to the Pan-Cancer group's December 2012 data freeze.
After combining copy number, exome sequence, methylation, and other information to get a list of so-called "selected functional events," or SFEs, authors of that study used a newly developed computational tool to classify tumors genomically at increasingly finer scales of resolution.
From the 479 SFEs detected in TCGA tumor samples, for example, the team found that tumors fell into two broad categories: one characterized by rampant copy number changes and another marked by a preponderance of somatic mutations.
While some tumors had intermediate levels of both genetic glitches, researchers reported, those at the extreme end of the copy number alteration spectrum tended to contain relatively few somatic mutations and vice versa.
These two broad groups contained dozens more cancer sub-classes, too, reflecting mutational events at play within the tumors. Rather than seeing clear genomic distinctions between tumors from different cancer types or tissues of origin, though, the team saw several instances of genomic groups and sub-groups housing multiple tumors types.
Such patterns highlight the potential of treating tumors in a genomics-based manner, according to the study's authors, who have already started mapping cross-tumor alterations that may be susceptible to targeted treatments.
"In future clinical trials, we envision that patients with a certain type of endometrial cancer, for example, may be enrolled in the same trial as patients with a subtype of colorectal cancer," Memorial Sloan-Kettering Cancer Center researcher Chris Sander, co-senior author on the oncogenic signature study, said in a statement, "and that patient selection for clinical trials can be guided by cancer genomics profiling in the clinic."
"This work is intended to help in the design of such trials," Sander added, "and the development of more personalized cancer therapies."
In another Nature Genetics study, meanwhile, members of the Pan-Cancer project from the Dana-Farber Cancer Institute, the Broad Institute, and other centers used copy number data for nearly 5,000 tumors to take a closer look at the nature of copy number changes within and across 11 cancer types.
That analysis revealed 140 somatic copy number alterations — 70 amplifications and 70 deletions — that appear to be recurrent across cancers from all of the types tested.
A few dozen of those overlapped with parts of the genome containing oft-mutated cancer genes, authors of that study found. But more than 100 of the recurrent amplifications or deletions affected parts of the genome that don't code for known tumor players. Instead, many of those copy number alterations affected sequences suspected of contributing to chromatin function or other forms of epigenetic regulation.
In addition to commentary articles in Nature Genetics outlining the strategy and data-storing scheme behind the Pan-Cancer effort, related articles are slated to appear in other journals this week. Dozens more papers from the Pan-Cancer arm of TCGA are reportedly in earlier stages of the publication pipeline.
Going forward, Pan-Cancer group members plan to continue analyzing data for tumors analyzed through TCGA — an effort that's expected to reveal CNV, oncogenic signatures, and other mutation profiles from across 20 cancer types or more.
"These initial papers are just the first step, and we expect much more to come from the Pan-Cancer Initiative," UCSC's Stuart said. "With the infrastructure now in place, we can scale up to look at more types of data, especially whole-genome sequencing data, and to include many more tumor types, including rare tumors."