At first glance, the two tumor subtypes seem to have little in common: one takes root in the ovaries and the other, in breast tissue.
Conventionally, tumors like these are referred to by their organ of origin and other basic characteristics such as the cell type that spawned them.
But through projects like The Cancer Genome Atlas (TCGA), researchers from the Broad Institute and elsewhere are taking a deeper look at cancer’s many forms and finding genomic similarities that cut across these classifications, as well as great diversity within single classes of cancer.
In a paper published online Sunday in Nature, the TCGA network describes discoveries gleaned from characterizing samples from 825 breast cancer patients.
Breast cancer is a diverse disease with four primary subtypes. The new research suggests that one of these subtypes - Basal-like breast cancer -is genetically distinct from other breast cancer subtypes, but shares many molecular characteristics with high-grade serous ovarian cancer, a form of cancer that is also notoriously difficult to treat.
“The underlying genomic alterations in Basal-breast cancer were different than they were in the other breast cancer subtypes,” said Andrew Cherniack, a research scientist in the Broad’s Cancer Program and an author of the paper. “Molecularly, it’s a different disease.”
The Broad Genomic Characterization Center, led by Matthew Meyerson and Stacey Gabriel, provides all genomic copy number data for the entire TCGA.
In cancer, large pieces of DNA can get amplified (copied and pasted multiple times) or deleted, resulting in distorted genomic function.
Copy number analysis allows researchers to detect these large additions and subtractions. This analysis, along with other lines of evidence, helped the team see how distinct Basal-breast cancer was from other forms of the disease and to discover its similarity to high-grade serous ovarian cancer, which was characterized in depth by the TCGA last year.
In addition to processing and analyzing samples sent to the Broad, researchers at the Broad oversee the system known as Firehose, which is used to analyze all of the TCGA data generated not only at the Broad but also at the other TCGA centers.
Gad Getz, the director of Cancer Genome Computational Analysis at the Broad, and his team developed Firehose, which incorporates tools developed by several groups including those from the Broad, such as the software package GenePattern.
“With a dozen centers involved in the TCGA project, tools like Firehose can help us coordinate and consolidate data analysis,” said Gordon Saksena, a senior software engineer in Getz’s group and an author of the paper.
Saksena continued, “It’s great to have this much data for testing, refining, and improving our models. And, with so many samples, we can identify more genes that appear to be under positive selection within the tumor, and thus worthy of further study.”
In addition to breast cancer and ovarian cancer, the TCGA has also published analyses on glioblastoma multiforme, colorectal adenocarcinoma, and lung squamous cell carcinoma.
“As we work on more TCGA projects, we can begin to make associations among subgroups of different types of cancer,” said Getz. “These are the kinds of connections that we would never have been able to make without this level of genetic detail or this quantity of samples.”
Mike Noble, a software engineering manager in Getz’s group and an author of the breast cancer paper, describes the work of the TCGA as creating maps, and likens the first major papers from the TCGA on each of form of cancer as markers along a trail.
"These marker papers we've published on brain, breast, and lung cancer are like the first sign posts erected along trails that will be blazed with these maps," said Noble.
Noble continued, "And it's really just the beginning: once you establish your map, you start building your roads, you build your signposts, and then everyone wants to start traveling."