Here and there, the genetic information of a cancer patient has helped a doctor find that person the right drug or steer her away from the wrong one. But the use of genetic knowledge to improve people’s health is in its infancy.
That’s why seven major cancer treatment centers in North America and Europe are pooling their patients’ data in a central repository, dubbed Project GENIE (Genomics, Evidence, Neoplasia, Information, Exchange), for their own researchers and doctors to use and, eventually, to open completely to all comers. “Right now genetic testing is getting done for individual patients and being returned to those patients’ physicians,” says Victor Velculescu, co-director of cancer biology and a professor at Johns Hopkins University who has helped push forward the field of cancer genomic analysis. “Those data are not being used, except to help that patient. This effort brings all data from all reports together and allows us to keep learning.”
The bet with this project—and writ larger, with the promise of so-called big data in all manner of medical fields, not just cancer—is that the more peoples’ information is pooled for comparison and analysis, the deeper the insights into health.
“It’s reasonable to think that it could have the records of 50,000 to 100,000 patients in several years,” says Charles Sawyers, one of the nation’s top cancer researchers and the chair of the human oncology and pathogenesis program at New York’s Memorial Sloan Kettering Cancer Center. “That’s a lot.”
Sawyers hinted earlier this year to Xconomy that the massive data sharing project was underway and could launch in the fall.
Sloan Kettering, Johns Hopkins’s Kimmel Cancer Center in Baltimore, Dana-Farber Cancer Institute in Boston, Princess Margaret Cancer Centre in Toronto, Vanderbilt-Ingram Cancer Center in Nashville, TN, Institut Gustave Roussy in France, and the Center for Personalized Cancer Treatment in the Netherlands are the seven participating institutions. Velculescu says the consortium could welcome more in due time.
One of the first projects GENIE could tackle is what Dana-Farber chief scientific officer Barrett Rollins calls a “meta” problem. Does decoding the genes of patients’ cancer actually lead to better health outcomes? Dana-Farber and other institutions have their anecdotal success stories; Farber has a handful of what it calls “Lazarus cases,” such as a young man with a seemingly intractable form of leukemia, who was at death’s door when a sequence of his cancer pointed toward treatment with the drug imatinib (Gleevec), according to Rollins. “Now he and his wife are having a baby,” says Rollins.
But no one has examined whether more sequencing, across large populations, is worthwhile. “This is one of our best hopes to demonstrate what we all intuitively feel is true,” says Rollins, noting that insurance companies aren’t keen on paying for tests that don’t have a body of positive health outcomes behind them.
The participants are all well equipped with the labs and machinery, not to mention the flow of patients, to contribute to the pool of data, but some are contributing more than others. Supported with philanthropic money that subsidizes its data efforts, Sloan Kettering sequences 410 genes from the tumor of some of its patients—mainly those with metastatic cancer. Dana-Farber looks at 405 genes, but it does so for every cancer patient. More financially constrained, Hopkins sequences about 50, and hopes to double that in the future, says Velculescu, but it relies on insurance and reimbursement to pay for its data work.
These disparities are a reminder that, while the cost of sequencing has plummeted in the genomic age, there are still financial limits to how much an institution can do. The differences also underscore a practical problem: How to combine all these data into one searchable pool? Not only are the data sets different sizes, but each institute might sequence a particular gene in different ways—focusing on one mutation over another, say, or sequencing the same site with varying levels of redundancy for quality control.
To help create a framework, GENIE has turned to Sage Bionetworks in Seattle, a nonprofit group that helps groups share scientific data. Part of the job is not just melding together existing data but accounting for future changes in each institution’s panels, or sets of genes, under examination. “These panels have to be living entities, driven by what the scientific
