Federated Network Could Share Data Across Cancer Centers
"This new network model provides the technology, legal agreements and standards needed to use de-identified data and tissue specimens across institutions."
- A new element in the national health IT infrastructure could be a system that facilitates data and biospecimen sharing among U.S. cancer centers. A team led by scientists from the University of Pittsburgh School of Medicine described in the Dec. 15 edition of Cancer Research a federated network that uses advanced text processing.
The work stems from previous development of the Text Information Extraction System (TIES) for language processing. The new study proposes the TIES Cancer Research Network (TCRN) as a model to promote translational research across all cancer centers. The study notes that as the need for personalized therapies and precision medicine increases, more sophisticated systems to facilitate the sharing of data and tissue samples will be essential.
“With the TCRN, we can study rare diseases and rare behaviors of common diseases much more effectively. Investigators may not have enough cases at a single institution to support a compelling study, but they can now aggregate and access data and biomaterials across multiple institutions,” said Rebecca Jacobson, MD, professor of biomedical informatics at Pitt School of Medicine.
Researchers from Pitt, Georgia Regents University Cancer Center, Roswell Park Cancer Institute and the Abramson Cancer Center at the University of Pennsylvania collaborated on the project.
In previous efforts, cancer researchers from various institutions typically worked through a centralized network in which one facility accepted all relevant data and materials and made it available to the others. As such, each study required its own technology infrastructure and agreements to operate.
The new study set out to create a single infrastructure that could be used for many studies — across many institutions — without moving any of the data. Member sites can access pathology data that are de-identified and processed with the TIES natural language processing system, which creates a repository of rich phenotype data linked to clinical biospecimens. TIES incorporates multiple security and privacy best practices that, combined with legal agreements, network policies, and procedures, enable regulatory compliance, according to the study.
“The centralized model cannot scale to a national network,” noted Jacobson. Every new study or new institution means more work for the central broker, and institutions don’t want to cede their authority to manage their own data. This new network model provides the technology, legal agreements and standards needed to easily use de-identified data and tissue specimens across institutions. You can think about it like a superhighway for data and biomaterial sharing, helping researchers get there much, much faster.”
TCRN now has active nodes at four member institutions, enabling searches across nearly 6 million cases and 2.5 million patients, the study states.
“Future advances in cancer research and precision medicine will require significant new national infrastructure, including more robust biorepositories that link human tissue to phenotypes and outcomes, especially as distinctions among molecular subtypes become increasingly refined,” the study authors conclude. “We envision that the TIES Cancer Research Network could provide the foundation for a national federated network for cancer data and biospecimen sharing, a goal that now seems within reach.”
