Intel to pilot cloud technology for sharing personalized cancer treatment
- 20 August, 2015 22:55
At 19, Eric Dishman began a fight with kidney cancer, and for 23 years he endured what he described as Russian roulette chemotherapy.
It wasn't until he had his DNA sequenced that doctors were able to administer a personalized treatment that placed Dishman's cancer into remission.
The problem is, even after his genome was uncovered, it took physicians seven months to come up with a treatment -- a period of time when Dishman said he was on "death's door."
Intel and the Knight Cancer Institute at Oregon Health & Science University (OHSU) this week launched a pilot network that will allow healthcare facilities to securely share genomic data for tailoring cancer research and personalized medicine.
With the new Collaborative Cancer Cloud, large amounts of patient genomic data from sites all over the world can be analyzed in a distributed way while preserving the privacy and security of the patient data at each site. The research cloud should scale "needle in the haystack" searches for personalized treatments that have worked on patients with similar cancers.
Instead of going through chemotherapy that can kill healthy cells along with cancerous cells, personalized or "precision" medicine treats patients as individuals based on their specific genome sequencing.
Key to the cancer cloud is Intel's plan to make open-source its Trusted Execution Technology in the first quarter of 2016. The firmware allows server CPUs to exchange encrypted data, which in turn protects patient privacy.
Dr. Brian Druker, director of OHSU's Knight Cancer Institute, said the Trusted Execution Technology will allow healthcare centers to maintain control of patient data, while also allowing clinics around the world to exchange data for genomic analysis.
"If we're seeing 5,000 new cancer patients a year... many of them may get sequenced three or four times a year, we won't be able to ship all that data to a central location," Druker said. "And that's just cancer. We have all our Alzheimer's, MS patients and cardiac patients. We'll be sequencing 1,000 patients a day in our clinics."
Dishman, who went on to became an Intel fellow and the general manager of Intel's Health & Life Sciences Group, said the problem with his precision cancer treatment was that he first had to wait three months while computers analyzed his genome to find the flaw in his DNA that could be addressed. The second four months were taken up by clinics shipping hard disk drives with a terabyte of genomic data back and forth across the country to tailor Dishman's treatment.
"Fortunately, I survived that process long enough for them to come back with an answer in time. When you're talking about advanced cancer patients it's unacceptable for this to take months... even weeks can have deadly consequences," Dishman said.
Today's hospitals aren't set up to share electronic health data. While the federal government has required healthcare facilities to roll out electronic healthcare records, the technology to share data even within hospital groups is either non-existent or in the nascent stage. And, for oncology and other specialized medical care, sharing data is even more difficult, as it calls for specialists around the country to first find each other and then share data about specific cancer types.
The first human genome took 13 years and $3 billion to produce. Today, geneticists can generate the same information in hours and for under $1,000.
But genomic data is still massive in size -- a terabyte or more of information.
"With less than 2% of cancer patients in the world being sequenced, it's already an incredibly difficult prospect to actually share the data," Druker said.
The Collaborative Cancer Cloud works by allowing physicians to send out secure electronic queries to other healthcare facilities seeking patients with cancers and genomics similar to their own patients. Once a match is discovered, all electronic patient data is made anonymous before the data is analyzed.
"The moment an answer is given back through the secure container, the virtual machine disappears and all the data used to come up with an answer to the query disappears," Dishman said. "So it's not like the cancer center in Texas can peer into raw cancer data at OHSU."
OHSU has plans to go live in the first quarter of 2016 with two other cancer centers to pilot the Collaborative Cancer Cloud technology. If successful, the plan is to then wrangle more healthcare centers into the network to share the collaborative technology.
"We hope dozens, if not hundreds of cancer centers will use ours or use the open-sourced technology to make their own data sharing platforms," Dishman said. "At the end of the day, we want to at least remove barriers that are keeping precision healthcare from scaling."