By Vivien Marx
A clinical genomics networking system under development at the Harvard Medical School-Partners HealthCare Center for Personalized Genetic Medicine passed a key milestone last week when it was used to electronically transmit the results of a patient's genetic test from a lab in Boston to a health center in Utah.
The system, called VariantWire, is envisioned as a networking hub that will allow labs to securely transfer genetic test results and integrate that data with electronic health records.
VariantWire has been under development for "some time, but this is the first instance it has actually been used to transfer a live test result," Sandy Aronson, director of IT at PCPGM, told BioInform. "There was a huge amount of validation testing" to ensure that the system functions and transfers the information accurately.
The results belonged to a Utah man who was tested at PCPGM for an inherited heart condition, hypertrophic cardiomyopathy. The results were transmitted to Intermountain Healthcare in Utah where they were incorporated into the patient's electronic health record.
"This effort will likely serve as a model for the rest of the nation," Aronson said. "It’s a major advance in the area of electronic medicine."
Building a Hub
Genetic testing involves many different labs and tests, but it is "infeasible to think that every lab is going to be connected to every provider, so it's kind of a scenario that calls out for a hub architecture," Aronson said.
Genetic test results are made up of the detected variants, "plus the interpretation of what any variants identified mean," he said.
In the case of last week's transmission, Grant Wood, senior IT strategist with Intermountain's clinical genetics institute, told BioInform that the test results comprised approximately "13 pages of solid data."
Aronson said that VariantWire is a "new piece of infrastructure" that is built on top of a "generic data-movement hub" called Ensemble sold by InterSystems, a privately held firm in Cambridge, Mass.
The VariantWire system is related to several other computational infrastructure projects underway at PCPGM, such as the Gateway for Integrated-Genomic Proteomic Applications and Data, or GIGPAD, which serves as a LIMS umbrella system. Other projects include the GeneInsight database and the Genomic Variant Interpretation Engine, or GVIE.
Genetic test results must go through preprocessing before they can hit the VariantWire hub, Aronson explained. First stop is GIGPAD, which "figures out what variants are present" in the patient's DNA. Then information is drawn from the GeneInsight database, which contains the "correlations that have been established between genetic variations and clinically relevant facts," Aronson said.
GVIE pulls together the variants from the lab system and gathers information from the knowledgebase. "Then it creates the report in narrative and structured form, and geneticists review it and make modifications if necessary," he said. That report is what is sent on to VariantWire.
Meanwhile at Intermountain, the IT staff "wrote software to receive that message and to store it" in its clinical-data repository, which serves as a back end for its electronic health record front-end reporting system, Wood said.
VariantWire transmitted the results using a specialized version of Health Level 7, a standard format for the exchange and integration of electronic health information. The connection between the two institutions grew out of their collaboration on the HL7 Clinical Genomics Workgroup, which developed the HL7 Genetic Variation messaging model, Wood said.
Both Partners and Intermountain are active in this working group, Aronson said. While the HL7 standard is constantly evolving, "we worked with them to formulate this message" and refine the standard.
[ pagebreak ]
One Interface, if Possible
The idea behind VariantWire is to create a "simplified interface network" between healthcare providers who are ordering tests and the labs that perform them, Wood said.
Without the hub, "every ordering provider would have to create their own interface into every individual lab, which could turn into hundreds of interfaces for just Intermountain alone," he said.
Alternatively, if there is agreement on using VariantWire, "I only would have to have one interface into VariantWire," Wood said, adding that he hopes VariantWire will become a standard for hospitals and labs.
In order to manage patients' genetic profiles, Aronson said, clinical decision support systems "need to work on structured data."
"You need to make sure that you are able to get information from the laboratory to the clinician's system in a manner that clinical decision-support systems will be able to use," he said. As a result, setting up the network to make that transfer smooth was a priority for the project.
For the Boston-Utah data transfer, the test had been ordered on paper but in a next phase that, too, could be handled through secure electronic routing, Wood said.
The system is still at a "very early stage," he said. The developers hope to eventually enable "a results review screen" for doctors that will pull data from clinical-data repository and present it as part of a patient's electronic health record. "That is not in place right in this minute yet."
Wood explained that having genomic data in a "structured and coded" computer-readable format will help with these plans, which also include adding other clinical decision-making tools in the area of pharmacogenomics, such as an alerting system for new prescriptions and dosage information.
"Then we want to mix in the family health history information," and tie in other clinical information, such as prescriptions and lab tests relevant to that patient, Wood said.
In order to connect to VariantWire, a hospital needs to have an electronic health record in place, Aronson said. He and his team would work with that institution "on connecting that EHR to VariantWire, he said, adding that EHR could be a homegrown system or a vendor system.
With a clinical data warehouse that houses information on more than 5.5 million patients over 20 years, Intermountain Healthcare has a clinical informatics tradition. The group has developed its own electronic health record "and has had it in place since the 1970s," Wood said.
The healthcare system covers the state of Utah and southeastern Idaho and includes 24 hospitals, more than 100 clinics, 700 physicians, and 3,500 affiliated physicians, he said.
The system has 650 full-time IT staff, most of whom are in hospital operations. Eight IT staffers were involved in the VariantWire project, which took 14 to 16 months to complete, Wood said.
Wood said that Intermountain has faced some challenges in garnering financial support for the project, which is one reason it "has taken a little bit longer than we would have liked."
For example, an NIH Challenge grant application for further development of the system "did not get funded," he said.
Intermountain is also partnering with GE Healthcare on the development of a clinical-information system, with the goal of releasing a product in around two years. While the VariantWire project is not part of the GE Healthcare partnership, Wood said that Intermountain has relayed its work with PCPGM to GE Healthcare and they "have every intention" to incorporate it into the final clinical information system under development.
In another partnership with Microsoft HealthVault, Intermountain is developing a "patient-facing" family health history program, Wood said.
As Intermountain continues to expand its EHR system to include genetic data, Wood anticipates new computational issues such as storage. "That is a big … question we're working on," he said.
He and his colleagues are discussing the types of data that must be stored in a repository. "Right now we're just doing single-gene or multi-gene tests, so storing that is no big deal at all," he said.
However, he noted, when a patient's DNA is sequenced in its entirety, that might be 100 to 200 gigabytes of data per person — a challenge for database design and storage that "we're still researching," he said.