CHICAGO (GenomeWeb) – After four years of planning and construction, the Netherlands consolidated all of its pediatric cancer care into a new hospital and research hub last year, the Princess Máxima Center for Pediatric Oncology.
Billed as the largest pediatric cancer center in Europe — though one under construction in Barcelona, Spain, may soon top it — the Princess Máxima Center in the Dutch city of Utrecht presented a rare opportunity to set up an informatics infrastructure from scratch, with no legacy software. Those in charge of building the informatics have looked to support the hospital's mission of tightly integrating clinical care and cancer research by creating a two-way data flow that includes genomic data, patient records, and medical literature.
The hospital consolidates the work of seven different facilities across the Netherlands that used to treat the nation's 600 or so annual cases of pediatric cancer, putting the Princess Máxima Center in the class of St. Jude Children's Research Hospital in the US, according to Patrick Kemmeren, a bioinformatician and systems biologist at the Utrecht hospital.
"At the same time, we also moved all the research into one building to really facilitate and improve the speed upon which we find new things in research and get that implemented into clinical care, and also optimize the clinical care," said Kemmeren.
Already, the massive new facility has shown promise as a center of excellence. Kemmeren, who joined the Princess Máxima Center during the construction phase in 2016, said that some cancer surgeons used to operate only once every couple of weeks. Now, in the consolidated hospital, they perform the same procedures three to four times per week.
"We have already saved a lot of time in operations themselves. Operations which took like four hours now take only two hours to do the same procedures because they're more experienced in doing these kinds of procedures," Kemmeren said.
Every patient who comes into the Princess Máxima Center gets their tumor genome sequenced. Phenotypic data is collected in electronic health records — which includes an Epic Systems EHR as well as a system called HiX, made by Dutch vendor ChipSoft — as well as in the hospital's laboratory information systems. The institution runs CliniSys Group's GLIMS for hematology-oncology diagnostics and shares a Deutsche Telekom Lifecare pathology system with the adjacent University Medical Center Utrecht.
Although Princess Máxima Center is new, it is dealing with the same interoperability issues that plague hospitals across the Netherlands, Europe, and beyond. Kemmeren described it as "probably the same kind of mess as [in] the States" in trying to combine genotype and phenotype data.
EHRs both within the Princess Máxima Center and from referring clinicians and hospitals still contain a lot of documentation in free text, according to Kemmeren. It is the same issue that Boston-based Dana-Farber Cancer Institute is facing, according to a presentation at the Bio-IT World conference in Boston last month after Kemmeren presented there.
Kemmeren and his bioinformatics team are having to transform the free text into a structured, machine-readable format. "It is still a lot of manual labor, but there is some natural language processing going on," he said.
Since the beginning, Princess Máxima Center has looked to intermingle clinical and research operations as much as possible.
The institution set up a platform intended to make both research and clinical data available to any user with proper credentials. "It's a two-way data transfer in that sense," he noted.
Because childhood cancers tend to be rare, researchers often struggle to find enough patient samples and build large enough cohorts to conduct meaningful studies.
In addressing this and other data challenges, the Princess Máxima Center brought in The Hyve, an open-source bioinformatics software development company that also is based in Utrecht.
The Hyve is helping with data aggregation, integration, harmonization, and matching. "They have many different data sources, of course, in the hospital," including the EHR, biobank data, sequencing data, and information on legacy studies on childhood cancer, said Ward Weistra, the company's bioinformatics project manager. "They needed to bring this all together in one data warehouse."
Researchers need to be able to determine if there are enough patients to create a suitable cohort or if there are samples available to study. The central warehouse serves as kind of a one-stop shop for this kind of information, with a user-friendly search interface, Weistra explained.
"We were looking for solutions that would help us in the research domain," said Kemmeren, who noted that the hospital deals with complex, multi-omic datasets. "That's why we tried to do this optimization step as early as possible."
Among the open platforms The Hyve builds on are cBioPortal for Cancer Genomics, the Memorial Sloan Kettering Cancer Center-developed system that helps researchers visualize and analyze large datasets from the Cancer Genome Atlas, as well as TranSmart, a data management system for translational biomedical research. The Hyve helped create the Glowing Bear version of TranSmart, with an updated interface that supports queries for longitudinal and sample data.
Princess Máxima Center came up with its own data model, taking into account that diagnoses often are associated with multiple samples and samples have multiple materials, according to Kemmeren.
"They came to us with this model and we translated that to TranSmart, which is a really flexible data model," Weistra said. He also noted that cBioPortal has fixed standards, which he said makes it easier to build visualizations.
The Hyve started with a pilot. During that phase, Kemmeren and his team identified several issues they wanted addressed in the platform, including the ability to request samples directly from the biobank.
"CBioPortal they kept kind of as is because it already had a lot of features that they needed, but for TranSmart, we developed some features with them," Weistra said. Users could already run queries on which patients have a specific tumor type and whether there are samples available, but The Hyve added the ability to subscribe to a query because data gets updated as often as nightly.
"This is basically the platform where we put everything together. What it will allow you to do is basically ask questions like, 'Okay, give me all patients that have, for instance, acute lymphomic leukemia,'" Kemmeren said.
A future step likely will be to build a direct query function for oncogenomics data, looking for mutations in specific genes and searching across data types, according to Weistra. The Hyve also is getting ready to launch a new feature as soon as this month to ask questions on the sample level in addition to the patient level, and Princess Máxima Center is one of the testing sites for that.