Aurora Health Care is collaborating with Oracle to develop software to mine data from tens of thousands of samples stored in Aurora's biorepository in order to help researchers find samples of particular interest for biomarker discovery or clinical trials.
Through the partnership, Aurora hopes to provide a more "strategic approach towards clinical trials" in which patients are selected to participate based on clinical information stored in the biorepository, called the Open-Source Robotic Biorepository & Informatics Technology, or ORBIT, which holds consenting patients’ DNA and is linked to patients' de-identified electronic medical records.
"We are focusing on what we call the secondary use of clinical data to help improve various data-management issues around managing and developing clinical trials more effectively," Matthew Tector, ORBIT's director, told BioInform. "[In] other areas we are hoping to improve upon our ability to do safety and efficacy studies over the long term and also the ability to compare the effectiveness of different therapeutic actions."
Randall Lambrecht, Aurora's vice president for research and academic relations, told BioInform that it "made sense" to partner with Oracle because "we have a huge amount of clinical information, and that’s not even genomic information. [W]e anticipate [that] in the next decade or two or three, as more genomic information is learned, that [will] become an inordinate amount of data."
He added that the deal with Oracle is a "good fit" because "they know how to manage data."
Aurora launched ORBIT in 2009 to create a library of blood samples, collected from consenting patients, linked to health information contained in EMRs.
At the time of the launch, Aurora had in place a Tecan robotics system, an in-house data warehouse, a system for electronic health records, and was looking to develop data-analysis systems to incorporate and manage genomic data (BI 04/24/2009).
Currently, the bank contains between 81,000 and 82,000 anonymized patient samples, an increase from 11,000 samples two years ago. Aurora hopes to bring the number to 500,000 over the next three to five years, Lambrecht said.
Furthermore, Aurora has hired a dedicated data analyst for ORBIT to help migrate data and put it into a format that’s easily evaluated.
At present, patients' genomic data isn’t included in ORBIT. However, as the genomics space matures and as sequencing and data-analysis technologies become more mainstream, Aurora plans to begin collecting this information. Such data is currently lacking in most clinical trials, Lambrecht noted, rendering them far less effective and potentially dangerous.
"Clearly we are missing a whole side of medicine when we don’t include a person's genetic makeup," he said. "I have always been amazed at the way clinical trials are conducted. It just seems, from a scientific methodology point of view, [that the process is] so shot-gun."
He added that most trials are conducted on patients who are selected based on safety and compliance criteria — not their likelihood of responding to the drug candidate.
Aurora envisions that the collaboration with Oracle will create a system under which pharmaceutical companies or medical researchers could mine the EMR data associated with ORBIT to select patients with specific criteria, such as poor response to a given therapy. Once matches are found, researchers could then request access to additional data or samples from consented patients in the biorepository for the purpose of conducting a clinical trial.
Initially, Lambrecht said, the partners will work on methods to organize and mange clinical data in such a way that interested drug or pharma companies looking to conduct trials can look at Aurora's population and determine if it is suited to the need — for example, whether it is the right mix of ages, or if there is enough of a population for the disease in question.
Eventually, Lambrecht said, it's likely that Aurora and Oracle will develop additional systems to handle genomic data analysis. To bring these systems to life, the healthcare provider hopes to partner with universities and commercial groups with expertise in genomic data.
On the commercial front, at least, Aurora may have its pick of multiple software options as last year several IT vendors began developing new capabilities for their systems for the purpose of integrating genomic information and clinical data.
For example GE Healthcare developed Qualibria, a clinical-knowledge platform that correlates genomic information with phenotypes, which is scheduled for general release sometime this year.
Meanwhile, Cerner, which supplied Aurora's electronic health records system, said it was developing a Clinical Bioinformatics Ontology that links vocabularies used for genomics and clinical practice (BI 04/02/2010).
Oracle could not be reached to comment on its part in the project in time for this report.
The Case for EHRs
Aurora isn't the only organization to recognize the value of electronic medical records to help stratify patient populations and select participants for drug trials.
Separately this week, the International Serious Adverse Events Consortium said that it will work with the HMO Research Network to use HMO electronic medical records to identify genetic markers linked to drug-related adverse events.
Each HMORN member "will use detailed clinical profiles to search for potential subjects to enroll into … serious adverse events research projects using their electronic medical record databases," iSAEC said in a statement.
Specifically, the iSAEC will use EMR data from nine HMOs to build research cohorts associated with the genetics of three drug-induced SAEs: hepatoxicity; serious skin rashes; and extreme weight gain in users of atypical antipsychotic medications.
HMORN members include HealthPartners Research Foundation; Kaiser Permanente Georgia; Marshfield Clinic Research Foundation; Group Health Collaborative; Geisinger Center for Health Studies; Henry Ford Health Care; Kaiser Permanente Southern California; Harvard Pilgrim Health Plan; and Kaiser Permanente Colorado.
This week's announcement signals the second phase of an ongoing collaboration between iSAEC and HMORN.
Arthur Holden, the chairman of the iSAEC, told BioInform that during the first phase of the project, the consortium applied a series of sorting algorithms that were developed in-house to mine data stored in a common data warehouse from six HMOs in the research network who agreed to participate. The goal was to identify patients who had experienced any of the three drug-induced SAEs.
Data from selected patients were entered into Cerner's web-based Discovery platform, making the data accessible to clinical experts to evaluate patients' suitability.
The second phase of the project, which is expected to last two years, will compare the genetic and clinical data from EMR-derived cases to healthy controls, with the goal of identifying genetic mutations that are associated with the each of the three drug-induced events, iSAEC said.
Have topics you'd like to see covered in BioInform? Contact the editor at uthomas [at] genomeweb [.] com.