This report has been updated to include corrections to Silvana Canevari's position at the institute and to clarify comments from Michele Torresani
IBM Research said this week that it has begun testing a prototype of an informatics platform that combines clinical knowledge with a patient's genomic information and electronic medical records to make personalized treatment suggestions.
Clinical Genomics, or Cli-G, was developed by researchers in IBM's Haifa, Israel, branch, and is currently being tested by researchers at Italy's Fondazione IRCCS National Cancer Institute who are using the tool to analyze data from patients with sarcoma and head and neck cancers
It combines cancer patients' clinical data — including information on prior treatments and outcomes — with genotype information, results from similar past cases, and standard clinical guidelines from organizations such as the National Comprehensive Cancer Network, Chalapathy Neti, director of healthcare transformation at IBM Research, explained to BioInform.
Cli-G's intended users are clinical researchers who can use it to generate treatment hypotheses that would then undergo additional testing and statistical validation before they can be established as standards for treating similar cancers, Neti explained.
As an illustration, a researcher looking to determine whether a patient with late-stage cancer would be a good candidate for chemotherapy could use the platform to create a virtual cohort of individuals with similar clinical and genetic characteristics; and then retrospectively analyze the treatment outcomes where chemotherapy was used, he said
The scientists could also mine the data for alternative treatments — such as a hysterectomy in the case of uterine cancer — that might have better outcomes, he said.
Ultimately, IBM expects the solution to provide physicians and administrators with a better picture of the patient-care process and reduce costs by helping clinicians make more effective treatment decisions.
Once physicians make a diagnosis, they will receive personalized
recommendations for their patients, based on medical information, automated
interpretation of pathology clinical guidelines, and information from past clinical cases that have been documented in the hospital information system.
In addition, the tool can provide administrators with an aggregated view of patient care, enabling them to evaluate performance and then use this knowledge to streamline processes for maximum safety. For example, they can drill down into the data to better understand what the guidelines were for recommendations, what succeeded, and whether treatment quality has improved as a result.
It isn't clear at this early stage how and when IBM will commercialize the tool, Neti said, though he noted that it is seeing "a lot of customer interest" in a clinical genomics platform.
If the cancer pilot is successful, Cli-G could move into similar pilot deployments at additional customer sites and then pass through an internal commercialization process, he said, but couldn’t comment on a possible commercialization timeline.
The Evolution of Cli-G
If and when IBM brings Cli-G to market, it won't be limited to cancer studies but should be applicable to other diseases that have strong genomic underpinnings, Neti said.
He told BioInform that the system has been brewing in the lab for several years and that earlier versions have been employed in at least two other research settings.
Although the Fondazione IRCCS National Cancer Institute is the only participant in this pilot study, Cli-G has been used in other efforts, such as the EuResist project, which was set up to develop a computational approach to managing antiretroviral drug resistance.
According to the project website, the EuResist system accepts HIV genotype and, optionally, a set of clinical data as input and predicts a response to common antiretroviral regimens. This helps HIV specialists choose the most effective drug cocktails for their patients. The project was funded by Abbott, Pfizer, and the European Commission's sixth framework program.
Cli-G was also used in Europe's HyperGenes project — funded by the European Commission's seventh framework program — which aims to construct a genetic-epidemiological model for complex diseases using as a disease model essential hypertension, which describes cases where the cause of high blood pressure is not known.
The current incarnation of the platform has moved beyond these projects to have a "greater emphasis" on support for "decisions at point of use," Neti said. Additionally, it provides a more comprehensive knowledgebase — combining internal and external data — from which treatment recommendations are made, he said.
Cli-G uses extensible markup language tags from the Clinical Genomics Level Seven — an extension of the Health Level Seven International standards — to represent patients' SNP data as well as clinical phenotype information, Neti explained.
The system then uses a set of analytical applications and information about other clinically similar patients as well as external information such as treatment guidelines to generate a hypothesis.
Currently, IBM and its collaborators at the Fondazione IRCCS National Cancer Institute have developed capabilities that allow users to scrutinize treatment decisions made in cohorts of earlier cases of sarcoma and head and neck cancer, Michele Torresani, an engineer in the institute's information and communication technology department, told BioInform.
However, capabilities that would allow physicians to assess treatment recommendations in the light of the findings from cohort analyses are still being developed, he said.
Meanwhile, the institute hopes to pull in resources from other oncology clinics and centers in its region.
Furthermore, the institute hopes, with IBM's help, to establish a virtual biobank that will hold data from samples from nearby oncology institutes and departments, Maria Grazia Daidone, director of its experimental oncology and molecular medicine department, told BioInform.
The partners also plan to put in place standard operating procedures for collecting data and obtaining patient consent, she said.
Meanwhile, the institute hopes to create a database that will hold functional genomic information from tumor samples, Silvana Canevari, the head of the functional genomics facility and the molecular therapies unit, told BioInform.
So far the researchers have collected data from about 1,000 patient samples, she said.
Data from both these efforts would eventually be fodder for analyses in Cli-G, she said.
The pilot is expected to last for one year and may also include analysis of other cancer types such as breast cancers, Torresani said.
Have topics you'd like to see covered in BioInform? Contact the editor at uthomas [at] genomeweb [.] com.