This article has been updated from a previous version to correct an error in the description of the effect of the EphB4 receptor in ovarian cancer patients. Survival in patients with tumors expressing the receptor decreased by two and a half years. It did not increase, as originally stated.
After several years in stealth mode, computational oncology firm Life Biosystems recently shared some initial results of its informatics-based approach to personalized medicine.
Earlier this month at the annual meeting of the American Association for Cancer Research in Orlando, Fla., Life Biosystems researchers presented two posters outlining recent advances in the company's R&D activities. The first poster described the in silico discovery and subsequent experimental validation of an alternative receptor for erythropoietin, or EPO, which is used to treat chemotherapy-induced anemia in cancer patients, but has recently been linked to tumor progression and shorter survival in some individuals.
The second poster outlined a computational platform the company has developed called MASE, for Molecular Analysis of Side Effect information. The system, which is based on data from the US Food and Drug Administration's Adverse Event Reporting System along with curated information about the biological mechanisms that underlie those side effects, can be used to predict the outcomes of specific combinations of drugs, targets, and metabolizing enzymes.
David Jackson, chief scientific officer of Life Biosystems, told BioInform that the posters are examples of a range of projects the company has been working on over the past several years, though all of its efforts involve "a strong emphasis on text-mining and bioinformatics and proprietary databases."
Jackson said that the EPO project grew out of an early system the company developed to track whether drugs had failed by monitoring stock prices of pharmaceutical companies. The assumption was that a sudden drop could signal an opportunity for the company to develop a response or safety marker for the failed compound. In 2007, Jackson said, the system picked up an $18 billion drop in Amgen's market cap, "which set off a red light."
It turned out that data had been published showing that Amgen's erythropoiesis-stimulating agents Epogen and Arenesp were linked to poorer survival in some cancer patients — a finding that was borne out over the next several years in additional studies. The US Food and Drug Administration ultimately added a black-box warning to the drugs' labels advising doctors of the potential safety risks and urging them to use the lowest possible dose.
During this time, Jackson noted, Amgen published several papers refuting the association between EPO receptor function and these adverse effects. Furthermore, he said, when Life scientists began looking into other indications for which EPO was being considered, they found that EPO derivatives were being developed as a cytoprotective agents, and were able to induce a cytoprotective effect without binding to the EPO receptor.
"This really strongly argued that there's another receptor out there," Jackson said.
The company then took a "holistic in silico approach" that focused on the secreted portion of the proteome "and asked, 'What proteins possess functional, structural, and regulatory features that are consistent with the role of a potential EPO receptor?'"
Using a combination of text-mining and traditional bioinformatics, the company culled the scientific literature, protein structure databases, and other resources in order to come up with a short list of potential alternative receptors. At the top of the list was a protein called ephrin type-B receptor, or EphB4. "One of the interesting things was that EphB4 was sitting directly beside EPO on the genome, which was highly striking," Jackson said.
The company then decided to test the hypothesis under the auspices of a 10-year collaboration agreement it signed with MD Anderson Cancer Center in 2008 that allows the company to conduct validation studies with investigators at the center. MD Anderson has the rights to publish the results and the company has the rights to any IP that is discovered [BI 4/25/2008].
The company partnered with MD Anderson's Anil Sood, who performed a series of in vitro and in vivo studies that demonstrated a direct interaction between EPO and Ephb4. In one study, Sood conducted siRNA studies in orthotopic mouse models of ovarian and breast cancer. As reported in the AACR poster, EPO receptor siRNA had no impact on EPO-stimulated tumor growth, while EphB4 siRNA "completely blocked EPO-stimulated tumor growth."
Next, the collaborators turned to clinical information that was available at MD Anderson. Sood's team identified a cohort of 175 ovarian cancer patients and retrospectively looked at the correlation between expression of EphB4, treatment with EPO, and patient outcome.
Jackson said that the findings indicate that in patients who don't express EphB4, "survival is not at all affected whether you give EPO or not." However, in patients with EphB4-positive tumors who received EPO, median survival decreased by about two and a half years.
[ pagebreak ]
Sood has also looked at breast cancer data, "and the results there were also confirmatory," Jackson said. "For patients who have this receptor expressed in their tumors, if they receive EPO, then it has very serious implications for their survival."
In a statement, Sood called the findings "significant," and said that EphB4 could serve as a "novel diagnostic biomarker and therapeutic target" that could help "reduce the risk of adverse events in anemic cancer patients and improve their overall treatment."
Jackson said that Life and the MD Anderson team are currently preparing a paper that describes how tumor progression is induced by the interaction of erythropoiesis-stimulating agents with the EphB4 receptor.
The company has also filed for patents on the expression of the EphB4 receptor as a theranostic for EPO treatment, but Jackson said that Life has not yet solidified its commercialization plans for the finding.
"Our strategy is to get the paper together and get it submitted to a high-profile journal and then to start the licensing discussions," he said. "We're very interested in outlicensing this IP."
Mining Adverse Events
While the EPO project grew out of a desire to answer a specific biological question, some of the company's other efforts aim to cast a wider net for hypothesis generation. Life's MASE system, for example, was developed to serve as an exploratory tool that could be of use in a range of applications — from guiding treatment decisions, to drug repositioning, to cost-benefit analysis.
"The motivation for developing MASE was … to learn from human information as opposed to information coming from in vitro models or model organism studies," Jackson said. "The idea here was to link the phenotypic information contained within the [FDA's] Adverse Event Reports to the molecular explanation as to what is actually going on from a mechanistic perspective."
Jackson noted that the FDA's Adverse Events Reports are publicly available, "but there's a lot of processing you have to do to clean up the data because it comes from pharmaceutical companies, doctors, even sometimes from patients."
In addition, much of the information in the records is unstructured, so the company had to do quite a bit of work to correlate misspelled and alternative drug names with the correct molecular compounds. On top of this foundation, which was culled from more than 1.9 million records in the FDA database, the company then associated all the reported drugs with known and predicted protein data as well as other molecular information.
The resulting system, according to the AACR poster, transforms adverse event information "from a purely drug-centric resource to one that emphasizes the functional mediators of drug activity within the patient system."
Life expects that MASE would be a useful tool to help doctors determine which therapy — or combination of therapies — is likely to be the safest option for a patient.
It also shows promise for drug repositioning, Jackson said. Life and MD Anderson recently completed a study that used MASE to look at the effects of several non-oncology drug classes on cancer patient outcomes and found that beta-blockers in combination with anti-cancer therapies "seem to provide a protective effect."
Jackson said that the collaborators have submitted this study for publication.
Life is still discussing "the optimal approach" to take with the system, though. "MASE is a platform that can head in many different directions," Jackson said.
"We're taking numerous different strategies on how we're developing the platform, because it is applicable to different players within the healthcare industry," he said. "It is of interest to doctors, it's of interest to us in terms of identifying novel IP, it's of interest to the pharma industry, it's potentially of interest to payors as well, because if you're able to look at specific drug combinations and look at the effects on outcome, well then outcomes such as hospitalizations have very important implications for cost."
In addition, the company plans to continue developing the system. One goal is to add the capability to incorporate a patient's molecular data into the system in order to provide personalized safety assessments.
"That's where the system is heading," Jackson said. "The version presented at AACR is the first version of the platform, and development is currently taking place that heads specifically in this direction — toward integrating patient-specific clinical molecular data."
Have topics you'd like to see covered in BioInform? Contact the editor at btoner [at] genomeweb [.] com.