The era of precision or personalized cancer medicine is driven by data, and many experts believe that the solutions to a lot of the remaining cancer mysteries may be hidden within this data. Radiation oncology physicist Todd McNutt is among them. Within a sea of data, the challenge is figuring out what information has the value to advance patient care and how to extract it.
“There is so much more data collected than is ever used,” says McNutt. To put some of this unused data to work in radiation therapy, he built—from the ground up—a complex, computerized data mining system. It is called Oncospace, and it scrutinizes and analyzes data from prior patients who received radiation treatment to improve the treatment of new patients. It evaluates the therapies that worked best for a particular cancer as well as those that resulted in less than favorable outcomes and generates an optimal treatment plan.
Creating this complex, interactive system has been a laborious, ten-year process for McNutt and colleagues, but it is rapidly gaining traction in the research and clinical setting. “The practice of cancer medicine naturally creates data,” he says, “but for the first time in history, we have the technology to sift and sort through this data in completely new ways."
“Todd has proven that large data warehouses of patient information collected from previously treated patients can be used to individualize treatment decisions for new patients,” says Theodore DeWeese, Director of Radiation Oncology and Molecular Radiation Sciences.
Oncospace does more than collect and store data. It takes informatics to the critical next level with the capability to perform interactive analysis that informs clinical decision- making. Radiation oncologist and head and neck cancer expert Harry Quon put the system McNutt designed to the test in clinical practice.
In working with radiation, the line between healing and harming is almost as narrow as the beam itself. Quon understands the consequences of crossing that line. His job is to develop the treatment plans that use radiation to destroy cancers in the head and neck without causing permanent damage to the dense anatomy surrounding the cancer. Patients want their disease cured. They do not want to be left unable to speak or eat, but these are some of toxic effects radiation treatment of head and neck cancers can cause.
This was also the reason McNutt saw these cancers as the ideal choice to put Oncospace to the test. Head and neck cancers are among the most difficult cancers for radiation physicists and oncologists to plan, often requiring as many as 20 treatment revisions as they work to design a treatment that hits the cancer with radiation but does not do damage to vital organs and glands, such as the voice box and salivary glands.
McNutt’s system provides the guidance that allows Quon and other clinicians to maximize the healing and minimize harm. It scours all of the data on head and neck cancer patients treated in the Kimmel Cancer Center, charts radiation dose distributions, toxicity, and other data in vividly colored computerized maps and graphs and reveals the optimal plan. At the same time, it takes into account and connects all of the variables—age, underlying health conditions, and other treatments patients are receiving and figures out how all of these variables relate and influence toxicities and response to treatment. “We can build predictive models of toxicities and other side effects based on data we have collected from prior patients, including indicators that a patient may be at higher risk for certain treatment toxicities and use this information to adjust the treatment plan,” explains McNutt.
“There is knowledge in the variations in toxicities and response that occur from patient to patient,” says Quon. “That type of analysis is not possible without the analytic capabilities of Oncospace. It does what no other tool can do and allows us to see unique relationships that otherwise would be hidden.”
As important as the data it stores and analyzes is the interface it uses to gather the data. McNutt worked closely with Quon and other members of the clinical care team, including nurses, speech pathologists, nutritionists—all of the specialists involved in the treatment of head and neck cancer patients—to develop web-based assessment forms so that all of the information collected by caregivers could be easily integrated into the clinical workflow and ultimately into the Oncospace database. “It requires some changes in habits and doing things a little bit differently than we were used to, but the reward gets people on board,” says Quon. “We have a tool that no one else has. As a result we’ve improved our patient care and doubled our head and neck practice.”
McNutt and Quon have proven that Oncospace improves treatment plan quality and reduces toxicities. Now they are using it to track and improve treatment outcomes and to advance research. McNutt says it is imperative that the data be tied to outcome, and he is among the first to take on the challenge.
More recently, they earned a grant from information technology giant Toshiba to incorporate imaging into the data collected.
A grant from Toshiba is playing a major role in providing funding and scientific expertise to help McNutt and team adapt the Oncospace system to incorporate data, including imaging, on disease response and status: Is the cancer stable? Has it progressed? Did it recur? Toshiba has developed a sensor system for computers that generates millions of data points on temperature, usage, and other factors to provide predictive models for hard drive failure. McNutt is hopeful that this data mining expertise can be applied to cancer medicine through Oncospace.
Some of the new data he hopes to use to enrich Oncospace is in CT imaging scans done in treatment simulation and guide how patients are positioned. These images are not currently used beyond that purpose, but inherent in these scans is information that shows how tumors are responding to treatment. If the scans could be incorporated automatically into Oncospace, it would allow them to track the history of the tumor during treatment. “Using Oncospace to analyze and quantify these daily images, we could potentially tell early on in the course of the treatment if the tumor is responding and change the treatment plan if necessary,” says McNutt. He says it is the radiation oncology version of the work being done in molecular genetics using genetic biomarkers to track and monitor the response of cancers to targeted drug therapies. “It is real time, in-treatment monitoring,” says McNutt. “The same way we used the system to relate the dose of radiation to the parotid salivary gland to the loss of gland function, we can use it to relate treatment plans to treatment responses,” says McNutt.
McNutt also hopes to gather from treating physicians’ notes in text. This is bigger challenge because text is not the language of computers, and for that reason, he says, many data mining systems are missing this critical clinical piece. “Physicians are trained to document records for communication, but not for data collection,” says McNutt. To incorporate patient outcomes in Oncospace, McNutt worked with clinicians to develop a new interface designed to extrapolate clinical information through a numerical ranking system caregivers use each time they see a patient.
As McNutt continues to expand the capabilities of the pioneering system he built, its success in head and neck cancer has made it the model for use in other cancer types, including lung, pancreatic, and prostate cancers. He is also planning to extend the use of Oncospace to other cancer centers in a novel endeavor that has never before been tried but offers to even more extensively realize the power of data. If the answers are in the data, than more data analyzed should lead to more rapid discovery of better roadmaps for care. Partner institutions would be given access to Oncospace technology and share their results with all of the others participating centers. McNutt says sharing the technology with other institutions will also allow many cancer types to be studied simultaneously.