On Target - The Data Quest
Issue No. 4
The Data Quest
Date: April 18, 2012
Radiation oncology physicist Todd McNutt had an intriguing idea. He imagined a system that would store clinical data from past patients treated with radiation therapy and use it to improve the quality and efficiency of the treatment of new patients.
Working with collaborators from Johns Hopkins Computer Science and Physics and Astronomy Programs, McNutt and team have created an analytical database he calls “Oncospace.”
It pulls together radiation therapy data in a complex computerized system, using
anatomy, radiation dose distributions, toxicity, and outcome data of past patients
to improve the therapy for those about to be treated. It enables the analysis of
the best outcomes and, conversely, those with less than favorable outcomes, to
create the optimal treatment plan.
“When a person is diagnosed with a serious illness, they typically seek out a physician with a wealth of experience in treating it,” says McNutt. Oncospace becomes another experienced member of the clinical team. It captures patients’ clinical experiences, not based on the physician’s memory, but on real data collected and organized in a computer database to reveal information about improving treatment and avoiding radiation toxicities to healthy tissue and organs, he says.Oncospace works by surveying the data from past patients to uncover similarities between tumors and their relationship to critical organs and tissue they want to spare from radiation. The system finds the set of critical organs from all patients in the system. This provides upfront information on the radiation dose distribution as well as any potential toxicity risks to the patient. The information is then used to automate and ensure quality in radiation treatment planning.
Just as cancer gene sequencing uses data from millions of DNA samples to establish patterns in gene expression, McNutt’s model uses data from patients treated with radiation therapy to reveal patterns in patient outcomes.
“Todd’s work is one of the first demonstrations of how we can develop large data warehouses of patient information collected from previously treated patients and use it to make individualized treatment decisions for new patients,” says Theodore DeWeese, director of Radiation Oncology and Molecular Radiation Sciences.
The goal with radiation therapy is to target the highest doses of radiation directly and precisely to tumors. Lower doses of radiation are aimed at high-risk tissue around the tumor and nearby lymph nodes that could potentially harbor small numbers of unseen cancer cells that if left untreated can allow the cancer to come back.
Equally important as hitting the tumor and high-risk tissue around it with radiation, is not hitting the critical structures, organs, and normal tissue that could be irreparably damaged by radiation. The spinal cord, brain stem, and esophagus are examples of critical structures. In head and neck cancers, the parotid gland is a critical structure.
Damage to the gland, McNutt says, can permanently destroy the function of salivary glands. The inability to produce saliva can substantially affect patients’ quality of life, causing them to have a chronically dry mouth and making it difficult to eat.
Oncospace was most recently studied in head and neck and pancreatic cancers, and investigators found that the method considerably improves treatment plan quality and sparing of critical organs. As a result, they are now ready to begin using it as part of standard, clinical practice for these cancers.
They have also begun to take on a bigger challenge, studying its use in lung cancer. With this cancer, the organ being treated is a critical structure. To further complicate matters, tumors can arise in either or both lungs and in many different areas of the organ. Although this work will take longer, McNutt is optimistic that the Oncospace data
analysis will also lead to improvements in radiation therapy for this
leading cancer killer.