A Case Western Reserve University-led team of scientists has used artificial intelligence (AI) to identify which patients with certain head and neck cancers would benefit from reducing the intensity of treatments such as radiation therapy and chemotherapy. Their research was published recently in the Journal of the National Cancer Institute.
The researchers used AI tools similar to those they developed over the last decade at the Center for Computational Imaging and Personal Diagnostics (CCIPD) at Case Western Reserve in Cleveland. In this case, they asked the computer to analyze digitized images of tissue samples that had been taken from 439 patients from six hospital systems with a type of head and neck cancer, known as human papillomavirus (HPV)-associated oropharyngeal squamous cell carcinoma (OPCSCC).
The computer program successfully identified a subset of patients who might have benefitted from a significantly reduced dose of radiation therapy. While that analysis was retrospective—meaning the computer analyzed data from patients in which the eventual outcome was already known—the researchers say their next step could be to test its accuracy in clinical trials.
The work was led by Anant Madabhushi, PhD, CCIPD director and the Donnell Institute Professor of Biomedical Engineering at the Case School of Engineering, along with Germán Corredor Prada, PhD, a research associate in the CCIPD lab.
‘Overtreating Patients’
Although most others with HPV-driven cancer would still benefit from aggressive treatment—along with patients whose cancer was unrelated to the virus—the researchers says their study revealed a significant group was receiving more aggressive therapy than they needed to achieve a favorable outcome.
Clinicians are not able to easily make that distinction simply from simply looking at the tissue scans, the researchers say. So virtually all patients with these cancers—regardless of whether HPV-driven or not—are treated with a full course of chemo and radiation.
“We have been overtreating many patients with chemotherapy and radiation that they do not need because we didn’t have a way to find out which patients would benefit from de-escalation,” Madabhushi says. “We’re saying that now we do—and that someday physicians could modulate the way we care for people and not just give the standard high dose of radiation to everyone who comes through the door.”
Madabhushi says that reducing radiation for these patients could also help lessen the “toxicity of radiation therapy,” meaning that they could experience fewer side effects such as dry mouth, swallowing dysfunction and taste changes.
“There are already national clinical trials ongoing investigating the reduction of radiation therapy and chemotherapy intensity in favorable HPV positive oropharynx cancer patients,” adds Shlomo Koyfman, MD, director of head and neck and skin cancer radiation at Cleveland Clinic and a study collaborator. “However, properly selecting the ideal patients for this treatment reduction has been a challenge. This imaging classifier can help us better select patients for these novel treatment paradigms.”
Collaboration Among 10 Institutions
Germán Corredor Prada says other researchers are already testing whether reducing the intensity of treatment can benefit some patients. But this new work, if validated in human trials, could provide a tool for physicians to make better decisions about who should get chemo or radiation, he says. “Maybe we can reduce intensity of treatment for some people and give them a better quality of life because chemo and radiation often have very strong side effects,” Corredor adds.
Nearly two dozen other scientists contributed, including six others from Case Western Reserve. “We have been able to visually assess patient’s tumor microscopically for a very long time, but now, with this technology, can actually extract meaningful information from the morphology for prognosis and prediction,” says James Lewis Jr., MD, a professor of pathology, microbiology and immunology at Nashville, Tenn.-based Vanderbilt University Medical Center and a research collaborator.
