- Human Biospecimens
- Biospecimen Contributors
June 25, 2021
Oral cancer claims nearly 10,000 American lives a year, and roughly two in five newly diagnosed patients die within five years of diagnosis.
Researchers, however, have hit on new a saliva-based early-detection approach now embedded in a commercial diagnostic device that the FDA has determined to be a breakthrough worthy of expedited development. The designation is a milestone in a decade of research by Chamindie Punyadeera, associate professor at Queensland University of Technology (QUT), who lost a brother-in-law within six months of being diagnosed with oral cancer.
“This test could save many lives because until now, early-stage oral cancer has been hard to detect because effective diagnostic tools have not been available,” said Punyadeera. “This has led to late diagnosis, a poor prognosis and low survival rates.”
Key roles for artificial intelligence and machine learning
QUT offers a summary of the research here, and an article is to be published in npj Genomic Medicine (pre-print). “For the first time, we demonstrate the potential clinical utility of an artificial intelligence/machine learning model for diagnosing oral cancers early, opening a new era of non-invasive diagnostics, enabling early intervention and improved patient outcomes,” write Punyadeera and her co-authors.
The findings have informed the development of a first-of-its-kind screening tool for oral cancer and throat cancer by Washington-based Viome, whose AI platform procured the FDA Breakthrough Device designation.
“Today’s standard of care to detect oral and throat cancer is severely outdated,” said Naveen Jain, CEO and founder of Viome. “Everyone relies on a primary care clinician to examine their mouths and look for lesions. This subjective and qualitative approach is a key reason why oral and throat cancer are detected at stage three or four, when many people cannot receive life-saving treatments. [W]e believe in the power of technology to help everyone stay healthier, do more, and live longer.”
The work relied on 433 saliva samples from patients with oral cancer, oral premalignant disorders, and normal controls. “[O]nce an early diagnostic test is available at scale, we can routinely improve the accuracy of our test as we collect more ‘real-world evidence’ to further train our machine learning models,” the researchers write.
At iSpecimen, we provide samples of saliva and other biofluids for liquid biopsy developers, as well as tissue and cells. Research organizations can specify patient age, gender, race, condition, severity, blood type, procedures, test results, outcomes, smoking status, family history, and more. We’re grateful for every opportunity we have to collaborate with inspiring innovators.
Learn about the iSpecimen Marketplace where you can browse millions of richly annotated, de-identified human tissue and biofluid biospecimens, in addition to hematopoietic and immune cell products and COVID-19 samples. You can join for free.