Quest Diagnostics is working with artificial-intelligence startup Paige to try to identify new biomarkers of cancer and other diseases and develop new software products, the companies said Tuesday.
The companies will apply machine learning from Paige, a spinoff of New York City-based Memorial Sloan Kettering Cancer Center that uses AI to help diagnose cancer patients, to analyze digitized pathology slides from Quest, including from the clinical laboratory company's AmeriPath and Dermpath pathology services businesses.
Paige, a startup based in New York City, was founded in 2017 and in 2019 was granted breakthrough-device designation by the Food and Drug Administration.
Quest and Paige plan to work together to develop software products that diagnose various cancers based on the biomarkers they discover, for which the companies will seek regulatory approval. Diagnosing cancer typically involves a pathologist inspecting tumor tissue on a pathology slide, but AI tools can potentially help by identifying patterns that are difficult for physicians to discern on their own.
The companies plan to sell the products to pathologists and oncologists to deliver "faster, more accurate and more informed clinical insights," said Leo Grady, chief executive officer at Paige, in a statement. Quest will also be able to use the software products as part of its pathology operations.
Quest and Paige will work together on research and development, a Paige spokesperson wrote in an email to Modern Healthcare, but Paige will build and commercialize the products.
It's too early to share a timeline for when products could launch, according to Paige.
Quest and Paige will initially focus their development work on tools for solid tumor cancers, including prostate, breast, colorectal and lung cancer.
As the companies work on developing software products, Quest and Paige also plan to license some of their findings to biopharmaceutical and research groups that are developing drugs.
Quest and Paige did not disclose financial details of the partnership, beyond acknowledging the agreement involves shared revenue for some product and commercial milestones.
Under the partnership, Quest will also add Paige's AI software to a digital pathology toolset for pathologists.
Secaucus, N.J.-based Quest posted $2.7 billion in revenue for 2021's first quarter, up 49.3% from the year-ago period, and $660 million in operating income, up 277.2%.
The company's top competitor, Labcorp, posted first quarter revenue of $4.2 billion.
Quest is one of the market leaders in the lab testing space, said Paddy Padmanabhan, founder and CEO of healthcare consultancy Damo Consulting. To continue to grow, Quest will have to create new products and services for its customer base, Padmanabhan said.
The partnership with Paige is part of Quest's push to draw out "actionable insights" as part of its advanced diagnostics business, said Kristie Dolan, vice president and general manager of Quest's oncology franchise. Quest's advanced diagnostics business focuses on genetics and precision medicine.
Roughly half of health systems and healthcare providers use services from Quest each year, according to the company.
That market footprint will help Quest create new software products, since developers need large datasets to train AI systems.
"Quest has a very, very large dataset of lab results," Padmanabhan said. "That potentially is the training data that a product like Paige needs to put it over the top."
Bringing advanced analytics and AI to bear in cancer care has been a popular area that companies are looking to tackle.
Oncology was a cornerstone of IBM Watson Health's high-profile healthcare AI effort, which came under fire in 2018 after reports it recommended erroneous and unsafe cancer treatments. And pharmaceutical company Roche in 2018 acquired oncology data company Flatiron Health for $1.9 billion.
But cancer diagnosis is difficult. Unlike back-office administrative functions that have integrated AI, such as claims processing, a cancer diagnosis has a direct effect on patient lives, Padmanabhan said.
"People have to tread very carefully, because the stakes are so high," he said. "For AI to become mainstream in high-stakes clinical conditions, it's going to be a gradual process."