Using AI to quickly spot the microscopic facets of cancer
CEO: Leo Grady
Based: New York City
The scoop: “There are a lot of patterns in tissue,” said Leo Grady, CEO of Paige, the Memorial Sloan-Kettering Cancer Center spinout working on artificial intelligence to assist in the diagnosis of cancer.
Paige’s programs are built on years’ worth of tumor slides digitized by the cancer center, plus millions more taken from its archives, and search for those patterns and trace them back to the genes that may affect the patient’s prognosis or treatment.
“For years, pathologists have been saying, ‘I can look at tissue and I can see there’s a BRCA mutation here. I can look and see that the patient is HER-2 positive, or that there's an EGFR mutation,’” Grady said in an interview.
“Yet when pathologists say this, they can see it but they can't quite describe what it is they're seeing,” he said. “They can't quantify it, they can't standardize it, they can't measure it.”
“So as a result, the patient still ends up with immunohistochemistry tests, or they end up with genomic profiling—but there is still some phenotypic expression in the tissue of this underlying molecular change,” Grady said.
The company's software scans for suspicious features across samples of breast and prostate tissue—with programs approved in Europe as clinical diagnostics and a breakthrough designation from the FDA—and flags them in real-time for a focused review.
“By using modern AI technology with large databases, the computer can go through these slides measure those patterns, identify them, quantify them and then reveal those to the pathologist—who may not recognize it as something they learned in medical school—but we can still prove and validate it, just like any other diagnostic test.”
What makes Paige fierce: The New York-based company—which has raised up to $195 million in venture capital funding since its 2018 debut—doesn’t plan to stop at what can simply be seen with a human or computer’s eye.
A partnership with the genomic testing company Agendia, launched last November, aims to integrate its digital pathology programs with the sequencing of breast tumor tissue to establish new cancer biomarkers derived by AI.
“This is really a first-of-its-kind diagnostic test, to be built out this way,” Grady said. “It will help patients know how well they’re going to benefit from chemotherapy.”
When time is an immediate factor, many patients may opt to pursue therapies—and risk the side effects—rather than wait for test results that may confirm whether or not it will work, he said. This system aims to provide results in 24 hours, and on a much wider and accessible scale.
“One of the things about AI-based tests is that you don't need special equipment,” Grady said. “All you need is a digital slide scanner and everyday pathology slides. And that’s what allows us to really broaden the footprint of hospitals that are going to be able to incorporate this technology.”
Paige’s work isn’t limited to clinical diagnostics, either. The company is also developing programs for clinical trials and drug development, in concert with biopharma partners such as Janssen’s oncology division.
“There’s been tremendous interest in leveraging Paige’s technology to look for biomarkers for different indications, or leveraging outcomes data that we have access to through our partners,” he said. “The space of this opportunity is really enormous.”
Investors: A $100 million series C round this past January raised funds from Casdin Capital and Johnson & Johnson Innovation, while a months-long series B round last year gathered up to $70 million from Goldman Sachs, Healthcare Venture Partners and others.