Bridging the parallel worlds of multiplex imaging and H&E microscopy
Researchers led by Ludwig Harvard’s Sandro Santagata, Peter Sorger, Jia-Ren Lin and Yu-An Chen described in a June Nature Cancer paper a platform technology that enables integration of the methods used in surgical pathology—the examination of hematoxylin and eosin (H&E) stained slides—with emerging research methods in multiplexed tissue imaging for visualization of multiple molecular markers in individual cells. The new platform, Orion, allowed both human experts and artificial intelligence algorithms to identify cellular and molecular features that predict progression-free survival of colorectal cancer (CRC) patients. Analyzing cancer specimens from 40 CRC patients, the researchers sifted through some 15,000 combinations of biomarkers to identify those most tightly linked to patient prognosis. They then applied them to samples from an additional cohort of CRC patients whose outcomes were known. Their biomarkers accurately predicted—just a one in 20 chance of being wrong—the likelihood of poor prognosis. The approach revealed interesting relationships between the molecular markers, cell morphology and tumor topography. One such finding indicated that inflammation—or immune activity—at the rim of the tumor is of pathological significance, while another revealed the molecular basis of a tissue morphology associated with a likelihood of metastasis.