Spatial transcriptomics has given cancer researchers a powerful way to see not only which genes are active in a tumor, but where that activity is taking place. Yet the resulting maps can be difficult to compare across tumors because each tumor microenvironment is compositionally and functionally heterogeneous.
Researchers at the University of Chicago now report a framework for comparing these spatial maps across tumor types. In a study titled “Pan-tumor spatial transcriptomics reveals conserved properties of tumor organization,” published in Cell Reports Medicine, the team analyzed spatial transcriptomics data from 262 solid tumors and found that tumor microenvironments could be organized into recurring, hierarchically structured multicellular regions, which the researchers call “spatial groups.”
Those spatial groups (SGs) appeared to capture recognizable biological domains, ranging from global tissue context to local cellular neighborhoods. When the researchers compared tumors through these groups, they found that the dominant axis separating tumors was the spatial heterogeneity of immune biology. In a small, independent retrospective cohort of 16 patients with non-small cell lung cancer treated with immune checkpoint blockade, the classification distinguished clinical responders from nonresponders and captured features associated with sensitivity to immunotherapy. “Together, these findings suggest that SGs may be important organizing domains of the TME that relate spatial structure, biological function, and response to therapy,” the authors wrote.
The work addresses a central problem in spatial biology: how to compare tumors at the level of tissue organization. Arjun Raman, MD, PhD, assistant professor of pathology at the University of Chicago and senior author of the study, likened the challenge to understanding a flock of birds. Individual cells behave in their own ways, but they also form collective structures that influence the behavior of the tumor as a whole.
“What we found out was we could describe tumors not as a composition of a whole bunch of cells, but like a flock of birds, where all the cells talk to each other and then create these subunits, and then subunits interact with each other to create meta subunits, and so on and so forth until you get the whole biopsy sample,” Raman said.
By treating those subunits as spatial groups, the researchers could compare tumor “floor plans” using a statistical and machine learning framework. The approach places tumors in a comparative latent space, where samples are arranged according to spatial similarity. In principle, that could allow researchers or clinicians to ask whether a newly profiled tumor resembles tumors previously associated with response or resistance to a given therapy.
Raman said the longer-term goal is to bring this kind of comparative spatial analysis closer to precision oncology. “You could have a person who comes in with their unicorn of a tumor,” he said. “You then perform profiling on it, put the data into the comparative space, and within a few hours you can see if they should or should not get regimen X.”
For now, the treatment-response finding remains early. The immunotherapy analysis involved only 16 non-small cell lung cancer cases, and the authors framed spatial groups as candidate organizing domains that may relate to tumor structure, biological function, and therapy response. Still, the study suggests that spatial transcriptomics could move beyond producing detailed tumor maps to providing a common language for comparing them.
