%20(1).jpg)
The paper, authored by Tomas Konecny, Nate Zadirako, Arpine Grigoryan, Melina Tamazyan, Sveta Mnatsakanyan, Luiza Stepanyan, Henry Loeffler-Wirth, Sean Bourdelais, Gabriel Mednick, Chloe Delepine, Dhan Chand, and Hans Binder, is titled "Transcriptomic profiling identifies immunotherapy-responsive phenotypes in microsatellite-stable metastatic colorectal cancer."
Microsatellite-stable (MSS) colorectal cancer accounts for the vast majority of metastatic colorectal cancer cases and has historically shown little benefit from immunotherapy. Using self-organizing map (SOM) machine learning, the team analyzed tumor biopsies from patients treated with botensilimab and balstilimab, two immune checkpoint inhibitors developed by Agenus, in a phase I clinical trial. The analysis revealed four distinct molecular tumor types, each with its own immune landscape: a liver-like type, a proliferative type, an immune-enriched type, and a mesenchymal-like type.
Patients whose tumors fell into the immune-enriched or mesenchymal-like categories showed better clinical responses and longer overall survival after treatment. The study also identified an immunophenotype axis driven by interferon-gamma signaling and APOBEC3 gene activity, which was not captured by standard measures of tumor mutation burden, suggesting that targeted, focal hypermutation rather than overall mutational load drives immune recognition in this cancer type. Paired biopsy analysis further showed that treatment actively shifted tumors toward more immune-active states.
"Being able to stratify MSS colorectal cancer tumors by their immune microenvironment brings us closer to selecting the right patients for the right therapy. Our molecular map of the tumor landscape gives us the resolution to do that from standard tumor biopsy data," said Hans Binder, ABI.
The work was conducted in collaboration with Agenus, a US-based biotech company, as part of the Agenus C-800-01 phase I clinical trial (NCT03860272). Tomas Konecny and Nate Zadirako are co-first authors; Dhan Chand (Agenus) and Hans Binder (ABI) are co-corresponding authors.
The article is available in open access. Read more here: https://doi.org/10.1038/s41388-026-03861-2
