Yan Leyfman, Medical Oncologist, Co-Founder and Executive Director of MedNews Week, shared a post on X:
“Can AI better predict who will benefit from immunotherapy?
Despite transforming cancer treatment, most patients still don’t respond to immune checkpoint inhibitors, and current biomarkers often work well in one cancer type but fail in others.
A new study introduces COMPASS, a pan-cancer AI foundation model trained on more than 10,000 tumors across 33 cancer types. Instead of relying on black-box predictions, COMPASS interprets tumor biology through 44 immune-related concepts, including immune cell states, tumor microenvironment interactions, and signaling pathways.
Across 16 independent clinical cohorts spanning 7 cancer types and 6 different immunotherapies, COMPASS outperformed 22 existing prediction methods, improving both predictive accuracy and precision. Patients predicted to respond also had significantly longer overall survival.
Perhaps most interesting, the model doesn’t just predict response-it provides biologically interpretable insights into why tumors may resist immunotherapy, highlighting mechanisms such as TGF-β signaling, endothelial exclusion, CD4⁺ T-cell dysfunction, and B-cell deficiency.
As oncology moves toward precision medicine, AI models like COMPASS may help improve patient selection while also uncovering new therapeutic targets and combination strategies.”
Title: Generalizable AI predicts immunotherapy outcomes across cancers and treatments
Authors: Wanxiang Shen, Intae Moon, Thinh H. Nguyen, Michelle M. Li, Yepeng Huang, Nitya Nair, Daniel Marbach, Marinka Zitnik
Read The Full Article
Other articles featuring Yan Leyfman on OncoDaily.