AI Blood Test Maps Tumor Neighborhoods and Predicts Immunotherapy Response

Overview

• Researchers from Stanford Medicine and Mayo Clinic reported a blood test called Liquid EcoTyper that reads DNA methylation in plasma to reconstruct the tumor microenvironment using AI.
• They defined nine recurring cellular “neighborhoods,” or spatial ecotypes, across many cancers using spatial transcriptomics and tools such as CytoSPACE and Spatial EcoTyper.
• The same ecotype patterns appeared in at least 17 cancer types, and several were linked to outcomes on immune checkpoint drugs, with T cell–rich SE7 and SE8 tied to benefit and SE4 tied to resistance.
• In matched samples, ecotype levels inferred from blood mirrored those in tumor biopsies and forecasted patient outcomes across immunotherapy cohorts involving more than 1,300 patients.
• The team has begun serial monitoring studies and reported blinded validation and active programs in melanoma, lung, bladder, and mesothelioma, while larger prospective trials are still needed before clinical use.