Agenus and Noetik presented new research at the 2026 American Society of Clinical Oncology Annual Meeting showing how artificial intelligence models trained on routine tumor pathology images may help predict patient response and survival outcomes in metastatic colorectal cancer. The findings highlight the growing role of AI-powered pathology and multimodal biomarker discovery in precision oncology, particularly for difficult-to-treat microsatellite stable (MSS) colorectal cancers where immunotherapy response rates have historically remained low.
At ASCO 2026, biotechnology company Agenus and AI oncology platform developer Noetik unveiled clinical data linking artificial intelligence-driven pathology analysis with treatment outcomes in patients receiving BOT/BAL therapy for microsatellite stable metastatic colorectal cancer (MSS mCRC).
The study focused on applying machine learning models to routine pre-treatment pathology images to identify predictive signals associated with therapeutic response and survival outcomes. Researchers said the approach could improve patient stratification for immunotherapy combinations in colorectal cancer, one of the largest and most treatment-resistant oncology categories globally.
The announcement reflects a broader transformation underway across oncology, where AI-powered biomarker discovery is becoming increasingly central to drug development, pathology workflows, and precision medicine strategies.
Unlike traditional biomarker testing that often relies on genomic sequencing or specialized assays, the Agenus-Noetik approach analyzes standard hematoxylin and eosin (H&E) tumor pathology slides using deep learning models trained to identify clinically relevant spatial and cellular patterns invisible to human interpretation alone.
The companies say this enables scalable, lower-friction biomarker analysis using pathology infrastructure already widely deployed across hospitals and cancer centers.
The research specifically examined outcomes tied to BOT/BAL, Agenus’ investigational immunotherapy combination targeting treatment-resistant MSS metastatic colorectal cancer. MSS colorectal tumors account for roughly 95% of colorectal cancer cases and have historically shown limited responsiveness to checkpoint inhibitor therapies that transformed treatment in other tumor types.
That has made MSS colorectal cancer one of the most important unsolved challenges in immuno-oncology.
According to the companies, Noetik’s AI platform identified image-derived biomarkers associated with both therapeutic response and overall survival in patients treated with BOT/BAL. The analysis used routine pathology slides collected prior to treatment initiation, suggesting AI-derived pathology features could potentially help identify which patients are more likely to benefit from therapy.
The findings also reinforce a rapidly expanding industry trend toward multimodal AI models in healthcare.
Rather than relying solely on genomic data, modern oncology AI systems increasingly combine pathology images, clinical records, radiology scans, molecular profiling, and treatment histories to improve predictive accuracy. Technology companies and healthcare platforms including Microsoft, Google, NVIDIA, and Tempus AI have all expanded investments in AI-driven healthcare infrastructure and medical foundation models over the past year.
The market opportunity is substantial.
According to Statista and Grand View Research estimates, the global AI in healthcare market is projected to exceed $180 billion by the early 2030s as hospitals, pharmaceutical companies, and research organizations adopt machine learning systems for diagnostics, drug discovery, and clinical decision support.
Pathology has emerged as one of the fastest-moving segments within that transformation.
Digital pathology adoption accelerated significantly following advances in cloud infrastructure, GPU computing, and medical imaging AI frameworks. Deep learning models can now process extremely large pathology datasets to identify tumor microenvironment patterns, immune infiltration signatures, and spatial biomarkers that may correlate with therapy response.
For pharmaceutical companies, those capabilities are increasingly valuable in clinical development.
AI-powered biomarker discovery may help drug developers improve patient selection, optimize trial enrollment, reduce development costs, and identify subpopulations more likely to benefit from experimental therapies. In oncology, where clinical trial failure rates remain high, predictive AI models could significantly improve development efficiency.
The Agenus-Noetik collaboration also highlights how AI companies are becoming more deeply integrated into biotech research pipelines.
Rather than serving purely as analytics vendors, AI platforms are increasingly operating as strategic clinical development partners, helping generate novel biomarkers and computational insights that can shape therapeutic positioning and regulatory strategies.
Industry analysts say these partnerships may become increasingly important as pharmaceutical companies race to operationalize generative AI, multimodal machine learning, and computational biology platforms across drug development workflows.
The ASCO presentation arrives during a period of heightened investor interest in AI-enabled healthcare companies.
Public markets and venture investors have poured billions into AI-driven healthcare startups focused on pathology, radiology, clinical documentation, drug discovery, and genomics infrastructure. However, healthcare AI adoption still faces regulatory, validation, interoperability, and clinical trust challenges.
One major issue is explainability.
Healthcare providers and regulators continue pushing for greater transparency around how AI models generate predictions, especially when systems influence treatment selection or patient care decisions. Companies operating in clinical AI environments increasingly need to demonstrate not only model performance but also reproducibility, fairness, and clinical utility.
For Agenus, the findings may strengthen the positioning of BOT/BAL within the competitive immuno-oncology landscape.
Checkpoint inhibitors transformed treatment in melanoma, lung cancer, and several other tumor types, but MSS colorectal cancer has remained a difficult frontier due to low immunogenicity and poor response rates. AI-guided biomarker strategies may help identify subsets of patients more likely to respond to emerging combination therapies.
The results also reinforce how AI is evolving from a research optimization tool into a core component of precision medicine infrastructure.
As pathology digitization expands and multimodal AI models mature, computational pathology platforms may increasingly shape how oncologists diagnose disease, select therapies, design trials, and evaluate treatment outcomes across large-scale healthcare ecosystems.
