AI tumor microenvironment analysis is becoming a major focus in precision oncology. New findings presented by Agenus and Noetik suggest artificial intelligence could improve patient selection for advanced immunotherapy treatments in metastatic colorectal cancer.
The retrospective study evaluated Noetik’s TARIO-2 platform in patients treated with botensilimab and balstilimab. Researchers analysed standard pretreatment H&E pathology images to identify tumour microenvironment patterns linked to treatment response and survival outcomes.
Importantly, the findings demonstrated meaningful predictive value in microsatellite stable metastatic colorectal cancer, commonly referred to as MSS mCRC. This disease subtype historically shows limited responsiveness to traditional checkpoint inhibitors.
How AI Tumor Microenvironment Analysis Supports Precision Immunotherapy
The TARIO-2 platform uses artificial intelligence to interpret spatial relationships inside tumour pathology images. Unlike conventional biomarker approaches, the model does not depend on a single molecular marker. Instead, it evaluates complex tumour microenvironment characteristics from routinely generated H&E slides.
Consequently, the technology may offer a more scalable method for identifying patients likely to benefit from next-generation immunotherapy combinations.
The study included 113 efficacy-evaluable patients enrolled in the C-800-01 Phase 1b trial. Researchers assessed several solid tumour cohorts, including:
- MSS metastatic colorectal cancer without active liver metastases
- Ovarian cancer
- Sarcomas
Among MSS mCRC patients, the AI-selected subgroup achieved a 64% response rate following BOT+BAL treatment. In contrast, the remaining cohort recorded only a 9% response rate.
Moreover, overall survival improved significantly in the AI-identified subgroup. Median survival was not reached, while the hazard ratio declined to 0.18 versus the broader patient population.
Why BOT+BAL Is Generating Industry Attention
Botensilimab and balstilimab represent a differentiated immunotherapy strategy developed by Agenus. The combination targets both CTLA-4 and PD-1 pathways using an Fc-enhanced mechanism designed to activate broader anti-tumour immune responses.
Traditional biomarkers such as PD-L1 expression and tumour mutational burden have not consistently predicted BOT+BAL activity. Therefore, AI-driven tumour microenvironment analysis could become increasingly important for patient stratification.
Botensilimab was specifically engineered to improve immune activation within “cold” tumours. These cancers often resist standard checkpoint inhibition therapies. According to the company, the therapy enhances T-cell priming, antigen presentation and myeloid cell activation.
Meanwhile, balstilimab acts as a fully human anti-PD-1 monoclonal antibody. The therapy blocks PD-1 interactions with PD-L1 and PD-L2 ligands.
AI-Based Biomarkers Could Reduce Complexity in Oncology Diagnostics
One notable advantage of the TARIO-2 platform involves accessibility. The model analyses standard pathology slides already generated during routine cancer diagnosis. Consequently, healthcare providers may avoid more expensive and operationally complex tissue profiling technologies.
This approach could improve scalability across oncology centres worldwide. Additionally, it may reduce barriers to biomarker adoption in community healthcare systems.
Ryan Dalton, Senior Computational Scientist at Noetik, stated that pathology images contain substantial biological information that remains difficult to interpret visually. He added that AI analysis may uncover tumour microenvironment patterns associated with stronger clinical benefit from BOT+BAL therapy.
Furthermore, Agenus executives described the findings as an important translational research milestone. The company plans prospective validation studies to confirm TARIO-2’s predictive capabilities in MSS colorectal cancer and broader solid tumour settings.
What These Findings Mean for the Future of Immuno-Oncology
The oncology sector increasingly relies on AI-enabled diagnostics to improve therapeutic precision. Therefore, studies linking pathology imaging with clinical outcomes could reshape future drug development strategies.
Importantly, MSS colorectal cancer remains one of the largest unmet needs in immunotherapy. Most patients receive limited benefit from currently approved checkpoint inhibitors. However, AI-guided patient selection may improve efficacy rates while supporting more personalised treatment strategies.
The results also reinforce broader industry trends toward multimodal biomarker development. Rather than relying exclusively on genomic sequencing, developers are now integrating imaging, spatial biology and machine learning tools into clinical decision-making.
If prospective studies confirm these findings, AI tumor microenvironment analysis could become a valuable companion diagnostic approach for advanced immunotherapy combinations.
