Mount Sinai Develops AI Prediction Tool for Cancer Patient Response to ICI Therapy

January 22, 2025

By Bio-IT World Team 

January 22, 2025 | Researchers at the Tisch Cancer Institute at Mount Sinai, in collaboration with Memorial Sloan Kettering Cancer Center (MSKCC), have developed SCORPIO, an artificial intelligence (AI) tool that can predict how well patients with cancer will respond to immune checkpoint inhibitor (ICI) therapy by analyzing routine blood tests and clinical data. The model uses machine learning to identify patterns in the data that correlate with treatment outcomes. The team wanted to create an accessible tool that was “ubiquitous, low-cost, and already part of clinical workflows” to provide an accessible and scalable solution for predicting ICI efficacy, according to Diego Chowell, Ph.D., study lead and assistant professor of Immunology and Immunotherapy, Oncological Sciences, and Artificial Intelligence and Human Health at the Icahn School of Medicine at Mount Sinai.  

“By ‘learning’ from thousands of patients, SCORPIO can predict immune responses and clinical outcomes with remarkable accuracy, offering insights without requiring complex or costly genomic testing,” he explains.  

So far, SCORPIO has shown great promise. Other FDA-approved biomarkers, such as tumor mutational burden (TMB) and PD-L1 immunostaining, are useful and provide insight, but PD-L1 is not universally available and requires the use of various platforms, antibodies, and quality assurance practices. TMB uses resource-intensive genomic profiling (Nature, DOI: https://doi.org/10.1038/s41591-024-03398-5). Because of how SCORPIO is designed, it is more accessible, cost-saving, and practical. Furthermore, the results of the study demonstrated that SCORPIO consistently outperformed the aforementioned biomarkers in predicting overall survival and treatment efficiency.  

“One surprising finding was SCORPIO’s strong performance in real-world cohorts compared to clinical trial datasets,” reports Chowell. This discovery highlights the model’s adaptability to diverse patient populations and healthcare environments. SCORPIO’s ability to reflect immune-inflamed phenotypes through routine blood test features added a layer of biological relevance that hadn’t been fully anticipated, he adds. 

It is important to mention that SCORPIO’s ability to predict clinical benefit from ICI varied across different cancer types and cohorts, which suggests it is a challenge for the model to accurately predict clinical benefit (Nature, DOI: https://doi.org/10.1038/s41591-024-03398-5).  

Nonetheless, the model’s cost effective, simple ways will help both physicians and patients. Physicians and clinicians will have a reliable tool that can be integrated into routine clinical workflows and support making ICI therapy decisions. SCORPIO can reduce unnecessary treatments, minimize toxicities, and optimize resource allocation in oncology practices. Patients can look forward to more personalized and equitable care, as well as reduced risk of undergoing ineffective treatments and improving outcomes and quality of life. 

Chowell and his team hope that the model can be brought to a global scale and made accessible for patients everywhere and anywhere, “making it a cornerstone of equitable precision oncology.” The team is currently working to collaborate with hospitals and cancer centers to prospectively validate SCORPIO in diverse clinical environments.