SAN DIEGO – A new AI model applied to routine pathology slides accurately predicts outcomes and response to immunotherapy in patients with metastatic non-small cell lung cancer (NSCLC). The study was presented at the American Association for Cancer Research (AACR) Annual Meeting.
“Immunotherapy has transformed cancer treatment, but only a subset of patients benefit from it, and predicting who will respond remains challenging,” said Rukhmini Bandyopadhyay, PhD, a postdoctoral fellow at The University of Texas (UT) MD Anderson Cancer Center. This study represents, to our knowledge, the first deep learning-based pathomics biomarker rigorously validated across international real-world cohorts and a Phase III randomized clinical trial, directly addressing one of the most urgent unmet needs in precision oncology: reliable patient selection and stratification for immunotherapy.
Pathomics applies computational and machine learning methods for high-throughput analysis of digital pathology images to extract large-scale data related to cell and tissue architecture linked to disease outcomes. Bandyopadhyay and colleagues developed a deep learning survival prediction model called Pathology-driven Immunotherapy Optimization (Path-IO), which identifies patients most likely to benefit from immunotherapy by studying patterns across tissue. The model combines imaging and clinical data to estimate whether a patient may have a higher or lower risk of poor outcomes from immunotherapy.
The platform was tested in a study that included 797 immune checkpoint inhibitor-treated NSCLC patients from UT MD Anderson, with external validation in 280 additional patients from Mayo Clinic, Gustave Roussy, and the Phase III Lung-MAP S1400I trial. The model reliably stratified patients into higher and lower risk groups, with high-risk patients at UT MD Anderson showing more than double the risk of death or disease progression compared to low-risk counterparts.
Model performance was evaluated using the concordance index (C-index), which measures how well each biomarker distinguishes between patients with different outcomes. Notably, Path-IO consistently outperformed PD-L1, the FDA-validated standard-of-care biomarker for guiding immunotherapy use in NSCLC patients, across both discovery and test cohorts. Path-IO achieved C-indices of 0.69 for overall survival and 0.65 for progression-free survival in the discovery cohort, compared to PD-L1’s limited performance.
Combining pathology-based predictions with radiomics and clinical data further improved the model’s performance, with the C-index increasing significantly. Given that the approach is designed for routine pathology slides, it can be integrated into existing clinical workflows without significant expense compared to other emerging technologies. However, as the study is retrospective, further investigation is necessary to predict not only which patients will benefit from immunotherapy but also what type they may respond to. Future directions include prospective validation and the integration of comprehensive molecular profiling to enhance predictive performance.
Use the database as your supply chain compass →
