AI Unlocks Genetic Clues for Precision Oncology

A USC-led study has made a major advance in precision oncology by analyzing genetic and clinical data from over 78,000 cancer patients across 20 different cancer types. Researchers identified nearly 800 genetic mutations that significantly influence how patients respond to treatments such as immunotherapy, chemotherapy, and targeted therapies. The team also found 95 genes closely associated with survival outcomes in cancers like breast, ovarian, skin, and gastrointestinal cancers. Using these findings, they developed a machine learning tool that can predict how patients with advanced lung cancer might respond to immunotherapy, helping clinicians make more personalized treatment decisions.

Key genes highlighted in the study include TP53, CDKN2A, and CDKN2B, which play crucial roles in determining treatment effectiveness and patient prognosis. The machine learning models were validated using real-world data, confirming the reliability of these genetic markers for guiding therapy choices. The research underscores the importance of genetic profiling in cancer care, allowing doctors to avoid ineffective treatments and focus on those most likely to benefit each patient.

By integrating large-scale genomic and clinical data with advanced computational methods, the study demonstrates how artificial intelligence can uncover complex mutation-treatment interactions that might otherwise go unnoticed. The findings pave the way for more precise, personalized cancer therapies and could expand access to effective treatments like immunotherapy for more patients. Published in Nature Communications, this work represents a significant step toward making individualized cancer treatment a clinical reality.