Comprehensive Summary
Amir Jassim and colleagues present RECODR (REsistance through COntext DRift), a computational method that detects treatment resistance by analyzing shifts in gene co-expression context rather than relying simply on gene expression levels. By utilizing an approach inspired by natural language processing, RECODR identifies context-dependent interactions of genes throughout the course of therapy. In this study, RECODR was applied to both preclinical and clinical tumor samples, and the method revealed resistance mechanisms that were not detectable through standard approaches. Choroid plexus carcinoma mouse models were also used to analyze gene profiles and tumorigenesis. RECODR predicted cell resistance against radiation monotherapy and combination therapy resistance.
Outcomes and Implications
The findings demonstrate that resistance can emerge not only from changes in gene activity but also from differing gene interaction contexts emphasizing the need to look beyond gene expression levels in cancer research. RECODR provides a framework for finding drivers of resistance and designing more effective combination therapies. By making resistance mechanisms more predictable, the method can be used to improve treatments, guide therapy decisions, and broaden the landscape for cancers with limited options including aggressive pediatric and breast cancers. RECODR has the potential to strengthen oncology by changing how resistance to therapy is identified and addressed in clinical practice.