Understanding evolutionary and ecological mechanisms of sotorasib resistance in KRAS G12C-mutant non-small cell lung cancer

KRAS G12C mutation is a major oncogenic driver in non-small cell lung cancer, with targeted therapy using sotorasib offering clinical promise. However, resistance inevitably emerges. Evolutionary therapies based on collateral sensitivity exploit molecular and pharmacologic vulnerabilities that arise in evolved resistant populations and provide a promising strategy to guide evolution-informed sequential or combination therapies to delay resistance. We developed a novel in vitro experimental evolution framework using a KRAS G12C-mutant H358 cell line. Multiple independent replicates were continuously exposed to 25nM sotorasib for around 3 months across 8 sequential treatment cycles, while independent control replicates were treated without drug. Most treated replicates (4/6) developed robust resistance at cycle 8. To map evolutionary shifts in drug response, we performed collateral sensitivity screening using a panel of 10 FDA-approved agents targeting pathways relevant to KRAS signaling. Drug sensitivity shifted over time and resistant replicates evolved cross-resistance to EGFR/HER2-targeted inhibitors (e.g. Gefitinib and Lapatinib). In contrast, a statistically significant increase in paclitaxel sensitivity was detected, suggesting a collateral vulnerability associated with altered cell-cycle regulation as a compensatory response to KRAS inhibition. Bulk RNA-Seq revealed clear transcriptional changes accompanying resistance acquisition. Gene set enrichment analysis identified upregulation of several hallmark pathways in resistant group, including epithelial-mesenchymal transition, mTORC1 signaling, and inflammatory signaling by the end of experiment. Moreover, further analysis revealed that while replicates initially respond similarly at cycle 4, their molecular profiles diverge over time and give rise to considerable heterogeneity by cycle 8, despite a shared evolutionary origin. Drug resistance is also shaped by dynamic ecological interactions, so we evaluated ecological vulnerabilities using a co-culture framework that quantifies ecological coefficients based on relative growth rates of resistant and sensitive cells. Preliminary assays show that several resistant populations outcompete sensitive cells even without drug, contradicting the expected fitness cost of resistance and suggesting that certain resistance mechanisms may enhance baseline fitness. Competition outcomes shifted across sotorasib concentrations, indicating that ecological interactions are dose-dependent, highlighting the potential for ecology-informed dosing strategies to reduce competitive release and delay resistance. Together, we provide novel insights into the eco-evolutionary processes underlying sotorasib resistance and identify vulnerabilities that may be leveraged to design therapies that delay or redirect resistance.