Abstract
A well-established genotype to phenotype relationship in genomic medicine is that activating KRAS mutations indicate resistance to anti-EGFR agents. We used a computational model of Ras signaling to investigate a confusing exception to this relationship whereby colorectal cancers with one specific, constitutively-active, mutant, KRAS G13D, respond to anti-EGFR agents. Our computational simulations of the biochemical processes that regulate Ras suggest EGFR inhibition reduces wild-type Ras activation in KRAS G13D mutant cancer cells more than in other KRAS mutant cancer cells. The model also reveals a non-intuitive, mutant-specific, dependency of wild-type Ras activation on EGFR. This dependency is determined by the interaction strength between a KRAS mutant and tumor suppressor neurofibromin. Our prospective experiments confirm this mechanism that arises from the systems-level regulation of Ras pathway signaling. Overall, our work demonstrates how systems approaches enable mechanism-based inference in genomic medicine.