Project Summary Colorectal cancer (CRC) is the fourth most common cancer, and the second leading cause of cancer death in the United States. Activating oncogenic mutations in KRAS or NRAS are found in >45% of CRC, driving tumor progression and influencing efficacy of both cytotoxic and targeted therapies; these mutations often co-occur with mutations of other oncogenes and tumor suppressors, including APC, TP53, BRAF, PI3KCA, and others. Importantly, structure-function analysis of relatively common RAS mutations in G12, G13, Q61, and other codons indicate these have non-equivalent transforming potential and modes of action. These differences can have clinical impact; for example, G12 mutations confer resistance to EGFR-targeting drugs, but G13D mutations do not. Our goal is to better understand the significance of KRAS mutations in CRC, and to leverage this work to better stratify CRC patients for selection between treatment options. In preliminary studies, using a dataset of 13,000 CRC specimens, we have investigated genomic patterns of variance in RAS-family proteins in CRC. From this recently published work, we have identified novel hotspot mutations, and revealed striking differences in RAS mutational pattern occurring in distinct patient subgroups, based on microsatellite instable (MSI) versus microsatellite stable (MSS) status, colon versus rectum as primary tumor subsite, patient age, and in one case, patient sex. In a second study, we have also identified age-related patterns of mutation of additional driver genes, including APC, BRAF, and FAM123B, characterizing distinct patient subgroups. However, we were limited by the study sample size from addressing a number of important questions; in particular, whether specific alleles of KRAS and NRAS segregated with mutations in additional driver genes in a manner that predicted signaling pathway dependency in specific patient subgroups. Our goal is to address these and other questions in detail, using information for ~34,000 CRC tumors provided by Foundation Medicine, and for ~8,400 CRC tumors we compiled from public datasets. To this end, we propose two Aims. In Aim 1, we will identify the complete repertoire of RAS hotspot mutations, including 3D hotspots. We will align hotspots to RAS structures, and establish their covariance with clinicopathological variables (including tumor subsite, age, sex, tumor mutation burden (TMB, including specific consideration of hypermutated tumors), and MSI/MSS status) and underlying targeting of trinucleotide contexts associated with specific mutational processes. We will analyze mutational patterns that differ between the two primary KRAS isoforms, KRAS4B and KRAS4A, given recent data suggesting an important specific role for KRAS4A. In Aim 2, we will elucidate gene, codon, and variant level analysis of co-occurrence of RAS mutations with other driver genes commonly mutated in CRC, and we will determine whether co-variance patterns differ based on clinicopat...