PROJECT SUMMARY/ABSTRACT Colorectal cancer (CRC) incidence and mortality vary greatly across population groups in the United States (US). This variability in CRC rates is marked by differences in the distribution of clinicopathologic factors (e.g., tumor site, age at CRC onset, consensus molecular subtypes) across population groups, which also has implications for outcomes. In exploring factors that may contribute to these observed differences, minimal consideration has been paid to the potential contribution of the microbiome. The presence of a gut microbiome (i.e., all of the bacteria in the gut) is common across all human populations, but the composition of the microbiome can vary greatly across populations. The composition of the gut microbiome has plausible implications for a variety of pertinent health outcomes, including CRC. Increasing evidence indicates that specific gut bacteria, or imbalances in bacterial populations, could play a role in the natural history of CRC. Prior studies have implicated a small number of candidate bacteria (e.g., Fusobacterium nucleatum; Fn) as contributors to CRC etiology and survival; however, our early studies suggest these candidates are unlikely to be significant contributors to population group CRC differences. Instead, we are identifying novel candidate bacteria that are disproportionately prevalent in tumors among population groups that experience higher CRC rates. The objective of this project is to test our overarching hypothesis that population differences in the tumor microbiome contribute to differences in CRC attributes and outcomes. In pursuit of this objective, we will leverage and build upon existing biospecimen and data resources, infrastructure, and preliminary findings accumulated through the Translational Research Program in Cancer Differences across Populations (TRPCDP). In Aim 1, using quantitative digital droplet PCR (ddPCR) assays, we will: (1a) test for differences in the prevalence and abundance of novel candidate bacteria in colorectal tumors across four population groups, and (1b) test for associations with risk of lethal CRC overall and within population groups. For Aim 2, we will expand our evaluation of (2a) these novel candidates and (2b) microbial 16S rRNA gene sequencing (16S rRNAseq) data to examine relationships with clinicopathologic factors, with a focus on attributes that differ in their distribution across population groups and have implications for outcomes. In Aim 3, we will use artificial intelligence, such as machine learning strategies, to identify tumor-associated microbial signatures in CRC associated with CRC-specific mortality and will assess differences in the distribution of those signatures across population groups. Combining all of the results from our aims can be used to inform more targeted cancer treatment strategies, particularly in populations experiencing the greatest burden of CRC.