Hang Yin

Supplemental Table 1. Associations of continuous (m2CLR) 16S microbiome with CRC-specific mortality (N=581)

Supplemental Figure 5 shows the m2CLR distribution (boxplots) of two MetaCyc pathways of interest stratified by population group.

Supplemental Figure 1 displays relative abundance of eight different phyla stratified by tissue macrodissection status, either macrodissection or scraping the entire slide.

Supplemental Figure 2 shows the distribution (boxplots) of an alpha diversity metric, Shannon entropy, stratified by CRC-specific mortality status.

Supplemental Table 4. Associations of Fusobacteriota phylum with CRC-specific mortality stratified by right & left-sided tumor location, overall & by population group

Supplemental Figure 4 shows the m2CLR distribution (boxplots) of the phylum Fusobacteriota, the genera Fusobacterium and Leptotrichia, stratified by population group. Due to high percentages of zeros the median of all boxplots is zero.

Supplemental Table 2. Associations of 16S microbiome positivity with CRC-specific mortality among those microsatellite instability low or microsatellite stable (N=415)

Supplemental Table 3. Associations of continuous (m2CLR) 16S Fusobacteriota phylum & genera with CRC-specific mortality stratified by population group

Supplemental Figure 3 shows two components of beta diversity, PCoA 1 and PCoA 2, as well as ellipses to compare if beta diversity using weighted unifrac was different by CRC-specific mortality status.

Background: The microbiome is an important component of the tumor microenvironment implicated in colorectal cancer. However, its relationship with colorectal cancer–specific mortality remains unclear. Methods: We included 581 participants with colorectal cancer (167 African American, 176 Alaska Native, 118 Hispanic, and 120 non-Hispanic White) from the Translational Research Program in Cancer Differences across Populations (TRPCDP). We sequenced the V4 region of the 16S rRNA bacterial gene using DNA extracted from formalin-fixed, paraffin-embedded tumors. A total of 204 participants died of colorectal cancer and 377 did not die of colorectal cancer. Participants who died of colorectal cancer were matched to participants who did not die of colorectal cancer during follow-up by age, sex, tumor site, tumor stage, year of diagnosis, and population group. Logistic regression estimated odds ratios (OR) and 95% confidence intervals (CI) for associations between bacterial presence and colorectal cancer–specific mortality, adjusting for matching factors and tissue macrodissection status. Results: Individuals who died from colorectal cancer were 1.71 times as likely to have bacteria from the Fusobacteriota phylum present in their tumors (OR = 1.71; 95% CI = 1.19–2.47). Associations with Fusobacteriota were strongest among African American participants (OR = 2.36; 95% CI = 1.14–4.99) compared with other populations; however, this difference was not statistically significant (OR range = 1.05–1.38; Pinteraction = 0.697). Candidate pathways of pyruvate fermentation to acetate and lactate II and peptidoglycan biosynthesis I were associated with higher odds of colorectal cancer death. Conclusions: Fusobacteriota was significantly associated with colorectal cancer–specific mortality with noted differences across populations. Impact: This finding highlights the tumor microbiome as a candidate for further investigation into colorectal cancer outcome disparities.