Re-purposing 16S rRNA gene sequence data from within case paired tumor biopsy and tumor-adjacent biopsy or fecal samples to identify microbial markers for colorectal cancer

Microbes colonizing colorectal cancer (CRC) tumors have the potential to affect disease, and vice-versa. The manner in which they differ from microbes in physically adjacent tissue or stool within the case in terms of both, taxonomy and biological activity remains unclear. In this study, we systematically analyzed previously published 16S rRNA sequence data from CRC patients with matched tumor:tumor-adjacent biopsies (n = 294 pairs, n = 588 biospecimens) and matched tumor biopsy:fecal pairs (n = 42 pairs, n = 84 biospecimens). Procrustes analyses, random effects regression, random forest (RF) modeling, and inferred functional pathway analyses were conducted to assess community similarity and microbial diversity across heterogeneous patient groups and studies. Our results corroborate previously reported association of increased Fusobacterium with tumor biopsies. Parvimonas and Streptococcus abundances were also elevated while Faecalibacterium and Ruminococcaceae abundances decreased in tumors relative to tumor-adjacent biopsies and stool samples from the same case. With the exception of these limited taxa, the majority of findings from individual studies were not confirmed by other 16S rRNA gene-based datasets. RF models comparing tumor and tumor-adjacent specimens yielded an area under curve (AUC) of 64.3%, and models of tumor biopsies versus fecal specimens exhibited an AUC of 82.5%. Although some taxa were shared between fecal and tumor samples, their relative abundances varied substantially. Inferred functional analysis identified potential differences in branched amino acid and lipid metabolism. Microbial markers that reliably occur in tumor tissue can have implications for microbiome based and microbiome targeting therapeutics for CRC.