Tissue Microbiome Profiling Identifies an Enrichment of Specific Enteric Bacteria in Opisthorchis viverrini Associated Cholangiocarcinoma

Abstract

Cholangiocarcinoma (CCA) is the primary cancer of the bile duct system. The role of bile duct tissue microbiomes in CCA tumorigenesis is unestablished. To address this, sixty primary CCA tumors and matched normals, from both liver fluke (Opisthorchis viverrini) associated (OVa, n=28) and non-O. viverrini associated (non-OVa, n=32) cancers, were profiled using high-throughput 16S rRNA sequencing. A distinct, tissue-specific microbiome dominated by the bacterial families Dietziaceae, Pseudomonadaceae and Oxalobacteraceae was observed in bile duct tissues. Systemic perturbation of the microbiome was noted in tumor and paired normal samples (vs non-cancer normals) for several bacterial families with a significant increase in Stenotrophomonas species distinguishing tumors vs paired normals. Comparison of parasite associated (OVa) vs non-associated (non-OVa) groups identified enrichment for specific enteric bacteria (Bifidobacteriaceae, Enterobacteriaceae and Enterococcaceae). One of the enriched families, Bifidobacteriaceae, was found to be dominant in the O. viverrini microbiome, providing a mechanistic link to the parasite. Functional analysis and comparison of CCA microbiomes revealed higher potential for producing bile acids and ammonia in OVa tissues, linking the altered microbiota to carcinogenesis. These results define how the unique microbial communities resident in the bile duct, parasitic infections and the tissue microenvironment can influence each other, and contribute to cancer.

Keywords: Cancer; Cholangiocarcinoma; Liver fluke; Microbiome.

Fig. 1

Tissue microbiome 16S rRNA profiles of the bile duct, gastric mucosa and liver. Barplots showing microbiome profiles of (a) bile duct tissue (4 out of 28 OVa and 4 out of 32 non-OVa) from CCA patients, (b) non-cancer hepatic tissue (n = 5), (c) non-cancer gastric mucosa (n = 4) and (d) bile fluid (n = 2) from CCA patients at family level resolution. Only families with mean relative abundance > 0.5% are shown. (e) Principle coordinates analysis based on Jensen-Shannon distance of the microbiome profiles (family level) of different tissue types (OVa: O. viverrini associated tissue).

Fig. 2

Comparison of O. viverrini associated and non-associated paired tumor-normal tissue microbiomes. (a) Boxplots of Yue-Clayton theta indices for quantifying similarity across tissue microbiomes (family level) within tumor-normal pairs, for O. viverrini associated (n = 28; OVa) tumors and non-O. viverrini associated (n = 32; non-OVa) tumors respectively. (b) Boxplots depicting the microbiome diversity of O. viverrini associated and non-associated tumor and adjacent normal tissues. (c) Boxplots showing the relative abundance of the 3 families identified to be enriched in O. viverrini associated (n = 28) vs non-O. viverrini (n = 32) associated tissues. All p-values were calculated using the Wilcoxon rank-sum test, where ***, ** and * represent FDR adjusted p-values < 0.01, 0.05 and 0.1 respectively.

Fig. 3

Composition of the O. viverrini microbiome based on whole metagenome profiling. Breakdown of the composition of the O. viverrini microbiome at family, genus and species levels. Only members with  > 0.5% abundance are shown.

Fig. 4

Functional analysis of the O. viverrini associated tissue microbiome. (a) Functional pathways that were identified to be differentially abundant in the O. viverrini associated (n = 28 pairs) and non-associated (n = 32 pairs) CCA tissue microbiomes. (b) Breakdown of bile salts into metabolic products implicated in carcinogenesis. (c) Boxplots of predicted abundances of the bsh gene in respective tissue microbiomes (OVa, n = 28; non-OVa, n = 32). All p-values were calculated using the Wilcoxon rank-sum test.