SO-26 Identification and quantification of the microbiome in colorectal cancer metastases

Abstract

Recent research has identified the existence of living bacteria in the tumor microenvironment, as well as cancer-specific bacteria in blood of patients. Early evidence suggested a role of the gut and tissue microbiome in the development of colorectal cancer (CRC). However, the part of this intra-tumor microbiome in promoting or aiding metastasis to distant organs remains unexplored. Interestingly, similar bacterial strains from CRC primary tumors have been identified in patient-matched liver metastases. Few mechanistic hypotheses have been proposed of how bacterial translocation to distant sites occurs, or whether the tumor-resident microbiome can influence the spreading or the behavior of metastases and their local immune response. We aimed to characterize the intra-tumor bacterial signature of metastatic colorectal cancer (mCRC) through a large-scale study with different cohorts from several biobanks in Belgium. We collected 378 frozen samples from 99 patients composed by 104 primary colorectal tumors (PT) and 99 patient-matched liver metastases (LMT), both associated with normal adjacent tissue (NAT) as control (n = 83 and 58, respectively). We added primary liver tumors (hepatocellular carcinoma (HCC, n=28) and cholangiocarcinoma (CGC, n=6)) from 27 patients as comparative cohort for LMT. The V4 region of the bacterial 16S rRNA gene was sequenced. The computational DADA2 pipeline was used to process the sequencing data and determine the abundance of amplicon sequence variants (ASV). We strictly applied sterile conditions from sample collection to sequencing, with negative and positive controls along the process to minimize contamination and misinterpretation of the results. Potential contaminants were removed bioinformatically by applying a series of stringent filters. Bacterial sequences were identified in all tissue types. Based on preliminary results, the principal component analysis based on bacterial composition of the samples revealed significant sample grouping (PERMANOVA, p-value = 1e-04) per tissue type (PT, LMT and HCC-CGC). The read number and bacterial composition of associated NAT were similar to their associated tumor tissue. PT, as expected by their colonic origin, presented higher bacterial biomass than LMT, yet the patient-specific microbiome composition in LMT was partially overlapping with PT, consisting with previously reported prevalent taxa. The number of shared ASVs between PT and LMT was significantly higher within than between patients (linear regression model with p-value = 4,22e-10), suggesting an individual bacterial transfer. This transfer appeared to be a probabilistic event, as it correlated with taxa abundance, and this transfer was more likely for the most abundant bacteria in the PT. Our preliminary results suggest the presence of a low bacterial biomass in LMT with characteristics closer to PT compared to primary tumors of the liver, suggesting a “per-tumor type” rather than “per-organ” bacterial signature and abundance. A within-patient bacterial transfer from PT to LMT seems to be observed for the most abundant taxa. These preliminary results are currently in a validation process. If our results are confirmed, this multicentric large-scale study would allow us to characterize and classify tumor tissue bacterial signature in mCRC to compare it with genomic and immune tumor features.