Microbiota as a new indicator of colorectal cancer (CRC) heterogeneity

3615 Background: Location and somatic gene signature of CRCs may impact prognosis and therapy response. The aim was to characterize colon Microbiota in CRC patients regarding location, gene markers and outcome. Methods: Patients (N = 173) signed consent for whole metagenome (shot gun sequencing on Illumina HiSeq2500) analysis of stool DNA: 72 CRC (53 sporadic-S, 19 Lynch-L), 87 asymptomatic subjects (normal colonoscopy), 14 first degree healthy relatives from Lynch families. "MOCAT" pipeline was used, library sorted (Phred quality score ¡20 Alientrimmer v0.4.0) after exclusion of < 35 nt, human genes or phage sequences. Quality sequences were aligned (REFMG.V13) and most abundant genes constructed (MBMA program v0.1). The Shaman program (shaman.c3bi.pasteur.fr) was used. The number of bacteria was estimated (REFMG program). The linear model (GLM) was implemented in the DESeq2 R kit. Dfferences between Control (N = 87) and CRCs (N = 69), between L (N = 19) and S CRCs(N = 50), and between LCRC (N = 19) and Healthy Lynch relatives were obtained after interaction of age, BMI and gender was considered (GLM model). The P < 0.1 value was retained after correction (Benjamini and Hochberg). The specific taxonomic composition of the control and CRC groups was subjected to random analysis ( Caret's R package) with two optimization parameters (precision and kappa) in the model. Results: There was no difference for gender, age (p = 0.08) and BMI (p = 0.187) in the L and S CRCs. Significant differences were observed between Normal and CRCs, C-CRC and L-CRC, L-CRC and first degree relatives based on the common component (similarity of sequences): 13 species differentiated Normal and CCRs, two were more prevalent in L-CRCs. The panels of bacteria linked with location, MSI, Ras mutations, methylation phenotypes and survival were identified. No significant link was observed with TNM Staging: I (N = 17, 2L and 15S), II (N = 12, 5L and 7S), III (N = 20, 10L, 10S), IV (N = 22.1L, 21S). DFS might be dysbiosis dependent. Conclusions: CRC dysbiosis is location-dependent. Several bacteria are associated with Ras mutation, MSI, and methylation status. They may directly or through therapies impact the prognosis. Microbiota signature should be taken in consideration in trials.