Meta-analysis Of Fecal And Tissue Microbiome Of Human Colorectal Cancer Based On Amplicon Data

Researches focusing on colorectal cancer fecal microbiome using next generation sequencing continue,increasing evidence has supported correlations between colorectal carcinoma(CRC)and fecal microbiome dysbiosis.Turbulent fecal and tissue microbiome dysbiosis of colorectal carcinoma and adenoma had been identified,and some taxa had been proven to be carcinogenic.However,large scale on-site and off-site(surrounding adjacent)tissue microbiome characterization of CRC was underrepresented.In this study,we intended to study the colorectal cancer surrounding tissue microbiome signature with different kinds of samples groups for comparison.16S rRNA datasets were retrieved from fifteen studies,summing up to 2099 samples.After systematic data filtering and bioinformatic preprocessing,with each taxon considered as a feature,multiple methods including differential abundance analysis,random forest classification,co-occurrence network analysis,and Dirichlet multinomial mixtures enterotype analysis were conducted to investigate the microbial signatures of CRC,colorectal adenomas(CRA),and healthy control groups.Detailed results were listed below:(1)Principal Coordinates Analysis showed that the distance among tissue sample groups was far than that in stools groups,and differential abundance analysis identified different degrees of microbial feature overlaps among various comparison strategies.All these results characterized the common and different microbiome composition patterns in colorectal cancer sample groups.(2)We also found that the microbiota ecologies of surrounding adjacent tissue of CRC and CRA were highly correlated to the on-site parts according to network analysis,suggesting high microbial similarity between colorectal carcinoma adjacent tissue microbiomes and on-site counterparts.(3)The dysbiosis of the off-site tissue of the colonic cancer is distinctive and predictive.The AUC was 80.7%,96.0%,and 95.8% for CRC versus control random forest models using stool,tissue,and adjacent tissue samples,and 69.9%,91.5%,and 89.5% for corresponding CRA models,respectively.Despite cohort heterogeneity,we still discovered that the microbiome dysbiosis in adjacent tissues could discriminate colorectal carcinoma from healthy controls effectively independently in individual cohort.Compared with healthy individuals,Fusobacterium enriched along the time series,while Ruminococcus,Blautia and Dorea were depleted in diseased groups.(4)According to the enterotype analysis,both the sample group and the cohort information could be reflected in the enterotype results.Furthermore,consistent with researches where datasets were from,based on the enterotype of tissue samples,we identified some cohortspecific microbial signatures like high abundance of lactococcus in China cohort.This might be the crux in addressing classification generalization problems,suggesting that cohort factors should be included in future study.In conclusion,this study pointed out the microbiome signature of colorectal cancer adjacent tissues was highly similar to that of the onsite tissue,highlighted microbiota dysbiosis in the surrounding tissue and extended our knowledge on the microbial profile of colorectal cancer tissue.It also suggested that microbial features variation of cancerous lesion adjacent tissue might help to reveal microbial etiology of colonic cancer and could ultimately be applied for diagnostic and screening purposes.