Integrated analysis of colorectal neoplasia tissue and gut microbiome associated serum metabolites for diagnosis of early-stage CRC and advanced adenoma.

Abstract

e15523 Background: Detection and excision of early-stage CRC or colorectal precancerous lesions (advanced adenoma) has been recognized as an effective way to reduce CRC induced mortality. Integrating signals from both tumoral tissues and microenvironment exhibit better accuracy for tumor detection, especially for early-stage tumors. Gut microbiome carried out profound effects on initiation and progression of colorectal cancer and adenoma, and our previous findings have also revealed that gut microbiome-associated serum metabolites exhibited potential for detecting colorectal neoplasia patients. Here, we carried out an integrated analysis of both early colorectal neoplasia tissues and related gut microbiome to define their combined effect on serum metabolome and develop diagnostic test for early-stage CRC and advanced adenoma. Methods: Integrated analysis of untargeted metabolomics of serum and tissue samples and metagenome sequencing of paired fecal samples were applied to define tissue metabolome and gut microbiome variations between normal and colorectal neoplasia patients. Selected metabolic biomarkers were quantified by targeted LC-MS detection, and a diagnostic model based on the integrated serum metabolite biomarkers was established and evaluated in an independent modeling cohort. Results: Consensus participants undergoing coloscopy were enrolled. The serum-fecal matched discovery cohort includes 120 individuals, ranging from normal, non-adenoma polyp individuals to adenoma and early-stage CRC patients, and among them, resected neoplastic tissue samples were also examined. Co-relation analysis was caried out to build associations of serum metabolome with either tissue metabolome or gut microbiome enriched in early neoplasia patients. More than 100 candidate metabolic biomarkers in serum were selected and transferred to targeted metabolomics detection in an independent modeling cohort including 696 individuals (265 normal individuals, 122 with hyperplastic polyps, 155 with low-risk colorectal adenoma were assigned to the negative group; and positive group includes 136 advanced colorectal adenoma and 18 early-stage CRC patients). After feature selection, an integrated diagnostic model was established to detect early CRC and advanced adenoma, yielded an area under the curve (AUC) of 0.89 in the training set and an AUC of 0.87 (66.7% sensitivity, 90.4% specificity) in the testing set. As for advanced adenoma, the model could also achieve an AUC of 0.84 in the testing set (59.7% sensitivity, 90.4% specificity), significantly higher than previously reported fecal based FIT-DNA test and blood based Septin 9 test. Conclusions: Integrating serum metabolic biomarkers associated with both tumor tissue and colorectal neoplasia related gut microbiota could effectively detect early colorectal cancer and advanced adenoma.