Integrated Analysis of Colorectal Cancer Reveals Cross-Cohort Gut Microbial Signatures and Associated Serum Metabolites

Renyuan Gao,Chunyan Wu,Yefei Zhu,Cheng Kong,Yin Zhu,Yaohui Gao,Xiaohui Zhang,Rong Yang,Hui Zhong,Xiao Xiong,Chunqiu Chen,Qian Xu,Huanlong Qin
DOI: https://doi.org/10.1053/j.gastro.2022.06.069
IF: 29.4
Gastroenterology
Abstract:Background & aims: Studies have reported abnormal gut microbiota or circulating metabolome associated with colorectal cancer (CRC), but it remains a challenge to capture the CRC-relevant features consistent across geographic regions. This is particularly the problem for metabolic traits of CRC because the analyses generally use different platforms and laboratory methods, which poses a barrier to cross-dataset examination. In light of this, we sought to elucidate the microbial and metabolic signatures of CRC with broad population relevance. Methods: In this integrated metagenomic (healthy controls [HC], n = 91; colorectal adenoma [CRA], n = 63; CRC, n = 71) and metabolomic (HC, n = 34; CRA, n = 31; CRC, n = 35) analysis, CRC-associated features and microbe-metabolite correlations were first identified from a Shanghai cohort. A gut microbial panel was trained in the in-house cohort and cross-validated in 7 published metagenomic datasets of CRC. The in-house metabolic connections to the cross-cohort microbial signatures were used as evidence to infer serum metabolites with potentially external relevance. In addition, a combined microbe-metabolite panel was produced for diagnosing CRC or adenoma. Results: CRC-associated alterations were identified in the gut microbiome and serum metabolome. A composite microbe-metabolite diagnostic panel was developed and yielded an area under the curve of 0.912 for adenoma and 0.994 for CRC. We showed that many CRC-associated metabolites were linked to cross-cohort gut microbiome signatures of the disease, including CRC-enriched leucylalanine, serotonin, and imidazole propionate; and CRC-depleted perfluorooctane sulfonate, 2-linoleoylglycerol (18:2), and sphingadienine. Conclusions: We generated cross-cohort metagenomic signatures of CRC, some of which linked to in-house CRC-associated serum metabolites. The microbial and metabolic shifts may have wide population relevance.
What problem does this paper attempt to address?