Identification of microbial markers across populations in early detection of colorectal cancer

Abstract

Several studies have investigated the association between microbial and colorectal cancer (CRC). However, the replicable markers for early stage adenoma diagnosis across multiple populations remain elusive. Here, a meta-analysis of six studies, comprising a total of 1057 fecal samples, was performed to identify candidate markers. By adjusting the potential confounders, 11 and 26 markers (P<0.05) were identified and separately applied into constructing Random Forest classifier models to discriminate adenoma from control, and adenoma from CRC, achieving robust diagnostic accuracy with AUC = 0.80 and 0.89, respectively. Moreover, these markers demonstrated high diagnostic accuracy in independent validation cohorts. Pooled functional analysis and targeted qRT-PCR based genetic profiles reveal that the altered microbiome triggers different pathways of ADP-heptose and menaquinone biosynthesis (P<0.05) in adenoma vs. control and adenoma vs. CRC sequences respectively. The combined analysis of heterogeneous studies confirm adenoma-specific but universal markers across multi-populations, which improves early diagnosis and prompt treatment of CRC.