Aging characteristics of colorectal cancer based on gut microbiota

Abstract

Abstract Background Aging is one of the factors leading to cancer. Gut microbiota is related to aging and colorectal cancer (CRC). Methods A total of 11 metagenomic data sets related to CRC were collected from the R package curated Metagenomic Data. After batch effect correction, healthy individuals and CRC samples were divided into three age groups. Ggplot2 and Microbiota Process packages were used for visual description of species composition and PCA in healthy individuals and CRC samples. LEfSe analysis was performed for species relative abundance data in healthy/CRC groups according to age. Spearman correlation coefficient of age‐differentiated bacteria in healthy individuals and CRC samples was calculated separately. Finally, the age prediction model and CRC risk prediction model were constructed based on the age‐differentiated bacteria. Results The structure and composition of the gut microbiota were significantly different among the three groups. For example, the abundance of Bacteroides vulgatus in the old group was lower than that in the other two groups, the abundance of Bacteroides fragilis increased with aging. In addition, seven species of bacteria whose abundance increases with aging were screened out. Furthermore, the abundance of pathogenic bacteria (Escherichia_coli, Butyricimonas_virosa, Ruminococcus_bicirculans, Bacteroides_fragilis and Streptococcus_vestibularis) increased with aging in CRCs. The abundance of probiotics (Eubacterium_eligens) decreased with aging in CRCs. The age prediction model for healthy individuals based on the 80 age‐related differential bacteria and model of CRC patients based on the 58 age‐related differential bacteria performed well, with AUC of 0.79 and 0.71, respectively. The AUC of CRC risk prediction model based on 45 disease differential bacteria was 0.83. After removing the intersection between the disease‐differentiated bacteria and the age‐differentiated bacteria from the healthy samples, the AUC of CRC risk prediction model based on remaining 31 bacteria was 0.8. CRC risk prediction models for each of the three age groups showed no significant difference in accuracy (young: AUC=0.82, middle: AUC=0.83, old: AUC=0.85). Conclusion Age as a factor affecting microbial composition should be considered in the application of gut microbiota to predict the risk of CRC.

Keywords

• aging
• artificial intelligence
• colorectal cancer
• gut microbiota
• metagenomic sequencing