Identification and predictive machine learning models construction of gut microbiota associated with lymph node metastasis in colorectal cancer

This study focuses on the significant role between gut microbiota and lymph node metastasis (LNM) in colorectal cancer (CRC). By conducting 16S rRNA sequencing on fecal samples from 147 CRC patients and combining it with the linear discriminant analysis effect size (LEfSe) algorithm, we successfully identified significant differences in the gut microbiota between patients with LNM and those with none lymph node metastasis (NLNM). Furthermore, using transcriptome data from 9 CRC patients, we constructed an immune cell infiltration matrix to deeply explore the biological functions associated with LNM. Eventually, using the characteristics of the gut microbiota associated with LNM, we developed random forest (RF) and multilayer perceptron (MLP) machine learning models to predict the lymph node metastasis status of CRC patients. We identified 21 differentially abundant gut microbes between the two groups, among which Bacteroides plebeius , significantly enriched in LNM group, is closely related to the upregulation of resting NK cells and chemokine CXCL8 expression, and this bacterial species is also positively correlated with the enhancement of fructose metabolism. The RF and MLP models constructed based on the LNM-associated gut microbiota showed good predictive efficacy in predicting LNM status in CRC. This study reveals that Bacteroides plebeius may play an important role in the progression of CRC, with its mechanism potentially involving changes in immune modulation and metabolic pathways. The classification model constructed based on gut microbiota characteristics can predict LNM status of CRC, providing a new perspective for personalized and precision treatment of CRC patients.