In Clinical & translational oncology : official publication of the Federation of Spanish Oncology Societies and of the National Cancer Institute of Mexico
BACKGROUND : Lymph node metastasis is the main metastatic mode of CRC. Lymph node metastasis affects patient prognosis.
OBJECTIVE : To screen differential intestinal bacteria for CRC lymph node metastasis and construct a prediction model.
METHODS : First, fecal samples of 119 CRC patients with lymph node metastasis and 110 CRC patients without lymph node metastasis were included for the detection of intestinal bacterial 16S rRNA. Then, bioinformatics analysis of the sequencing data was performed. Community structure and composition analysis, difference analysis, and intragroup and intergroup correlation analysis were conducted between the two groups. Finally, six machine learning models were used to construct a prediction model for CRC lymph node metastasis.
RESULTS : The community richness and the community diversity at the genus level of the two groups were basically consistent. A total of 12 differential bacteria (Agathobacter, Catenibacterium, norank_f__Oscillospiraceae, Lachnospiraceae_FCS020_group, Lachnospiraceae_UCG-004, etc.) were screened at the genus level. Differential bacteria, such as Agathobacter, Catenibacterium, norank_f__Oscillospiraceae, and Lachnospiraceae_FCS020_group, were more associated with no lymph node metastasis in CRC. In the discovery set, the RF model had the highest prediction accuracy (AUC = 1.00, 98.89% correct, specificity = 55.21%, sensitivity = 55.95%). In the test set, SVM model had the highest prediction accuracy (AUC = 0.73, 72.92% correct, specificity = 69.23%, sensitivity = 88.89%). Lachnospiraceae_FCS020_group was the most important variable in the RF model. Lachnospiraceae_UCG - 004 was the most important variable in the SVM model.
CONCLUSION : CRC lymph node metastasis is closely related to intestinal bacteria. The prediction model based on intestinal bacteria can provide a new evaluation method for CRC lymph node metastasis.
Yinhang Wu, Jing Zhuang, Jie Zhou, Yin Jin, Xinyue Wu, Yifei Song, Zhiqing Fan, Wei Wu, Shuwen Han
2023-Jan-12
Colorectal cancer (CRC), Intestinal bacteria, Lymph node metastasis, Machine learning algorithm, Prediction model
