Background: Colorectal cancer (CRC) is a malignancy of the digestive system with high incidence rate and mortality, and reliable diagnostic and prognostic markers for CRC are still lacking. Glutamine metabolism is crucial to the occurrence and development of CRC. However, no research has systematically analyzed the biological role of glutamine metabolism-related genes (GMRGs) in CRC.
Methods: We downloaded gene expression data and clinical data of CRC patients from the TCGA database. The UCSC database downloads pan-cancer gene expression data and prognosis data. Characteristic GMRGs were screened out using differential analysis and two types of machine learning (SVM-REF and RandomForest). Single-cell RNA-sequencing data from CRC patients were downloaded from GEO data. ROC curve was used to evaluate the diagnostic value. Kaplan-Meier method and univariate Cox regression analysis were used to evaluate the prognostic value. The oncopredict package is used to calculate IC50 values for common drugs in CRC patients.
Results: A total of 31 differentially expressed GMRGs were identified, 9 of which were identified as characteristic GMRGs. Further evaluation of diagnostic and prognostic value finally identified GPT as the most important GMRGs in CRC. scRNA-seq analysis revealed that GPT is almost exclusively expressed in epithelial cells. GPT expression is closely related to the tumor microenvironment and can effectively distinguish the sensitivity of different CRC patients to clinical drugs. In addition, pan-cancer analysis showed that GPT is an excellent diagnostic and prognostic marker for multiple cancers.
Conclusions: GPT is a reliable diagnostic, prognostic marker and therapeutic target in CRC.
Keywords: GPT; colorectal cancer; machine learning; scRNA-seq; tumor microenvironment.