Tumor budding is a long-established independent adverse prognostic marker for colorectal cancer (CRC), yet assessment of tumor budding was not reproducible. Therefore, development of precise diagnostic approaches to tumor budding is in demand. In this study, we first performed bioinformatic analysis in our single-center CRC patients’ cohort (n = 84) and identified tumor budding-associated hub genes using the weighted gene co-expression network analysis (WGCNA). A machine learning methodology was used to identify hub genes and construct a prognostic signature. Nomogram model was used to identified hub genes score for tumor budding, and the receiver operating characteristic (ROC) curve and calibration plot indicated high accuracy and stability of hub gene score for predicted the prognosis of CRC. The association between budding-associated hub genes and score and prognosis of CRC were further verified in TCGA CRC cohort (n = 342). Then gene set enrichment analysis (GSEA) and gene set variation analysis (GSVA) were applied to explore the signaling pathways related to the tumor budding and validated by immunohistochemistry (IHC) of our clinical samples. Subsequently, immune infiltration analysis demonstrated that there was a high correlation between hub genes score and M2-like macrophages infiltrated in tumor tissue. In addition, somatic mutation and chemotherapeutic response prediction were analyzed based on the risk signature. In summary, we established a tumor budding diagnostic molecular model, which can improve tumor budding assessment and provides a promising novel molecular marker for immunotherapy and prognosis of CRC.
Read full abstract