In The journal of gene medicine
BACKGROUND : A growing number of studies have shown that inflammation-related components of the tumor microenvironment (TME) affect the clinical outcomes of cancer patients, and advances in radiomics may help predict survival and prognosis.
METHODS : We performed a systematic analysis of inflammation-related genes (IRGs) in clear cell renal cell carcinoma (ccRCC) from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO). and map their interaction network to assess the specific relationship between these DEIRGs and inflammation. The association between DEIRGs and prognosis was discussed and further validated using consensus cluster analysis. Next, we constructed IRGs-related risk score from the collected information and validated the prognostic value of the model using Kaplan-Meier survival analysis and receiver operating characteristic (ROC) analysis. CT images corresponding to the TCGA-ccRCC cohort were obtained from the Cancer Imaging Archive (TCIA) database for radiomics signature extraction.
RESULTS : We screened for prognostic IRGs and found that they were positively correlated with inflammatory cells in the tumor microenvironment associated with tumor progression and metastasis, such as activated CD8+ cells, myeloid-derived suppressor cells, and neutrophils. The impact of IRGs on the prognosis of ccRCC patients was also verified. Furthermore, using these DEGs, we successfully constructed a risk signature and validated its good prognosis assessment for patients. Furthermore, radiomics-based prognostic models performed better than those using risk signatures or clinical characteristics.
CONCLUSIONS : IRG-related risk scores play an important role in assessing the prognosis and improving the management of patients with ccRCC. Through this feature, the infiltration of immune cells in the TME can be predicted. Furthermore, noninvasive radiomics signatures showed satisfactory performance in predicting ccRCC prognosis.
Yang Yu, Huang Hang, Liang Haote
2023-Mar-03
Enhanced CT Image, machine learning, renal clear cell carcinoma, transcriptome mapping, tumor environment
