AUTHOR=Xie Lin-Ying , Huang Han-Ying , Hao Yu-Lei , Yu Miaomiao , Zhang Wenju , Wei Enwei , Gao Chunfeng , Wang Chang , Zeng Lei
TITLE=Development and validation of a tumor immune cell infiltration-related gene signature for recurrence prediction by weighted gene co-expression network analysis in prostate cancer
JOURNAL=Frontiers in Genetics
VOLUME=14
YEAR=2023
URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2023.1067172
DOI=10.3389/fgene.2023.1067172
ISSN=1664-8021
ABSTRACT=Introduction: Prostate cancer (PCa) is the second most common malignancy in men. Despite multidisciplinary treatments, patients with PCa continue to experience poor prognoses and high rates of tumor recurrence. Recent studies have shown that tumor-infiltrating immune cells (TIICs) are associated with PCa tumorigenesis.
Methods: The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) datasets were used to derive multi-omics data for prostate adenocarcinoma (PRAD) samples. The CIBERSORT algorithm was used to calculate the landscape of TIICs. Weighted gene co-expression network analysis (WGCNA) was performed to determine the candidate module most significantly associated with TIICs. LASSO Cox regression was applied to screen a minimal set of genes and construct a TIIC-related prognostic gene signature for PCa. Then, 78 PCa samples with CIBERSORT output p-values of less than 0.05 were selected for analysis. WGCNA identified 13 modules, and the MEblue module with the most significant enrichment result was selected. A total of 1143 candidate genes were cross-examined between the MEblue module and active dendritic cell-related genes.
Results: According to LASSO Cox regression analysis, a risk model was constructed with six genes (STX4, UBE2S, EMC6, EMD, NUCB1 and GCAT), which exhibited strong correlations with clinicopathological variables, tumor microenvironment context, antitumor therapies, and tumor mutation burden (TMB) in TCGA-PRAD. Further validation showed that the UBE2S had the highest expression level among the six genes in five different PCa cell lines.
Discussion: In conclusion, our risk-score model contributes to better predicting PCa patient prognosis and understanding the underlying mechanisms of immune responses and antitumor therapies in PCa.