Immunogenic cell death (ICD) plays a vital role in tumor progression and immune response. However, the integrative role of ICD-related genes and subtypes in the tumor microenvironment (TME) in prostate cancer (PCa) remains unknown.
The sample data were obtained from The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), and Memorial Sloan Kettering Cancer Center (MSKCC) prostate cancer-related databases. We first divided the subtypes based on ICD genes from 901 PCa patients and then identified the prognosis- related genes (PRGs) between different ICD subtypes. Subsequently, all the patients were randomly split into the training and test groups. We developed a risk signature in the training set by least absolute shrinkage and selection operator (LASSO)–Cox regression. Following this, we verified this prognostic signature in both the training test and external test sets. The relationships between the different subgroups and clinical pathological characteristics, immune infiltration characteristics, and mutation status of the TME were examined. Finally, the artificial neural network (ANN) and fundamental experiment study were constructed to verify the accuracy of the prognostic signature.
We identified two ICD clusters with immunological features and three gene clusters composed of PRGs. Additionally, we demonstrated that the risk signature can be used to evaluate tumor immune cell infiltration, prognostic status, and an immune checkpoint inhibitor. The low-risk group, which has a high overlap with group C of the gene cluster, is characterized by high ICD levels, immunocompetence, and favorable survival probability. Furthermore, the tumor progression genes selected by the ANN also exhibit potential associations with risk signature genes.
This study identified individuals with high ICD levels in prostate cancer who may have more abundant immune infiltration and revealed the potential effects of risk signature on the TME, immune checkpoint inhibitor, and prognosis of PCa.