Identification of Enzalutamide-Related Genes for Prognosis and Immunotherapy in Prostate Adenocarcinoma

Abstract

Enzalutamide is classified as a novel antiandrogen medication; however, the majority of patients ultimately develop resistance to it. Consequently, conducting an in-depth investigation into potential targets of enzalutamide is essential for addressing the drug resistance observed in patients and for facilitating the discovery of new therapeutic targets. The SwissTargetPrediction database was used to identify targets linked to enzalutamide and to assess these targets in the prostate adenocarcinoma (PRAD) dataset sourced from the TCGA database. By employing various datasets and applying different machine learning methods for clustering, researchers constructed and validated both diagnostic and prognostic models for PRAD. A correlation analysis with the androgen receptor revealed TDP1 as the gene most significantly associated with enzalutamide. In addition, this study examined the relationship between TDP1 and immune infiltration. The expression levels of TDP1 and its prognostic correlation in PRAD patients were validated through immunofluorescence staining of 60 PRAD tissue specimens. Cluster analysis revealed a notable correlation among the 24 genes related to enzalutamide with regard to both prognosis and immune infiltration in PRAD patients. The diagnostic model, which incorporates various machine learning techniques, exhibits robust predictive ability for PRAD diagnosis, while the prognostic model employing the LASSO algorithm has also shown encouraging outcomes. Among the various prognostic genes linked to enzalutamide, TDP1 stands out as an important indicator of prognosis. Furthermore, immunofluorescence experiments confirmed that an increased expression of TDP1 is associated with a worse prognosis in patients with PRAD. Our results underscore the substantial potential of TDP1 as a novel diagnostic and prognostic biomarker for individuals diagnosed with PRAD.

Keywords: biomarker; enzalutamide; gene function; machine learning; prognosis.

Figure 1

Discovery of enzalutamide-associated molecular targets in the TCGA-PRAD cohort. (a) Computational prediction of putative enzalutamide targets via SwissTargetPrediction. (b) Heatmap depicting expression patterns of enzalutamide-associated transcript profiles in tumor vs. adjacent tissues. (c) Forest plot illustrating PFI associations for enzalutamide-linked genes. (d) Correlation network analysis of enzalutamide-associated gene interactions. (e) Molecular docking validation of enzalutamide's ligand–receptor binding with candidate targets.

Figure 2

Enzalutamide-related genes play an important role in PRAD. (a) Expression analysis of enzalutamide-related genes in different T-stages. (b) Expression analysis of enzalutamide-related genes in different N-stages. (c) Expression analysis of enzalutamide-related genes in different PSA scores. (d) Expression analysis of enzalutamide-related genes in different Gleason scores. (e, f) Functional analysis of enzalutamide-related prognostic genes.

Figure 3

Analysis of clusters derived from genes related to enzalutamide. (a) Evaluation of both performance and stability regarding the clusters utilizing various techniques. (b) Consensus representation of NMF clustering. (c, d) Variations in survival across different clusters. (e, f) Variations in the expression levels of enzalutamide-associated genes among the distinct clusters.

Figure 4

Genes associated with enzalutamide demonstrate a significant correlation with immune infiltration in PRAD. (a, b) An examination of enzalutamide-associated genes in relation to PRAD immune infiltration utilizing the ssGSEA algorithm. (c) A heatmap displaying varying levels of immune cell infiltration. (d–g) Variations in the distribution of distinct subgroups across different pathological stages of PRAD. (h, i) Gene enrichment assessment of two identified clusters.

Figure 5

The combination of the Ridge algorithm is regarded as the optimal choice for developing diagnostic models. (a) An analysis of the predictive capabilities of enzalutamide-related genes in the diagnosis of PRAD patients. (b) AUC scores for diagnostic models generated through different algorithm combinations. (c) The number of genes incorporated into diagnostic models produced with various algorithm combinations.

Figure 6

Development of a prognostic model. (a) A forest plot depicting the prognosis of PFI concerning genes linked to enzalutamide. (b, c) Development of prognostic models employing the LASSO algorithm. (d) A heatmap representing the expression levels of genes included in the prognostic model sourced from the GSE116918 dataset. (e) Differences in survival rates identified between high-risk and low-risk patient cohorts within the GSE116918 dataset. (f) The importance of the risk score from the GSE116918 dataset in predicting prognosis for individuals diagnosed with PRAD. (g) A heatmap that reveals the expression levels of genes constituting the prognostic model from the TCGA-PRAD dataset. (h) Prognostic variations recorded between high-risk and low-risk groups in the TCGA-PRAD dataset. (i) The relevance of the risk score from the TCGA-PRAD dataset in evaluating prognosis for patients with PRAD.

Figure 7

The prognostic risk score shows a significant correlation with immune infiltration in PRAD. (a, b) An analysis of the prognostic risk score concerning immune infiltration in PRAD was conducted using the ssGSEA methodology. (c) A heatmap displays diverse levels of immune cell infiltration. (d–g) Variations in the distribution of various subgroups across distinct pathological stages of PRAD are presented. (h, i) Gene enrichment analysis was performed on two different groups.

Figure 8

TDP1 is the most essential gene among enzalutamide-related genes. (a–c) The correlation between enzalutamide-related genes and AR was analyzed. (d) An examination of the relationship between TDP1 and immune infiltration in PRAD. (e–f) Evaluative analysis of TDP1's functions.

Figure 9

TDP1 is highly expressed in PRAD. (a, b) Variations in TDP1 expression within PRAD. (c) ROC curve analysis indicating the diagnostic predictive value of TDP1. (d) Survival analysis KM curve reflecting invasion-free survival associated with TDP1 in PRAD.