Background/Objectives: Testicular germ cell tumors (TGCTs) are uniquely curable with cisplatin-based therapies even when widely metastatic; however, cisplatin resistance does occur, resulting in very poor prognosis. The mechanisms to explain TGCT hypersensitivity to cisplatin and mechanisms of resistance are not well-understood. Methods: The global transcriptional response to acute cisplatin treatment (24 h after a 6 h pulse of cisplatin) was assessed in three parental embryonal carcinoma TGCT cells lines compared to multiple isogenic, stable, cisplatin-resistant clonal lines from these parental cells. Results: Cisplatin treatment of parental cells consistently showed a more robust overall transcriptional response to cisplatin compared to their cisplatin-resistant cellular counterparts for a common set of genes and pathways including the upregulation of genes associated with histone modifications and p53, EMT, and KRAS signaling and the downregulation of genes normally upregulated by MYC. Focusing on genes exclusively altered in parental cells revealed upregulated genes known to be induced by p53 and downregulated by MYC and the transferrin receptor, TFRC1. Several of these p53/MYC/TFRC1 targets were associated with a higher instance of disease-free survival in a cohort of TGCT patients. Conclusions: Cisplatin resistance in TGCT cells is associated with a diminished alteration in cisplatin-responsive genes, especially genes known to be regulated by p53, MYC and TFRC1, that may be linked to cisplatin hypersensitivity and survival in TGCTs.

Background/Objectives: Cancer is a chronic and heterogeneous disease, possessing molecular variation within a single type, resulting in its molecular subtypes. Cancer molecular subtyping offers biological insights into cancer variability, facilitating the development of personalized medicines. Various models have been proposed for cancer molecular subtyping, utilizing the high-dimensional transcriptomic, genomic, or proteomic data. The issue of data scarcity, characterized by high feature dimensionality and a limited sample size, remains a persistent problem.The objective of this research is to propose a deep learning framework, DeepCMS, that leverages the capabilities of feed-forward neural networks, gene set enrichment analysis, and feature selection to construct a well-representative subset of the feature space, thereby producing promising results. Methods: The gene expression data were transformed into enrichment scores, resulting in over 22,000 features. From those, the top 2000 features were selected, and deep learning was applied to these features. The encouraging outcomes indicate the efficacy of the proposed framework in terms of defining a well-representative feature space and accurately classifying cancer molecular subtypes. Results: DeepCMS consistently outperformed state-of-the-art models in aggregated accuracy, sensitivity, specificity, and balanced accuracy. The aggregated metrics surpassed 0.90 for all efficiency measures on independent test datasets, showing the generalizability and robustness of our framework. Although developed using colon cancer’s gene expression data, this approach may be applied to any gene expression data; a case study is also devised for illustration. Conclusions: Overall, the proposed DeepCMS framework enables the accurate and robust classification of cancer molecular subtypes using a compact and informative feature set, facilitating improved precision in oncology applications.

Targeted therapies have become pivotal in modern clinical oncology, driven by a molecularlevel understanding of cancer’s intricacies, its progression, and innovative research and technology. Personalized and targeted treatments hinge on identifying key genes, hub genes, or biomarkers. Protein-protein interaction (PPI) networks are instrumental in understanding the molecular basis of diseases. While existing literature has identified significant genes based on network centrality, investigations based on community-aware centrality have been notably absent. This omission matters because disease networks frequently display modular structures, necessitating a new perspective. This study bridges the gap between network centrality and community-based investigations. Indeed, in modular networks, node influence can be categorized into two types: local impact within its community, determined by intra-community connections, and global effect on nodes in other communities, established through inter-community links. This concept extends conventional centrality measures to networks with a community structure. Initially, we performed a comparative analysis of seven PPI networks related to cancer and noncancerous conditions. We explore the correlation between classical network centralities and their equivalents at the global (inter-community) and local (intra-community) levels. Notably, we consistently observed a high correlation between network degree and local degree centrality in all PPIs, leading us to select local degree centrality for further investigation. Pronounced modularity characterizes prostate and cervical disease networks. Consequently, we investigate these networks to identify key genes at the local community level and validate them by examining their expression levels. Variations in gene expression between cancerous and non-cancerous tissues bolster our findings. We identify a novel set of genes as potential key players in prostate and cervical cancer. Specifically, in cervical cancer, the top genes at the mesoscopic level include AKT1, CDK2, BRCA1, VEGFA, SRC, PSMD14, MRPL3, TP53, and NUP37. Meanwhile, the top genes identified in prostate cancer are FOS, TP53, UBA52, HLA-B, TSPO, and CD19. Although we focus on cancer data, our methodology’s versatility makes it applicable to other disease networks, opening avenues to identify key genes as potential drug targets.