Integrative Structural and Network Analysis Identifies CDK9 as a Candidate Therapeutic Target in Short-Term Survivor High-Grade Serous Ovarian Cancer

Abstract

Background: High-grade serous ovarian cancer (HGSC) exhibits marked clinical heterogeneity, with Short-Term (ST) survivors demonstrating intrinsic resistance to standard chemotherapy and poor outcomes. Understanding tumor-intrinsic transcriptional vulnerabilities in ST-HGSC represents a critical unmet need for precision therapeutic development. This study integrates structural bioinformatics, network topology analysis, and pharmacogenomics to identify druggable master regulators sustaining the aggressive ST phenotype. Methods: We analyzed transcriptomic data from Kotnik et al. (2023) comparing ST and Long-Term (LT) HGSC survivors, identifying 325 significantly upregulated differentially expressed genes (DEGs) in the ST cohort (log₂FC >1.5, adjusted P<0.05). The ESR1-CCDC170 fusion protein structure was predicted using AlphaFold2 v2.3.1 to assess ligand-binding domain integrity. A protein-protein interaction (PPI) network was constructed from ST-specific DEGs using STRING (confidence >0.4), and master regulatory hubs were identified using the Maximal Clique Centrality (MCC) algorithm in CytoHubba. Survival analysis was performed using Kaplan-Meier Plotter (n=1,435 serous ovarian cancer patients). Functional enrichment of hub neighborhoods was conducted using NetworkSuite v2.1 (Sequensolutions), integrating Gene Ontology, Reactome, WikiPathways, and KEGG databases. Pharmacogenomic repurposing was performed using L1000CDS2, and drug-target interactions were validated through structural analysis. Results: AlphaFold2 modeling demonstrated that ESR1-CCDC170 retains the DNA-binding domain (pLDDT >85) but completely lacks the C-terminal ligand-binding domain, providing structural evidence for endocrine therapy resistance. Network analysis identified Cyclin-Dependent Kinase 9 (CDK9) as the top-ranked hub (degree=17, highest MCC score), exhibiting both high connectivity and betweenness centrality. Survival analysis confirmed that high CDK9 expression significantly predicted poor overall survival (median OS: 42.5 vs 51.2 months; HR=1.28, 95% CI: 1.10-1.51; log-rank P=0.0088). Functional enrichment of CDK9's 17 first-degree interactors revealed profound overrepresentation in RNA Polymerase II transcriptional elongation pathways (FDR<10⁻¹⁰), supporting a model of elongation-dependent transcriptional addiction. Network stability analysis across multiple STRING confidence thresholds (0.4-0.9) confirmed CDK9's robust central position. Pharmacogenomic analysis identified transcriptional inhibitors as signature-reversing compounds, and structural validation confirmed stable CDK9-alvocidib interaction. Conclusion: Our integrative computational analysis identifies CDK9 as a candidate master regulator sustaining transcriptional addiction in ST-HGSC, potentially driven by fusion-mediated constitutive transcriptional activation. These findings support CDK9 inhibition as a rational precision therapeutic strategy for this high-risk patient subgroup. Experimental validation through functional studies, including CDK9 knockdown experiments, drug sensitivity assays in HGSC cell lines, and patient-derived xenograft models, is essential to confirm these computational predictions and advance toward clinical translation.