Abstract Background: Biochemical recurrence (BCR) after radical prostatectomy for localized or locally advanced prostate adenocarcinoma is clinically heterogeneous, and current risk categories only partially capture the underlying biology. Multi-cohort transcriptomic integration may clarify molecular distinctions between European Association of Urology (EAU) low- versus high-risk BCR. Objective: To identify reproducible gene-expression profile associated with high-risk BCR across independent datasets and to determine whether low-risk BCR tumors differ biologically from non-recurrent disease. Methods: Transcriptomic data were analyzed from TCGA-PRAD (n=479), GSE220095 (BCR-No=101, BCR-EAU-Low=45, BCR-EAU-High=30), GSE70768 (No=93, Low=7, High=12), GSE54460 (No=48, Low=29, High=21), and GSE70769 (No=49, Low=21, High=24). Cohorts were reclassified into and adapted EAU-like BCR groups. RNA-seq data were log-transformed. All matrices were then harmonized, filtered, and aligned to curated metadata. Differential expression was performed with limma on R software using contrasts High-vs-No, High-vs-Low, and Low-vs-No. Pathway enrichment was evaluated using preranked GSEA (GO/Reactome). Results: Across cohorts, high-risk BCR tumors consistently upregulated cell-cycle and DNA-replication programs, including G2/M checkpoint activation, sister chromatid segregation, replication-stress signaling, and mitotic spindle assembly. TCGA-PRAD high-risk tumors showed strong activation of ATR-dependent replication-stress pathways. GSE220095 high-risk cases exhibited enrichment of mitotic checkpoint and pre-replicative complex assembly mechanisms. In GSE70768, key differentially expressed genes included PAGE4, PGM5, SERPINA3, and TOP2A, indicating loss of androgen-regulated stress-response programs and increased proliferative drive. GSE70769 and GSE54460 similarly converged on cell-cycle deregulation, chromatin remodeling, and ECM/EMT-related signaling. The absence of DEGs between low- and high-risk BCR groups—despite marked differences between high-risk and No-BCR—suggests that low-risk tumors form a heterogeneous intermediate state, sharing baseline expression with No-BCR cases while lacking the proliferative and replication-stress programs. Computational workflow and writing support was provided by an AI-based language (ChatGPT, OpenAI). Conclusions: High-risk BCR tumors show a strong, reproducible transcriptional signature driven by replication stress and proliferative pathways, whereas low-risk BCR tumors are molecularly similar to non-recurrent diseases. These results provide biological validation of the EAU risk stratification system and highlight the potential of transcriptomic profiling to refine surveillance and treatment decisions after prostatectomy. Citation Format: Sabrina Yepes-Rodriguez, Rafael Parra-Medina. Multi-cohort transcriptomic profiling identifies pathways driving high-risk biochemical recurrence in prostate cancer abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 7911.
Yepes-Rodríguez et al. (Fri,) studied this question.