Integrated analysis identifies ELF3-CREB1 co-activation as a prognostic driver of high-risk pancreatic cancer

Background: Patients diagnosed with pancreatic ductal adenocarcinoma (PDAC) face a dismal prognosis, with only about 13% surviving five years. Transcriptional factors play a vital role in PDAC survival and prognosis by defining tumor aggressiveness, therapeutic resistance, and clinical outcome. Key regulators such as CREB1 have previously been linked with tumor progression and shorter survival. While CREB1 signaling driving PDAC is known, the systematic characterization of active Transcription Factor (TF) associated with CREB1, and their clinical relevance is limited. In this study, we aimed to identify critical TF activity associated with CREB1 and assess their combinatorial prognostic impact in PDAC. Methods: We utilized the VIPER (Virtual Inference of Protein-activity by Enriched Regulon) algorithm with high-confidence DoRothEA regulons (A-C) to infer TF activity across 182 TCGA-PAAD RNA-seq samples (178 Primary Tumors [TP] and 4 Normal Tissues [NT]). Differential Activity Analysis (DAA) using limma identified TFs dysregulated between TP and NT. Survival analyses utilized Kaplan-Meier and Cox Proportional Hazards modeling on the TP cohorts, focusing on the interaction between two key dysregulated TFs, CREB1 and ELF3 (ETS transcription factor 3). R version 4.3.0 was used for all analyses, including GSVA (v1.50.5) and VIPER (v1.36.0). Differential expression and pathway analyses were performed using the R packages limma (v3.58.1), GSVA (v1.50.5), msigdbr, gplots, and ggplot2. Results: DAA identified 44 significantly dysregulated TFs (FDR < 0.05). The epithelial lineage regulator ELF3 showed the highest activation in tumors, while the lymphoid regulator PAX5 (paired box 5) was highly repressed. Although single-TF prognostic tests were non-significant, combinatorial analysis revealed a strong context-dependent effect. The simultaneous High CREB1 and High ELF3 activity defined a uniquely aggressive subgroup with a markedly shorter median overall survival (272 days) compared to the lowest-risk combination (492 days). Cox modeling confirmed a significant synergistic interaction between the two TFs (HR =2.31, p = 0.049). Conclusion: This analysis reveals that the highly activated epithelial driver ELF3 acts synergistically with CREB1 to define a high-risk prognostic signature in PDAC. Our findings underscore that TF networks, rather than single factors, are crucial for patient stratification and represent compelling, context-specific therapeutic targets in PDAC.