Functional variants in ATF4 and PMAIP1 reveal mitochondrial apoptosis dependencies driving pancreatic cancer progression.

Pancreatic cancer is one of the most lethal malignancies worldwide, characterized by late diagnosis, aggressive progression, and poor survival. Dysregulation of regulated cell death (RCD) pathways, including apoptosis, ferroptosis, necroptosis, and mitochondrial dysfunction, contributes to tumor survival, therapy resistance, and immune evasion. Understanding the molecular mechanisms underlying these processes is critical for identifying prognostic biomarkers and therapeutic targets. Publicly available pancreatic cancer gene expression datasets (GSE227567 and GSE275246) were analyzed to identify differentially expressed RCD and mitochondrial genes. FRGs, APGs, NRGs, and MGs were intersected with the DEGs, followed by functional enrichment, protein-protein interaction (PPI) network construction, and hub gene prioritization. Co-expression, immune infiltration, and pathway activity analyses were performed across pathological stages and pan-cancer datasets. Missense SNPs in top hub genes ATF4 and PMAIP1 were evaluated for structural and energetic impact using HOPE and DynaMut. Analysis identified 448 FRGs, 73 APGs, 32 NRGs, and 930 MGs in GSE227567, and 123 APGs, 0 FRGs, 36 NRGs, and 577 MGs in GSE275246. Cross-dataset overlap was highest for APGs (116 genes), while other RCD and mitochondrial genes were largely dataset-specific. Filtering and PPI network analysis prioritized nine hub genes (ATF4, PMAIP1, BCL2L1, ATF3, TRADD, SRC, SFN, KCNMA1, GUK1). Functional enrichment highlighted mitochondrial metabolism, oxidative phosphorylation, and apoptosis pathways, with integration into PI3K-AKT, MAPK, p53, and VEGF signaling. Immune infiltration analysis revealed myeloid-enriched tumor microenvironments in mutant contexts. Pan-cancer and stage-wise expression profiling indicated that ATF4 and PMAIP1 exhibit consistent stage-dependent modulation. Survival analysis showed that high PMAIP1 expression correlates with poor prognosis (HR = 1.26, p = 0.00185), while ATF4 shows a protective trend (HR = 0.84, p = 0.0145). SNP analysis revealed six missense variants in each gene, with ATF4 variants predominantly destabilizing and affecting the bZIP domain, whereas PMAIP1 variants were mostly surface-exposed and modulatory. Our integrative analysis identifies ATF4 and PMAIP1 as key RCD- and mitochondrial-associated genes in pancreatic cancer, with functional SNPs that influence prognosis and may serve as therapeutic targets. These findings provide insights into mitochondrial-driven apoptosis, tumor progression, and potential avenues for precision medicine interventions.