Utilizing Single-Cell and Transcriptomic Data to Identify Mitochondrial Pathway-Associated Prognostic Genes and Their...

Mitochondrial abnormalities correlate closely with multiple cancers, but the role of genes associated with mitochondrial pathways in ovarian cancer (OC) remains unclear. This study aimed to identify OC prognosis-related mitochondrial pathway-associated genes at single-cell and transcriptome levels. Public datasets (GSE184880, GSE54388, GSE18520, TCGA-OV) were retrieved. Using GSE184880, potential cell subpopulations and their tumor-control differentially expressed genes (DEGs) were identified, then intersected with GSE54388 key module genes to yield candidate genes. Univariate Cox regression and LASSO analyses were performed to screen prognostic genes, based on which a prognostic risk model was constructed and validated. Functional and localization analyses of prognostic genes, regulatory network construction, immune infiltration analysis, drug prediction, and expression verification were conducted. Pseudotime, cell communication and expression analyses were implemented at single-cell level. To account for patient-level variation, mixed-effects models using the MAST method were applied. Targeted GSEA focusing on mitochondrial functional pathways was performed. Clinical correlation, independent prognostic analysis, and computational functional inference were conducted. Eight cell types were observed by cell clustering and annotation, among which endothelial cells, fibroblasts, NK cells, and tissue stem cells were observed as potential cell subpopulations. TK1, VWF, CPXM1, WIPF3, APOLD1, MGST2, and PPA1 were identified as prognostic genes, and risk models constructed based on them had good performance and universality. Mixed-effects model analysis confirmed that 1200 DEGs remained significant after adjusting for patient-level variation, supporting the robustness of single-cell findings. These prognostic genes were found to be robust predictors of patient survival with inconsistent expression patterns in vitro, and may play important roles in the differentiation and development of key cells. Targeted GSEA revealed that mitochondrial respiration (electron transport chain/oxidative phosphorylation) and mitophagy were significantly suppressed in the high-risk group, while mitochondrial fusion/fission and ROS pathways did not reach significance. PPA1 was highly expressed in various cell types in the GSE184880 dataset. In the GSE54388 dataset, CPXM1 and WIPF3 were substantially downregulated in the tumor group. However, other prognostic genes displayed the opposite expression pattern. Single-cell functional inference showed that TK1 and APOLD1 were highly expressed in G2/M phase, while MGST2, PPA1, VWF, CPXM1, and WIPF3 were enriched in G1 phase. VWF showed the strongest positive correlation with the PI3K-AKT pathway. Clinical validation confirmed the independent prognostic value of the signature in advanced-stage OC (FIGO III/IV). Seven mitochondrial pathway-associated prognostic genes were identified in OC, with inconsistent survival and in vitro expression patterns, providing novel references for exploring potential therapeutic targets.