A novel defined pyroptosis-related gene signature for predicting the prognosis of ovarian cancer

Ovarian cancer (OC) has high recurrence and mortality with poor 5-year survival, largely due to late diagnosis and limited effectiveness of current treatments (surgery and platinum-based chemotherapy). There is an urgent need for reliable prognostic models and new therapeutic targets. Pyroptosis is an inflammatory form of programmed cell death executed primarily by gasdermin family proteins, leading to membrane pore formation, cell swelling, rupture, and release of proinflammatory cytokines. Pyroptosis components (inflammasomes, gasdermins, cytokines) are implicated in tumorigenesis, invasion, and metastasis, and pyroptosis can modulate antitumor immunity. Despite emerging evidence, the specific roles and prognostic implications of pyroptosis-related genes in OC remain unclear. This study aimed to characterize pyroptosis-related gene expression differences between normal ovary and OC, define OC subtypes based on these genes, construct and validate a pyroptosis-related prognostic gene signature, and explore associations with the tumor immune microenvironment.

Data sources and cohorts: Gene expression data from TCGA ovarian cancer (tumor, n=379) and GTEx normal ovarian tissues (n=88) were pooled to compare expression of 33 known pyroptosis-related genes. For prognostic modeling, 374 TCGA OC cases with complete survival data were used. An external validation cohort of 380 OC patients from GEO (GSE140802) was employed. Expression data were normalized using the scale function when specified. Differential expression and interaction analyses: Differentially expressed genes (DEGs) among the 33 pyroptosis-related genes were identified between normal and tumor tissues (statistical significance threshold P < 0.01). Protein–protein interaction (PPI) analysis was performed with a minimum required interaction score of 0.9 to identify hub genes. Correlation networks among pyroptosis genes were constructed. Consensus clustering: Using the 31 identified DEGs, consensus clustering of 379 TCGA OC patients was conducted with k ranging from 2 to 4; k=2 was selected to define two clusters. Clinical characteristics (grade, age, survival status) and overall survival were compared between clusters. Prognostic model construction: Univariate Cox regression in TCGA screened survival-associated genes among the pyroptosis-related set (cutoff P < 0.2). LASSO Cox regression (R package glmnet) with cross-validation determined the optimal penalty parameter λ and selected genes to build the prognostic signature. The 7-gene risk score was calculated as a linear combination of scaled expression values weighted by LASSO-derived coefficients: risk score = (−0.187AIM2) + (0.068PLCG1) + (0.097ELANE) + (−0.143PIVK) + (−0.086CASP3) + (−0.033CASP6) + (0.130*GSDMA). Patients were dichotomized at the TCGA median risk score into low- and high-risk groups. Validation and performance assessment: Principal component analysis (PCA) based on the 7 genes assessed separation of risk groups. Kaplan–Meier survival analysis and log-rank tests compared overall survival between risk groups. Time-dependent ROC analysis (R packages survival, survminer, timeROC) evaluated predictive performance at 1-, 2-, and 3-year time points. The TCGA-derived median risk score cutoff was applied to the GEO cohort for external validation with analogous analyses. Independent prognostic analysis: Univariate and multivariable Cox regression models assessed whether the risk score independently predicted survival, adjusting for age and grade (TCGA) or age and FIGO stage (GEO). Functional enrichment and immune analyses: Using TCGA, DEGs between low- and high-risk groups were identified with |log2FC| ≥ 1 and FDR < 0.05 (limma). GO and KEGG enrichment analyses (clusterProfiler) characterized biological processes and pathways. Immune infiltration and pathway activity were evaluated by single-sample GSEA (ssGSEA) implemented via the gsva package for 16 immune cell types and 13 immune-related pathways, comparing enrichment scores between risk groups in both TCGA and GEO.

• Of 33 pyroptosis-related genes examined, 31 were differentially expressed between normal ovary (n=88) and OC tumors (n=379) (all P < 0.01). Downregulated in tumors: PRKCA, GSDMB, SCAF1, PIVK, CASP3, NOD1, PLCG1, NLRP1, GSDME, ELANE, TIRAP, CASP4, GSDMD. Upregulated in tumors: GPX4, NLRP7, NLRP2, CASP2, CASP6, TNF, IL1B, IL18, CASP8, NLRP6, GSDMA, GSDMC, PYCARD, CASP5, AIM2, NOD2, NLRC4, NLRP3. PPI analysis (score ≥0.9) highlighted hub genes including CASP1, PYCARD, NLRC4, NLRP1, CASP5, NLRP3, CASP8, AIM2.
• Consensus clustering of 379 TCGA OC patients using the 31 DEGs produced two clusters (k=2), but clusters showed little difference in clinical features and no significant OS difference (P = 0.841).
• A 7-gene LASSO Cox signature (AIM2, PLCG1, ELANE, PIVK, CASP3, CASP6, GSDMA) stratified 374 TCGA OC patients into high- and low-risk groups with significantly different OS (P < 0.001). PCA showed clear separation. Time-dependent AUCs: 1-year 0.628, 2-year 0.662, 3-year 0.607.
• External validation in GEO (GSE140802; n=380) using the TCGA median cutoff: low-risk group had longer OS (P = 0.014) with AUCs of 0.766 (1-year), 0.655 (2-year), and 0.584 (3-year).
• Risk score independently predicted survival in both cohorts. Univariate HRs: TCGA HR 3.285 (95% CI 1.973–5.467), GEO HR 2.613 (95% CI 1.319–5.175). Multivariable HRs: TCGA HR 3.059 (95% CI 1.836–5.095), GEO HR 2.770 (95% CI 1.374–5.583).
• Between high- and low-risk groups in TCGA, 115 DEGs were identified (66 upregulated, 49 downregulated in high-risk). GO/KEGG enrichment implicated immune response, chemokine-mediated signaling, and inflammatory cell recruitment.
• Immune analyses (ssGSEA) showed the high-risk group had broadly lower infiltration of immune cells (notably CD8+ T cells, NK cells, neutrophils, Th1/Th2 cells, TILs, Tregs) and reduced activity across 12 of 13 immune-related pathways (except type II IFN response). Similar patterns were observed in GEO; additionally, DCs, iDCs, and macrophages were enriched and type II IFN responses were downregulated in the low-risk group compared with the high-risk group.

The study demonstrates that pyroptosis-related genes are extensively dysregulated in OC and are closely linked to tumor immunity. Although clustering by pyroptosis-related DEGs did not yield prognostically distinct subtypes, a refined 7-gene pyroptosis-related signature effectively stratified patients by survival and functioned as an independent prognostic factor. Biological interpretation of the signature suggests complex roles for its components: AIM2 can act context-dependently and was upregulated in tumors yet associated with better survival within the low-risk group; PLCG1 may negatively regulate pyroptosis and was linked to worse outcomes; ELANE can activate GSDMD and neutrophil pyroptosis; PIVK (DFNB59/pejvakin) may act as a tumor suppressor; CASP3 and CASP6, while central to apoptosis, also intersect with pyroptosis pathways (e.g., CASP3-mediated GSDME cleavage), potentially enhancing chemosensitivity; GSDMA likely serves as a pyroptosis executor and was associated with poorer prognosis. Functional and immune analyses indicate that high-risk patients have reduced antitumor immune cell infiltration and diminished immune pathway activity, consistent with impaired antitumor immunity driving poorer outcomes. The unexpected higher Treg proportion in the low-risk group may reflect a role in tempering excessive inflammation induced by pyroptosis or heterogeneity among Treg subtypes in OC. Overall, the findings link pyroptosis-related gene expression to immune microenvironment alterations that impact OC prognosis.

Pyroptosis-related genes are widely and differentially expressed in ovarian cancer compared with normal ovary. A 7-gene pyroptosis-related signature (AIM2, PLCG1, ELANE, PIVK, CASP3, CASP6, GSDMA) reliably stratifies OC patients into prognostic groups and serves as an independent predictor of overall survival, validated across TCGA and GEO cohorts. Differential expression between risk groups is enriched in immune and inflammatory pathways, and high-risk patients exhibit attenuated antitumor immune infiltration and pathway activity. These results provide a novel, immune-related pyroptosis gene signature for OC prognosis and a foundation for future studies exploring mechanistic links between pyroptosis and the tumor immune microenvironment, with potential implications for targeted therapies and immunomodulation.

The study relies on retrospective analysis of public datasets (TCGA, GTEx, GEO) and lacks experimental validation to confirm that identified regulators exert the same roles in OC-specific pyroptosis pathways. Mechanistic interactions among signature genes and their direct effects on pyroptosis and immunity in OC were not experimentally demonstrated. Some observations (e.g., Treg distribution) require deeper phenotypic resolution to reconcile with prior reports. Clinical heterogeneity and potential batch effects, despite normalization, may influence results.