Identification and Computational validation of ferroptosis suppressor genes as therapeutic targets in pancreatic cancer Authors: Nitin Kumar Singh, Maninder Singh, Adil Husain, Lina I Alnajjar, Firoz Ahmad, Siddhartha Kumar Mishra, Nawaf Alshamamri, Mohd Saeed, Safia Obaidur Rab, Varsha Gupta Published: 2026 ABSTRACT AbstractPancreatic cancer remains one of the most lethal malignancies, characterised by aggressive progression, metabolic adaptation, and resistance to therapy. Ferroptosis, an iron-dependent form of regulated cell death driven by lipid peroxidation, has emerged as a critical mechanism influencing tumour survival and therapeutic response. However, the role of ferroptosis suppressor genes (FSGs) in pancreatic cancer remains incompletely understood. In this study, FSGs were systematically retrieved from FerrDb V2 and subjected to multi-step filtering to identify a curated set of 196 protein-coding genes. Genomic alteration analysis using cBioPortal across 21 pancreatic cancer studies (n = 5189 samples) identified seven recurrently altered genes (TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, and ENPP2) with alteration frequencies ≥ 3%. Co-occurrence analysis revealed predominantly cooperative alteration patterns among these genes, suggesting functional complementarity. Survival analysis demonstrated that alterations in FSGs were significantly associated with reduced overall survival in pancreatic cancer, with several genes (ECH1, ZFP36, SIRT2, and ENPP2) showing particularly strong adverse prognostic effects. In contrast, no significant survival associations were observed in oesophageal and gastric cancers, indicating a tumour-specific dependency on ferroptosis-related mechanisms. KEGG pathway enrichment analysis of the broader FSG set revealed significant involvement in pathways related to metabolic regulation (AMPK–mTOR signalling), autophagy, hypoxia response (HIF-1 signalling), and oncogenic signalling (PI3K–Akt pathway). Integration of these findings suggests that ferroptosis suppressor genes contribute to pancreatic cancer progression by promoting metabolic adaptation and resistance to oxidative stress. In conclusion, this study identifies key ferroptosis suppressor genes with prognostic relevance in pancreatic cancer and highlights their integration within critical metabolic and stress-response pathways. The tumour-specific nature of these associations underscores the importance of biological context and supports the potential of FSGs as prognostic biomarkers and potential therapeutic targets in ferroptosis-based strategies.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12672-026-05269-6.Keywords: Ferroptosis suppressor genes, Pancreatic cancer (PDAC), Survival analysis, Genomic alterations, Metabolic reprogramming, KEGG pathway enrichment Introduction Pancreatic cancer, predominantly pancreatic ductal adenocarcinoma (PDAC), remains one of the most lethal solid malignancies worldwide because it is often diagnosed late, progresses rapidly, and responds poorly to currently available therapies. Global estimates for 2022 indicate roughly 510,992 new cases and 467,409 deaths, underscoring its exceptionally high mortality burden (https://www.wcrf.org/preventing-cancer/cancer-statistics/pancreatic-cancer-statistics/, https://gco.iarc.who.int/media/globocan/factsheets/cancers/13-pancreas-fact-sheet.pdf). Even with advances in systemic therapy and perioperative care, long-term survival remains poor, and PDAC continues to be defined by marked biological aggressiveness, early dissemination, and profound therapeutic resistance [12, 13]. These features highlight the need to identify biologically meaningful molecular determinants that may improve prognostic stratification and reveal new therapeutic vulnerabilities [12, 13]. Ferroptosis has emerged as a relevant candidate process in this context. It is an iron-dependent form of regulated cell death driven by the accumulation of lipid peroxides and is mechanistically distinct from apoptosis, necroptosis, and other cell death programmes [3, 21]. Because ferroptosis is tightly linked to redox control, membrane lipid composition, mitochondrial metabolism, and antioxidant defence, it has attracted growing attention in cancer research as both a tumour-suppressive mechanism and a potential therapeutic target [23, 26]. In pancreatic cancer specifically, ferroptosis has been increasingly implicated in tumour progression, treatment response, and microenvironmental adaptation, suggesting that impaired ferroptotic susceptibility may contribute to PDAC persistence and aggressiveness [9, 11]. This possibility is especially relevant in PDAC because pancreatic tumours exist in a biologically hostile environment characterised by hypoxia, nutrient limitation, dense desmoplasia, autophagy dependence, and extensive metabolic rewiring [6, 8]. These adaptations are not merely background features of the disease; they are central to tumour survival. Hypoxia-associated metabolic reprogramming supports pancreatic cancer cell fitness under stress, while autophagy has been shown to be required for pancreatic tumour growth and maintenance [6, 22]. Because ferroptosis is closely influenced by oxidative stress handling, lipid metabolism, and stress-response pathways, genes that suppress ferroptosis may acquire particular importance in PDAC [3, 26]. Among the available resources for studying ferroptosis biology, FerrDb V2 provides a curated catalogue of ferroptosis regulators and disease associations assembled from the published literature. The database includes more than 600 gene regulators, among which 238 are classified as ferroptosis suppressors, making it a useful foundation for systematic interrogation of this gene class [25]. However, although ferroptosis-related signatures have been explored in several malignancies, the subset of ferroptosis suppressor genes (FSGs) has not been comprehensively prioritised in pancreatic cancer using an integrated framework that combines genomic alteration profiling, survival analysis, pathway interpretation, and cross-cancer comparison [11, 25]. In particular, it remains unclear which recurrently altered FSGs are most relevant to pancreatic cancer outcome and whether their clinical significance is specific to PDAC or shared across related gastrointestinal tumours. To address this gap, the present study systematically curated FSGs from FerrDb V2 and evaluated their genomic alterations in pancreatic cancer using large-scale public cancer genomics resources. Recurrently altered genes were then prioritised and assessed for co-occurrence patterns, prognostic relevance, and pathway context, with oesophageal/gastric cancers analysed as a comparative gastrointestinal cohort. By integrating alteration frequency, survival impact, and KEGG pathway enrichment, this study aimed to identify ferroptosis suppressor genes with potential biological and clinical relevance in pancreatic cancer and to clarify whether their importance reflects a broader gastrointestinal cancer phenomenon or a more pancreas-weighted dependency. Materials and methods Data collection and filtering of ferroptosis suppressor genes FSGs were retrieved from the FerrDb V2 database, a curated repository of ferroptosis regulators, on 5 April 2026 (http://www.zhounan.org/ferrdb/current/). The downloaded dataset initially comprised 348 entries corresponding to experimentally supported ferroptosis suppressors. To ensure data quality and relevance to human pancreatic cancer, a stepwise filtering strategy was applied. First, duplicate entries were removed, yielding 238 unique genes. Subsequently, only Homo sapiens entries were retained. Genes lacking a valid HUGO Gene Nomenclature Committee (HGNC) identifier were excluded to ensure standardised gene annotation. In addition, only protein-coding genes were retained, as downstream analyses focused on biologically actionable and therapeutically relevant targets. After these filtering steps, 197 genes remained. The filtered gene list was then cross-referenced with the cBioPortal database to ensure compatibility with downstream cancer genomic analyses. One gene, ARNTL, was not present in cBioPortal and was therefore excluded. Finally, a total of 196 ferroptosis suppressor genes were retained for subsequent analyses. Genomic alteration analysis of ferroptosis suppressor genes across cancer types Genomic alteration profiles of the curated 196 FSGs were analysed using the cBioPortal database (https://www.cbioportal.org/) [2, 5]. Pancreatic cancer datasets were identified by querying studies related to the pancreas, yielding a total of 21 studies comprising 5,189 tumour samples. The 196 FSGs were queried collectively using the “Query by Gene” function, and genomic alterations—including mutation, structural variant, amplification, deep deletion, and multiple alteration s —were assessed. The results were visualised using OncoPrint representations, where samples were organised based on gene-wise alteration patterns and types of genomic events. The percentage of samples exhibiting alterations for each gene was extracted. To identify genes with potential biological and clinical relevance, a threshold of ≥ 3% alteration frequency across samples was applied. Based on this criterion, seven FSGs (TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, and ENPP2) were shortlisted for further analysis. To evaluate whether these alterations were consistent across other gastrointestinal malignancies, a comparative analysis was performed in oesophageal and gastric cancer datasets within cBioPortal. A total of 22 studies comprising 5,623 samples were analysed using the same gene set and methodology. Alteration frequencies for these genes were extracted for cross-cancer comparison. Mutual exclusivity and co-occurrence analysis To investigate potential functional relationships among the selected FSGs, mutual exclusivity and co-occurrence analyses were performed using the cBioPortal platform. All possible pairwise combinations among the seven selected genes (n = 21 pairs) were analysed across the queried samples. The analysis evaluated whether genomic alterations in one gene were statistically associated with alterations in another gene within the same tumour samples. The strength and direction of association between gene pairs were quantified using the log2 odds ratio (log2 OR). A positive log2 OR indicates a tendency towards co-occurrence, whereas a negative or zero log2 OR suggests mutual exclusivity. Statistical significance was assessed using a two-sided Fisher’s exact test, and p-values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) correction. Associations were considered statistically significant at q < 0.05, with stronger associations highlighted at q < 0.001. Survival analysis of ferroptosis suppressor gene alterations Survival analysis was performed using the cBioPortal platform to evaluate the prognostic significance of genomic alterations in FSGs across cancer cohorts. Patients were stratified into groups based on the presence or absence of genomic alterations in the selected genes. The altered group comprised samples harbouring at least one genomic alteration (mutation, copy number alteration, or structural variation) in any of the queried genes, whereas the unaltered group included samples without detectable alterations in these genes. Kaplan–Meier survival curves were generated to estimate overall survival (OS), defined as the time from diagnosis to death or last follow-up. Survival differences between groups were assessed using the log-rank test, which evaluates the null hypothesis that there is no difference in survival distributions between groups. Hazard ratios (HRs) with 95% confidence intervals (CIs) were computed to quantify the relative risk of death associated with gene alterations. An HR > 1 indicates poorer survival in the altered group, whereas an HR < 1 indicates improved survival. Statistical significance was considered at p < 0.05. Pairwise survival comparisons were performed between: (1) Unaltered group vs. combined altered group. (2) Unaltered group vs. individual gene-altered groups (TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, ENPP2). Analyses were conducted separately for pancreatic cancer and oesophageal/gastric cancer cohorts. KEGG pathway enrichment analysis Functional enrichment analysis was performed to investigate the biological pathways associated with FSGs. The curated list of 196 genes was analysed using the DAVID (Database for Annotation, Visualization and Integrated Discovery) (https://davidbioinformatics.nih.gov/) [7, 18]. The analysis was conducted using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database, with Homo sapiens selected as the reference organism. Enrichment significance was assessed using a modified Fisher’s exact test (EASE score). To account for multiple hypothesis testing, the Benjamini–Hochberg false discovery rate (FDR) correction was applied. Pathways with FDR < 0.05 were considered statistically significant. Significantly enriched pathways were ranked based on FDR and gene count, and biologically relevant pathways were grouped into functional categories, including metabolic regulation, stress response, and cancer-related signalling pathways. Results Curation and refinement of ferroptosis suppressor gene set A total of 348 ferroptosis suppressor gene entries were initially obtained from FerrDb V2. Removal of duplicate records reduced the dataset to 238 unique genes. Subsequent filtering based on species specificity (Homo sapiens), HGNC annotation, and protein-coding status further refined the dataset to 197 genes. Cross-validation with the cBioPortal database resulted in the exclusion of one gene (ARNTL) due to its absence, yielding a final curated set of 196 ferroptosis suppressor genes (Table S1 and Table S2). This systematic filtering ensured that the final gene set was human-specific, well annotated, and compatible with cancer genomics resources, thereby providing a robust foundation for downstream expression and functional analyses. Workflow of the study was depicted in Fig. 1. Workflow of the study. Schematic representation of the study design illustrating the identification and analysis of ferroptosis suppressor genes (FSGs). Genes were retrieved from FerrDb V2 and filtered to obtain 196 protein-coding genes. Genomic alteration analysis using cBioPortal identified seven key genes with ≥ 3% alteration frequency in pancreatic cancer. These genes were further analysed for co-occurrence patterns, survival associations, and cross-cancer validation. KEGG pathway enrichment analysis provided biological insights into the involvement of metabolic, stress-response, and cancer-related pathways, highlighting tumour-specific ferroptosis dependency in pancreatic cancer. HR is Hazard ratio Genomic alteration landscape of ferroptosis suppressor genes in pancreatic cancer Genomic alteration analysis of 196 ferroptosis suppressor genes across 21 pancreatic cancer studies (n = 5,189 samples) revealed a heterogeneous alteration landscape, with alteration frequencies ranging from 0% to 63%. Application of a ≥ 3% alteration frequency threshold identified seven genes (TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, and ENPP2) as recurrently altered in pancreatic cancer (Figure S1, Table 1 and Table S3). Among these, TP53 exhibited the highest alteration frequency (63%), indicating its dominant role in pancreatic tumorigenesis. The remaining genes showed moderate alteration frequencies, ranging from 3% to 5%, suggesting potential involvement in ferroptosis regulation within the pancreatic cancer context. These findings highlight a subset of ferroptosis suppressor genes with recurrent genomic alterations that may contribute to tumour progression. Table 1Alteration frequencies of selected ferroptosis suppressor genes in pancreatic cancerGeneAltered SamplesFrequency (%) TP53 322563 HSF1 865 PARP10 845 ZFP36 634 SIRT2 583 ECH1 553 ENPP2 543Alteration frequency represents the percentage of tumour samples exhibiting at least one genomic alteration (mutation, copy number variation, or mRNA expression change) in the queried gene Alteration frequencies of selected ferroptosis suppressor genes in pancreatic cancer Alteration frequency represents the percentage of tumour samples exhibiting at least one genomic alteration (mutation, copy number variation, or mRNA expression change) in the queried gene Cross-cancer comparison of altered ferroptosis suppressor genes in oesophageal and gastric cancers To assess the broader relevance of the identified genes, alteration analysis was extended to oesophageal and gastric cancers across 22 studies (n = 5,623 samples). The selected seven FSGs demonstrated comparable alteration patterns, with frequencies ranging from approximately 2.5% to 63%. Consistent with pancreatic cancer findings, TP53 remained the most frequently altered gene (63%), reinforcing its central role across gastrointestinal malignancies. Other genes, including ENPP2, PARP10, and HSF1, showed moderate alteration frequencies (7–9%), while ZFP36, ECH1, and SIRT2 exhibited lower but consistent alteration rates (~ 2.5-3%) (Table 2). This cross-cancer analysis suggests that a subset of ferroptosis suppressor genes exhibits conserved alteration patterns across gastrointestinal cancers, supporting their potential as broadly relevant therapeutic targets. Table 2Alteration frequencies of selected ferroptosis suppressor genes in oesophageal and gastric cancersGeneAltered SamplesFrequency (%) TP53 351863 ENPP2 3299 PARP10 2888 HSF1 2457 ZFP36 932.7 ECH1 912.6 SIRT2 882.5Alteration frequencies were calculated across 22 studies (n = 5623 samples) using cBioPortal. Values reflect the proportion of samples with genomic alterations Alteration frequencies of selected ferroptosis suppressor genes in oesophageal and gastric cancers Alteration frequencies were calculated across 22 studies (n = 5623 samples) using cBioPortal. Values reflect the proportion of samples with genomic alterations Cancer type-specific alteration patterns of selected ferroptosis suppressor genes To further characterise the genomic landscape of the selected FSGs, cancer type–specific alteration patterns were analysed across different subtypes of pancreatic and oesophageal/gastric cancers using cBioPortal. Pancreatic cancer In pancreatic cancer subtypes, TP53 exhibited a predominantly mutation-driven alteration profile, with mutations representing the major alteration type, followed by occasional deep deletions. This pattern was consistent across multiple pancreatic cancer subtypes, highlighting the well-established role of TP53 mutations in pancreatic tumorigenesis. In contrast, HSF1 and PARP10 demonstrated alteration patterns dominated by gene amplification, followed by mutations and rare deep deletions, particularly in pancreatic adenocarcinoma. Similarly, ZFP36 showed a mixed alteration profile, with both mutations and amplifications contributing substantially across pancreatic cancer subtypes. For SIRT2, amplification emerged as the most frequent alteration, while mutations were primarily observed in pancreatic neuroendocrine tumours, suggesting subtype-specific variation in its genomic alterations. ECH1 exhibited a predominance of amplification events, especially in pancreatic adenocarcinoma, with minimal contribution from other alteration types. ENPP2 displayed a dual pattern characterised by both mutations and amplifications, indicating potential involvement of multiple genomic mechanisms. Overall, amplification and mutation were the predominant alteration types across most genes, with variations in their relative contributions depending on the gene and cancer subtype (Fig. 2). Cancer type–specific alteration patterns of ferroptosis suppressor genes in pancreatic cancer. Bar plots showing the distribution of genomic alteration types (mutation, structural variant, amplification, deep deletion, and multiple alterations) for selected ferroptosis suppressor genes across pancreatic cancer subtypes. Each panel represents an individual gene, and bars indicate alteration frequencies within specific tumour types. Mutation and amplification represent the dominant alteration mechanisms across most genes Oesophageal and gastric cancers In oesophageal and gastric cancers, TP53 again showed a mutation-dominated profile across all major subtypes, reinforcing its conserved role across gastrointestinal malignancies. For HSF1, PARP10, ZFP36, and ECH1, gene amplification was the predominant alteration, followed by mutations and occasional deep deletions. Notably, multiple concurrent alterations were observed in specific tumour types, particularly in stomach adenocarcinoma and oesophagogastric cancers, suggesting increased genomic instability in these subtypes. SIRT2 exhibited a mixed alteration pattern, with both mutations and amplifications contributing substantially, while deep deletions were specifically observed in oesophageal squamous cell carcinoma, stomach adenocarcinoma, and oesophagogastric cancer. Similarly, ECH1 showed amplification as the dominant alteration, with additional contributions from mutations and deep deletions in selected cancer subtypes. For ENPP2, both mutation and amplification events were prominent, with multiple alterations observed in stomach adenocarcinoma, indicating potential co-occurrence of genomic events. These findings demonstrate that while core alteration patterns are conserved, the distribution and complexity of genomic alterations vary across cancer subtypes, reflecting tumour-specific genomic contexts (Fig. 3). Cancer type–specific alteration patterns of ferroptosis suppressor genes in oesophageal and gastric cancers. Bar plots illustrating genomic alteration profiles of selected ferroptosis suppressor genes across oesophageal and gastric cancer subtypes. Amplification and mutation are the predominant alteration types, with subtype-specific occurrences of deep deletions and multiple alterations, particularly in stomach adenocarcinoma and oesophagogastric cancers Co-occurrence patterns among ferroptosis suppressor genes in pancreatic cancer Mutual exclusivity analysis of the seven selected ferroptosis suppressor genes in pancreatic cancer revealed a strong predominance of co-occurrence patterns among gene pairs. A total of 11 gene pairs demonstrated significant co-occurrence (log2 OR > 3, q < 0.001), indicating a non-random tendency for these genes to be altered together within the same tumour samples. Notable co-occurring gene pairs included HSF1–PARP10, SIRT2–ECH1, ZFP36–SIRT2, ZFP36–ECH1, PARP10–ENPP2, and multiple interactions involving HSF1 and PARP10, suggesting that these genes may participate in interconnected regulatory mechanisms (Table 3 and Table S4). Importantly, no statistically significant mutual exclusivity patterns were observed among the analysed gene pairs. This absence of exclusivity suggests that these genes are unlikely to function as redundant alternatives and instead may operate in complementary or synergistic pathways that collectively contribute to pancreatic tumorigenesis. Table 3Significant co-occurring gene pairs among ferroptosis suppressor genes in pancreatic cancerGene AGene BLog2 ORq-valueInterpretation HSF1 PARP10 > 3< 0.001Co-occurrence SIRT2 ECH1 > 3< 0.001Co-occurrence ZFP36 SIRT2 > 3< 0.001Co-occurrence ZFP36 ECH1 > 3< 0.001Co-occurrence PARP10 ENPP2 > 3< 0.001Co-occurrence HSF1 ENPP2 > 3< 0.001Co-occurrence HSF1 SIRT2 > 3< 0.001Co-occurrence HSF1 ECH1 > 3< 0.001Co-occurrence PARP10 SIRT2 > 3< 0.001Co-occurrence PARP10 ECH1 > 3< 0.001Co-occurrence PARP10 ZFP36 > 3< 0.001Co-occurrenceLog2 OR log2 odds ratio. Positive values indicate co-occurrence. q-values were obtained using Benjamini–Hochberg FDR correction. Only statistically significant pairs (q < 0.05) are shown Significant co-occurring gene pairs among ferroptosis suppressor genes in pancreatic cancer Log2 OR log2 odds ratio. Positive values indicate co-occurrence. q-values were obtained using Benjamini–Hochberg FDR correction. Only statistically significant pairs (q < 0.05) are shown Co-occurrence patterns in oesophageal and gastric cancers A similar analysis performed in oesophageal and gastric cancers identified six gene pairs exhibiting significant co-occurrence (log2 OR > 3, q < 0.001), including HSF1–PARP10, HSF1–ENPP2, PARP10–ENPP2, SIRT2–ECH1, ZFP36–SIRT2, and ZFP36–ECH1 (Table 4 and Table S5). The recurrence of these co-occurrence relationships across multiple cancer types suggests a conserved pattern of coordinated genomic alterations among these ferroptosis suppressor genes. As observed in pancreatic cancer, no significant mutual exclusivity was detected. Table 4Significant co-occurring gene pairs in oesophageal and gastric cancersGene AGene BLog2 ORq-valueInterpretation HSF1 PARP10 > 3< 0.001Co-occurrence HSF1 ENPP2 > 3< 0.001Co-occurrence PARP10 ENPP2 > 3< 0.001Co-occurrence SIRT2 ECH1 > 3< 0.001Co-occurrence ZFP36 SIRT2 > 3< 0.001Co-occurrence ZFP36 ECH1 > 3< 0.001Co-occurrenceLog2 OR log2 odds ratio. Positive values indicate co-occurrence. q-values were obtained using Benjamini–Hochberg FDR correction. Only statistically significant pairs (q < 0.05) are shown Significant co-occurring gene pairs in oesophageal and gastric cancers Log2 OR log2 odds ratio. Positive values indicate co-occurrence. q-values were obtained using Benjamini–Hochberg FDR correction. Only statistically significant pairs (q < 0.05) are shown The predominance of co-occurrence over mutual exclusivity indicates that these ferroptosis suppressor genes may not act as alternative drivers but rather contribute additively or synergistically to tumour progression. Such coordinated alterations may enhance tumour cell survival by simultaneously modulating multiple ferroptosis-related pathways. These findings support the hypothesis that targeting combinations of ferroptosis suppressor genes, rather than individual genes alone, may offer improved therapeutic potential. However, further functional validation is required to confirm these interactions. Prognostic significance of ferroptosis suppressor gene alterations in pancreatic cancer Kaplan–Meier survival analysis was performed to evaluate the prognostic impact of genomic alterations in FSGs in pancreatic cancer. Patients harbouring alterations in any of the selected FSGs (altered group) exhibited significantly reduced overall survival compared to the unaltered group. The survival curves showed early and sustained separation, indicating a consistent adverse prognostic effect. The altered group demonstrated a significantly increased risk of mortality (HR = 1.643, 95% CI: 1.521–1.776; log-rank p < 0.001), confirming the overall negative impact of FSG alterations on patient outcomes (Fig. 4). Kaplan–Meier survival analysis of ferroptosis suppressor gene alterations in pancreatic cancer. Kaplan–Meier survival curves showing overall survival in pancreatic cancer patients stratified by alteration status of ferroptosis suppressor genes. The combined altered group (red) shows significantly reduced survival compared to the unaltered group (blue). Additional panels illustrate gene-specific survival comparisons for TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, and ENPP2. Several genes, particularly ECH1, ZFP36, SIRT2, and ENPP2, demonstrate strong adverse prognostic effects, with hazard ratios exceeding 2.0 To further dissect the contribution of individual genes, pairwise survival analyses were conducted between the unaltered group and each gene-altered subgroup (Fig. 4). Among the analysed genes, TP53, the most frequently altered gene, showed a significant association with poor survival (HR = 1.640, 95% CI: 1.518–1.772; p < 0.001), closely mirroring the survival effect observed in the combined altered group. This suggests that TP53 alterations may represent a major driver of the overall survival disadvantage. In contrast, several lower-frequency genes exhibited stronger hazard ratios, indicating a disproportionately high impact on survival despite their lower prevalence. Notably, ECH1 (HR = 2.264), ZFP36 (HR = 2.199), SIRT2 (HR = 2.143), and ENPP2 (HR = 2.102) were associated with the highest risk estimates, all exceeding a two-fold increase in mortality risk. These genes showed steep early declines in survival curves, suggesting association with aggressive disease behaviour. Intermediate effects were observed for HSF1 (HR = 1.829) and PARP10 (HR = 1.784), which also demonstrated significant associations with reduced survival (Table 5). Table 5Summary of survival analysis for ferroptosis suppressor genes in pancreatic cancerGeneAltered SamplesHR95% CIp-valueq-valueInterpretationAltered (combined)32641.6431.521–1.776< 0.001< 0.001Worse survival TP53 32251.6401.518–1.772< 0.001< 0.001Core driver HSF1 861.8291.107–3.0210.0010.002Moderate risk PARP10 841.7841.077–2.9560.0020.005Moderate risk ZFP36 632.1991.201–4.029< 0.001< 0.001Strong risk SIRT2582.1431.208–3.801< 0.001< 0.001Strong risk ECH1 552.2641.191–4.305< 0.001< 0.001Strongest risk ENPP2 542.1021.262–3.4993.14e-05< 0.001Strong riskHR Hazard ratio, CI confidence interval. HR > 1 indicates increased risk of death. p-values derived from log-rank test, q-values derived from BH-FDR correction Summary of survival analysis for ferroptosis suppressor genes in pancreatic cancer HR Hazard ratio, CI confidence interval. HR > 1 indicates increased risk of death. p-values derived from log-rank test, q-values derived from BH-FDR correction Collectively, these findings reveal a two-tier prognostic structure: High-frequency driver: TP53, which largely accounts for the population-level survival effect. Low-frequency, high-impact modifiers: ECH1, ZFP36, SIRT2, ENPP2 (strongest effects) and HSF1, PARP10 (moderate effects). This pattern suggests that while TP53 alterations contribute broadly to tumour progression, a subset of ferroptosis suppressor genes may define high-risk molecular subgroups with particularly poor outcomes. The predominance of strong hazard ratios among multiple ferroptosis suppressor genes indicates that dysregulation of ferroptosis-related pathways may play a critical role in pancreatic cancer progression. The consistent early separation of survival curves across multiple genes further supports their potential relevance in aggressive disease phenotypes. However, given the relatively smaller number of altered cases for several genes, these findings should be interpreted with caution and require validation in independent pancreatic cancer cohorts. Prognostic significance of ferroptosis suppressor gene alterations in oesophageal and gastric cancers Kaplan–Meier survival analysis was performed to evaluate the prognostic impact of genomic alterations in FSGs in oesophageal and gastric cancers. In contrast to pancreatic cancer, no significant difference in overall survival was observed between patients harbouring alterations in any of the selected FSGs (altered group) and those without alterations (unaltered group). The survival curves of the two groups showed substantial overlap throughout the follow-up period, indicating comparable survival outcomes (HR = 0.978, 95% CI: 0.890–1.074; log-rank p = 0.635) (Fig. 5). Kaplan–Meier survival analysis of ferroptosis suppressor gene alterations in oesophageal and gastric cancers. Kaplan–Meier survival curves showing overall survival in oesophageal and gastric cancer patients stratified by alteration status of ferroptosis suppressor genes. The altered group (red) and unaltered group (blue) show no significant difference in survival (HR = 0.978, 95% CI: 0.890–1.074; log-rank p = 0.635). Additional panels illustrate gene-specific comparisons, all of which demonstrate no significant association with survival To further assess the contribution of individual genes, survival comparisons were performed between the unaltered group and each gene-altered subgroup (Fig. 5). Consistent with the overall analysis, none of the individual genes demonstrated a statistically significant association with overall survival. Hazard ratios for all genes were close to unity, and confidence intervals consistently crossed 1.0, indicating the absence of a measurable prognostic effect. Specifically, TP53 (HR = 0.978), HSF1 (HR = 0.957), and PARP10 (HR = 0.981) showed no evidence of increased mortality risk. Similarly, genes that exhibited strong prognostic effects in pancreatic cancer ZFP36 (HR = 1.124), SIRT2 (HR = 1.086), ECH1 (HR = 1.033), and ENPP2 (HR = 0.985) also showed no significant association with survival in this cohort (Table 6). Table 6Summary of survival analysis for ferroptosis suppressor genes in oesophageal and gastric cancersGeneAltered SamplesHR95% CIp-valueq-valueInterpretationAltered (combined)37340.9780.890–1.0740.6350.762No significant effect TP53 35180.9780.889–1.0750.6410.77No significant effect HSF1 2450.9570.759–1.2070.7160.716No significant effect PARP10 2880.9810.788–1.2210.8620.862No significant effect ZFP36 931.1240.778–1.6260.5100.510No significant effect SIRT2 881.0860.738–1.5970.6650.665No significant effect ECH1 911.0330.717–1.4900.8580.858No significant effect ENPP2 3290.9850.802–1.2090.8860.886No significant effectHR Hazard ratio, CI confidence interval. HR values close to 1.0 indicate no significant difference in survival between groups. p-values derived from the log-rank test. q-values derived from BH-FDR correction Summary of survival analysis for ferroptosis suppressor genes in oesophageal and gastric cancers HR Hazard ratio, CI confidence interval. HR values close to 1.0 indicate no significant difference in survival between groups. p-values derived from the log-rank test. q-values derived from BH-FDR correction Collectively, these findings indicate a uniform lack of prognostic significance for ferroptosis suppressor gene alterations in oesophageal and gastric cancers. The absence of consistent survival differences across both combined and gene-specific analyses suggests that these alterations do not play a major role in determining patient outcomes in this cancer type. The lack of survival association contrasts sharply with the findings in pancreatic cancer, where multiple ferroptosis suppressor genes demonstrated strong and consistent prognostic effects. This divergence suggests that the clinical impact of ferroptosis-related alterations is highly context-dependent, varying across tumour types. These observations imply that ferroptosis suppressor genes may have limited prognostic utility in oesophageal and gastric cancers, despite their biological relevance in other malignancies. Further studies are required to determine whether these genes contribute to tumour biology through mechanisms not directly reflected in survival outcomes. Cross-cancer comparison of prognostic impact of ferroptosis suppressor gene alterations To evaluate whether the prognostic significance of FSG alterations is consistent across tumour types, a comparative survival analysis was performed between pancreatic cancer and oesophageal/gastric cancers. A striking contrast was observed between the two cancer types. In pancreatic cancer, alterations in FSGs were consistently associated with significantly reduced overall survival, both at the combined gene level and across multiple individual genes. In contrast, no significant survival differences were observed in oesophageal and gastric cancers for either the combined altered group or any individual gene. The combined altered group in pancreatic cancer demonstrated a significant increase in mortality risk (HR = 1.643, p < 0.001), whereas in oesophageal/gastric cancers, the corresponding analysis showed no effect (HR = 0.978, p = 0.635) (Table 7). This divergence indicates that the overall prognostic impact of FSG alterations is not conserved across cancer types. Table 7Cross-cancer comparison of hazard ratios for ferroptosis suppressor genesGenePancreatic Cancer HROesophageal/Gastric HR TP53 1.6400.978 HSF1 1.8290.957 PARP10 1.7840.981 ZFP36 2.1991.124 SIRT2 2.1431.086 ECH1 2.2641.033 ENPP2 2.1020.985 Cross-cancer comparison of hazard ratios for ferroptosis suppressor genes At the individual gene level, all seven analysed genes (TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, and ENPP2) exhibited a consistent pattern of significant survival association in pancreatic cancer, with hazard ratios ranging from approximately 1.6 to 2.3. In contrast, none of these genes demonstrated a statistically significant association with survival in oesophageal/gastric cancers, with hazard ratios consistently close to unity and confidence intervals crossing 1.0. Notably, genes that showed the strongest prognostic impact in pancreatic cancer ECH1, ZFP36, SIRT2, and ENPP2 did not exhibit any measurable survival effect in oesophageal/gastric cancers. Similarly, TP53, which largely accounted for the population-level survival effect in pancreatic cancer, showed no prognostic relevance in the latter cohort. These findings highlight a tumour-type–specific dependency on ferroptosis-related pathways. In pancreatic cancer, the consistent and strong association between FSG alterations and reduced survival suggests that dysregulation of ferroptosis suppression may play a critical role in tumour progression and patient outcomes. In contrast, the absence of a survival association in oesophageal/gastric cancers suggests that these tumours may rely on alternative biological pathways, or that ferroptosis-related alterations do not represent a dominant determinant of clinical outcome in this context. Based on the observed patterns, FSG alterations in pancreatic cancer can be broadly categorised into two functional groups: High-frequency driver: TP53, contributing to the overall survival effect at the population level. Low-frequency, high-impact modifiers: ECH1, ZFP36, SIRT2, ENPP2 (strongest effects) and HSF1, PARP10 (moderate effects). This layered structure suggests that ferroptosis suppression may contribute to tumour progression through both broad driver mechanisms and subtype-specific high-risk alterations. Collectively, these results demonstrate that the prognostic relevance of ferroptosis suppressor gene alterations is highly context-dependent, with strong and consistent effects observed in pancreatic cancer but not in oesophageal/gastric cancers. This divergence underscores the importance of cancer-type–specific evaluation when assessing the clinical significance of molecular alterations. KEGG pathway enrichment analysis of ferroptosis suppressor genes KEGG pathway enrichment analysis of 196 ferroptosis suppressor genes revealed significant enrichment in pathways related to cellular metabolism, stress adaptation, and oncogenic signalling, indicating a broad functional involvement of these genes in tumour biology. Among the most significantly enriched pathways were the AMPK signalling pathway and the mTOR signalling pathway, both of which are central regulators of cellular energy homeostasis and metabolic reprogramming. Key genes such as SIRT1, PRKAA2, MTOR, and PIK3CA were involved in these pathways, highlighting the role of ferroptosis suppressor genes in regulating metabolic adaptation (Table S6). The enrichment of the autophagy pathway and longevity-regulating pathway indicates a strong association between ferroptosis suppression and cellular stress-response mechanisms. These pathways are known to regulate oxidative stress, lipid metabolism, and cell survival under adverse conditions. Additionally, enrichment of the HIF-1 signalling pathway suggests a role for hypoxia-driven metabolic reprogramming, a key feature of tumour microenvironments that influences ferroptosis sensitivity. Several canonical cancer-related pathways, including the PI3K–Akt signalling pathway, TNF signalling pathway, and pathways in cancer, were significantly enriched. These pathways involve critical oncogenic regulators such as TP53, PIK3CA, STAT3, and RELA, indicating that ferroptosis suppressor genes are closely integrated with tumour progression mechanisms. Importantly, several of the seven ferroptosis suppressor genes identified through survival analysis (TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, and ENPP2) were found to be functionally associated with the enriched pathways. TP53 was prominently represented in multiple cancer-related pathways, including PI3K–Akt signalling and pathways in cancer, consistent with its central role in tumour progression. TP53 was prominently represented in multiple cancer-related pathways, including PI3K–Akt signalling and pathways in cancer, consistent with its central role in tumour progression. SIRT2 was associated with metabolic regulation through AMPK-related pathways, linking it to cellular energy homeostasis and ferroptosis sensitivity. SIRT2 was associated with metabolic regulation through AMPK-related pathways, linking it to cellular energy homeostasis and ferroptosis sensitivity. ECH1, involved in fatty acid metabolism, aligns with metabolic pathways relevant to lipid peroxidation, a key process in ferroptosis. ECH1, involved in fatty acid metabolism, aligns with metabolic pathways relevant to lipid peroxidation, a key process in ferroptosis. ZFP36 and HSF1 are implicated in stress and inflammatory signalling pathways, supporting their role in cellular adaptation and tumour survival. ZFP36 and HSF1 are implicated in stress and inflammatory signalling pathways, supporting their role in cellular adaptation and tumour survival. PARP10 and ENPP2 are associated with pathways involved in DNA damage response and lipid signalling, respectively, further contributing to tumour progression mechanisms. PARP10 and ENPP2 are associated with pathways involved in DNA damage response and lipid signalling, respectively, further contributing to tumour progression mechanisms. These observations indicate that the prognostically relevant genes are not isolated entities but are embedded within broader biological networks identified through pathway enrichment analysis. Although KEGG pathway analysis provides a general overview of biological processes, integration with survival findings highlights their potential relevance in pancreatic cancer. Several enriched pathways, including AMPK–mTOR signalling, autophagy, and HIF-1 signalling, are known to play critical roles in pancreatic tumour biology. Pancreatic cancer is characterised by a highly metabolically active and hypoxic microenvironment, requiring tumour cells to adapt through metabolic reprogramming and stress-response mechanisms. The enrichment of these pathways suggests that ferroptosis suppressor genes may contribute to tumour progression by supporting metabolic adaptation, resistance to oxidative stress, and enhanced survival under hypoxic conditions. The strong association of these genes with poor survival specifically in pancreatic cancer further supports their functional relevance in this tumour type, indicating that ferroptosis suppression may be an important mechanism underlying tumour aggressiveness. Discussion This study systematically evaluated FSGs and identified a subset of seven genes (TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, and ENPP2) that are recurrently altered and prognostically relevant in pancreatic cancer. The most notable finding is the consistent association between FSG alterations and reduced overall survival in pancreatic cancer, contrasted by the absence of a significant survival effect in oesophageal/gastric cancers, indicating a strong tumour-context dependency. Among the identified genes, TP53 showed the highest alteration frequency and largely recapitulated the survival effect observed in the combined altered group. This is consistent with its well-established role as a core driver of pancreatic cancer progression and poor prognosis [19]. However, several lower-frequency genes, including ECH1, ZFP36, SIRT2, and ENPP2, demonstrated higher hazard ratios, suggesting that they may define smaller but more aggressive molecular subgroups. This layered structure supports a model in which ferroptosis suppression contributes to tumour progression through both broad driver effects (TP53) and subtype-specific high-risk modifiers. The predominance of co-occurrence patterns among these genes further suggests that they function in complementary rather than redundant pathways, consistent with the multi-layered adaptive biology of pancreatic cancer, which relies on coordinated regulation of metabolism, stress response, and survival signalling [20]. Gene-specific interpretations align with existing literature. HSF1 has been implicated in pancreatic tumorigenesis, invasion, and metastasis, particularly through stress-adaptive transcriptional programmes [14]. SIRT2 is involved in metabolic regulation and oncogenic signalling, including modulation of KRAS activity, supporting its role in tumour fitness [4]. ENPP2 (autotaxin) contributes to tumour–stroma interactions and has been shown to promote pancreatic cancer progression through lysophosphatidate signalling [1]. For PARP10, ZFP36, and ECH1, the direct evidence in pancreatic cancer is comparatively limited. PARP10 has been linked to replication stress tolerance and tumour proliferation [17], while ZFP36 family proteins regulate inflammatory and oncogenic transcripts [16]. ECH1 is involved in fatty-acid metabolism, which is mechanistically relevant to ferroptosis through lipid peroxidation pathways [24]. In this context, these genes should be considered promising candidates requiring further validation, rather than fully established pancreatic drivers. KEGG pathway analysis provided important mechanistic context by showing enrichment in pathways related to AMPK–mTOR signalling, autophagy, HIF-1 signalling, and PI3K–Akt signalling, which collectively regulate metabolism, hypoxia adaptation, and stress response. These pathways are highly relevant to pancreatic cancer, a tumour type characterised by hypoxia, nutrient limitation, and dependence on metabolic reprogramming [15]. In particular, autophagy has been shown to be essential for pancreatic tumour growth and survival under metabolic stress (https://research.birmingham.ac.uk/en/publications/pancreatic-cancers-require-autophagy-for-tumor-growth/). The enrichment of these pathways supports a model in which FSGs contribute to tumour progression by enhancing metabolic flexibility and resistance to oxidative stress, thereby suppressing ferroptosis. The absence of a comparable survival effect in oesophageal/gastric cancers is a key finding that strengthens the study. Despite similar alteration frequencies, these genes did not show prognostic significance, suggesting that their functional impact depends on the biological context of the tumour microenvironment. This aligns with the concept that pancreatic cancer exhibits a unique dependence on ferroptosis-related metabolic and stress-response pathways. Lyu et al., [11]. From a translational perspective, these findings suggest that ferroptosis suppressor genes may serve as prognostic markers and potential therapeutic targets in pancreatic cancer. While TP53 captures the dominant survival signal, genes such as ECH1, SIRT2, and ENPP2 may help identify high-risk subgroups and provide avenues for targeted intervention, particularly in strategies aimed at modulating ferroptosis and metabolic vulnerability. Liu et al., [10]. This study has several limitations. First, the analyses are based on publicly available retrospective datasets and therefore identify associations rather than direct cause–and–effect relationships. Second, survival analyses were performed without adjustment for additional clinical variables or key driver mutations such as KRAS, TP53, CDKN2A, and SMAD4, which may introduce potential confounding effects. Third, external validation in independent cohorts was not performed, and therefore, the prognostic findings should be considered exploratory. Fourth, the study did not include experimental validation; functional ferroptosis assays, rescue experiments, and gene perturbation studies were not conducted and should be addressed in future work. In addition, pathway enrichment analysis reflects the broader ferroptosis suppressor gene landscape rather than the shortlisted genes alone, and thus provides biological context rather than mechanistic proof. Furthermore, the analyses were based on bulk tumour datasets, and the cellular source of gene expression (tumour versus stromal compartments) could not be resolved, which may influence interpretation. Taken together, these limitations indicate that the findings should be interpreted as hypothesis-generating and require further validation through experimental studies and independent clinical datasets. In summary, this study demonstrates that ferroptosis suppressor gene alterations are associated with poor survival in pancreatic cancer and are embedded within key metabolic and stress-response pathways. The tumour-specific nature of these effects highlights the importance of context in interpreting molecular alterations and supports the potential of ferroptosis-targeted strategies in pancreatic cancer. Conclusion In this study, we systematically identified and characterised ferroptosis suppressor genes in pancreatic cancer and demonstrated their clinical and biological relevance through integrative genomic, survival, and pathway analyses. A subset of seven genes (TP53, HSF1, PARP10, ZFP36, SIRT2, ECH1, and ENPP2) was found to be recurrently altered and significantly associated with poor overall survival in pancreatic cancer. Importantly, the prognostic impact of these genes was not observed in oesophageal and gastric cancers, highlighting a tumour-specific dependency on ferroptosis-related mechanisms. This cross-cancer divergence underscores the importance of biological context in determining the clinical relevance of molecular alterations. Pathway enrichment analysis further revealed that ferroptosis suppressor genes are embedded within key regulatory networks involving metabolic reprogramming, autophagy, hypoxia response, and oncogenic signalling, which are central to pancreatic cancer progression. Integration of these findings suggests that ferroptosis suppression contributes to tumour survival by promoting metabolic adaptation and resistance to oxidative stress. Collectively, this study provides a comprehensive framework linking ferroptosis suppressor gene alterations to pancreatic cancer prognosis and underlying biological pathways. These findings highlight the potential of selected ferroptosis suppressor genes as prognostic indicators and potential therapeutic targets, and warrant further experimental validation to explore their role in ferroptosis-based treatment strategies. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgements We would like to acknowledge the Princess Nourah bint Abdulrahman University Researchers Supporting Project (PNURSP2026R341), Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia. The authors are also grateful to the Deanship of Research and Graduate Studies, King Khalid University, Abha, Saudi Arabia, for financially supporting this work through the Large Research Group Project under grant no. R.G.P.2/136/47. Author contributions N.K.S., M.S., and V.G. conceptualised and designed the study. N.K.S. and M. S. performed data acquisition and bioinformatics analysis and drafted the original manuscript. S.K.M., V.G., N.A., M.S., L. I. A and S.O.R. provided supervision and critical guidance throughout the study. F. A. and A.H. contributed to data interpretation and methodological validation. L. I. A and S.O.R. provided supervision. All authors reviewed, revised, and approved the final version of the manuscript. Funding This research was funded by the Princess Nourah bint Abdulrahman University Researchers Supporting Project (PNURSP2026R341), Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia. The authors also acknowledge the financial support provided by the Deanship of Research and Graduate Studies at King Khalid University, Abha, Saudi Arabia, through the Large Research Group Project under grant number R.G.P.2/136/47. Data availability No new experimental datasets were generated in this study. Publicly available datasets from FerrDb V2, cBioPortal and DAVID were analysed, as described in the Materials and Methods section. Declarations Ethics approval and consent to participate This study was entirely based on publicly available datasets. No human participants or animals were directly involved. Therefore, ethical approval and consent to participate were not required. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Nitin Kumar Singh and Maninder Singh have contributed equally and should be considered as first authors. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement No new experimental datasets were generated in this study. Publicly available datasets from FerrDb V2, cBioPortal and DAVID were analysed, as described in the Materials and Methods section.