Molecular and Structural Reprogramming of Gastric Cancer Revealed by Systems-Level Transcriptomic Analysis

Other names Location 1 H3C12 1.6 8356 H3 clustered histone 12 MCC (1), MNC (3), Degree (3) H3/j; H3C1; H3C2; H3C3; H3C4; H3C6; H3C7; H3C8; H3FJ; H3C10; H3C11; HIST1H3J 6p22.1 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 9 2 H3-4 1.19 8290 H3.4 histone, cluster member MCC (2), MNC (2), Degree (2) H3t; H3.4; H3/g; H3FT; H3C16; HIST3H3 1q42.13 3 PAX2 1.3 5076 paired box 2 MCC (3) FSGS7; PAPRS; PAX-2 10q24.31 4 HOXD11 1.6 3237 homeobox D11 MCC (4), DMNC (1) HOX4; HOX4F 2q31.1 5 HOXD13 1.4 3239 homeobox D13 MCC (5) BDE; SPD; BDSD; SPD1; HOX4I 2q31.1 6 PRKACG 1.6 5568 protein kinase cAMP-activated catalytic subunit gamma MNC (1), Degree (1) KAPG; PKACg; BDPLT19 9q21.11 7 FGF8 1.02 2253 fibroblast growth factor 8 MNC (4), Degree (4) HH6; AIGF; KAL6; FGF-8; HBGF-8 10q24.32 8 IFNG 1.03 3458 interferon gamma MNC (5), Degree (5) IFG; IFI; IMD69 12q15 9 OR8J1 3.7 219477 olfactory receptor family 8 subfamily J member 1 DMNC (2) OR11-183 11q12.1 10 OR5B21 2.22 219968 olfactory receptor family 5 subfamily B member 21 DMNC (3) - 11q12.1 11 HOXC12 3.06 3228 homeobox C12 DMNC (4) HOX3; HOC3F; HOX3F 12q13.13 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 10 12 IFNL2 3.02 282616 interferon lambda 2 DMNC (5) IL28A; IFNL2a; IFNL3a; IL-28A 19q13.2 144 A second, focused network built from the top 200 DEGs (100 up, 100 down) yielded 93 nodes and 145 318 edges (Fig 1) and highlighted 13 hubs (HOXC8, HOXC9, HOXA11, HOXC11, HOXA13, 146 WT1, ADIPOQ, GCG, VSTM2A, ATP4A, KCNE2, PTF1A) with both strong expression shifts 147 and network centrality, supporting their potential as diagnostic/prognostic biomarkers (Table 2). 148 This dual-network strategy separates broadly actionable targets from clinically tractable 149 biomarkers by integrating systems-level topology with differential expression magnitude. 150 Table 2. Hub genes from the PPI network of the top 200 DEGs, including ranking methods and 151 log2 fold change (logFC) values. # Node LogFC Gene ID Gene Description Rank Ranking

Other names Location 1 H3C12 1.6 8356 H3 clustered histone 12 1, 1, 1 MCC, MNC, Degree H3/j; H3C1; H3C2; H3C3; H3C4; H3C6; H3C7; H3C8; H3FJ; H3C10; H3C11; HIST1H3J 6p22.1 2 HOXC8 1.6 3224 homeobox C8 2 MCC HOX3; HOX3A 12q13.13 3 HOXC9 2.1 3225 homeobox C9 3 MCC HOX3; HOX3B 12q13.13 4 HOXA11 2.2 3207 homeobox A11 4 MCC HOX1; HOX1I; RUSAT1 7p15.2 5 HOXC11 1.7 3227 homeobox C11 5, 1 MCC, DMNC HOX3H 12q13.13 6 HOXA13 1.5 3209 homeobox A13 2 DMNC HOX1; HOX1J 7p15.2 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 11 7 PTF1A -1.2 256297 pancreas associated transcription factor 1a 3 DMNC p48; PACA; PAGEN2; bHLHa29; PTF1- p48 10p12.2 8 VSTM2A -1.2 222008 V-set and transmembrane domain containing 2A 4 DMNC VSTM2 7p11.2 9 KCNE2 -1.2 9992 potassium voltage- gated channel subfamily E regulatory subunit 2 5 DMNC LQT5; LQT6; ATFB4; MIRP1 21q22.11 10 GCG -1.2 2641 glucagon 2, 2 MNC, Degree GLP1; GLP2; GRPP; GLP-1 2q24.2 11 WT1 1.4 7490 WT1 transcription factor 3, 4 MNC, Degree GUD; AWT1; WAGR; WT-1; WT33; NPHS4; WIT-2 11p13 12 ADIPOQ -1.2 9370 adiponectin, C1Q and collagen domain containing 3, 5 MNC, Degree ACDC; ADPN; APM1; APM-1; GBP28; ACRP30; ADIPQTL1 3q27.3 13 ATP4A -1.2 495 ATPase H+/K+ transporting subunit alpha 3, 3 MNC, Degree ATP6A 19q13.12 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 12 152 3.3. Functional enrichment analysis 153 Functional enrichment analysis was performed to understand the biological role of identified up- 154 and downregulated genes. GO terms and pathway associations were analyzed in two sets 155 separately to disclose different functional profiles of both gene sets (Fig 2 and 3). 156 Upregulated genes (n = 85) were enriched for embryonic/developmental programs and immune 157 differentiation (e.g., morphogenesis terms and myeloid dendritic cell activation), along with 158 transcriptional regulation (sequence-specific DNA binding) and multiple solute/anion transport 159 activities. Reactome analysis similarly highlighted organic anion and vitamin/nucleoside transport, 160 kidney developmental pathways, and drug-metabolism modules (e.g., ciprofloxacin/atorvastatin 161 ADME, bile acid metabolism), supporting a phenotype of developmental reactivation, 162 transcriptional control, and altered transport/metabolic capacity. 163 Downregulated genes were predominantly enriched for ion homeostasis and membrane transport, 164 particularly sodium/potassium handling (transmembrane transport, export/import, membrane 165 repolarization), and localized to vesicular/endosomal–Golgi compartments and transport 166 complexes (e.g., clathrin-coated vesicles, Na⁺/K⁺ -ATPase complexes). Functional terms also 167 indicated reduced channel and transporter activities, with Reactome showing decreased innate 168 immune defense (defensins/antimicrobial peptides), aquaporin and ion channel transport, and 169 neuroendocrine signaling pathways (acetylcholine release, incretin/glucagon signaling), consistent 170 with loss of normal physiological and immune functions. 171 Overall, GC displayed a shift toward developmental reprogramming, transcriptional activation, 172 and altered solute/drug metabolism (upregulated genes), alongside suppression of epithelial 173 ion/transport homeostasis and innate immunity (downregulated genes), underscoring tissue 174 remodeling and functional derangement relevant to biomarker and target discovery. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 13 175 3.4. Cluster analysis and reactome pathway enrichment 176 We further enriched the top four clusters obtained from IPCA into Reactome pathways to gain 177 deeper insights into the modular organization of the gene network and to identify distinct 178 biological themes. These clusters were mainly enriched for FGFR signaling and its canonical 179 downstream pathways, developmental biology, and transcriptional regulation, showing that they 180 are biologically coherent and functionally specialized (Fig 4, and Table 3). 181 Table 3. Classification of Reactome Pathways Enriched Across Top Four Clusters. Category Reactome Pathway Clusters FGFR Signaling FGFR1 Ligand Binding and Activation 1, 2, 3 FGFR1c Ligand Binding and Activation 4 FGFR2 Ligand Binding and Activation 1, 2, 3 FGFR2c Ligand Binding and Activation 4 FGFR3 Ligand Binding and Activation 4 FGFR3b Ligand Binding and Activation 4 FGFR3c Ligand Binding and Activation 4 FGFR4 Ligand Binding and Activation 4 FGFR2 Mutant Receptor Activation 1, 2, 3 FGFR3 Mutant Receptor Activation 4 Activated Point Mutants of FGFR2 1, 2, 3 Activated Point Mutants of FGFR1 4 Activated Point Mutants of FGFR3 4 Signaling by FGFR2 in Disease 2, 3 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 14 FGFRL1 Modulation of FGFR1 Signaling 4 Negative Regulation of FGFR1 Signaling 1, 2, 3 Negative Regulation of FGFR2 Signaling 1, 2, 3 Canonical Signaling Cascades SHC-mediated Cascade FGFR1 1, 2, 3 SHC-mediated Cascade FGFR2 1, 2, 3 FRS-mediated FGFR1 Signaling 1, 2, 3 FRS-mediated FGFR2 Signaling 1, 2, 3 Phospholipase C-mediated Cascade FGFR1 1, 2, 3 Phospholipase C-mediated Cascade; FGFR2 1, 2, 3 Phospholipase C-mediated Cascade; FGFR3 4 PI-3K Cascade FGFR1 1, 2, 3 PI-3K Cascade FGFR2 1, 2, 3 PI3K Cascade 2, 3 IRS-mediated Signalling 3 Downstream Signaling of Activated FGFR1 1, 2, 3 Downstream Signaling of Activated FGFR2 1, 2, 3 Developmental and Organogenesis Developmental Biology 1, 2, 4 Gastrulation 1, 2, 4 Kidney Development 1 Nephron Development 4 Formation of the Ureteric Bud 1 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 15 Formation of Intermediate Mesoderm 4 Formation of the Anterior Neural Plate 4 Formation of the Posterior Neural Plate 4 Specification of the Neural Plate Border 4 Transcriptional Regulation Regulation of Gene Expression in Early Pancreatic Precursor Cells 4 Regulation of Expression of SLITs and ROBOs 4 Matrix Remodeling and Invasion Activation of Matrix Metalloproteinases 3 182 183 Across all clusters, FGFR signaling emerged as a consistent hallmark, including ligand-driven 184 activation of FGFR1–4 and isoforms (FGFR1c/2c/3b/3c), alongside evidence of oncogenic 185 FGFR2 mutant activation and regulatory counterbalance via negative modulators (e.g., FGFRL1). 186 Enrichment extended to major downstream cascades, SHC/FRS-mediated signaling, PLC, and 187 PI3K/IRS pathways, implicating proliferation, survival, motility, and metabolic regulation. 188 Developmental programs were also reactivated (e.g., gastrulation, kidney/nephron and ureteric bud 189 development, neural plate formation), suggesting increased cellular plasticity and possible EMT- 190 related microenvironmental shifts. Additional cluster-specific themes included disrupted 191 transcriptional control (early pancreatic precursor programs; SLIT/ROBO regulation) and matrix 192 remodeling via metalloproteinase activation. Overall, the IPCA clusters define a coherent network 193 in which FGFR signaling links developmental reprogramming, intracellular signaling, 194 transcriptional regulation, and invasive behavior in gastric cancer. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 16 195 3.5. Analysis of tumor grade-specific gene expression and potential biomarker utility 196 Gene expression analysis across tumor grades in stomach adenocarcinoma (STAD) uncovered a 197 dynamic transcriptional landscape, with several genes exhibiting significant and grade-specific 198 expression changes (Fig 5). These alterations offer insight into molecular mechanisms underlying 199 tumor progression and differentiation, and highlight potential diagnostic and prognostic 200 biomarkers. 201 Grade-stratified analysis showed increasing expression of 202 HOXA11/HOXA13/HOXC8/HOXC9/HOXC11 and HIST1H3J from Grade 1 to Grades 2–3, 203 consistent with developmental/epigenetic reprogramming. In contrast, gastric lineage markers 204 (ATP4A, KCNE2, PTF1A, VSTM2A) were strongly downregulated early and remained low 205 across grades, indicating sustained loss of epithelial identity. WT1 rose mainly in Grades 2–3, 206 while ADIPOQ was repressed early with modest recovery and GCG declined gradually. Together, 207 these patterns support a panel where HOX/HIST1H3J (±WT1) reflect progression, and 208 ATP4A/KCNE2/PTF1A/VSTM2A (±ADIPOQ) mark early dedifferentiation. 209 3.6. Expression of genes in STAD based on nodal metastasis status and biomarker potential 210 When TCGA-STAD samples were examined according to nodal metastasis stage (N0–N3), a clear 211 and recurring alteration appeared among the 13 studied genes, pointing to their possible value as 212 biomarkers for identifying and tracking disease (Fig 6). 213 Early Detection and Diagnostic Biomarkers: The HOX cluster genes (HOXA11, HOXA13, 214 HOXC8, HOXC9, and HOXC11) together with HIST1H3J, showed marked and statistically 215 strong overexpression across every nodal category when compared with normal gastric tissue 216 (most with P ≤ 10⁻ ⁹). The elevated expression was already present in node-negative tumors, 217 pointing to early engagement of these transcription factors in tumor development rather than .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 17 218 secondary activation during metastasis. Seen in this light, their persistent activity may serve as an 219 early warning sign for primary tumors that have yet to reach the lymphatic system. 220 Progression and Metastatic Biomarkers: While the expression of HOX genes and HIST1H3J did 221 not show a progressive increase with higher nodal stages (N1–N3), certain genes demonstrated 222 patterns suggestive of progression markers. Expression of WT1 rose sharply in N0 and N1 tumors 223 (P around 10⁻⁶–10⁻⁷) and climbed even higher by the N3 stage, which may point to its role in 224 promoting or tracking advanced nodal spread. The smaller but steady changes seen for KCNE2 225 and PTF1A across the same stages might reflect the gradual erosion of differentiation as tumors 226 become more metastatic, an observation that deserves closer study for its potential prognostic 227 value. 228 Loss of Differentiation Markers: Gastric lineage genes (ATP4A, KCNE2, PTF1A, VSTM2A) and 229 ADIPOQ were markedly downregulated in both node-negative and metastatic tumors (all P < 0.01 230 vs normal), indicating early, sustained dedifferentiation and supporting their use as negative 231 diagnostic markers independent of nodal status. GCG showed no nodal-group differences and can 232 be excluded from the core biomarker set. Overall, the data support a two-tier framework: early 233 activation markers (HOX-cluster genes, HIST1H3J), persistent loss-of-identity markers (ATP4A, 234 KCNE2, PTF1A, VSTM2A, ADIPOQ), and WT1 as a potential progression marker associated 235 with nodal spread. 236 3.7. Tumor stage–specific gene expression and implications for biomarker discovery in STAD 237 As we looked across tumor stages in STAD, certain genes stood out for how sharply their 238 expression shifted; patterns that could eventually guide biomarker discovery and treatment design. 239 Several genes, among them HOXA11, HOXA13, HOXC8, HOXC9, HOXC11, HIST1H3J, WT1, .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 18 240 and the differentiation markers ATP4A, KCNE2, VSTM2A, and PTF1A, showed stage-specific 241 shifts in activity, suggesting value for classifying tumors or refining clinical decisions (Fig 7). 242 Across STAD stages, HIST1H3J and multiple HOX-cluster genes are consistently upregulated 243 (often P < 10⁻¹²), with early onset and sustained expression, suggesting roles in tumor 244 establishment and maintenance and potential utility as early molecular markers. In contrast, gastric 245 differentiation genes ATP4A, KCNE2, VSTM2A, and PTF1A are persistently downregulated 246 across all stages versus normal tissue, indicating stable loss of parietal/ductal identity and 247 supporting their use as diagnostic markers of malignant transformation. WT1 increases with stage, 248 reaching stronger significance in advanced disease, consistent with a progression-associated 249 biomarker. Together, these patterns define a stable STAD signature, early HOX/HIST1H3J 250 activation with sustained repression of gastric identity genes, while WT1 may aid risk stratification 251 and monitoring. 252 3.8. Survival analysis of candidate genes in STAD 253 We used Kaplan–Meier analysis to examine how the 13 candidate genes relate to patient survival 254 in STAD. Of all the genes tested, only ADIPOQ showed a significant link with overall survival (p 255 = 0.012); the others showed no clear association (p > 0.05) (Fig 8). 256 Patients were stratified by ADIPOQ expression (high, n = 100 vs low/intermediate, n = 292). 257 Kaplan–Meier analysis showed poorer survival in the high-expression group, suggesting ADIPOQ 258 as a negative prognostic marker in STAD. As adiponectin modulates metabolism and 259 inflammation, this pattern supports a context-dependent role in the tumor microenvironment. 260 Clinically, ADIPOQ may aid risk stratification, but independent cohort validation and mechanistic 261 studies are needed. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 19 262 4. Discussion 263 In this work, we took a broad look at the transcriptome of STAD to map gene expression changes 264 and their biological and clinical implications. By combining a full set of roughly 6,500 DEGs with 265 a focused analysis of the top 200, two clear interaction modules began to emerge from the data. 266 The first includes histone genes, immune regulators, and kinases that occupy central positions in 267 the protein–protein interaction network, making them attractive targets for therapeutic disruption. 268 The second group includes transcription factors, epithelial genes, and metabolic regulators that 269 show striking expression differences. These features make them appealing candidates for 270 developing both diagnostic and prognostic markers. 271 We found that HOXA11, HOXA13, HOXC8, HOXC9, and HOXC11 were consistently more 272 active across tumor grades, stages, and nodal groups, a result that fits well with earlier evidence of 273 HOX gene reactivation in gastrointestinal cancers [20]. Their expression tended to climb as tumors 274 became less differentiated and more advanced, most noticeably in the aggressive forms. This 275 steady rise links them to both loss of cellular identity and increasing malignancy. The histone gene 276 HIST1H3J showed a comparable pattern, its levels rose step by step with tumor grade and stage, 277 which aligns with reports of broad epigenetic remodeling [21]. These genes seem to switch on 278 early in tumor formation and remain active as the disease develops, helping to sustain 279 transcriptional activity that supports growth and survival. 280 Conversely, the observed downregulation of gastric lineage markers (ATP4A, KCNE2, VSTM2A, 281 PTF1A) from early stages supports the idea that dedifferentiation is an initiating event in gastric 282 tumorigenesis [22, 23]. Our observations agree with earlier reports showing that genes tied to acid 283 secretion and epithelial differentiation are already dampened in the early stages of gastric cancer. 284 In particular, several studies have noted reduced expression of ATP4A and ATP4B, both of which .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 20 285 have been examined as potential diagnostic markers [22]. ADIPOQ also showed dynamic 286 regulation, with an early decrease followed by partial reactivation in advanced tumors; recent work 287 suggests that low adiponectin levels or expression correlate with more aggressive GC features 288 [24]. Survival analysis revealed that high ADIPOQ expression correlates with poorer patient 289 outcomes, further highlighting its potential as a context-dependent prognostic biomarker [24]. 290 Our enrichment analysis pointed to a set of molecular patterns that mirror what other 291 transcriptomic studies have described in gastric cancer. The genes showing higher expression were 292 mostly involved in transcriptional control, immune signaling, and developmental processes, 293 similar to earlier reports describing a reactivation of embryonic and inflammatory pathways during 294 tumor growth [25]. By contrast, the genes showing reduced activity were mostly linked to ion 295 transport, vesicle movement, and antimicrobial defense. Their decline mirrors the loss of normal 296 gastric cell function that tends to accompany malignant transformation [26]. Overall, the data 297 suggest a broad reorganization of the tumor transcriptome, shifting away from the physiological 298 roles of the stomach toward a more adaptable, development-like and metabolically flexible state, 299 a trend that recent single-cell and spatial transcriptomic analyses of gastric and gastrointestinal 300 cancers have also captured [25, 26]. Our cluster-based pathway enrichment further emphasized the 301 centrality of FGFR signaling and its downstream cascades. This finding is in agreement with recent 302 clinical and preclinical studies demonstrating FGFR dysregulation as a driver of oncogenic 303 transcriptional programs in gastric cancer [27, 28]. 304 A key strength of this study is its layered design, combining stringent DEG filtering with network 305 modeling, functional annotation, and clinicopathological interpretation to identify clinically 306 relevant targets. Our findings are consistent with recent bulk and single-cell GC studies reporting 307 metabolic reprogramming, immune microenvironment changes, and loss of gastric differentiation .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 21 308 [7, 29]. Limitations include reliance on TCGA-STAD only, lack of experimental validation, and 309 the inability of transcriptomics to capture post-transcriptional/post-translational regulation; future 310 work should validate results in independent cohorts and integrate additional omics layers. 311 Going forward, it will be important to check how well these biomarkers perform in other patient 312 cohorts and to see whether they can be tracked in easier-to-obtain materials like blood or gastric 313 fluid. Testing how HOX genes, WT1, and FGFR-pathway effectors behave when perturbed in cells 314 or organoids should also help clarify what roles they actually play in tumor biology. Additionally, 315 the relationship between ADIPOQ expression and the immune microenvironment warrants 316 investigation to decipher its dual role in metabolism and immunoregulation. 317 In summary, our study delineates two functionally distinct gene modules in gastric cancer: a 318 regulatory core enriched in HOX and histone genes that may be exploited for early diagnosis and 319 a set of suppressed differentiation genes marking the loss of gastric identity. These results outline 320 a molecular framework that could guide biomarker-based patient classification and help shape new 321 therapeutic strategies in STAD. They also highlight how combining network-level analysis with 322 clinical data can move the field closer to truly precise cancer care. 323 5. Conclusion 324 Using an integrated transcriptomic and network-based framework (differential expression, PPI 325 mapping, hub ranking, and pathway enrichment), we identified a coordinated STAD signature 326 marked by upregulation of developmental/immune programs and repression of epithelial 327 differentiation and transport genes. Dual-network analysis prioritized a focused biomarker set— 328 HOX genes, HIST1H3J, ATP4A, KCNE2, and PTF1A—with potential utility for early detection, 329 monitoring, and risk stratification, and highlighted actionable pathways relevant to precision .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 22 330 oncology. Key limitations are reliance on TCGA-STAD data and lack of experimental validation; 331 future work should validate these markers in independent cohorts and test mechanisms in vitro/in 332 vivo, ideally integrating proteomic/epigenomic and spatial transcriptomic layers to refine 333 heterogeneity and clinical translation. 334 Abbreviations 335 BP: Biological Process; CC: Cellular Component; DEG: Differentially Expressed Gene; FGFR: 336 Fibroblast Growth Factor Receptor; GC: Gastric Cancer; GO: Gene Ontology; HOX: Homeobox 337 Gene; KEGG: Kyoto Encyclopedia of Genes and Genomes; MF: Molecular Function; PPI: 338 Protein–Protein Interaction; RNA-seq: RNA Sequencing; STAD: Stomach Adenocarcinoma; 339 TCGA: The Cancer Genome Atlas; UALCAN: University of Alabama at Birmingham Cancer 340 Data Analysis Portal. 341 Declarations 342 Funding 343 The authors received no specific funding for this work. 344 Ethics statement 345 This study used only publicly available, de-identified data and did not require ethics approval. 346 Consent for publication 347 Not applicable. 348 Data Availability 349 TCGA-STAD RNA-seq and clinical data are publicly available from the Genomic Data Commons 350 (GDC). All processed data supporting the findings of this study are available at GitHub .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 23 351 (https://github.com/negmot/tcga-stad-network-analysis) and have been archived on Zenodo (DOI: 352 10.5281/zenodo.18537980). 353 Competing interests 354 The authors have no competing interests to declare that are relevant to the content of this article. 355 Author contributions 356 N.M.D.: Conceptualization, Supervision, Data curation, Formal analysis, Visualization, Writing- 357 original draft. 358 M.S.R.R.: Data curation, Investigation, Writing-review & editing. 359 Both authors: Final approval of the manuscript. 360 Supporting information 361 Supplementary File 1. Full differential expression results tables (RAR/XLSX). 362 References 363 1. Chen Y, Jia K, Xie Y, Yuan J, Liu D, Jiang L, et al. The current landscape of gastric 364 cancer and gastroesophageal junction cancer diagnosis and treatment in China: a comprehensive 365 nationwide cohort analysis. J Hematol Oncol. 2025;18(1):42. doi: 10.1186/s13045-025-01698-y. 366 2. Mousavi SE, Ilaghi M, Elahi Vahed I, Nejadghaderi SA. Epidemiology and 367 socioeconomic correlates of gastric cancer in Asia: results from the GLOBOCAN 2020 data and 368 projections from 2020 to 2040. Sci Rep. 2025;15(1):6529. doi: 10.1038/s41598-025-90064-6. 369 3. Lin J-L, Lin J-X, Lin G-T, Huang C-M, Zheng C-H, Xie J-W, et al. Global incidence and 370 mortality trends of gastric cancer and predicted mortality of gastric cancer by 2035. BMC Public 371 Health. 2024;24(1):1763. doi: 10.1186/s12889-024-19104-6. 372 4. Shin WS, Xie F, Chen B, Yu P, Yu J, To KF, et al. Updated Epidemiology of Gastric 373 Cancer in Asia: Decreased Incidence but Still a Big Challenge. Cancers (Basel). 374 2023;15(9):2639. doi: 10.3390/cancers15092639. 375 5. Zhang Z, Wang J, Song N, Shi L, Du J. The global, regional, and national burden of 376 stomach cancer among adolescents and young adults in 204 countries and territories, 1990–2019: 377 A population-based study. Front Public Health. 2023;11:1079248. doi: 378 10.3389/fpubh.2023.1079248. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 24 379 6. Zhang H, Yang W, Tan X, He W, Zhao L, Liu H, et al. Long-term relative survival of 380 patients with gastric cancer from a large-scale cohort: a period-analysis. BMC Cancer. 381 2024;24(1):1420. doi: 10.1186/s12885-024-13141-5. 382 7. Lin X, Yang P, Wang M, Huang X, Wang B, Chen C, et al. Dissecting gastric cancer 383 heterogeneity and exploring therapeutic strategies using bulk and single-cell transcriptomic 384 analysis and experimental validation of tumor microenvironment and metabolic interplay. Front 385 Pharmacol. 2024;15:1355269. doi: 10.3389/fphar.2024.1355269. 386 8. Mottaghi-Dastjerdi N, Ghorbani A, Montazeri H, Guzzi PH. A systems biology approach 387 to pathogenesis of gastric cancer: gene network modeling and pathway analysis. BMC 388 Gastroenterol. 2023;23(1):248. doi: 10.1186/s12876-023-02891-4. 389 9. Khoshdel F, Mottaghi-Dastjerdi N, Yazdani F, Salehi S, Ghorbani A, Montazeri H, et al. 390 CTGF, FN1, IL-6, THBS1, and WISP1 genes and PI3K-Akt signaling pathway as prognostic and 391 therapeutic targets in gastric cancer identified by gene network modeling. Discov Oncol. 392 2024;15(1):344. doi: 10.1007/s12672-024-01225-4. 393 10. Rahimi-Farsi N, Shahbazi T, Ghorbani A, Mottaghi-Dastjerdi N, Yazdani F, Mohseni P, 394 et al. Network-based analysis of candidate oncogenes and pathways in hepatocellular carcinoma. 395 Biochem Biophys Rep. 2025;43:102086. doi: 10.1016/j.bbrep.2025.102086. 396 11. Alvani A, Mottaghi-Dastjerdi N, Gholami A, Ghorbani A, Pazhoohesh Z, Pajdam M, et 397 al. Identification of indirect pathways enhancing the biocompatibility of DOX/GO/Fe3O4 398 nanomaterials in Glioblastoma: Gene network modeling and pathway analysis. Biochem Biophys 399 Res Commun. 2025;773:152077. doi: 10.1016/j.bbrc.2025.152077. 400 12. Rahimi-Farsi N, Ghorbani A, Mottaghi-Dastjerdi N, Shahbazi T, Bostanian F, Mohseni 401 P, et al. Comprehensive system biology analysis of microRNA-101-3p regulatory network 402 identifies crucial genes and pathways in hepatocellular carcinoma. J Genet Eng Biotechnol. 403 2025;23(1):100471. doi: 10.1016/j.jgeb.2025.100471. 404 13. Rahimi H, Farzadifar E, Mottaghi-Dastjerdi N, Ghorbani A, Amerizadeh F, Pasdar A. 405 Integrative Systems Biology Analysis of MicroRNA Regulation in Wnt Signaling Pathway 406 During Breast Cancer Progression. Int J Cancer Manag. 2025;18(18):e164192. doi: 407 10.5812/ijcm-160657. 408 14. Saeidi M, Pasdar A, Rahmani F, Ghorbani A, Mottaghi N, Amerizadeh F. A Prospective 409 to Regulatory Role of miRNAs on Wnt/β-catenin Signaling and Its Crosstalk to the Other 410 Cellular Pathways in Tumorigenesis of Glioblastoma by a Systems Biology Approach. Int J 411 Cancer Manag. 2025;18(1):e156834. doi: 10.5812/ijcm-156834. 412 15. Yazdani F, Mottaghi-Dastjerdi N, Shahbazi B, Ahmadi K, Ghorbani A, Soltany-Rezaee- 413 Rad M, et al. Identification of key genes and pathways involved in T-DM1-resistance in OE-19 414 esophageal cancer cells through bioinformatics analysis. Heliyon. 2024;10(18):e37451. doi: 415 10.1016/j.heliyon.2024.e37451. 416 16. Soltany-Rezaee-Rad M, Mottaghi-Dastjerdi N, Setayesh N, Roshandel G, Ebrahimifard 417 F, Sepehrizadeh Z. Overexpression of FOXO3, MYD88, and GAPDH identified by suppression 418 subtractive hybridization in esophageal cancer is associated with autophagy. Gastroenterol Res 419 Pract. 2014;2014(1):185035. doi: 10.1155/2014/185035. 420 17. Mottaghi-Dastjerdi N, Soltany-Rezaee-Rad M, Sepehrizadeh Z, Roshandel G, 421 Ebrahimifard F, Setayesh N. Genome expression analysis by suppression subtractive 422 hybridization identified overexpression of Humanin, a target gene in gastric cancer 423 chemoresistance. DARU J Pharm Sci. 2014;22(1):1-7. doi: 10.1186/2008-2231-22-14. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 25 424 18. Mottaghi-Dastjerdi N, Soltany-Rezaee-Rad M, Sepehrizadeh Z, Roshandel G, 425 Ebrahimifard F, Setayesh N. Identification of novel genes involved in gastric carcinogenesis by 426 suppression subtractive hybridization. Hum Exp Toxicol. 2015;34(1):3-11. doi: 427 10.1177/0960327114532386. 428 19. Li M, Li D, Tang Y, Wu F, Wang J. CytoCluster: a cytoscape plugin for cluster analysis 429 and visualization of biological networks. Int J Mol Sci. 2017;18(9):1880. doi: 430 10.3390/ijms18091880. 431 20. Li Z, Lu T, Chen Z, Yu X, Wang L, Shen G, et al. HOXA11 promotes lymphatic 432 metastasis of gastric cancer via transcriptional activation of TGFβ1. iScience. 433 2023;26(8):107346. doi: 10.1016/j.isci.2023.107346. 434 21. Wang Y, Liu H, Zhang M, Xu J, Zheng L, Liu P, et al. Epigenetic reprogramming in 435 gastrointestinal cancer: biology and translational perspectives. MedComm. 2024;5(9):e670. doi: 436 10.1002/mco2.670. 437 22. Chen Q, Wang Y, Liu Y, Xi B. ESRRG, ATP4A, and ATP4B as Diagnostic Biomarkers 438 for Gastric Cancer: A Bioinformatic Analysis Based on Machine Learning. Front Physiol. 439 2022;13:905523. doi: 10.3389/fphys.2022.905523. 440 23. Akhtar A, Hameed Y, Ejaz S, Abdullah I. Identification of gastric cancer biomarkers 441 through in-silico analysis of microarray based datasets. Biochem Biophys Rep. 2024;40:101880. 442 doi: 10.1016/j.bbrep.2024.101880. 443 24. Cheng M, Hajime O, Yan S, Quan Y, Caroline Nadia F, Wu Y, et al. The role of 444 adiponectin in gastric cancer. J Cancer Metastasis Treat. 2023;9:33. doi: 10.20517/2394- 445 4722.2023.23. 446 25. Sun Y, Nie W, Xiahou Z, Wang X, Liu W, Liu Z, et al. Integrative single-cell and spatial 447 transcriptomics uncover ELK4-mediated mechanisms in NDUFAB1+ tumor cells driving gastric 448 cancer progression, metabolic reprogramming, and immune evasion. Front Immunol. 449 2025;16:1591123. doi: 10.3389/fimmu.2025.1591123. 450 26. Chen H, Jing C, Shang L, Zhu X, Zhang R, Liu Y, et al. Molecular characterization and 451 clinical relevance of metabolic signature subtypes in gastric cancer. Cell Rep. 452 2024;43(7):114424. doi: 10.1016/j.celrep.2024.114424. 453 27. Lau DK, Collin JP, Mariadason JM. Clinical developments and challenges in treating 454 FGFR2-driven gastric cancer. Biomedicines. 2024;12(5):1117. doi: 455 10.3390/biomedicines12051117. 456 28. Edirisinghe O, Ternier G, Alraawi Z, Suresh Kumar TK. Decoding FGF/FGFR 457 Signaling: Insights into Biological Functions and Disease Relevance. Biomolecules. 458 2024;14(12):1622. doi: 10.3390/biom14121622. 459 29. Cai X, Yang J, Guo Y, Yu Y, Zheng C, Dai X. Re-analysis of single cell and spatial 460 transcriptomics data reveals B cell landscape in gastric cancer microenvironment and its 461 potential crosstalk with tumor cells for clinical prognosis. J Transl Med. 2024;22(1):807. doi: 462 10.1186/s12967-024-05606-9. 463 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint 26 464 Figure Legends 465 Fig 1. PPI network of the top 200 DEGs, with nodes colored by expression (red: upregulated, 466 blue: downregulated). Constructed using STRING in Cytoscape (confidence score: 0.15). 467 Fig 2. Functional enrichment analysis of upregulated genes. Top GO terms (BP, CC, MF) and 468 Reactome pathways enriched among 85 upregulated GC genes, ranked by -Log 10FDR using 469 ENRICHR. 470 Fig 3. Functional enrichment of downregulated genes. Top GO terms and Reactome pathways for 471 94 downregulated GC genes, ranked by -Log 10FDR using ENRICHR. 472 Fig 4. Heatmap of Reactome pathway enrichment across top-ranked clusters. The presence of a 473 pathway in a given cluster is indicated in blue, and absence is shown in white. Data are based on 474 the top 20 enriched Reactome pathways for each of the top four PPI network clusters identified 475 using the IPCA algorithm in CytoCluster. 476 Fig 5. Grade-specific expression of 13 candidate genes in GC. Boxplots show transcript per 477 million (TPM) values across normal and cancer grades 1-3 (TCGA data via UALCAN). 478 Fig 6. Transcript expression levels of selected genes in GC patients based on nodal metastasis 479 status. 480 Fig 7. Stage-specific expression of 13 candidate genes in GC. Boxplots show transcript per 481 million (TPM) values across normal tissue and cancer stages 1-4 (TCGA data via UALCAN). 482 Fig 8. Survival analysis of ADIPOQ in GC patients. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 19, 2026. ; https://doi.org/10.64898/2026.02.18.706548doi: bioRxiv preprint