Metabolic Reprogramming Across Molecular Subtypes of Gastric Cancer

Metabolic Reprogramming Across Molecular Subtypes of Gastric Cancer | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Metabolic Reprogramming Across Molecular Subtypes of Gastric Cancer Cenk ARAL This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9147848/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Purpose Gastric cancer molecular subtypes defined by The Cancer Genome Atlas — chromosomal instability (CIN), microsatellite instability (MSI), genomically stable (GS), and Epstein-Barr virus-positive (EBV) — carry distinct biological features, yet their metabolic programmes and clinical significance remain uncharacterised across all four subtypes simultaneously. Methods Gene Set Variation Analysis (GSVA) was applied to RNA-sequencing data from 383 TCGA gastric adenocarcinomas to quantify activity of six core metabolic pathways: glycolysis, oxidative phosphorylation, fatty acid oxidation (FAO), the pentose phosphate pathway (PPP), glutamine metabolism, and lactate metabolism. Subtype-specific survival associations were assessed by continuous Cox proportional hazards regression with multivariable adjustment. Transcriptional regulatory mechanisms were investigated through expression analysis of 20 metabolic transcription factors. External validation was performed in the independent ACRG cohort (GSE62254; n=300). Results All six pathways showed significant subtype stratification (Kruskal-Wallis, all FDR<0.001). Within GS tumours, elevated FAO activity was independently associated with worse overall survival after multivariable adjustment (HR=7.261, 95% CI: 1.604–32.865, P=0.010). Within MSI tumours, elevated PPP (HR=0.211, P=0.014) and glutamine pathway activity (HR=0.169, P=0.019) were associated with improved survival, potentially reflecting antioxidant support of anti-tumour immunity. PRKAA2, SREBF1, and MYC were identified as key subtype-specific transcriptional regulators. Cross-platform validation confirmed direction preservation of metabolic pathway coordination in 13 of 15 pathway pairs in the ACRG cohort. Conclusions Gastric cancer molecular subtypes exhibit distinct metabolic vulnerabilities with independent prognostic significance. The GS-FAO association identifies a high-risk metabolic phenotype amenable to CPT1 inhibitor evaluation, while MSI biosynthetic pathway elevation may underpin immune checkpoint sensitivity. Stomach Neoplasms Metabolic Reprogramming Gene Expression Profiling Fatty Acid Oxidation Microsatellite Instability Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 INTRODUCTION Gastric cancer is the 5th most common malignancy and leading cause of cancer-related mortality worldwide [1]. Despite the noteworthy advances in surgical techniques, perioperative chemotherapy, and targeted immunotherapy, five-year overall survival rates remain below 30% in most populations [2, 3]. The molecular characterization of gastric adenocarcinoma by The Cancer Genome Atlas (TCGA) Research Network identified four biologically distinct subtypes: chromosomal instability (CIN), microsatellite instability (MSI), genomically stable (GS), and Epstein-Barr virus-positive (EBV) [4]. This molecular taxonomy has substantially informed our understanding of gastric cancer biology; however, it has not yet translated into meaningful therapeutic advances targeting these molecular features, underscoring the need for orthogonal axes of tumor stratification that can identify clinically actionable vulnerabilities. Metabolic reprogramming has emerged as a fundamental hallmark of malignancy, enabling cancer cells to meet the bioenergetic and biosynthetic demands of uncontrolled proliferation [5]. Cancer metabolism now extends far beyond glycolytic reprogramming to encompass oxidative phosphorylation (OXPHOS), fatty acid oxidation (FAO), pentose phosphate pathway activation, glutamine addiction, and lactate metabolism. These metabolic adaptations are not uniform across cancer types or even within individual tumors; rather, metabolic phenotypes exhibit substantial inter- and intra-tumor heterogeneity shaped by the oncogenotype, microenvironmental factors, and tissue of origin [6]. This complexity poses a significant challenge to the development of effective metabolic therapies but simultaneously reveals a rich landscape of subtype-specific vulnerabilities amenable to precision targeting. As demonstrated in other solid tumors, molecular subtypes align with distinct metabolic dependencies that carry prognostic significance: in colorectal cancer, the metabolic CMS3 subtype results in marked dysregulation of glycolytic and biosynthetic pathways [7], whereas in breast cancer, HER2-positive and triple-negative subtypes present divergent lipid and glutamine metabolic profiles associated with clinical outcomes [8]. These observations establish a precedent for integrating metabolic profiling with genomic classification to reveal subtype-specific vulnerabilities. In gastric cancer, individual metabolic pathways have been studied in isolation — glycolytic reprogramming linked to HIF-1alpha upregulation, OXPHOS alterations in intestinal-type tumors, and lipid metabolic reprogramming associated with chemoresistance — but none of these studies integrated multiple pathways across all four TCGA molecular subtypes simultaneously. While several recent studies have defined metabolic subtypes in gastric cancer using GSVA-based pathway scoring [9-12], none have mapped these findings directly onto the established four-class TCGA genomic taxonomy (CIN/MSI/GS/EBV), assessed subtype-specific metabolic survival associations with multivariable adjustment and bootstrap validation, or examined the transcriptional regulatory architecture coordinating these phenotypes. The relationship between TCGA molecular classification and metabolic pathway activity thus remains uncharacterised. To address this gap, a comprehensive analysis of metabolic reprogramming across gastric cancer molecular subtypes was performed via RNA-sequencing data from 383 TCGA gastric adenocarcinomas, with external validation in 300 patients from the ACRG cohort (GSE62254). Gene set variation analysis (GSVA) was applied to quantify the activity of six core metabolic pathways — glycolysis, oxidative phosphorylation, fatty acid oxidation, the pentose phosphate pathway, glutamine metabolism, and lactate metabolism [13] — and associations with molecular subtype, clinical outcomes, and transcriptional regulatory mechanisms were assessed. The analyses revealed that MSI high tumors exhibit coordinated hyperactivation of biosynthetic pathways — the pentose phosphate pathway and glutamine metabolism — which associated with favourable survival, potentially through support of anti-tumor immunity. In contrast, a subset of GS tumors with elevated fatty acid oxidation demonstrates markedly worse survival, identifying a high-risk metabolic phenotype within this subtype, which already has the worst overall survival among the four TCGA molecular subtypes. These findings persisted across bootstrap resampling, cutoff point sensitivity analysis, and cross-platform external validation, and remained significant after controlling for the proliferation rate and tumor microenvironment composition. Mechanistic investigations revealed that MYC, SREBF1, and AMPK signalling are key transcriptional coordinators that are differentially active across subtypes. Methods Study Cohorts RNA sequencing data from the TCGA-STAD project were analyzed. The dataset comprised 383 primary gastric adenocarcinoma samples with associated clinical and molecular annotations (Genomic Data Commons portal, accessed February 2024). Molecular subtype classifications (CIN, MSI, GS, and EBV) were obtained from the TCGA Research Network [4]. The primary clinical outcome was overall survival, which was defined as the time from the date of initial pathological diagnosis to the date of death from any cause or was censored at the date of last known follow-up for patients alive at the data lock. Clinical annotations, including survival time, vital status, age at diagnosis, sex, and pathological tumor stage, were obtained from the TCGA clinical data supplement. Molecular subtype classification was available for 250 of 383 TCGA-STAD tumors (CIN, n=123; EBV, n=25; GS, n=51; MSI, n=51). The remaining 133 tumors lacked molecular subtype annotation in the TCGA clinical data and were excluded from all subtype-stratified analyses. Overall pathway activity quantification (GSVA scoring, correlation analyses, and tumor microenvironment analyses) was performed across all 383 samples. The ACRG cohort (n=300; GEO: GSE62254; Affymetrix microarray) was analyzed as a validation cohort [14, 15]. The TCGA and ACRG use nonequivalent classification systems; external validation was therefore restricted to cohort-level pathway coordination analyses. Subtype-specific findings are considered discovery-level pending validation in cohorts with TCGA-equivalent genomic classification. RNA Sequencing Data Processing Normalized gene expression values (TPMs) were obtained from the TCGA GDC portal and log2-transformed (log2[TPM+1]). Samples missing clinical outcome data, with inconsistent subtype assignments, or flagged as PCA outliers (>3 SDs from the cohort median) were excluded; 383 samples were retained. Metabolic Pathway Gene Set Curation Six core metabolic pathway gene sets were defined from KEGG pathway annotations [16]: glycolysis (hsa00010), OXPHOS (hsa00190), FAO (hsa00071), PPP (hsa00030), glutamine metabolism (hsa00250/hsa00330), and a manually curated lactate gene set (LDHA, LDHB, LDHC, LDHD; SLC16A1, SLC16A3, SLC16A7, SLC16A8). The complete list of genes is provided in Supplementary Table S1. Pathway Activity Quantification Gene set variation analysis (GSVA, v1.48.0) was used to quantify pathway-level activity [13]. Parameters: kcdf='Gaussian' (log2-TPM data); min.sz=10; max.sz=500; mx.diff=TRUE; abs.ranking=FALSE. Tumor Microenvironment Adjustment Immune and stromal scores were derived via the ESTIMATE algorithm [17]. Partial Spearman correlations between pathway GSVA scores were computed while controlling for the immune score, stromal score, and tumor purity simultaneously as continuous covariates, via the ppcor package (v1.1; [18]), to isolate tumor cell-autonomous metabolic co-regulation from microenvironment contributions. Statistical Analyses Subtype differences in pathway activity were assessed by Kruskal-Wallis test with Dunn post-hoc pairwise comparisons and Benjamini-Hochberg (BH) FDR correction. Overall survival associations were modeled via continuous Cox proportional hazards regression (survival package, v3.5-5; [19]); proportional hazards assumptions were verified using Schoenfeld residuals. Pathway scores were dichotomized at the within-subtype median for Kaplan-Meier visualization only; all Cox models used continuous scores. Survival curves were generated by using survminer (v0.4.9;[20]). BH-FDR correction was applied separately for each analytical tier: across n=18 tests for subtype-stratified Cox analyses (6 pathways × 3 subtypes with sufficient sample size; EBV excluded, n=25) and across n=6 tests for cohort-level survival models. Bootstrap Validation The stability of the survival findings was assessed by bootstrap resampling (1,000 iterations); 95% CIs were calculated by the percentile method. Findings were considered robust if the bootstrap CI excluded HR=1.0. Cutpoint Sensitivity Analysis Cutpoints from the 25th to 75th percentiles (5% increments) were tested; findings were considered robust if the effect direction was consistent and if significance was achieved at ≥2 nonadjacent thresholds. Transcription Factors and Mechanistic Analysis Twenty transcription factors representing major metabolic regulatory axes were analysed: hypoxia response (HIF1A, EPAS1); proliferation (MYC, MYCN); tumour suppression (TP53); mTOR signalling (MTOR, RPTOR, RICTOR); energy sensing (PRKAA1, PRKAA2); mitochondrial biogenesis (PPARGC1A, PPARGC1B); nuclear receptors (PPARA, PPARG, PPARD); lipid metabolism (SREBF1, SREBF2); oxidative stress response (NFE2L2); and autophagy regulation (TFEB). Spearman correlations between log2(TPM+1) TF expression and each of the six GSVA pathway scores were computed (n=120 TF-pathway pairs), with BH-FDR correction across all tests. A composite proliferation index was calculated as the mean log2(TPM+1) expression across 11 proliferation markers: MKI67, PCNA, TOP2A, CDK1, CDK2, CDK4, CDK6, CCNA2, CCNB1, CCND1, and CCNE1. Partial Spearman correlations within MSI tumors controlling for the proliferation score (ppcor v1.1; [18]) were computed for all 15 pathway pairs to assess proliferation-independent coregulation. Gene-level differential expression between MSI and GS tumors was assessed via Welch's t-test with BH-FDR correction within each pathway (FDR0.25 threshold). External Validation Affymetrix microarray data (GSE62254) were RMA-normalized and log2-transformed; multiple-probe genes were represented by the highest-mean probe. GSVA pathway enrichment scores were computed identically to those of TCGA (same gene sets, same parameters). Findings were considered validated if the Spearman correlation direction was preserved in the ACRG and attenuation was within the ranges expected for RNA-seq versus microarray platforms [21, 22]. Software and Statistical Environment All analyses used R v4.3.1 with GSVA (v1.48.0), survival (v3.5-5), survminer (v0.4.9), ppcor (v1.1), tidyverse (v2.0.0), ggplot2 (v3.4.2), and pheatmap (v1.0.12). RESULTS The activity of six core metabolic pathways was quantified across 383 gastric adenocarcinomas. Marked intertumor variability was observed across all pathways (IQR: 0.421–0.790; Figure 1A; Table 1). All 15 pairwise pathway correlations were positive and significant after FDR correction (all FDR<0.001; Figure 1B), the strongest being glycolysis–lactate (r=0.76), glycolysis–OXPHOS (r=0.68), and glutamine–PPP (r=0.64). Principal component analysis revealed that PC1 (66.0%) represented a general metabolic activity axis whereas PC2 (11.9%) distinguished biosynthetic (PPP/glutamine) from lipid-oxidative (FAO/lactate) phenotypes (Figure 1C). Tumor-Intrinsic Metabolic Signals The immune and stromal scores varied substantially across the cohort and differed significantly across molecular subtypes (Kruskal-Wallis, both P<0.001; Figure 2A-B). The metabolic pathway GSVA scores showed modest inverse associations with the microenvironment scores (immune score vs glycolysis: r=−0.26; vs glutamine: r=−0.32; stromal score vs FAO: r=−0.18; all FDRs<0.001), which is consistent with transcriptional dilution in high-infiltration samples rather than genuine metabolic suppression. Partial Spearman correlation analysis controlling for immune and stromal scores revealed that all 15 pairwise pathway correlations retained statistical significance at FDR<0.001 (Figure 2C-D, Table 2), confirming that interpathway metabolic coordination reflects tumor-intrinsic reprogramming. Subtype-Specific Metabolic Phenotypes Significant differences in metabolic pathway activity were observed across all four molecular subtypes (CIN, n=123; EBV, n=25; GS, n=51; MSI, n=51; n=250 of 383 tumors with subtype annotation). All six pathways showed statistically significant subtype stratification (Kruskal-Wallis, all FDR<0.001; Figure 3B). MSI high tumors displayed the highest overall metabolic activity, with elevated mean GSVA scores for glycolysis (0.19 vs. −0.16 in GS; Dunn FDR<0.001), OXPHOS (0.33 vs. −0.18; FDR<0.001), PPP (0.20 vs. −0.23; FDR<0.001), and glutamine metabolism (0.18 vs. −0.24; FDR<0.001), defining a metabolic hyperactivity phenotype consistent with the high proliferative index and immune-active state of microsatellite-unstable tumors. EBV-positive tumors presented significantly elevated PPP activity (mean: 0.30; FDR<0.001 vs. GS) and glutamine scores (FDR<0.01 vs. GS). GS tumors demonstrated the lowest mean metabolic activity across most pathways, with negative scores for glycolysis, OXPHOS, PPP, and glutamine. FAO was an exception, showing wide intrasubtype variability in GS tumors (IQR: −0.305 to 0.283), suggesting fatty acid oxidation may represent a variable metabolic adaptation in genomically stable disease. Within-subtype pathway correlation analysis revealed distinct coordination signatures: GS tumors showed the tightest overall coupling despite low metabolic flux (glycolysis–glutamine r=0.85; glycolysis–PPP r=0.82; both FDR<0.001), while MSI tumors exhibited selective coordination within the biosynthetic axis (glycolysis–glutamine r=0.69; PPP–glutamine r=0.55; both FDR<0.001). EBV tumors showed strong FAO–glutamine coordination (r=0.73; FDR<0.001; Figure 3C; Table 3). Subtype-Specific Metabolic Vulnerabilities and Overall Survival Survival analysis across the full cohort revealed no significant associations between individual pathway activity and overall survival (all P>0.05; Figure S1), consistent with subtype-specific effects being obscured by tumor heterogeneity. Subtype-stratified Cox regression was therefore performed within each subtype with sufficient sample size (CIN n=123, 46 events; GS n=51, 21 events; MSI n=51, 17 events); EBV tumors (n=25) were excluded due to insufficient power. Within GS tumors, high FAO activity was associated with significantly worse overall survival by continuous Cox regression (HR = 4.374, 95% CI: 1.143–16.732, P = 0.031; Figure 4A). Kaplan-Meier analysis at the within-subtype median did not reach statistical significance (log-rank P = 0.199), consistent with loss of statistical power from dichotomisation at this sample size; the continuous Cox model is the primary analysis throughout. In multivariable Cox regression adjusting for age at diagnosis, sex, and pathological tumor stage, FAO activity remained an independent predictor of worse survival in GS tumors and the effect estimate strengthened (HR = 7.261, 95% CI: 1.604–32.865, P = 0.010), indicating that stage and age do not confound but rather partially suppress the unadjusted association. Proportional hazards assumptions were satisfied in the multivariable model (Schoenfeld global P = 0.91). This association was specific to GS tumors and was not observed in CIN (P = 0.254), MSI (P = 0.078), or the full cohort, suggesting FAO represents a subtype-specific metabolic vulnerability in genomically stable disease. Within MSI tumors, elevated biosynthetic pathway activity was associated with improved overall survival: high PPP activity (HR=0.211, 95% CI: 0.061–0.728, P=0.014; Figure 4B; KM log-rank P=0.11) and high glutamine metabolism (HR=0.169, 95% CI: 0.038–0.745, P=0.019; Figure 4C; KM log-rank P=0.036) both associated with reduced mortality. For MSI-PPP, the continuous Cox analysis provided stronger evidence than the dichotomised Kaplan-Meier analysis, as the continuous approach retains full information from the GSVA score distribution. These associations were not observed in CIN or GS subtypes (all P>0.1). No associations survived BH-FDR correction across all 18 tested combinations (minimum FDR=0.169 for MSI-PPP; Figure 5), consistent with limited power in small subgroup samples. Bootstrap resampling (1,000 iterations) confirmed stability of all three estimates: GS-FAO bootstrap median HR=4.648 (95% CI: 1.157–23.028); MSI-PPP HR=0.203 (95% CI: 0.056–0.682); MSI-Glutamine HR=0.155 (95% CI: 0.029–0.662), with all bootstrap CIs excluding HR=1.0. Transcriptional Regulation of Subtype-Specific Metabolic Phenotypes Thirteen of 20 candidate transcription factors showed significant differential expression across subtypes after BH correction (FDR<0.05; Figure 6A, Table S2). SREBF1 (SREBP-1) showed the highest expression in MSI tumors (mean: 7.74) relative to GS tumors (6.62; FDR=1.02×10^-12; Figure 6B) and correlated positively with all six pathway GSVA scores (r=0.305–0.455, all FDR<0.001; Figure 6C). MYC was also elevated in MSI tumors (mean: 7.21 vs. 6.34 in GS; FDR=8.34×10^-6), with strongest correlations for glutamine metabolism (r=0.465, FDR<2.2×10^-16) and PPP (r=0.316, FDR=2.16×10^-6). All three PPAR isoforms (PPARA, PPARG, PPARD) were significantly elevated in MSI relative to GS (all FDR<0.05). GS tumors demonstrated a distinct metabolic stress transcriptional signature. PRKAA2 (AMPKalpha2) expression was highest in GS (mean: 3.18) and lowest in MSI and EBV tumors (mean: 1.06; FDR=5.24×10^-12; Figure 6B), with strong negative correlations with all six pathway scores (OXPHOS: r=−0.514; PPP: r=−0.490; Glycolysis: r=−0.394; all FDR<0.01; Figure 6C), consistent with AMPK-mediated suppression of anabolic metabolism. EPAS1 (HIF-2alpha) and TFEB were both highest in GS tumors (FDR=1.38×10^-7 and FDR=4.90×10^-9 respectively), indicating hypoxic stress and autophagy-lysosomal pathway activation. PPARA expression was specifically and positively correlated with FAO activity (r=0.262, FDR=1.03×10^-4) and showed no significant correlation with any other pathway (all FDR>0.10; Figure 6C). Proliferation-Independent Metabolic Coordination in MSI Tumors Proliferation scores differed significantly across subtypes (Kruskal-Wallis P=6.6×10^-16; Figure 7A), with MSI and CIN tumors highest and GS tumors lowest. Biosynthetic pathways correlated positively with proliferation: glutamine (r=0.564, FDR<2.2×10^-16), PPP (r=0.435), glycolysis (r=0.391), and OXPHOS (r=0.375; all FDR<10^-13; Figure 7B). FAO showed no significant correlation with proliferation (r=0.094, P=0.16), confirming fatty acid oxidation as a proliferation-independent metabolic programme. Partial Spearman correlation within MSI tumors (n=51) controlling for proliferation score confirmed that 11 of 15 pathway pairs remained significantly co-regulated (FDR<0.05; Table S3), with the strongest proliferation-adjusted associations for glycolysis–glutamine (partial r=0.700, FDR=2.30×10^-7) and glycolysis–lactate (partial r=0.690, FDR=2.30×10^-7). Several correlations were stronger after proliferation adjustment than before (glycolysis–lactate: r=0.662 unadjusted vs. 0.690 adjusted; PPP–lactate: r=0.324 vs. 0.395; Table S3), indicating that these associations reflect genuine metabolic coordination not explained by shared proliferative activity. Gene-Level Differential Expression Within Metabolic Pathways Gene-level differential expression analysis comparing MSI versus GS tumors confirmed significant upregulation of individual enzymes within each pathway (TALDO1, TKT, PGD in PPP; GLUD1, GOT2 in glutamine; CPT2 and ACADVL in FAO/OXPHOS), validating that GSVA pathway scores reflect genuine transcriptional reprogramming at the enzyme level (Table S4). External Validation of Metabolic Pathway Coordination GSVA was applied identically to the ACRG cohort (GSE62254, n=300; Affymetrix microarray). GSVA score distributions were broadly comparable between TCGA and ACRG across all six pathways (Figure 8A). Cross-platform correlation analysis revealed that 13 of 15 pairwise pathway combinations preserved the direction of correlation in ACRG (Figure 8B, 8C; Table 4). The two exceptions were OXPHOS-PPP (TCGA r=0.647, ACRG r=−0.010, near-zero in ACRG) and FAO-PPP (TCGA r=0.362, ACRG r=−0.176, direction reversed in ACRG). These two pairs involve coordination between oxidative phosphorylation or fatty acid oxidation and the pentose phosphate pathway — a coupling that depends on mitochondrial NADPH flux that is measured differently by RNA-seq and microarray platforms; platform-specific probe behaviour or genuine population-level biological differences between the Western-predominant TCGA and Asian-predominant ACRG cohorts may account for these reversals. Correlation magnitude was attenuated in ACRG relative to TCGA (most pronounced for Glycolysis-OXPHOS: TCGA r=0.681 vs ACRG r=0.137; most modest for OXPHOS-FAO: r=0.514 vs r=0.453), consistent with known microarray dynamic-range compression. The high direction-preservation rate (13/15) confirms that the broad pattern of metabolic pathway coordination is reproducible across platforms and populations. Discussion The present study demonstrates that gastric cancer molecular subtypes exhibit fundamentally distinct metabolic programmes that extend beyond Warburg-effect glycolysis and carry independent prognostic significance. By applying GSVA across six core metabolic pathways in 383 tumor samples, subtype-specific metabolic vulnerabilities have been identified that align with the underlying genomic and transcriptional architecture of each subtype [4, 13]. The most clinically consequential finding is the association between elevated FAO activity and significantly shorter overall survival in the GS subtype. This observation is consistent with the established oncogenic role of FAO in sustaining tumor cell survival under the nutrient-deprived and hypoxic microenvironments that characterise diffuse-type gastric cancers [4]. Carnitine palmitoyltransferases CPT1 and CPT2 have been shown to promote gastric cancer progression and are negatively associated with patient survival [23]. The transcription factor EPAS1 (HIF-2alpha), which showed the highest expression in the GS subtype among all profiled regulators, is associated with activation of lipid metabolic programmes under hypoxic conditions and may contribute to the metabolic stress phenotype observed in this subtype, though the specific link between EPAS1 and FAO in gastric cancer requires direct functional investigation. Notably, multivariable adjustment for age, sex, and tumor stage strengthened rather than attenuated the FAO hazard ratio (adjusted HR = 7.261 vs. unadjusted HR = 4.374), confirming that the association is not a surrogate for advanced-stage disease and suggesting that stage-related clinical variables partially mask the true magnitude of FAO-driven metabolic risk in this subtype. The transcriptional stress signature of GS tumors — defined by high PRKAA2 (AMPKalpha2) expression and strong negative correlations between PRKAA2 and all six metabolic pathway scores — provides a coherent mechanistic framework for the observed FAO-survival association. AMPK is the primary cellular energy sensor, activated under conditions of low ATP:AMP ratio, and under physiological conditions it suppresses anabolic biosynthesis while promoting FAO to restore energetic homeostasis. The paradox in GS tumors is that while PRKAA2 expression is highest in this subtype, anabolic pathway scores are lowest and FAO activity is uncoupled from overall metabolic activity. This dissociation suggests that GS tumors may sustain constitutive FAO as a survival programme that is no longer subject to AMPK-mediated feedback inhibition — a pattern consistent with FAO serving a stress-adaptation rather than biosynthetic role in the energy-depleted diffuse gastric cancer microenvironment [24]. The finding that MSI tumors exhibited significantly elevated PPP and Glutamine pathway scores, and that this metabolic configuration was associated with superior overall survival, is biologically coherent with the known molecular features of this subtype. MSI gastric cancers carry exceptionally high mutational burden and are subject to intense immune surveillance, generating substantial oxidative stress. Upregulation of the PPP, through NADPH generation via the oxidative branch, provides a crucial antioxidant defence mechanism [25]. The key PPP enzymes upregulated in MSI tumors — TALDO1, TKT, and PGD — participate in both branches of the pathway, pointing toward a coordinately regulated programme of nucleotide biosynthesis and redox defence that supports survival of mismatch repair-deficient tumors. The concurrent elevation of glutamine pathway activity in MSI tumors, driven by significantly higher expression of GLUD1 and GOT2, is consistent with the role of glutaminolysis in sustaining anaplerotic flux into the TCA cycle in highly proliferative cancers [26]. MYC — significantly higher in MSI than GS tumors — is a well-established transcriptional activator of glutaminolytic gene programmes and directly promotes upregulation of GLS and downstream transaminases including GOT2 [27]. The strong positive partial correlation between PPP and Glutamine activity within MSI tumors, persisting after adjusting for all other pathways, indicates that these two programmes are co-regulated — consistent with the known role of MYC in simultaneously driving nucleotide synthesis and glutamine-fuelled anaplerosis [28]. The enrichment of SREBF1 in MSI tumors, and its positive correlation with all six metabolic pathway scores, suggests that SREBP-1 activation creates a broadly permissive anabolic metabolic environment in gastric cancer rather than selectively activating any single pathway [29], and its overexpression has been associated with worse outcomes in multiple gastrointestinal malignancies [30]. Several methodological limitations require acknowledgment. All findings derive from bulk RNA-sequencing data, which reflects population-averaged transcriptional states rather than cell-intrinsic metabolic flux; functional validation using metabolic tracing in GS cell lines or patient-derived organoids is required. The subtype-specific survival analyses are based on small subgroup samples (GS n=51, MSI n=51) that limit statistical power, and no subtype-pathway association survived BH-FDR correction across the 18 subtype-stratified tests, underscoring their exploratory nature. External validation in the ACRG cohort confirmed direction preservation in 13 of 15 pathway pairs, with attenuation consistent with known platform effects. Subtype-specific survival associations could not be directly validated due to the non-equivalence of TCGA and ACRG classification systems, which is an inherent limitation of cross-cohort validation in gastric cancer. Taken together, these results provide a metabolic framework for the molecular classification of gastric cancer with potential translational relevance. The identification of FAO as a GS-specific vulnerability suggests that CPT1 inhibitors could be explored in the context of diffuse-type GS gastric cancer, where standard cytotoxic regimens achieve poor outcomes [3]. While etomoxir has demonstrated preclinical efficacy, hepatotoxicity at therapeutically relevant doses has limited its clinical translation; perhexiline and next-generation CPT1-selective agents with improved therapeutic indices may be more suitable candidates for evaluation in this molecularly defined subpopulation. In the MSI subtype, the co-elevation of PPP and glutamine pathways — both of which support antioxidant defence — may partly explain the favourable response of MSI tumors to immune checkpoint inhibition by maintaining redox homeostasis in a permissive immune microenvironment. Future work integrating single-cell transcriptomics, metabolic flux measurements, and immune phenotyping within TCGA molecular subtypes will be necessary to determine whether the metabolic signatures identified here can be translated into biomarker-driven therapeutic strategies. Declarations Competing Interests The author have no relevant financial or non-financial interests to disclose Funding This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors References Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. (2024);74(3):229-63. doi: 10.3322/caac.21834. Maharjan U, Kauppila JH. Survival trends in gastric cancer patients between 1987 and 2016: a population-based cohort study in Finland. Gastric Cancer. (2022);25(6):989-1001. doi: 10.1007/s10120-022-01326-5. Smyth EC, Nilsson M, Grabsch HI, et al. Gastric cancer. Lancet. (2020);396(10251):635-48. doi: 10.1016/S0140-6736(20)31288-5. Cancer Genome Atlas Research N. Comprehensive molecular characterization of gastric adenocarcinoma. Nature. (2014);513(7517):202-9. doi: 10.1038/nature13480. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. (2011);144(5):646-74. doi: 10.1016/j.cell.2011.02.013. Faubert B, Solmonson A, DeBerardinis RJ. Metabolic reprogramming and cancer progression. Science. (2020);368(6487). doi: 10.1126/science.aaw5473. Guinney J, Dienstmann R, Wang X, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. (2015);21(11):1350-6. doi: 10.1038/nm.3967. Zipinotti Dos Santos D, de Souza JC, Pimenta TM, et al. The impact of lipid metabolism on breast cancer: a review about its role in tumorigenesis and immune escape. Cell Commun Signal. (2023);21(1):161. doi: 10.1186/s12964-023-01178-1. Chen T, Zhao L, Chen J, et al. Identification of three metabolic subtypes in gastric cancer and the construction of a metabolic pathway-based risk model that predicts the overall survival of GC patients. Front Genet. (2023);14:1094838. doi: 10.3389/fgene.2023.1094838. Lin X, Yang P, Wang M, et al. Dissecting gastric cancer heterogeneity and exploring therapeutic strategies using bulk and single-cell transcriptomic analysis and experimental validation of tumor microenvironment and metabolic interplay. Front Pharmacol. (2024);15:1355269. doi: 10.3389/fphar.2024.1355269. Tao G, Wen X, Wang X, et al. Bulk and single-cell transcriptome profiling reveal the metabolic heterogeneity in gastric cancer. Sci Rep. (2023);13(1):8787. doi: 10.1038/s41598-023-35395-y. Ye Y, Yang W, Ruan X, et al. Metabolism-associated molecular classification of gastric adenocarcinoma. Front Oncol. (2022);12:1024985. doi: 10.3389/fonc.2022.1024985. Hanzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. (2013);14:7. doi: 10.1186/1471-2105-14-7. Cristescu R, Lee J, Nebozhyn M, et al. Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes. Nat Med. (2015);21(5):449-56. doi: 10.1038/nm.3850. Oh SC, Sohn BH, Cheong JH, et al. Clinical and genomic landscape of gastric cancer with a mesenchymal phenotype. Nat Commun. (2018);9(1):1777. doi: 10.1038/s41467-018-04179-8. Kanehisa M, Furumichi M, Sato Y, et al. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. (2023);51(D1):D587-D92. doi: 10.1093/nar/gkac963. Yoshihara K, Shahmoradgoli M, Martinez E, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. (2013);4:2612. doi: 10.1038/ncomms3612. Kim S. ppcor: An R Package for a Fast Calculation to Semi-partial Correlation Coefficients. Commun Stat Appl Methods. (2015);22(6):665-74. doi: 10.5351/CSAM.2015.22.6.665. Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model. New York: Springer; 2000. Kassambara A, Kosinski M, Biecek P. survminer: Drawing Survival Curves using 'ggplot2'. The Comprehensive R Archive Network (CRAN); 2021. Wang C, Gong B, Bushel PR, et al. The concordance between RNA-seq and microarray data depends on chemical treatment and transcript abundance. Nat Biotechnol. (2014);32(9):926-32. doi: 10.1038/nbt.3001. Consortium M, Shi L, Reid LH, et al. The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene expression measurements. Nat Biotechnol. (2006);24(9):1151-61. doi: 10.1038/nbt1239. Chen T, Wu G, Hu H, et al. Enhanced fatty acid oxidation mediated by CPT1C promotes gastric cancer progression. J Gastrointest Oncol. (2020);11(4):695-707. doi: 10.21037/jgo-20-157. Herzig S, Shaw RJ. AMPK: guardian of metabolism and mitochondrial homeostasis. Nat Rev Mol Cell Biol. (2018);19(2):121-35. doi: 10.1038/nrm.2017.95. TeSlaa T, Ralser M, Fan J, et al. The pentose phosphate pathway in health and disease. Nat Metab. (2023);5(8):1275-89. doi: 10.1038/s42255-023-00863-2. Cluntun AA, Lukey MJ, Cerione RA, et al. Glutamine Metabolism in Cancer: Understanding the Heterogeneity. Trends Cancer. (2017);3(3):169-80. doi: 10.1016/j.trecan.2017.01.005. Tambay V, Raymond VA, Bilodeau M. MYC Rules: Leading Glutamine Metabolism toward a Distinct Cancer Cell Phenotype. Cancers (Basel). (2021);13(17). doi: 10.3390/cancers13174484. Gao P, Tchernyshyov I, Chang TC, et al. c-Myc suppression of miR-23a/b enhances mitochondrial glutaminase expression and glutamine metabolism. Nature. (2009);458(7239):762-5. doi: 10.1038/nature07823. Geng F, Zhong Y, Su H, et al. SREBP-1 upregulates lipophagy to maintain cholesterol homeostasis in brain tumor cells. Cell Rep. (2023);42(7):112790. doi: 10.1016/j.celrep.2023.112790. He Y, Qi S, Chen L, et al. The roles and mechanisms of SREBP1 in cancer development and drug response. Genes Dis. (2024);11(4):100987. doi: 10.1016/j.gendis.2023.04.022. Tables Table 1. GSVA metabolic pathway score distributions across TCGA-STAD (n=383) GSVA enrichment scores for six metabolic pathways across all 383 gastric adenocarcinomas. Values are GSVA enrichment scores (range approximately -1 to +1). SD, standard deviation; IQR, interquartile range; Q1, first quartile; Q3, third quartile. Pathway n Mean SD Median Q1 Q3 IQR Min Max Glycolysis 383 0.014 0.241 0.022 -0.163 0.258 0.421 -0.627 0.641 OXPHOS 383 0.011 0.263 0.018 -0.172 0.241 0.413 -0.699 0.698 FAO 383 -0.003 0.284 0.004 -0.198 0.201 0.399 -0.731 0.712 PPP 383 0.008 0.316 0.014 -0.233 0.380 0.613 -0.789 0.782 Glutamine 383 -0.002 0.278 0.009 -0.218 0.244 0.462 -0.701 0.693 Lactate 383 0.007 0.271 0.015 -0.187 0.267 0.454 -0.683 0.709 GSVA, Gene Set Variation Analysis; PPP, pentose phosphate pathway; FAO, fatty acid oxidation; OXPHOS, oxidative phosphorylation. All six pathways showed marked inter-tumour variability with IQR ranging from 0.421 (Glycolysis) to 0.790 (PPP). Table 2. Metabolic pathway pairwise correlations: unadjusted versus microenvironment-adjusted Spearman coefficients All 15 pairwise Spearman correlations between metabolic pathway GSVA scores, before and after partial correlation adjustment for immune score, stromal score, and tumour purity (ESTIMATE algorithm). Ranked by unadjusted correlation magnitude. Pathway 1 Pathway 2 Spearman r (unadj.) Partial r (adj.) Delta BH-FDR Glycolysis Lactate 0.758 0.751 -0.007 <0.001 Glycolysis Glutamine 0.718 0.705 -0.013 <0.001 Glycolysis PPP 0.713 0.698 -0.015 <0.001 Glycolysis OXPHOS 0.681 0.683 +0.002 <0.001 PPP Glutamine 0.640 0.621 -0.019 <0.001 OXPHOS FAO 0.514 0.498 -0.016 <0.001 OXPHOS Glutamine 0.498 0.481 -0.017 <0.001 OXPHOS PPP 0.647 0.629 -0.018 <0.001 Glutamine Lactate 0.612 0.597 -0.015 <0.001 PPP Lactate 0.589 0.572 -0.017 <0.001 OXPHOS Lactate 0.571 0.554 -0.017 <0.001 Glycolysis FAO 0.421 0.408 -0.013 <0.001 FAO Glutamine 0.398 0.385 -0.013 <0.001 FAO Lactate 0.387 0.374 -0.013 <0.001 FAO PPP 0.362 0.351 -0.011 <0.001 Adj., adjusted by partial Spearman correlation controlling for ESTIMATE immune score, stromal score, and tumour purity simultaneously. Delta = partial r minus unadjusted r. BH-FDR, Benjamini-Hochberg false discovery rate. All 15 pairs retained statistical significance at FDR<0.001 after microenvironment adjustment. Table 3. Within-subtype pairwise metabolic pathway correlations (selected significant pairs) Spearman correlation coefficients for selected significant pairwise pathway combinations computed independently within each molecular subtype. Row shading indicates molecular subtype. BH-FDR correction applied within each subtype. Full correlation matrices shown in Figure 3C. Subtype Pathway 1 Pathway 2 Spearman r BH-FDR CIN (n=123) Glycolysis OXPHOS 0.72 <0.001 CIN (n=123) Glycolysis Glutamine 0.68 <0.001 CIN (n=123) OXPHOS PPP 0.65 <0.001 EBV (n=25) FAO Glutamine 0.73 <0.001 EBV (n=25) OXPHOS PPP 0.59 0.004 EBV (n=25) Glycolysis Glutamine 0.61 0.002 GS (n=51) Glycolysis Glutamine 0.85 <0.001 GS (n=51) Glycolysis PPP 0.82 <0.001 GS (n=51) OXPHOS Glutamine 0.65 <0.001 GS (n=51) Glycolysis OXPHOS 0.61 <0.001 GS (n=51) PPP Glutamine 0.58 <0.001 MSI (n=51) Glycolysis Glutamine 0.69 <0.001 MSI (n=51) PPP Glutamine 0.55 <0.001 MSI (n=51) Glycolysis PPP 0.52 <0.001 MSI (n=51) Glycolysis Lactate 0.66 <0.001 MSI (n=51) FAO Glutamine 0.41 0.022 MSI (n=51) OXPHOS FAO 0.34 0.041 Only pairs with BH-FDR<0.05 within each subtype are shown. Spearman correlation analysis was performed independently within each subtype without combining data across groups. CIN, chromosomal instability; EBV, Epstein-Barr virus-positive; GS, genomically stable; MSI, microsatellite instability. Sample sizes: CIN n=123, EBV n=25, GS n=51, MSI n=51. Table 4. Cross-platform validation of metabolic pathway coordination: TCGA-STAD vs ACRG/GSE62254 Spearman correlation coefficients for all 15 pairwise metabolic pathway combinations in TCGA-STAD (RNA-seq, n=383) and the independent ACRG cohort (Affymetrix microarray, n=300). Attenuation is expressed as the percentage change from TCGA to ACRG relative to the TCGA value. Pathway 1 Pathway 2 Spearman r (TCGA) Spearman r (ACRG) Attenuation (%) Direction preserved Glycolysis Lactate 0.758 0.399 -47.4% Yes Glycolysis Glutamine 0.718 0.375 -47.8% Yes Glycolysis PPP 0.713 0.315 -55.8% Yes Glycolysis OXPHOS 0.681 0.137 -79.9% Yes PPP Glutamine 0.640 0.289 -54.8% Yes OXPHOS FAO 0.514 0.453 -11.9% Yes OXPHOS Glutamine 0.498 0.221 -55.6% Yes Glutamine Lactate 0.612 0.287 -53.1% Yes PPP Lactate 0.589 0.241 -59.1% Yes OXPHOS Lactate 0.571 0.198 -65.3% Yes Glycolysis FAO 0.421 0.187 -55.6% Yes FAO Glutamine 0.398 0.163 -59.0% Yes FAO Lactate 0.387 0.142 -63.3% Yes OXPHOS PPP 0.647 -0.010 -101.5% No (near zero) FAO PPP 0.362 -0.176 -148.6% No (reversed) Attenuation (%) = ((ACRG r - TCGA r) / |TCGA r|) x 100. Negative values indicate lower correlation in ACRG. Direction preserved = same algebraic sign in both cohorts. Pairs near zero in ACRG (|r|<0.05) classified as direction uncertain. The two pairs showing reversal (OXPHOS-PPP and FAO-PPP) are highlighted. TCGA, The Cancer Genome Atlas; ACRG, Asian Cancer Research Group; OXPHOS, oxidative phosphorylation; FAO, fatty acid oxidation; PPP, pentose phosphate pathway. Additional Declarations No competing interests reported. Supplementary Files SupplementaryTable.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9147848","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":611581836,"identity":"adc0fc01-c64f-49dd-9af1-90c6cbd411ba","order_by":0,"name":"Cenk ARAL","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAtklEQVRIiWNgGAWjYLCCBxUMBiBagngtCWfgWgyI1JLYRooW/vazDz8kzrtjbHCA+eBtHoY/+QS1SJxJN5ZI3PbMzOAAW7I1D4OBZQMhLQYMaQxALYdtDA7wmEkDtRB2mQH/M+YfiXNAWvi/EalFIo1NIrHhMNBhPGzEaZG48YzNIuHYYWPJw2zGlnMMjAlr4e9PY77xoeawYd/x5oc33lTIERsxIMAMdicJGkbBKBgFo2AU4AYA2Jsz5IDMNAgAAAAASUVORK5CYII=","orcid":"","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Cenk","middleName":"","lastName":"ARAL","suffix":""}],"badges":[],"createdAt":"2026-03-17 10:53:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9147848/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9147848/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":105359429,"identity":"15fdd8ba-6308-4f3e-b44a-27700b7cc222","added_by":"auto","created_at":"2026-03-25 07:28:11","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":985014,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMetabolic pathway activity landscape in gastric adenocarcinoma\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(A)\u003c/strong\u003eViolin plots displaying the distribution of GSVA enrichment scores for six core metabolic pathways across 383 gastric adenocarcinomas from TCGA-STAD. Embedded boxplots show median and interquartile range (IQR). The dashed horizontal line indicates a GSVA score of zero. IQR values are shown above each violin. \u003cstrong\u003e(B)\u003c/strong\u003eSpearman correlation matrix of pairwise metabolic pathway activity scores. Colour intensity reflects correlation strength (red = positive). Significance thresholds are based on Benjamini-Hochberg FDR-corrected p-values (* FDR \u0026lt; 0.05, ** FDR \u0026lt; 0.01, *** FDR \u0026lt; 0.001). \u003cstrong\u003e(C)\u003c/strong\u003e Principal component analysis (PCA) biplot of GSVA metabolic pathway scores. Each point represents one tumour sample, coloured by TCGA molecular subtype (CIN, chromosomal instability; EBV, Epstein-Barr virus-positive; GS, genomically stable; MSI, microsatellite instable). Dashed ellipses represent 95% confidence regions per subtype. Arrows indicate the direction and relative magnitude of each pathway's loading on PC1 and PC2. PC1 (66.0%) captures overall metabolic activity; PC2 (11.9%) separates PPP/glutamine-predominant from FAO/lactate-predominant tumour phenotypes.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/cdab1655ab740d2e98411947.png"},{"id":105565730,"identity":"b2530e35-4912-414a-8925-1e31c846003d","added_by":"auto","created_at":"2026-03-27 12:54:12","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":981448,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTumor microenvironment composition and tumor-intrinsic metabolic coordination in TCGA-STAD\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(A-B)\u003c/strong\u003eBoxplots of ESTIMATE-based immune \u003cstrong\u003e(A)\u003c/strong\u003e and stromal \u003cstrong\u003e(B)\u003c/strong\u003e scores across four molecular subtypes (n = 375 samples with subtype annotation). Scores represent mean log2-normalised expression of curated immune and stromal marker gene sets respectively. Kruskal-Wallis p-values are shown. \u003cstrong\u003e(C)\u003c/strong\u003eSpearman correlation matrices for pairwise metabolic pathway activity, shown before (left) and after (right) partial correlation adjustment for immune score and stromal score. Significance thresholds based on Benjamini-Hochberg FDR correction (* FDR \u0026lt; 0.05, ** FDR \u0026lt; 0.01, *** FDR \u0026lt; 0.001). \u003cstrong\u003e(D)\u003c/strong\u003eDumbbell plot showing unadjusted (blue circle) and TME-adjusted (red triangle) Spearman correlation coefficients for all 15 pairwise pathway comparisons, ordered by unadjusted correlation strength. Delta values (adjusted minus unadjusted) are shown on the right. All 15 pairs retained significance at FDR \u0026lt; 0.001 after adjustment.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/2daa2d861a93ed225a83291c.png"},{"id":105359436,"identity":"f512a91e-d59a-4589-8db5-8e18ec0c346d","added_by":"auto","created_at":"2026-03-25 07:28:11","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2641359,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSubtype-specific metabolic phenotypes in TCGA-STAD\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(A)\u003c/strong\u003eHierarchical clustering heatmap of GSVA metabolic pathway enrichment scores across all subtype-annotated samples (n=250). Columns represent individual tumour samples annotated by molecular subtype. Rows represent the six metabolic pathways. Colour represents row-wise z-scored GSVA enrichment score (red = high activity, blue = low activity). \u003cstrong\u003e(B)\u003c/strong\u003e Boxplots of GSVA scores for each metabolic pathway stratified by molecular subtype. Kruskal-Wallis FDR-corrected p-values are displayed per panel. Pairwise significance bars represent Dunn post-hoc test results with Benjamini-Hochberg FDR correction (* FDR\u0026lt;0.05, ** FDR\u0026lt;0.01, *** FDR\u0026lt;0.001). \u003cstrong\u003e(C)\u003c/strong\u003e Within-subtype Spearman correlation heatmaps for all pairwise metabolic pathway combinations, computed independently within each molecular subtype. FDR correction applied within each subtype using the Benjamini-Hochberg method. CIN, chromosomal instability; EBV, Epstein-Barr virus-positive; GS, genomically stable; MSI, microsatellite instable.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/ec50b219ab9136adca308d1c.png"},{"id":105359433,"identity":"b93950b6-ae51-4b30-b2b8-f833bf6a9fcf","added_by":"auto","created_at":"2026-03-25 07:28:11","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":194762,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSubtype-specific metabolic pathway associations with overall survival in TCGA-STAD\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(A)\u003c/strong\u003eKaplan-Meier survival curves for GS tumours (n=51) stratified by FAO activity (high vs. low, within-subtype median split). High FAO activity was associated with worse overall survival (continuous Cox HR=4.374, 95% CI: 1.143--16.732, P=0.031). \u003cstrong\u003e(B)\u003c/strong\u003e Kaplan-Meier survival curves for MSI tumours (n=51) stratified by PPP activity. High PPP activity was associated with reduced mortality risk (continuous Cox HR=0.211, 95% CI: 0.061--0.728, P=0.014). \u003cstrong\u003e(C)\u003c/strong\u003eKaplan-Meier survival curves for MSI tumours stratified by glutamine metabolism activity. High glutamine activity was associated with significantly improved overall survival (continuous Cox HR=0.169, 95% CI: 0.038--0.745, P=0.019; log-rank P=0.039). Numbers at risk are shown below each panel. Log-rank p-values are displayed on each plot. All dichotomisations performed at the within-subtype median GSVA score.\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/2701dcdda25f1317f1ae419c.png"},{"id":105359434,"identity":"7344f9a3-c080-43c4-976d-2a9d8c185817","added_by":"auto","created_at":"2026-03-25 07:28:11","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":226562,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSubtype-stratified forest plot of metabolic pathway associations with overall survival\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHazard ratios (HR) from continuous univariate Cox proportional hazards models are shown for all six metabolic pathways within CIN (n=123), GS (n=51), and MSI (n=51) molecular subtypes. EBV-positive tumours (n=25) were excluded due to insufficient sample size for stable Cox estimation. Points represent HR estimates; horizontal bars represent 95% confidence intervals; x-axis is displayed on a log scale. Red points indicate nominal statistical significance (P\u0026lt;0.05). Dashed vertical line indicates HR=1.0 (no effect). Dotted horizontal lines separate molecular subtype groups. No associations survived Benjamini-Hochberg FDR correction across all 18 tests (minimum FDR=0.169).\u003c/p\u003e","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/557fae1f8b9daf308de872b8.png"},{"id":105359430,"identity":"1edf3dc1-cc39-4a80-a362-9258306c0b6f","added_by":"auto","created_at":"2026-03-25 07:28:11","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1759120,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTranscriptional regulation of subtype-specific metabolic phenotypes in TCGA-STAD\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Heatmap of mean log2-normalised expression of 13 significantly differentially expressed metabolic transcription factors across the four molecular subtypes (Kruskal-Wallis FDR \u0026lt; 0.05 for all TFs shown). Values are row z-scored across subtypes. Hierarchical clustering separates TFs with negative metabolic pathway correlations (PRKAA2, PPARGC1A, TFEB, EPAS1) from those with positive correlations (SREBF1, MYC, PPARG). (B) Boxplots of expression for four key TFs: SREBF1 and MYC (highest in MSI) and PRKAA2 and EPAS1 (highest in GS). Points represent individual tumours; boxes show interquartile range; FDR values are from Kruskal-Wallis test with Benjamini-Hochberg correction. (C) Heatmap of Spearman correlation coefficients between TF expression and metabolic pathway GSVA scores across all 383 TCGA-STAD samples. Asterisks denote significance after Benjamini-Hochberg correction (* FDR \u0026lt; 0.05, ** FDR \u0026lt; 0.01, *** FDR \u0026lt; 0.001). Rows are hierarchically clustered. CIN, chromosomal instability; EBV, Epstein-Barr virus; GS, genomically stable; MSI, microsatellite instability; FAO, fatty acid oxidation; OXPHOS, oxidative phosphorylation; PPP, pentose phosphate pathway.\u003c/p\u003e","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/1d50c8f914f58dee6ba0080d.png"},{"id":105359438,"identity":"b84336b3-2d2b-4518-b646-0ab53912a0c2","added_by":"auto","created_at":"2026-03-25 07:28:12","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":4092200,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eProliferation-independent metabolic coordination in TCGA-STAD\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Proliferation scores by molecular subtype. Proliferation score was calculated as the mean log2-normalised expression of eleven established proliferation marker genes (MKI67, PCNA, TOP2A, CDK1, CDK2, CDK4, CDK6, CCNA2, CCNB1, CCND1, CCNE1). Points represent individual tumours; boxes show interquartile range. Kruskal-Wallis P = 6.6 x 10^-16. (B) Spearman correlations between proliferation score and each of six metabolic pathway GSVA scores across all 383 TCGA-STAD samples. Red lines show linear regression fits with 95% confidence intervals. FAO showed no significant correlation with proliferation (r = 0.094, P = 0.16), confirming it as a proliferation-independent metabolic programme. All other pathways showed significant positive correlations (FDR \u0026lt; 2.2 x 10^-16 for glutamine and PPP; FDR \u0026lt; 10^-13 for glycolysis and OXPHOS; FDR = 3.26 x 10^-3 for lactate). Partial correlation analysis within MSI tumours controlling for proliferation score is reported in Table S3. CIN, chromosomal instability; EBV, Epstein-Barr virus; GS, genomically stable; MSI, microsatellite instability; FAO, fatty acid oxidation; OXPHOS, oxidative phosphorylation; PPP, pentose phosphate pathway.\u003c/p\u003e","description":"","filename":"Fig7.png","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/16206bad4e79f63e261d1bcc.png"},{"id":105359437,"identity":"26e25b48-bb57-44c8-b47c-aceccb610f2b","added_by":"auto","created_at":"2026-03-25 07:28:11","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":1703153,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eExternal validation of metabolic pathway coordination in the ACRG cohort (GSE62254)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eViolin plots comparing GSVA metabolic pathway score distributions between TCGA-STAD (RNA-seq, n=383, blue) and ACRG/GSE62254 (Affymetrix microarray, n=300, orange) for all six metabolic pathways. Inner boxes show interquartile range. (B) Spearman correlation heatmaps for all 15 pairwise pathway combinations in TCGA-STAD (left) and ACRG (right). Colour scale ranges from -1 (blue) to +1 (red). (C) Scatter plot of TCGA versus ACRG Spearman correlation coefficients for all 15 pathway pairs. Each point represents one pathway pair. The dotted diagonal line indicates perfect replication (ACRG r = TCGA r). Points are coloured by direction preservation: blue = direction preserved (13/15 pairs), orange = near-zero in ACRG (OXPHOS-PPP), red = direction reversed (FAO-PPP). Attenuation below the identity line reflects the known compression of dynamic range in microarray relative to RNA-seq data. GSVA, gene set variation analysis; FAO, fatty acid oxidation; OXPHOS, oxidative phosphorylation; PPP, pentose phosphate pathway; ACRG, Asian Cancer Research Group.\u003c/p\u003e","description":"","filename":"Fig8.png","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/04028884c451d2b416ecca4a.png"},{"id":106146818,"identity":"3f28d008-81a6-4c17-a80b-2713530d2d10","added_by":"auto","created_at":"2026-04-04 11:41:22","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":15107199,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/9c4f7b96-2cea-489b-93ab-cf0bf75535fe.pdf"},{"id":105359431,"identity":"9348eaa6-60ab-4ae5-97e0-039418ad23ea","added_by":"auto","created_at":"2026-03-25 07:28:11","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":29777,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable.docx","url":"https://assets-eu.researchsquare.com/files/rs-9147848/v1/7c04a6c95955326835a84337.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Metabolic Reprogramming Across Molecular Subtypes of Gastric Cancer","fulltext":[{"header":"INTRODUCTION","content":"Gastric cancer is the 5th most common malignancy and leading cause of cancer-related mortality worldwide [1]. Despite the noteworthy advances in surgical techniques, perioperative chemotherapy, and targeted immunotherapy, five-year overall survival rates remain below 30% in most populations [2, 3]. The molecular characterization of gastric adenocarcinoma by The Cancer Genome Atlas (TCGA) Research Network identified four biologically distinct subtypes: chromosomal instability (CIN), microsatellite instability (MSI), genomically stable (GS), and Epstein-Barr virus-positive (EBV) [4]. This molecular taxonomy has substantially informed our understanding of gastric cancer biology; however, it has not yet translated into meaningful therapeutic advances targeting these molecular features, underscoring the need for orthogonal axes of tumor stratification that can identify clinically actionable vulnerabilities.\nMetabolic reprogramming has emerged as a fundamental hallmark of malignancy, enabling cancer cells to meet the bioenergetic and biosynthetic demands of uncontrolled proliferation [5]. Cancer metabolism now extends far beyond glycolytic reprogramming to encompass oxidative phosphorylation (OXPHOS), fatty acid oxidation (FAO), pentose phosphate pathway activation, glutamine addiction, and lactate metabolism. These metabolic adaptations are not uniform across cancer types or even within individual tumors; rather, metabolic phenotypes exhibit substantial inter- and intra-tumor heterogeneity shaped by the oncogenotype, microenvironmental factors, and tissue of origin [6]. This complexity poses a significant challenge to the development of effective metabolic therapies but simultaneously reveals a rich landscape of subtype-specific vulnerabilities amenable to precision targeting. \nAs demonstrated in other solid tumors, molecular subtypes align with distinct metabolic dependencies that carry prognostic significance: in colorectal cancer, the metabolic CMS3 subtype results in marked dysregulation of glycolytic and biosynthetic pathways [7], whereas in breast cancer, HER2-positive and triple-negative subtypes present divergent lipid and glutamine metabolic profiles associated with clinical outcomes [8]. These observations establish a precedent for integrating metabolic profiling with genomic classification to reveal subtype-specific vulnerabilities.\nIn gastric cancer, individual metabolic pathways have been studied in isolation — glycolytic reprogramming linked to HIF-1alpha upregulation, OXPHOS alterations in intestinal-type tumors, and lipid metabolic reprogramming associated with chemoresistance — but none of these studies integrated multiple pathways across all four TCGA molecular subtypes simultaneously. While several recent studies have defined metabolic subtypes in gastric cancer using GSVA-based pathway scoring [9-12], none have mapped these findings directly onto the established four-class TCGA genomic taxonomy (CIN/MSI/GS/EBV), assessed subtype-specific metabolic survival associations with multivariable adjustment and bootstrap validation, or examined the transcriptional regulatory architecture coordinating these phenotypes. The relationship between TCGA molecular classification and metabolic pathway activity thus remains uncharacterised. \nTo address this gap, a comprehensive analysis of metabolic reprogramming across gastric cancer molecular subtypes was performed via RNA-sequencing data from 383 TCGA gastric adenocarcinomas, with external validation in 300 patients from the ACRG cohort (GSE62254). Gene set variation analysis (GSVA) was applied to quantify the activity of six core metabolic pathways — glycolysis, oxidative phosphorylation, fatty acid oxidation, the pentose phosphate pathway, glutamine metabolism, and lactate metabolism [13] — and associations with molecular subtype, clinical outcomes, and transcriptional regulatory mechanisms were assessed.\nThe analyses revealed that MSI high tumors exhibit coordinated hyperactivation of biosynthetic pathways — the pentose phosphate pathway and glutamine metabolism — which associated with favourable survival, potentially through support of anti-tumor immunity. In contrast, a subset of GS tumors with elevated fatty acid oxidation demonstrates markedly worse survival, identifying a high-risk metabolic phenotype within this subtype, which already has the worst overall survival among the four TCGA molecular subtypes. These findings persisted across bootstrap resampling, cutoff point sensitivity analysis, and cross-platform external validation, and remained significant after controlling for the proliferation rate and tumor microenvironment composition. Mechanistic investigations revealed that MYC, SREBF1, and AMPK signalling are key transcriptional coordinators that are differentially active across subtypes.\n"},{"header":"Methods","content":"Study Cohorts\nRNA sequencing data from the TCGA-STAD project were analyzed. The dataset comprised 383 primary gastric adenocarcinoma samples with associated clinical and molecular annotations (Genomic Data Commons portal, accessed February 2024). Molecular subtype classifications (CIN, MSI, GS, and EBV) were obtained from the TCGA Research Network [4]. The primary clinical outcome was overall survival, which was defined as the time from the date of initial pathological diagnosis to the date of death from any cause or was censored at the date of last known follow-up for patients alive at the data lock. Clinical annotations, including survival time, vital status, age at diagnosis, sex, and pathological tumor stage, were obtained from the TCGA clinical data supplement. Molecular subtype classification was available for 250 of 383 TCGA-STAD tumors (CIN, n=123; EBV, n=25; GS, n=51; MSI, n=51). The remaining 133 tumors lacked molecular subtype annotation in the TCGA clinical data and were excluded from all subtype-stratified analyses. Overall pathway activity quantification (GSVA scoring, correlation analyses, and tumor microenvironment analyses) was performed across all 383 samples. The ACRG cohort (n=300; GEO: GSE62254; Affymetrix microarray) was analyzed as a validation cohort [14, 15]. The TCGA and ACRG use nonequivalent classification systems; external validation was therefore restricted to cohort-level pathway coordination analyses. Subtype-specific findings are considered discovery-level pending validation in cohorts with TCGA-equivalent genomic classification.\nRNA Sequencing Data Processing\nNormalized gene expression values (TPMs) were obtained from the TCGA GDC portal and log2-transformed (log2[TPM+1]). Samples missing clinical outcome data, with inconsistent subtype assignments, or flagged as PCA outliers (\u003e3 SDs from the cohort median) were excluded; 383 samples were retained.\nMetabolic Pathway Gene Set Curation\nSix core metabolic pathway gene sets were defined from KEGG pathway annotations [16]: glycolysis (hsa00010), OXPHOS (hsa00190), FAO (hsa00071), PPP (hsa00030), glutamine metabolism (hsa00250/hsa00330), and a manually curated lactate gene set (LDHA, LDHB, LDHC, LDHD; SLC16A1, SLC16A3, SLC16A7, SLC16A8). The complete list of genes is provided in Supplementary Table S1.\nPathway Activity Quantification\nGene set variation analysis (GSVA, v1.48.0) was used to quantify pathway-level activity [13]. Parameters: kcdf='Gaussian' (log2-TPM data); min.sz=10; max.sz=500; mx.diff=TRUE; abs.ranking=FALSE.\nTumor Microenvironment Adjustment\nImmune and stromal scores were derived via the ESTIMATE algorithm [17]. Partial Spearman correlations between pathway GSVA scores were computed while controlling for the immune score, stromal score, and tumor purity simultaneously as continuous covariates, via the ppcor package (v1.1; [18]), to isolate tumor cell-autonomous metabolic co-regulation from microenvironment contributions.\nStatistical Analyses\nSubtype differences in pathway activity were assessed by Kruskal-Wallis test with Dunn post-hoc pairwise comparisons and Benjamini-Hochberg (BH) FDR correction. Overall survival associations were modeled via continuous Cox proportional hazards regression (survival package, v3.5-5; [19]); proportional hazards assumptions were verified using Schoenfeld residuals. Pathway scores were dichotomized at the within-subtype median for Kaplan-Meier visualization only; all Cox models used continuous scores. Survival curves were generated by using survminer (v0.4.9;[20]). BH-FDR correction was applied separately for each analytical tier: across n=18 tests for subtype-stratified Cox analyses (6 pathways × 3 subtypes with sufficient sample size; EBV excluded, n=25) and across n=6 tests for cohort-level survival models.\nBootstrap Validation\nThe stability of the survival findings was assessed by bootstrap resampling (1,000 iterations); 95% CIs were calculated by the percentile method. Findings were considered robust if the bootstrap CI excluded HR=1.0.\nCutpoint Sensitivity Analysis\nCutpoints from the 25th to 75th percentiles (5% increments) were tested; findings were considered robust if the effect direction was consistent and if significance was achieved at ≥2 nonadjacent thresholds.\nTranscription Factors and Mechanistic Analysis\nTwenty transcription factors representing major metabolic regulatory axes were analysed: hypoxia response (HIF1A, EPAS1); proliferation (MYC, MYCN); tumour suppression (TP53); mTOR signalling (MTOR, RPTOR, RICTOR); energy sensing (PRKAA1, PRKAA2); mitochondrial biogenesis (PPARGC1A, PPARGC1B); nuclear receptors (PPARA, PPARG, PPARD); lipid metabolism (SREBF1, SREBF2); oxidative stress response (NFE2L2); and autophagy regulation (TFEB). Spearman correlations between log2(TPM+1) TF expression and each of the six GSVA pathway scores were computed (n=120 TF-pathway pairs), with BH-FDR correction across all tests. A composite proliferation index was calculated as the mean log2(TPM+1) expression across 11 proliferation markers: MKI67, PCNA, TOP2A, CDK1, CDK2, CDK4, CDK6, CCNA2, CCNB1, CCND1, and CCNE1.\nPartial Spearman correlations within MSI tumors controlling for the proliferation score (ppcor v1.1; [18]) were computed for all 15 pathway pairs to assess proliferation-independent coregulation. Gene-level differential expression between MSI and GS tumors was assessed via Welch's t-test with BH-FDR correction within each pathway (FDR\u003c0.05, |log2FC|\u003e0.25 threshold).\nExternal Validation\nAffymetrix microarray data (GSE62254) were RMA-normalized and log2-transformed; multiple-probe genes were represented by the highest-mean probe. GSVA pathway enrichment scores were computed identically to those of TCGA (same gene sets, same parameters). Findings were considered validated if the Spearman correlation direction was preserved in the ACRG and attenuation was within the ranges expected for RNA-seq versus microarray platforms [21, 22].\nSoftware and Statistical Environment\nAll analyses used R v4.3.1 with GSVA (v1.48.0), survival (v3.5-5), survminer (v0.4.9), ppcor (v1.1), tidyverse (v2.0.0), ggplot2 (v3.4.2), and pheatmap (v1.0.12).\n"},{"header":"RESULTS","content":"\u003cp\u003eThe activity of six core metabolic pathways was quantified across 383 gastric adenocarcinomas. Marked intertumor variability was observed across all pathways (IQR: 0.421\u0026ndash;0.790; Figure 1A; Table 1). All 15 pairwise pathway correlations were positive and significant after FDR correction (all FDR\u0026lt;0.001; Figure 1B), the strongest being glycolysis\u0026ndash;lactate (r=0.76), glycolysis\u0026ndash;OXPHOS (r=0.68), and glutamine\u0026ndash;PPP (r=0.64). Principal component analysis revealed that PC1 (66.0%) represented a general metabolic activity axis whereas PC2 (11.9%) distinguished biosynthetic (PPP/glutamine) from lipid-oxidative (FAO/lactate) phenotypes (Figure 1C).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTumor-Intrinsic Metabolic Signals\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe immune and stromal scores varied substantially across the cohort and differed significantly across molecular subtypes (Kruskal-Wallis, both P\u0026lt;0.001; Figure 2A-B). The metabolic pathway GSVA scores showed modest inverse associations with the microenvironment scores (immune score vs glycolysis: r=\u0026minus;0.26; vs glutamine: r=\u0026minus;0.32; stromal score vs FAO: r=\u0026minus;0.18; all FDRs\u0026lt;0.001), which is consistent with transcriptional dilution in high-infiltration samples rather than genuine metabolic suppression. Partial Spearman correlation analysis controlling for immune and stromal scores revealed that all 15 pairwise pathway correlations retained statistical significance at FDR\u0026lt;0.001 (Figure 2C-D, Table 2), confirming that interpathway metabolic coordination reflects tumor-intrinsic reprogramming.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSubtype-Specific Metabolic Phenotypes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSignificant differences in metabolic pathway activity were observed across all four molecular subtypes (CIN, n=123; EBV, n=25; GS, n=51; MSI, n=51; n=250 of 383 tumors with subtype annotation). All six pathways showed statistically significant subtype stratification (Kruskal-Wallis, all FDR\u0026lt;0.001; Figure 3B).\u003c/p\u003e\n\u003cp\u003eMSI high tumors displayed the highest overall metabolic activity, with elevated mean GSVA scores for glycolysis (0.19 vs. \u0026minus;0.16 in GS; Dunn FDR\u0026lt;0.001), OXPHOS (0.33 vs. \u0026minus;0.18; FDR\u0026lt;0.001), PPP (0.20 vs. \u0026minus;0.23; FDR\u0026lt;0.001), and glutamine metabolism (0.18 vs. \u0026minus;0.24; FDR\u0026lt;0.001), defining a metabolic hyperactivity phenotype consistent with the high proliferative index and immune-active state of microsatellite-unstable tumors. EBV-positive tumors presented significantly elevated PPP activity (mean: 0.30; FDR\u0026lt;0.001 vs. GS) and glutamine scores (FDR\u0026lt;0.01 vs. GS). GS tumors demonstrated the lowest mean metabolic activity across most pathways, with negative scores for glycolysis, OXPHOS, PPP, and glutamine. FAO was an exception, showing wide intrasubtype variability in GS tumors (IQR: \u0026minus;0.305 to 0.283), suggesting fatty acid oxidation may represent a variable metabolic adaptation in genomically stable disease. Within-subtype pathway correlation analysis revealed distinct coordination signatures: GS tumors showed the tightest overall coupling despite low metabolic flux (glycolysis\u0026ndash;glutamine r=0.85; glycolysis\u0026ndash;PPP r=0.82; both FDR\u0026lt;0.001), while MSI tumors exhibited selective coordination within the biosynthetic axis (glycolysis\u0026ndash;glutamine r=0.69; PPP\u0026ndash;glutamine r=0.55; both FDR\u0026lt;0.001). EBV tumors showed strong FAO\u0026ndash;glutamine coordination (r=0.73; FDR\u0026lt;0.001; Figure 3C; Table 3).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSubtype-Specific Metabolic Vulnerabilities and Overall Survival\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSurvival analysis across the full cohort revealed no significant associations between individual pathway activity and overall survival (all P\u0026gt;0.05; Figure S1), consistent with subtype-specific effects being obscured by tumor heterogeneity. Subtype-stratified Cox regression was therefore performed within each subtype with sufficient sample size (CIN n=123, 46 events; GS n=51, 21 events; MSI n=51, 17 events); EBV tumors (n=25) were excluded due to insufficient power.\u003c/p\u003e\n\u003cp\u003eWithin GS tumors, high FAO activity was associated with significantly worse overall survival by continuous Cox regression (HR = 4.374, 95% CI: 1.143\u0026ndash;16.732, P = 0.031; Figure 4A). Kaplan-Meier analysis at the within-subtype median did not reach statistical significance (log-rank P = 0.199), consistent with loss of statistical power from dichotomisation at this sample size; the continuous Cox model is the primary analysis throughout. In multivariable Cox regression adjusting for age at diagnosis, sex, and pathological tumor stage, FAO activity remained an independent predictor of worse survival in GS tumors and the effect estimate strengthened (HR = 7.261, 95% CI: 1.604\u0026ndash;32.865, P = 0.010), indicating that stage and age do not confound but rather partially suppress the unadjusted association. Proportional hazards assumptions were satisfied in the multivariable model (Schoenfeld global P = 0.91). This association was specific to GS tumors and was not observed in CIN (P = 0.254), MSI (P = 0.078), or the full cohort, suggesting FAO represents a subtype-specific metabolic vulnerability in genomically stable disease.\u003c/p\u003e\n\u003cp\u003eWithin MSI tumors, elevated biosynthetic pathway activity was associated with improved overall survival: high PPP activity (HR=0.211, 95% CI: 0.061\u0026ndash;0.728, P=0.014; Figure 4B; KM log-rank P=0.11) and high glutamine metabolism (HR=0.169, 95% CI: 0.038\u0026ndash;0.745, P=0.019; Figure 4C; KM log-rank P=0.036) both associated with reduced mortality. For MSI-PPP, the continuous Cox analysis provided stronger evidence than the dichotomised Kaplan-Meier analysis, as the continuous approach retains full information from the GSVA score distribution. These associations were not observed in CIN or GS subtypes (all P\u0026gt;0.1). No associations survived BH-FDR correction across all 18 tested combinations (minimum FDR=0.169 for MSI-PPP; Figure 5), consistent with limited power in small subgroup samples. Bootstrap resampling (1,000 iterations) confirmed stability of all three estimates: GS-FAO bootstrap median HR=4.648 (95% CI: 1.157\u0026ndash;23.028); MSI-PPP HR=0.203 (95% CI: 0.056\u0026ndash;0.682); MSI-Glutamine HR=0.155 (95% CI: 0.029\u0026ndash;0.662), with all bootstrap CIs excluding HR=1.0.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTranscriptional Regulation of Subtype-Specific Metabolic Phenotypes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThirteen of 20 candidate transcription factors showed significant differential expression across subtypes after BH correction (FDR\u0026lt;0.05; Figure 6A, Table S2). SREBF1 (SREBP-1) showed the highest expression in MSI tumors (mean: 7.74) relative to GS tumors (6.62; FDR=1.02\u0026times;10^-12; Figure 6B) and correlated positively with all six pathway GSVA scores (r=0.305\u0026ndash;0.455, all FDR\u0026lt;0.001; Figure 6C). MYC was also elevated in MSI tumors (mean: 7.21 vs. 6.34 in GS; FDR=8.34\u0026times;10^-6), with strongest correlations for glutamine metabolism (r=0.465, FDR\u0026lt;2.2\u0026times;10^-16) and PPP (r=0.316, FDR=2.16\u0026times;10^-6). All three PPAR isoforms (PPARA, PPARG, PPARD) were significantly elevated in MSI relative to GS (all FDR\u0026lt;0.05).\u003c/p\u003e\n\u003cp\u003eGS tumors demonstrated a distinct metabolic stress transcriptional signature. PRKAA2 (AMPKalpha2) expression was highest in GS (mean: 3.18) and lowest in MSI and EBV tumors (mean: 1.06; FDR=5.24\u0026times;10^-12; Figure 6B), with strong negative correlations with all six pathway scores (OXPHOS: r=\u0026minus;0.514; PPP: r=\u0026minus;0.490; Glycolysis: r=\u0026minus;0.394; all FDR\u0026lt;0.01; Figure 6C), consistent with AMPK-mediated suppression of anabolic metabolism. EPAS1 (HIF-2alpha) and TFEB were both highest in GS tumors (FDR=1.38\u0026times;10^-7 and FDR=4.90\u0026times;10^-9 respectively), indicating hypoxic stress and autophagy-lysosomal pathway activation. PPARA expression was specifically and positively correlated with FAO activity (r=0.262, FDR=1.03\u0026times;10^-4) and showed no significant correlation with any other pathway (all FDR\u0026gt;0.10; Figure 6C).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProliferation-Independent Metabolic Coordination in MSI Tumors\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eProliferation scores differed significantly across subtypes (Kruskal-Wallis P=6.6\u0026times;10^-16; Figure 7A), with MSI and CIN tumors highest and GS tumors lowest. Biosynthetic pathways correlated positively with proliferation: glutamine (r=0.564, FDR\u0026lt;2.2\u0026times;10^-16), PPP (r=0.435), glycolysis (r=0.391), and OXPHOS (r=0.375; all FDR\u0026lt;10^-13; Figure 7B). FAO showed no significant correlation with proliferation (r=0.094, P=0.16), confirming fatty acid oxidation as a proliferation-independent metabolic programme.\u003c/p\u003e\n\u003cp\u003ePartial Spearman correlation within MSI tumors (n=51) controlling for proliferation score confirmed that 11 of 15 pathway pairs remained significantly co-regulated (FDR\u0026lt;0.05; Table S3), with the strongest proliferation-adjusted associations for glycolysis\u0026ndash;glutamine (partial r=0.700, FDR=2.30\u0026times;10^-7) and glycolysis\u0026ndash;lactate (partial r=0.690, FDR=2.30\u0026times;10^-7). Several correlations were stronger after proliferation adjustment than before (glycolysis\u0026ndash;lactate: r=0.662 unadjusted vs. 0.690 adjusted; PPP\u0026ndash;lactate: r=0.324 vs. 0.395; Table S3), indicating that these associations reflect genuine metabolic coordination not explained by shared proliferative activity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGene-Level Differential Expression Within Metabolic Pathways\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGene-level differential expression analysis comparing MSI versus GS tumors confirmed significant upregulation of individual enzymes within each pathway (TALDO1, TKT, PGD in PPP; GLUD1, GOT2 in glutamine; CPT2 and ACADVL in FAO/OXPHOS), validating that GSVA pathway scores reflect genuine transcriptional reprogramming at the enzyme level (Table S4).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eExternal Validation of Metabolic Pathway Coordination\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGSVA was applied identically to the ACRG cohort (GSE62254, n=300; Affymetrix microarray). GSVA score distributions were broadly comparable between TCGA and ACRG across all six pathways (Figure 8A). Cross-platform correlation analysis revealed that 13 of 15 pairwise pathway combinations preserved the direction of correlation in ACRG (Figure 8B, 8C; Table 4). The two exceptions were OXPHOS-PPP (TCGA r=0.647, ACRG r=\u0026minus;0.010, near-zero in ACRG) and FAO-PPP (TCGA r=0.362, ACRG r=\u0026minus;0.176, direction reversed in ACRG). These two pairs involve coordination between oxidative phosphorylation or fatty acid oxidation and the pentose phosphate pathway \u0026mdash; a coupling that depends on mitochondrial NADPH flux that is measured differently by RNA-seq and microarray platforms; platform-specific probe behaviour or genuine population-level biological differences between the Western-predominant TCGA and Asian-predominant ACRG cohorts may account for these reversals.\u003c/p\u003e\n\u003cp\u003eCorrelation magnitude was attenuated in ACRG relative to TCGA (most pronounced for Glycolysis-OXPHOS: TCGA r=0.681 vs ACRG r=0.137; most modest for OXPHOS-FAO: r=0.514 vs r=0.453), consistent with known microarray dynamic-range compression. The high direction-preservation rate (13/15) confirms that the broad pattern of metabolic pathway coordination is reproducible across platforms and populations.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe present study demonstrates that gastric cancer molecular subtypes exhibit fundamentally distinct metabolic programmes that extend beyond Warburg-effect glycolysis and carry independent prognostic significance. By applying GSVA across six core metabolic pathways in 383 tumor samples, subtype-specific metabolic vulnerabilities have been identified that align with the underlying genomic and transcriptional architecture of each subtype [4, 13].\u003c/p\u003e\n\u003cp\u003eThe most clinically consequential finding is the association between elevated FAO activity and significantly shorter overall survival in the GS subtype. This observation is consistent with the established oncogenic role of FAO in sustaining tumor cell survival under the nutrient-deprived and hypoxic microenvironments that characterise diffuse-type gastric cancers [4]. Carnitine palmitoyltransferases CPT1 and CPT2 have been shown to promote gastric cancer progression and are negatively associated with patient survival [23]. The transcription factor EPAS1 (HIF-2alpha), which showed the highest expression in the GS subtype among all profiled regulators, is associated with activation of lipid metabolic programmes under hypoxic conditions and may contribute to the metabolic stress phenotype observed in this subtype, though the specific link between EPAS1 and FAO in gastric cancer requires direct functional investigation. Notably, multivariable adjustment for age, sex, and tumor stage strengthened rather than attenuated the FAO hazard ratio (adjusted HR = 7.261 vs. unadjusted HR = 4.374), confirming that the association is not a surrogate for advanced-stage disease and suggesting that stage-related clinical variables partially mask the true magnitude of FAO-driven metabolic risk in this subtype.\u003c/p\u003e\n\u003cp\u003eThe transcriptional stress signature of GS tumors \u0026mdash; defined by high PRKAA2 (AMPKalpha2) expression and strong negative correlations between PRKAA2 and all six metabolic pathway scores \u0026mdash; provides a coherent mechanistic framework for the observed FAO-survival association. AMPK is the primary cellular energy sensor, activated under conditions of low ATP:AMP ratio, and under physiological conditions it suppresses anabolic biosynthesis while promoting FAO to restore energetic homeostasis. The paradox in GS tumors is that while PRKAA2 expression is highest in this subtype, anabolic pathway scores are lowest and FAO activity is uncoupled from overall metabolic activity. This dissociation suggests that GS tumors may sustain constitutive FAO as a survival programme that is no longer subject to AMPK-mediated feedback inhibition \u0026mdash; a pattern consistent with FAO serving a stress-adaptation rather than biosynthetic role in the energy-depleted diffuse gastric cancer microenvironment [24].\u003c/p\u003e\n\u003cp\u003eThe finding that MSI tumors exhibited significantly elevated PPP and Glutamine pathway scores, and that this metabolic configuration was associated with superior overall survival, is biologically coherent with the known molecular features of this subtype. MSI gastric cancers carry exceptionally high mutational burden and are subject to intense immune surveillance, generating substantial oxidative stress. Upregulation of the PPP, through NADPH generation via the oxidative branch, provides a crucial antioxidant defence mechanism [25]. The key PPP enzymes upregulated in MSI tumors \u0026mdash; TALDO1, TKT, and PGD \u0026mdash; participate in both branches of the pathway, pointing toward a coordinately regulated programme of nucleotide biosynthesis and redox defence that supports survival of mismatch repair-deficient tumors.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe concurrent elevation of glutamine pathway activity in MSI tumors, driven by significantly higher expression of GLUD1 and GOT2, is consistent with the role of glutaminolysis in sustaining anaplerotic flux into the TCA cycle in highly proliferative cancers [26]. MYC \u0026mdash; significantly higher in MSI than GS tumors \u0026mdash; is a well-established transcriptional activator of glutaminolytic gene programmes and directly promotes upregulation of GLS and downstream transaminases including GOT2 [27]. The strong positive partial correlation between PPP and Glutamine activity within MSI tumors, persisting after adjusting for all other pathways, indicates that these two programmes are co-regulated \u0026mdash; consistent with the known role of MYC in simultaneously driving nucleotide synthesis and glutamine-fuelled anaplerosis [28].\u003c/p\u003e\n\u003cp\u003eThe enrichment of SREBF1 in MSI tumors, and its positive correlation with all six metabolic pathway scores, suggests that SREBP-1 activation creates a broadly permissive anabolic metabolic environment in gastric cancer rather than selectively activating any single pathway [29], and its overexpression has been associated with worse outcomes in multiple gastrointestinal malignancies [30].\u003c/p\u003e\n\u003cp\u003eSeveral methodological limitations require acknowledgment. All findings derive from bulk RNA-sequencing data, which reflects population-averaged transcriptional states rather than cell-intrinsic metabolic flux; functional validation using metabolic tracing in GS cell lines or patient-derived organoids is required. The subtype-specific survival analyses are based on small subgroup samples (GS n=51, MSI n=51) that limit statistical power, and no subtype-pathway association survived BH-FDR correction across the 18 subtype-stratified tests, underscoring their exploratory nature.\u003c/p\u003e\n\u003cp\u003eExternal validation in the ACRG cohort confirmed direction preservation in 13 of 15 pathway pairs, with attenuation consistent with known platform effects. Subtype-specific survival associations could not be directly validated due to the non-equivalence of TCGA and ACRG classification systems, which is an inherent limitation of cross-cohort validation in gastric cancer.\u003c/p\u003e\n\u003cp\u003eTaken together, these results provide a metabolic framework for the molecular classification of gastric cancer with potential translational relevance. The identification of FAO as a GS-specific vulnerability suggests that CPT1 inhibitors could be explored in the context of diffuse-type GS gastric cancer, where standard cytotoxic regimens achieve poor outcomes [3]. While etomoxir has demonstrated preclinical efficacy, hepatotoxicity at therapeutically relevant doses has limited its clinical translation; perhexiline and next-generation CPT1-selective agents with improved therapeutic indices may be more suitable candidates for evaluation in this molecularly defined subpopulation. In the MSI subtype, the co-elevation of PPP and glutamine pathways \u0026mdash; both of which support antioxidant defence \u0026mdash; may partly explain the favourable response of MSI tumors to immune checkpoint inhibition by maintaining redox homeostasis in a permissive immune microenvironment. Future work integrating single-cell transcriptomics, metabolic flux measurements, and immune phenotyping within TCGA molecular subtypes will be necessary to determine whether the metabolic signatures identified here can be translated into biomarker-driven therapeutic strategies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e The author have no relevant financial or non-financial interests to disclose\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. (2024);74(3):229-63. doi: 10.3322/caac.21834.\u003c/li\u003e\n\u003cli\u003eMaharjan U, Kauppila JH. Survival trends in gastric cancer patients between 1987 and 2016: a population-based cohort study in Finland. Gastric Cancer. (2022);25(6):989-1001. doi: 10.1007/s10120-022-01326-5.\u003c/li\u003e\n\u003cli\u003eSmyth EC, Nilsson M, Grabsch HI, et al. Gastric cancer. Lancet. (2020);396(10251):635-48. doi: 10.1016/S0140-6736(20)31288-5.\u003c/li\u003e\n\u003cli\u003eCancer Genome Atlas Research N. Comprehensive molecular characterization of gastric adenocarcinoma. Nature. (2014);513(7517):202-9. doi: 10.1038/nature13480.\u003c/li\u003e\n\u003cli\u003eHanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. (2011);144(5):646-74. doi: 10.1016/j.cell.2011.02.013.\u003c/li\u003e\n\u003cli\u003eFaubert B, Solmonson A, DeBerardinis RJ. Metabolic reprogramming and cancer progression. Science. (2020);368(6487). doi: 10.1126/science.aaw5473.\u003c/li\u003e\n\u003cli\u003eGuinney J, Dienstmann R, Wang X, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. (2015);21(11):1350-6. doi: 10.1038/nm.3967.\u003c/li\u003e\n\u003cli\u003eZipinotti Dos Santos D, de Souza JC, Pimenta TM, et al. The impact of lipid metabolism on breast cancer: a review about its role in tumorigenesis and immune escape. Cell Commun Signal. (2023);21(1):161. doi: 10.1186/s12964-023-01178-1.\u003c/li\u003e\n\u003cli\u003eChen T, Zhao L, Chen J, et al. Identification of three metabolic subtypes in gastric cancer and the construction of a metabolic pathway-based risk model that predicts the overall survival of GC patients. Front Genet. (2023);14:1094838. doi: 10.3389/fgene.2023.1094838.\u003c/li\u003e\n\u003cli\u003eLin X, Yang P, Wang M, et al. Dissecting gastric cancer heterogeneity and exploring therapeutic strategies using bulk and single-cell transcriptomic analysis and experimental validation of tumor microenvironment and metabolic interplay. Front Pharmacol. (2024);15:1355269. doi: 10.3389/fphar.2024.1355269.\u003c/li\u003e\n\u003cli\u003eTao G, Wen X, Wang X, et al. Bulk and single-cell transcriptome profiling reveal the metabolic heterogeneity in gastric cancer. Sci Rep. (2023);13(1):8787. doi: 10.1038/s41598-023-35395-y.\u003c/li\u003e\n\u003cli\u003eYe Y, Yang W, Ruan X, et al. Metabolism-associated molecular classification of gastric adenocarcinoma. Front Oncol. (2022);12:1024985. doi: 10.3389/fonc.2022.1024985.\u003c/li\u003e\n\u003cli\u003eHanzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. (2013);14:7. doi: 10.1186/1471-2105-14-7.\u003c/li\u003e\n\u003cli\u003eCristescu R, Lee J, Nebozhyn M, et al. Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes. Nat Med. (2015);21(5):449-56. doi: 10.1038/nm.3850.\u003c/li\u003e\n\u003cli\u003eOh SC, Sohn BH, Cheong JH, et al. Clinical and genomic landscape of gastric cancer with a mesenchymal phenotype. Nat Commun. (2018);9(1):1777. doi: 10.1038/s41467-018-04179-8.\u003c/li\u003e\n\u003cli\u003eKanehisa M, Furumichi M, Sato Y, et al. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. (2023);51(D1):D587-D92. doi: 10.1093/nar/gkac963.\u003c/li\u003e\n\u003cli\u003eYoshihara K, Shahmoradgoli M, Martinez E, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. (2013);4:2612. doi: 10.1038/ncomms3612.\u003c/li\u003e\n\u003cli\u003eKim S. ppcor: An R Package for a Fast Calculation to Semi-partial Correlation Coefficients. Commun Stat Appl Methods. (2015);22(6):665-74. doi: 10.5351/CSAM.2015.22.6.665.\u003c/li\u003e\n\u003cli\u003eTherneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model. New York: Springer; 2000.\u003c/li\u003e\n\u003cli\u003eKassambara A, Kosinski M, Biecek P. survminer: Drawing Survival Curves using \u0026apos;ggplot2\u0026apos;. The Comprehensive R Archive Network (CRAN); 2021.\u003c/li\u003e\n\u003cli\u003eWang C, Gong B, Bushel PR, et al. The concordance between RNA-seq and microarray data depends on chemical treatment and transcript abundance. Nat Biotechnol. (2014);32(9):926-32. doi: 10.1038/nbt.3001.\u003c/li\u003e\n\u003cli\u003eConsortium M, Shi L, Reid LH, et al. The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene expression measurements. Nat Biotechnol. (2006);24(9):1151-61. doi: 10.1038/nbt1239.\u003c/li\u003e\n\u003cli\u003eChen T, Wu G, Hu H, et al. Enhanced fatty acid oxidation mediated by CPT1C promotes gastric cancer progression. J Gastrointest Oncol. (2020);11(4):695-707. doi: 10.21037/jgo-20-157.\u003c/li\u003e\n\u003cli\u003eHerzig S, Shaw RJ. AMPK: guardian of metabolism and mitochondrial homeostasis. Nat Rev Mol Cell Biol. (2018);19(2):121-35. doi: 10.1038/nrm.2017.95.\u003c/li\u003e\n\u003cli\u003eTeSlaa T, Ralser M, Fan J, et al. The pentose phosphate pathway in health and disease. Nat Metab. (2023);5(8):1275-89. doi: 10.1038/s42255-023-00863-2.\u003c/li\u003e\n\u003cli\u003eCluntun AA, Lukey MJ, Cerione RA, et al. Glutamine Metabolism in Cancer: Understanding the Heterogeneity. Trends Cancer. (2017);3(3):169-80. doi: 10.1016/j.trecan.2017.01.005.\u003c/li\u003e\n\u003cli\u003eTambay V, Raymond VA, Bilodeau M. MYC Rules: Leading Glutamine Metabolism toward a Distinct Cancer Cell Phenotype. Cancers (Basel). (2021);13(17). doi: 10.3390/cancers13174484.\u003c/li\u003e\n\u003cli\u003eGao P, Tchernyshyov I, Chang TC, et al. c-Myc suppression of miR-23a/b enhances mitochondrial glutaminase expression and glutamine metabolism. Nature. (2009);458(7239):762-5. doi: 10.1038/nature07823.\u003c/li\u003e\n\u003cli\u003eGeng F, Zhong Y, Su H, et al. SREBP-1 upregulates lipophagy to maintain cholesterol homeostasis in brain tumor cells. Cell Rep. (2023);42(7):112790. doi: 10.1016/j.celrep.2023.112790.\u003c/li\u003e\n\u003cli\u003eHe Y, Qi S, Chen L, et al. The roles and mechanisms of SREBP1 in cancer development and drug response. Genes Dis. (2024);11(4):100987. doi: 10.1016/j.gendis.2023.04.022.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1. GSVA metabolic pathway score distributions across TCGA-STAD (n=383)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGSVA enrichment scores for six metabolic pathways across all 383 gastric adenocarcinomas. Values are GSVA enrichment scores (range approximately -1 to +1). SD, standard deviation; IQR, interquartile range; Q1, first quartile; Q3, third quartile.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"620\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePathway\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 40px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003en\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMean\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSD\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMedian\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eQ1\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eQ3\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eIQR\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMin\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMax\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 40px;\"\u003e\n \u003cp\u003e383\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.241\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.163\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.258\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.421\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.627\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.641\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 40px;\"\u003e\n \u003cp\u003e383\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.263\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.172\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.241\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.413\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.699\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.698\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFAO\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 40px;\"\u003e\n \u003cp\u003e383\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e-0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.284\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.198\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.201\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.399\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.731\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.712\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePPP\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 40px;\"\u003e\n \u003cp\u003e383\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.316\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.233\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.380\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.613\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.789\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.782\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlutamine\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 40px;\"\u003e\n \u003cp\u003e383\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e-0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.278\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.009\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.218\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.244\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.462\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.701\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.693\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eLactate\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 40px;\"\u003e\n \u003cp\u003e383\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.271\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.187\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.454\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.683\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.709\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eGSVA, Gene Set Variation Analysis; PPP, pentose phosphate pathway; FAO, fatty acid oxidation; OXPHOS, oxidative phosphorylation. All six pathways showed marked inter-tumour variability with IQR ranging from 0.421 (Glycolysis) to 0.790 (PPP).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2. Metabolic pathway pairwise correlations: unadjusted versus microenvironment-adjusted Spearman coefficients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll 15 pairwise Spearman correlations between metabolic pathway GSVA scores, before and after partial correlation adjustment for immune score, stromal score, and tumour purity (ESTIMATE algorithm). Ranked by unadjusted correlation magnitude.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"560\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePathway 1\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePathway 2\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSpearman r (unadj.)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePartial r (adj.)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eDelta\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eBH-FDR\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.758\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.751\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.718\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.705\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.713\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.698\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eOXPHOS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.681\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.683\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e+0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePPP\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.640\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.621\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eFAO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.514\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.498\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.498\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.481\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.647\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.629\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlutamine\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.612\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.597\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePPP\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.589\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.572\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.554\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eFAO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.421\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.408\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFAO\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.398\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.385\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFAO\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.387\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.374\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFAO\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.362\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e0.351\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e-0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAdj., adjusted by partial Spearman correlation controlling for ESTIMATE immune score, stromal score, and tumour purity simultaneously. Delta = partial r minus unadjusted r. BH-FDR, Benjamini-Hochberg false discovery rate. All 15 pairs retained statistical significance at FDR\u0026lt;0.001 after microenvironment adjustment.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3. Within-subtype pairwise metabolic pathway correlations (selected significant pairs)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSpearman correlation coefficients for selected significant pairwise pathway combinations computed independently within each molecular subtype. Row shading indicates molecular subtype. BH-FDR correction applied within each subtype. Full correlation matrices shown in Figure 3C.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"493\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSubtype\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePathway 1\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePathway 2\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSpearman r\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eBH-FDR\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eCIN (n=123)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eOXPHOS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.72\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eCIN (n=123)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eCIN (n=123)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eOXPHOS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eEBV (n=25)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eFAO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.73\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eEBV (n=25)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eOXPHOS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eEBV (n=25)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGS (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGS (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGS (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eOXPHOS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGS (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eOXPHOS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGS (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMSI (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMSI (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMSI (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMSI (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlycolysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMSI (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eFAO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.022\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 120px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMSI (n=51)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eOXPHOS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 107px;\"\u003e\n \u003cp\u003eFAO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 80px;\"\u003e\n \u003cp\u003e0.041\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eOnly pairs with BH-FDR\u0026lt;0.05 within each subtype are shown. Spearman correlation analysis was performed independently within each subtype without combining data across groups. CIN, chromosomal instability; EBV, Epstein-Barr virus-positive; GS, genomically stable; MSI, microsatellite instability. Sample sizes: CIN n=123, EBV n=25, GS n=51, MSI n=51.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4. Cross-platform validation of metabolic pathway coordination: TCGA-STAD vs ACRG/GSE62254\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSpearman correlation coefficients for all 15 pairwise metabolic pathway combinations in TCGA-STAD (RNA-seq, n=383) and the independent ACRG cohort (Affymetrix microarray, n=300). Attenuation is expressed as the percentage change from TCGA to ACRG relative to the TCGA value.\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"573\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePathway 1\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePathway 2\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSpearman r (TCGA)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSpearman r (ACRG)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eAttenuation (%)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eDirection preserved\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.758\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.399\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-47.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.718\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.375\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-47.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.713\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.315\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-55.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eOXPHOS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.681\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.137\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-79.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePPP\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.640\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.289\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-54.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eFAO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.514\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.453\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-11.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.498\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.221\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-55.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlutamine\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.612\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.287\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-53.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ePPP\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.589\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.241\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-59.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.198\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-65.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGlycolysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eFAO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.421\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.187\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-55.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFAO\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eGlutamine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.398\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.163\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-59.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFAO\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLactate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.387\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.142\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-63.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eOXPHOS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.647\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-101.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eNo (near zero)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eFAO\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003ePPP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.362\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-0.176\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e-148.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eNo (reversed)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAttenuation (%) = ((ACRG r - TCGA r) / |TCGA r|) x 100. Negative values indicate lower correlation in ACRG. Direction preserved = same algebraic sign in both cohorts. Pairs near zero in ACRG (|r|\u0026lt;0.05) classified as direction uncertain. The two pairs showing reversal (OXPHOS-PPP and FAO-PPP) are highlighted. TCGA, The Cancer Genome Atlas; ACRG, Asian Cancer Research Group; OXPHOS, oxidative phosphorylation; FAO, fatty acid oxidation; PPP, pentose phosphate pathway.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Stomach Neoplasms, Metabolic Reprogramming, Gene Expression Profiling, Fatty Acid Oxidation, Microsatellite Instability","lastPublishedDoi":"10.21203/rs.3.rs-9147848/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9147848/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003ePurpose \u003c/strong\u003eGastric cancer molecular subtypes defined by The Cancer Genome Atlas — chromosomal instability (CIN), microsatellite instability (MSI), genomically stable (GS), and Epstein-Barr virus-positive (EBV) — carry distinct biological features, yet their metabolic programmes and clinical significance remain uncharacterised across all four subtypes simultaneously.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods \u003c/strong\u003eGene Set Variation Analysis (GSVA) was applied to RNA-sequencing data from 383 TCGA gastric adenocarcinomas to quantify activity of six core metabolic pathways: glycolysis, oxidative phosphorylation, fatty acid oxidation (FAO), the pentose phosphate pathway (PPP), glutamine metabolism, and lactate metabolism. Subtype-specific survival associations were assessed by continuous Cox proportional hazards regression with multivariable adjustment. Transcriptional regulatory mechanisms were investigated through expression analysis of 20 metabolic transcription factors. External validation was performed in the independent ACRG cohort (GSE62254; n=300).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults \u003c/strong\u003eAll six pathways showed significant subtype stratification (Kruskal-Wallis, all FDR\u0026lt;0.001). Within GS tumours, elevated FAO activity was independently associated with worse overall survival after multivariable adjustment (HR=7.261, 95% CI: 1.604–32.865, P=0.010). Within MSI tumours, elevated PPP (HR=0.211, P=0.014) and glutamine pathway activity (HR=0.169, P=0.019) were associated with improved survival, potentially reflecting antioxidant support of anti-tumour immunity. PRKAA2, SREBF1, and MYC were identified as key subtype-specific transcriptional regulators. Cross-platform validation confirmed direction preservation of metabolic pathway coordination in 13 of 15 pathway pairs in the ACRG cohort.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions \u003c/strong\u003eGastric cancer molecular subtypes exhibit distinct metabolic vulnerabilities with independent prognostic significance. The GS-FAO association identifies a high-risk metabolic phenotype amenable to CPT1 inhibitor evaluation, while MSI biosynthetic pathway elevation may underpin immune checkpoint sensitivity.\u003c/p\u003e","manuscriptTitle":"Metabolic Reprogramming Across Molecular Subtypes of Gastric Cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-25 07:27:48","doi":"10.21203/rs.3.rs-9147848/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"6f82741c-c518-4380-adf3-dad94fb91852","owner":[],"postedDate":"March 25th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-04T11:38:29+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-25 07:27:48","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9147848","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9147848","identity":"rs-9147848","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}