Advances in molecular mechanism of gastric cancer peritoneal metastasis

Abstract

Peritoneal metastasis represents one of the most frequent patterns of metastasis in gastric cancer and is associated with an exceedingly poor prognosis, demonstrated by a five-year survival rate of less than 2%. While treatment options remain limited for affected patients, advances in targeted therapy and immunotherapy continue to emerge. In this review, we synthesize published molecular biology research to summarize the diverse mechanisms underlying gastric cancer peritoneal metastasis. Our aim is to enhance the understanding of this process and to identify key molecules, signaling pathways, and cellular immune components involved, thereby offering new perspectives for precise diagnosis and treatment.

Key Words: Gastric cancer; Peritoneal metastasis; Mutation; Epithelial-mesenchymal transition; Extracellular matrix remodeling; Tumor microenvironment; Hypoxia; Anoikis resistance; Metabolic reprogramming; Epigenetics

Core Tip: Gastric cancer peritoneal metastasis involves multiple complex biological processes. These include genetic mutations, epithelial-mesenchymal transition, extracellular matrix remodeling, tumor microenvironment interactions, hypoxia-driven responses, anoikis resistance, metabolic reprogramming, and epigenetic regulation. This review systematically elucidates the synergistic interactions among these networks. It also highlights their potential as diagnostic biomarkers and therapeutic targets.

INTRODUCTION

Gastric cancer (GC) is a malignant tumor of the digestive tract. It ranks as the fifth most common cancer globally and the fourth leading cause of cancer death[1]. Metastasis is the primary cause of cancer-related mortality in GC patients, with the peritoneum serving as a favorable site for GC cell growth and the formation of metastatic lesions[2]. More than 50% of GC patients have peritoneal metastasis (PM) at the time of death[3]. Despite recent advances in treatment, such as cytoreductive surgery plus hyperthermic intraperitoneal chemotherapy, PM remains the leading cause of GC related death[4,5]. Additionally, the application of treatments such as cytoreductive surgery and hyperthermic intraperitoneal chemotherapy is limited to selected patients with a low peritoneal tumor burden. The high recurrence rate further indicates an urgent need for more effective systemic and targeted therapies. Currently, there is a lack of reliable early diagnostic biomarkers for peritoneal micrometastases. Moreover, validated therapeutic targets remain absent. These limitations mean that most patients with advanced GCPM are incurable. Therefore, elucidating the underlying molecular mechanisms is essential to overcoming these clinical bottlenecks.

Recent years have witnessed rapid progress in elucidating the molecular landscape of GCPM, fueled by developments in molecular biology. This has illuminated the complex dynamics between tumor cells and their peritoneal milieu, establishing the key contributions of diverse cytokines, chemokines, and signaling pathways in orchestrating metastasis[6-8]. At the same time, the process of epithelial-mesenchymal transition (EMT) of tumor cells has also been confirmed to be closely related to GCPM and the expression changes of key genes and proteins regulating EMT can affect the invasion ability of tumor cells[9,10]. In addition, metabolic reprogramming of tumor cells, abnormal epigenetic mechanisms and changes in immune microenvironment also play an important role in GC peritoneal metastasis (GCPM). This review aims to synthesize the latest research on the molecular mechanisms of GCPM, with the goal of informing both scientific understanding and clinical practice. By systematically summarizing these advances, we seek not only to elucidate the underlying biology of GCPM but also to uncover new diagnostic and therapeutic opportunities, thereby supporting collaborative efforts in preventing and treating this condition.

STRUCTURAL, FUNCTIONAL, ANATOMICAL CHARACTERISTICS OF THE PERITONEUM AND THE NATURAL FORMATION OF GCPM

The peritoneum is the largest and most intricately distributed serous membrane in the human body. It is composed of a single layer of mesothelial cells and a small amount of connective tissue. It is divided into the parietal peritoneum and the visceral peritoneum. The parietal peritoneum lines the inner surface of the abdominal wall, while the visceral peritoneum covers the surfaces of the abdominal organs[11]. The peritoneum has three core physiological functions based on its unique anatomy[11,12]: (1) Mechanical support and protection: It envelops organs, reduces friction, and adipose pads provide cushioning and insulation; (2) Metabolism and lubrication: Mesothelial cells secrete serous fluid for lubrication, and its semipermeable nature allows fluid and solute transport; and (3) Immune defense: It acts as a critical immune barrier where resident immune cells eliminate pathogens via coordinated inflammation. Peritoneal dysfunction can lead to microenvironmental imbalance, resulting in pathological conditions such as peritonitis, ascites, and peritoneal carcinomatosis. In modern oncology, the peritoneum is regarded as a common “soil” for metastatic dissemination, while the process by which “seeds” (cancer cells) successfully “take root and germinate” involves multiple factors and represents a key focus and challenge in current research. The anatomical basis for the peritoneum’s susceptibility to metastasis is mainly reflected in four aspects. First, its extensive surface area: The total area of the human peritoneum reaches 1.8 square meters, providing a vast “landing zone” for circulating tumor cells and serving as a critical physical interface for their initial adhesion and subsequent clonal proliferation[11]. Second, its well-developed vascular and lymphatic networks: This dense vascular network, essential for physiological transport, is exploited by cancer cells under pathological conditions, acting as a natural “channel” for them to directly reach the peritoneal surface and establish metastatic lesions[13]. Third, the immune niche reversal of milky spot structures: The peritoneal surface is dotted with secondary lymphoid organs known as “milky spots”, which are composed of immune cells including macrophages, T cells, and B cells[14]. This “outpost” is originally responsible for immune defense. However, its unique immune niche is often co-opted by cancer cells, turning it into a “hotspot” that supports cancer cell adhesion and early colonization. This process is therefore recognized as the initial step in PM. This discovery reveals the functional reversal of immune structures in tumor metastasis, yet the specific roles of distinct immune cell subsets within milky spots during the initial stages of metastasis remain to be systematically elucidated. Fourth, the anatomical channel of the subperitoneal space: The deeper layer beneath the peritoneum is not a solid tissue but rather contains a potential space, the subperitoneal space, composed of loose, fat-rich connective tissue that harbors embedded nerves, blood vessels, and lymphatic networks[15]. Under physiological conditions, this space provides cushioning and connectivity; under pathological conditions, it unfortunately offers convenient pathways for cancer cell invasion and room for growth, thereby functioning as a critical step for local tumor dissemination. This anatomical feature explains why PM, once established, often presents as diffuse dissemination; however, the structural differences in the subperitoneal space across various regions and their impact on metastatic propensity still require further investigation. Furthermore, the omentum is packed with metabolically active adipocytes, forming a prominent visceral fat depot. This unique lipid-rich microenvironment is often exploited by tumor cells under pathological conditions, transforming the omentum from a protective organ into a “fertile soil” and “hotbed” that promotes metastatic progression[16]. This understanding provides important clues for elucidating the organotropism of GCPM, yet the complex metabolic crosstalk mechanisms between adipocytes and tumor cells still require in-depth investigation. In summary, the susceptibility of the peritoneum as a metastatic “soil” arises from the synergistic interplay of multiple anatomical factors, including its extensive surface area, well-developed vascular and lymphatic networks, the immune niche reversal of milky spots, the anatomical channel of the subperitoneal space, and the lipid microenvironment of the omentum. These structural features collectively constitute the biological foundation for the survival, colonization, and dissemination of GC cells within the peritoneal cavity, and also provide an important tissue context for subsequent investigations into molecular mechanisms. It has been documented that the occurrence of PM is considerably elevated in ovarian cancer and GC. Other cancer types, including colorectal, hepatic, and pancreatic carcinomas, exhibit a lower propensity for such metastasis, as summarized in Table 1.

Table 1 Incidence of peritoneal metastasis in major abdominal and gynecological cancers.

Type of cancer	Incidence of peritoneal metastases	Ref.
Ovarian cancer	> 75%	[187]
Gastric cancer	33%-66%	[188,189]
Colorectal cancer	5%-15%	[190]
Liver cancer	2%-18%	[191]
Pancreatic cancer	10%	[192]

This table summarizes the widely reported frequency of peritoneal metastasis in different primary tumor types, highlighting that ovarian and gastric cancers exhibit the highest tropism for the peritoneum.

From this, it can be seen that GC cells have extremely strong characteristics of peritoneal invasion and a tendency to metastasize. However, GCPM is a complex and multi-stage process that involves the interconnection of several aspects (Figure 1): (1) Tumor cell detachment: The detachment of GC cells from the primary lesion is the initial step of PM. The direct pathway of PM involves tumor cells invading the gastric wall, traversing the serous membrane, and entering the peritoneal cavity, where they spread and implant to establish secondary lesions. This pathogenic cascade may be facilitated by a synergy between augmented invasive capabilities of the tumor cells and diminished expression of intercellular adhesion molecules[17,18]. Furthermore, dissociated GC cells can enter the vasculature and, via hematogenous circulation, reach the peritoneal capillary network. The rich vascularity of the peritoneum facilitates the arrest and invasion of these cells into the peritoneal tissue, leading to the formation of metastatic deposits[19,20]. In addition, lymphatic metastasis constitutes a prevalent pattern of spread in GC. Cells that have broken away from the primary tumor first invade the regional gastric lymphatics and are carried through the lymphatic system to establish secondary tumors in distant nodes. These node-based deposits can subsequently serve as a source for peritoneal seeding. Furthermore, direct invasion of peritoneal lymphatic vessels by tumor cells, followed by their extravasation, also represents a direct mechanism for the formation of peritoneal metastatic lesions[21]; (2) Anoikis resistance: When cells lose contact with the extracellular matrix (ECM) or adjacent cells, they may trigger a programmed cell death process, which is known as anoikis[22]. Following detachment, tumor cells disseminated into body cavities or vessels are exposed to anoikis, a form of cell death initiated by the loss of proper matrix attachment. GC cells, however, circumvent this intrinsically programmed demise by activating pro-survival signaling pathways, including YAP/TEA domain transcription factor 1 (TEAD1)[23]. This capability to resist anoikis is crucial for their survival in suspension and significantly contributes to their potential for PM; (3) Adhesion: Free GC cells adhere to the peritoneal surface by interacting with peritoneal mesothelial cells at suitable locations. Adhesion molecules such as E-cadherin, N-cadherin[10] and integrin[24] play an important role in this process. In addition, ECM components such as fibronectin and laminin also provide a matrix for tumor cells to adhere to[25]; (4) Invasion: GC cells adhering to the peritoneum will further invade the peritoneal tissue, destroy the mesothelial cell layer and basement membrane, and infiltrate into deep tissues. This process involves the secretion of enzymes such as matrix metalloproteinases (MMPs)[26,27] and cathepsins[9,28], which degrade the ECM and provide a channel for GC cell invasion; (5) Proliferation: The ability of unlimited proliferation is one of the typical characteristics of tumor cells. GC cells that invade the subperitoneal tissues promote cell proliferation by autocrine activating factors[29] or stimulating relevant interstitial paracrine factors in the tumor microenvironment (TME) and integrating with their corresponding receptors[30]. During this process, tumor-associated macrophages (TAMs) and cancer-associated fibroblasts (CAFs) play an indispensable role[31,32]; and (6) Angiogenesis: After GC cells form small metastatic foci within the peritoneum, they will secrete angiogenic factors[33,34], inducing the formation of new blood vessels. These new blood vessels provide oxygen and nutrients to the tumor cells, promoting their further growth and spread.

Figure 1 The stepwise process of gastric cancer peritoneal metastasis. This schematic illustrates the sequential steps by which gastric cancer cells disseminate from the primary tumor and establish metastatic lesions in the peritoneum. The process involves six critical phases: (1) Shedding: Gastric cancer cells detach from the primary tumor by downregulating intercellular adhesion molecules; (2) Anoikis resistance: Detached cells survive in the peritoneal fluid by activating pro-survival pathways; (3) Adhesion: Free-floating cells adhere to the peritoneal mesothelial layer via adhesion molecules; (4) Invasion: Adhered cells secrete matrix metalloproteinases and cathepsins to degrade the extracellular matrix, invading the subperitoneal space; (5) Proliferation: Invaded cells proliferate, forming micrometastases; and (6) Angiogenesis: The growing metastatic foci secrete angiogenic factors (e.g., vascular endothelial growth factor), inducing new blood vessel formation to supply nutrients and oxygen.

RESEARCH ON THE MECHANISM OF PM OF GC

Mutation

Targeted therapies and immunotherapies have emerged in recent years. However, patients with GCPM still face a dismal prognosis. Under standard systemic chemotherapy, their median overall survival is less than one year[35]. Improving clinical outcomes is hindered by two major barriers. First, there is a lack of reliable genomic biomarkers to guide treatment selection. Second, effective targeted strategies specifically for the peritoneal metastatic route remain absent. Therefore, identifying key driver mutations is critically important. Understanding their route-specific enrichment patterns also holds significant translational value. This knowledge can improve patient stratification and facilitate the development of precision therapies. Gene mutations are core driving events in tumor initiation and progression, primarily characterized by inactivation of tumor suppressor genes, activation of proto-oncogenes, and defects in DNA repair genes. Although familial aggregation is an important risk factor for GC, one study shows that 37% of GC patients have a family history in first-degree relatives[36], suggesting that sporadic mutations play a dominant role in the pathogenesis of GC. Genomic analysis of clinical patient tissue specimens based on metastatic routes reveals important distinctions[37]. Primary gastric tumors of the chromosomal instability subtype predominantly undergo hematogenous metastasis. Their metastatic lesions are frequently enriched in activating amplifications of pathways such as RTK/RAS, phosphatidylinositide 3-kinase (PI3K)/protein kinase B (AKT), and MYC. In contrast, the genomically stable subtype is closely associated with PM. This subtype is characterized by small insertions, deletions, or nonsense mutations within protein-coding sequences that result in loss of protein function. This study provides an important framework for understanding the molecular basis of GC metastasis. However, its limited sample size and the fact that some patients received preoperative chemotherapy warrant further validation. Future studies should involve larger cohorts integrated with multi-omics data.

Hereditary diffuse GC is an autosomal dominant syndrome driven by germline mutations in the CDH1 gene. The encoded E-cadherin not only mediates intercellular adhesion but also participates in critical processes such as mitotic spindle orientation and mechanotransduction[38]. Inactivation of CDH1 leads to aberrant spatial positioning of daughter cells following cell division, progressively disrupting the architecture of the gastric mucosal epithelium and promoting cell shedding, thereby serving as a sustained driver of tumor initiation and progression[39,40]. A clinical multi-omics analysis further confirmed that the expression profiles of CDH1, ELF3, and PIGR are closely associated with the PM potential of primary tumors[41]. A preclinical study investigated the inhibitory effect of defactinib on the FAK signaling pathway[42]. This pathway is activated by CDH1 loss. The study also explored potential combination targets in diffuse GC. The results confirmed the therapeutic efficacy of defactinib. They also highlighted the promising potential of combination targeted therapy. Interestingly, a recent large scale cohort study confirmed that signet ring cells (SRCs) are nearly universal in CDH1 carriers[43]. However, most lesions remain confined to stage pT1aN0. Very few progress to advanced GC. Therefore, SRCs should not be equated with clinically actionable cancer. This is especially important when mucosal abnormalities such as ulcers are absent. This understanding helps avoid overtreatment. It also provides critical evidence based support for preserving the stomach in eligible patients. Therefore, CDH1 mutation status should be incorporated into germline testing for hereditary diffuse GC families. This testing helps guide decisions about prophylactic gastrectomy. In addition, CDH1 expression levels in primary tumors can serve as a routine immunohistochemical marker. This marker helps assess the risk of PM. TP53 mutations lead to loss of cell cycle regulation and apoptotic function, and are significantly associated with the risk of liver metastasis and PM in GC[44,45]. Molecular subtyping studies based on whole-exome sequencing have classified GC into four subtypes, among which subtypes 3 and 4, characterized by ARID1A deficiency, account for a higher proportion of genomically stable-type tumors and exhibit a significantly higher incidence of PM compared to other subtypes. Further in vitro and preclinical mechanistic studies have shown that ARID1A deficiency activates the mammalian target of rapamycin (mTOR) signaling pathway and upregulates SRY box 9 (SOX9) expression, conferring enhanced migratory and invasive capabilities to GC cells[46,47]. A prospective study further demonstrated the prognostic value of baseline TP53 mutation status in circulating tumor DNA (ctDNA)[48]. The study focused on GCPM patients receiving intraperitoneal and systemic combination chemotherapy. Patients with TP53 mutations had significantly worse progression-free and overall survival than those with wild-type TP53. The median overall survival was 15.78 months for the mutation group, compared to 29.26 months for the wild-type group. This finding has direct clinical utility. Baseline ctDNA testing for TP53 mutations could be implemented in routine practice. It can help stratify GCPM patients who are receiving intraperitoneal chemotherapy. Patients with wild-type TP53 are more likely to achieve long-term survival. In contrast, those with mutant TP53 may require alternative or combination regimens.

Through in vivo genome-wide CRISPR/Cas9 screening (a preclinical study), NF2 and RASA1 were identified as metastasis suppressor genes in GC. Their loss enhances tumor stemness and metastatic capacity through synergistic activation of the Wnt and YAP signaling pathways[49]. In the NF2/RASA1 deficient model, combined use of Bcl-2 and YAP inhibitors showed a synthetic lethal effect. A preliminary clinical trial in patients with gastroesophageal cancer showed that the BCL-2 inhibitor AT101, when combined with chemotherapy, significantly enhanced antitumor activity[50]. This combination strategy achieved notable clinical complete responses. It also led to encouraging overall survival benefits. Another preclinical study revealed a drug resistance mechanism in HER2 positive GC[51]. It also identified a strategy to reverse this resistance. Specifically, long term treatment with trastuzumab induced YAP transcription in cancer cells. This induction occurred through chromatin remodeling. YAP transcription was identified as a key factor in drug resistance. Combining trastuzumab with verteporfin, a YAP inhibitor, produced an effective synergistic effect. This combination successfully reversed the drug resistance process. These observations warrant further validation in GCPM specific models and clinical trials. Key unanswered questions include the prevalence of NF2/RASA1 deficiency in GCPM patients and the optimal combination regimen for clinical testing. A recent retrospective study characterized an “immune-activated” microenvironment specific to responders among patients with GCPM receiving chemoimmunotherapy. This signature included frequent ERBB2/ERBB3 mutations, high expression of CTLA-4 and HLA-DQB1, abundant CD8+ T cell infiltration, and a low frequency of CDH1 mutations[52]. A multidimensional biomarker model was constructed based on these features. Clinically, this multidimensional model could be developed as a composite score. The model incorporates ERBB2/ERBB3 mutations, CTLA-4/HLA-DQB1 expression, and CD8+ T cell infiltration. This score can predict chemoimmunotherapy response. Moreover, it offers superior performance over current biomarkers like programmed death ligand 1 (PD-L1) or tumor mutation burden alone.

Current research on gene mutations associated with GC metastasis still has limitations, such as limited sample sizes and a lack of prospective validation cohorts. Future efforts require large-scale, multicenter, prospective studies that integrate dynamic genomic monitoring with single-cell multi-omics technologies to elucidate the temporal sequence and clonal evolutionary patterns of driver mutations during PM development. Composite biomarker models that integrate genomic features with immune microenvironment phenotypes are expected to become a critical breakthrough in achieving precise subtyping and personalized treatment for GC (Table 2 summarizes the key genetic mutations implicated in GCPM, their alteration types, affected signaling pathways, functional consequences, and clinical implications).

Table 2 Key genetic mutations and their signaling pathways in gastric cancer peritoneal metastasis.

Gene	Alteration type	Affected pathway	Function	Clinical implication
CDH1	Deletion mutations	Cell adhesion and LGN	Randomization of cell division direction and increased invasiveness	Potential targeting direction for cell division
TP53	Deletion mutations	p53 signaling and apoptosis	Loss of cell cycle control and apoptosis	ctDNA TP53 mutation predicts poor response to chemotherapy
ARID1A	Deletion mutations	mTOR/SOX9	Enhanced migration and invasion	Potential target for mTOR inhibitors
NF2/RASA1	Deletion mutations	Wnt and YAP	Increased tumor stemness and metastatic capacity	Synthetic lethality with combined Bcl-2 and YAP inhibitors
ERBB2/ERBB3	Activating mutation	-	Enhanced TMB, CD8+ T cell infiltration, and TME immunological activity	Predict response to chemoimmunotherapy

This table summarizes the major genetic alterations implicated in gastric cancer peritoneal metastasis, their affected signaling pathways, functional consequences, and clinical implications. It highlights mutations that guide patient stratification, predict treatment response, or represent potential therapeutic targets. LGN: Leu-Gly-Asn repeat-enriched protein; ctDNA: Circulating tumor DNA; mTOR: Mammalian target of rapamycin; SOX9: SRY box 9; TMB: Tumor mutation burden.

EMT and ECM remodeling

Therapies targeting tumor invasion and metastasis remain a major unmet need in GCPM. Current systemic chemotherapy has limited efficacy against peritoneal dissemination, partly due to the physical barrier formed by the dense ECM and the inherent plasticity of tumor cells undergoing EMT. Moreover, no anti-metastatic drugs specifically targeting EMT or ECM remodeling have been approved for GCPM to date. Understanding the molecular mechanisms driving these processes is therefore critical for developing novel biomarkers to predict metastatic risk and identifying actionable targets to overcome treatment resistance.

EMT and ECM remodeling are core processes that reciprocally drive and synergistically advance GCPM. EMT confers migratory and invasive capabilities to tumor cells, while ECM remodeling provides structural support and signaling guidance for this process. Together, they constitute the biological foundation enabling tumor cells to breach the primary site and disseminate to the peritoneum.

The core driving role of EMT: EMT is a central manifestation of cellular phenotypic plasticity, enabling epithelial cells to acquire mesenchymal-like migratory and invasive capabilities by downregulating epithelial markers (such as E-cadherin) and upregulating mesenchymal markers (such as N-cadherin and vimentin)[53]. This transformation process is precisely regulated by a core network of transcription factors, including SNAIL, ZEB, and TWIST, serving as a critical “switch” that initiates the metastatic cascade in tumor cells[54,55]. In GCPM, the activation of EMT not only promotes tumor cell detachment from the primary site but also enhances their survival and colonization capacity within the peritoneal cavity. Multiple molecules have been confirmed to participate in GCPM by regulating EMT. Preclinical experimental evidence demonstrated that cystatin SN (CST1) inhibits ferroptosis by stabilizing the glutathione peroxidase 4 protein[56]. CST1 also simultaneously upregulates N-cadherin and Snail while downregulating E-cadherin. Together, these effects synergistically drive EMT and tumor invasion. Clinically, elevated CST1 expression in GC tissues correlates with deeper tumor invasion, higher rates of PM, and poor overall survival[57]. The innovation of this study lies in establishing, for the first time, a functional link between ferroptosis and EMT in the context of GCPM, providing a theoretical basis for combination targeting strategies. However, the positioning of CST1 within the upstream regulatory network and its synergistic mechanisms with other EMT drivers remain to be elucidated. A preclinical study demonstrated that biglycan (BGN) secreted by GC cells binds to Toll-like receptor 2/4 receptors on peritoneal mesothelial cells[7]. This binding activates the nuclear factor-kappa B pathway and induces the transformation of mesothelial cells into CAF-like cells. These CAF-like cells subsequently secrete fibronectin attachment protein (FAP), which acts on tumor cells to form a BGN/FAP/signal transducer and activator of transcription 3 (STAT3) positive feedback loop. This loop continuously activates EMT and promotes PM. Clinical cohort studies have confirmed that BGN is often overexpressed in GC tissues. Moreover, high BGN expression is significantly associated with disease recurrence and poor prognosis in patients with advanced disease[58]. This study is the first to reveal a bidirectional signaling loop formed between tumor cells and mesothelial cells through the BGN/FAP axis, deepening the mechanistic understanding of the “seed and soil” theory. Combined detection of BGN and FAP holds promise as a biomarker panel for predicting the risk of GCPM. However, whether this mechanism is universally present in other peritoneal metastatic tumors remains to be validated. Furthermore, the E3 ubiquitin ligase MIB1 promotes EMT and maintenance of tumor stemness by degrading DDX3X[59]; apolipoprotein C2 (APOC2) activates the PI3K/AKT/mTOR pathway via the CD36 receptor, upregulating the expression of mesenchymal markers and MMPs[60]; and CD44 acts as a signaling integration platform, cooperating with STAT3 to initiate EMT transcriptional reprogramming[61]. Among these, the study on APOC2 is the first to directly link lipid metabolism regulation to EMT, expanding the dimensional role of metabolic reprogramming in GCPM. Notably, an exploratory clinical trial in patients with advanced gastrointestinal cancer showed that lipid metabolism is closely associated with treatment response and prognosis[62]. This finding suggests that APOC2 has translational potential as a therapeutic target. The discovery regarding CD44 reveals the dual function of the same molecule in “adhesive anchoring” and “signal integration”, although the fine structural basis of its interaction with STAT3 remains to be elucidated.

Pro-metastatic mechanisms of ECM remodeling: The ECM is a complex network composed of fibrous proteins, proteoglycans, and various signaling molecules, providing structural support for cells and mediating mechanosensing and signal transduction[63]. In the TME, the ECM undergoes profound pathological remodeling, characterized by excessive collagen deposition, increased matrix stiffness, and aberrant crosslinking. These physicochemical changes themselves serve as critical pro-invasive signals[64,65]. MMPs are core effector molecules in ECM remodeling. High expression of MMP-1, MMP-7, and MMP-9 is significantly associated with the risk of GCPM and poor prognosis[26,66,67]. MMPs directly degrade the ECM barrier and release growth factors sequestered within the matrix, synergistically activating multiple pro-metastatic signaling pathways[68]. The value of MMPs as biomarkers for GCPM has been widely recognized. However, their upstream regulatory networks are complex. These networks include non-coding RNAs and epigenetic modifications. Furthermore, therapeutic strategies targeting MMPs have yet to achieve breakthroughs. A major obstacle is the issue of side effects. Lysyl oxidase (LOX) increases matrix stiffness by catalyzing collagen crosslinking, and its aberrant high expression is closely associated with poor prognosis in GC patients[69]. Notably, LOX secreted by tumor cells can reach distant organs via the circulatory system, pre-constructing a favorable “niche” for metastasis (preclinical research)[70]. CAF-derived LOX can also activate the transforming growth factor-β (TGF-β)-insulin-like growth factor-1 (IGF-1) signaling axis, synergistically regulating tumor cell motility, EMT, and glucose metabolic reprogramming (preclinical research)[71]. This study was the first to propose that LOX participates in metastasis through “remote pre-conditioning of the niche”, expanding the understanding of ECM remodeling beyond its traditional confinement to the local microenvironment. However, the stable form of LOX in the circulatory system and targeted delivery strategies still require in-depth investigation. Single-cell sequencing further revealed that LOX-positive fibroblasts and M2-type macrophages form a stable adjacent network in GC tissues, synergistically constructing an immunosuppressive microenvironment via interleukin-6 (IL-6) signaling[72]. The innovation of this study lies in uncovering the spatial co-localization and functional synergy between CAFs and TAMs at single-cell resolution, proposing a vicious cycle model of “hypoxia-LOX+ fibroblasts-IL6-monocytes-M2 macrophages”. However, the causal chain of this model still requires further validation through lineage tracing or conditional gene knockout animal models.

MMPs and LOX are not only prognostic biomarkers but also promising therapeutic targets. LOX is particularly attractive because it functions both locally and systemically to pre-condition metastatic niches. Targeting LOX with small molecule inhibitors (e.g., β-aminopropionitrile) or blocking the LOX/TGFβ/IGF1 axis may simultaneously reduce matrix stiffness, impair EMT, and relieve immunosuppression. However, the safety of chronic ECM modulation and the risk of off-target effects remain key barriers to clinical translation.

Synergistic interplay between EMT and ECM remodeling: There exists a close bidirectional regulatory relationship between EMT and ECM remodeling. On one hand, upon EMT activation, tumor cells actively secrete molecules such as MMPs and LOX to remodel the ECM, clearing a path for their migration. On the other hand, the remodeled ECM, through mechanotransduction and biochemical signaling, reciprocally reinforces the EMT phenotype in tumor cells, establishing a self-amplifying vicious cycle.

Taking anoikis resistance as an example, GC cells in suspension stabilize SOX2 via FERMT2, activating fibronectin 1 transcription and guiding fibronectin matrix remodeling, thereby coupling survival signaling with microenvironmental reprogramming to systemically drive PM[73]. This study revealed the core role of the FERMT2-SOX2-fibronectin 1 axis in anoikis resistance, yet the specific molecular mechanism by which it inhibits SOX2 ubiquitination, such as the identity of the E3 ligase involved, remains unknown. Similarly, the molecular chaperone glucose-regulated protein 94 (GRP94) influences intracellular reactive oxygen species homeostasis by regulating the YAP/TEAD1 pathway, modulating the balance between cell survival and death[23]. This study was the first to report the role of the GRP94-YAP/TEAD1 functional axis in GCPM, providing a novel intervention node for targeting the YAP pathway. However, the upstream activation signals of GRP94 in PM remain unclear. Both in vitro and in vivo preclinical studies provided evidence that macrophage-derived cathepsin L (CTSL) degrades ECM components[9]. CTSL also promotes macrophage polarization toward the M2 phenotype. Through these mechanisms, CTSL further enhances EMT and tumor invasion. This study reveals the bridging role of immune cell derived proteases. These proteases connect ECM remodeling and EMT. The findings also suggest that CTSL may serve as a dual effect target. This target would simultaneously affect both tumor cells and the immune microenvironment. A first-in-human phase I/II trial evaluated bexmarilimab in patients with specific advanced or metastatic solid tumors[74]. The goal was to explore its potential for inducing anti-tumor immune responses. Spatial transcriptomics was performed on biopsy samples taken before and after treatment. The results showed that treatment response coincided with two key events: Intratumoral macrophage phenotype switching and induction of adaptive immune responses. These findings suggest that combining TAM reprogramming with CTSL-targeted inhibition has translational potential. Another preclinical mechanistic study demonstrated that GC cell-derived exosomal miR-106a is delivered to peritoneal mesothelial cells[75]. This activates the TGF-β signaling pathway, inducing mesothelial-mesenchymal transition. Consequently, peritoneal integrity is disrupted and a pre-metastatic niche is established. This study systematically elucidated, for the first time, the formation mechanism of the pre-metastatic niche in PM from the perspective of the tripartite interaction among “exosomes, mesothelial cells, and tumor cells”. However, the molecular basis of exosome targeting to peritoneal mesothelial cells, such as the surface integrin profile, remains to be further elucidated. A clinical cohort study reported significantly elevated plasma miR-106a levels in GC patients compared to healthy controls[76]. Its expression was closely linked to tumor differentiation, lymph node metastasis, TNM stage, and tumor size. This finding is consistent with the observation that miR-106a is secreted by GC cells via exosomes. Thus, miR-106a holds potential as a diagnostic and prognostic biomarker for GCPM. In addition, combining miR-106a inhibition with TGF-β inhibition may offer an effective therapeutic strategy for these patients. EMT and ECM remodeling form a mutually driving and synergistically amplified regulatory network in GCPM. Current research has revealed the core roles of several key molecules (e.g., BGN, FAP, LOX, CTSL) and signaling pathways (e.g., TGF-β, PI3K/AKT, YAP/TEAD1) within this network. This work has also provided potential biomarkers (e.g., combined BGN/FAP detection) and therapeutic targets (e.g., LOX, CTSL, or the TGF-β pathway) for clinical translation. Table 3 summarizes the key molecules and pathways involved in EMT and ECM remodeling in GCPM, their mechanisms of action, functional consequences, and clinical potential. However, existing evidence is mostly derived from retrospective studies with small sample sizes or in vitro experiments, lacking validation in large-scale prospective cohorts. Moreover, most studies focus on single molecules or pathways. As a result, several key issues remain unaddressed. These include the overall regulatory mechanisms of the EMT-ECM interaction network, heterogeneity among different molecular subtypes, and the safety and efficacy of targeting strategies. Future efforts should integrate single-cell multi-omics, spatial transcriptomics, and organoid models to dissect the dynamic evolution of this network from spatiotemporal dimensions, and to advance the clinical translation of biomarkers and combination therapeutic strategies.

Table 3 Summary of key molecules and signaling pathways driving epithelial-mesenchymal transition and extracellular matrix remodeling in gastric cancer peritoneal metastasis.

Category	Key molecule/pathway	Mechanism of action	Functional	Clinical potential
EMT drivers	CST1	Stabilizes GPX4 to inhibit ferroptosis; upregulates N-cadherin/Snail, downregulates E-cadherin	Synergistically drives EMT and tumor invasion	Prognostic biomarker and combination targeting strategy
	BGN/FAP/STAT3 loop	BGN binds TLR2/4 on mesothelial cells → NF-κB activation → mesothelial-to-CAF transition; CAFs secrete FAP → forms positive feedback loop	Continuous EMT activation	Prediction of biomarkers and therapeutic targets
	APOC2/CD36	Activates PI3K/AKT/mTOR pathway via CD36 receptor	Upregulates mesenchymal markers and MMPs	Therapeutic target
	CD44/STAT3	Signaling integration platform cooperating with STAT3	Initiates EMT transcriptional reprogramming	Therapeutic target
ECM remodeling	MMPs	Degrade ECM barrier and release matrix-bound growth factors	Promote invasion and metastasis	Prognostic biomarkers
	LOX	Catalyzes collagen crosslinking → increased matrix stiffness; remote pre-conditioning of metastatic niche	Promotes invasion, EMT, and immunosuppression	Prognostic biomarker and therapeutic target
EMT-ECM crosstalk	FERMT2/SOX2/FN1	Stabilizes SOX2 → activates FN1 transcription → guides fibronectin matrix remodeling	Promotes tumor cells survival, invasion, and metastasis	Therapeutic target
	GRP94/YAP/TEAD1	Upregulates IL6 expression → activates YAP/TEAD1 pathway	Anoikis resistance; promotes migration and invasion	Therapeutic target
	CTSL, macrophage-derived	Degrades ECM components; promotes M2 macrophage polarization	Promotes EMT and invasion	Dual-effect target
	Exosomal miR-106a	Delivered to mesothelial cells → activates TGF-β signaling → induces MMT	Disrupts peritoneal integrity; establishes pre-metastatic niche	Diagnostic/prognostic biomarker and therapeutic target

This table catalogs the major molecular regulators of epithelial-mesenchymal transition and extracellular matrix remodeling in gastric cancer peritoneal metastasis. It summarizes their mechanisms of action, functional consequences, and translational potential. EMT: Epithelial-mesenchymal transition; CST1: Cystatin SN; GPX4: Glutathione peroxidase 4; BGN: Biglycan; FAP: Fibronectin attachment protein; STAT3: Signal transducer and activator of transcription 3; TLR2/4: Toll-like receptor 2/4; NF-κB: Nuclear factor-kappa B; CAF: Cancer-associated fibroblast; APOC2: Apolipoprotein C2; PI3K: Phosphatidylinositide 3-kinase; AKT: Protein kinase B; mTOR: Mammalian target of rapamycin; MMPs: Matrix metalloproteinases; ECM: Extracellular matrix; LOX: Lysyl oxidase; FERMT2: Fermitin family homolog-2; SOX2: SRY box 2; FN1: Fibronectin 1; GRP94: Glucose-regulated protein 94; TEAD1: TEA domain transcription factor 1; CTSL: Cathepsin L; TGF: Transforming growth factor; MMT: Mesothelial-mesenchymal transition.

Hypoxia and anoikis resistance: Dual drivers of PM

Hypoxia is a common feature of peritoneal metastases and contributes to poor treatment response. Anti-angiogenic therapies targeting vascular endothelial growth factor (VEGF) (e.g., bevacizumab) have shown limited efficacy in GCPM, partly due to hypoxia-induced compensatory mechanisms and enhanced tumor cell survival pathways. Similarly, conventional chemotherapy often fails to eradicate disseminated tumor cells in the peritoneal cavity because these cells acquire anoikis resistance, allowing them to survive in suspension. Understanding how hypoxia and anoikis resistance synergistically drive GCPM progression is therefore critical for developing novel therapeutic strategies that overcome these adaptive survival mechanisms.

Hypoxia is a hallmark feature of the solid TME, while anoikis resistance is a critical capability enabling tumor cells to survive after detachment from the ECM. In GCPM, the hypoxic microenvironment not only directly drives tumor cell invasion and angiogenesis but also, by enhancing anoikis resistance, creates conditions for the suspension survival and distant colonization of tumor cells within the peritoneal cavity. These two factors synergize to constitute the dual drivers of GCPM progression.

Formation and core regulation of the hypoxic microenvironment: The rapid proliferation and high metabolic activity of tumor cells lead to a sharp increase in oxygen consumption. When metabolic oxygen demand exceeds the oxygen supply capacity of the vascular system, a hypoxic microenvironment is established[77]. Under hypoxic conditions, the central responsive molecule hypoxia-inducible factor-1α (HIF-1α) becomes stabilized and activated. It then transcriptionally regulates a series of downstream genes, including VEGF, lactate dehydrogenase, and MMPs[78-80]. This process promotes angiogenesis, metabolic reprogramming, and invasive metastasis.

The stability of HIF-1α is precisely regulated by an oxygen-dependent ubiquitin-proteasome degradation pathway[81]. A preclinical mechanistic study made a key discovery. Researchers identified a functional hypoxia response element in the Snail promoter[82]. This finding provided evidence that HIF-1α directly transcriptionally regulates Snail to drive EMT. This study was the first to directly link hypoxic signaling with a core EMT transcription factor, deepening the understanding of the mechanisms by which hypoxia promotes metastasis. Further preclinical multi-level studies revealed that under hypoxic conditions, HIF-1α directly activates miR-574-3p transcription[83]. This miR-574-3p inhibits CUL2 expression, which blocks VCBCR complex formation and prevents HIF-1α degradation. Together, these events establish a self-amplifying positive feedback loop. Similarly, hypoxia-induced miR-301a-3p can be transferred between cells via exosomes, targeting and inhibiting PHD3, similarly resulting in positive feedback activation of HIF-1α[84]. These studies have uncovered the non-coding RNA regulatory mechanisms underlying the self-maintenance and propagation of hypoxic signaling in GCPM, providing new targets for intervening in the hypoxia pathway. However, the selective mechanisms governing the intercellular transfer of miR-301a-3p remain to be elucidated.

Tipifarnib is a farnesyltransferase inhibitor originally designed to block Ras signaling. A recent innovative study systematically evaluated the anti-tumor effects of low dose tipifarnib in GC cells[85]. The study uncovered a novel mechanism: Tipifarnib inhibits the mTOR pathway, thereby downregulating HIF-1α expression. This discovery offers a new theoretical foundation for targeting HIF-1α signaling in GC treatment. IDF-11774 is an orally active small molecule inhibitor targeting HIF-1α. Its anti-tumor activity has been demonstrated in colorectal, lung, thyroid, and melanoma cancers. However, its therapeutic potential in GC remains unreported. A recent study reported for the first time that IDF-11774 promotes HIF-1α degradation in GC cells[86]. It does so by activating the ubiquitin-proteasome system, thereby exerting anti-tumor effects. Although this study lacked in vivo animal validation, the identified molecular mechanism offers a clear direction and theoretical basis for future investigations. Discoidin domain receptor 1 (DDR1) is an ECM receptor that plays a central role in stabilizing HIF-1α protein. It does this by inhibiting HIF-1α ubiquitination and degradation. The DDR1 inhibitor 7h benzamide demonstrated significant anti-tumor effects in PDX and organoid models[87]. It effectively suppressed tumor growth, angiogenesis, and metastasis. These findings indicate strong translational potential. This strategy may address the resistance commonly seen with current anti-angiogenic therapies. Thus, combining a DDR1 inhibitor with anti-angiogenic treatment represents a potential therapeutic approach for GCPM. Nevertheless, further preclinical and clinical studies are needed to validate its safety and efficacy.

Molecular mechanisms of anoikis resistance: Anoikis is a form of programmed cell death triggered by the loss of contact between cells and the ECM or adjacent cells, serving as a critical natural barrier against tumor metastasis. GC cells acquire anoikis resistance through multiple mechanisms, enabling them to survive within the peritoneal cavity and form metastatic lesions. A molecular mechanism study using both in vitro and in vivo models demonstrated that GRP94 activates the YAP/TEAD1 pathway by upregulating IL6 expression[23]. This activation promotes anoikis resistance in GC cells and enhances their migratory and invasive capabilities. This study was the first to report the role of the GRP94-YAP/TEAD1 functional axis in GCPM, providing a novel intervention node for targeting the YAP pathway. However, the upstream activation signals of GRP94 in PM remain unclear. Encouragingly, a team has successfully developed a fully human monoclonal antibody targeting GRP94, named K101.1[88]. Studies in cell and animal models showed that this antibody exhibits potent anti-tumor efficacy and a favorable safety profile. Mechanistic investigations revealed that K101.1 induces GRP94 internalization, which downregulates its expression on the cell surface. This in turn inhibits tumor growth and angiogenesis, while promoting tumor cell apoptosis. Preclinical mechanistic evidence showed that nuclear-localized myosin-9 (MYH9) induces β-catenin transcription by specifically recognizing the CTNNB1 promoter sequence, activating the Wnt/β-catenin pathway and contributing to anoikis resistance[89]. In addition, the research team successfully identified staurosporine as a MYH9 inhibitor and validated its therapeutic potential in cell and conditional transgenic mouse models. The study found that staurosporine inhibits MYH9 phosphorylation, which leads to downregulation of Wnt target gene expression and effectively suppresses tumor growth and metastasis. These findings suggest that staurosporine, or targeting MYH9 itself, could be a potential therapeutic strategy for GCPM. Another preclinical study demonstrated that FERMT2 promotes the expression of fibronectin and integrin α5/β1 by stabilizing SOX2, guiding ECM remodeling to form an anti-anoikis barrier[73]. Notably, the study further revealed that TGF-β1/TGF-β-RI and FERMT2 form a positive feedback loop in GC cells under suspension culture conditions. Treatment with SB431542, a TGF-β-RI inhibitor, effectively reversed the upregulation of FERMT2 and TGF-β-RI expression induced by suspension culture. In vivo experiments demonstrated that combining SB431542 administration with FERMT2 knockdown significantly inhibited the development of GCPM. This study revealed a survival strategy in which tumor cells actively remodel the microenvironment to adapt to suspension survival, yet the specific mechanism by which FERMT2 inhibits SOX2 ubiquitination remains to be elucidated. Adipocytes in the TME upregulate PITPNC1 expression in GC cells by secreting IL-6 and tumor necrosis factor-α. PITPNC1 in turn promotes the expression of CD36 and CPT1B, activates the fatty acid oxidation pathway, and thereby enhances the ability of GC cells to adapt to oxidative stress after detachment from the ECM[90]. This study was the first to directly link adipocyte metabolic signaling to anoikis resistance, expanding the dimensional role of metabolic reprogramming in GCPM. Moreover, the study further found that Etomoxir, a CPT1 inhibitor, significantly inhibited fatty acid oxidation in both cell experiments and in vivo animal models. This led to the reversal of anoikis resistance and effectively reduced metastasis. The transcription factor C/EBPβ regulates PDGFB expression, activating the mitogen-activated protein kinase (MAPK)/ERK pathway through autocrine and paracrine mechanisms, thereby simultaneously enhancing anoikis resistance and angiogenesis[91].

Synergistic interplay between hypoxia and anoikis resistance: Hypoxia and anoikis resistance form a close synergistic relationship in GCPM. Hypoxia not only directly activates HIF-1α-mediated pro-survival signaling but also indirectly enhances anoikis resistance through the regulation of multiple key molecules. A preclinical mechanistic study demonstrated that HIF-1α transcriptionally activates long noncoding RNA (lncRNA)-PMAN. This lncRNA-PMAN then promotes the nucleocytoplasmic translocation of ELAVL1, enhances SLC7A11 mRNA stability, and inhibits ferroptosis, thereby promoting the colonization and growth of GC cells within the peritoneal cavity[92]. This study revealed a novel mechanism by which hypoxia enhances anoikis resistance through the inhibition of ferroptosis. Targeting the HIF-1α/PMAN/ELAVL1 axis holds promise for synergistic effects with ferroptosis inducers. However, the precise mechanism by which PMAN regulates the subcellular localization of ELAVL1 remains to be elucidated. Under hypoxic conditions, exosomal miR-301a-3p secreted by GC cells enters recipient cells and targets PHD3[84]. This blocks HIF-1α degradation and forms a positive feedback loop. Simultaneously, it induces a mesenchymal-like transformation, thereby significantly enhancing peritoneal metastatic capacity. Furthermore, this study confirmed that serum exosomal miR-301a-3p levels were significantly higher in GC patients with PM compared to healthy volunteers and GC patients without PM. This finding suggests that serum exosomal miR-301a-3p could serve as a potential non-invasive diagnostic biomarker for GCPM, providing a new theoretical basis and technical tool for the application of liquid biopsy in the early detection of this condition.

Notably, hypoxia can activate the YAP/TEAD1 pathway, and the activation of this pathway is a key step in GRP94-mediated anoikis resistance[23]. Hypoxia-induced metabolic reprogramming generates large amounts of lactate. This lactate, catalyzed by valyl-tRNA synthetase 1 (AARS1), mediates protein lactylation and subsequently regulates the activity of pathways such as YAP/TEAD1[93]. Together, these events form a multi-level synergistic network of “hypoxia–metabolic reprogramming-epigenetic regulation-anoikis resistance”. Furthermore, HIF-1α upregulated by hypoxia promotes PDGFB expression[94], which not only participates in anoikis resistance but also promotes angiogenesis via the MAPK/ERK pathway[91], further functionally coupling hypoxia, anoikis resistance, and angiogenesis.

Hypoxia and anoikis resistance constitute a mutually driving and synergistically amplified regulatory network in GCPM. As a central molecule, HIF-1α not only directly regulates angiogenesis and metabolic reprogramming but also enhances the anoikis resistance of GC cells through multiple mechanisms. These include the regulation of ferroptosis, exosome-mediated transmission, and the YAP pathway. In this way, HIF-1α provides a dual safeguard for tumor cell survival and colonization within the peritoneal cavity. However, existing evidence is largely derived from in vitro experiments or small-sample clinical studies. Several key issues therefore remain to be addressed, including the overall regulatory mechanisms of the interactive network between hypoxia and anoikis resistance, heterogeneity among different molecular subtypes, and the safety and efficacy of targeting the HIF-1α or YAP pathways. Therapeutic strategies targeting the hypoxic microenvironment (such as HIF-1α inhibitors and anti-VEGF therapy) have shown limited efficacy due to tumor heterogeneity and compensatory mechanisms, while interventions solely targeting anoikis resistance also face the risk of drug resistance. Future efforts should integrate single-cell multi-omics and spatial transcriptomics to dissect the dynamic evolution of this network from spatiotemporal dimensions, and explore combination therapeutic strategies targeting the hypoxia-anoikis resistance interface, aiming to provide new theoretical foundations for precision intervention in GCPM (Figure 2 outlines the complex interaction network formed by hypoxia and anoikis resistance as core drivers in GCPM progression).

Figure 2 Schematic diagram of the synergistic interplay between hypoxia and anoikis resistance in gastric cancer peritoneal metastasis. Under hypoxic conditions, Hypoxia-inducible factor 1alpha (HIF-1α) is stabilized and activated. It promotes angiogenesis, epithelial-mesenchymal transition, and inhibits ferroptosis. Positive feedback loops involving miR-574-3p/CUL2 and miR-301a-3p sustain HIF-1α activation. Following detachment from the extracellular matrix, gastric cancer cells activate multiple pro-survival pathways. Hypoxia and anoikis resistance are connected through shared nodes. YAP/TEA domain transcription factor 1 is activated by both hypoxia and Glucose-regulated protein 94. PDGFB is co-regulated by HIF-1α and C/EBPβ. Extracellular matrix detachment: Loss of contact between tumor cells and the extracellular matrix. DDR1: Discoidin domain receptor 1; HIF-1α: Hypoxia-inducible factor 1alpha; VEGF: Vascular endothelial growth factor; EMT: Epithelial-mesenchymal transition; lncRNA: Long non-coding RNA; ECM: Extracellular matrix; GRP94: Glucose-regulated protein 94; TEAD1: TEA domain transcription factor 1; MYH9: Myosin-9; SOX2: SRY box 2; FN1: Fibronectin 1.

TME: The synergistic network of immunosuppression and angiogenesis

Immunotherapy, particularly immune checkpoint inhibitors (e.g., anti-programmed cell death 1/PD-L1), has shown limited efficacy in GCPM patients, with response rates below 20% in most clinical trials. Similarly, anti-angiogenic agents (e.g., bevacizumab, ramucirumab) provide only modest survival benefits. A major barrier to improving these outcomes is the complex and immunosuppressive TME. The TME is a dynamic ecosystem composed of tumor cells, stromal cells, immune cells, vascular endothelial cells, and the ECM[95]. During GCPM, the TME establishes a dual barrier of immunosuppression and aberrant angiogenesis through synergistic crosstalk among multiple cell types. These two factors mutually reinforce each other, collectively driving tumor progression and therapeutic resistance. Understanding the synergistic interactions among these TME components is therefore critical for developing rational combination strategies.

CAFs - from stromal support to functional heterogeneity: CAFs are the most abundant and functionally complex stromal cell population in the GCPM microenvironment, and their infiltration density is significantly positively correlated with tumor size, depth of invasion, lymph node metastasis, and PM[31,96]. CAFs participate in the GCPM process through four major pathways[30]: Establishing an immunosuppressive microenvironment [secreting TGF-β and IL-6, expressing PD-L1, recruiting myeloid-derived suppressor cell (MDSCs)]; maintaining tumor cell survival and stemness [fibroblast growth factor (FGF) signaling inhibiting apoptosis, exosomal miR-522 conferring ferroptosis resistance]; promoting angiogenesis and metabolic adaptation (galectin-1 activating VEGF/VEGFR2, LOX enhancing the Warburg effect, CPT1A optimizing fatty acid oxidation); and driving invasion and metastasis (SLIT2 remodeling the cytoskeleton, secreting MMPs to degrade the ECM). Although this study systematically elucidated the multi-dimensional pro-tumor mechanisms of CAFs, it was primarily based on correlational analyses and did not distinguish functional subsets of CAFs. Subsequent studies have revealed that CAFs exhibit high heterogeneity[97,98], with certain subsets [such as antigen-presenting CAFs (apCAFs)] even exerting anti-tumor effect, suggesting that the umbrella term “CAFs” may obscure critical functional differences.

At the mechanical interaction level, bidirectional signaling mediated by Src kinase occurs between GC cells and stromal fibroblasts[99]. This signaling activates the Rho/ROCK pathway, which promotes enhanced cytoskeletal contractility and drives ECM mechanical remodeling. Helicobacter pylori activates the nuclear factor-kappa B pathway through CagA and VacA[100]. This leads to upregulation of PIEZO1 expression in GC cells and activation of the YAP1-CTGF axis, thereby promoting CAF infiltration and the formation of a stiffened matrix. In addition, this study successfully identified procyanidin C1 as a novel small molecule inhibitor of CTGF. Furthermore, in vivo experiments confirmed that its combination with 5-fluorouracil produced synergistic anti-tumor effects. These two preclinical studies respectively revealed the roles of mechanosignaling and pathogen-driven inflammation in CAF activation. However, the upstream activating signals of Src kinase remain unclear, the mechanotransduction mechanisms mediated by PIEZO1 require further investigation, and the studies primarily focused on the α-smooth muscle actin+ CAF subset without addressing other subtypes.

Regarding the cellular source diversity of CAFs, BGN secreted by GC cells binds to Toll-like receptor 2/4 on the surface of peritoneal mesothelial cells, inducing their transdifferentiation into CAF-like cells. These CAF-like cells subsequently secrete FAP, promoting GC cell proliferation, migration, and EMT, forming a closed-loop “tumor cell-mesothelial cell-tumor cell” signaling pathway[7]. Importantly, this study confirmed that both the FAP inhibitor and the anti-BGN antibody demonstrated significant anti-tumor effects in preclinical models. This finding was the first to reveal that mesothelial cells represent a significant source of CAFs, and that BGN and FAP may serve as potential predictive biomarkers and therapeutic target for GCPM. However, the generalizability of BGN to other cancer types has not been validated, and the functional differences between mesothelial cell-derived CAFs and CAFs from other sources remain to be compared. Furthermore, IL-6 secreted by CAFs in peritoneal metastases activates the JAK/STAT3 pathway, synergistically driving tumor progression, immunosuppression, and chemotherapy resistance. Tocilizumab, an IL-6 receptor antibody, has shown significant therapeutic potential in CAF-rich preclinical models[101], although the precise mechanism of anti-IL-6R antibody action requires further elucidation.

More importantly, by integrating spatial transcriptomics and single-cell RNA sequencing, a subset of apCAFs marked by CD74, HLA-DRA, and other markers was identified. These apCAFs are spatially enriched around tertiary lymphoid structures and are closely associated with favorable patient prognosis and enhanced response to immunotherapy[97]. In contrast, preclinical evidence showed that growth differentiation factor 15-driven myCAFs secreted by GC cells form a dense physical barrier surrounding tumor cells, impeding CD8+ T cell infiltration and creating an immune “desert” microenvironment[98]. Targeting growth differentiation factor 15 may relieve myCAF-mediated immune exclusion and enhance the sensitivity of tumors to immunotherapy. Multi-omics analysis revealed that, compared with primary gastric tumors, iCAFs and myCAFs are significantly enriched in peritoneal metastases, driving malignant progression through regulation of EMT, angiogenesis, and YAP/TAZ activity[41,102]. These studies have elucidated CAF heterogeneity from new dimensions of cellular functional states and spatial niches, challenging the traditional paradigm that CAFs primarily mediate immunosuppression and laying the foundation for “state-specific targeting” therapy. However, several issues remain unresolved. The antigen-presenting function of apCAFs has yet to be validated through in vitro experiments. The mechanisms underlying their spatial enrichment remain unclear. Whether the formation of the myCAF barrier is regulated by hypoxia or other signals is not yet understood. Furthermore, the mechanisms governing the functional transition between iCAFs and myCAFs at different stages of GCPM remain to be elucidated.

TAMs - from polarization dichotomy to lineage dynamics: TAMs predominantly exhibit an M2-like pro-tumor phenotype in GCPM, synergistically driving tumor progression by secreting immunosuppressive factors such as IL-10 and TGF-β, expressing immune checkpoint molecules including PD-L1, and promoting angiogenesis and ECM remodeling[95,103,104]. TAMs in malignant ascites from patients with advanced GC display significant M2-like polarization, characterized by high IL-10 expression and low tumor necrosis factor-α expression, which is closely associated with poor prognosis[105].

Multiple preclinical studies have demonstrated that GC cells induce M2 polarization through multiple signaling pathways: In the hypoxic microenvironment, the C-X-C motif chemokine ligand 8 (CXCL8)-CXC chemokine receptor 1 and 2 axis activates the JAK/STAT1 pathway, upregulating IL-10 to form a positive feedback loop[106]; DKK1 binds to the CKAP4 receptor on macrophages, activating the PI3K-AKT axis to drive M2 polarization[107]; POU1F1 induces M2 polarization via the CXCL12/CXC chemokine receptor 4 axis and enhances its pro-angiogenic function[108]. Among these findings, it is particularly noteworthy that preclinical studies have confirmed combining an anti-DKK1 antibody with an anti-programmed cell death 1 antibody significantly enhances anti-tumor efficacy. After combination therapy, the proportion of M1-like macrophages increased markedly, while the proportion of M2-like macrophages decreased significantly. This suggests that targeting DKK1 reshapes the immune microenvironment primarily by inducing macrophage reprogramming rather than depletion. However, the relative contribution of this mechanism in GCPM, as well as its synergistic effects with other immune cells, remains to be systematically investigated.

In the construction of the pre-metastatic niche, TGF-β1 secreted by M2-type TAMs plays a central role[109]. Macrophage-derived CTSL not only reinforces M2 polarization but also promotes E-cadherin degradation and EMT[9]. Furthermore, TAMs form functional synergy with mesothelial cells: Mesothelial cells secrete fibronectin, which enhances cancer cell adhesion and invasion upon induction by TAM-derived TGF-β, while TAMs promote cancer cell breaching of the mesothelial cell layer through high expression of ECM remodeling-related genes[110]. These preclinical studies have revealed the central role of TAMs in the pre-metastatic niche, yet the interactive mechanisms between TAMs and mesothelial cells require further in-depth elucidation. Nevertheless, inducing phenotypic reprogramming of TAMs remains a potentially effective strategy for reshaping the pre-metastatic niche and inhibiting tumor metastasis. Several clinical trials have demonstrated that targeting molecules such as CD40, CD137, interferon-γ, IL-4, IL-15/IL-15Rα, TGF-β, and histamine H1R can reprogram TAMs from a pro-tumor to an anti-tumor phenotype[111]. This reprogramming reshapes the tumor immune microenvironment and improves the efficacy of immune checkpoint inhibitors. Single-cell trajectory analysis revealed that, in the ascites of GCPM patients, TAMs undergo a dynamic lineage transition from CTSL+ to C1Q+: CTSL+ TAMs establish the pre-metastatic microenvironment and mediate tumor cell “homing”, whereas C1Q+ TAMs drive tumor cell proliferation, immune evasion, and therapeutic resistance. The latter upregulate PD-L1 and NECTIN2 expression on tumor cells through the secretion of complement C1q, establishing a dual-protective immune-privileged microenvironment[112]. Based on these findings, the research team functionally validated a C1 inhibitor in vitro cell models. The results showed that C1 inhibitor treatment significantly reduced PD-L1 and NECTIN2 expression on GC cell surfaces. Thus, C1q is a key mediator of tumor immune evasion, and targeting it with a C1 inhibitor effectively reverses this immunosuppressive effect. This study was the first to reveal the lineage transition trajectory of TAMs through single-cell sequencing, breaking through the traditional M1/M2 binary classification framework. However, the lineage transition was primarily based on pseudotime analysis and lacks direct validation using lineage-tracing mouse models. The specific binding molecules of complement C1q with receptors on the tumor cell surface have not yet been identified, and the signaling pathways through which it upregulates PD-L1 and NECTIN2 remain to be elucidated.

Furthermore, secreted phosphoprotein 1 (SPP1)+ TAMs highly express immune checkpoint molecules such as TIM-3, CD276, and LGALS9, enhancing local immunosuppression in the tumor by inhibiting T cell function[113]. Spatial analysis reveals that they form a close anatomical proximity with GREM1+ CAFs, with active molecular crosstalk[114]. THBS2+ myCAFs promote the recruitment and polarization of SPP1+ TAMs by secreting complement C3, establishing a CAF-macrophage synergistic immunosuppressive niche. In vivo experiments using a subcutaneous xenograft model further demonstrated that a C3aR antagonist blocks the C3-C3aR1 axis. This effectively disrupts the pathological stromal-myeloid interaction network and significantly enhances the anti-tumor immune response to immune checkpoint inhibitors[113]. This finding systematically revealed, for the first time, the spatial interaction and functional synergy between THBS2+ CAFs and SPP1+ TAMs, providing a novel strategy for combination therapy in GCPM. However, the study is primarily based on correlative analyses and lacks causal validation, and humanization of C3aR antagonists along with clinical safety assessments have not yet been conducted.

Angiogenesis - synergistic contributions of TME cells and therapeutic targets: Tumor angiogenesis is a core process in the progression of GCPM. When solid tumors exceed 1-2 mm³ in volume, they release large amounts of pro-angiogenic factors to induce neovascularization[115,116], with the VEGF/VEGFR pathway being the most representative[117]. Additionally, pathways such as PDGF/PDGFR, FGF/FGFR, ANGPT/Tie, HIF-1α and TGF-βare also widely involved[118-121].

TME cells make significant contributions to angiogenesis: TAMs activate the “angiogenic switch” by secreting factors such as VEGF[122], TGF-β[123], MMP-9[124]; CAFs promote angiogenesis through paracrine CCL2 activating the STAT3 pathway in endothelial cells[125], or via HGF activating the PI3K/AKT and ERK1/2 pathways[126]. A noteworthy finding from these preclinical studies is that CAF-derived HGF has three pro-angiogenic functions in GC. It promotes endothelial-dependent angiogenesis, tumor cell-dependent vasculogenic mimicry, and mosaic vessels formed by endothelial-tumor cell mixtures. In vitro experiments confirmed that both PI3K and MEK/ERK inhibitors effectively suppress these three vascular patterns. Although in vivo validation is still lacking, the strategy of targeting HGF along with its downstream PI3K/AKT and ERK1/2 pathways shows promise in overcoming resistance to conventional anti-angiogenic therapy. This could become a new direction for anti-angiogenic treatment in GC. Single-cell sequencing and spatial transcriptomics analyses revealed that the VCAN+ TAM subset is significantly enriched in the GCPM microenvironment, directly driving endothelial cell proliferation and tube formation through the secretion of pro-angiogenic factors[127]. Notably, the VCAN+ TAM subset was significantly expanded during chemotherapy resistance in GCPM, with greater expansion than in the immunotherapy-resistant group. This suggests a specific role in chemotherapy resistance. Single-cell transcriptomic data also revealed that pro-angiogenic and immunosuppressive functions are coupled within this same TAM subset. Thus, targeting VCAN or its downstream pathways may offer a novel strategy to reverse chemotherapy resistance. Nevertheless, these conclusions are based on correlative transcriptomic analyses and lack direct functional validation. Moreover, the lineage relationship between VCAN+ TAMs and other TAM subsets remains unclear and warrants further study.

Furthermore, endothelial cell-specific molecule 1 (ESM1) from GC cells binds to the c-Met receptor on vascular endothelial cells, which activates the MAPK pathway and upregulates HIF-1α, VEGFA, and MMP-9 expression[34]. This leads to systematic regulation of tumor angiogenesis. To validate the in vivo potential of the ESM1/c-Met axis as a therapeutic target, the team tested an ESM1 neutralizing antibody and a c-Met inhibitor in a PM model. Both strategies significantly suppressed tumor angiogenesis and PM. The study also found markedly elevated serum ESM1 levels in GC patients. Thus, targeting ESM1 may serve as both a novel therapeutic approach for GCPM and an early diagnostic biomarker for PM, offering broad clinical potential. However, the upstream and downstream regulatory relationships of ESM1 within the angiogenesis network remain to be elucidated. GC cells with anoikis-resistant properties, through C/EBPβ-mediated sustained paracrine PDGFB signaling, activate the MAPK/ERK pathway in vascular endothelial cells, promoting neovascularization[91], revealing an intrinsic link between anoikis resistance and angiogenesis. Nevertheless, the relative contribution of PDGFB signaling to angiogenesis in GCPM remains to be quantified.

Adipocytes - a hub for metabolic-immune-vascular interactions: The peritoneum is a unique organ rich in adipose tissue, and adipocytes play an active regulatory role in GCPM. GC cells secrete LAMC1, which enhances their own motility through autocrine signaling and drives the differentiation of preadipocytes into mature adipocytes via paracrine signaling. The latter release free fatty acids that reshape the metabolic programming of GC cells, establishing a pre-metastatic niche[128]. Importantly, serum enzyme-linked immunosorbent assay in an independent clinical cohort revealed that high LAMC1 levels effectively predicted PM with high diagnostic accuracy. Thus, LAMC1 may serve as a potential serum biomarker for GCPM. CXCL2 secreted by omental adipocytes activates the AKT-HIF1α-VEGFA signaling axis, systematically promoting tumor angiogenesis and growth[6]. Notably, this study used an in situ omentectomy model to provide direct evidence that the omentum promotes GCPM. Thus, omentectomy may help reduce PM risk, offering a theoretical basis for optimizing surgical strategies. Furthermore, urine analysis showed that GC patients with PM had significantly elevated urinary CXCL2 levels, suggesting that urinary CXCL2 could be a potential non-invasive diagnostic biomarker for GCPM. In the peritoneal lipid-rich environment, GC cells stabilize ORAI2 mRNA through NSUN2-mediated m⁵C methylation, activating PI3K-AKT survival signaling and enhancing proliferation, migration, and peritoneal colonization capacity[129]. These preclinical studies systematically elucidated, for the first time, the active regulatory role of adipocytes in GCPM, elevating their function from traditional metabolic support to that of multi-dimensional regulators. However, the specific receptor through which LAMC1 drives preadipocyte differentiation has yet to be identified; the clinical translational value of serum LAMC1 as a biomarker has only been validated in small cohort studies; the sensitivity and specificity of urinary CXCL2 as a biomarker need to be validated in larger sample cohorts; and the upstream regulatory signals and downstream target spectrum of NSUN2 remain to be elucidated.

Synergistic interaction between immunosuppression and angiogenesis - the hypoxia-metabolism axis: The synergistic interaction between immunosuppression and angiogenesis is primarily mediated through the hypoxia-metabolism axis. Hypoxia upregulates glycolytic gene expression via HIF-1α, enhancing lactate production[130,131]. Lactate accumulation directly suppresses CD8+ T cell function, induces M2 polarization of TAMs, and promotes MDSC recruitment[132-134], establishing a complete causal chain of “hypoxia-metabolic reprogramming-immunosuppression”. However, the specific molecular mechanisms by which lactate suppresses T cell function (such as lactylation) have yet to be thoroughly validated in GCPM. The combination of anti-angiogenic therapy and immunotherapy has emerged as an important direction for GCPM treatment. Notably, the C3-C3aR1 axis activated by THBS2+ myCAFs promotes the enrichment of SPP1+ TAMs, leading to resistance to immune checkpoint inhibitors, while blockade of the C3-C3aR1 axis significantly enhances the efficacy of immunotherapy[113], providing a novel strategy for combination therapy.

Components within the TME, including CAFs, TAMs, and adipocytes, synergistically construct a dual barrier of immunosuppression and aberrant angiogenesis through complex signaling networks, collectively driving GCPM progression (Figure 3 provides a schematic overview of the key components and interaction networks within the TME of GCPM). Current research still faces the following limitations: The heterogeneity and functional plasticity of TME cell subsets have not been systematically resolved; the mechanisms underlying the interplay between immunosuppression and angiogenesis are largely based on correlational analyses, lacking direct evidence of causality; optimization of combination therapeutic strategies requires further clinical validation; research on the mechanistic roles of adipocytes is still in its early stages; and most studies have small sample sizes and lack prospective validation. Future efforts should integrate single-cell multi-omics, spatial transcriptomics, and multiplex immunofluorescence technologies to precisely delineate the spatial distribution and functional synergy of TME cell subsets from spatiotemporal dimensions. Humanized mouse models and organoid co-culture systems should be established to validate causal relationships of key molecular pathways. Prospective clinical trials are needed to explore optimal combination strategies for anti-angiogenic agents and immune checkpoint inhibitors. Novel intervention approaches targeting the CAF-TAM interface (such as the C3-C3aR1 axis) and the adipocyte-tumor cell interface should be developed, with the goal of providing more effective precision treatment strategies for patients with GCPM (Table 4 summarizes the key components of the TME in GCPM, their pro-tumor mechanisms, intercellular interactions, and potential therapeutic targets).

Figure 3 Outline of tumor microenvironment in peritoneal metastasis of gastric cancer. M1 macrophages, dendritic cells, and T cells are innate immune cells that fight tumors. Gastric cancer cells induce effector T cells into an exhausted state by expressing relevant immune checkpoints, recruiting and promoting the polarization of M2 macrophages, and modifying regulatory T cell and myeloid-derived suppressor cells to suppress immune responses. Immune cells normally secrete cytokines to suppress tumor growth, however, suppressor T cells cannot produce these cytokines. In addition, gastric cancer cells induce the differentiation of cancer-associated fibroblasts and promote the maturation of preadipocytes, shaping the aggressive phenotype and immune tolerance of tumor cells. Anoxia is a core feature of peritoneal metastasis, which can drive epithelial-mesenchymal transition, angiogenesis, ferroptosis resistance, and acidic tumor microenvironment. EMT: Epithelial-mesenchymal transition; ICOS: Inducible T-cell co-stimulator; IL: Interleukin; PD-L1: Programmed death ligand 1; TGF: Transforming growth factor; CTSL: Cathepsin L; GDF15: Growth differentiation factor 15; ECM: Extracellular matrix; FAP: Fibronectin attachment protein; VEGF: Vascular endothelial growth factor; HIF-1α: Hypoxia-inducible factor 1alpha; CAFs: Cancer-associated fibroblasts; MDSC: Myeloid-derived suppressor cell; FFAs: Free fatty acids.

Table 4 Tumor microenvironment components in gastric cancer peritoneal metastasis: Functions, interactions, and therapeutic opportunities.

TME component	Subset/key factor	Mechanisms	Key interactions	Potential therapeutic targets
CAFs	General CAFs	Immunosuppression; tumor survival/stemness; angiogenesis; invasion	Interact with tumor cells, immune cells, and endothelial cells	FAP inhibitors; IL-6R antibody
	apCAFs (CD74+, HLA-DRA+)	Antigen-presenting function	Spatially enriched around tertiary lymphoid structures	Prognostic biomarker; predict immunotherapy response
	myCAFs (GDF15-driven)	Form dense physical barrier; impede CD8+ T cell infiltration	Interact with tumor cells	Target GDF15 to relieve immune exclusion
TAMs	M2-like TAMs	Immunosuppression; angiogenesis; ECM remodeling	Interact with tumor cells, immune cells, CAFs and endothelial cells	TAM reprogramming (anti-DKK1, anti-IL10, anti-POU1F1)
	CTSL+ TAMs	Establish pre-metastatic microenvironment; mediate tumor cell “homing”	Dynamic lineage transition to C1Q+ TAMs	Target CTSL
	C1Q+ TAMs	Drive proliferation, immune evasion, therapeutic resistance	Dynamic lineage transition from C1Q+TAMs	C1 inhibitor
	SPP1+ TAMs	Immunosuppression	Interact with THBS2+ CAFs	C3aR antagonists
	VCAN+ TAM	Angiogenesis, immunosuppression and chemotherapy resistance	-	Target VCAN
Adipocytes	-	Secrete CXCL2 → AKT-HIF1α-VEGFA axis (angiogenesis); LAMC1 drives preadipocyte differentiation → releases free fatty acids (metabolic reprogramming); NSUN2/ORAI2 activates PI3K-AKT (proliferation/migration)	Interact with tumor cells	Omentectomy; serum LAMC1 as biomarker; urinary CXCL2 as biomarker; FAO inhibitors
Angiogenesis	VEGF/VEGFR; ESM1/c-Met; CAF-derived HGF	VEGF/VEGFR primary pathway; ESM1 activates MAPK → HIF-1α/VEGFA/MMP-9; CAF-derived HGF promotes multiple vascular patterns	-	Anti-VEGF/VEGFR (bevacizumab, ramucirumab); ESM1 antibody; c-Met inhibitors; HGF inhibitors PI3K/MEK/ERK inhibitors

This table summarizes the key components of the tumor microenvironment in gastric cancer peritoneal metastasis, including cancer-associated fibroblasts, tumor-associated macrophages, angiogenic factors, and adipocytes. It highlights their pro-tumor mechanisms, intercellular interactions, and potential therapeutic targets for clinical translation. TME: Tumor microenvironment; CAFs: Cancer-associated fibroblasts; TAMs: Tumor-associated macrophages; FAP: Fibronectin attachment protein; IL-6R: Interleukin-6 receptor; apCAFs: Antigen-presenting cancer-associated fibroblasts; GDF-15: Growth differentiation factor 15; DKK1: Dickkopf-1; IL10: Interleukin-10; POU1F1: POU class 1 homeobox1; CTSL: Cathepsin L; SPP1: Secreted phosphoprotein 1; THBS2: Thrombospondin 2; VCAN: Versican; CXCL2: C-X-C motif chemokine ligand 2; FAO: Fatty acid oxidation; AKT: Protein kinase B; HIF1α: Hypoxia-inducible factor 1alpha; VEGFA: Vascular endothelial growth factor A; LAMC1: Laminin gamma 1; NSUN1: NOL1/NOP2/sun family; PI3K: Phosphatidylinositide 3-kinase; VEGF: Vascular endothelial growth factor; VEGFR: Vascular endothelial growth factor receptor; ESM1: Endothelial cell-specific molecule 1; HGF: Hepatocyte growth factor; MAPK: Mitogen-activated protein kinase; MMP-9: Matrix metallopeptidase 9.

Epigenetics

Epigenetics refers to the study of heritable changes in gene expression that occur without alterations in the DNA sequence. Its core molecular mechanisms encompass four levels: Non-coding RNA regulation, DNA methylation modification, covalent histone modification, and chromatin spatial conformation remodeling, which together constitute a sophisticated and complex gene expression regulatory network[135,136]. Epigenetic modifications are inherently reversible, a property that provides a theoretical foundation for the development of therapeutic strategies targeting epigenetic regulators[137]. Epigenetic drugs, including DNA methyltransferase inhibitors (e.g., azacitidine, decitabine) and histone deacetylase inhibitors (e.g., vorinostat, romidepsin), have been approved for hematologic malignancies but show limited efficacy in solid tumors[138,139]. In GCPM, the clinical application of these agents faces two major barriers: First, their lack of selectivity and off-target toxicity; second, the complexity of the epigenetic regulatory network that enables tumor cells to develop resistance. Understanding the specific epigenetic mechanisms driving GCPM is therefore critical for identifying selective targets and developing rational combination strategies, particularly with immunotherapy and targeted therapy.

Non-coding RNAs - from molecular sponges to pre-metastatic niche construction: Non-coding RNAs are among the most active molecular families in the epigenetic regulation of GCPM. A recent preclinical study found that circular RNA circPTBP3 is significantly elevated in plasma exosomes of patients with GCPM. It recruits the transcription factor TFAP2B to the SGK1 promoter region in mesothelial cells, activating SGK1 transcription and inducing mesothelial-mesenchymal transition, thereby driving PM[140]. This study is the first to demonstrate that the exosomal circular RNA circPTBP3 plays a key role in GCPM. Plasma exosomal circPTBP3 not only serves as a liquid biopsy biomarker with high diagnostic efficacy, but the circPTBP3/TFAP2B/SGK1 axis it mediates also represents a potential therapeutic target for GCPM. However, the clinical utility of plasma exosomal circPTBP3 as a liquid biopsy biomarker requires validation in large-scale, multicenter studies. Additionally, the downstream signaling network of SGK1 in GC cells, especially its interaction with classical EMT pathways, warrants further exploration. Corresponding to circRNAs, various lncRNAs and microRNAs (miRNAs) participate in GCPM by regulating mesothelial-mesenchymal transition. GC cell-derived miR-21-5p targets SMAD7, relieving inhibition of the TGF-β/Smad pathway. This drives transdifferentiation of peritoneal mesothelial cells into myofibroblasts and promotes peritoneal fibrosis[141]. In a PM model, a miR-21-5p inhibitor significantly suppressed the formation of metastatic nodules. This suggests that exosomal miR-21-5p may serve as a promising diagnostic biomarker and therapeutic target for GCPM. miR-301a-3p and miR-93-5p enhance the proliferation, invasion, and EMT capabilities of GC cells through regulation of the HIF-1α and STAT3 pathways[84,142]. Conversely, some miRNAs exert tumor-suppressive roles; for example, miR-22 inhibits ECM remodeling and EMT by suppressing MMP14 and Snail expression[143]. These preclinical studies reveal the dual roles of non-coding RNAs in GCPM, yet most evidence is derived from in vitro experiments or small-sample clinical studies. The expression differences and functional specificities of these non-coding RNAs across distinct GCPM molecular subtypes remain to be systematically elucidated.

In the field of lncRNAs, large-scale clinical cohort analyses have shown that lnc-RFNG-1 and lnc-TRIM28-14 are significantly upregulated in GCPM tissues. Their expression is negatively correlated with shortened overall survival in patients, suggesting their potential as novel diagnostic biomarkers[144]. Preclinical evidence has shown that the classic lncRNA HOTAIR relieves inhibition of HER2 by competitively binding to miR-331-3p[145]. This activates downstream HER2 signaling and promotes GC cell proliferation, apoptosis resistance, and migration. This finding provides an important addition to the molecular mechanisms underlying HER2-positive GC, suggesting that HER2 overexpression may arise from fine-tuned regulation at the post-transcriptional network level, and offers a new theoretical explanation for resistance to clinical HER2-targeted therapy. However, whether HOTAIR participates in gastric carcinogenesis through other miRNAs or non-ceRNA mechanisms (such as chromatin modification), and whether there exists a bidirectional positive feedback loop between HOTAIR and HER2, remain scientific questions that urgently need to be addressed.

A pioneering study revealed the role of RNA m⁶A modification in GCPM: The methyltransferase METTL3 methylates RAB27A mRNA, promoting exosome biogenesis and release[146]. The exosome-enriched miR-17-92 cluster targets and inhibits SRCIN1. This activates SRC signaling in peritoneal macrophages, promotes the release of immunosuppressive factors, and suppresses T cell function. Together, these events construct an immunosuppressive pre-metastatic niche. This study was the first to link three critical processes in a sequential manner: RNA modification, exosome biogenesis, and pre-metastatic niche establishment. This provides a novel target for GCPM immunotherapy. However, the specific roles of other immune components beyond macrophages (such as neutrophils and MDSCs) within the pre-metastatic niche in GCPM remain an area of insufficient research. Encouragingly, a recent preclinical study reported the anti-tumor activity of the METTL3 inhibitor STM2457 in GC[147].

Furthermore, lncRNAs participate in GCPM through metabolic reprogramming. LINC00924 regulates alternative splicing of Mnk2 by binding to hnRNPC, activating the p38/peroxisome proliferator-activated receptor alpha (PPARα) pathway and enhancing fatty acid oxidation to support the survival of GC cells upon matrix detachment[148]. TENM3-AS1 enhances FASN expression through interaction with hnRNPK, promoting fatty acid synthesis[149]. In the regulation of ferroptosis, hypoxia-induced lncRNA-PMAN promotes glutathione synthesis by enhancing SLC7A11 mRNA stability[92]. This inhibits ferroptosis and facilitates GC cell peritoneal colonization. BDNF-AS promotes FBXW7 promoter methylation by recruiting WDR5, inhibiting VDAC3 ubiquitin-mediated degradation, disrupting iron ion homeostasis, and enabling ferroptosis evasion[150]. These preclinical findings reveal the diverse mechanisms by which lncRNAs support GCPM through metabolic reprogramming, offering new insights for targeting metabolism-epigenetics crosstalk. However, the fine-tuned regulatory networks between lncRNAs and metabolic enzymes still require systematic elucidation. Table 5 summarizes the diverse non-coding RNAs implicated in GCPM and their translational potential as biomarkers and therapeutic targets.

Table 5 Summary of non-coding RNAs involved in gastric cancer peritoneal metastasis.

ncRNA	Location	Association with GC	Targeted genes and pathways	Application
CircPTBP3[140]	Plasma with GC, peritoneal metastasis and cell lines	Promote MMT	SGK1	Diagnostic and therapeutic, potential
lncRNA BDNF-AS[150]	Primary and peritoneal metastatic GC tissue	Inhibit ferroptosis	FBXW7	Prognosis and therapeutic, potential
lncRNA SEMA3B-AS1[193]	Primary and peritoneal metastatic GC tissue	Promote ubiquitination	FBXW7	Tumor suppressor
lncRNA-PMAN[92]	Peritoneal metastatic GC tissue	Inhibit ferroptosis	SLC7A11	Prognosis and therapeutic, potential
LINC00924[148]	GC cells	Lipid metabolism remodeling	Mnk2	Potential therapeutic effector
lncRNA CASC15[194]	Primary and peritoneal metastatic GC tissue	Promote EMT	JNK and p38	Predicting recurrence
lncRNA SNHG12[195]	Extracellular vesicles	Promote HPMC apoptosis and MMT	MAPK/ERK	Potential therapeutic target
lncRNA CTSLP4[196]	GC tissue	Suppress MMT	Snail	Tumor suppressor
lncRNA CCAT1[197]	GC tissue	Promote proliferation and metastasis	Bmi-1	Potential therapeutic target
lncRNA-HOXA11[198]	GC cells	Promote MMT	PDGF BB and TGF β1	Potential therapeutic target
miR-193a-3p[128]	Primary and peritoneal metastatic	Promote proliferation and metastasis	LAMC1	Tumor suppressor
miR-466[199]	GC tissue	Inhibit proliferation, migration and invasion	Unknown	Tumor suppressor
has-let-7 g-3p and has-miR-10395-3p[200]	Samples from peritoneal lavage	Promote metastasis and chemotherapy resistance	Unknown	Predicting peritoneal metastasis and the efficacy of chemotherapy
miR-106a[75]	Exosome	Promote MMT	Smad7 and TIMP2	Potential therapeutic target
miR-136[201]	GC tissue	Inhibit migration and invasion	HOXC10	Tumor suppressor
miR-493[202]	Exosome	Promote chemotherapy resistance	MAD2 L1	Predicting the efficacy of chemotherapy
miR-214[203]	GC cells	Promote proliferation, migration and invasion	PTEN	Potential therapeutic target
miR-21-5p[141]	Exosome	Promote MMT	SMAD7	Potential therapeutic target
miR-106a[204]	Exosome	Promote MMT	SMAD7	Potential therapeutic target
miR-22[143]	GC tissue	Inhibit ECM remodeling and EMT	MMP14 and Snail	Tumor suppressor
miR-30a[205]	GC tissue	Inhibit migration	FAPα	Tumor suppressor
miR-93-5p[142]	GC tissue	Promote proliferation, migration and invasion	IFNAR1	Potential therapeutic target
miR-544[206]	Extracellular vesicles	Promote migration and invasion	PLZF	Potential diagnostic and therapeutic targets
miR-196a-5p[207]	Exosome	Promote MMT	FBXO45	Potential therapeutic target
miR-199a-3p[208]	Primary and peritoneal metastatic GC tissue	Promote EMT	DDR2	Potential therapeutic target
miR-338-3p[209]	GC tissue	Inhibit migration	PTP1B	Tumor suppressor
miR-370[210]	GC tissue	Promote EMT	UQCRC2	Potential therapeutic target

This table catalogs the diverse functions of circular RNAs, long non-coding RNAs, and microRNAs in gastric cancer peritoneal metastasis. It details their cellular or exosomal location, associated mechanisms (e.g., promoting mesothelial-mesenchymal transition, inhibiting ferroptosis), targeted genes/pathways, and their potential applications as diagnostic biomarkers or therapeutic targets. ncRNA: Non-coding RNA; GC: Gastric cancer; circRNAs: Circular RNAs; MMT: Mesothelial-mesenchymal transition; SGK1: Serum and glucocorticoid-regulated kinase 1; lncRNAs: Long non-coding RNAs; BDNF: Brain-derived neurotrophic factor; miRNAs: MicroRNAs; EMT: Epithelial-mesenchymal transition; JNK: Jun N-terminal kinase; HPMC: Human peritoneal mesothelial cell; MAPK: Mitogen-activated protein kinase; PDGF BB: Platelet-derived growth factor-BB; TGF: Transforming growth factor; LAMC1: Laminin gamma 1; TIMP2: Tissue inhibitor of metalloproteinases 2; HOXC10: Homeodomain-containing gene 10; PTEN: Phosphatase and tensin homolog; MMP14: Matrix metallopeptidase 14; FAPα: Fibroblast activation protein alpha; IFNAR1: Interferon (alpha and beta) receptor 1; PLZF: Promyelocytic leukemia zinc finger protein; DDR2: Discoidin domain receptor 2; PTP1B: Protein tyrosine phosphatase 1B; UQCRC2: Ubiquinol-cytochrome c reductase core protein 2.

DNA methylation - molecular switches and clinical translation potential: DNA methylation determines the “on” or “off” state of genes by regulating the methylation status of gene promoter regions, playing a critical role in GCPM. During tumorigenesis, aberrant hypermethylation of promoter CpG islands leads to transcriptional silencing of tumor suppressor genes[151,152]. In GC, hypermethylation of promoters of genes such as CDH1, p16, and RUNX3 is not only closely associated with tumor proliferation, invasion, and metastasis but also significantly impacts patient prognosis[153]. CDH1 methylation is positively correlated with tumor size, depth of invasion, lymph node metastasis, and distant metastasis, and may serve as a potential molecular marker for assessing malignant progression in GC[154].

Aberrant activation of the PI3K/AKT/mTOR pathway is one of the core events in the malignant progression of GC. Preclinical mechanism studies have shown that hypermethylation of promoter regions in tumor suppressor genes such as phosphatase and tensin homolog leads to their transcriptional silencing. This relieves inhibition of the PI3K/AKT/mTOR pathway and synergistically promotes autophagy dysregulation, EMT, apoptosis inhibition, and chemotherapy resistance[155]. Although inhibitors targeting the PI3K/AKT/mTOR pathway have shown some promise in clinical trials, their efficacy as monotherapy is limited[156]. Given the inherent reversibility of epigenetic modifications, combining DNA methyltransferase inhibitors with pathway inhibitors, conventional chemotherapy, or molecular targeted agents has emerged as a highly promising therapeutic strategy. This approach has the potential to reverse malignant phenotypes and overcome resistance induced by monotherapy. Notably, METTL3 inhibitors have shown promising anti-tumor effects in GC[147].

A recent study found that METTL9 enhances respiratory chain complex I activity by mediating histidine methylation of mitochondrial proteins such as NDUFB3[157]. This reshapes cellular energy metabolism and promotes peritoneal dissemination in scirrhous GC. Hypermethylation of the THBS1 gene promoter is strongly correlated with peritoneal dissemination, accelerated tumor progression, and shortened overall survival in patients[158]. It may also serve as a molecular indicator for assessing GC aggressiveness and predicting GCPM risk. These preclinical studies are the first to directly connect DNA methylation with mitochondrial metabolism and peritoneal dissemination, thereby expanding the functional dimensions of epigenetic regulation in GCPM. However, future research is needed to clarify several issues: The complete substrate spectrum of METTL9, its molecular mechanisms in regulating mitochondrial function, and its clinical translational potential.

Histone modifications - from acetylation to the functional spectrum of lactylation: Histone modifications precisely regulate gene transcription by altering chromatin spatial conformation and accessibility. Histone acetylation is dynamically regulated by HATs and HDACs. HATs catalyze the addition of acetyl groups, which neutralizes the positive charge of histones, leading to chromatin relaxation and gene activation. In contrast, HDACs remove acetyl groups, restoring chromatin compaction and suppressing gene expression[159]. In GCPM, HDAC3 epigenetically silences lncRNA-LET, relieving its inhibition of miR-548k, thereby promoting GC cell proliferation, invasion, and migration while inhibiting apoptosis[160]. Small molecule inhibitors targeting HDAC3 can restore the epithelial phenotype and significantly inhibit tumor cell invasion and GCPM[161]. Furthermore, HDACs mediate the “permissive” effect of interferon-γ on PD-L1 expression, providing a theoretical basis for combination therapy targeting HDACs and immune checkpoints[162]. These studies indicate that HDACs play a central role in GCPM, but the selectivity and safety of HDAC inhibitors in clinical applications still require optimization.

Histone phosphorylation dynamically regulates chromatin structure and gene expression through the addition of negatively charged phosphate groups. Persistent activation of the STAT3 signaling pathway in the GC microenvironment is mediated by MSK1. This activation induces histone H3 phosphorylation, which reshapes the chromatin landscape and co-activates NFAT signaling. Consequently, this drives aberrant expression of pro-inflammatory cytokine networks and reinforces malignant transformation of gastric epithelial cells[163]. In vivo experiments using a subcutaneous xenograft model further confirmed that an MSK1 inhibitor significantly suppressed tumor growth. Histone methylation also participates in GCPM regulation: SETD1A forms a transcriptional complex with HIF1α, catalyzing H3K4 methylation and activating transcription of the glycolytic genes HK2 and PFK2, thereby reprogramming energy metabolism[164].

In recent years, lactylation has been identified as a novel epigenetic mechanism. CAF-secreted LOX activates the TGFβ/IGF1 axis, driving glycolytic reprogramming in GC cells and causing lactate accumulation. Lactate subsequently induces histone H3K18 lactylation, which increases chromatin accessibility at the PD-L1 promoter and activates PD-L1 transcription[71]. In addition, primary cell experiments confirmed that dasatinib inhibits LOX expression in CAFs. This indirectly suppresses EMT, glycolysis, and PD-L1 expression in GC cells. This mechanism suggests that dasatinib and immune checkpoint inhibitors may have synergistic anti-tumor effects. AARS1, as a novel lactyltransferase, directly catalyzes the lactylation of YAP/TEAD1 using lactate and ATP, activating the transcription of downstream oncogenes[93]. Notably, the YAP/TEAD1 pathway is one of the key pathways involved in anoikis resistance and PM of GC cells[23]. These preclinical findings reveal a cascade regulatory axis of “metabolic reprogramming-lactylation-immunity/metastasis”, providing a new strategy for targeting the metabolism-epigenetics interface. Nevertheless, AARS1 also functions as a tRNA synthetase, which is essential for protein synthesis. Therefore, systemic inhibition of AARS1 may cause severe toxicity. In the future, selective inhibitors should be developed to specifically target its lactyltransferase activity without affecting its tRNA synthetase function. Moreover, elevated levels of histone H3K18 lactylation in GC are associated with poor prognosis[165], suggesting its potential value as a prognostic biomarker.

Epigenetic mechanisms, through a multi-layered network involving non-coding RNA regulation, DNA methylation, and histone modifications, deeply participate in the GCPM process. Non-coding RNAs play a central role in GCPM as molecular sponges, regulators of metabolic reprogramming, and architects of the pre-metastatic niche. DNA methylation influences malignant tumor phenotypes by regulating the expression of tumor suppressor genes and key pathway genes. Histone modifications, ranging from acetylation, phosphorylation, and methylation to the newly discovered lactylation, constitute a dynamically reversible and finely tuned regulatory network. Current research limitations include: First, the cross-regulatory mechanisms among non-coding RNAs, DNA methylation, and histone modifications have not been systematically elucidated; second, most studies are based in vitro experiments or small-sample clinical cohorts, lacking large-scale prospective validation; third, the clinical translation of epigenetic-targeted drugs still faces challenges such as selectivity, safety, and resistance; fourth, the functional spectrum and regulatory networks of novel modifications such as lactylation in GCPM require further exploration. Future efforts should integrate multi-omics technologies (such as single-cell epigenomics and spatial transcriptomics) to systematically dissect the spatiotemporal dynamics of epigenetic regulatory networks. Patient-derived organoid and humanized mouse models should be established to validate the clinical translational potential of key targets. Combination strategies combining epigenetic drugs with immunotherapy and targeted therapy should be explored, with the goal of providing more effective precision treatment options for patients with GCPM.

Metabolic reprogramming: Synergistic networks of energy metabolism and metastatic adaptation

Metabolic reprogramming is one of the hallmark features of cancer and is closely associated with the onset and progression of GCPM. During GCPM progression, tumor cells develop a unique energy supply system adapted to the metastatic microenvironment through coordinated regulation of three core pathways: Glycolysis, lipid metabolism, and amino acid metabolism. This metabolic remodeling not only provides bioenergy and biosynthetic precursors for tumor cells but also forms a self-reinforcing cycle with the immunosuppressive microenvironment, collectively driving metastatic progression. Targeting tumor metabolism has become a promising therapeutic strategy. Several metabolic inhibitors have entered preclinical and clinical trials for solid tumors. For example, pyruvate kinase M2 (PKM2) activators and fatty acid synthase (FASN) inhibitors have shown preliminary efficacy in preclinical and clinical stages, respectively[166,167]. However, their application in GCPM remains largely unexplored. Two major barriers hinder clinical translation: First, the remarkable metabolic plasticity of cancer cells allows them to switch between metabolic pathways, leading to adaptive resistance; second, the complex interplay between tumor metabolism and the immunosuppressive microenvironment limits the efficacy of single-agent metabolic interventions. Understanding the specific metabolic dependencies of GCPM cells is therefore critical for identifying actionable nodes and developing rational combination strategies.

Glycolytic reprogramming - the Warburg effect and non-metabolic functions: Even under conditions of ample oxygen availability, GCPM cells exhibit a pronounced dependence on the glycolytic pathway, a phenomenon known as the Warburg effect. The activity and expression levels of key glycolytic enzymes, such as hexokinase, PKM2, and lactate dehydrogenase A, are significantly upregulated in GC cells. This not only accelerates glucose catabolism to provide abundant ATP but also promotes the accumulation of biosynthetic precursors like pentose phosphate and serine, meeting the demands of rapid proliferation[168,169]. Clinical cohort analysis has indicated that aberrant overexpression of PKM2 is closely associated with enhanced tumor invasiveness and poor patient prognosis[170]. In addition to its classical glycolytic function, preclinical evidence has shown that PKM2 can translocate to the nucleus. It then acts as a co-activator to regulate multiple signaling pathways, thereby promoting GC cell proliferation, invasion, and metastasis[171-173]. These studies reveal the critical role of the non-metabolic functions of metabolic enzymes in tumor progression, expanding our understanding of metabolic reprogramming. However, the precise regulatory mechanisms governing PKM2 nuclear translocation and the functional differences across distinct GCPM subtypes remain to be elucidated.

Modulating the ubiquitin-proteasome degradation pathway of the hnRNPA1 protein can effectively block the alternative splicing of pyruvate kinase pre-mRNA towards the PKM2 isoform, reversing the Warburg effect and significantly inhibiting the glycolytic activity and peritoneal metastatic capacity of GC cells[174]. LINC00589 exerts tumor-suppressive effects by promoting ubiquitin-mediated degradation of hnRNPA1, according to mechanistic studies. Using this mechanism, the team delivered a LINC00589-encoding plasmid via polyethyleneimine-modified mesoporous silica nanoparticles. They demonstrated its therapeutic potential against GCPM in vitro and in vivo models. This finding offers a new strategy for targeting tumor metabolism, but its safety and efficacy in vivo still need further validation.

Lipid metabolic reprogramming - dual drivers of de novo synthesis and oxidative utilization: During GCPM progression, tumor cells exhibit a pronounced metabolic dependency on fatty acids. Clinical cohort analysis has revealed that FASN, the key rate-limiting enzyme of the fatty acid synthesis pathway, is specifically upregulated in peritoneal metastatic lesions[175]. Its high expression is significantly associated with poor patient prognosis. These findings suggest that FASN may serve as a predictive marker for peritoneal dissemination and an independent prognostic factor in GC. Preclinical mechanistic studies have shown that the PI3K-AKT-mTOR complex 1 pathway upregulates monounsaturated fatty acid synthesis. This inhibits phospholipid peroxidation and blocks ferroptosis, thereby conferring ferroptosis resistance to cancer cells[176]. Based on these findings, combining an mTOR complex 1 inhibitor with a ferroptosis inducer induced near-complete tumor regression in vivo. This combination offers a promising therapeutic strategy for cancer patients with PI3K-AKT-mTOR pathway mutations. Fatty acids are essential building blocks of the cell membrane bilayer. They also influence tumor cell proliferation, survival, and metastasis by participating in second messenger generation and dynamically regulating membrane receptors[177].

In addition to fatty acid synthesis, fatty acid oxidation serves as an important supplementary pathway for tumor energy metabolism and also plays a critical role in GCPM. LINC00924 binds to hnRNPC and regulates the alternative splicing of Mnk2 pre-mRNA. This inhibits p38 MAPK signaling and activates PPARα transcriptional activity. Consequently, it synergistically enhances fatty acid oxidation and uptake, providing energetic support for GCPM progression[148]. Notably, tumor cells can actively take up free fatty acids from the peritoneal microenvironment, substantially increasing fatty acid oxidation flux through exogenous supplementation[128,129]. These studies reveal the central role of lncRNA-mediated lipid metabolic reprogramming in GCPM. Importantly, preclinical model studies have confirmed that PPARα inhibitors significantly suppress tumor growth and PM. This is achieved by inhibiting fatty acid oxidation and uptake. However, the cross-regulatory networks among lipid metabolism, glycolysis, and amino acid metabolism remain to be systematically elucidated.

Amino acid metabolic reprogramming - glutamine dependency and immunometabolic crosstalk: Glutamine is one of the primary energy sources for tumor cells, and its uptake and metabolism are significantly enhanced during the malignant progression of GC[178,179]. Glutamine is converted to glutamate to participate in the tricarboxylic acid cycle, providing energy and biosynthetic intermediates for cells, while also regulating cellular redox status and signal transduction[180]. Furthermore, α-ketoglutarate generated from glutamine metabolism serves as a cofactor for histone demethylases, influencing gene expression and thereby modulating cell invasion and metastatic capacity[181-183].

Of particular importance, tumor cells actively secrete arginine to induce polarization of TAMs toward an M2-like pro-tumor phenotype. These reprogrammed macrophages highly express arginase 1, depleting arginine in the microenvironment and directly leading to functional exhaustion of CD8+ T cells[184]. This finding reveals a direct causal link between metabolic reprogramming and immunosuppression, providing a new perspective for targeting the metabolism-immunity interface. However, the spatial dynamics of arginine metabolism at different stages of GCPM and its quantitative relationship with T cell exhaustion still require in-depth exploration.

Neural-tumor metabolic crosstalk - an emerging regulatory dimension: Recent studies have revealed a significant interaction between tumors and the nervous system. Tumor cells actively recruit neural progenitor cells into the TME. They also secrete neurotrophic factors like nerve growth factor and brain-derived neurotrophic factor. Collectively, these activities create a microenvironment that supports nerve fiber infiltration. This phenomenon is positively correlated with tumor invasion depth and metastatic potential[185]. A groundbreaking study further challenges traditional understanding. Neurons can directly “donate” functional mitochondria to tumor cells through intercellular mitochondrial transfer. This significantly enhances the oxidative phosphorylation capacity of tumor cells and provides exceptional bioenergetic support for their metastatic process[186]. Encouragingly, chemical denervation with BoNT/A (botulinum toxin) in preclinical models led to a significant downregulation of metabolism-related pathways in cancer cells, particularly the tricarboxylic acid cycle, as shown by transcriptomic analysis. This finding suggests that denervation strategies can effectively reduce mitochondrial transfer and may have therapeutic potential. And this study is the first to reveal a previously unrecognized metabolic symbiotic relationship in neural-tumor interactions, offering a novel perspective for understanding the metabolic adaptation in GCPM. However, the specific mechanisms of mitochondrial transfer (tunneling nanotubes? extracellular vesicles?) and its functional contribution in GCPM still require further validation.

GC cells develop a unique energy supply system adapted to the metastatic microenvironment through the coordinated activation of three fundamental metabolic pathways: Glycolysis, fatty acid metabolism, and glutaminolysis. The Warburg effect not only enables rapid ATP generation but also reshapes the TME through lactate accumulation. De novo fatty acid synthesis and oxidation pathways both fulfill the biosynthetic demands for membrane structures and meet sustained energy requirements. Glutamine metabolism maintains malignant progression by sustaining tricarboxylic acid cycle activity and facilitating epigenetic regulation. Importantly, metabolic reprogramming forms a self-reinforcing cycle with the immunosuppressive microenvironment, wherein metabolites directly modulate immune cell function, while emerging mechanisms such as neural-tumor mitochondrial transfer further broaden the scope of metabolic adaptation (Table 6 summarizes the three core metabolic pathways involved in metabolic reprogramming in GCPM, their key enzymes, mechanisms of action, functional consequences, and therapeutic opportunities). However, current understanding of metabolic reprogramming remains largely based on static, single-pathway analyses. The cross-regulatory networks among the three major metabolic pathways and their dynamic evolutionary patterns at different stages of GCPM have yet to be systematically elucidated. The bidirectional interactive mechanisms between metabolic remodeling, immunosuppression, and epigenetic modification remain largely at the correlational level, lacking direct causal evidence. The functional contribution and therapeutic potential of neural-tumor metabolic crosstalk, an emerging field, are yet to be thoroughly explored. Overcoming these bottlenecks requires a shift from single-pathway analysis to multi-dimensional, dynamic research frameworks. This involves three key approaches. First, leverage advanced technologies such as single-cell metabolomics, spatial metabolomics, and live-cell metabolic imaging to precisely map metabolic remodeling in GCPM across time and space. Second, establish patient-derived organoid and humanized mouse models to validate the feasibility and safety of key metabolic nodes as therapeutic targets. Third, actively explore synergistic combinations of metabolic interventions (e.g., PKM2 inhibitors, FASN inhibitors, and glutamine antagonists) with immunotherapy and epigenetic therapy. Together, these strategies can advance metabolic targeted therapy from the lab to the clinic, offering new hope for patients with GCPM.

Table 6 Key pathways and mechanisms in metabolic reprogramming in gastric cancer peritoneal metastasis.

Metabolic pathway	Key molecule/enzyme	Mechanism of action	Functional	Clinical potential
Glycolysis (Warburg effect)	PKM2	Accelerated glucose metabolism; translocates to nucleus as co-activator	Promotes proliferation, invasion, metastasis	Target PKM2 (PKM2 inhibitors or PKM2 activators)
	LINC00589	Regulates alternative splicing of pyruvate kinase pre-mRNA toward PKM2 isoform	Drives Warburg effect	LINC00589 delivery via PMSNs
Lipid metabolism	FASN	Accelerated fatty acid anabolism	Promote proliferation, survival, and metastasis	FASN inhibitors
	PI3K-AKT-mTORC1	Upregulates monounsaturated fatty acid synthesis; inhibits phospholipid peroxidation	Confers ferroptosis resistance	mTORC1 inhibitor combined with ferroptosis inducer
	LINC00924/hnRNPC/Mnk2	Regulates alternative splicing of Mnk2 → inhibits p38 MAPK → activates PPARα	Enhances fatty acid oxidation and uptake	PPARα inhibitors
Amino acid metabolism	Glutamine	Converted to glutamate → enters TCA cycle; regulates redox status and signal transduction	Promotes proliferation, invasion, metastasis	Glutaminase inhibitors
Neural-tumor crosstalk	Mitochondrial transfer	Neurons donate functional mitochondria to tumor cells via intercellular transfer	Enhances oxidative phosphorylation; provides bioenergetic support for metastasis	Chemical denervation

This table summarizes the core metabolic pathways involved in gastric cancer peritoneal metastasis: Glycolysis, lipid metabolism, and amino acid metabolism. It highlights key enzymes, mechanisms of action, functional consequences, and therapeutic opportunities. PKM2: Pyruvate kinase M2; PMSNs: Polyethyleneimine-modified mesoporous silica nanoparticles; FASN: Fatty acid synthase; PI3K: Phosphatidylinositide 3-kinase; AKT: Protein kinase B; mTORC1: Mechanistic target of rapamycin complex 1; MAPK: Mitogen-activated protein kinase; PPARα: Peroxisome proliferator-activated receptor alpha; TCA: Tricarboxylic acid.

CONCLUSION

Current research has systematically elucidated the complex biological basis of GCPM from multiple dimensions. These include genetic mutations, epigenetic reprogramming, metabolic adaptation, and dynamic remodeling of the TME. These advances have begun to translate into clinical opportunities at three levels. First, several candidates have shown promise as diagnostic and prognostic biomarkers. ctDNA mutations (TP53, CDH1), exosomal non-coding RNAs (circPTBP3, miR-301a-3p), and methylated DNA markers (THBS1) may enable early detection of peritoneal micrometastases and risk stratification. Serum proteins such as LAMC1 and urinary CXCL2 also offer non-invasive diagnostic potential. Second, actionable nodes have been identified as therapeutic targets across multiple pathways. These include the YAP/TEAD1 axis (via GRP94 or verteporfin), the C3-C3aR1 axis for CAF-TAM crosstalk, metabolic enzymes (PKM2, FASN, glutaminase), and epigenetic regulators (HDAC3, METTL3). Third, preclinical evidence supports combination strategies to overcome resistance. These include pairing metabolic inhibitors with immunotherapy, HDAC inhibitors with immune checkpoint blockade, and anti-angiogenic agents with C3aR antagonists.

Despite these advances, several key gaps remain: (1) How can we reliably detect peritoneal micrometastases before they become clinically evident? (2) What is the temporal sequence and clonal evolution of driver mutations during GCPM progression? (3) How can we overcome the functional heterogeneity and plasticity of CAFs and TAMs? These features are likely the basis of resistance to current therapies; (4) Can the novel mechanisms of neural-tumor metabolic crosstalk be safely targeted? and (5) Most existing evidence comes from preclinical models or small retrospective cohorts. Large-scale prospective validation is lacking. Addressing these questions will require integrated multi-omics analyses, patient-derived models, and biomarker-driven clinical trials. Such efforts are essential to translate molecular insights into precision interventions. Only then can we establish effective multi-target combination strategies and ultimately improve clinical outcomes for patients with GCPM.

ACKNOWLEDGEMENTS

I would like to express my deep gratitude for the FiG DRAW 2.0 online drawing page for providing creative elements for the production of images for this review.