Gene Signature Linked to Metastasis May Predict Gastric Cancer Chemotherapy Response

Gastric cancer remains one of the world’s most lethal malignancies, and the failure of chemotherapy in many patients is often traced to two intertwined processes: resistance to the drugs themselves and the spread of tumor cells to distant organs. Now, a team of researchers from the Royan Institute for Stem Cell Biology and Technology, the University of Science and Culture, Shahid Beheshti University, and collaborating clinical centers in Tehran, together with the Princess Margaret Cancer Centre in Toronto, has identified a distinctive pattern of gene and microRNA co-expression that appears in both metastatic gastric cancer tissues and in three-dimensional tumor models exposed to cisplatin and 5-fluorouracil, the backbone drugs of gastric cancer chemotherapy. The study, published in Molecular Biology Reports, suggests that the molecular programs driving metastatic behavior and those activated during chemotherapy exposure may share more common ground than previously appreciated.

The investigation began with a computational sweep of pretreatment tumor samples from a publicly available gastric cancer cohort, GSE14210, obtained from the NCBI Gene Expression Omnibus repository. Because these samples were collected before any chemotherapy was administered, they offered a clean snapshot of the tumor’s baseline molecular state. Using differential expression analysis, the researchers identified 414 genes that were significantly upregulated in pretreatment tumors. They then mapped these genes onto protein interaction networks built with the STRING database, a tool that links proteins based on known and predicted functional associations. From this network, they extracted 63 highly connected hub genes, defined as nodes with a degree of at least ten, meaning each interacts with ten or more other proteins in the network. Twenty-one of these hub genes were also upregulated when tumor tissue was compared against normal gastric tissue using the GEPIA web platform, which draws on The Cancer Genome Atlas and Genotype-Tissue Expression datasets.

Hub genes are the molecular equivalent of central train stations in a city’s transit network: they coordinate the flow of information across multiple pathways simultaneously. By prioritizing these hubs, the researchers narrowed a list of hundreds of candidate genes down to a manageable set likely to exert broad influence over tumor behavior. They then applied a target-centric microRNA prioritization strategy, searching for microRNAs whose predicted and validated targets overlapped with the hub genes. MicroRNAs are short, non-coding RNA molecules, roughly 18 to 24 nucleotides long, that bind to complementary sequences on messenger RNAs and either degrade them or block their translation into proteins. Because a single microRNA can regulate dozens or even hundreds of targets, they act as master rheostats, fine-tuning gene expression thresholds and conferring robustness on biological processes.

With the computational candidates in hand, the team moved to the laboratory bench. They performed quantitative reverse transcription polymerase chain reaction, or qRT-PCR, on metastatic and non-metastatic gastric cancer tissues obtained from patients through the Iranian National Tumor Bank and fresh surgical specimens collected at collaborating hospitals. In metastatic tissues, six molecules stood out with statistically significant changes: the chemokine CXCL8, the matrix metalloproteinase MMP9, the transcription factor STAT1, and three microRNAs, miR-17-5p, miR-24-3p, and miR-124-3p, all significantly increased. Meanwhile, miR-145-5p, a microRNA with known tumor-suppressive functions, was significantly decreased. The chemokine CCL5, though a plausible candidate based on prior literature, did not reach statistical significance in this comparison.

CXCL8, also known as interleukin-8, is a pro-inflammatory chemokine that recruits immune cells and promotes angiogenesis, tumor growth, and invasion. MMP9 is an enzyme that digests collagen and other extracellular matrix components, clearing a physical path for tumor cells to migrate through tissue barriers. STAT1 is a transcription factor traditionally associated with interferon signaling and antiviral defense, but accumulating evidence shows it can also amplify metastatic potential and confer resistance to therapy in several cancer types. The fact that all three genes, together with the three upregulated microRNAs, showed significant increases specifically in metastatic tissues suggests they may form a coordinated expression module linked to the spread of gastric cancer.

The researchers then turned to a three-dimensional cell culture system. They grew MKN-45 gastric cancer cells as spheroids, compact aggregates that better mimic the architecture, nutrient gradients, and cell-cell interactions of real tumors than conventional flat monolayer cultures. These spheroids were exposed to cisplatin and 5-fluorouracil, a platinum-based DNA crosslinking agent and an antimetabolite that disrupts thymidine synthesis, respectively. After drug treatment, the same six molecules that had been elevated in metastatic tissues, CXCL8, MMP9, STAT1, miR-17-5p, miR-24-3p, and miR-124-3p, showed significant expression changes in the drug-exposed spheroids. CCL5 and miR-145-5p did not change significantly in this setting, mirroring the pattern seen in the tissue comparison.

Perhaps the most intriguing observation came from cyclic drug exposure. When the MKN-45 spheroids were repeatedly treated with cisplatin and 5-fluorouracil over multiple rounds, they regrew after each cycle, consistent with an adaptive drug-tolerant phenotype rather than outright cell death. This phenomenon, in which a subset of tumor cells enters a reversible, quiescent state that allows survival under chemotherapy pressure, has been described in several cancer types and is thought to contribute to minimal residual disease and eventual relapse. The fact that the same gene and microRNA signature appeared in both metastatic tissues and drug-tolerant spheroids raises the possibility that metastatic competence and chemotherapy tolerance share overlapping molecular mechanisms.

As a feasibility observation, the team included a single patient-derived spheroid model, created from tumor cells obtained directly from a gastric cancer patient. Unlike the MKN-45 spheroids, this patient-derived model showed limited recovery after drug exposure, suggesting that the adaptive drug-tolerant phenotype may vary considerably between individual tumors. Patient-derived organoids and spheroids are increasingly recognized as valuable tools for personalized drug screening, and their clinical relevance in gastric cancer has been demonstrated in recent studies. However, a single model cannot establish generalizability, and the researchers are careful to note this limitation.

The microRNA findings deserve particular attention. miR-17-5p, a member of the miR-17-92 cluster, has been shown to promote proliferation in gastric cancer cells by targeting SOCS6, a suppressor of cytokine signaling. miR-24-3p has a dual role in cancer, acting as an oncomiR in some contexts by targeting BCL2L11, a pro-apoptotic protein, and as a tumor suppressor in others by targeting RegIV. miR-124-3p, traditionally regarded as a neuron-specific microRNA, has emerged as a tumor suppressor in several cancers, including esophageal squamous cell carcinoma, where it inhibits proliferation and invasion by downregulating STAT3. miR-145-5p, the only microRNA that decreased in metastatic tissues, targets YES and STAT1 in colon cancer cells and suppresses migration and invasion through the SP1/NF-κB pathway in esophageal cancer. The coordinated dysregulation of these four microRNAs, three up and one down, suggests they may act together to shift the balance of the CXCL8-MMP9-STAT1 module toward a more aggressive, drug-tolerant state.

The authors are appropriately cautious in interpreting their findings. They emphasize that the study identifies context-dependent candidate co-expression patterns across metastatic tissues and chemotherapy-exposed three-dimensional models, but does not establish a shared causal mechanism. Functional validation in independent cohorts and additional patient-derived models will be required before any of these molecules can be considered reliable biomarkers of treatment response or metastatic risk. Nevertheless, the convergence of computational network analysis, patient tissue validation, and three-dimensional drug exposure experiments represents a methodologically rigorous approach to a problem that has long frustrated oncologists: why some gastric cancers respond to cisplatin and 5-fluorouracil while others do not. If future studies confirm that the CXCL8-MMP9-STAT1 module and its associated microRNAs truly bridge metastatic behavior and chemotherapy tolerance, they could open new avenues for predicting treatment outcomes and designing combination therapies that simultaneously target both processes, potentially transforming the management of one of the world’s most challenging cancers.

Subject of Research: gastric cancer chemotherapy response and metastasis gene signature

Article Title: Integrated in silico and in vitro analysis identifies a co-expression pattern associated with cisplatin/5-FU treatment response and metastatic features in gastric cancer

Article References: Iranmehr, Z., Sharahi, H. Y., Eslahchi, C., Ghaemi, F., Wadji, M. B., Miri, S. R., & Ebrahimi, M. (2026). Integrated in silico and in vitro analysis identifies a co-expression pattern associated with cisplatin/5-FU treatment response and metastatic features in gastric cancer. Molecular Biology Reports, 53(1), Article 1731. https://doi.org/10.1007/s11033-026-12888-3

Image Credits: AI Generated

DOI: 10.1007/s11033-026-12888-3

Keywords: Integrated, silico, vitro, analysis, identifies, co-expression, pattern, associated, cisplatin, treatment, response, metastatic