Editorial reviews multi-omics evidence on metastatic cancer ecosystems

Multi-omics research reveals that metastatic tumors are not simply collections of malignant cells but active ecosystems in which cancer cells, immune cells, fibroblasts and extracellular matrix interact to support growth, evade immunity and resist treatment. An editorial in Frontiers in Oncology draws together 15 studies using single-cell transcriptomics, spatial profiling, proteomics, epigenomics, radiomics and functional validation to describe this post-metastatic tumor microenvironment. The central conclusion is that advanced cancer may require simultaneous attention to the metastatic “seed” and the surrounding “soil,” although most of the proposed molecular targets and clinical models still need prospective validation.

The metastatic environment can conceal biologically important subclones that are diluted by traditional bulk-omics methods, which average signals across many cell populations. The studies identify a series of molecular hubs at the tumor-stroma interface, including SERPINE2, a factor reported to be consistently overexpressed in metastatic tumors. Mechanistic work summarized by the editorial links SERPINE2 to stabilization of epidermal growth factor receptor through reduced c-Cbl-mediated ubiquitination, sustained STAT3 and ERK signaling, and increased ATM-driven DNA repair associated with radioresistance. As a secreted factor, SERPINE2 also polarizes macrophages toward an M2 phenotype, activates cancer-associated fibroblasts and remodels the extracellular matrix. These findings nominate a molecule with both tumor-intrinsic and stromal functions, but they do not establish that targeting it will improve outcomes in patients.

Metabolic adaptation emerged as another major theme. In hepatocellular carcinoma, TRIM25, an E3 ubiquitin ligase, was associated with suppression of ferroptosis through lower ferrous iron and lipid peroxidation and stronger antioxidant defenses. Knocking down TRIM25 sensitized cells to ferroptotic death, while high expression correlated with worse survival in The Cancer Genome Atlas and International Cancer Genome Consortium cohorts. In angioimmunoblastic T-cell lymphoma, a different study implicated lysosomal alpha-glucosidase, or GAA, a glycogen-degrading enzyme, in poor chemotherapy response. GAA-high tumors showed enhanced glycolysis, reduced oxidative phosphorylation and signatures of exhausted CD8-positive T cells, with results validated in independent Gene Expression Omnibus cohorts. Together, the studies suggest that altered fuel use may connect metabolic vulnerability, immune dysfunction and resistance across solid and hematologic malignancies, although the lymphoma findings were correlative and based on a relatively small sample.

Neutrophils and epigenetic regulation provide a further route by which the metastatic niche may reorganize. A study of lung adenocarcinoma found that neutrophil extracellular traps carrying IL-17A promoted H2BC4 acetylation through recruitment of the histone acetyltransferase p300. This molecular change was linked to epithelial-mesenchymal transition and brain metastasis. Disrupting neutrophil extracellular traps with PAD4 inhibitors or neutralizing IL-17A reduced metastasis in mouse models, providing functional evidence in that preclinical setting. The work does not demonstrate the same effect in human patients.

Single-cell and spatial analyses sharpen the view by showing that important cells occupy specialized niches. In pancreatic ductal epithelial cells, PTPRR was reported to be almost exclusively expressed and elevated in metastatic lesions, with high expression correlated with poor survival. PTPRR-high cells formed niches with inflammatory cancer-associated fibroblasts through endothelin, EGF and PDGF signaling and engaged TGF-beta, CCL and CXCL pathways associated with immunosuppression. In papillary thyroid carcinoma, tumors without thyroiditis showed extracellular-matrix remodeling and activation of PI16-positive fibroblasts, while tumors accompanied by thyroiditis instead displayed immune pathways without comparable fibroblast infiltration. A 17-gene machine-learning model predicted lymph-node metastasis in this setting, and Mendelian randomization identified SHISA5 as a potentially causal gene. The contrast between the two thyroid-cancer groups also demonstrates that a shared histological label can conceal biologically distinct microenvironments.

Several projects translated molecular observations into prediction tools. A five-gene model for esophageal squamous cell carcinoma included BID, CBX3, ECT2, KIF14 and SORBS2 and achieved an area under the receiver operating characteristic curve, or AUC, of 0.99; SORBS2 was associated with myofibroblasts and behaved as a protective factor. In gastric cancer, the extracellular-matrix-related ribosomal protein RPS29 was identified as a hub associated with migration and epithelial-mesenchymal transition, while high expression was linked to sensitivity to trifluoperazine. A pan-cancer model based on dynamic declines in CD8-positive T cells, combined with neutrophil-to-lymphocyte ratio, treatment line and cancer type, produced time-dependent AUCs of 0.727, 0.827 and 0.853 at 2.5, 3.5 and 5.7 months. A separate gastric cancer model combined muscle attenuation, neutrophil-to-lymphocyte ratio and weight loss to predict response to chemotherapy plus PD-1 blockade, reaching an AUC of 0.728.

Imaging studies attempted to make tumor ecosystems visible without direct molecular sampling. In advanced differentiated thyroid carcinoma, a 99mTc-FAPI-YQ3 SPECT/CT study was compared with FDG PET/CT. The two tracers showed concordance in only 52.4% of lesions: 42.7% were FDG-positive but FAPI-negative, while 4.9% were FDG-negative but FAPI-positive. The authors interpreted these differences as compatible with distinct metabolic and stromal phenotypes, and a radiomics model based on the imaging data reached an AUC of 0.951. They also cautioned that FAPI uptake reflects an imaging phenotype rather than direct proof of fibroblast activation protein expression. Separate CT-based radiomics work in non-small cell lung cancer connected an imaging score with hypoxia, epithelial-mesenchymal transition and immune exclusion, offering another route between imaging and tumor biology. The thyroid study was especially preliminary, with only 17 participants and no immunohistochemical validation of FAP expression.

Two reviews broadened the analysis beyond individual targets. One synthesized evidence that lactate accumulation, lipid-metabolism rewiring, ferroptosis, cuproptosis and the tryptophan-IDO1-kynurenine axis can contribute to immune suppression, arguing that multi-omics is needed to decode their interactions. The other examined KLF5 across cancers, finding that it was upregulated in most tumor types and associated with poor prognosis; experimental knockdown suppressed proliferation and migration in glioblastoma and bladder cancer models. Such pan-cancer analyses can identify shared biological patterns, but they also show why shared molecules may have tissue-specific functions. The editorial specifically notes that SERPINE2 could behave differently in lung and liver metastases and that PTPRR may vary with KRAS mutation status, limiting the feasibility of a uniform strategy across metastatic cancers.

Despite the breadth of the research, the evidence remains uneven. Much of the mechanistic work is correlative or comes from cells and animals, while large, independent and prospective patient cohorts are scarce. Drug-development challenges are also substantial. Although neutralizing antibodies or ligand traps could conceptually target extracellular factors such as SERPINE2 and PTPRR, their pharmacokinetics, penetration into metastatic tissue and toxicity against normal stroma remain unknown. The dynamic CD8-positive T-cell model also requires external validation in multicenter prospective studies, including in neoadjuvant or adjuvant settings where measurable lesions may be absent. For GAA in angioimmunoblastic T-cell lymphoma, functional experiments that manipulate expression are still needed to establish causality. These constraints mean that the reported AUCs and candidate targets should be treated as research leads rather than ready-made clinical tests or treatments.

The editorial’s forward-looking priorities are prospective validation, integration and actionable deployment. Multi-omics signatures will need testing in multicenter trials before they can guide treatment reliably. Combining spatial measurements with longitudinal imaging may allow researchers to track microenvironmental change during therapy rather than relying on a static biopsy. Artificial-intelligence systems could integrate these datasets, but the authors caution that they should progress beyond identifying molecular clusters to producing predictive models that can be tested at the bedside. The 15 articles do not demonstrate a new cure for advanced cancer; instead, they document a shift in how the disease is being investigated, from viewing metastasis as a genetic event alone to studying it as an ecosystem disorder shaped by immunity, metabolism, structure and treatment pressure.

Subject of Research: Multi-omics analysis of the tumor microenvironment after cancer metastasis

Article Title: Editorial: The insights of multi-omics into the microenvironment after tumor metastasis: a paradigm shift in molecular targeting modeling and immunotherapy for advanced cancer patients, volume II

Article References: Xie, M., Chang, H., Zhao, X., Huang, J., Tang, H., Gopinath, D., Xiao, X., & Wang, C. (2026). Editorial: The insights of multi-omics into the microenvironment after tumor metastasis: a paradigm shift in molecular targeting modeling and immunotherapy for advanced cancer patients, volume II. Frontiers in Oncology, 16, Article 1986038. https://doi.org/10.3389/fonc.2026.1986038

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1986038

Keywords: cancer metastasis, tumor microenvironment, multi-omics, single-cell transcriptomics, spatial profiling, immunotherapy, metabolic reprogramming, epigenetic regulation, radiomics, precision oncology, Editorial, insights