Gastric Cancer Assembloids and Stromal Drug Response

Patient-derived organoids preserve important features of tumor epithelium, but they commonly omit the stromal populations that shape inflammation, extracellular matrix remodeling, invasion, and treatment response. The 2025 study by Shapira-Netanelov and colleagues addresses this limitation by building gastric cancer assembloids from matched tumor organoids and stromal cell subpopulations. The full study is available in Cancers.

Study Background and Research Question

Gastric cancer remains a heterogeneous disease with substantial variation in molecular features and therapeutic response. The reference study notes that gastric cancer is the fifth most diagnosed carcinoma and the second leading cause of cancer-related death worldwide, while five-year survival for locally advanced, unresectable, or metastatic disease remains below 10% in the cited clinical context. These figures, reported and discussed in the reference paper, underscore why predictive models must account for more than cancer-cell genotype alone.

Most organoid workflows enrich for epithelial tumor cells because epithelial expansion is comparatively straightforward and supports high-throughput assays. However, cancer-associated fibroblasts, mesenchymal cells, endothelial cells, and other stromal populations can modify growth-factor availability, matrix architecture, inflammatory signaling, and drug penetration. The central research question was therefore whether stromal populations isolated from the same patient tumor could be incorporated into organoid cultures to create a more physiologically relevant, patient-specific model of gastric cancer biology and therapeutic response.

Key Innovation from the Reference Study

The study’s main innovation is the matched assembloid design. Rather than adding generic fibroblasts or commercially sourced stromal cells to an epithelial organoid, the investigators generated tumor organoids and several stromal populations from the same gastric cancer specimen. The stromal compartment included tumor-derived mesenchymal stem cells, fibroblasts, and endothelial cells, each expanded under a growth medium selected to support the relevant population.

These components were then recombined in an optimized co-culture medium. This design preserves a degree of patient-specific cellular history that is lost when tumor epithelium and stroma originate from unrelated donors. It also makes the assembloid suitable for comparing different epithelial-to-stromal compositions. In practical terms, the model allows investigators to ask whether a treatment effect is intrinsic to tumor cells or is altered by the cellular and extracellular environment surrounding them.

Methods and Experimental Design Insights

Tumor tissue was first dissociated into viable cell populations. The resulting cells were expanded in media supporting organoids, mesenchymal stem cells, fibroblasts, or endothelial cells. After expansion, the populations were co-cultured in an assembloid medium designed to maintain the major cellular compartments together. This staged approach is important because direct outgrowth from a heterogeneous tumor sample may otherwise allow one rapidly proliferating population to dominate the culture.

The investigators used immunofluorescence to examine epithelial and stromal biomarker expression, confirming that the assembloids retained multiple cellular identities. RNA sequencing was used to compare transcriptomic states, while cell viability assays measured responses to therapeutic agents. The study also evaluated different organoid-to-stromal ratios, enabling assessment of how model composition affects molecular phenotype and drug sensitivity. Together, these readouts connect structure, cell identity, gene expression, and pharmacology rather than treating viability as an isolated endpoint.

Protocol Parameters

• Matched specimen design: Literature-backed: derive tumor organoids and stromal subpopulations from the same patient tumor whenever the research question concerns patient-specific tumor–stroma interactions.
• Compartment-specific expansion: Literature-backed: use tailored growth media for organoid, mesenchymal stem-cell, fibroblast, and endothelial-cell expansion before recombination, as performed in the reference study.
• Assembloid assembly: Literature-backed: use a co-culture medium optimized to support the constituent cell types. The paper supports testing more than one organoid-to-stromal ratio rather than assuming that a single composition represents the tumor.
• Identity validation: Literature-backed: verify epithelial and stromal marker expression by immunofluorescence before interpreting treatment effects.
• Transcriptomic profiling: Literature-backed: use RNA sequencing to examine inflammatory cytokines, extracellular matrix remodeling, and tumor-progression programs that may not be apparent from morphology alone.
• Drug-response testing: Workflow recommendation: compare matched monocultures and assembloids under the same exposure and viability-readout conditions, then interpret discordant responses as evidence for microenvironmental modulation rather than as a simple technical failure.

Core Findings and Why They Matter

The optimized co-culture conditions produced assembloids that more closely reproduced the cellular heterogeneity of the primary tumors than epithelial organoid cultures alone. Immunofluorescence supported the presence of both epithelial and stromal compartments. This is a foundational result: without confirming compartmental retention, a nominal co-culture cannot be assumed to model tumor–stroma biology.

Compared with monocultures, the assembloids showed higher expression of inflammatory cytokines, extracellular matrix remodeling factors, and genes associated with tumor progression. These changes were observed across organoid and stromal compositions, indicating that the surrounding cells were not passive structural additives. They actively altered the transcriptional state of the model. For researchers, this provides a mechanistic rationale for including stromal cells when studying invasion, resistance, biomarker expression, or treatment-induced adaptation.

Drug screening revealed patient- and drug-specific variability. Some agents retained activity in both organoid and assembloid settings, whereas others became less effective after stromal components were incorporated. The latter observation is especially important for preclinical oncology research because an epithelial-only screen may overestimate treatment efficacy. A compound that appears potent against tumor cells in isolation may encounter altered signaling, matrix-associated protection, or other stromal influences in a more complex model.

The findings do not establish that every assembloid will predict clinical outcome, but they demonstrate that model architecture can change the biological interpretation of a drug response. The system may therefore help identify resistance mechanisms, prioritize combination strategies, and distinguish patient-specific sensitivity patterns before more resource-intensive studies.

Comparison with Existing Internal Articles

Two related internal resources discuss fluoropyrimidine workflows in complex tumor models: an assembloid-focused workflow discussion and a tumor–stroma modeling discussion. Their practical value is complementary to the reference paper: they emphasize compound handling, assay planning, and tumor–stroma applications, whereas Shapira-Netanelov et al. provide the primary evidence for constructing matched gastric cancer assembloids and demonstrating stromal effects on gene expression and drug response.

The distinction matters when interpreting translational claims. The reference study validates the model concept and its biological consequences, but it does not by itself validate a particular chemotherapy in every patient-derived assembloid. Conversely, a compound-centered workflow should not be presented as evidence that the compound will retain activity after stromal incorporation unless it has been tested in the matched system.

Limitations and Transferability

The model improves biological relevance but also introduces experimental complexity. Patient-derived cultures may differ in the efficiency of organoid formation, the abundance of recoverable stromal populations, and the stability of each compartment during serial passage. Selective media can enrich desired populations, but culture conditions may also reshape cellular states relative to the original tumor. Consequently, biomarker validation and transcriptomic checks are necessary for every donor-derived model rather than being treated as one-time characterization steps.

Drug-response comparisons also require careful normalization. Differences in cell number, proliferation rate, matrix exposure, and relative stromal abundance can influence viability measurements independently of pharmacological activity. The study’s use of matched monocultures, assembloids, multiple cellular compositions, immunofluorescence, RNA sequencing, and viability assays provides a strong framework, but replication across additional tumors and orthogonal functional endpoints would be needed to define predictive performance.

Why this cross-domain matters, maturity, and limitations

Applying this framework beyond gastric cancer, including to colon cancer research, is a reasonable experimental extension because the central principle is tissue-specific matching of tumor and stromal compartments. However, such transfer is a hypothesis rather than a result established by the reference study. Stromal composition, epithelial lineage, matrix dependence, and drug metabolism may differ between gastric and colorectal tumors. Similarly, using an assembloid to investigate tumor-targeted drug delivery requires direct measurements of compound distribution, cellular exposure, and response across compartments; the paper demonstrates microenvironmental modulation, not targeted delivery itself.

The most mature conclusion is therefore methodological: matched assembloids can reveal treatment effects that epithelial monocultures miss. The next stage is comparative validation across patients, stromal ratios, treatment schedules, and clinically relevant response endpoints while preserving clear separation between observed evidence and proposed application.

Research Support Resources

For workflows that evaluate a fluoropyrimidine prodrug in matched organoid–stroma systems, researchers can use Capecitabine (SKU A8647), also known as N4-pentyloxycarbonyl-5'-deoxy-5-fluorocytidine. The APExBIO product information describes enzymatic conversion to 5-fluorouracil and reports apoptosis induction via Fas-dependent pathway in engineered LS174T colon cancer cells. Those observations can inform exploratory preclinical oncology research, including colon cancer research and tumor-targeted drug delivery assays, but dosing, exposure time, and endpoint selection should be optimized for each assembloid and should not be inferred directly from the gastric cancer study.