CAF-Macrophage Crosstalk Marks a Resistant Niche in Gastric Cancer Peritoneal Metastasis
Gastric cancer peritoneal metastasis GCPM remains one of the most difficult settings in gastrointestinal oncology. Even as immune checkpoint blockade has altered treatment expectations across multiple malignancies, activity in this disease has stayed limited, reflecting a tumor microenvironment built to exclude or blunt anti-tumor immunity. In this setting, the biology of resistance is not an abstract concept. It is visible at the level of individual stromal and immune cells, where cellular neighborhoods can either support immune attack or actively suppress it.
A single-cell transcriptomic analysis of tumor samples from a phase II trial of sintilimab combined with chemotherapy in patients with GCPM now points to a specific resistant ecosystem. The work assembled a high-resolution transcriptional atlas and identified a stroma–myeloid niche enriched for THBS2+ matrix cancer-associated fibroblasts and SPP1+ tumor-associated macrophages in resistant tumors. Within that compartment, the immune system appears to be pushed into a non-productive state by complement-mediated signaling rather than by a single isolated pathway.
The finding matters because it moves resistance away from a vague description of “cold” tumors and into a named cellular circuit.
Complement Signaling Links Fibroblasts to Macrophage Polarization
The resistant niche was not simply crowded with fibroblasts and macrophages; it was organized around a defined ligand-receptor axis. THBS2+ matrix cancer-associated fibroblasts secreted complement component C3, which then engaged C3aR, encoded by C3AR1, on macrophages. That interaction built a C3-C3aR signaling loop that helped maintain the suppressive macrophage state within the tumor microenvironment.
In practical immunologic terms, this is the kind of communication that can shift macrophages away from antigen-presenting, T-cell-supportive behavior and toward a program that sustains immune evasion. A macrophage population dominated by SPP1+ cells is not a passive bystander. It can shape extracellular matrix remodeling, dampen T-cell access, and reinforce the stromal architecture that makes checkpoint blockade less effective.
Resistant Tumors Carried a Distinct Stromal-Myeloid Architecture
The single-cell atlas created in this trial provides an important map of what resistant disease looks like at baseline tissue resolution. Instead of a uniform suppressive mass, the tumor microenvironment appeared as a tightly coordinated niche in which fibroblasts and myeloid cells reinforced one another. THBS2-positive matrix fibroblasts likely contributed not only structural support but also biochemical instruction, and the macrophages receiving those signals adopted a phenotype compatible with immune suppression.
This architecture helps explain why checkpoint blockade can fail even when a PD-1 inhibitor is paired with chemotherapy. Cytotoxic treatment may increase antigen release and inflammatory stress, but if the stromal network continues to recruit, educate, and retain suppressive myeloid cells, the T-cell arm of anti-tumor immunity may never gain sustained traction. The problem is not simply a lack of immune stimulation. It is the presence of a local circuit that intercepts and redirects it.
Why This Pathway Could Blunt Sintilimab Activity
Checkpoint blockade depends on functional immune effector cells entering the tumor, recognizing antigen, and remaining active long enough to produce clinical effect. A fibroblast-macrophage-complement axis can interrupt that sequence at multiple points. Fibroblasts set up the niche, complement signaling amplifies the macrophage program, and macrophages then help preserve a microenvironment hostile to T-cell activity.
In GCPM, where peritoneal spread already creates major spatial and biologic barriers to therapy, this kind of suppressive network may be especially consequential. The peritoneal cavity is not a neutral compartment; it can favor dissemination, matrix remodeling, and immune exclusion. A complement-driven stromal-myeloid hub adds another layer of resistance by converting local cellular cross-talk into a self-reinforcing shield around the tumor.
Single-Cell Data Turn a Broad Problem Into a Targetable Circuit
Single-cell transcriptomics is valuable here because bulk profiling would blur the relationship between fibroblasts and macrophages into a generic inflammatory signal. By separating cells individually, investigators could identify the distinct THBS2-positive fibroblast population and the SPP1-positive macrophage population, then connect them through complement signaling. The result is a mechanistic model that is far more precise than the usual description of immunosuppression.
This level of resolution also creates a rational basis for biomarker work. If this axis is validated in larger cohorts, markers tied to THBS2, C3, C3AR1, or SPP1 could help identify patients less likely to benefit from checkpoint blockade alone and more likely to need combination strategies that disrupt the stromal-myeloid niche. Such an approach would not replace immunotherapy; it would address one of the reasons immunotherapy fails to work well in this disease.
A Resistant Microenvironment Still Dominates the Clinical Picture
The most important message is blunt: in GCPM, resistance can be written into the tissue architecture itself. When fibroblasts and macrophages form a complement-mediated immunosuppressive hub, the tumor does more than hide from the immune system. It recruits the very cells that help maintain that hidden state.
For clinicians, this reinforces why responses to immune checkpoint blockade in peritoneal metastatic gastric cancer remain difficult to predict and often disappointing. For translational researchers, it points to a concrete mechanism worth interrupting rather than a diffuse immune defect. The remaining challenge is to determine how durable this circuit is across patients and whether breaking the C3-C3aR link can restore meaningful sensitivity to immunotherapy without simply shifting suppression into another pathway.