Checkpoint immunotherapy has transformed treatment for many cancers, but cholangiocarcinoma (CCA) has proven largely resistant. Despite tumors often expressing PD-L1, anti-PD-1 therapies produce responses in fewer than 10% of unselected CCA patients, suggesting that the CCA immune environment suppresses immune activation through mechanisms beyond the PD-1 checkpoint alone.
The tumor microenvironment of CCA is characterized by abundant immunosuppressive cells including tumor-associated macrophages (TAMs) and myeloid-derived suppressor cells (MDSCs). These cells express PD-L1 themselves and actively prevent cytotoxic T cells from attacking the tumor.
The central finding is that TAMs, not tumor cells, are the primary source of PD-L1 in CCA. Blocking TAMs with a CSF1R inhibitor reduces CCA tumor burden, but this benefit is blunted by compensatory accumulation of granulocytic MDSCs (G-MDSCs). Only dual inhibition of both cell types effectively potentiates anti-PD-1 therapy.
Clinical relevance is immediate: this provides a mechanistic rationale for combination clinical trials targeting TAMs and MDSCs alongside checkpoint inhibitors in CCA patients.
Syngeneic orthotopic mouse model was used throughout the study. The model was created by directly instilling active oncogenes (YAP and myristoylated Akt) into the biliary tree of C57BL/6 mice, generating tumors that recapitulate the histology and gene expression of human CCA. Derived cell lines (SB cells) can be re-implanted orthotopically into immunocompetent mice.
Human tissue analysis confirmed the mouse findings: PD-L1 expression was assessed by co-immunostaining with CD68 (macrophage marker) and CK19 (bile duct epithelial marker) in 33 and 18 resected human CCA specimens respectively.
Genetic and pharmacological tools were combined: Pd-l1 knockout mice revealed the relative contribution of host versus tumor PD-L1; CSF1R inhibitor (PLX5622) depleted TAMs; and anti-Ly6G antibody depleted G-MDSCs, allowing precise dissection of each cell population's role.
Single-cell RNA sequencing of immune cells from CCA tumors identified specific G-MDSC subsets including an ApoE-expressing subset that compensatorily expanded when TAMs were depleted.
PD-L1 on macrophages, not tumor cells was the primary driver of immune suppression: co-immunostaining of human CCA tissue showed PD-L1 co-localizing far more with CD68-positive macrophages than with CK19-positive cancer cells. This finding was replicated in the mouse model.
Host PD-L1 controls CCA progression - implanting PD-L1-positive SB tumor cells into Pd-l1 knockout mice significantly reduced tumor burden compared to wild-type hosts, even though the tumor cells themselves expressed PD-L1. This counterintuitive result confirmed that immune cell PD-L1 (not tumor cell PD-L1) is the dominant driver of immune escape.
PD-L1 transfer from cancer cells to macrophages was discovered: even in Pd-l1 knockout mice, PD-L1-positive TAMs appeared in tumors. Conditioned medium from PD-L1-expressing SB cells increased PD-L1 levels on macrophages from Pd-l1 knockout mice, revealing a direct paracrine or vesicle-mediated transfer mechanism.
CD8+ T cell infiltration was suppressed in wild-type hosts with intact PD-L1 on immune cells but was dramatically enhanced in Pd-l1 knockout mice, confirming that macrophage PD-L1 is the functional barrier to effective anti-tumor immunity in CCA.
TAM depletion with CSF1R inhibitor reduced CCA tumor burden in the mouse model, confirming that TAMs functionally promote tumor growth. However, the benefit was incomplete - tumors still grew, suggesting an alternative immunosuppressive mechanism was taking over.
Compensatory G-MDSC expansion was the explanation: when CSF1R inhibitor depleted TAMs, granulocytic myeloid-derived suppressor cells (G-MDSCs, identified by Ly6G expression) dramatically expanded within the tumor. G-MDSCs suppress anti-tumor immunity by releasing reactive oxygen species and inhibitory enzymes that disable cytotoxic T cells.
ApoE-expressing G-MDSC subset was identified by single-cell RNA sequencing as a specific G-MDSC population that preferentially expanded after TAM depletion. This subset had a particularly potent immunosuppressive gene expression profile.
Dual TAM plus G-MDSC inhibition using CSF1R inhibitor plus anti-Ly6G antibody was required to achieve meaningful tumor reduction and was necessary to sensitize tumors to anti-PD-1 checkpoint blockade - neither single agent achieved this effect.
Triple combination rationale: anti-PD-1 + CSF1R inhibitor + G-MDSC targeting represents the strategy most likely to overcome CCA's immunosuppressive barrier based on this study's findings. Multiple CSF1R inhibitors (pexidartinib, cabiralizumab) and anti-Ly6G-equivalent human antibodies are clinically available for combination testing.
Biomarker-guided patient selection - quantifying TAM and G-MDSC density in patient tumor biopsies before treatment could identify patients most likely to have the TAM-dependent immunosuppression described in this study and thus most likely to benefit from the combination approach.
TOPAZ-1 trial context - the recent approval of durvalumab (anti-PD-L1) plus chemotherapy for biliary tract cancer shows that the immune compartment can be engaged in CCA. This study suggests that removing TAM and G-MDSC barriers could dramatically expand the fraction of responding patients beyond what chemotherapy combinations achieve.
Tumor microenvironment remodeling as a strategy - rather than solely targeting immune checkpoints on T cells, the data support therapies that reshape the tumor microenvironment by eliminating immunosuppressive myeloid cells before activating anti-tumor T cell immunity.
ApoE G-MDSC targeting requires development of specific approaches to eliminate this subpopulation identified by scRNA-seq. Understanding what signals drive ApoE G-MDSC expansion after TAM depletion could reveal a targetable pathway upstream of the compensatory resistance.
Clinical biomarker development - establishing that the TAM/G-MDSC balance in patient biopsies predicts anti-PD-1 response requires prospective biomarker studies in ongoing CCA immunotherapy trials.
Mechanisms of PD-L1 transfer from cancer cells to macrophages deserve further investigation - is this transfer via exosomes, direct cell contact, or soluble factors? The answer could reveal additional therapeutic targets.
Extension to perihilar and distal CCA - this study focused on intrahepatic CCA. Whether the same TAM-dominated immunosuppression operates in perihilar and distal CCA, which have different anatomical and molecular profiles, requires dedicated investigation.