Targeting stroma-mediated T-cell exclusion and functional exhaustion in pancreatic ductal adenocarcinoma through CXCR4 and PD-1 blockade

Abstract

Background:

Pancreatic ductal adenocarcinoma (PDAC) is one of the deadliest malignancies to date and characterized by a unique immunosuppressive and highly desmoplastic tumor microenvironment (TME). These features drive profound T-cell dysfunction and maintain high resistance to current immunotherapy. By recapitulating the complex 3D architecture of human PDAC, we demonstrate the key immunosuppressive mechanisms that drive T-cell dysfunction within the tumor microenvironment.

Method:

3D PDAC spheroids—generated from PANC-1 cells alone or together with primary pancreatic stellate cells (PSC)—were infiltrated with primary human T-cells from healthy donors, allowing controlled analysis of T-cell infiltration, activation, and checkpoint regulation. Furthermore, patient-derived spheroids (PDS) composed of primary tumor cells and cancer-associated fibroblasts were infiltrated with autologous T-cells.

Results:

Infiltrated T-cells exhibited a pronounced exhaustion signature, including strong upregulation of PD-1, LAG-3, and CTLA-4, closely mirroring the phenotype of tumor-infiltrating lymphocytes (TIL) isolated from PDAC patient samples. Incorporation of pancreatic stellate cells (PSC) generated a fibrotic barrier around the tumor cells that markedly restricted T-cell infiltration, modeling the desmoplastic TME characteristic of PDAC. From a mechanistic perspective, stromal CXCL12–CXCR4 enhanced T-cell exclusion: pharmacological CXCR4 blockade with AMD3100 significantly enhanced T-cell infiltration into PSC-containing spheroids. Furthermore, treatment with the anti-PD-1 monoclonal antibody pembrolizumab partially restored the effector cell function of T-cells within the 3D system. These results demonstrate that this minimalistic platform is capable of capturing complex, cytokine- and stroma-driven immunomodulation typically observed only in advanced organoid or in vivo systems. Crucially, key immunological features—including T-cell exhaustion, stromal exclusion, and therapeutic responsiveness—were fully reproduced in PDS. PDS reproduced patient-specific T-cell suppression patterns and therapeutic responses, underscoring the translational relevance of the platform.

Conclusion:

Together, our findings identify critical determinants of T-cell dysfunction in PDAC and introduce a versatile, animal-free 3D model that powerfully captures hallmark immune-evasion mechanisms in PDAC. This system provides a scalable and mechanistically faithful tool for dissecting TME-driven immune suppression and for accelerating the functional evaluation of immunotherapeutic strategies, including patient-tailored approaches.

1 Introduction

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal malignancies, with a five-year survival rate of only about 13% (1, 2). Late diagnosis, a high metastatic potential, and a heterogeneous tumor microenvironment (TME) are key contributors to this poor prognosis (1, 3). The PDAC TME consists of various cell types, including immune cells, endothelial cells, neuronal cells and cancer-associated fibroblasts (CAF) (4–6). CAF promote the tumor progression through the release of immunosuppressive cytokines and chemokines like TGF-β, CXCL10, and CXCL12, and produce large amounts of extracellular matrix (ECM), resulting in a dense and desmoplastic stroma (4, 7).

In addition to CAF, the PDAC TME contains diverse immune cell populations, however, these are predominantly immunosuppressive, including tumor associated macrophages (TAM), myeloid-derived suppressor cells (MDSC), and regulatory T-cells (Treg). TAM are mainly polarized towards the immunosuppressive M2 phenotype contributing to tumor progression by enhancing immunosuppression and chemoresistance (8, 9). MDSC impair T-cell function directly by upregulating programmed death-ligand 1 (PD-L1) to inhibit T-cell activation, and indirectly via an IL-10 dependent release of TGF-β, driving the expansion of Treg (10, 11). Treg (FoxP3⁺, CD4⁺, CD25⁺) secrete the anti-inflammatory cytokines IL-10 and TGF-β and express the checkpoint molecule cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), thereby inhibiting effector T-cell function and polarization of TAM towards M2 (12). Treg inversely correlate with CD8⁺ T-cell infiltration and are associated with poor clinical outcomes (13). Treg activate a tolerogenic phenotype in dendritic cells, characterized by downregulation of the costimulatory molecules CD86 and CD40, MHC class II, and IFN-y, leading to suppressed activation of CD8⁺ T-cells (10). Accordingly, elevated Treg frequencies are considered a negative prognostic biomarker in PDAC (12, 13). Besides Treg, CD4⁺ T helper (Th) cells are largely polarized toward a Th2 phenotype, promoting immunosuppression through the release of anti-inflammatory cytokines such as IL-4 and IL-13, which polarize anti-inflammatory M2 macrophages and support tumor cell metabolism through increased glycolysis via JAK-STAT signaling (14, 15). In addition, CD4⁺ Th17 cells have also been described in the TME of PDAC patients, where they are associated with a poorer prognosis by promoting the formation of pancreatic intraepithelial neoplasia (PanIN) and enhancing the stem cell properties of tumor cells (10, 16, 17). Th17 secrete IL-17, which increases fibrosis, neovascularization and myeloid cell recruitment, enhancing survival and growth of tumor cells through the activation of the gp130-JAK2-STAT3 pathway (17–19). Cytotoxic lymphocytes (CD8⁺ T-cells, (CTL)) are crucial for the elimination of tumor cells, mediating killing through the secretion of granzyme B, perforin, IFN-γ, and TNF-α, or of mediating FasL-induced apoptosis (12). In PDAC, however, effector T-cells are largely excluded from the TME and confined to the periphery within the stroma-rich areas surrounding the tumor (12). This exclusion is driven by the CAF-derived chemokine CXCL12, which acts as a specific ligand for the G protein-coupled receptor CXCR4 in T-cells, promoting their chemotactic migration into stroma-rich areas (20, 21). Furthermore, T-cells in the PDAC TME exhibit an exhausted phenotype characterized by increased expression of checkpoint molecules, reduced proliferative capacity, and diminished effector function (12, 22, 23). Key checkpoint molecules elevated in these T-cells include programmed cell death protein-1 (PD-1), lymphocyte-activation gene 3 (LAG-3), and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) (24, 25). PD-1 engages its ligands PD-L1 and PD-L2, expressed on tumor cells and other TME-resident T-cells, attenuating T-cell activation and promoting T-cell apoptosis (26). CTLA-4 competes with CD28 for binding to B7 ligands on antigen-presenting cells, thereby preventing co-stimulatory signaling required for full T-cell activation and driving an immunosuppressive state (27, 28). LAG-3 interacts with MHC class II molecules, interfering with effective T-cell receptor (TCR) signaling and dampening T-cell activation (29). Collectively, these mechanisms reduce effector function of CD4⁺ T-cells and impair the cytotoxic activity of CD8⁺ T-cells against tumor cells (28, 29).

Advances in immunotherapy have focused on reactivating tumor-infiltrating T-cells, by blocking checkpoint molecules with monoclonal antibodies (11, 30). The first FDA approved immune checkpoint inhibitor (ICI) was the CTLA-4 blocker ipilimumab for metastatic melanoma in 2011 (31). Since then, numerous ICI targeting CTLA-4, PD-1, or PD-L1, including pembrolizumab, nivolumab, durvalumab, atezolizumab, and avelumab, have been approved for various malignancies, such as non-small cell lung cancer, urothelial carcinoma, Hodgkin lymphoma, and hepatocellular carcinoma (32–37). In PDAC, several ICI have been or are currently under clinical investigated, including the PD-1 inhibitors nivolumab or pembrolizumab, the PD-L1 inhibitor durvalumab, and the CTLA-4 inhibitor ipilimumab (38–41). However, as monotherapies, these antibodies have failed to demonstrate an increase in overall survival (OS) and progression free survival (PFS) compared to chemotherapy, and led to grade 3 treatment-related adverse events in up to 55% of the patients (26, 42–44). To overcome this limited efficacy, multiple clinical trials have evaluated ICI in combination with other therapeutics. These include combination with standard chemotherapy regimens (FOLFIRINOX or gemcitabine plus nab-paclitaxel) alongside pembrolizumab, nivolumab, ipilimumab, or durvalumab (45–47), as well as combinations with PARP-inhibitors (48, 49), or TME-modulating agents such as the CXCR4 antagonist motixafortide (BL-8040), an indoleamine 2,3-dioxygenase 1 [IDO1] inhibitor (50, 51), and a focal adhesion kinase (FAK) inhibitor (52). To date, no broadly effective immunotherapeutic strategy is available for the vast majority of PDAC patients with microsatellite-stable tumors, although pembrolizumab is FDA-approved and effective for the rare (1.2%) microsatellite instability-high (MSI-H) and deficient mismatch repair (dMMR) patient subset (26, 53, 54).

To elucidate mechanisms underlying T-cell dysfunction in the PDAC TME, we examine the composition, spatial localization and functional state of tumor-infiltrating lymphocytes in PDAC patients. Clinical analyses have frequently reported an altered CD4⁺/CD8⁺ T-cell ratio in both the tumor microenvironment and in peripheral blood from PDAC patients (12, 55). The TME is often characterized by a relative enrichment of CD4⁺ T-cell subsets, including Th2-polarized and regulatory Treg, alongside a reduced cytotoxic CD8⁺ T-cell compartment (12). In the peripheral blood, an increased expansion of regulatory and memory CD4⁺ T-cell populations has been described (55). The overall increased CD4/CD8 ration is less consistent compared to the TME and the shift is associated more with an increase of specific CD4⁺ subpopulations (e.g. Treg) rather than a reduction of CD8 T-cells (55). This suggest a systemic shift towards a CD4 dominant immunity, however it remains unclear how the peripheral immune alterations reflect the intratumoral T-cell composition. In addition, tumor-infiltrating T-cells commonly exhibit elevated expression of multiple inhibitory receptors, including PD-1, CTLA-4, and LAG-3, indicative of an exhausted phenotype (24). We therefore investigate the effect of infiltration into PDAC spheroids on T-cell checkpoint expression. Beyond functional exhaustion, spatial immune exclusion represents another hallmark of the PDAC microenvironment. Cytotoxic T-cells are frequently restricted to stromal regions or the tumor periphery, while the tumor core remains poorly infiltrated (20). This phenomenon has been attributed in part to the dense desmoplastic stroma and the activity of CAF, which can promote T-cell exclusion via the CXCL12–CXCR4 axis (21). By mimicking the spatial composition of the TME we analyzed the role of CAF on the infiltration of T-cells.

While these immunologic alterations have been well documented, they have been predominantly studied in isolation in patient tumor samples, mouse models and in vitro models (56–58). Conventional 2D models and classic patient-derived organoids lack the spatial organization and cellular composition of the TME and are therefore of limited utility for investigating of T-cell infiltration and immunotherapeutic responses. As a result, the interplay between checkpoint molecule upregulation, the accumulation of Treg, and the stroma-mediated exclusion of T-cells, and how these collectively contribute to immunotherapy resistance in PDAC remains poorly understood. Therefore, we aim to recapitulate these key immunosuppressive mechanisms simultaneously to drive T-cell dysfunction within the TME using the complex 3D architecture of human PDAC. This model was specifically designed to recapitulate the key mechanisms of immune evasion in PDAC, thereby providing a more physiologically relevant and predictive platform for the assessment of immunotherapeutic treatments and patient-specific responses than current in vitro models.

2 Material and methods

2.1 Cell culture

The PANC-1 cell line was obtained from the American Type Culture Collection (ATCC) and maintained in Dulbecco’s Modified Eagle Medium (DMEM; #41965039, Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS; #10270-106, Gibco) and 1% penicillin–streptomycin (Pen-Strep; #15140122, Gibco). Pancreatic stellate cells (PSC) were kindly provided by Dr. Erkan (Koç University Hospital, Turkey). Ethical approval for the use of PSC was obtained from the Biomedical Sciences Ethics Committee of Koç University, and written informed consent was received from all participating patients. PSC were isolated from individuals diagnosed with pancreatic ductal adenocarcinoma and cultured in DMEM/F12 medium (#11320074, Gibco) containing 20% FBS and 1% Pen-Strep under sterile conditions. All cell cultures were maintained at 37 °C and 5% CO₂.

2.2 Isolation of T-cells

Peripheral blood mononuclear cells (PBMC) were isolated from buffy coats obtained from anonymized healthy donors (Düsseldorf University Hospital, Germany) who had provided material approved for in vitro research. Autologous T-cells from PDAC patients were isolated in the same manner from patient’s PBMC. The isolation was performed using a Ficoll-Paque density gradient centrifugation method. Briefly, buffy coats were diluted with D-PBS and carefully layered over Histopaque (#10771, Sigma-Aldrich, St. Louis, MO, USA). Following centrifugation at 850 × g for 20 min with minimal acceleration and deceleration, the PBMC layer was collected. To remove residual erythrocytes, cells were incubated for 10 min in ammonium chloride lysis buffer, then washed with Dulbecco’s Phosphate Buffered Saline (D-PBS), and the cell count was determined.

T-cells were subsequently isolated from PBMC using Magnetic-Activated Cell Sorting (MACS) with biotinylated CD4 (#130-113-224, Miltenyi Biotec, Bergisch Gladbach, Germany) and CD8 (#130-110-676, Miltenyi Biotec) antibodies, followed by anti-biotin microbeads (#130-090-485, Miltenyi Biotec) according to the manufacturer’s instructions, employing the MidiMACS separator (#130-042-302, Miltenyi Biotec). The purified T-cells were maintained in RPMI medium (#21875-034, Gibco) supplemented with 10% FBS, 1% penicillin–streptomycin, and 50 ng/mL IL-2 (#11340025, ImmunoTools, Friesoythe, Germany). For T-cell activation and expansion, CD3/CD28 Dynabeads (#11161D, Thermo Fisher Scientific, Waltham, MA, USA) were added at a 1:2 bead-to-cell ratio.

2.3 Isolation and cultivation of patient-derived tumor cells, T-cells and CAF

PDAC tissue and full blood were obtained during resection at Pius-Hospital Oldenburg, Germany. The use of human tissue samples was approved by the ethics committee at Carl von Ossietzky University Oldenburg and confirmed by Heinrich Heine University Düsseldorf (study number 2022-058). All patients provided written informed consent prior to tissue collection. Clinicopathological characteristics of the included PDAC patients are summarized in Supplementary Table 1. All patients underwent surgical resection without prior neoadjuvant treatment. For tissue dissociation, the tumor specimen was washed in DPBS supplemented with 2% (v/v) FBS, and carefully trimmed to remove residual adipose tissue. The remaining tumor tissue was finely minced into fragments smaller than 1 cm³ and enzymatically digested in 15 mL of digestion solution, containing 0.6 mg/mL Collagenase IV (#17104–019 Gibco), 0.1 mg/mL DNAse (#10104159001 Roche, Basel, Switzerland) and 2% FBS in RPMI (#21875–034 Gibco) at 37 °C with constant agitation for 45–60 min. The resulting suspension was filtered through a 70 µm cell strainer, and remaining tissue fragments were mechanically dissociated using a syringe plunger. The filter was rinsed with DPBS containing 2% FBS, and the cell suspension was centrifuged at 550 g for 7 min. The cell pellet was resuspended in DMEMF12 containing 10% FBS for tumor cell cultivation and 20% FBS for CAF cultivation. Pure cell lines were obtained via differential trypsinization (59). Characterization of patient-derived cells was performed via immunofluorescence staining. Cancer cell were stained for vimentin (#sc-6260, Santa Cruz Biotechnology, Dallas, TX, USA), pan-cytokeratin (#130-133-439, Miltenyi Biotec), cytokeratin 7 (#130-115-446, Miltenyi Biotec), E-cadherin (#130-111-992, Miltenyi Biotec), CD44 (#130-113-900, Miltenyi Biotec), Epcam (#14-9326-82, Thermo Fisher), Ki67 (#MAB7617, R&D Systems, Minnneapolis, MN, USA), and N-cadherin (#ab98952, Abcam, Cambridge, England). Patient-derived CAF were immunofluorescence labeled for αSMA (#19245S, Cell Signaling, Danvers, MA, USA). For flow cytometric analysis T-cells from the tumor tissue were isolated from the digested tumor tissue using MACS with a biotinylated CD45 antibody (#130-110-630, Miltenyi Biotec), followed by anti-biotin microbeads (#130-090-485, Miltenyi Biotec) according to the manufacturer’s instructions. Isolated immune cells were stained for FACS analysis as described below.

2.4 Spheroid formation

For the generation of mono-culture spheroids composed of PANC-1 cells, further referred to as PANC-1 spheroids, 1500 PANC-1 cells in 50 µL DMEM with 2.5% (v/v) Matrigel (#356230, Corning, NY, USA) were added into an ultra-low attachment coated 96-well plate (#650970, Greiner Bio-One, Kremsmünster, Austria) and centrifuged at 1500 rpm for 5 min. Co-culture spheroids were generated according to our previous protocol (60). Patient-derived spheroids (PDS) composed of patient-derived tumor cells and primary CAF accordingly.

2.5 Immunofluorescence staining of tumor slices

Formalin-fixed, paraffin-embedded (FFPE) PDAC tumor sections (1.5 µm) were kindly generated at the Institute of Pathology from the University Hospital Düsseldorf. All samples were anonymized and their use was approved by the ethics committee of the Heinrich Heine University Düsseldorf (study number 2022-2170). Tissue sections were deparaffinized using xylene and a graded ethanol series, followed by antigen retrieval in Tris(hydroxymethyl)aminomethane–ethylenediaminetetraacetic acid (Tris-EDTA) buffer (pH 9) for 15 min at 100 °C. After washing in Tris-buffered saline containing Tween 20 (TBS-T), sections were blocked for 1 h at 20 °C in phosphate-buffered saline (PBS) supplemented with 5% bovine serum albumin (BSA) and 0.01% Tween 20. Primary and secondary antibodies were diluted in blocking buffer and incubated with the sections for 1 h at 20 °C in the dark. Following incubation, slides were washed, mounted using fluorescence mounting medium (#GM30411-2, Agilent Technologies, Santa Clara, CA, USA), and allowed to dry overnight. Image acquisition was performed using the CellVoyager™ CQ1 High-Content Imaging System (#90ZA00673, Yokogawa Electric Corporation, Musashino, Japan).

2.6 Quantitative analysis of tumor slices

FFPE tissue sections of 10 PDAC patients from the Institute of Pathology, University Hospital Düsseldorf, were subjected to immunofluorescence staining for CD4 and CD8, followed by nuclear counterstaining with DAPI. 40 images per patient were captured at 40x magnification, with Z-stacks acquired to ensure accurate detection of cellular structures across all focal planes.

Image analysis was performed using the CellProfiler software (version 4.2.8, Supplementary Figure 1). To improve segmentation accuracy, individual image planes were analyzed instead of maximum-intensity projections, thereby preserving spatial resolution and minimizing signal overlap in densely packed areas. Nuclei were identified based on DAPI staining using a deep learning-based segmentation approach. Specifically, the Cellpose algorithm was applied via a CellProfiler plugin to enable accurate instance segmentation of individual nuclei despite variable staining intensity and high tissue density (61). Following nuclear segmentation, the expression of cell-associated markers was determined by quantifying the CD4 and CD8 fluorescence signals in the vicinity of each nucleus. Cells were classified as CD4⁺ or CD8⁺ based on marker-specific thresholds for fluorescence intensity, which were applied consistently across all samples. The total number of CD4⁺ and CD8⁺ T-cells was counted per image and summarized for each patient. Finally, the percentage distribution of CD4⁺ and CD8⁺ T-cells was calculated individually for each patient based on the total number of counted T-cells.

2.7 Immunofluorescence staining of spheroids

Spheroids were stained according to the MACS Clearing Kit protocol (#130-126-719, Miltenyi Biotec). Briefly, spheroids were fixed in 4% paraformaldehyde (PFA) for 1 h at room temperature (RT). Following fixation, samples were permeabilized for 6 h prior to incubation with 100 µL of primary antibody solution containing CD4 (#ab133616, Abcam), α-SMA (#19245S, Cell Signaling), pan-cytokeratin (#130-133-439, Miltenyi Biotec), CD8 (#130-110-815, Miltenyi Biotec) or CD45 (#130-110-638, Miltenyi Biotec) for 24 h at 4 °C. Subsequently, spheroids were incubated with the secondary antibody solution comprising Alexa Fluor 647 donkey (H+L) anti-mouse (#A-31571, Invitrogen) and Alexa Fluor 488 donkey (H+L) anti-rabbit (#A-21206, Invitrogen), Alexa Flour 405 donkey (H+L) anti mouse (#A-48257, Invitrogen) supplemented with 2 µg/mL DAPI (#D9542, Sigma-Aldrich), for 24 h at 4 °C. Optical clearing was performed by sequential incubation in ethanol solutions of increasing concentration (50%, 70%, and 100%) containing 2% Tween, each for 2 h, with the final 100% step carried out overnight. This was followed by incubation in clearing solution for 6 h. Cleared spheroids were then transferred to an imaging plate (#89626, ibidi GmbH, Gräfelfing, Germany) and imaged using the CQ1 High-Content Spinning Disk System (#90ZA00673, Yokogawa). 3D images were visualized using the 3D viewer plugin (62) in in the Fiji Software (63).

2.8 Quantitative real-time PCR

To assess chemokine and effector molecule expression, quantitative real-time PCR (qRT-PCR) was performed on T-cells, PANC-1 cells, pancreatic stellate cells (PSC), PANC-1 spheroids, co-culture spheroids, as well as patient-derived tumor cells, cancer-associated fibroblasts (CAF), and corresponding spheroids. The following primers were utilized to determine the mRNA levels of CXCL12, CXCR4, IFN-γ, Granzyme B, Perforin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (forward, 5’-TGCACCACCAACTGCTTAGC-3’; reverse, 5’- GGCATGGACTGTGGTCATGAG-3’), CXCL12 (forward 5´-CTCAACACTCCAAACTGTGCCC-3´; reverse 5´- CTCCAGGTACTCCTGAATCCAC-3´), CXCR4 (forward 5´- CTCCTCTTTGTCATCACGCTTCC -3´; reverse 5´- GGATGAGGACACTGCTGTAGAG-3´), IFN-γ (forward 5´- GAGTGTGGAGACCATCAAGGAAG -3´; reverse 5´- TGCTTTGCGTTGGAC ATTCAAGTC -3´), Granzyme B (forward 5´- CGACAGTACCATTGAGTTGTGCG -3´; reverse 5´- TTCGTCCATAGGAGACAATGCCC-3´), Perforin (forward 5´- ACT CAC AGG CAG CCA ACT TTG C -3´; reverse 5´- CTCTTGAAGTCAGGGTGCAGCG-3´). 2D cell lines were cultivated for 72 h and spheroids for 96 h in RPMI medium. The qRT-PCR was performed as previously described by us (64). Relative gene expression was calculated using the ΔΔCt method, with normalization to the housekeeping gene GAPDH. Quantitative data analysis was conducted using qPCRsoft 4.1 software (Analytik Jena, Jena, Germany).

2.9 Flow cytometry of T-cells

Checkpoint molecule expression in T-cells was analyzed by flow cytometry. PDAC spheroids and patient-derived spheroids were collected following infiltration with either PBMC-derived T-cells or autologous T-cells. Subsequently, spheroids were enzymatically dissociated into single-cell suspensions using Accutase (A1110501, Gibco). Cells were subsequently stained according to the manufacturers’ protocols using the following fluorochrome-conjugated antibodies: CD45-VioGreen (#130-110-638, Miltenyi Biotec), CD8-FITC (#130-110-815, Miltenyi Biotec), CD4-VioBright-V600 (#130-129-579, Miltenyi Biotec), CD25-APC (#130-113-284, Miltenyi Biotec), CD69-VioBlue (#130-112-799, Miltenyi Biotec), CD223-APCVio770 (#130-130-284, Miltenyi Biotec), CD279-PerCP 5.5 (#367410, BioLegend, San Diego, CA, USA), CD3-AF700 (#367410, BioLegend), CD152-PEVio770 (#130-130-455, Miltenyi Biotec). Data acquisition was performed using a Cytek^® Northern Lights flow cytometer (Cytek Biosciences, Fremont, CA, USA), and subsequent analyses were carried out with FlowJo software (version 10.9.0; BD, Ashland, OR, USA).

2.10 Semi-automatic spheroid analysis pipeline

Initially the three-channel raw image data (Supplementary Figure 2A) were split into individual files per channel are denoted as nuclei (nucleus marker DAPI), T-cells (CD45 T-cell marker) and shell (fibroblast marker α-SMA) in the subsequent paragraphs. All three channels were normalized using a quantile-based normalization strategy by stretching the intensity range to the 0.01 and 0.99 quantiles and clipping values to a range of [0,1] (Supplementary Figure 2B). T-cells were segmented using the TWANG algorithm (65) using the parameters σ_min = 3, σ_max = 4 for the Laplacian-of-Gaussian Scale-Space Maximum Intensity Projection (LoGSSMP). To suppress false positive detections in the background, only detections with an average intensity exceeding μ + 2σ were kept, where μ is the mean intensity and σ is the standard deviation of the LoGSSMP intensity. The TWANG segmentation parameters were set to σ_grad = 3, σ_kernel = 2, r^s = 1.3. Exemplary results of the detection are depicted in Supplementary Figure 2C.

The shell interior region was segmented in a semi-automatic fashion. A dedicated graphical user interface was used to annotate selected slices of the shell image channel. In the lateral image orientation (XY), annotations were performed at slice intervals of 25 and polygons were manually drawn along the inner edge of the interface between interior and surrounding shell in slices where the interior was properly visible. To faithfully reconstruct the 3D shape, two additional orthogonal views that were located in the center of the lateral annotations were annotated. Finally, the polygon points of all slice annotations were used to compute the 3D convex hull that served as an approximation of the interior region of the spheroid (Supplementary Figure 2D).

For the final quantifications, T-cells were assigned either to the inside of the spheroid or to the outside region. If a T-cell’s segmentation intersected more than 50% with the semi-automatically created shell images, it was counted to the interior region, otherwise to the outside region and in addition to the absolute counts, the fraction of inside vs. outside T-cells was measured in percent. To further characterize the localization of the T-cells with respect to the spheroid center, the Euclidean distance of each T-cell to the centroid of the shell’s interior region was computed and plotted as separate normalized histograms of the distance distributions of the inside and the outside T-cells (Supplementary Figure 2E).

2.11 Statistical analysis

For statistical and graphical analysis, the GraphPad Prism 8.0.2 software (GraphPad Software, Boston, MA, USA) was used. One-way analysis of variance (ANOVA) and unpaired t-test were used to test the statistical data, with p-values ≤ 0.05 considered statistically significant.

3 Results

To investigate the intratumoral distribution of T-cell subsets, immunofluorescence staining for CD8 and CD4 was performed on PDAC patient tumor slices. As shown in Figures 1A, B, a higher abundance of CD4⁺ T-cells than CD8⁺ T-cells was observed. Image-based analysis of tumor slices from 10 PDAC patients was performed to quantify the CD4⁺ and CD8⁺ frequencies (Supplementary Figure 1). As shown in Figure 1C, the frequency of CD4⁺ T-cells was 4.3-fold higher compared to CD8⁺ T-cells.

Figure 1

To further validate these observations in fresh PDAC samples, flow cytometric analysis was conducted to determine the frequencies of CD4⁺ and CD8⁺ T-cells isolated from either tumor tissue or peripheral blood of five PDAC patients. Consistent with the immunofluorescence data, tumor-derived samples exhibited a significantly elevated frequency of CD4⁺ T-cells, corresponding to a 2.6-fold increase relative to CD8⁺ T-cells. Notably, a similar shift in the CD4/CD8 ratio toward CD4⁺ T-cell predominance was detected in peripheral blood mononuclear cells (PBMC) from the same patients, with a 3.5-fold increase. To exclude age-related effects as a potential confounding factor, PBMC from age-matched healthy donors were analyzed. As shown in Figure 1D, PBMC from the control cohort displayed a 2.6-fold higher frequency of CD8⁺ T-cells compared to PDAC patients, indicating that the altered CD4/CD8 ratio is associated with disease status rather than with age.

Beyond reduction of cytotoxicT-cells, the TME of PDAC actively promotes T-cell dysfunction, which is characterized by the upregulation of immune checkpoint molecules (25). To evaluate the dysfunction of T-cells, checkpoint molecule expression was analyzed by flow cytometry in T-cells isolated from tumor tissue and peripheral blood of PDAC patients. As shown in Figure 2, tumor-infiltrating T-cells exhibited a significantly increased expression of immune checkpoint molecules compared to peripheral blood-derived T-cells from the same patients. In CD4⁺ T-cells (Figure 2A) expression levels of CTLA-4, PD-1, and LAG-3 were elevated by 2.4-fold, 5.7-fold, and 2.3-fold, respectively, relative to CD4⁺ T-cells from PBMC. A comparable expression pattern was observed in CD8⁺ T-cells, with a 2.4-fold increase in CTLA-4, a 2.4-fold increase in PD-1, and a 1.8-fold increase in LAG-3 expression (Figure 2B).

Figure 2

In addition, tumor-infiltrating CD8⁺ T-cells displayed a 2.6-fold higher expression of the e activation marker CD69 compared to their peripheral counterparts (Figure 2B). Collectively, these findings highlight the pronounced dysfunctional phenotype acquired by T-cells within the PDAC tumor microenvironment.

To enable the functional evaluation of novel immunotherapeutic strategies for PDAC, it is essential to recapitulate the immunosuppressive mechanisms operating within the TME in vitro. We therefore established a simplified, human 3D spheroid platform designed mimic T-cell checkpoint upregulation and immune exclusion observed in PDAC patients.

PANC-1 spheroids composed of PANC-1 tumor cells alone or in combination with PSC were generated and infiltrated with CD3⁺ T-cells isolated from healthy donor PBMC for 72 hours. As shown in Figure 3, T-cell infiltration into both spheroid types resulted in a pronounced induction of immune checkpoint expression. In CD4⁺ T-cells (Figure 3A), co-culture spheroids induced a 2.4-fold increase in CTLA-4, a 7.5-fold increase in PD-1, and a 1.4-fold increase in LAG-3 expression, whereas PANC-1 spheroids induced corresponding increases of 3.5-fold (CTLA-4), 6.2-fold (PD-1), and 1.3-fold (LAG-3). In CD8⁺ T-cells (Figure 3B), co-culture spheroids led to a 3.7-fold increase in PD-1, and a 1.4-fold increase in LAG-3 expression, while PANC-1 spheroids induced corresponding increases 3.6-fold (PD-1), and 1.3-fold (LAG-3). No significant upregulation of CTLA-4, or CD69 was observed in CD8⁺ T-cells under these conditions. Collectively, these data demonstrate that despite its simplicity, the co-culture 3D PDAC model robustly reproduces key features of checkpoint upregulation observed in infiltrating T-cells in PDAC tumors (Figure 2). Furthermore, the frequency of FoxP3⁺ CD25⁺ CD4⁺ T-cells increased 3.4-fold following infiltration into PANC-1 spheroids and 3.0-fold in co-culture spheroids, indicating an expansion of immunosuppressive regulatory T-cell populations (Figure 3C).

Figure 3

CAF are considered to be key mediators of T-cell exclusion in the PDAC TME through secretion of the chemokine CXCL12, which binds the chemokine receptor CXCR4 on T-cells sequestering them within the peritumoral stroma (21). To assess whether this stromal-driven immune exclusion could be recapitulated in vitro, T-cell infiltration was compared between PANC-1 monospheroids and PSC-containing co-culture spheroids.

As illustrated in Figure 4, T-cells efficiently and homogeneously infiltrated monospheroids. In contrast, in PSC-containing co-culture spheroids, T-cells predominantly accumulated at the stromal outer layer, failing to penetrate the spheroid core. These findings indicate that stromal fibroblasts establish both a physical and functional barrier, restricting T-cell infiltration.

Figure 4

Consistent with this observation, expression analysis demonstrated significantly elevated CXCL12 levels in PSC and PSC-containing spheroids compared with PANC-1 tumor cells (Supplementary Figure 3). In parallel, CXCR4 expression was enriched on CD3⁺ T-cells, indicating a CXCL12–CXCR4 axis–dependent mechanism underlying T-cell retention at the stromal compartment.

Patient-derived systems provide higher translational relevance by preserving tumor heterogeneity, authentic stromal signaling, and autologous immune interactions that cannot be captured by conventional cell line–based models. Therefore, a patient-derived spheroid (PDS) model was established, composed of patient-derived tumor cells and CAF, and subsequently infiltrated with autologous T-cells. Primary tumor cells were characterized by immunofluorescence staining using established tumor cell markers (59), while CAF identity was confirmed by α-SMA expression (Supplementary Figure 4). As shown in Figure 5, patient-derived tumor cells and CAF self-assemble into co-culture spheroids in a manner comparable to the cell line-based model, resulting in the formation of a distinct fibrotic shell surrounding a tumor cell-rich core. Maximum intensity projections are provided in Supplementary Figure 5. Notably, compared to the cell line–based model, the fibrotic stromal compartment was even more pronounced in the patient-derived system.

Figure 5

Furthermore, T-cells were largely restricted from infiltrating the tumor cell core by the fibrotic shell (Figure 6A), thereby recapitulating the T-cell exclusion previously observed in the cell line-based co-culture spheroids These findings are further supported by elevated CXCL12 expression in patient-derived CAF and patient-derived spheroids compared to primary tumor cells (Supplementary Figure 3B).

Figure 6

In addition to immune exclusion, increased expression of immune checkpoint molecules was observed upon T-cell infiltration (Figures 6B, C). In infiltrated CD4⁺ T-cells, PD-1 expression increased 6.0-fold, and LAG-3 expression increased 1.9-fold. The upregulation of CTLA-4 and CD69 did not reach statistical significance in CD4⁺ T-cells. In CD8⁺ T-cells, PD-1 and LAG-3 expression increased 3.5-fold and 2.1-fold, respectively, whereas CTLA-4, and CD69 did not show significant changes. Collectively, these results recapitulate the checkpoint molecule expression signature previously observed in tumor infiltrating T-cells (Figure 2). Furthermore, the frequence of FoxP3⁺ CD25⁺ CD4⁺ T-cells increased 7.3-fold upon infiltration into PDS (Figure 6D), indicating an expansion of the regulatory T-cell populations.

To overcome T-cell exclusion, several preclinical studies have applied CXCR4 antagonists, such as AMD3100, to disrupt the CXCL12–CXCR4 axis, thereby enhancing T-cell infiltration into the TME (21, 51, 66). Moreover, Bockorny et al. demonstrated in a clinical trial that pharmacological CXCR4 inhibition increased intratumoral T-cell infiltration in PDAC patients (51).

To assess whether this phenomenon could be recapitulated in our models, co-culture spheroids and PDS were treated with AMD3100 for 24 h immediately following T-cell addition. As shown in Figure 7A, AMD3100 treatment resulted in increased T-cell infiltration into the co-culture spheroids. Notably, T-cells displayed a more homogeneous spatial distribution throughout the spheroid rather than being restricted to the fibrotic shell. A comparable effect was observed in patient-derived spheroids in the presence of autologous T-cells (Figure 7B). Quantitative image analysis was performed to determine the frequency of T-cells within the tumor core (Supplementary Figure 2). The analysis further confirmed a significant enhancement of T-cell infiltration upon CXCR4 blockade in both co-culture spheroids and PDS compared to untreated controls (Figures 7D, E). In the co-culture spheroids, the percentage of T-cells in the tumor cell core increased from 3.0% to 16.2% (± 2.65%) following treatment with AMD3100 (Figure 7D). In the PDS, the proportion of T-cells infiltrating the tumor core increased from 2.3% to 6.7% (± 1.75%) (Figure 7E).

Figure 7

Collectively, these findings indicate that inhibition of the CXCR4–CXCL12 axis effectively restores T-cell infiltration into the tumor cell region, further supporting the role of CAF-driven CXCL12 signaling as a clinically relevant mechanism of immune exclusion in PDAC.

Counteracting T-cell dysfunction and restoring effector function through immune checkpoint blockade targeting PD-1 has been widely employed to disrupt PD-1 engagement with its ligands (26). However, in PDAC, clinical application of ICI, including the anti–PD-1 monoclonal antibody pembrolizumab, has demonstrated limited therapeutic benefit, with the exception of the small group of MSI-H/dMMR patients (26, 39, 48, 51, 54).

To evaluate the functional impact of PD-1 blockade in our 3D PDAC models, we quantified IFN-γ, Granzyme B and perforin mRNA expression in infiltrating T-cells following 72 hours of pembrolizumab treatment. As shown in Figure 8A, treatment with 1 µg/mL and 10 µg/mL pembrolizumab resulted in a 1.6-fold and 2.0-fold increase in IFN-γ mRNA expression, respectively, in T-cells infiltrating PANC-1 spheroids. In contrast, T-cells infiltrating co-culture spheroids containing PSC did not exhibit a significant increase in IFN-γ expression upon PD-1 blockade. In PDS pembrolizumab induced a 5.0-fold and 3.4-fold increase in IFN-γ expression at 1 µg/mL and 10 µg/mL, respectively. Furthermore, treatment with pembrolizumab enhanced the expression of cytotoxic effector molecules in PDS. Granzyme B expression (Figure 8B) increased 2.6-fold following treatment with 1 µg/mL and 2.5-fold with 10 µg/mL pembrolizumab, while perforin expression (Figure 8C) was elevated 2.8-fold and 4.7-fold at the corresponding concentrations. In contrast, granzyme B and perforin expression were not significantly altered in T-cells infiltrating PANC-1 or co-culture spheroids upon treatment. Moreover, flow cytometric analysis revealed increased protein expression of the effector molecules IFN-γ and granzyme B in stimulated CD8⁺ T-cells following co-culture with PDS (Supplementary Figure 6). The frequency of IFN-γ⁺ CD8⁺ T-cells increased from 12.5% to 17.2%, while granzyme B expression increased from 65.0% to 70.6% (Supplementary Figures 6A, B). Perforin expression showed a tendency towards increased expression; however, this difference did not reach statistical significance (Supplementary Figure 6C). To determine whether these effects were attributable to altered PD-1 surface expression, flow cytometric analysis were performed following 72 hours of treatment. As illustrated in Supplementary Figure 7 pembrolizumab did not affect PD-1 surface levels in either CD4⁺ or CD8⁺ T-cells. These findings indicate that the observed increase in effector molecule expression results from functional PD-1 blockade rather than modulation of PD-1 receptor expression.

Figure 8

4 Discussion

T-cell dysfunction within the TME of PDAC is multifactorial and emerges from the interplay of multiple immunosuppressive mechanisms. Central processes are the accumulation of immunosuppressive Treg, a shift towards a CD4⁺-dominant immune profile, the stromal-mediated exclusion of effector T-cells from the tumor core, and a pronounced upregulation of multiple immune checkpoint molecules, including LAG-3, CTLA-4, and PD-1 (3, 12). Together, these mechanisms form a spatially and functionally distinct immune microenvironment that limits effective T-cell infiltration and promotes a functionally exhausted phenotype (12). In line with these reports, we observed a shift in the CD4/CD8 ratio toward CD4⁺ T-cells within the TIL, supporting the hypothesis of preferential accumulation of CD4⁺ subsets in the TME of PDAC. Furthermore, consistent with previous studies tumor-infiltrating T-cells exhibited a significant upregulation of the immune checkpoint molecules CTLA-4, LAG-3, and PD-1 (26, 67).

In line with the findings of Rodríguez et al. (55), we also detected an increased frequency of CD4⁺ T-cells in the peripheral blood of PDAC patients compared to age-matched healthy donors. Our data, derived from matched tumor tissue and peripheral blood samples from the same patients, provide evidence for a coordinated shift toward a CD4⁺-dominant immune profile at both systemic and intratumoral levels, suggesting a bidirectional interaction between tumor-derived immunomodulatory signals and systemic immune cell composition. To the best of our knowledge, this is the first study to directly demonstrate such a parallel alteration in the TME and peripheral blood within individual PDAC patients. However, the mechanistic relationship between these compartments remains to be elucidated further.

By recapitulating the basic spatial and stromal architecture of the PDAC TME in our PANC-1 spheroid models, we were able to reproduce key mechanisms underlying T-cell dysfunction observed in patients, including checkpoint upregulation, increased Treg frequencies, and stromal-mediated exclusion of T-cells. However, as PANC-1 spheroids lack an autologous T-cell response, we infiltrated patient-derived spheroids (PDS) with autologous T-cells. In contrast to previously described patient-derived organoids, our approach includes CAF-mediated stromal compartments, recapitulating T-cell–stromal interactions including the named hallmarks of T-cell dysfunction in PDAC (56, 68, 69). Here, we demonstrate that infiltration of autologous T-cells into PDS results in a pronounced upregulation of immune checkpoint molecules, increased accumulation of Treg, and spatial restriction of T-cells to the stromal compartment. Importantly, our study highlights a progressive increase in physiological relevance across the applied in vitro models. This difference likely reflects the increasing restoration of stromal complexity across the models, whereby CAF-derived signaling and spatial organization critically shape T-cell access and functional activation (70). In contrast, PDS more closely resemble the complexity of the human PDAC tumor microenvironment by preserving patient-specific stromal architecture, tumor heterogeneity, and autologous immune interactions. Consequently, functional T-cell responses were most pronounced and most clinically aligned in the PDS model, suggesting that stromal complexity and cellular origin critically determine the extent of immunological modulation. This further emphasize the relevance of autologous, patient-derived systems for evaluating patient-specific responses to cell-based therapeutic approaches for PDAC.

It has been demonstrated that T-cell exclusion from the tumor core in PDAC is mediated by CAF via CXCL12–CXCR4 signaling (21). To overcome this mechanism, the FDA-approved CXCR4 antagonist AMD3100 has been employed in several studies, which collectively demonstrated increased T-cell infiltration in vitro, in vivo, and in clinical trials (21, 51, 66, 71). These findings underscore the translational relevance of the CXCL12–CXCR4 axis as a key driver of stromal-mediated T-cell exclusion in PDAC. In our model, disruption of the CXCL12–CXCR4 axis with AMD3100 reduced chemokine-driven T-cell retention and thereby significantly increased T-cell infiltration into the tumor core region in both co-culture spheroids and PDS. In PDS, this effect was less pronounced compared to that in PANC-1 co-culture spheroids. The reduced proportion of T-cells within the tumor core in PDS likely reflects a migration-restrictive effect imposed by the denser fibrotic stromal compartment, which limits T-cell penetration despite chemokine axis blockade. However, this structural organization more accurately reflects the spatial architecture of the PDAC TME in vivo, in which the fibrotic component can comprise up to 80% of the TME (6).

ICI represent an effective strategy to counteract T-cell exhaustion and restore effector function (11). In PDAC, several ICI have been evaluated in both monotherapy and combination regimens (11, 26, 53). However, their clinical efficacy has remained very limited, with the exception of the rare (approximately 1.2%) MSI-H/dMMR patient subgroup, in which pembrolizumab has been approved by the FDA (54). Here, we demonstrate an increased expression of the effector molecules IFN-γ, Granzyme B, and perforin in autologous T-cells infiltrating PDS following pembrolizumab treatment. As reported by James et al., this effect is patient-dependent and warrants further investigation in additional PDAC patients using established PDS models (72). In contrast, no significant changes in Granzyme B or perforin expression were observed in T-cells infiltrating cell line–based spheroids. For IFN-γ, only T-cells infiltrating PANC-1 spheroids exhibited an increase upon treatment with pembrolizumab. These findings are consistent with those of Daunke et al., who, using cell line–based PDAC spheroids, similarly observed no significant effect of pembrolizumab on effector molecule expression (73). Taken together, these results support the hypothesis that blockade of the PD-1/PD-L1 axis alone is insufficient to overcome the complex immunosuppressive TME in PDAC (53). In this context, our study provides a basis for further investigations into the efficacy of ICI in physiologically relevant preclinical PDAC models. Novel checkpoint targets such as LAG-3, T-cell immunoglobulin and mucin domain-containing protein 3 (TIM-3), and T-cell immunoreceptor with Ig and ITIM domains (TIGIT) are currently under investigation (11), and may enhance therapeutic efficacy when used as combinatorial co-inhibitory targets alongside PD-1/PD-L1-directed therapies.

Overall, our findings indicate that immune evasion in PDAC is driven by both T-cell exclusion and checkpoint-associated T-cell dysfunction. Importantly, the CXCL12–CXCR4 axis and PD-1 signaling appear to contribute to distinct yet complementary mechanisms of immunosuppression. While CXCR4-mediated signaling restricts T-cell infiltration into the tumor microenvironment, PD-1 signaling impairs T-cell effector function. Consequently, a combined intervention targeting the PD-1/PD-L1 signaling pathway and the CXCR4–CXCL12 axis may represent a strategy to overcome key mechanisms of immune evasion in PDAC by simultaneously enhancing T-cell infiltration and restoring cytotoxic activity. To support this concept, Bockorny et al. clinically evaluated the combination of the CXCR4 antagonist motixafortide with pembrolizumab and reported increased intratumoral T-cell infiltration, as well as an overall response rate of 32% and a median duration of response of 7.8 months (51). Overall, our 3D co-culture model provides a suitable platform for combinatorial drug testing and may facilitate further investigation of the synergistic potential of PD-1 blockade and CXCR4 inhibition in PDAC.

We acknowledge that this study has several limitations, which highlight important directions for future research. First, the inclusion of a relatively small patient cohort (n = 5) may limit the robustness and generalizability of our findings. Further studies should include larger patient cohorts to further validate the patient-derived findings of this study including the expected heterogeneity of patient-specific responses. Nevertheless, our results are consistent with those reported in larger patient cohorts (23, 24). Second, we present initial proof-of-concept data for PDS as a novel experimental platform and for the CXCR4-CXCL21 axis as one putative therapeutic approach for ICI co-treatment. Further studies incorporating a larger number of patient-specific models will be required to evaluate patient-specific response differences. Furthermore, a more comprehensive investigation of the signaling pathways underlying the distinct mechanisms of T-cell dysfunction including transcriptomic analysis may enable the identification of novel therapeutic targets beyond the CXCR4-CXCL12 axis to overcome immune suppression in PDAC. To this end, the incorporation of additional cellular components of the TME including other immune cell types or neuronal structures will be essential (74).

Overall, our results support a multifaceted and spatially organized model of T-cell dysfunction in PDAC, in which immune evasion arises from the coordinated interplay of multiple mechanisms rather than isolated signaling pathways. Immune dysfunction is not driven by a single dominant process but instead emerges from the interaction of stromal-mediated T-cell exclusion via the CXCL12–CXCR4 axis, immune checkpoint upregulation leading to functional exhaustion, and the accumulation of regulatory T-cell populations that further reinforce immunosuppression (6, 75). Within this framework, CAF act as central orchestrators by shaping both the physical architecture of the TME and its immunological landscape (5, 6). This spatial and functional compartmentalization not only restricts T-cell infiltration but also limits effective effector function, thereby contributing to the pronounced resistance of PDAC to current immunotherapeutic strategies (75).

In summary, we identified key determinants of T-cell dysfunction in PDAC, including immune checkpoint upregulation, stromal exclusion, and the accumulation of regulatory T-cells. Despite its simplicity, our 3D models are able to recapitulate these hallmark mechanisms of immune evasion in PDAC, providing a scalable and robust tool for studying TME-driven immunosuppression and functional characterization of immunotherapeutic approaches. However, additional experimental patient-derived models recapitulating both the complex spatial organization and the full functional and multicelluar complexity of the TME are essential to dissect these interconnected mechanisms and to evaluate future therapeutic strategies for PDAC.

References

• 1

  MackSKoesslerTBichardPFrossardJ. Pancreatic cancer: epidemiology, risk factors, and prevention. Onco. (2025) 5:37. doi: 10.3390/onco5030037

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2

  SiegelRLKratzerTBGiaquintoANSungHJemalA. Cancer statistics, 2025. CA Cancer J Clin. (2025) 75:10–45. doi: 10.3322/caac.21871

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3

  HartupeeCNagaloBMChabuCYTesfayMZColeman-BarnettJWestJTet al. Pancreatic cancer tumor microenvironment is a major therapeutic barrier and target. Front Immunol. (2024) 15:1287459. doi: 10.3389/fimmu.2024.1287459

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4

  WuYZhangCJiangKWernerJBazhinAVD'HaeseJGet al. The role of stellate cells in pancreatic ductal adenocarcinoma: targeting perspectives. Front Oncol. (2020) 10:621937. doi: 10.3389/fonc.2020.621937

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5

  ThomasDRadhakrishnanP. Pancreatic stellate cells: the key orchestrator of the pancreatic tumor microenvironment. Adv Exp Med Biol. (2020) 1234:57–70. doi: 10.1007/978-3-030-37184-5_5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  BräutigamKSkokKSzymonskiKRiftCVKaramitopoulouE. Tumor immune microenvironment in pancreatic ductal adenocarcinoma revisited - exploring the "Space. Cancer Lett. (2025) 622:217699. doi: 10.1016/j.canlet.2025.217699

  • CrossRef
  • Google Scholar

• 7

  NortonJFosterDChintaMTitanALongakerM. Pancreatic cancer associated fibroblasts (CAF): under-explored target for pancreatic cancer treatment. Cancers (Basel). (2020) 12(5):1347. doi: 10.3390/cancers12051347

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 8

  WeizmanNKrelinYShabtay-OrbachAAmitMBinenbaumYWongRJet al. Macrophages mediate gemcitabine resistance of pancreatic adenocarcinoma by upregulating cytidine deaminase. Oncogene. (2014) 33:3812–9. doi: 10.1038/onc.2013.357

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  YangSLiuQLiaoQ. Tumor-associated macrophages in pancreatic ductal adenocarcinoma: origin, polarization, function, and reprogramming. Front Cell Dev Biol. (2020) 8:607209. doi: 10.3389/fcell.2020.607209

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10

  HuberMBrehmCUGressTMBuchholzMAlashkar AlhamweBvon StrandmannEPet al. The immune microenvironment in pancreatic cancer. Int J Mol Sci. (2020) 21(19):7307. doi: 10.3390/ijms21197307

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11

  ThankachanSVJayaramanVDattaLApthiSZamanBFGurunathanRet al. Tumor microenvironment: recent advances in immunotherapies of pancreatic cancer. Med (Kaunas). (2025) 61:1776. doi: 10.3390/medicina61101776

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12

  GoulartMRStasinosKFinchamREADelvecchioFRKocherHM. T cells in pancreatic cancer stroma. World J Gastroenterol. (2021) 27:7956–68. doi: 10.3748/wjg.v27.i46.7956

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  ProttiMPDe MonteL. Immune infiltrates as predictive markers of survival in pancreatic cancer patients. Front Physiol. (2013) 4:210. doi: 10.3389/fphys.2013.00210

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14

  KerkSAPapagiannakopoulosTShahYMLyssiotisCA. Metabolic networks in mutant KRAS-driven tumours: tissue specificities and the microenvironment. Nat Rev Cancer. (2021) 21:510–25. doi: 10.1038/s41568-021-00375-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15

  DeyPLiJZhangJChaurasiyaSStromAWangHet al. Oncogenic KRAS-driven metabolic reprogramming in pancreatic cancer cells utilizes cytokines from the tumor microenvironment. Cancer Discov. (2020) 10:608–25. doi: 10.1158/2159-8290.cd-19-0297

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16

  WangSLiZHuG. Prognostic role of intratumoral IL-17A expression by immunohistochemistry in solid tumors: a meta-analysis. Oncotarget. (2017) 8:66382–91. doi: 10.18632/oncotarget.18807

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  HegdeSKrisnawanVEHerzogBHZuoCBredenMAKnolhoffBLet al. Dendritic cell paucity leads to dysfunctional immune surveillance in pancreatic cancer. Cancer Cell. (2020) 37:289–307.e9. doi: 10.1016/j.ccell.2020.02.008

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18

  LiuWWangXWuW. Role and functional mechanisms of IL-17/IL-17R signaling in pancreatic cancer (review). Oncol Rep. (2024) 52:144. doi: 10.3892/or.2024.8803

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19

  McAllisterFBaileyJMAlsinaJNirschlCJSharmaRFanHet al. Oncogenic Kras activates a hematopoietic-to-epithelial IL-17 signaling axis in preinvasive pancreatic neoplasia. Cancer Cell. (2014) 25:621–37. doi: 10.1016/j.ccr.2014.03.014

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20

  Ene-ObongAClearAJWattJWangJFatahRRichesJCet al. Activated pancreatic stellate cells sequester CD8+ T cells to reduce their infiltration of the juxtatumoral compartment of pancreatic ductal adenocarcinoma. Gastroenterology. (2013) 145:1121–32. doi: 10.1053/j.gastro.2013.07.025

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21

  GargBGiriBModiSSethiVCastroIUmlandOet al. NFκB in pancreatic stellate cells reduces infiltration of tumors by cytotoxic T cells and killing of cancer cells, via up-regulation of CXCL12. Gastroenterology. (2018) 155:880–891.e8. doi: 10.1053/j.gastro.2018.05.051

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22

  PapalamprosAVailasMNtostoglouKChiloechesMLSakellariouSChouliariNVet al. Unique spatial immune profiling in pancreatic ductal adenocarcinoma with enrichment of exhausted and senescent T cells and diffused CD47-SIRPα expression. Cancers (Basel). (2020) 12:1825. doi: 10.3390/cancers12071825

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23

  WanSZhaoEWeissingerDKrantzBAWerbaGFreemanDet al. Tumor infiltrating T cell states and checkpoint inhibitor expression in hepatic and pancreatic Malignancies. Front Immunol. (2023) 14:1067352. doi: 10.3389/fimmu.2023.1067352

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24

  YousufSQiuMVoith von VoithenbergLHulkkonenJMacinkovicISchulzARet al. Spatially resolved multi-omics single-cell analyses inform mechanisms of immune dysfunction in pancreatic cancer. Gastroenterology. (2023) 165:891–908.e14. doi: 10.1053/j.gastro.2023.05.036

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 25

  SakaDGökalpMPiyadeBCevikNCArik SeverEUnutmazDet al. Mechanisms of T-cell exhaustion in pancreatic cancer. Cancers (Basel). (2020) 12:2274. doi: 10.3390/cancers12082274

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26

  BalsanoRZanusoVPirozziARimassaLBozzarelliS. Pancreatic ductal adenocarcinoma and immune checkpoint inhibitors: the gray curtain of immunotherapy and spikes of lights. Curr Oncol. (2023) 30:3871–85. doi: 10.3390/curroncol30040293

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27

  ChenL. Co-inhibitory molecules of the B7-CD28 family in the control of T-cell immunity. Nat Rev Immunol. (2004) 4:336–47. doi: 10.1038/nri1349

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28

  LiHBYangZHGuoQQ. Immune checkpoint inhibition for pancreatic ductal adenocarcinoma: limitations and prospects: a systematic review. Cell Commun Signal. (2021) 19:117. doi: 10.1186/s12964-021-00789-w

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29

  LongLZhangXChenFPanQPhiphatwatcharaPZengYet al. The promising immune checkpoint LAG-3: from tumor microenvironment to cancer immunotherapy. Genes Cancer. (2018) 9:176–89. doi: 10.18632/genesandcancer.180

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30

  JuYXuDLiaoMMSunYBaoWDYaoFet al. Barriers and opportunities in pancreatic cancer immunotherapy. NPJ Precis Oncol. (2024) 8:199. doi: 10.1038/s41698-024-00681-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31

  HodiFSO'DaySJMcDermottDFWeberRWSosmanJAHaanenJBet al. Improved survival with ipilimumab in patients with metastatic melanoma. N Engl J Med. (2010) 363:711–23. doi: 10.1056/nejmoa1003466

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  WangJYangTXuJ. Therapeutic development of immune checkpoint inhibitors. Adv Exp Med Biol. (2020) 1248:619–49. doi: 10.1007/978-981-15-3266-5_23

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 33

  DangTOOgunniyiABarbeeMSDrilonA. Pembrolizumab for the treatment of PD-L1 positive advanced or metastatic non-small cell lung cancer. Expert Rev Anticancer Ther. (2016) 16:13–20. doi: 10.1586/14737140.2016.1123626

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34

  KimJKwiatkowskiDMcConkeyDJMeeksJJFreemanSSBellmuntJet al. The Cancer Genome Atlas expression subtypes stratify response to checkpoint inhibition in advanced urothelial cancer and identify a subset of patients with high survival probability. Eur Urol. (2019) 75:961–4. doi: 10.1016/j.eururo.2019.02.017

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35

  KasamonYLde ClaroRAWangYShenYLFarrellATPazdurRet al. FDA approval summary: nivolumab for the treatment of relapsed or progressive classical Hodgkin lymphoma. Oncologist. (2017) 22:585–91. doi: 10.1634/theoncologist.2017-0004

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36

  KudoM. Targeted and immune therapies for hepatocellular carcinoma: predictions for 2019 and beyond. World J Gastroenterol. (2019) 25:789–807. doi: 10.3748/wjg.v25.i7.789

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37

  Arafat HossainM. A comprehensive review of immune checkpoint inhibitors for cancer treatment. Int Immunopharmacol. (2024) 143:113365. doi: 10.1016/j.intimp.2024.113365

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38

  TsujikawaTCrocenziTDurhamJNSugarEAWuAAOnnersBet al. Evaluation of cyclophosphamide/GVAX pancreas followed by listeria-mesothelin (CRS-207) with or without nivolumab in patients with pancreatic cancer. Clin Cancer Res. (2020) 26:3578–88. doi: 10.1158/1078-0432.ccr-19-3978

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39

  KatzMHGPetroniGRBauerTReilleyMJWolpinBMStuckyCCet al. Multicenter randomized controlled trial of neoadjuvant chemoradiotherapy alone or in combination with pembrolizumab in patients with resectable or borderline resectable pancreatic adenocarcinoma. J Immunother Cancer. (2023) 11:e007586. doi: 10.1136/jitc-2023-007586

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40

  GrossNEZhangZMitchellJTCharmsazSHernandezAGCoyneEMet al. Phosphodiesterase-5 inhibition collaborates with vaccine-based immunotherapy to reprogram myeloid cells in pancreatic ductal adenocarcinoma. JCI Insight. (2024) 9(18):e179292. doi: 10.1172/jci.insight.179292

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  O'ReillyEMOhDYDhaniNRenoufDJLeeMASunWet al. Durvalumab with or without tremelimumab for patients with metastatic pancreatic ductal adenocarcinoma: a phase 2 randomized clinical trial. JAMA Oncol. (2019) 5:1431–8. doi: 10.1001/jamaoncol.2019.1588

  • CrossRef
  • Google Scholar

• 42

  SharmaPHu-LieskovanSWargoJARibasA. Efficacy and tolerability of tremelimumab in patients with metastatic pancreatic ductal adenocarcinoma. J Clin Oncol. (2017) 168(4):707–23. doi: 10.1200/jco.2018.36.4_suppl.470

  • CrossRef
  • Google Scholar

• 43

  BrahmerJRTykodiSSChowLQHwuWJTopalianSLHwuPet al. Safety and activity of anti-PD-L1 antibody in patients with advanced cancer. N Engl J Med. (2012) 366:2455–65. doi: 10.1056/nejmoa1200694

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 44

  RoyalRELevyCTurnerKMathurAHughesMKammulaUSet al. Phase 2 trial of single agent ipilimumab (anti-CTLA-4) for locally advanced or metastatic pancreatic adenocarcinoma. J Immunother. (2010) 33:828–33. doi: 10.1097/cji.0b013e3181eec14c

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 45

  KamathSDKalyanAKircherSNimeiriHFoughtAJBensonAet al. Ipilimumab and gemcitabine for advanced pancreas cancer: a phase Ib study. Am Soc Clin Oncol. (2016) 25(5):e808–e815. doi: 10.1200/jco.2016.34.15_suppl.e15747

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46

  MaJSunDWangJHanCQianYChenGet al. Immune checkpoint inhibitors combined with chemotherapy for the treatment of advanced pancreatic cancer patients. Cancer Immunol Immunother. (2020) 69:365–72. doi: 10.1007/s00262-019-02452-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 47

  WeissGJBlaydornLBeckJBornemann-KolatzkiKUrnovitzHSchützEet al. Phase Ib/II study of gemcitabine, nab-paclitaxel, and pembrolizumab in metastatic pancreatic adenocarcinoma. Invest New Drugs. (2018) 36:96–102. doi: 10.1007/s10637-017-0525-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48

  ReissKAMickRTeitelbaumUO'HaraMSchneiderCMassaRet al. Niraparib plus nivolumab or niraparib plus ipilimumab in patients with platinum-sensitive advanced pancreatic cancer: a randomised, phase 1b/2 trial. Lancet Oncol. (2022) 23:1009–20. doi: 10.1016/s1470-2045(22)00369-2

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49

  PishvaianMJWangHHeARHwangJJSmagloBGKimSSet al. A phase I/II study of veliparib (ABT-888) in combination with 5-fluorouracil and oxaliplatin in patients with metastatic pancreatic cancer. Clin Cancer Res. (2020) 26:5092–101. doi: 10.1158/1078-0432.ccr-20-1301

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50

  NaingAPowderlyJDFalchookGCreelanBNemunaitisJLutzkyJet al. Abstract CT177: Epacadostat plus durvalumab in patients with advanced solid tumors: preliminary results of the ongoing, open-label, phase I/II ECHO-203 study. Cancer Res. (2018) 78:CT177–7. doi: 10.1158/1538-7445.am2018-ct177

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51

  BockornyBSemenistyVMacarullaTBorazanciEWolpinBMStemmerSMet al. BL-8040, a CXCR4 antagonist, in combination with pembrolizumab and chemotherapy for pancreatic cancer: the COMBAT trial. Nat Med. (2020) 26:878–85. doi: 10.1038/s41591-020-0880-x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52

  Wang-GillamALockhartACTanBRSureshRLimKHRatnerLet al. Phase I study of defactinib combined with pembrolizumab and gemcitabine in patients with advanced cancer. Am Soc Clin Oncol. (2018) 36(15_suppl):2561. doi: 10.1200/JCO.2018.36.15_suppl.2561

  • CrossRef
  • Google Scholar

• 53

  Al-KhinjiAAl-KorbiNAl-KuwariSAl-HorAMaloucheD. Immune checkpoint inhibitors in pancreatic adenocarcinoma: a systematic review and meta analysis of clinical outcomes. Front Oncol. (2025) 15:1569884. doi: 10.3389/fonc.2025.1569884

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 54

  MarabelleAO'MalleyDMHendifarAEAsciertoPAMotola-KubaDPenelNet al. Pembrolizumab in microsatellite-instability-high and mismatch-repair-deficient advanced solid tumors: updated results of the KEYNOTE-158 trial. Nat Cancer. (2025) 6:253–8. doi: 10.1038/s43018-024-00894-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55

  RodriguezEZwartESAffandiAAVerhoeffJde KokMBoydLNCet al. In-depth immune profiling of peripheral blood mononuclear cells in patients with pancreatic ductal adenocarcinoma reveals discriminative immune subpopulations. Cancer Sci. (2024) 115:2170–83. doi: 10.1111/cas.16147

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 56

  LahusenACaiJSchirmbeckRWellsteinAKlegerASeufferleinTet al. A pancreatic cancer organoid-in-matrix platform shows distinct sensitivities to T cell killing. Sci Rep. (2024) 14:9377. doi: 10.1038/s41598-024-60107-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 57

  AjinaRWeinerLM. T-cell immunity in pancreatic cancer. Pancreas. (2020) 49:1014–23. doi: 10.1097/mpa.0000000000001621

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 58

  BliseKESivagnanamSBettsCBBetreKKirchbergerNTateBJet al. Machine learning links T-cell function and spatial localization to neoadjuvant immunotherapy and clinical outcome in pancreatic cancer. Cancer Immunol Res. (2024) 12:544–58. doi: 10.1158/2326-6066.cir-23-0873

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 59

  ChaiCTangHMiaoXSuYLiLYuCet al. Establishment and characterization of the PDAC-X3 cell line: a novel Chinese-origin pancreatic ductal adenocarcinoma cell line. Hum Cell. (2024) 37:1578–92. doi: 10.1007/s13577-024-01100-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60

  XieBHMundorfVBetzJReimcheIErkanMBüdeyriIet al. Pseudopterosin and O-methyltylophorinidine suppress cell growth in a 3D spheroid co-culture model of pancreatic ductal adenocarcinoma. Bioengineering. (2020) 7(2):57. doi: 10.3390/bioengineering7020057

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 61

  StringerCWangTMichaelosMPachitariuM. Cellpose: a generalist algorithm for cellular segmentation. Nat Methods. (2021) 18:100–6. doi: 10.1038/s41592-020-01018-x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 62

  SchmidBSchindelinJCardonaALongairMHeisenbergM. A high-level 3D visualization API for Java and ImageJ. BMC Bioinf. (2010) 11:274. doi: 10.1186/1471-2105-11-274

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 63

  SchindelinJArganda-CarrerasIFriseEKaynigVLongairMPietzschTet al. Fiji: an open-source platform for biological-image analysis. Nat Methods. (2012) 9:676–82. doi: 10.1038/nmeth.2019

  • CrossRef
  • Google Scholar

• 64

  SchiedlauskeKDeipenbrockAPfliegerMHamacherAHänselJKassackMUet al. Novel histone deacetylase (HDAC) inhibitor induces apoptosis and suppresses invasion via E-cadherin upregulation in pancreatic ductal adenocarcinoma (PDAC). Pharmaceuticals. (2024) 17:752. doi: 10.3390/ph17060752

  • CrossRef
  • Google Scholar

• 65

  StegmaierJOtteJCKobitskiABartschatAGarciaANienhausGUet al. Fast segmentation of stained nuclei in terabyte-scale, time resolved 3D microscopy image stacks. PloS One. (2014) 9:e90036. doi: 10.1371/journal.pone.0090036

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 66

  FeigCJonesJOKramanMWellsRJDeonarineAChanDSet al. Targeting CXCL12 from FAP-expressing carcinoma-associated fibroblasts synergizes with anti-PD-L1 immunotherapy in pancreatic cancer. Proc Natl Acad Sci USA. (2013) 110:20212–7. doi: 10.1073/pnas.1320318110

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 67

  BrouwerTPde VriesNLAbdelaalTKrogRTLiZRuanoDet al. Local and systemic immune profiles of human pancreatic ductal adenocarcinoma revealed by single-cell mass cytometry. J Immunother Cancer. (2022) 10:e004638. doi: 10.1136/jitc-2022-004638

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 68

  BoilèveACartryJGoudarziNBedjaSMathieuJRRBaniMAet al. Organoids for functional precision medicine in advanced pancreatic cancer. Gastroenterology. (2024) 167:961–976.e13. doi: 10.1053/j.gastro.2024.05.032

  • CrossRef
  • Google Scholar

• 69

  MorceauFEl-KhouryVLeeKBernaMJKwonYJ. Pancreatic cancer organoids: modeling disease and guiding therapy. Cancers (Basel). (2025) 17:3850. doi: 10.3390/cancers17233850

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 70

  DeipenbrockAWilmesBESommermannTAbdoNMoustakasKRaaschMet al. Modelling of the multicellular tumor microenvironment of pancreatic ductal adenocarcinoma (PDAC) on a fit-for-purpose biochip for preclinical drug discovery. Lab Chip. (2025) 25(9):2168–81. doi: 10.1039/d4lc01016g

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 71

  Gok YavuzBGunaydinGGedikMEKosemehmetogluKKarakocDOzgurFet al. Cancer associated fibroblasts sculpt tumour microenvironment by recruiting monocytes and inducing immunosuppressive PD-1(+) TAMs. Sci Rep. (2019) 9:3172. doi: 10.1038/s41598-019-39553-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 72

  NealJTLiXZhuJGiangarraVGrzeskowiakCLJuJet al. Organoid modeling of the tumor immune microenvironment. Cell. (2018) 175:1972–1988.e16. doi: 10.1016/j.cell.2018.11.021

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 73

  DaunkeTBeckingerSRahnSKrügerSHecklSSchäferHet al. Expression and role of the immune checkpoint regulator PD-L1 in the tumor-stroma interplay of pancreatic ductal adenocarcinoma. Front Immunol. (2023) 14:1157397. doi: 10.3389/fimmu.2023.1157397

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 74

  ShakibaMTuvesonDA. Macrophages and fibroblasts as regulators of the immune response in pancreatic cancer. Nat Immunol. (2025) 26:678–91. doi: 10.1038/s41590-025-02134-6

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 75

  MercantiLSindacoMMazzoneMDi MarcantonioMCPiscioneMMuraroRet al. PDAC, the influencer cancer: cross-talk with tumor microenvironment and connected potential therapy strategies. Cancers (Basel). (2023) 15:2923. doi: 10.3390/cancers15112923

  • Pubmed Abstract
  • CrossRef
  • Google Scholar