Spatially resolved analysis of TGF/BMP signalling in pancreatic ductal adenocarcinoma by digital pathology identifies patient subgroups with adverse outcome

Spatially resolved analysis of TGF/BMP signalling in pancreatic ductal adenocarcinoma by digital pathology identifies patient subgroups with adverse outcome | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Spatially resolved analysis of TGF/BMP signalling in pancreatic ductal adenocarcinoma by digital pathology identifies patient subgroups with adverse outcome Konstantin Bräutigam, Philipp Zens, Stefan Reinhard, Jessica L Rohrbach, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6506146/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 18 Aug, 2025 Read the published version in BMC Cancer → Version 1 posted 8 You are reading this latest preprint version Abstract Background Transforming Growth Factor (TGF) and Bone Morphogenetic Protein (BMP) signalling critically influence pancreatic ductal adenocarcinoma (PDAC) progression, with TGF-B paradoxically exerting both tumour-promoting and -suppressive effects. Parallel to this observation, the specific context-dependent, spatial dynamics of these pathways and their interaction with the tumour microenvironment (TME) remain poorly understood. Methods We performed a spatially resolved analysis of PDAC on a multi-region tissue microarray cohort of 117 curatively resected PDAC specimens consisting of tumour centre (TC), tumour front (TF), and stromal(-predominant) tissue cores each. Protein (ID1, pSMAD2) and mRNA (TGF-A, TGF-B1/2, BMP4, GREM1) expression were assessed in each tissue compartment by immunohistochemistry and in situ hybridization, respectively, quantified by digital image analysis, and correlated with clinicopathologic features. Results ID1 was significantly overexpressed in PDAC cells compared to associated stroma (p < 0.01), while pSMAD2 was largely absent in PDAC cells, but preserved among associated stroma compartments, particularly in TF cores (p = 0.04). Higher stromal GREM1 signal correlated with reduced overall tumoural ID1 protein expression (p = 0.02), and TGF-B2 high /TGF-A low stroma was significantly associated with worse survival (p < 0.01). Intratumoural TGF-B2 was inversely correlated with stromal pSMAD2 expression (p = 0.03) and was associated with lymph node involvement (p = 0.02). FOXP3 + regulatory T-cells were significantly reduced in TGF-B2 high tumours (p = 0.04), while higher tumoural TGF-B1 exhibited a trend towards increased FOXP3 + cells (p = 0.08). Conclusions Our spatial analysis reveals intratumoural heterogeneity of TGF/BMP signalling and its significance for PDAC progression. Notably, stromal TGF-B2 emerges as a prognostic biomarker, while TGF-B1 and ID1 are implicated in adverse clinical and pathologic features. These findings highlight the importance of TGF/BMP signalling niches in the TME with implications for PDAC biology and can inform the development of future therapeutic strategies. Transforming Growth Factors Bone Morphogenetic Proteins Stromal Cells Pancreatic Neoplasms AI (Artificial Intelligence) GREM1 protein Inhibitor of Differentiation Protein 1 (ID1) Spatial Analysis Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Transforming Growth Factor (TGF) and Bone Morphogenetic Protein (BMP) signalling play a crucial role in cancer development and progression ( 1 , 2 ). In pancreatic ductal adenocarcinoma (PDAC), one of the most lethal cancer types with increasing incidence ( 3 ), TGF-B is known to play a paradoxical role, i.e., it can be tumour-promoting (especially in late-stage, metastatic PDAC) or -suppressive ( 4 , 5 ). Promisingly, blocking TGF-B1/B2 enhances sensitivity to combination chemotherapy in PDAC ( 6 ), and silencing TGF-B2 expression inhibits tumour proliferation in PDAC cell lines ( 7 ) and can boost gemcitabine chemosensitivity ( 8 ). TGF-B is a well-known inducer of epithelial-mesenchymal transition (EMT) ( 4 ), a process that alters cytoskeletal structure and cellular motility, often enhancing the metastatic potential of cancer cells, and driving a more aggressive phenotype ( 9 , 10 ) with nuclear abnormalities and chromatin remodelling ( 11 ). High-grade tumour budding, as a morphological “correlate” of EMT ( 12 ), has been shown to be a strong predictor of worse outcome in PDAC ( 13 , 14 ). Despite the established roles of BMPs and Suppressor of Mothers against Decapentaplegic (SMAD) proteins in regulating EMT and invasiveness, their specific influence on tumour budding in PDAC remains poorly understood ( 15 ). The tumour-promoting effect of TGF-B is pronounced in the context of SMAD4 loss ( 4 ), a gene involved in pancreatogenesis and frequently inactivated in PDAC ( 16 ). SMAD2, a downstream intracellular mediator activated by TGF-B, forms a heterodimeric complex with SMAD4 ( 17 ) (Fig. 1 ). After nuclear translocation, the SMAD2/SMAD4 complex controls gene expression of various processes, such as apoptosis, cell cycle arrest or EMT ( 18 ). TGF-B also has a profound effect on the tumour microenvironment (TME) in PDAC, e.g., by inducing heterogeneity of cancer-associated fibroblasts (CAFs) ( 19 ). TGF-A signalling through the epidermal growth factor receptor (EGFR) has been shown to activate pancreatic stellate cells and may contribute to chronic pancreatitis and pancreatic carcinogenesis ( 20 ). Crosstalk between the TGF-B and BMP pathways exists: Upon binding to BMP receptors, BMPs first activate SMAD1/5/8 and then SMAD4 as the common downstream protein in both pathways (Fig. 1 ). BMP signalling induces the expression of Inhibitor of DNA binding 1 (ID1), which is known to be upregulated in pancreatitis and PDAC ( 21 , 22 ) and is believed to have tumour-promoting effects by enabling bypassing of TGF-B signalling ( 23 ). Protein overexpression of ID1 has been described as a negative prognostic predictor associated with higher tumour (T)- and nodal (N)-stage as well as higher microvessel density in PDAC ( 24 ). Interestingly, ID1/ID3 knockdown seems to inhibit PDAC metastasis in human pancreatic cancer cells ( 25 ). Gremlin1 (GREM1) is a BMP antagonist with regulatory function on cellular heterogeneity in PDAC ( 26 ). Its inactivation promotes EMT and lower GREM1 levels are associated with late-stage disease. Recent evidence suggests that GREM1 also influences the TME by driving a fibrogenic activated stroma ( 27 ) with higher macrophage counts ( 28 ). An additional microenvironmental factor is TGF-B-induced immune escape, e.g., by recruiting FOXP3 + T-cells ( 29 , 30 ), and thereby turning the tumour into a “cold cancer” with worse outcome. The creation of an immunosuppressive, tumour-permissive (modulatory) niche is supported by the typical context of (excessive) desmoplasia in PDAC, analogous to wound healing ( 31 ). In this work, we spatially resolve and comprehensively quantify key signalling molecules of both, TGF and BMP, pathways in PDAC and the immediate environment of invading cancer cells using digital pathology and a combined protein and mRNA expression profiling approach. Materials and methods Patient cohort and PDAC tissue microarray (TMA) Our tissue microarray (TMA) contains 117 cases of curatively resected PDAC treated at the Department of Visceral Surgery of Inselspital Bern, including thirteen with neoadjuvant treatment (cohort details in Table 1 , Table S1 and in ( 32 )), and a mean (median; range) follow-up of 794.6 (567; 3 to 3000) days. To account for tumor heterogeneity, each patient case is represented by three tissue cores selected by an expert pathologist (MW) on morphologic grounds from different tumour regions, core diameter 0.6 mm/area 0.28 mm 2 : Tumour front (TF) to account for the infiltrative capacity of the cancer, i.e., the outermost tumour periphery, which is further characterised by histological hallmarks of invasion, e.g. tumour buds and single cell growth; Tumour centre (TC) sampled from the central tumour mass; stroma(-predominant) (in total n = 351 tissue cores). Table 1 Clinico-pathological details of the PDAC cohort (n = 104) without neoadjuvant treatment (neoadjuvant in Table S1 ). a The Grade 4 (G4) tumour represents an undifferentiated carcinoma, anaplastic type. ADM : Acinar-to-Ductal Metaplasia; BD : Tumour Budding; ITBCC : International Tumour Budding Consensus Conference; IPMN : Intraductal Papillary Mucinous Neoplasm; NA : not available; PanIN : Pancreatic Intraepithelial Neoplasia; SD : Standard Deviation. UICC : Union Internationale Contre le Cancer Characteristic N = 104 1 Age (years) 68.6 (SD 9.6) Gender Female 45 / 104 (43.3%) Male 59 / 104 (56.7%) Deceased 78 / 103 (75.7%) NA 1 Overall Survival (days) 801.3 (SD 674.2) NA 2 Type of resection Left resection 11 / 104 (10.6%) Left resection and adrenalectomy 1 / 104 (1.0%) Left resection, colectomy and adrenalectomy 1 / 104 (1.0%) Total pancreatectomy 16 / 104 (15.4%) Whipple procedure 75 / 104 (72.1%) Size invasive tumour (mm) 35.3 (SD 11.9, range 12 to 95) Intrapancreatic location invasive tumour Head 81 / 104 (77.9%) Head and body 1 / 104 (1.0%) Head, body and tail 2 / 104 (1.9%) Tail 8 / 104 (7.7%) Body 7 / 104 (6.7%) Body and tail 5 / 104 (4.8%) T-Stage (UICC 8th) 1c 5 / 104 (4.8%) 2 69 / 104 (66.3%) 3 30 / 104 (28.8%) N-Stage (UICC 8th) 0 19 / 104 (18.3%) 1 42 / 104 (40.4%) 2 43 / 104 (41.3%) M-Stage (UICC 8th) 0 103 / 104 (99.0%) 1 1 / 104 (1.0%) Lymphatic vessel infiltration (L1) 82 / 104 (78.8%) Blood vessel infiltration (V) 0 12 / 104 (11.5%) 1 89 / 104 (85.6%) 2 3 / 104 (2.9%) Perineural infiltration (Pn1) 101 / 104 (97.1%) Resection status (R) 0 64 / 104 (61.5%) 1 31 / 104 (29.8%) 2 2 / 104 (1.9%) x 7 / 104 (6.7%) Grading (G) 1 12 / 104 (11.5%) 2 51 / 104 (49.0%) 3 40 / 104 (38.5%) 4 a 1 / 104 (1.0%) Tumour Budding (ITBCC) BD0 2 / 104 (1.9%) BD1 29 / 104 (27.9%) BD2 31 / 104 (29.8%) BD3 42 / 104 (40.4%) Total Lymph Node (LN) yield 30.0 (12.0) Positive LN 4.6 (SD 5.3) Presence of Extracapsular Extension 46 / 104 (44.2%) 1 Mean (SD); n / N (%) Immunohistochemistry (IHC) Primary antibodies were incubated for 30 min at room temperature and used as follows: Anti-Id1 (ab203202, rabbit polyclonal, Abcam, Cambridge, United Kingdom, dilution: 1:200, retrieval: Citrate buffer pH 6, 20 min); CD8 (M710301, mouse monoclonal, clone C8/144B, DAKO/Agilent Technologies, CA, United States, dilution 1:200, retrieval: TrisEDTA, 20 min); CD68 (M087601, clone PG-M1, monoclonal mouse, DAKO/Agilent Technologies, dilution: 1:200,Tris EDTA, 30 min); CD163 (NCL-CD163, clone 10D6, mouse monoclonal, Leica Biosystems, Newcastle upon Tyne, United Kingdom, dilution: 1:400, retrieval: Tris EDTA, 20 min); FOXP3 (14-477-82, mouse monoclonal, clone 236A/E7, Invitrogen, MA, United States, dilution: 1:50, retrieval: Citrate buffer pH 6, 30 min); pSMAD2 (3108S, rabbit monoclonal, Cell Signaling Technology, London, United Kingdom, dilution: 1:200, retrieval: Citrate buffer pH 6, 20 min). Antibody detection was performed with the BOND Polymer Refine DAB kit (Leica Biosystems) using 3,3-diaminobenzidine as a brown chromogen. The samples were counterstained with hematoxylin. TMA slides were scanned on a 3DHISTECH Pannoramic 250 Flash II scanner (3DHISTECH, Budapest, Hungary). Protein expression was rigorously validated by three board-certified pathologists (KB, VHK and MW) and by including appropriate on-slide controls. Messenger RNA (mRNA) in-situ hybridisation (ISH) mRNA transcripts of TGF-A (reg 58, reference 313138 Hs-TGFA), TGF-B1 (reg 52, reference 400888 Hs-TGFB1), TGF-B2 (reg 54, reference 489248 Hs-TGFB2), BMP ligand BMP4 (reg 56, reference 454308 Hs-BMP4), and BMP antagonist GREM1 (reg 46, reference 312831 Hs-GREM1) were visualized by applying RNAscope (2.5 LS DUPLEX, Reagent Kit-BROWN) chromogenic probes (ACD BioTechne, Newark, CA, USA) on the TMA sections described above using TrisEDTA epitope retrieval buffer (95°, 20 min) and ACD enzyme for 15 min, together with a negative control probe (2.5 LS Duplex “DapB”). Digital Image Analysis (DIA) DIA was performed using HALO AI™ (Indica Labs, NM, USA). The digitized TMA slides were de-arrayed into individual spot images of each tissue sample, which were then linked to the corresponding clinical annotations. After the first visual review, seven tissue cores with insufficient tissue and/or artifacts were excluded from further analysis. To localize and quantify tumour and stromal tissue, a deep neural network algorithm was trained based on a pre-trained DenseNet network (Figs. 2 A and 2 B). Graphical overlays were generated for each tissue class and the classification accuracy was visually verified. The total area of each tissue class was quantified in square millimetres (mm²). Nuclear segmentation and staining quantification were performed in both the tumour and stromal compartments. For nuclear segmentation, the HALO™ AI pre-trained convolutional neural network was used and fine-tuned with application-specific training examples. Pixels were classified as positive if the staining intensity exceeded the internal controls, as validated by pathologist review (KB, VHK). The total area of vital tissue in the tumour and stromal compartments, the number of ID1-positive and pSMAD2-positive cells ( Cytonuclear v2.0.9 module), and mRNA transcripts of TGFA, TGFB1, TGFB2, BMP4 and GREM1 ( ISH IHC v3.2.5 module) were recorded for analysis with clinicopathologic features. Marker quantification subproperties such as cytoplasmic radius, contrast thresholds, optical density, segmentation aggressiveness, and nuclear staining intensity were optimized based on expert pathologist review. Nuclear roundness (from 0 to 1, with 1 being perfectly round) and cytoplasmic radius were measured by HALO™ Image Analysis Software. Tumour budding and immune cell quantification The grade of tumour budding was assigned by consensus (BD0 to BD3) ( 14 ) for each tumour using representative hematoxylin and eosin (H&E) slides. In addition, tumour budding was individually counted at the tissue core level for each individual marker by an expert pathologist (MW) in a blinded, independent and randomized manner using the browser-based online TMA analysis tool "Scorenado" ( 33 ). CD8 + T-cells, FOXP3 + T-regulatory cells and macrophages (CD68 + and CD163 + ) have been binarily classified into “low” and “high” per tissue core (MW). Validation using The Cancer Genome Atlas (TCGA) The findings were independently validated using open-source, publicly available data from The Cancer Genome Atlas (TCGA). mRNA data (pancreatic adenocarcinoma (PAAD), n = 177) were accessed and plots were generated using the online tool “UALCAN: A portal for facilitating tumor subgroup gene expression and survival analyses” ( https://ualcan.path.uab.edu/cgi-bin/ualcan-res.pl ) ( 34 ). Statistical analysis For statistical analyses, only tissue cores with at least 25 tumour and stromal cells (corresponding to approximately the 25th percentile of epithelial cell counts in the raw data output) were used (thereby losing all tissue cores from a total of five patients). In the non-neoadjuvant cases (n = 104), the PDAC cell count ranged from 25 to 1,605 cells per tissue core (median: 218; mean: 305; interquartile range: 290), while the stromal cell count ranged from 42 to 1,862 cells per tissue core (median: 638; mean: 712; interquartile range: 430). Unless stated otherwise, cases involving neoadjuvant therapy were excluded from downstream statistics. Statistical analyses were performed using the R statistical computing environment version 4.4.1 (RStudio version 2024.04.2). All p-values were calculated for two-tailed tests with significance set at p < 0.05. Statistical analyses included Wilcoxon signed-rank tests for pairwise comparisons, Spearman's rank correlation coefficient to assess associations, Kaplan-Meier survival analysis with log-rank tests for time-to-event data, and chi-squared tests to assess relationships between categorical variables. Results ID1 is highly expressed, while pSMAD2 protein mostly absent in PDAC parenchyma ID1 protein was highly expressed among epithelial PDAC cells compared to the surrounding stroma (p < 0.01, non-neoadjuvant cases) in both TC and TF (Figs. 2 A, S1 and S8 A, Table 2 ). Mean ID1 protein expression was even more pronounced in the neoadjuvant-treated cases (both TC and TF, Table S2 ). pSMAD2 protein expression was mostly absent in PDAC parenchyma, aligning well with TCGA mRNA data (reduced transcript counts compared to normal parenchyma, Figure S8 B ), with significantly higher preservation in the stroma (p < 0.01, non-neoadjuvant cases) (Figure S2 A , Tables 2 and S2 ). Mean pSMAD2 expression in PDAC was significantly lower in the TC than in TF (p = 0.02) while mean stromal pSMAD2 protein expression was significantly higher in the TF (p = 0.04) (non-neoadjuvant cases, Figure S2 B ). Both PDAC parenchyma and stroma were mostly BMP4 low (non-neoadjuvant cases, mean 0.25 mRNA transcripts per tumour cell) with a few parenchymal BMP4 high cases (maximum of 6.3 transcripts/tumour cell (TF) in a ATM-mutant pT3 pN1 duct adenocarcinoma ( 35 , 36 ) with adjuvant FOLFIRINOX ( 37 ) and > 5 years overall survival; hyperactive BMP4 signalling described in PDAC with ATM loss ( 38 ); respective tissue core in Figure S1 ) (Table 2 ). Table 2 Expression of TGF- and BMP-signalling molecules in PDAC and associated stroma, each for Tumour Centre (TC) and Tumour Front (TF) (non-neoadjuvant only, n = 82 analysed, neoadjuvant in Table S2 ). Protein expression (ID1, pSMAD2 immunohistochemistry) in percent (%), mRNA transcripts (in-situ hybridisation) per PDAC cell and stromal cell respectively (including only tissue cores with at least 25 stromal and at least 25 cancer epithelial cells). SD : Standard Deviation. Region PDAC_ID1 Stroma_ID1 PDAC_pSMAD2 Stroma_pSMAD2 TC 63.8% (0-100, SD = 22.7%) 22.7% (3.7–77.7, SD = 13.1%) 16.3% (0-100, SD = 21.3%) 15.6% (0.7–87.8, SD = 18.9%) TF 60.5% (0-100, SD = 24.4%) 23.9% (3.1–72.6, SD = 12.9%) 16.5% (0-100, SD = 26.4%) 23.7% (0-96.7, SD = 25.1%) Region PDAC_GREM1 Stroma_GREM1 PDAC_TGFA Stroma_TGFA PDAC_TGFB1 Stroma_TGFB1 PDAC_TGFB2 Stroma_TGFB2 PDAC_BMP4 Stroma_BMP4 TC 0.7 (0-5.5, SD = 1.1) 0.9 (0-6.4, SD = 1.2) 1.3 (0.2–3.4, SD = 0.7) 1.5 (0-3.2, SD = 0.7) 1.8 (0.1–8.5, SD = 1.6) 0.9 (0-2.4, SD = 0.6) 0.8 (0.1–2.3, SD = 0.4) 1.2 (0.4–3.1, SD = 0.5) 0.4 (0-2.4, SD = 0.4) 0.3 (0-1.3, SD = 0.2) TF 0.6 (0-3.9, SD = 0.7) 1.1 (0-10.9, SD = 1.6) 1.3 (0.2–5.6, SD = 0.9) 1.5 (0.2–3.3, SD = 0.8) 1.8 (0.2–6.1, SD = 1.2) 0.9 (0-2.2, SD = 0.5) 0.8 (0.1–3.1, SD = 0.5) 1.1 (0.2–2.3, SD = 0.4) 0.5 (0-6.3, SD = 0.9) 0.4 (0-1.1, SD = 0.3) Strong GREM1 signalling in tumour-associated stroma and robust TGF signalling across compartments GREM1, TGF-A and TGF-B2 mRNA counts per cell were significantly higher in the stromal compartment compared to the PDAC cells (non-neoadjuvant only, Figs. 2 B-D, Table 2 ). On the contrary, TGF-B1 mRNA transcript counts were significantly higher in the PDAC compartment (p < 0.01). There were no significant differences between TC and TF in transcript counts for the three investigated TGF ligands, BMP4 and GREM1. Both among the tumoural and stromal compartment, transcript counts of TGF-A, TGF-B1 and -B2 showed strong positive correlations (Figs. 3 A and 3 B, Table S3 ). In PDAC parenchyma, higher GREM1 transcript counts were negatively correlated with ID1 protein expression (rho= -0.146, p = 0.03), consistent with biological priors. When comparing tumoural and stromal metrics, higher stromal GREM1 transcript numbers correlated with higher stromal TGF-A and -B1 transcript counts (rho = 0.40, p < 0.01; rho = 0.42, p < 0.01, Fig. 3 C, Table S3 ). Higher tumoural BMP4 correlated with stromal TGF-A transcripts (rho = 0.42, p < 0.01). Low GREM1 signalling and high ID1 expression define an aggressive PDAC phenotype While GREM1 low PDACs showed a trend towards perineural invasion (TC: p = 0.06, Fig. 4 A) and blood vessel infiltration (V0 vs. V2: p = 0.06, Fig. 4 B), expression of any of the signalling molecules in PDAC (TC and TF) was not significantly associated with the presence of extracapsular extension of lymph node (LN) metastases, lymphovascular (“L1”) or perineural invasion (“Pn1”), tumour stage or tumour grade. Tumour size correlated weakly with ID1 expression in the TF (p = 0.03, R 2 = 0.06; Figure S3 A ), but not in the TC (p = 0.14). N2-stage PDAC showed lower tumoural ID1 expression than N1-stage (p = 0.02; Fig. 4 C, N0 vs. N1/N2: p = 0.73, TCGA: Figure S8 C ). Other molecules exhibited no significant associations with tumour size, number of positive LN ( Figure S3 C and S3D ) or N-stage. Lower average GREM1 transcript counts per stromal (TC: p < 0.01; TF: p = 0.07, t-test) and tumoural cell (TC: p < 0.01; TF: ns, t-test) in the TC spots, significantly associated with presence of metachronous distant metastasis in follow-up. TGF-B2 transcript counts per cell in PDAC parenchyma correlated significantly with the absolute number of positive lymph nodes (TF, p = 0.02, Figure S3 B ). TCGA data support this finding (non-significant trend, Figure S8 D ) and reveal higher tumoral transcript counts in N1-stage PDAC. TGF-A low (p = 0.06, Wilcoxon rank sum; p = 0.09, Fisher test) and -B1 low stroma (p = 0.08, Wilcoxon rank sum; p = 0.07, Fisher test) showed a statistically non-significant trend towards higher tumour budding (BD2 and BD3, TC, Figure S4 ). When quantifying tumour buds per individual tissue core, we found no significant differences in tumour bud counts for any of the signalling molecules studied ( Table S4 ). This held true whether we compared the average marker expression in PDAC versus stroma or between TC and TF per tissue core. Heterogeneity of immune infiltrate and sparsity of FOXP3 + T-regulatory cells in TGF-B2 high PDAC Immune cell counts were heterogeneously distributed among the different tumour regions. CD3 had the greatest heterogeneity (43.3% of patients had consistent expression patterns (low, high) across tissue cores) in all three tumour regions (TC, TF, stromal-predominant tissue cores), while CD20 had the least (85.5); this pattern persisted when considering only TC and TF (CD3: 63.5%, CD20: 88.5%) ( Table S5 ). TGF-B is known for its direct suppressive effect on T-cell function and induction of T-regulatory cells ( 30 , 39 ) and the TF (invasive margin) is known as a site of pronounced immune density in PDAC ( 40 , 41 ). In our cohort, we observe a statistically non-significant trend towards higher FOXP3 counts in the TF of TGF-B1 high tumours (p = 0.08, Figure S5 A ; TC: p = 0.29). Stromal TGF-B1 copies did not associate with FOXP3 levels (TC: p = 1; TF: p = 0.64). Higher stromal and intratumoural TGF-B2 transcripts, significantly associated with lower FOXP3 counts in the TC (p = 0.04 and p = 0.05 respectively, Figure S5 A ). ID1 has been shown to suppress CD8 + T-cell infiltration ( 42 ). Here, CD8 high tissue cores were not associated with ID1 expression levels (TC: p = 0.35, TF: p = 0.17, Figure S5 B ). No significant associations between TGF-B1 or TGF-B2 (in either compartment) and CD8 + T-cell infiltration were detected (TC/TF: all non-significant). While BMP4 has been shown to induce M2-polarization of macrophages in bladder cancer ( 43 ), we did not observe significant differences in CD68 + or CD163 + macrophage levels between BMP4 low and - high PDAC (TC: p = 0.85; TF: p = 0.52). Also, GREM1 transcripts have been described to be associated with increased M1- and M2-macrophages in PDAC ( 28 , 44 ). Binary (by mean) stratification of BMP4 (CD68, TC: p = 0.24; TF p = 0.71; CD163, TC: p = 0.41; TF: p = 0.71) or GREM1 (CD68, TC: p = 0.17; TF: p = 0.35; CD163, TC: p = 0.31; TF: p = 0.56) mRNA transcript counts did not show significant correlations with macrophage levels ( Figure S5 C and -D ). Stromal TGF-A and -B2 linked to worse PDAC survival Lower number of stromal TGF-B2 copies in the TC were significantly associated with worse survival (p = 0.02, neoadjuvant cases excluded, Fig. 5 A, multivariate Cox model in Figure S6 A ). A higher number of stromal TGF-A in the TF showed a statistically non-significant trend towards worse survival (p = 0.069). Interestingly, when both stromal TGF-A and -B2 were combined, the poorest overall survival was seen in the TGF-A low (TF)-TGFB2 high (TC) stroma group (p < 0.01, Fig. 5 B), highlighting the complex interaction between these molecules and their non-linear relationship. Comparison of stromal TGF-A and -B2 groups within the same tumour region did not show statistical significance (TC/TC: p = 0.06; TF/TF: p = 0.21). Binary survival analysis (low versus high) of bulk mRNA TCGA data did not reveal any significant differences in outcomes for the investigated TGFs ( Figure S8 E ). Neither TGF-B1, BMP4 and GREM1 transcript levels nor ID1 and pSMAD2 protein expression were significantly associated with survival in either tumour or stromal compartments ( Table S5 ). “Mesenchymal geometry” and stromal proportion does not correlate with PDAC outcome To investigate EMT and its impact on tumour aggressiveness in PDAC, we examined nuclear roundness as a morphological EMT surrogate, i.e., more spindle-shaped, less round morphology in the mesenchymal cell spectrum. Nuclear circularity is decreased during EMT induction ( 45 ) and, for instance, in breast cancer a “rounded to deformed morphology” has been described after EMT ( 46 ). Here, the automatically computed roundness of PDAC nuclei was between 0.61 and 0.80 (mean of 0.73 among all tissue cores, neoadjuvant cases excluded). There was no significant correlation of nuclear roundness with N-stage in this cohort (n = 102, non-neoadjuvant, TC). Further, there was no significant difference in survival comparing less round to rounder PDAC nuclei (p = 0.83, TC) ( Figure S6 C ). TGF-B is a known (sometimes SMAD-dependent) inducer of tissue fibrosis ( 47 ). In PDAC, the typically dense desmoplastic stroma is traditionally considered a negative prognostic factor and is thought to form a tumour protective "niche" and physical barrier that can reduce drug penetration and prevent immune infiltration. Here, the amount of stroma was not associated with worse survival (neoadjuvant patients excluded, stratified by the mean stromal percentage per tissue core, TF: p = 0.23; TC: p = 0.32, Figure S6 B ). Discussion The results of this study on TGF- and BMP-signalling in PDAC align well with the known microenvironmental ( 48 , 49 ), transcriptomic ( 50 ) and cellular heterogeneity of PDAC across space, time and anatomical regions ( 40 , 51 , 52 ). Recent evidence suggests a site-specific TME ( 53 ) with intratumoural heterogeneity on single-cell level ( 54 ) involving microenvironmental transitions in metastatic progression of PDAC ( 52 ). In our spatially-resolved analysis investigating multiple tissue cores per PDAC case from different morphological tumour regions to account for tissue heterogeneity, we reveal differences in TGF/BMP signalling and, consistent with previous literature ( 55 – 57 ), in immune cell distribution across tumour regions. A chronic inflammatory state during manifestation of PDAC paralleled by dysregulation of the associated microenvironment might provide a cancer-promoting milieu and allow tumour cells to exploit physiological functional niches during tumour progression (“niche hijacking”) ( 58 , 59 ). In mice, SMAD2 and − 3 have been revealed to be crucial molecules that trigger a TGF-B-induced regulatory T-cell response ( 60 ). We observe contradictory results in TGF-B signalling: While TGF-B2 high PDAC significantly associated with less FOXP3 + T-regulatory cell infiltration in the TC, TGF-B1 high PDAC showed the opposite trend. Higher FOXP3 + counts in the proximity of TGF-B1 high PDAC align well with EMT ( 61 ), however, tumoural TGF-B1 mRNA counts themselves were not associated with adverse survival. Notably, here, TGF-B transcript counts were not associated with CD8 + T-cell levels. In combined analysis of the tumour regions, TGF-A low (TF)/TGFB2 high (TC) stroma was associated with worse overall survival. Interestingly, in the TCGA data, there were no significant outcome differences in compartment-agnostic tumoral TGF mRNA transcripts (high versus -low, n = 177), further highlighting the pivotal relevance of spatially resolved expression analysis. Recent data suggests negative LN to be a predictor of chemotherapy (gemcitabine plus capecitabine) efficacy ( 62 ). Here, in the TF, a cancer compartment associated with infiltrative properties ( 63 ), elevated tumoural TGF-B2 transcripts correlated with a higher number of lymph node metastases, in line with early reports on the role of TGF-B isoforms on PDAC progression ( 64 ) and bulk mRNA TCGA data. The TF has been characterised as a site of elevated immune interaction and immune cell density ( 41 , 65 ), with upregulation of inflammatory pathways ( 40 , 66 ) and pronounced tumour aggressiveness, for example in the form of tumour budding ( 67 ). Our findings regarding ID1 expression and TGF-B2 transcript counts in the tumour further emphasise its aggressive biology. Although our regression analyses revealed few significant associations (with outliers largely determining the significance of factors such as lymph node metastases and tumour size), we believe that examining larger cohorts would reinforce these findings. For example, neither tumoural nor stromal BMP4 showed any trend in association with the number of lymph node metastases or tumour size. This further highlights the complexity of pathway activity and the importance of possible downstream dysregulation. Higher TGF-B2 mRNA levels have been reported to be associated with worse overall survival when present in a macrophage-poor TME ( 68 ). It appears that TGF-B2 high PDAC and TGF-B2 low stroma define an aggressive, pro-metastatic PDAC. The reported effect of TGF-B2 receptor overexpression on survival in PDAC has been variable ( 69 , 70 ). In mice, TGF-B2 receptor neutralization resulted in increased differentiation and decreased metastasis of PDAC ( 71 ). This suggests that biological behaviour is strongly dictated by regional compartments/functional neighbourhoods rather than mere pathway activation. mRNA transcript counts of all TGF-ligands strongly correlated among tumour and stroma, possibly underscoring excessive fibrogenic signalling in PDAC (aberrant “wound healing” ( 72 )). Interestingly, GREM1, TGF-A and TGF-B2 transcripts were significantly enriched in the juxtatumoural stroma as compared to PDAC parenchyma. The strong stromal predilection of GREM1 is in line with previously reported high GREM1-levels in (cancer-associated) pancreatic fibroblasts ( 28 , 44 ). Corroborating previous results in SMAD4 ( 70 ), SMAD2 protein expression levels were not associated with worse survival. Consistent with proficient downstream BMP-signalling, GREM1 high PDAC showed less ID1 expression. As reported ( 24 ), we could confirm robust ID1 protein expression in PDAC, regardless of tumour region and in line with TCGA data. In our cohort, ID1 high tumours were significantly larger, but there was no further significant association with conventional histopathologic risk predictors or with overall survival. Ultimately, the BMP pathway also leads to cell cycle arrest. ID1 is a downstream protein of this pathway and is, herein, associated with larger tumour size. Consequently, aberrant, context-dependent pathway signalling that bypasses physiological activity (e.g. via upstream molecules) is likely to occur in PDAC. Paradoxically, tumoural ID1 expression was significantly lower in N2- than N1-stage PDAC. ID1 has been shown to exert an immunosuppressive effect by promoting myeloid-derived suppressor cell expansion and downregulating CD8 + T-cells ( 42 ), while the density of CD8 + T-cells in the TC has been demonstrated to have prognostic validity in PDAC ( 73 ). In our cohort, a high CD8 + T-cell infiltrate did not correlate with a higher percentage of tumoural or stromal ID1 expression. Of note, this was independent of the tumour region (TC or TF), despite previous evidence suggesting that there is less CD8 infiltration and more immunosuppression in the TC ( 41 , 65 ). GREM1 protein expression has been described to coincide with M1/M2 macrophage enrichment in PDAC ( 28 , 44 ), a finding we cannot reproduce in our binary stratification using mRNA transcript counts and CD68/CD163 immunohistochemistry. Despite the “different faces” of GREM1 function ( 74 ), loss of GREM1 promotes metastasis ( 26 ) and a fibrogenic stromal microenvironment ( 28 ). In this work, lower stromal and tumoural GREM1 transcript numbers were significantly associated with distant tumour recurrence, supporting the role of GREM1 loss for metastatic seeding. Interestingly, tumour budding counts did not differ significantly between tumour regions or between expression levels for any of the molecules analysed. TGF-B secretion by stromal cells enhances metastatic capacity in colorectal cancer ( 75 ), while in PDAC, it induces a myofibroblastic phenotype of cancer-associated fibroblasts via EGFR/ERBB2 signalling, promoting metastasis ( 76 ). Targeting TGF signalling in PDAC has already shown promising results in combination therapy ( 6 , 8 ). Here, we observed trends toward higher tumour budding in TGF-A low and -B1 low stroma. We found no distant tumour recurrence in TGF-B1 high or -B2 high PDAC. The TGF-B “paradox” ( 77 , 78 ), whereby TGF-B exhibits tumour-suppressive effects in the early stages of PDAC but tumour-promoting effects in the late stages, remains to be deciphered. In our cohort, we observed no significant differences in TGF transcript counts according to T-stage. However, the association of TGF-B2 high stroma with a higher number of lymph node metastases confirms tumour promotion in later stages and the relevance of (context-specific) TGF-B targeting ( 79 ). A spatial shift in TGF signalling, for example from tumour to stroma with hijacked wound healing physiology ( 80 ), might be a mechanistic explanation for the reverse biological effects of TGF-B in late-stage disease. We could not objectify the typical “geometry” of EMT, such as less nuclear roundness, in our cohort. This is partly surprising, as we had expected a more spindle-shaped morphology in more aggressive PDAC, as nuclei become more elongated in mesenchymal-type cells ( 81 , 82 ). Given the recent studies showing pronounced EMT in the TF, we would have expected a more distinct EMT phenotype ( 67 , 83 ). Nevertheless, multiple other factors, e.g. tissue pressure ( 84 ) and cellular migration ( 85 ), can shape nuclear morphology which we do not consider here, as they are not easily quantifiable. Recent evidence supports an "amoeboid" PDAC phenotype within the EMT spectrum, associated with disease progression and enriched at the invasive front, which retains conventional cellular geometry and appears TGF-β-induced ( 86 – 88 ). Contrary to previous works (e.g. ( 31 , 89 )), the proportion of stroma content per tissue core did not correlate with overall survival in our study. The alignment of collagen fibrils, matrix stiffness, the stromal composition ( 90 – 92 ) and the local immune context ( 93 ) might outweigh the mere amount of stroma in terms of negative prognosis ( 94 , 95 ). The stroma itself is dynamic ( 96 ) and susceptible to various microenvironmental cues, and dense desmoplasia alone does not impede T-cell infiltration ( 97 ). Limitations of our study include a relatively small sample size, which is also a limitation of the TCGA validation. In addition, not all patients were represented with three tissue cores, and some cores were lost due to deep TMA sectioning. Additionally, core exclusion was necessary because of the rather low number of tumour cells per core due to infiltrative growth, low tumour cell density, and the strong desmoplastic reaction typical of PDAC. To address discrepancies in expression due to intratumoural heterogeneity, “topographic” TMAs have been proposed ( 98 ). These capture a multitude of biological regions of interest, such as desmoplasia, inflammation and vascular invasion. The uniformly high biological aggressiveness of PDAC, characterized by vascular and perineural invasion in most tumours (e.g., only n = 3 non-neoadjuvant patients without perineural invasion), limited the power of group comparisons. In a few cases, the classifier showed inconsistent performance in identifying single cells and in detecting poorly formed carcinoma glands and areas of tumour budding. Although multiple markers of interest were analysed, each analysis was performed single-plex, using different sections, which affected the spatial analysis. Deeper insights into the true immune-stroma-cancer crosstalk could be achieved by more multiplexing with simultaneous visualization of RNA and proteins, e.g. by immunofluorescence on the same tissue slide. Further validation and extension of our findings would be possible through future studies using spatially resolved (multiplex) technologies at the whole tissue level. In this work, we spatially quantified key TGF- and BMP-signalling molecules in tumour regions and the juxtatumoural stroma. We highlight the importance of stromal signalling as a prognostic factor for overall survival and corroborate ID1 high , TGF-B2 high and GREM1 low tumours as more aggressive. Abbreviations AI Artificial Intelligence BMP Bone Morphogenetic Protein CAF Cancer-associated Fibroblast DIA Digital Image Analysis EGFR Epidermal Growth Factor Receptor EMT Epithelial-mesenchymal Transition GREM1 Gremlin1 ID1 Inhibitor of Differentiation Protein 1 IHC Immunohistochemistry ISH In-situ Hybridisation mRNA messenger RNA PDAC Pancreatic Ductal Adenocarcinoma pSMAD2 phosphoSMAD2 SMAD Suppressor of Mothers against Decapentaplegic TC Tumour Centre TF Tumour Front TGF Transforming Growth Factor TMA Tissue Microarray TME Tumor Microenvironment Declarations Ethics approval and consent to participate The study was approved by the Ethics Committee of Bern (CEC ID2020-00498) and conducted in accordance with the ethical standards of the Declaration of Helsinki. Informed consent to participate was obtained from all study participants. Competing Interests The authors declare no conflict of interest. Funding KB is supported by the Swiss National Science Foundation (P500PM_217647/1). MW is supported by the Foundation for Clinical-Experimental Cancer Research (“longévité”). VHK gratefully acknowledges funding by the Swiss National Science Foundation (P2SKP3_168322/1 and P2SKP3_168322/2), the Werner-Hedy Berger Janser Foundation and the Promedica Foundation. Author Contribution KB generated, analysed, and interpreted the data and drafted the manuscript. VHK and MW jointly designed and supervised the study, and interpreted data. PZ contributed data, JLR put together and characterized the PDAC cohort. SR provided the scoring software, MW scored and SR processed the data. BG and ASW provided surgical specimens and clinical data. SJL provided important resources and intellectual input. All authors revised the manuscript and approved the final version. Acknowledgement KB dedicates his contributions to his late uncle, A Olympos.The authors acknowledge the Translational Research Unit at the Institute of Tissue Medicine and Pathology, University of Bern, Switzerland, for excellent technical support, and Tissuebank Bern for providing human tissue samples. Data Availability Available upon reasonable request from the corresponding author. References Derynck R, Turley SJ, Akhurst RJ. TGFβ biology in cancer progression and immunotherapy. Nat Rev Clin Oncol. 2021;18(1):9–34. Guo X, Wang XF. Signaling cross-talk between TGF-β/BMP and other pathways. Cell Res. 2009;19(1):71–88. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin. 2024;74(1):12–49. David CJ, Huang YH, Chen M, Su J, Zou Y, Bardeesy N, et al. TGF-β Tumor Suppression through a Lethal EMT. Cell. 2016;164(5):1015–30. Hussain SM, Kansal RG, Alvarez MA, Hollingsworth TJ, Elahi A, Miranda-Carboni G, et al. Role of TGF-β in pancreatic ductal adenocarcinoma progression and PD-L1 expression. Cell Oncol. 2021;44(3):673–87. Qiang L, Hoffman MT, Ali LR, Castillo JI, Kageler L, Temesgen A, et al. Transforming Growth Factor-β Blockade in Pancreatic Cancer Enhances Sensitivity to Combination Chemotherapy. Gastroenterology. 2023;165(4):874–e89010. Schlingensiepen KH, Jaschinski F, Lang SA, Moser C, Geissler EK, Schlitt HJ, et al. Transforming growth factor-beta 2 gene silencing with trabedersen (AP 12009) in pancreatic cancer. Cancer Sci. 2011;102(6):1193–200. Ma MJ, Shi YH, Liu ZD, Zhu YQ, Zhao GY, Ye JY, et al. N6-methyladenosine modified TGFB2 triggers lipid metabolism reprogramming to confer pancreatic ductal adenocarcinoma gemcitabine resistance. Oncogene. 2024;43(31):2405–20. Lupo F, Pezzini F, Pasini D, Fiorini E, Adamo A, Veghini L, et al. Axon guidance cue SEMA3A promotes the aggressive phenotype of basal-like PDAC. Gut. 2024;73(8):1321–35. Nieto MA, Huang RYJ, Jackson RA, Thiery JP. EMT: 2016. Cell. 2016;166(1):21–45. Perelli L, Zhang L, Mangiameli S, Giannese F, Mahadevan KK, Peng F et al. Evolutionary fingerprints of epithelial-to-mesenchymal transition. Nature. 2025;1–10. De Smedt L, Palmans S, Andel D, Govaere O, Boeckx B, Smeets D, et al. Expression profiling of budding cells in colorectal cancer reveals an EMT-like phenotype and molecular subtype switching. Br J Cancer. 2017;116(1):58–65. Karamitopoulou E, Zlobec I, Born D, Kondi-Pafiti A, Lykoudis P, Mellou A, et al. Tumour budding is a strong and independent prognostic factor in pancreatic cancer. Eur J Cancer. 2013;49(5):1032–9. Karamitopoulou E, Wartenberg M, Zlobec I, Cibin S, Worni M, Gloor B, et al. Tumour budding in pancreatic cancer revisited: validation of the ITBCC scoring system. Histopathology. 2018;73(1):137–46. Gordon KJ, Kirkbride KC, How T, Blobe GC. Bone morphogenetic proteins induce pancreatic cancer cell invasiveness through a Smad1-dependent mechanism that involves matrix metalloproteinase–2. Carcinogenesis. 2009;30(2):238–48. Bardeesy N, Cheng Khung, Berger JH, Chu GC, Pahler J, Olson P, et al. Smad4 is dispensable for normal pancreas development yet critical in progression and tumor biology of pancreas cancer. Genes Dev. 2006;20(22):3130–46. Wu JW, Fairman R, Penry J, Shi Y. Formation of a Stable Heterodimer between Smad2 and Smad4*. J Biol Chem. 2001;276(23):20688–94. Papageorgis P. TGFβ Signaling in Tumor Initiation, Epithelial-to-Mesenchymal Transition, and Metastasis. J Oncol. 2015;2015(1):587193. Biffi G, Oni TE, Spielman B, Hao Y, Elyada E, Park Y, et al. IL1-Induced JAK/STAT Signaling Is Antagonized by TGFβ to Shape CAF Heterogeneity in Pancreatic Ductal Adenocarcinoma. Cancer Discov. 2019;9(2):282–301. Tahara H, Sato K, Yamazaki Y, Ohyama T, Horiguchi N, Hashizume H, et al. Transforming growth factor-α activates pancreatic stellate cells and may be involved in matrix metalloproteinase–1 upregulation. Lab Invest. 2013;93(6):720–32. Tang Y, Zhang S, Li J, Wu C, Fan Q. ID1 marks the tumorigenesis of pancreatic ductal adenocarcinoma in mouse and human. Sci Rep. 2022;12(1):13555. Raghunathan M, Scully KM, Wehrmaker A, Murad R, Lowy AM, Maurer HC et al. Bone Morphogenetic Protein (BMP) signaling upregulates expression of ID1 and ID3 in pancreatitis and pancreatic ductal adenocarcinoma [Internet]. bioRxiv; 2023 [cited 2024 May 29]. p. 2023.09.01.555987. Available from: https://www.biorxiv.org/content/ 10.1101/2023.09.01.555987v2 Huang YH, Hu J, Chen F, Lecomte N, Basnet H, David CJ, et al. ID1 Mediates Escape from TGFβ Tumor Suppression in Pancreatic Cancer. Cancer Discov. 2020;10(1):142–57. Georgiadou D, Sergentanis TN, Sakellariou S, Filippakis GM, Zagouri F, Vlachodimitropoulos D, et al. VEGF and Id–1 in pancreatic adenocarcinoma: Prognostic significance and impact on angiogenesis. Eur J Surg Oncol. 2014;40(10):1331–7. Shuno Y, Tsuno NH, Okaji Y, Tsuchiya T, Sakurai D, Nishikawa T, et al. Id1/Id3 Knockdown Inhibits Metastatic Potential of Pancreatic Cancer. J Surg Res. 2010;161(1):76–82. Lan L, Evan T, Li H, Hussain A, Ruiz EJ, Zaw Thin M, et al. GREM1 is required to maintain cellular heterogeneity in pancreatic cancer. Nature. 2022;607(7917):163–8. Staloch D, Gao X, Liu K, Xu M, Feng X, Aronson JF, et al. Gremlin is a key pro-fibrogenic factor in chronic pancreatitis. J Mol Med. 2015;93(10):1085–93. Tindall RR, Faraoni EY, Li J, Zhang Y, Ting SM, Okeugo B et al. Increased Gremlin1 Expression in Pancreatic Ductal Adenocarcinoma Promotes a Fibrogenic Stromal Microenvironment. Pancreas. 2024;10.1097/MPA.0000000000002378. Lainé A, Labiad O, Hernandez-Vargas H, This S, Sanlaville A, Léon S, et al. Regulatory T cells promote cancer immune-escape through integrin αvβ8-mediated TGF-β activation. Nat Commun. 2021;12(1):6228. Fu S, Zhang N, Yopp AC, Chen D, Mao M, Chen D, et al. TGF-β Induces Foxp3 + T-Regulatory Cells from CD4 + CD25 – Precursors. Am J Transpl. 2004;4(10):1614–27. Bulle A, Lim KH. Beyond just a tight fortress: contribution of stroma to epithelial-mesenchymal transition in pancreatic cancer. Signal Transduct Target Ther. 2020;5(1):1–12. Pati P, Karkampouna S, Bonollo F, Compérat E, Radić M, Spahn M et al. Accelerating histopathology workflows with generative AI-based virtually multiplexed tumour profiling. Nat Mach Intell. 2024;1–17. 14(th). European congress on digital Pathology. J Pathol Inf. 2019;10:32. Chandrashekar DS, Bashel B, Balasubramanya SAH, Creighton CJ, Ponce-Rodriguez I, Chakravarthi BVSK, et al. UALCAN: A Portal for Facilitating Tumor Subgroup Gene Expression and Survival Analyses. Neoplasia. 2017;19(8):649–58. Armstrong SA, Schultz CW, Azimi-Sadjadi A, Brody JR, Pishvaian MJ. ATM Dysfunction in Pancreatic Adenocarcinoma and Associated Therapeutic Implications. Mol Cancer Ther. 2019;18(11):1899–908. Martino C, Pandya D, Lee R, Levy G, Lo T, Lobo S, et al. ATM-Mutated Pancreatic Cancer: Clinical and Molecular Response to Gemcitabine/Nab-Paclitaxel After Genome-Based Therapy Resistance. Pancreas. 2020;49(1):143. Kondo T, Kanai M, Kou T, Sakuma T, Mochizuki H, Kamada M, et al. Association between homologous recombination repair gene mutations and response to oxaliplatin in pancreatic cancer. Oncotarget. 2018;9(28):19817–25. Russell R, Perkhofer L, Liebau S, Lin Q, Lechel A, Feld FM, et al. Loss of ATM accelerates pancreatic cancer formation and epithelial–mesenchymal transition. Nat Commun. 2015;6(1):7677. Shevach EM, Davidson TS, Huter EN, DiPaolo RA, Andersson J. Role of TGF-β in the Induction of Foxp3 Expression and T Regulatory Cell Function. J Clin Immunol. 2008;28(6):640–6. Karamitopoulou E, Wenning AS, Acharjee A, Aeschbacher P, Marinoni I, Zlobec I et al. Spatial Heterogeneity of Immune Regulators Drives Dynamic Changes in Local Immune Responses, Affecting Disease Outcomes in Pancreatic Cancer. Clin Cancer Res. 2024;OF1–12. Schoumacher C et al. Derangère,Valentin, Gaudillière-Le Dain, Gwladys, Huppe, Titouan, Rageot, David, Ilie, Alis,. CD3-CD8 immune score associated with a clinical score stratifies PDAC prognosis regardless of adjuvant or neoadjuvant chemotherapy. OncoImmunology. 2024;13(1):2294563. Papaspyridonos M, Matei I, Huang Y, do Rosario Andre M, Brazier-Mitouart H, Waite JC, et al. Id1 suppresses anti-tumour immune responses and promotes tumour progression by impairing myeloid cell maturation. Nat Commun. 2015;6(1):6840. Martínez VG, Rubio C, Martínez-Fernández M, Segovia C, López-Calderón F, Garín MI, et al. BMP4 Induces M2 Macrophage Polarization and Favors Tumor Progression in Bladder Cancer. Clin Cancer Res. 2017;23(23):7388–99. Davis JM, Cheng B, Drake MM, Yu Q, Yang B, Li J, et al. Pancreatic stromal Gremlin 1 expression during pancreatic tumorigenesis. Genes Dis. 2022;9(1):108–15. Leggett SE, Sim JY, Rubins JE, Neronha ZJ, Williams EK, Wong IY. Morphological single cell profiling of the epithelial–mesenchymal transition. Integr Biol. 2016;8(11):1133–44. Garcia MA, Yin Z, Roumeliotis TI, Butera F, Wang L, Rickman R et al. Epithelial-Mesenchymal Plasticity is regulated by inflammatory signalling networks coupled to cell morphology [Internet]. bioRxiv; 2022 [cited 2024 May 30]. p. 689737. Available from: https://www.biorxiv.org/content/ 10.1101/689737v3 Frangogiannis NG. Transforming growth factor–β in tissue fibrosis. J Exp Med. 2020;217(3):e20190103. Klein L, Tu M, Krebs N, Urbach L, Grimm D, Latif MU, et al. Spatial tumor immune heterogeneity facilitates subtype co-existence and therapy response in pancreatic cancer. Nat Commun. 2025;16(1):335. Bräutigam K, Skok K, Szymonski K, Rift CV, Karamitopoulou E. Tumor Immune Microenvironment in Pancreatic Ductal Adenocarcinoma revisited - Exploring the Space. Cancer Lett. 2025;217699. Williams HL, Dias Costa A, Zhang J, Raghavan S, Winter PS, Kapner KS, et al. Spatially Resolved Single-Cell Assessment of Pancreatic Cancer Expression Subtypes Reveals Co-expressor Phenotypes and Extensive Intratumoral Heterogeneity. Cancer Res. 2023;83(3):441–55. Grünwald BT, Devisme A, Andrieux G, Vyas F, Aliar K, McCloskey CW, et al. Spatially confined sub-tumor microenvironments in pancreatic cancer. Cell. 2021;184(22):5577–e559218. Khaliq AM, Rajamohan M, Saeed O, Mansouri K, Adil A, Zhang C et al. Spatial transcriptomic analysis of primary and metastatic pancreatic cancers highlights tumor microenvironmental heterogeneity. Nat Genet. 2024;1–11. Raghavan S, Winter PS, Navia AW, Williams HL, DenAdel A, Lowder KE, et al. Microenvironment drives cell state, plasticity, and drug response in pancreatic cancer. Cell. 2021;184(25):6119–e613726. Peng J, Sun BF, Chen CY, Zhou JY, Chen YS, Chen H, et al. Single-cell RNA-seq highlights intra-tumoral heterogeneity and malignant progression in pancreatic ductal adenocarcinoma. Cell Res. 2019;29(9):725–38. Lundgren S, Elebro J, Heby M, Nodin B, Leandersson K, Micke P, et al. Quantitative, qualitative and spatial analysis of lymphocyte infiltration in periampullary and pancreatic adenocarcinoma. Int J Cancer. 2020;146(12):3461–73. Schalck A, Sakellariou-Thompson D, Forget MA, Sei E, Hughes TG, Reuben A, et al. Single-Cell Sequencing Reveals Trajectory of Tumor-Infiltrating Lymphocyte States in Pancreatic Cancer. Cancer Discov. 2022;12(10):2330–49. Wartenberg M, Cibin S, Zlobec I, Vassella E, Eppenberger-Castori S, Terracciano L, et al. Integrated Genomic and Immunophenotypic Classification of Pancreatic Cancer Reveals Three Distinct Subtypes with Prognostic/Predictive Significance. Clin Cancer Res. 2018;24(18):4444–54. Bhatia R, Bhyravbhatla N, Kisling A, Li X, Batra SK, Kumar S. Cytokines chattering in pancreatic ductal adenocarcinoma tumor microenvironment. Semin Cancer Biol. 2022;86:499–510. Suijkerbuijk SJE, van Rheenen J. From good to bad: Intravital imaging of the hijack of physiological processes by cancer cells. Dev Biol. 2017;428(2):328–37. Takimoto T, Wakabayashi Y, Sekiya T, Inoue N, Morita R, Ichiyama K, et al. Smad2 and Smad3 Are Redundantly Essential for the TGF-β–Mediated Regulation of Regulatory T Plasticity and Th1 Development. J Immunol. 2010;185(2):842–55. Wartenberg M, Zlobec I, Perren A, Koelzer VH, Gloor B, Lugli A, et al. Accumulation of FOXP3 + T-cells in the tumor microenvironment is associated with an epithelial-mesenchymal-transition-type tumor budding phenotype and is an independent prognostic factor in surgically resected pancreatic ductal adenocarcinoma. Oncotarget. 2015;6(6):4190–201. Palmer DH, Jackson R, Springfeld C, Ghaneh P, Rawcliffe C, Halloran CM et al. Pancreatic Adenocarcinoma: Long-Term Outcomes of Adjuvant Therapy in the ESPAC4 Phase III Trial. J Clin Oncol. 2024;0(0):JCO.24.01118. Ju Y, Xu D, Liao Mmiao, Sun Y, Bao W, dai, Yao F, et al. Barriers and opportunities in pancreatic cancer immunotherapy. Npj Precis Oncol. 2024;8(1):199. Friess H, Yamanaka Y, Büchler M, Ebert M, Beger HG, Gold LI, et al. Enhanced expression of transforming growth factor β isoforms in pancreatic cancer correlates with decreased survival. Gastroenterology. 1993;105(6):1846–56. Zwart ES, van Ee T, Affandi AJ, Boyd LNC, Rodriguez E, den Haan JMM, et al. Spatial immune composition of tumor microenvironment in patients with pancreatic cancer. Cancer Immunol Immunother. 2023;72(12):4385–97. Karamitopoulou E, Wenning AS, Acharjee A, Zlobec I, Aeschbacher P, Perren A, et al. Spatially restricted tumour-associated and host-associated immune drivers correlate with the recurrence sites of pancreatic cancer. Gut. 2023;72(8):1523–33. Alver TN, Bergholtz H, Holm MB, Dorg LT, Skrede ML, Kure EH, et al. Spatial Transcriptomics Reveals Cancer and Stromal Cell Heterogeneity Between Center and Invasive Front of Pancreatic Cancer. Mod Pathol. 2025;38(6):100726. Qazi S, Chang WH, Lee C, Trieu V. Abstract B070: TGFB2 mRNA levels prognostically interact with Interferon-alpha receptor activation of IRF9 and IFI27, and makers for tumor-associated macrophages impacting overall survival in PDAC. Cancer Res. 2024;84(17Supplement2):B070. Wagner M, Kleeff J, Friess H, Büchler MW, Korc M. Enhanced Expression of the Type II Transforming Growth Factor-β Receptor Is Associated with Decreased Survival in Human Pancreatic Cancer. Pancreas. 1999;19(4):370. Javle M, Li Y, Tan D, Dong X, Chang P, Kar S, et al. Biomarkers of TGF-β Signaling Pathway and Prognosis of Pancreatic Cancer. PLoS ONE. 2014;9(1):e85942. Ostapoff KT, Cenik BK, Wang M, Ye R, Xu X, Nugent D, et al. Neutralizing Murine TGFβR2 Promotes a Differentiated Tumor Cell Phenotype and Inhibits Pancreatic Cancer Metastasis. Cancer Res. 2014;74(18):4996–5007. Deyell M, Garris CS, Laughney AM. Cancer metastasis as a non-healing wound. Br J Cancer. 2021;124(9):1491–502. Masugi Y, Abe T, Ueno A, Fujii-Nishimura Y, Ojima H, Endo Y, et al. Characterization of spatial distribution of tumor-infiltrating CD8 + T cells refines their prognostic utility for pancreatic cancer survival. Mod Pathol. 2019;32(10):1495–507. Moustakas A, Löhr JM, Heuchel RL. Cellular heterogeneity in pancreatic cancer: the different faces of gremlin action. Signal Transduct Target Ther. 2022;7(1):1–3. Calon A, Espinet E, Palomo-Ponce S, Tauriello DVF, Iglesias M, Céspedes MV, et al. Dependency of Colorectal Cancer on a TGF-β-Driven Program in Stromal Cells for Metastasis Initiation. Cancer Cell. 2012;22(5):571–84. Mucciolo G, Henríquez JA, Jihad M, Teles SP, Manansala JS, Li W, et al. EGFR-activated myofibroblasts promote metastasis of pancreatic cancer. Cancer Cell. 2024;42(1):101–e11811. Shen W, Tao G, qing, Zhang Y, Cai B, Sun J, Tian Z. qiang. TGF-β in pancreatic cancer initiation and progression: two sides of the same coin. Cell Biosci. 2017;7(1):39. Principe DR, Doll JA, Bauer J, Jung B, Munshi HG, Bartholin L, et al. TGF-β: Duality of Function Between Tumor Prevention and Carcinogenesis. JNCI J Natl Cancer Inst. 2014;106(2):djt369. Hosein AN, Dougan SK, Aguirre AJ, Maitra A. Translational advances in pancreatic ductal adenocarcinoma therapy. Nat Cancer. 2022;3(3):272–86. Deng Z, Fan T, Xiao C, Tian H, Zheng Y, Li C, et al. TGF-β signaling in health, disease and therapeutics. Signal Transduct Target Ther. 2024;9(1):1–40. Walters B, Uynuk-Ool T, Rothdiener M, Palm J, Hart ML, Stegemann JP, et al. Engineering the geometrical shape of mesenchymal stromal cells through defined cyclic stretch regimens. Sci Rep. 2017;7(1):6640. Leggett SE, Hruska AM, Guo M, Wong IY. The epithelial-mesenchymal transition and the cytoskeleton in bioengineered systems. Cell Commun Signal. 2021;19(1):32. Maiques O, Sallan MC, Laddach R, Pandya P, Varela A, Crosas-Molist E, et al. Matrix mechano-sensing at the invasive front induces a cytoskeletal and transcriptional memory supporting metastasis. Nat Commun. 2025;16(1):1394. Villeneuve C, Hashmi A, Ylivinkka I, Lawson-Keister E, Miroshnikova YA, Pérez-González C, et al. Mechanical forces across compartments coordinate cell shape and fate transitions to generate tissue architecture. Nat Cell Biol. 2024;26(2):207–18. Stöberl S, Flommersfeld J, Kreft MM, Benoit M, Broedersz CP, Rädler JO. Nuclear deformation and dynamics of migrating cells in 3D confinement reveal adaptation of pulling and pushing forces. Sci Adv. 2024;10(34):eadm9195. Georgouli M, Herraiz C, Crosas-Molist E, Fanshawe B, Maiques O, Perdrix A, et al. Regional Activation of Myosin II in Cancer Cells Drives Tumor Progression via a Secretory Cross-Talk with the Immune Microenvironment. Cell. 2019;176(4):757–e77423. Samain R, Maiques O, Monger J, Lam H, Candido J, George S, et al. CD73 controls Myosin II–driven invasion, metastasis, and immunosuppression in amoeboid pancreatic cancer cells. Sci Adv. 2023;9(42):eadi0244. Graziani V, Rodriguez-Hernandez I, Maiques O, Sanz-Moreno V. The amoeboid state as part of the epithelial-to-mesenchymal transition programme. Trends Cell Biol. 2022;32(3):228–42. Rhim AD, Oberstein PE, Thomas DH, Mirek ET, Palermo CF, Sastra SA, et al. Stromal Elements Act to Restrain, Rather Than Support, Pancreatic Ductal Adenocarcinoma. Cancer Cell. 2014;25(6):735–47. Shields MA, Dangi-Garimella S, Redig AJ, Munshi HG. Biochemical role of the collagen-rich tumour microenvironment in pancreatic cancer progression. Biochem J. 2011;441(2):541–52. Mascharak S, Guo JL, Foster DS, Khan A, Davitt MF, Nguyen AT, et al. Desmoplastic stromal signatures predict patient outcomes in pancreatic ductal adenocarcinoma. Cell Rep Med. 2023;4(11):101248. Ogawa Y, Masugi Y, Abe T, Yamazaki K, Ueno A, Fujii-Nishimura Y, et al. Three Distinct Stroma Types in Human Pancreatic Cancer Identified by Image Analysis of Fibroblast Subpopulations and Collagen. Clin Cancer Res. 2021;27(1):107–19. Tan X, Rosin M, Appinger S, Deierl JC, Reichel K, Coolsen M, et al. Stroma and lymphocytes identified by deep learning are independent predictors for survival in pancreatic cancer. Sci Rep. 2025;15(1):9415. Bolm L, Zghurskyi P, Lapshyn H, Petrova E, Zemskov S, Vashist YK, et al. Alignment of stroma fibers, microvessel density and immune cell populations determine overall survival in pancreatic cancer—An analysis of stromal morphology. PLoS ONE. 2020;15(7):e0234568. Drifka CR, Loeffler AG, Mathewson K, Keikhosravi A, Eickhoff JC, Liu Y, et al. Highly aligned stromal collagen is a negative prognostic factor following pancreatic ductal adenocarcinoma resection. Oncotarget. 2016;7(46):76197–213. Erkan M, Michalski CW, Rieder S, Reiser–Erkan C, Abiatari I, Kolb A, et al. The Activated Stroma Index Is a Novel and Independent Prognostic Marker in Pancreatic Ductal Adenocarcinoma. Clin Gastroenterol Hepatol. 2008;6(10):1155–61. Carstens JL, Correa de Sampaio P, Yang D, Barua S, Wang H, Rao A, et al. Spatial computation of intratumoral T cells correlates with survival of patients with pancreatic cancer. Nat Commun. 2017;8(1):15095. Bingham V, Harewood L, McQuaid S, Craig SG, Revolta JF, Kim CS, et al. Topographic analysis of pancreatic cancer by TMA and digital spatial profiling reveals biological complexity with potential therapeutic implications. Sci Rep. 2024;14(1):11361. Supplementary Files Fig.S1.AllMarkers.pdf Fig.S2.Aggression.pdf Fig.S3.Regression.pdf Fig.S4.Budding1.pdf Fig.S5.Immune.pdf Fig.S6.Survivalother.tif Fig.S7.Roundness1.pdf Fig.S8TCGA.pdf SupplementaryTables.docx SupplementaryFiguresLegends.docx Cite Share Download PDF Status: Published Journal Publication published 18 Aug, 2025 Read the published version in BMC Cancer → Version 1 posted Editorial decision: Revision requested 16 Jul, 2025 Reviews received at journal 15 Jul, 2025 Reviewers agreed at journal 10 Jul, 2025 Reviews received at journal 10 Jul, 2025 Reviewers agreed at journal 09 Jul, 2025 Reviewers invited by journal 08 Jul, 2025 Submission checks completed at journal 18 Jun, 2025 First submitted to journal 18 Jun, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6506146","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":477470298,"identity":"93cbbdab-a6a6-49fb-87aa-154e44e12f9c","order_by":0,"name":"Konstantin Bräutigam","email":"data:image/png;base64,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","orcid":"","institution":"Centre for Evolution and Cancer, Institute of Cancer Research, London, UK","correspondingAuthor":true,"prefix":"","firstName":"Konstantin","middleName":"","lastName":"Bräutigam","suffix":""},{"id":477470299,"identity":"139b3818-0538-4add-86b1-1c1c6f2e23f4","order_by":1,"name":"Philipp Zens","email":"","orcid":"","institution":"Institute of Tissue Medicine and Pathology, University of Bern, Bern, Switzerland","correspondingAuthor":false,"prefix":"","firstName":"Philipp","middleName":"","lastName":"Zens","suffix":""},{"id":477470300,"identity":"32a7df6b-04d4-4225-ba9c-e535a63cfbd2","order_by":2,"name":"Stefan Reinhard","email":"","orcid":"","institution":"Institute of Tissue Medicine and Pathology, University of Bern, Bern, Switzerland","correspondingAuthor":false,"prefix":"","firstName":"Stefan","middleName":"","lastName":"Reinhard","suffix":""},{"id":477470301,"identity":"3343af30-9709-42c2-b858-bc1b17fa2334","order_by":3,"name":"Jessica L Rohrbach","email":"","orcid":"","institution":"Institute of Tissue Medicine and Pathology, University of Bern, Bern, Switzerland","correspondingAuthor":false,"prefix":"","firstName":"Jessica","middleName":"L","lastName":"Rohrbach","suffix":""},{"id":477470302,"identity":"411617c2-210f-41ba-956e-fd2da39a676f","order_by":4,"name":"Simon J Leedham","email":"","orcid":"","institution":"Intestinal Stem Cell Biology Lab, Wellcome Centre for Human Genetics, University of Oxford, Oxford, UK;","correspondingAuthor":false,"prefix":"","firstName":"Simon","middleName":"J","lastName":"Leedham","suffix":""},{"id":477470303,"identity":"45def268-b462-4703-bb77-8b0567a4e8e9","order_by":5,"name":"Anna S Wenning","email":"","orcid":"","institution":"Department of Visceral Surgery and Medicine, Inselspital Bern, Bern University Hospital, Bern, University of Bern, Bern, Switzerland","correspondingAuthor":false,"prefix":"","firstName":"Anna","middleName":"S","lastName":"Wenning","suffix":""},{"id":477470304,"identity":"d4f07fe3-5dec-49d3-86f3-d24a288b0223","order_by":6,"name":"Beat Gloor","email":"","orcid":"","institution":"Department of Visceral Surgery and Medicine, Inselspital Bern, Bern University Hospital, Bern, University of Bern, Bern, Switzerland","correspondingAuthor":false,"prefix":"","firstName":"Beat","middleName":"","lastName":"Gloor","suffix":""},{"id":477470305,"identity":"d40e5d06-e3c7-41a2-9988-8ab0d4c7fa8b","order_by":7,"name":"Viktor H Koelzer","email":"","orcid":"","institution":"Department of Pathology and Molecular Pathology, University Hospital Zurich, University of Zurich, Zurich, Switzerland","correspondingAuthor":false,"prefix":"","firstName":"Viktor","middleName":"H","lastName":"Koelzer","suffix":""},{"id":477470306,"identity":"e1a3064e-f45b-41e9-8033-f7bb5104a52e","order_by":8,"name":"Martin Wartenberg","email":"","orcid":"","institution":"Institute of Tissue Medicine and Pathology, University of Bern, Bern, Switzerland","correspondingAuthor":false,"prefix":"","firstName":"Martin","middleName":"","lastName":"Wartenberg","suffix":""}],"badges":[],"createdAt":"2025-04-22 16:23:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6506146/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6506146/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12885-025-14751-3","type":"published","date":"2025-08-18T15:57:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":85620298,"identity":"fce26a78-fa6c-44cf-b589-5c6c278f53e8","added_by":"auto","created_at":"2025-06-29 15:11:18","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":4155369,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTransforming Growth Factor (TGF)-A, -B and Bone Morphogenetic Protein (BMP) signalling. A\u003c/strong\u003e TGF-A acts as a ligand for the epidermal growth factor receptor (EGFR), leading to receptor dimerization and autophosphorylation, which activates key signalling pathways such as JAK-STAT, MAPK, and PI3K-AKT. This activation promotes essential cellular processes such as (cell) proliferation and angiogenesis. \u003cstrong\u003eB\u003c/strong\u003e SARA (Smad Anchor for Receptor Activation) facilitates the recruitment of SMAD2/3 to the TGF-B receptor complex. Upon ligand binding, SMAD2/3 becomes phosphorylated (\"P\"). Once activated, phosphorylated SMAD2/3 dissociates from the receptor and forms a heteromeric complex with SMAD4, which translocates to the nucleus to activate transcriptional programs that regulate processes such as epithelial-mesenchymal transition (EMT), apoptosis, and cell cycle arrest, while also inducing the expression of DNA-binding protein inhibitor (ID1) protein. In BMP signalling, ligands such as BMP4 bind to their respective receptors, leading to the phosphorylation of SMAD1/5/8, which then associates with SMAD4 to form a transcriptionally active complex that also translocates to the nucleus. SMAD4 serves as a common mediator for both the TGF-B and BMP signalling pathways. Gremlin1 (GREM1) functions as a BMP antagonist, binding to BMPs and inhibiting their interaction with BMP receptors, thereby modulating BMP signalling during development and maintenance of tissue homeostasis. Created in\u003cem\u003e Biorender.com\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Fig.1.TGFBMPSignaling.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/c30007840c3cedd386fa8b2c.jpeg"},{"id":85619814,"identity":"0036bac5-ab89-4846-af88-fd3cbae5bac8","added_by":"auto","created_at":"2025-06-29 15:03:17","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":6724712,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDigital Image Analysis (DIA) and examples of expression of key markers across spatial compartments\u003c/strong\u003e. \u003cstrong\u003eA, B \u003c/strong\u003eImmunohistochemistry (IHC): High ID1 protein expression in PDAC parenchyma.\u003cstrong\u003e \u003c/strong\u003emRNA in-situ hybridisation (ISH): Strong stromal/juxtatumoural GREM1 signalling with high mRNA transcript counts (inset). Extended in \u003cstrong\u003eFigure S1\u003c/strong\u003e. \u003cem\u003eBlack\u003c/em\u003e and \u003cem\u003epurple: \u003c/em\u003estroma,\u003cem\u003e Blue: \u003c/em\u003eartifact,\u003cem\u003e Red: \u003c/em\u003ePDAC parenchyma, \u003cem\u003eYellow:\u003c/em\u003enecrosis. \u003cem\u003eScale bar\u003c/em\u003e in panel A: 200μm. \u003cstrong\u003eC, left\u003c/strong\u003e TGF-A signalling in PDAC epithelium (arrow, asterisk) and stroma (arrowhead) with significantly higher mRNA copies per cell in tumour stroma. \u003cstrong\u003eC, right\u003c/strong\u003e, example of TGF-A mRNA ISH in the PDAC TME \u003cstrong\u003eD, left \u003c/strong\u003eSignificantly stronger TGF-B2 signalling in tumour stroma (arrowheads) compared to PDAC epithelium; \u003cstrong\u003eD, right \u003c/strong\u003eexample of TGF-B2 mRNA ISH in the PDAC TME (arrows). \u003cem\u003eTME\u003c/em\u003e: Tumour Microenvironment.\u003c/p\u003e","description":"","filename":"Fig.2.DIAKeyMarkers.png","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/59c2d0c74ef9da82aedc1212.png"},{"id":85619825,"identity":"d8fce24d-a1cb-4cfc-9ac5-b889ee9f3fab","added_by":"auto","created_at":"2025-06-29 15:03:18","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1036168,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eInter- and intra-compartment correlation. \u003c/strong\u003eMarker expression (neoadjuvant cases excluded) in PDAC parenchyma (\u003cstrong\u003eA\u003c/strong\u003e), juxtatumoural stroma (\u003cstrong\u003eB\u003c/strong\u003e) and tumour parenchyma versus stroma (\u003cstrong\u003eC\u003c/strong\u003e). Scale indicates Spearman-Rho correlation coefficient and direction of association.\u003c/p\u003e","description":"","filename":"Fig.3A.png","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/d6c21035799ff47af87e4bc8.png"},{"id":85619854,"identity":"ce9f825b-243d-4461-901b-551cc2af6617","added_by":"auto","created_at":"2025-06-29 15:03:19","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":947533,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHigher tumour aggression in ID1\u003c/strong\u003e\u003csup\u003e\u003cstrong\u003ehigh\u003c/strong\u003e\u003c/sup\u003e\u003cstrong\u003e, TGF-B2\u003c/strong\u003e\u003csup\u003e\u003cstrong\u003ehigh\u003c/strong\u003e\u003c/sup\u003e\u003cstrong\u003e and GREM1\u003c/strong\u003e\u003csup\u003e\u003cstrong\u003elow\u003c/strong\u003e\u003c/sup\u003e\u003cstrong\u003e PDAC.\u003c/strong\u003e \u003cstrong\u003eA\u003c/strong\u003e Less perineural invasion and (\u003cstrong\u003eB\u003c/strong\u003e) less (micro- (V1) and macroscopic (V2)) blood vessel infiltration in GREM1\u003csup\u003elow\u003c/sup\u003e PDAC (TC). \u003cstrong\u003eC \u003c/strong\u003eHigher ID1 protein expression (in %) in lymph-node positive PDAC (\u003cstrong\u003eA-C\u003c/strong\u003e: neoadjuvant cases excluded).\u003c/p\u003e","description":"","filename":"Fig.4.Tumoraggression.png","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/5b1167cb0e803a64cad1325c.png"},{"id":85619860,"identity":"dc87fb32-9489-4512-8df3-7128c6022d76","added_by":"auto","created_at":"2025-06-29 15:03:19","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":693170,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStromal TGF-A and -B2 define subgroups of worse survival in PDAC.\u003c/strong\u003e \u003cstrong\u003eA \u003c/strong\u003e\u0026nbsp;TGF-B2\u003csup\u003elow\u003c/sup\u003e (TC) stroma and (\u003cstrong\u003eB\u003c/strong\u003e) TGF-A\u003csup\u003elow\u003c/sup\u003e(TF)-TGF-B2\u003csup\u003ehigh\u003c/sup\u003e (TC) stroma significantly associated with worse survival (low vs. high stratified by mean, neoadjuvant cases excluded, other outcome analyses in \u003cstrong\u003eFigure S6B and S6C\u003c/strong\u003e).\u003c/p\u003e","description":"","filename":"Fig.5A.png","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/8f8f2a92be5005312ae4f44a.png"},{"id":89847121,"identity":"9bdf1ae4-2860-49b7-b6b6-0b4590fa0457","added_by":"auto","created_at":"2025-08-25 16:40:39","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":15122348,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/a852cead-5c3a-44f1-b81c-b41fb7efec10.pdf"},{"id":85619852,"identity":"db0cacbd-c9eb-41e1-b0bc-de12d4e7cf91","added_by":"auto","created_at":"2025-06-29 15:03:19","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":8469434,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S1.AllMarkers.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/a2b9aa3da438293248acceaa.pdf"},{"id":85619846,"identity":"c5ce1fa8-7371-46b6-b1f7-e639092fb703","added_by":"auto","created_at":"2025-06-29 15:03:19","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":139901,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S2.Aggression.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/d1f459ed0f90abd92a95f777.pdf"},{"id":85619817,"identity":"9d35c6aa-fc51-4178-9a22-d0820ee487d4","added_by":"auto","created_at":"2025-06-29 15:03:17","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":609012,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S3.Regression.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/bd9726912e0fd863c22d9998.pdf"},{"id":85620295,"identity":"0a9b1139-fa53-4f34-92c8-3447ad6c9825","added_by":"auto","created_at":"2025-06-29 15:11:17","extension":"pdf","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":1292811,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S4.Budding1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/18bc51424bd15d55757d8068.pdf"},{"id":85619862,"identity":"342054a7-9f20-4dbf-906a-d41cf82e3d25","added_by":"auto","created_at":"2025-06-29 15:03:20","extension":"pdf","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":2393282,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S5.Immune.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/ab68998e5ec38cae7a7334c4.pdf"},{"id":85619833,"identity":"c89fca85-7a48-4eed-ad6b-face88c694f6","added_by":"auto","created_at":"2025-06-29 15:03:18","extension":"tif","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":2807428,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S6.Survivalother.tif","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/b24f24697cf8a34b00b4ef3e.tif"},{"id":85620394,"identity":"f56a1e1d-0b51-458a-b745-223bf4035edf","added_by":"auto","created_at":"2025-06-29 15:19:19","extension":"pdf","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":1073038,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S7.Roundness1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/ac1abc688012d68bb3fa73a4.pdf"},{"id":85619818,"identity":"80ad655e-195c-4c77-bf33-4c9c63680568","added_by":"auto","created_at":"2025-06-29 15:03:17","extension":"pdf","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":1092949,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S8TCGA.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/6a4f97824fcbd4bcefb40f67.pdf"},{"id":85620296,"identity":"54d2cdc1-68b8-44f7-a4ec-3717dceff19f","added_by":"auto","created_at":"2025-06-29 15:11:18","extension":"docx","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":34197,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTables.docx","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/3ba0d8b9ae1f299022e17744.docx"},{"id":85619849,"identity":"1271abbb-20de-4b46-ad75-475c80ec534f","added_by":"auto","created_at":"2025-06-29 15:03:19","extension":"docx","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":15508,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFiguresLegends.docx","url":"https://assets-eu.researchsquare.com/files/rs-6506146/v1/db9bd3668e8b855ef53e848d.docx"}],"financialInterests":"","formattedTitle":"Spatially resolved analysis of TGF/BMP signalling in pancreatic ductal adenocarcinoma by digital pathology identifies patient subgroups with adverse outcome","fulltext":[{"header":"Introduction","content":"\u003cp\u003eTransforming Growth Factor (TGF) and Bone Morphogenetic Protein (BMP) signalling play a crucial role in cancer development and progression (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). In pancreatic ductal adenocarcinoma (PDAC), one of the most lethal cancer types with increasing incidence (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e), TGF-B is known to play a paradoxical role, i.e., it can be tumour-promoting (especially in late-stage, metastatic PDAC) or -suppressive (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Promisingly, blocking TGF-B1/B2 enhances sensitivity to combination chemotherapy in PDAC (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e), and silencing TGF-B2 expression inhibits tumour proliferation in PDAC cell lines (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e) and can boost gemcitabine chemosensitivity (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTGF-B is a well-known inducer of epithelial-mesenchymal transition (EMT) (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e), a process that alters cytoskeletal structure and cellular motility, often enhancing the metastatic potential of cancer cells, and driving a more aggressive phenotype (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e) with nuclear abnormalities and chromatin remodelling (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). High-grade tumour budding, as a morphological \u0026ldquo;correlate\u0026rdquo; of EMT (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e), has been shown to be a strong predictor of worse outcome in PDAC (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). Despite the established roles of BMPs and Suppressor of Mothers against Decapentaplegic (SMAD) proteins in regulating EMT and invasiveness, their specific influence on tumour budding in PDAC remains poorly understood (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). The tumour-promoting effect of TGF-B is pronounced in the context of \u003cem\u003eSMAD4\u003c/em\u003e loss (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e), a gene involved in pancreatogenesis and frequently inactivated in PDAC (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). SMAD2, a downstream intracellular mediator activated by TGF-B, forms a heterodimeric complex with SMAD4 (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). After nuclear translocation, the SMAD2/SMAD4 complex controls gene expression of various processes, such as apoptosis, cell cycle arrest or EMT (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). TGF-B also has a profound effect on the tumour microenvironment (TME) in PDAC, e.g., by inducing heterogeneity of cancer-associated fibroblasts (CAFs) (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). TGF-A signalling through the epidermal growth factor receptor (EGFR) has been shown to activate pancreatic stellate cells and may contribute to chronic pancreatitis and pancreatic carcinogenesis (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eCrosstalk between the TGF-B and BMP pathways exists: Upon binding to BMP receptors, BMPs first activate SMAD1/5/8 and then SMAD4 as the common downstream protein in both pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). BMP signalling induces the expression of Inhibitor of DNA binding 1 (ID1), which is known to be upregulated in pancreatitis and PDAC (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e) and is believed to have tumour-promoting effects by enabling bypassing of TGF-B signalling (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Protein overexpression of ID1 has been described as a negative prognostic predictor associated with higher tumour (T)- and nodal (N)-stage as well as higher microvessel density in PDAC (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). Interestingly, ID1/ID3 knockdown seems to inhibit PDAC metastasis in human pancreatic cancer cells (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). Gremlin1 (GREM1) is a BMP antagonist with regulatory function on cellular heterogeneity in PDAC (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e). Its inactivation promotes EMT and lower GREM1 levels are associated with late-stage disease. Recent evidence suggests that GREM1 also influences the TME by driving a fibrogenic activated stroma (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e) with higher macrophage counts (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). An additional microenvironmental factor is TGF-B-induced immune escape, e.g., by recruiting FOXP3\u003csup\u003e+\u003c/sup\u003e T-cells (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e), and thereby turning the tumour into a \u0026ldquo;cold cancer\u0026rdquo; with worse outcome. The creation of an immunosuppressive, tumour-permissive (modulatory) niche is supported by the typical context of (excessive) desmoplasia in PDAC, analogous to wound healing (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn this work, we spatially resolve and comprehensively quantify key signalling molecules of both, TGF and BMP, pathways in PDAC and the immediate environment of invading cancer cells using digital pathology and a combined protein and mRNA expression profiling approach.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003ePatient cohort and PDAC tissue microarray (TMA)\u003c/h2\u003e \u003cp\u003eOur tissue microarray (TMA) contains 117 cases of curatively resected PDAC treated at the Department of Visceral Surgery of Inselspital Bern, including thirteen with neoadjuvant treatment (cohort details in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, \u003cb\u003eTable \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e and in (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e)), and a mean (median; range) follow-up of 794.6 (567; 3 to 3000) days. To account for tumor heterogeneity, each patient case is represented by three tissue cores selected by an expert pathologist (MW) on morphologic grounds from different tumour regions, core diameter 0.6 mm/area 0.28 mm\u003csup\u003e2\u003c/sup\u003e: Tumour front (TF) to account for the infiltrative capacity of the cancer, i.e., the outermost tumour periphery, which is further characterised by histological hallmarks of invasion, e.g. tumour buds and single cell growth; Tumour centre (TC) sampled from the central tumour mass; stroma(-predominant) (in total n\u0026thinsp;=\u0026thinsp;351 tissue cores).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eClinico-pathological details of the PDAC cohort (n\u0026thinsp;=\u0026thinsp;104) without neoadjuvant treatment (neoadjuvant in \u003cb\u003eTable \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e). \u003csup\u003ea\u003c/sup\u003eThe Grade 4 (G4) tumour represents an undifferentiated carcinoma, anaplastic type. \u003cem\u003eADM\u003c/em\u003e: Acinar-to-Ductal Metaplasia; \u003cem\u003eBD\u003c/em\u003e: Tumour Budding; \u003cem\u003eITBCC\u003c/em\u003e: International Tumour Budding Consensus Conference; \u003cem\u003eIPMN\u003c/em\u003e: Intraductal Papillary Mucinous Neoplasm; \u003cem\u003eNA\u003c/em\u003e: not available; \u003cem\u003ePanIN\u003c/em\u003e: Pancreatic Intraepithelial Neoplasia; \u003cem\u003eSD\u003c/em\u003e: Standard Deviation. \u003cem\u003eUICC\u003c/em\u003e: Union Internationale Contre le Cancer\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN\u0026thinsp;=\u0026thinsp;104\u003csup\u003e1\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (years)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e68.6 (SD 9.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e45 / 104 (43.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e59 / 104 (56.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeceased\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e78 / 103 (75.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOverall Survival (days)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e801.3 (SD 674.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType of resection\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLeft resection\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11 / 104 (10.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLeft resection and adrenalectomy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 / 104 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLeft resection, colectomy and adrenalectomy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 / 104 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal pancreatectomy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e16 / 104 (15.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhipple procedure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e75 / 104 (72.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSize invasive tumour (mm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e35.3 (SD 11.9, range 12 to 95)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntrapancreatic location invasive tumour\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHead\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e81 / 104 (77.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHead and body\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 / 104 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHead, body and tail\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 / 104 (1.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTail\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8 / 104 (7.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBody\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 / 104 (6.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBody and tail\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5 / 104 (4.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eT-Stage (UICC 8th)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1c\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5 / 104 (4.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e69 / 104 (66.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30 / 104 (28.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eN-Stage (UICC 8th)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19 / 104 (18.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e42 / 104 (40.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e43 / 104 (41.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eM-Stage (UICC 8th)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e103 / 104 (99.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 / 104 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLymphatic vessel infiltration (L1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e82 / 104 (78.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlood vessel infiltration (V)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12 / 104 (11.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e89 / 104 (85.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3 / 104 (2.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerineural infiltration (Pn1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e101 / 104 (97.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResection status (R)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e64 / 104 (61.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31 / 104 (29.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 / 104 (1.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 / 104 (6.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrading (G)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12 / 104 (11.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e51 / 104 (49.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40 / 104 (38.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 / 104 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumour Budding (ITBCC)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBD0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 / 104 (1.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBD1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e29 / 104 (27.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBD2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31 / 104 (29.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBD3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e42 / 104 (40.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal Lymph Node (LN) yield\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30.0 (12.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePositive LN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.6 (SD 5.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresence of Extracapsular Extension\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e46 / 104 (44.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003e\u003csup\u003e1\u003c/sup\u003eMean (SD); n / N (%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eImmunohistochemistry (IHC)\u003c/h3\u003e\n\u003cp\u003ePrimary antibodies were incubated for 30 min at room temperature and used as follows: Anti-Id1 (ab203202, rabbit polyclonal, Abcam, Cambridge, United Kingdom, dilution: 1:200, retrieval: Citrate buffer pH 6, 20 min); CD8 (M710301, mouse monoclonal, clone C8/144B, DAKO/Agilent Technologies, CA, United States, dilution 1:200, retrieval: TrisEDTA, 20 min); CD68 (M087601, clone PG-M1, monoclonal mouse, DAKO/Agilent Technologies, dilution: 1:200,Tris EDTA, 30 min); CD163 (NCL-CD163, clone 10D6, mouse monoclonal, Leica Biosystems, Newcastle upon Tyne, United Kingdom, dilution: 1:400, retrieval: Tris EDTA, 20 min); FOXP3 (14-477-82, mouse monoclonal, clone 236A/E7, Invitrogen, MA, United States, dilution: 1:50, retrieval: Citrate buffer pH 6, 30 min); pSMAD2 (3108S, rabbit monoclonal, Cell Signaling Technology, London, United Kingdom, dilution: 1:200, retrieval: Citrate buffer pH 6, 20 min). Antibody detection was performed with the BOND Polymer Refine DAB kit (Leica Biosystems) using 3,3-diaminobenzidine as a brown chromogen. The samples were counterstained with hematoxylin. TMA slides were scanned on a 3DHISTECH Pannoramic 250 Flash II scanner (3DHISTECH, Budapest, Hungary). Protein expression was rigorously validated by three board-certified pathologists (KB, VHK and MW) and by including appropriate on-slide controls.\u003c/p\u003e\n\u003ch3\u003eMessenger RNA (mRNA) in-situ hybridisation (ISH)\u003c/h3\u003e\n\u003cp\u003emRNA transcripts of TGF-A (reg 58, reference 313138 Hs-TGFA), TGF-B1 (reg 52, reference 400888 Hs-TGFB1), TGF-B2 (reg 54, reference 489248 Hs-TGFB2), BMP ligand BMP4 (reg 56, reference 454308 Hs-BMP4), and BMP antagonist GREM1 (reg 46, reference 312831 Hs-GREM1) were visualized by applying RNAscope (2.5 LS DUPLEX, Reagent Kit-BROWN) chromogenic probes (ACD BioTechne, Newark, CA, USA) on the TMA sections described above using TrisEDTA epitope retrieval buffer (95\u0026deg;, 20 min) and ACD enzyme for 15 min, together with a negative control probe (2.5 LS Duplex \u0026ldquo;DapB\u0026rdquo;).\u003c/p\u003e\n\u003ch3\u003eDigital Image Analysis (DIA)\u003c/h3\u003e\n\u003cp\u003eDIA was performed using HALO AI\u0026trade; (Indica Labs, NM, USA). The digitized TMA slides were de-arrayed into individual spot images of each tissue sample, which were then linked to the corresponding clinical annotations. After the first visual review, seven tissue cores with insufficient tissue and/or artifacts were excluded from further analysis. To localize and quantify tumour and stromal tissue, a deep neural network algorithm was trained based on a pre-trained DenseNet network (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Graphical overlays were generated for each tissue class and the classification accuracy was visually verified. The total area of each tissue class was quantified in square millimetres (mm\u0026sup2;). Nuclear segmentation and staining quantification were performed in both the tumour and stromal compartments. For nuclear segmentation, the HALO\u0026trade; AI pre-trained convolutional neural network was used and fine-tuned with application-specific training examples. Pixels were classified as positive if the staining intensity exceeded the internal controls, as validated by pathologist review (KB, VHK). The total area of vital tissue in the tumour and stromal compartments, the number of ID1-positive and pSMAD2-positive cells (\u003cem\u003eCytonuclear\u003c/em\u003e v2.0.9 module), and mRNA transcripts of TGFA, TGFB1, TGFB2, BMP4 and GREM1 (\u003cem\u003eISH IHC\u003c/em\u003e v3.2.5 module) were recorded for analysis with clinicopathologic features. Marker quantification subproperties such as cytoplasmic radius, contrast thresholds, optical density, segmentation aggressiveness, and nuclear staining intensity were optimized based on expert pathologist review. Nuclear roundness (from 0 to 1, with 1 being perfectly round) and cytoplasmic radius were measured by HALO\u0026trade; Image Analysis Software.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eTumour budding and immune cell quantification\u003c/h3\u003e\n\u003cp\u003eThe grade of tumour budding was assigned by consensus (BD0 to BD3) (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e) for each tumour using representative hematoxylin and eosin (H\u0026amp;E) slides. In addition, tumour budding was individually counted at the tissue core level for each individual marker by an expert pathologist (MW) in a blinded, independent and randomized manner using the browser-based online TMA analysis tool \"Scorenado\" (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). CD8\u003csup\u003e+\u003c/sup\u003e T-cells, FOXP3\u003csup\u003e+\u003c/sup\u003e T-regulatory cells and macrophages (CD68\u003csup\u003e+\u003c/sup\u003e and CD163\u003csup\u003e+\u003c/sup\u003e) have been binarily classified into \u0026ldquo;low\u0026rdquo; and \u0026ldquo;high\u0026rdquo; per tissue core (MW).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eValidation using The Cancer Genome Atlas (TCGA)\u003c/h2\u003e \u003cp\u003eThe findings were independently validated using open-source, publicly available data from The Cancer Genome Atlas (TCGA). mRNA data (pancreatic adenocarcinoma (PAAD), n\u0026thinsp;=\u0026thinsp;177) were accessed and plots were generated using the online tool \u0026ldquo;UALCAN: A portal for facilitating tumor subgroup gene expression and survival analyses\u0026rdquo; (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://ualcan.path.uab.edu/cgi-bin/ualcan-res.pl\u003c/span\u003e\u003cspan address=\"https://ualcan.path.uab.edu/cgi-bin/ualcan-res.pl\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e)\u003c/span\u003e (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eFor statistical analyses, only tissue cores with at least 25 tumour and stromal cells (corresponding to approximately the 25th percentile of epithelial cell counts in the raw data output) were used (thereby losing all tissue cores from a total of five patients). In the non-neoadjuvant cases (n\u0026thinsp;=\u0026thinsp;104), the PDAC cell count ranged from 25 to 1,605 cells per tissue core (median: 218; mean: 305; interquartile range: 290), while the stromal cell count ranged from 42 to 1,862 cells per tissue core (median: 638; mean: 712; interquartile range: 430). Unless stated otherwise, cases involving neoadjuvant therapy were excluded from downstream statistics. Statistical analyses were performed using the R statistical computing environment version 4.4.1 (RStudio version 2024.04.2). All p-values were calculated for two-tailed tests with significance set at p\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Statistical analyses included Wilcoxon signed-rank tests for pairwise comparisons, Spearman's rank correlation coefficient to assess associations, Kaplan-Meier survival analysis with log-rank tests for time-to-event data, and chi-squared tests to assess relationships between categorical variables.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003eID1 is highly expressed, while pSMAD2 protein mostly absent in PDAC parenchyma\u003c/h2\u003e\n \u003cp\u003eID1 protein was highly expressed among epithelial PDAC cells compared to the surrounding stroma (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, non-neoadjuvant cases) in both TC and TF (Figs. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eA, \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e and \u003cspan class=\"InternalRef\"\u003eS8\u003c/span\u003eA, Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). Mean ID1 protein expression was even more pronounced in the neoadjuvant-treated cases (both TC and TF, \u003cstrong\u003eTable \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/strong\u003e). pSMAD2 protein expression was mostly absent in PDAC parenchyma, aligning well with TCGA mRNA data (reduced transcript counts compared to normal parenchyma, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS8\u003c/span\u003eB\u003c/strong\u003e), with significantly higher preservation in the stroma (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, non-neoadjuvant cases) \u003cstrong\u003e(Figure \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003eA\u003c/strong\u003e, Tables \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e and \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e). Mean pSMAD2 expression in PDAC was significantly lower in the TC than in TF (p\u0026thinsp;=\u0026thinsp;0.02) while mean stromal pSMAD2 protein expression was significantly higher in the TF (p\u0026thinsp;=\u0026thinsp;0.04) (non-neoadjuvant cases, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003eB\u003c/strong\u003e). Both PDAC parenchyma and stroma were mostly BMP4\u003csup\u003elow\u003c/sup\u003e (non-neoadjuvant cases, mean 0.25 mRNA transcripts per tumour cell) with a few parenchymal BMP4\u003csup\u003ehigh\u003c/sup\u003e cases (maximum of 6.3 transcripts/tumour cell (TF) in a ATM-mutant pT3 pN1 duct adenocarcinoma (\u003cspan class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e36\u003c/span\u003e) with adjuvant FOLFIRINOX (\u003cspan class=\"CitationRef\"\u003e37\u003c/span\u003e) and \u0026gt;\u0026thinsp;5 years overall survival; hyperactive BMP4 signalling described in PDAC with ATM loss (\u003cspan class=\"CitationRef\"\u003e38\u003c/span\u003e); respective tissue core in \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/strong\u003e) (Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\u0026nbsp;\u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eExpression of TGF- and BMP-signalling molecules in PDAC and associated stroma, each for Tumour Centre (TC) and Tumour Front (TF) (non-neoadjuvant only, n\u0026thinsp;=\u0026thinsp;82 analysed, neoadjuvant in \u003cstrong\u003eTable \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/strong\u003e). Protein expression (ID1, pSMAD2 immunohistochemistry) in percent (%), mRNA transcripts (in-situ hybridisation) per PDAC cell and stromal cell respectively (including only tissue cores with at least 25 stromal and at least 25 cancer epithelial cells). \u003cem\u003eSD\u003c/em\u003e: Standard Deviation.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"5\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eRegion\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePDAC_ID1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eStroma_ID1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePDAC_pSMAD2\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eStroma_pSMAD2\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e63.8% (0-100, SD\u0026thinsp;=\u0026thinsp;22.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22.7% (3.7\u0026ndash;77.7, SD\u0026thinsp;=\u0026thinsp;13.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16.3% (0-100, SD\u0026thinsp;=\u0026thinsp;21.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15.6% (0.7\u0026ndash;87.8, SD\u0026thinsp;=\u0026thinsp;18.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e60.5% (0-100, SD\u0026thinsp;=\u0026thinsp;24.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.9% (3.1\u0026ndash;72.6, SD\u0026thinsp;=\u0026thinsp;12.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16.5% (0-100, SD\u0026thinsp;=\u0026thinsp;26.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.7% (0-96.7, SD\u0026thinsp;=\u0026thinsp;25.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cdiv class=\"gridtable\"\u003e\u003cbr\u003e\n \u003ctable id=\"Taba\" border=\"1\"\u003e\n \u003ccolgroup cols=\"11\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eRegion\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePDAC_GREM1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eStroma_GREM1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePDAC_TGFA\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eStroma_TGFA\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePDAC_TGFB1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eStroma_TGFB1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePDAC_TGFB2\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eStroma_TGFB2\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePDAC_BMP4\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eStroma_BMP4\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.7 (0-5.5, SD\u0026thinsp;=\u0026thinsp;1.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.9 (0-6.4, SD\u0026thinsp;=\u0026thinsp;1.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.3 (0.2\u0026ndash;3.4, SD\u0026thinsp;=\u0026thinsp;0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.5 (0-3.2, SD\u0026thinsp;=\u0026thinsp;0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.8 (0.1\u0026ndash;8.5, SD\u0026thinsp;=\u0026thinsp;1.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.9 (0-2.4, SD\u0026thinsp;=\u0026thinsp;0.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.8 (0.1\u0026ndash;2.3, SD\u0026thinsp;=\u0026thinsp;0.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.2 (0.4\u0026ndash;3.1, SD\u0026thinsp;=\u0026thinsp;0.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.4 (0-2.4, SD\u0026thinsp;=\u0026thinsp;0.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.3 (0-1.3, SD\u0026thinsp;=\u0026thinsp;0.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.6 (0-3.9, SD\u0026thinsp;=\u0026thinsp;0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.1 (0-10.9, SD\u0026thinsp;=\u0026thinsp;1.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.3 (0.2\u0026ndash;5.6, SD\u0026thinsp;=\u0026thinsp;0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.5 (0.2\u0026ndash;3.3, SD\u0026thinsp;=\u0026thinsp;0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.8 (0.2\u0026ndash;6.1, SD\u0026thinsp;=\u0026thinsp;1.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.9 (0-2.2, SD\u0026thinsp;=\u0026thinsp;0.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.8 (0.1\u0026ndash;3.1, SD\u0026thinsp;=\u0026thinsp;0.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.1 (0.2\u0026ndash;2.3, SD\u0026thinsp;=\u0026thinsp;0.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.5 (0-6.3, SD\u0026thinsp;=\u0026thinsp;0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.4 (0-1.1, SD\u0026thinsp;=\u0026thinsp;0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\u003cbr\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003eStrong GREM1 signalling in tumour-associated stroma and robust TGF signalling across compartments\u003c/h2\u003e\n \u003cp\u003eGREM1, TGF-A and TGF-B2 mRNA counts per cell were significantly higher in the stromal compartment compared to the PDAC cells (non-neoadjuvant only, Figs. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eB-D, Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). On the contrary, TGF-B1 mRNA transcript counts were significantly higher in the PDAC compartment (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01). There were no significant differences between TC and TF in transcript counts for the three investigated TGF ligands, BMP4 and GREM1. Both among the tumoural and stromal compartment, transcript counts of TGF-A, TGF-B1 and -B2 showed strong positive correlations (Figs. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eA and \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eB, \u003cstrong\u003eTable \u003cspan class=\"InternalRef\"\u003eS3\u003c/span\u003e\u003c/strong\u003e). In PDAC parenchyma, higher GREM1 transcript counts were negatively correlated with ID1 protein expression (rho= -0.146, p\u0026thinsp;=\u0026thinsp;0.03), consistent with biological priors. When comparing tumoural and stromal metrics, higher stromal GREM1 transcript numbers correlated with higher stromal TGF-A and -B1 transcript counts (rho\u0026thinsp;=\u0026thinsp;0.40, p\u0026thinsp;\u0026lt;\u0026thinsp;0.01; rho\u0026thinsp;=\u0026thinsp;0.42, p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eC, \u003cstrong\u003eTable \u003cspan class=\"InternalRef\"\u003eS3\u003c/span\u003e\u003c/strong\u003e). Higher tumoural BMP4 correlated with stromal TGF-A transcripts (rho\u0026thinsp;=\u0026thinsp;0.42, p\u0026thinsp;\u0026lt;\u0026thinsp;0.01).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003ch2\u003eLow GREM1 signalling and high ID1 expression define an aggressive PDAC phenotype\u003c/h2\u003e\n \u003cp\u003eWhile GREM1\u003csup\u003elow\u003c/sup\u003e PDACs showed a trend towards perineural invasion (TC: p\u0026thinsp;=\u0026thinsp;0.06, Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eA) and blood vessel infiltration (V0 vs. V2: p\u0026thinsp;=\u0026thinsp;0.06, Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eB), expression of any of the signalling molecules in PDAC (TC and TF) was not significantly associated with the presence of extracapsular extension of lymph node (LN) metastases, lymphovascular (\u0026ldquo;L1\u0026rdquo;) or perineural invasion (\u0026ldquo;Pn1\u0026rdquo;), tumour stage or tumour grade.\u003c/p\u003e\n \u003cp\u003eTumour size correlated weakly with ID1 expression in the TF (p\u0026thinsp;=\u0026thinsp;0.03, R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.06; \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS3\u003c/span\u003eA\u003c/strong\u003e), but not in the TC (p\u0026thinsp;=\u0026thinsp;0.14). N2-stage PDAC showed lower tumoural ID1 expression than N1-stage (p\u0026thinsp;=\u0026thinsp;0.02; Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eC, N0 vs. N1/N2: p\u0026thinsp;=\u0026thinsp;0.73, TCGA: \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS8\u003c/span\u003eC\u003c/strong\u003e). Other molecules exhibited no significant associations with tumour size, number of positive LN (\u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS3\u003c/span\u003eC\u003c/strong\u003e and \u003cstrong\u003eS3D\u003c/strong\u003e) or N-stage. Lower average GREM1 transcript counts per stromal (TC: p\u0026thinsp;\u0026lt;\u0026thinsp;0.01; TF: p\u0026thinsp;=\u0026thinsp;0.07, t-test) and tumoural cell (TC: p\u0026thinsp;\u0026lt;\u0026thinsp;0.01; TF: ns, t-test) in the TC spots, significantly associated with presence of metachronous distant metastasis in follow-up. TGF-B2 transcript counts per cell in PDAC parenchyma correlated significantly with the absolute number of positive lymph nodes (TF, p\u0026thinsp;=\u0026thinsp;0.02, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS3\u003c/span\u003eB\u003c/strong\u003e). TCGA data support this finding (non-significant trend, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS8\u003c/span\u003eD\u003c/strong\u003e) and reveal higher tumoral transcript counts in N1-stage PDAC.\u003c/p\u003e\n \u003cp\u003eTGF-A\u003csup\u003elow\u003c/sup\u003e (p\u0026thinsp;=\u0026thinsp;0.06, Wilcoxon rank sum; p\u0026thinsp;=\u0026thinsp;0.09, Fisher test) and -B1\u003csup\u003elow\u003c/sup\u003e stroma (p\u0026thinsp;=\u0026thinsp;0.08, Wilcoxon rank sum; p\u0026thinsp;=\u0026thinsp;0.07, Fisher test) showed a statistically non-significant trend towards higher tumour budding (BD2 and BD3, TC, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS4\u003c/span\u003e\u003c/strong\u003e). When quantifying tumour buds per individual tissue core, we found no significant differences in tumour bud counts for any of the signalling molecules studied (\u003cstrong\u003eTable \u003cspan class=\"InternalRef\"\u003eS4\u003c/span\u003e\u003c/strong\u003e). This held true whether we compared the average marker expression in PDAC versus stroma or between TC and TF per tissue core.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\n \u003ch2\u003eHeterogeneity of immune infiltrate and sparsity of FOXP3\u003csup\u003e+\u003c/sup\u003e T-regulatory cells in TGF-B2\u003csup\u003ehigh\u003c/sup\u003e PDAC\u003c/h2\u003e\n \u003cp\u003eImmune cell counts were heterogeneously distributed among the different tumour regions. CD3 had the greatest heterogeneity (43.3% of patients had consistent expression patterns (low, high) across tissue cores) in all three tumour regions (TC, TF, stromal-predominant tissue cores), while CD20 had the least (85.5); this pattern persisted when considering only TC and TF (CD3: 63.5%, CD20: 88.5%) (\u003cstrong\u003eTable \u003cspan class=\"InternalRef\"\u003eS5\u003c/span\u003e\u003c/strong\u003e).\u003c/p\u003e\n \u003cp\u003eTGF-B is known for its direct suppressive effect on T-cell function and induction of T-regulatory cells (\u003cspan class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e39\u003c/span\u003e) and the TF (invasive margin) is known as a site of pronounced immune density in PDAC (\u003cspan class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e41\u003c/span\u003e). In our cohort, we observe a statistically non-significant trend towards higher FOXP3 counts in the TF of TGF-B1\u003csup\u003ehigh\u003c/sup\u003e tumours (p\u0026thinsp;=\u0026thinsp;0.08, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS5\u003c/span\u003eA\u003c/strong\u003e; TC: p\u0026thinsp;=\u0026thinsp;0.29). Stromal TGF-B1 copies did not associate with FOXP3 levels (TC: p\u0026thinsp;=\u0026thinsp;1; TF: p\u0026thinsp;=\u0026thinsp;0.64). Higher stromal and intratumoural TGF-B2 transcripts, significantly associated with lower FOXP3 counts in the TC (p\u0026thinsp;=\u0026thinsp;0.04 and p\u0026thinsp;=\u0026thinsp;0.05 respectively, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS5\u003c/span\u003eA\u003c/strong\u003e). ID1 has been shown to suppress CD8\u003csup\u003e+\u003c/sup\u003e T-cell infiltration (\u003cspan class=\"CitationRef\"\u003e42\u003c/span\u003e). Here, CD8\u003csup\u003ehigh\u003c/sup\u003e tissue cores were not associated with ID1 expression levels (TC: p\u0026thinsp;=\u0026thinsp;0.35, TF: p\u0026thinsp;=\u0026thinsp;0.17, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS5\u003c/span\u003eB\u003c/strong\u003e). No significant associations between TGF-B1 or TGF-B2 (in either compartment) and CD8\u003csup\u003e+\u003c/sup\u003e T-cell infiltration were detected (TC/TF: all non-significant). While BMP4 has been shown to induce M2-polarization of macrophages in bladder cancer (\u003cspan class=\"CitationRef\"\u003e43\u003c/span\u003e), we did not observe significant differences in CD68\u003csup\u003e+\u003c/sup\u003e or CD163\u003csup\u003e+\u003c/sup\u003e macrophage levels between BMP4\u003csup\u003elow\u003c/sup\u003e and -\u003csup\u003ehigh\u003c/sup\u003e PDAC (TC: p\u0026thinsp;=\u0026thinsp;0.85; TF: p\u0026thinsp;=\u0026thinsp;0.52). Also, GREM1 transcripts have been described to be associated with increased M1- and M2-macrophages in PDAC (\u003cspan class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e44\u003c/span\u003e). Binary (by mean) stratification of BMP4 (CD68, TC: p\u0026thinsp;=\u0026thinsp;0.24; TF p\u0026thinsp;=\u0026thinsp;0.71; CD163, TC: p\u0026thinsp;=\u0026thinsp;0.41; TF: p\u0026thinsp;=\u0026thinsp;0.71) or GREM1 (CD68, TC: p\u0026thinsp;=\u0026thinsp;0.17; TF: p\u0026thinsp;=\u0026thinsp;0.35; CD163, TC: p\u0026thinsp;=\u0026thinsp;0.31; TF: p\u0026thinsp;=\u0026thinsp;0.56) mRNA transcript counts did not show significant correlations with macrophage levels (\u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS5\u003c/span\u003eC and -D\u003c/strong\u003e).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\n \u003ch2\u003eStromal TGF-A and -B2 linked to worse PDAC survival\u003c/h2\u003e\n \u003cp\u003eLower number of stromal TGF-B2 copies in the TC were significantly associated with worse survival (p\u0026thinsp;=\u0026thinsp;0.02, neoadjuvant cases excluded, Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eA, multivariate Cox model in \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS6\u003c/span\u003eA\u003c/strong\u003e). A higher number of stromal TGF-A in the TF showed a statistically non-significant trend towards worse survival (p\u0026thinsp;=\u0026thinsp;0.069). Interestingly, when both stromal TGF-A and -B2 were combined, the poorest overall survival was seen in the TGF-A\u003csup\u003elow\u003c/sup\u003e(TF)-TGFB2\u003csup\u003ehigh\u003c/sup\u003e(TC) stroma group (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eB), highlighting the complex interaction between these molecules and their non-linear relationship. Comparison of stromal TGF-A and -B2 groups within the same tumour region did not show statistical significance (TC/TC: p\u0026thinsp;=\u0026thinsp;0.06; TF/TF: p\u0026thinsp;=\u0026thinsp;0.21). Binary survival analysis (low versus high) of bulk mRNA TCGA data did not reveal any significant differences in outcomes for the investigated TGFs (\u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS8\u003c/span\u003eE\u003c/strong\u003e). Neither TGF-B1, BMP4 and GREM1 transcript levels nor ID1 and pSMAD2 protein expression were significantly associated with survival in either tumour or stromal compartments (\u003cstrong\u003eTable \u003cspan class=\"InternalRef\"\u003eS5\u003c/span\u003e\u003c/strong\u003e).\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026ldquo;Mesenchymal geometry\u0026rdquo; and stromal proportion does not correlate with PDAC outcome\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eTo investigate EMT and its impact on tumour aggressiveness in PDAC, we examined nuclear roundness as a morphological EMT surrogate, i.e., more spindle-shaped, less round morphology in the mesenchymal cell spectrum. Nuclear circularity is decreased during EMT induction (\u003cspan class=\"CitationRef\"\u003e45\u003c/span\u003e) and, for instance, in breast cancer a \u0026ldquo;rounded to deformed morphology\u0026rdquo; has been described after EMT (\u003cspan class=\"CitationRef\"\u003e46\u003c/span\u003e). Here, the automatically computed roundness of PDAC nuclei was between 0.61 and 0.80 (mean of 0.73 among all tissue cores, neoadjuvant cases excluded). There was no significant correlation of nuclear roundness with N-stage in this cohort (n\u0026thinsp;=\u0026thinsp;102, non-neoadjuvant, TC). Further, there was no significant difference in survival comparing less round to rounder PDAC nuclei (p\u0026thinsp;=\u0026thinsp;0.83, TC) (\u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS6\u003c/span\u003eC\u003c/strong\u003e). TGF-B is a known (sometimes SMAD-dependent) inducer of tissue fibrosis (\u003cspan class=\"CitationRef\"\u003e47\u003c/span\u003e). In PDAC, the typically dense desmoplastic stroma is traditionally considered a negative prognostic factor and is thought to form a tumour protective \u0026quot;niche\u0026quot; and physical barrier that can reduce drug penetration and prevent immune infiltration. Here, the amount of stroma was not associated with worse survival (neoadjuvant patients excluded, stratified by the mean stromal percentage per tissue core, TF: p\u0026thinsp;=\u0026thinsp;0.23; TC: p\u0026thinsp;=\u0026thinsp;0.32, \u003cstrong\u003eFigure \u003cspan class=\"InternalRef\"\u003eS6\u003c/span\u003eB\u003c/strong\u003e).\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe results of this study on TGF- and BMP-signalling in PDAC align well with the known microenvironmental (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e), transcriptomic (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e) and cellular heterogeneity of PDAC across space, time and anatomical regions (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e). Recent evidence suggests a site-specific TME (\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e) with intratumoural heterogeneity on single-cell level (\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e) involving microenvironmental transitions in metastatic progression of PDAC (\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e). In our spatially-resolved analysis investigating multiple tissue cores per PDAC case from different morphological tumour regions to account for tissue heterogeneity, we reveal differences in TGF/BMP signalling and, consistent with previous literature (\u003cspan additionalcitationids=\"CR56\" citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e), in immune cell distribution across tumour regions. A chronic inflammatory state during manifestation of PDAC paralleled by dysregulation of the associated microenvironment might provide a cancer-promoting milieu and allow tumour cells to exploit physiological functional niches during tumour progression (\u0026ldquo;niche hijacking\u0026rdquo;) (\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn mice, SMAD2 and \u0026minus;\u0026thinsp;3 have been revealed to be crucial molecules that trigger a TGF-B-induced regulatory T-cell response (\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e). We observe contradictory results in TGF-B signalling: While TGF-B2\u003csup\u003ehigh\u003c/sup\u003e PDAC significantly associated with less FOXP3\u003csup\u003e+\u003c/sup\u003e T-regulatory cell infiltration in the TC, TGF-B1\u003csup\u003ehigh\u003c/sup\u003e PDAC showed the opposite trend. Higher FOXP3\u003csup\u003e+\u003c/sup\u003e counts in the proximity of TGF-B1\u003csup\u003ehigh\u003c/sup\u003e PDAC align well with EMT (\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e), however, tumoural TGF-B1 mRNA counts themselves were not associated with adverse survival. Notably, here, TGF-B transcript counts were not associated with CD8\u003csup\u003e+\u003c/sup\u003e T-cell levels. In combined analysis of the tumour regions, TGF-A\u003csup\u003elow\u003c/sup\u003e (TF)/TGFB2\u003csup\u003ehigh\u003c/sup\u003e (TC) stroma was associated with worse overall survival. Interestingly, in the TCGA data, there were no significant outcome differences in compartment-agnostic tumoral TGF mRNA transcripts (high versus -low, n\u0026thinsp;=\u0026thinsp;177), further highlighting the pivotal relevance of spatially resolved expression analysis. Recent data suggests negative LN to be a predictor of chemotherapy (gemcitabine plus capecitabine) efficacy (\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e). Here, in the TF, a cancer compartment associated with infiltrative properties (\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e), elevated tumoural TGF-B2 transcripts correlated with a higher number of lymph node metastases, in line with early reports on the role of TGF-B isoforms on PDAC progression (\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e) and bulk mRNA TCGA data. The TF has been characterised as a site of elevated immune interaction and immune cell density (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e), with upregulation of inflammatory pathways (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e) and pronounced tumour aggressiveness, for example in the form of tumour budding (\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e). Our findings regarding ID1 expression and TGF-B2 transcript counts in the tumour further emphasise its aggressive biology. Although our regression analyses revealed few significant associations (with outliers largely determining the significance of factors such as lymph node metastases and tumour size), we believe that examining larger cohorts would reinforce these findings. For example, neither tumoural nor stromal BMP4 showed any trend in association with the number of lymph node metastases or tumour size. This further highlights the complexity of pathway activity and the importance of possible downstream dysregulation. Higher TGF-B2 mRNA levels have been reported to be associated with worse overall survival when present in a macrophage-poor TME (\u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e). It appears that TGF-B2\u003csup\u003ehigh\u003c/sup\u003e PDAC and TGF-B2\u003csup\u003elow\u003c/sup\u003e stroma define an aggressive, pro-metastatic PDAC. The reported effect of TGF-B2 receptor overexpression on survival in PDAC has been variable (\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e, \u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e). In mice, TGF-B2 receptor neutralization resulted in increased differentiation and decreased metastasis of PDAC (\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e). This suggests that biological behaviour is strongly dictated by regional compartments/functional neighbourhoods rather than mere pathway activation.\u003c/p\u003e \u003cp\u003emRNA transcript counts of all TGF-ligands strongly correlated among tumour and stroma, possibly underscoring excessive fibrogenic signalling in PDAC (aberrant \u0026ldquo;wound healing\u0026rdquo; (\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e)). Interestingly, GREM1, TGF-A and TGF-B2 transcripts were significantly enriched in the juxtatumoural stroma as compared to PDAC parenchyma. The strong stromal predilection of GREM1 is in line with previously reported high GREM1-levels in (cancer-associated) pancreatic fibroblasts (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e). Corroborating previous results in SMAD4 (\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e), SMAD2 protein expression levels were not associated with worse survival. Consistent with proficient downstream BMP-signalling, GREM1\u003csup\u003ehigh\u003c/sup\u003e PDAC showed less ID1 expression. As reported (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e), we could confirm robust ID1 protein expression in PDAC, regardless of tumour region and in line with TCGA data. In our cohort, ID1\u003csup\u003ehigh\u003c/sup\u003e tumours were significantly larger, but there was no further significant association with conventional histopathologic risk predictors or with overall survival. Ultimately, the BMP pathway also leads to cell cycle arrest. ID1 is a downstream protein of this pathway and is, herein, associated with larger tumour size. Consequently, aberrant, context-dependent pathway signalling that bypasses physiological activity (e.g. via upstream molecules) is likely to occur in PDAC. Paradoxically, tumoural ID1 expression was significantly lower in N2- than N1-stage PDAC. ID1 has been shown to exert an immunosuppressive effect by promoting myeloid-derived suppressor cell expansion and downregulating CD8\u003csup\u003e+\u003c/sup\u003e T-cells (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e), while the density of CD8\u003csup\u003e+\u003c/sup\u003e T-cells in the TC has been demonstrated to have prognostic validity in PDAC (\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e). In our cohort, a high CD8\u003csup\u003e+\u003c/sup\u003e T-cell infiltrate did not correlate with a higher percentage of tumoural or stromal ID1 expression. Of note, this was independent of the tumour region (TC or TF), despite previous evidence suggesting that there is less CD8 infiltration and more immunosuppression in the TC (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e). GREM1 protein expression has been described to coincide with M1/M2 macrophage enrichment in PDAC (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e), a finding we cannot reproduce in our binary stratification using mRNA transcript counts and CD68/CD163 immunohistochemistry. Despite the \u0026ldquo;different faces\u0026rdquo; of GREM1 function (\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e), loss of GREM1 promotes metastasis (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e) and a fibrogenic stromal microenvironment (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). In this work, lower stromal and tumoural GREM1 transcript numbers were significantly associated with distant tumour recurrence, supporting the role of GREM1 loss for metastatic seeding.\u003c/p\u003e \u003cp\u003eInterestingly, tumour budding counts did not differ significantly between tumour regions or between expression levels for any of the molecules analysed. TGF-B secretion by stromal cells enhances metastatic capacity in colorectal cancer (\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e), while in PDAC, it induces a myofibroblastic phenotype of cancer-associated fibroblasts via EGFR/ERBB2 signalling, promoting metastasis (\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e). Targeting TGF signalling in PDAC has already shown promising results in combination therapy (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). Here, we observed trends toward higher tumour budding in TGF-A\u003csup\u003elow\u003c/sup\u003e and -B1\u003csup\u003elow\u003c/sup\u003e stroma. We found no distant tumour recurrence in TGF-B1\u003csup\u003ehigh\u003c/sup\u003e or -B2\u003csup\u003ehigh\u003c/sup\u003e PDAC. The TGF-B \u0026ldquo;paradox\u0026rdquo; (\u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e77\u003c/span\u003e, \u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e78\u003c/span\u003e), whereby TGF-B exhibits tumour-suppressive effects in the early stages of PDAC but tumour-promoting effects in the late stages, remains to be deciphered. In our cohort, we observed no significant differences in TGF transcript counts according to T-stage. However, the association of TGF-B2\u003csup\u003ehigh\u003c/sup\u003e stroma with a higher number of lymph node metastases confirms tumour promotion in later stages and the relevance of (context-specific) TGF-B targeting (\u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e79\u003c/span\u003e). A spatial shift in TGF signalling, for example from tumour to stroma with hijacked wound healing physiology (\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e80\u003c/span\u003e), might be a mechanistic explanation for the reverse biological effects of TGF-B in late-stage disease.\u003c/p\u003e \u003cp\u003eWe could not objectify the typical \u0026ldquo;geometry\u0026rdquo; of EMT, such as less nuclear roundness, in our cohort. This is partly surprising, as we had expected a more spindle-shaped morphology in more aggressive PDAC, as nuclei become more elongated in mesenchymal-type cells (\u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e81\u003c/span\u003e, \u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e82\u003c/span\u003e). Given the recent studies showing pronounced EMT in the TF, we would have expected a more distinct EMT phenotype (\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e, \u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e83\u003c/span\u003e). Nevertheless, multiple other factors, e.g. tissue pressure (\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e) and cellular migration (\u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e85\u003c/span\u003e), can shape nuclear morphology which we do not consider here, as they are not easily quantifiable. Recent evidence supports an \"amoeboid\" PDAC phenotype within the EMT spectrum, associated with disease progression and enriched at the invasive front, which retains conventional cellular geometry and appears TGF-β-induced (\u003cspan additionalcitationids=\"CR87\" citationid=\"CR86\" class=\"CitationRef\"\u003e86\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR88\" class=\"CitationRef\"\u003e88\u003c/span\u003e). Contrary to previous works (e.g. (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR89\" class=\"CitationRef\"\u003e89\u003c/span\u003e)), the proportion of stroma content per tissue core did not correlate with overall survival in our study. The alignment of collagen fibrils, matrix stiffness, the stromal composition (\u003cspan additionalcitationids=\"CR91\" citationid=\"CR90\" class=\"CitationRef\"\u003e90\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR92\" class=\"CitationRef\"\u003e92\u003c/span\u003e) and the local immune context (\u003cspan citationid=\"CR93\" class=\"CitationRef\"\u003e93\u003c/span\u003e) might outweigh the mere amount of stroma in terms of negative prognosis (\u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e94\u003c/span\u003e, \u003cspan citationid=\"CR95\" class=\"CitationRef\"\u003e95\u003c/span\u003e). The stroma itself is dynamic (\u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e96\u003c/span\u003e) and susceptible to various microenvironmental cues, and dense desmoplasia alone does not impede T-cell infiltration (\u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e97\u003c/span\u003e). Limitations of our study include a relatively small sample size, which is also a limitation of the TCGA validation. In addition, not all patients were represented with three tissue cores, and some cores were lost due to deep TMA sectioning. Additionally, core exclusion was necessary because of the rather low number of tumour cells per core due to infiltrative growth, low tumour cell density, and the strong desmoplastic reaction typical of PDAC. To address discrepancies in expression due to intratumoural heterogeneity, \u0026ldquo;topographic\u0026rdquo; TMAs have been proposed (\u003cspan citationid=\"CR98\" class=\"CitationRef\"\u003e98\u003c/span\u003e). These capture a multitude of biological regions of interest, such as desmoplasia, inflammation and vascular invasion. The uniformly high biological aggressiveness of PDAC, characterized by vascular and perineural invasion in most tumours (e.g., only n\u0026thinsp;=\u0026thinsp;3 non-neoadjuvant patients without perineural invasion), limited the power of group comparisons. In a few cases, the classifier showed inconsistent performance in identifying single cells and in detecting poorly formed carcinoma glands and areas of tumour budding. Although multiple markers of interest were analysed, each analysis was performed single-plex, using different sections, which affected the spatial analysis. Deeper insights into the true immune-stroma-cancer crosstalk could be achieved by more multiplexing with simultaneous visualization of RNA and proteins, e.g. by immunofluorescence on the same tissue slide. Further validation and extension of our findings would be possible through future studies using spatially resolved (multiplex) technologies at the whole tissue level.\u003c/p\u003e \u003cp\u003eIn this work, we spatially quantified key TGF- and BMP-signalling molecules in tumour regions and the juxtatumoural stroma. We highlight the importance of stromal signalling as a prognostic factor for overall survival and corroborate ID1\u003csup\u003ehigh\u003c/sup\u003e, TGF-B2\u003csup\u003ehigh\u003c/sup\u003e and GREM1\u003csup\u003elow\u003c/sup\u003e tumours as more aggressive.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eArtificial Intelligence\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBMP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBone Morphogenetic Protein\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCAF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eCancer-associated Fibroblast\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDIA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDigital Image Analysis\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eEGFR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eEpidermal Growth Factor Receptor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eEMT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eEpithelial-mesenchymal Transition\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGREM1\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGremlin1\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eID1\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eInhibitor of Differentiation Protein 1\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eIHC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eImmunohistochemistry\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eISH\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eIn-situ Hybridisation\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003emRNA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003emessenger RNA\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePDAC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePancreatic Ductal Adenocarcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003epSMAD2\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ephosphoSMAD2\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSMAD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eSuppressor of Mothers against Decapentaplegic\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTumour Centre\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTumour Front\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTGF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTransforming Growth Factor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTMA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTissue Microarray\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTME\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTumor Microenvironment\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":" \u003ch2\u003eEthics approval and consent to participate\u003c/h2\u003e \u003cp\u003eThe study was approved by the Ethics Committee of Bern (CEC ID2020-00498) and conducted in accordance with the ethical standards of the Declaration of Helsinki. Informed consent to participate was obtained from all study participants.\u003c/p\u003e \u003ch2\u003eCompeting Interests\u003c/h2\u003e \u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eKB is supported by the Swiss National Science Foundation (P500PM_217647/1). MW is supported by the Foundation for Clinical-Experimental Cancer Research (\u0026ldquo;long\u0026eacute;vit\u0026eacute;\u0026rdquo;). VHK gratefully acknowledges funding by the Swiss National Science Foundation (P2SKP3_168322/1 and P2SKP3_168322/2), the Werner-Hedy Berger Janser Foundation and the Promedica Foundation.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eKB generated, analysed, and interpreted the data and drafted the manuscript. VHK and MW jointly designed and supervised the study, and interpreted data. PZ contributed data, JLR put together and characterized the PDAC cohort. SR provided the scoring software, MW scored and SR processed the data. BG and ASW provided surgical specimens and clinical data. SJL provided important resources and intellectual input. All authors revised the manuscript and approved the final version.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eKB dedicates his contributions to his late uncle, A Olympos.The authors acknowledge the Translational Research Unit at the Institute of Tissue Medicine and Pathology, University of Bern, Switzerland, for excellent technical support, and Tissuebank Bern for providing human tissue samples.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eAvailable upon reasonable request from the corresponding author.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eDerynck R, Turley SJ, Akhurst RJ. TGFβ biology in cancer progression and immunotherapy. Nat Rev Clin Oncol. 2021;18(1):9\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo X, Wang XF. Signaling cross-talk between TGF-β/BMP and other pathways. Cell Res. 2009;19(1):71\u0026ndash;88.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSiegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin. 2024;74(1):12\u0026ndash;49.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavid CJ, Huang YH, Chen M, Su J, Zou Y, Bardeesy N, et al. TGF-β Tumor Suppression through a Lethal EMT. Cell. 2016;164(5):1015\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHussain SM, Kansal RG, Alvarez MA, Hollingsworth TJ, Elahi A, Miranda-Carboni G, et al. Role of TGF-β in pancreatic ductal adenocarcinoma progression and PD-L1 expression. Cell Oncol. 2021;44(3):673\u0026ndash;87.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQiang L, Hoffman MT, Ali LR, Castillo JI, Kageler L, Temesgen A, et al. Transforming Growth Factor-β Blockade in Pancreatic Cancer Enhances Sensitivity to Combination Chemotherapy. Gastroenterology. 2023;165(4):874\u0026ndash;e89010.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchlingensiepen KH, Jaschinski F, Lang SA, Moser C, Geissler EK, Schlitt HJ, et al. Transforming growth factor-beta 2 gene silencing with trabedersen (AP 12009) in pancreatic cancer. Cancer Sci. 2011;102(6):1193\u0026ndash;200.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMa MJ, Shi YH, Liu ZD, Zhu YQ, Zhao GY, Ye JY, et al. N6-methyladenosine modified TGFB2 triggers lipid metabolism reprogramming to confer pancreatic ductal adenocarcinoma gemcitabine resistance. Oncogene. 2024;43(31):2405\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLupo F, Pezzini F, Pasini D, Fiorini E, Adamo A, Veghini L, et al. Axon guidance cue SEMA3A promotes the aggressive phenotype of basal-like PDAC. Gut. 2024;73(8):1321\u0026ndash;35.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNieto MA, Huang RYJ, Jackson RA, Thiery JP. EMT: 2016. Cell. 2016;166(1):21\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerelli L, Zhang L, Mangiameli S, Giannese F, Mahadevan KK, Peng F et al. Evolutionary fingerprints of epithelial-to-mesenchymal transition. Nature. 2025;1\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDe Smedt L, Palmans S, Andel D, Govaere O, Boeckx B, Smeets D, et al. Expression profiling of budding cells in colorectal cancer reveals an EMT-like phenotype and molecular subtype switching. Br J Cancer. 2017;116(1):58\u0026ndash;65.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaramitopoulou E, Zlobec I, Born D, Kondi-Pafiti A, Lykoudis P, Mellou A, et al. Tumour budding is a strong and independent prognostic factor in pancreatic cancer. Eur J Cancer. 2013;49(5):1032\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaramitopoulou E, Wartenberg M, Zlobec I, Cibin S, Worni M, Gloor B, et al. Tumour budding in pancreatic cancer revisited: validation of the ITBCC scoring system. Histopathology. 2018;73(1):137\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGordon KJ, Kirkbride KC, How T, Blobe GC. Bone morphogenetic proteins induce pancreatic cancer cell invasiveness through a Smad1-dependent mechanism that involves matrix metalloproteinase\u0026ndash;2. Carcinogenesis. 2009;30(2):238\u0026ndash;48.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBardeesy N, Cheng Khung, Berger JH, Chu GC, Pahler J, Olson P, et al. Smad4 is dispensable for normal pancreas development yet critical in progression and tumor biology of pancreas cancer. Genes Dev. 2006;20(22):3130\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu JW, Fairman R, Penry J, Shi Y. Formation of a Stable Heterodimer between Smad2 and Smad4*. J Biol Chem. 2001;276(23):20688\u0026ndash;94.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePapageorgis P. TGFβ Signaling in Tumor Initiation, Epithelial-to-Mesenchymal Transition, and Metastasis. J Oncol. 2015;2015(1):587193.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBiffi G, Oni TE, Spielman B, Hao Y, Elyada E, Park Y, et al. IL1-Induced JAK/STAT Signaling Is Antagonized by TGFβ to Shape CAF Heterogeneity in Pancreatic Ductal Adenocarcinoma. Cancer Discov. 2019;9(2):282\u0026ndash;301.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTahara H, Sato K, Yamazaki Y, Ohyama T, Horiguchi N, Hashizume H, et al. Transforming growth factor-α activates pancreatic stellate cells and may be involved in matrix metalloproteinase\u0026ndash;1 upregulation. Lab Invest. 2013;93(6):720\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTang Y, Zhang S, Li J, Wu C, Fan Q. ID1 marks the tumorigenesis of pancreatic ductal adenocarcinoma in mouse and human. Sci Rep. 2022;12(1):13555.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRaghunathan M, Scully KM, Wehrmaker A, Murad R, Lowy AM, Maurer HC et al. Bone Morphogenetic Protein (BMP) signaling upregulates expression of ID1 and ID3 in pancreatitis and pancreatic ductal adenocarcinoma [Internet]. bioRxiv; 2023 [cited 2024 May 29]. p. 2023.09.01.555987. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.biorxiv.org/content/\u003c/span\u003e\u003cspan address=\"https://www.biorxiv.org/content/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/2023.09.01.555987v2\u003c/span\u003e\u003cspan address=\"10.1101/2023.09.01.555987v2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang YH, Hu J, Chen F, Lecomte N, Basnet H, David CJ, et al. ID1 Mediates Escape from TGFβ Tumor Suppression in Pancreatic Cancer. Cancer Discov. 2020;10(1):142\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGeorgiadou D, Sergentanis TN, Sakellariou S, Filippakis GM, Zagouri F, Vlachodimitropoulos D, et al. VEGF and Id\u0026ndash;1 in pancreatic adenocarcinoma: Prognostic significance and impact on angiogenesis. Eur J Surg Oncol. 2014;40(10):1331\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShuno Y, Tsuno NH, Okaji Y, Tsuchiya T, Sakurai D, Nishikawa T, et al. Id1/Id3 Knockdown Inhibits Metastatic Potential of Pancreatic Cancer. J Surg Res. 2010;161(1):76\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLan L, Evan T, Li H, Hussain A, Ruiz EJ, Zaw Thin M, et al. GREM1 is required to maintain cellular heterogeneity in pancreatic cancer. Nature. 2022;607(7917):163\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStaloch D, Gao X, Liu K, Xu M, Feng X, Aronson JF, et al. Gremlin is a key pro-fibrogenic factor in chronic pancreatitis. J Mol Med. 2015;93(10):1085\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTindall RR, Faraoni EY, Li J, Zhang Y, Ting SM, Okeugo B et al. Increased Gremlin1 Expression in Pancreatic Ductal Adenocarcinoma Promotes a Fibrogenic Stromal Microenvironment. Pancreas. 2024;10.1097/MPA.0000000000002378.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLain\u0026eacute; A, Labiad O, Hernandez-Vargas H, This S, Sanlaville A, L\u0026eacute;on S, et al. Regulatory T cells promote cancer immune-escape through integrin αvβ8-mediated TGF-β activation. Nat Commun. 2021;12(1):6228.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFu S, Zhang N, Yopp AC, Chen D, Mao M, Chen D, et al. TGF-β Induces Foxp3\u0026thinsp;+\u0026thinsp;T-Regulatory Cells from CD4\u0026thinsp;+\u0026thinsp;CD25\u0026thinsp;\u0026ndash;\u0026thinsp;Precursors. Am J Transpl. 2004;4(10):1614\u0026ndash;27.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBulle A, Lim KH. Beyond just a tight fortress: contribution of stroma to epithelial-mesenchymal transition in pancreatic cancer. Signal Transduct Target Ther. 2020;5(1):1\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePati P, Karkampouna S, Bonollo F, Comp\u0026eacute;rat E, Radić M, Spahn M et al. Accelerating histopathology workflows with generative AI-based virtually multiplexed tumour profiling. Nat Mach Intell. 2024;1\u0026ndash;17.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e14(th). European congress on digital Pathology. J Pathol Inf. 2019;10:32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChandrashekar DS, Bashel B, Balasubramanya SAH, Creighton CJ, Ponce-Rodriguez I, Chakravarthi BVSK, et al. UALCAN: A Portal for Facilitating Tumor Subgroup Gene Expression and Survival Analyses. Neoplasia. 2017;19(8):649\u0026ndash;58.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArmstrong SA, Schultz CW, Azimi-Sadjadi A, Brody JR, Pishvaian MJ. ATM Dysfunction in Pancreatic Adenocarcinoma and Associated Therapeutic Implications. Mol Cancer Ther. 2019;18(11):1899\u0026ndash;908.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartino C, Pandya D, Lee R, Levy G, Lo T, Lobo S, et al. ATM-Mutated Pancreatic Cancer: Clinical and Molecular Response to Gemcitabine/Nab-Paclitaxel After Genome-Based Therapy Resistance. Pancreas. 2020;49(1):143.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKondo T, Kanai M, Kou T, Sakuma T, Mochizuki H, Kamada M, et al. Association between homologous recombination repair gene mutations and response to oxaliplatin in pancreatic cancer. Oncotarget. 2018;9(28):19817\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRussell R, Perkhofer L, Liebau S, Lin Q, Lechel A, Feld FM, et al. Loss of ATM accelerates pancreatic cancer formation and epithelial\u0026ndash;mesenchymal transition. Nat Commun. 2015;6(1):7677.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShevach EM, Davidson TS, Huter EN, DiPaolo RA, Andersson J. Role of TGF-β in the Induction of Foxp3 Expression and T Regulatory Cell Function. J Clin Immunol. 2008;28(6):640\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaramitopoulou E, Wenning AS, Acharjee A, Aeschbacher P, Marinoni I, Zlobec I et al. Spatial Heterogeneity of Immune Regulators Drives Dynamic Changes in Local Immune Responses, Affecting Disease Outcomes in Pancreatic Cancer. Clin Cancer Res. 2024;OF1\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchoumacher C et al. Derang\u0026egrave;re,Valentin, Gaudilli\u0026egrave;re-Le Dain, Gwladys, Huppe, Titouan, Rageot, David, Ilie, Alis,. CD3-CD8 immune score associated with a clinical score stratifies PDAC prognosis regardless of adjuvant or neoadjuvant chemotherapy. OncoImmunology. 2024;13(1):2294563.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePapaspyridonos M, Matei I, Huang Y, do Rosario Andre M, Brazier-Mitouart H, Waite JC, et al. Id1 suppresses anti-tumour immune responses and promotes tumour progression by impairing myeloid cell maturation. Nat Commun. 2015;6(1):6840.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMart\u0026iacute;nez VG, Rubio C, Mart\u0026iacute;nez-Fern\u0026aacute;ndez M, Segovia C, L\u0026oacute;pez-Calder\u0026oacute;n F, Gar\u0026iacute;n MI, et al. BMP4 Induces M2 Macrophage Polarization and Favors Tumor Progression in Bladder Cancer. Clin Cancer Res. 2017;23(23):7388\u0026ndash;99.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavis JM, Cheng B, Drake MM, Yu Q, Yang B, Li J, et al. Pancreatic stromal Gremlin 1 expression during pancreatic tumorigenesis. Genes Dis. 2022;9(1):108\u0026ndash;15.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeggett SE, Sim JY, Rubins JE, Neronha ZJ, Williams EK, Wong IY. Morphological single cell profiling of the epithelial\u0026ndash;mesenchymal transition. Integr Biol. 2016;8(11):1133\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGarcia MA, Yin Z, Roumeliotis TI, Butera F, Wang L, Rickman R et al. Epithelial-Mesenchymal Plasticity is regulated by inflammatory signalling networks coupled to cell morphology [Internet]. bioRxiv; 2022 [cited 2024 May 30]. p. 689737. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.biorxiv.org/content/\u003c/span\u003e\u003cspan address=\"https://www.biorxiv.org/content/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/689737v3\u003c/span\u003e\u003cspan address=\"10.1101/689737v3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFrangogiannis NG. Transforming growth factor\u0026ndash;β in tissue fibrosis. J Exp Med. 2020;217(3):e20190103.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKlein L, Tu M, Krebs N, Urbach L, Grimm D, Latif MU, et al. Spatial tumor immune heterogeneity facilitates subtype co-existence and therapy response in pancreatic cancer. Nat Commun. 2025;16(1):335.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBr\u0026auml;utigam K, Skok K, Szymonski K, Rift CV, Karamitopoulou E. Tumor Immune Microenvironment in Pancreatic Ductal Adenocarcinoma revisited - Exploring the Space. Cancer Lett. 2025;217699.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilliams HL, Dias Costa A, Zhang J, Raghavan S, Winter PS, Kapner KS, et al. Spatially Resolved Single-Cell Assessment of Pancreatic Cancer Expression Subtypes Reveals Co-expressor Phenotypes and Extensive Intratumoral Heterogeneity. Cancer Res. 2023;83(3):441\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGr\u0026uuml;nwald BT, Devisme A, Andrieux G, Vyas F, Aliar K, McCloskey CW, et al. Spatially confined sub-tumor microenvironments in pancreatic cancer. Cell. 2021;184(22):5577\u0026ndash;e559218.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhaliq AM, Rajamohan M, Saeed O, Mansouri K, Adil A, Zhang C et al. Spatial transcriptomic analysis of primary and metastatic pancreatic cancers highlights tumor microenvironmental heterogeneity. Nat Genet. 2024;1\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRaghavan S, Winter PS, Navia AW, Williams HL, DenAdel A, Lowder KE, et al. Microenvironment drives cell state, plasticity, and drug response in pancreatic cancer. Cell. 2021;184(25):6119\u0026ndash;e613726.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeng J, Sun BF, Chen CY, Zhou JY, Chen YS, Chen H, et al. Single-cell RNA-seq highlights intra-tumoral heterogeneity and malignant progression in pancreatic ductal adenocarcinoma. Cell Res. 2019;29(9):725\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLundgren S, Elebro J, Heby M, Nodin B, Leandersson K, Micke P, et al. Quantitative, qualitative and spatial analysis of lymphocyte infiltration in periampullary and pancreatic adenocarcinoma. Int J Cancer. 2020;146(12):3461\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchalck A, Sakellariou-Thompson D, Forget MA, Sei E, Hughes TG, Reuben A, et al. Single-Cell Sequencing Reveals Trajectory of Tumor-Infiltrating Lymphocyte States in Pancreatic Cancer. Cancer Discov. 2022;12(10):2330\u0026ndash;49.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWartenberg M, Cibin S, Zlobec I, Vassella E, Eppenberger-Castori S, Terracciano L, et al. Integrated Genomic and Immunophenotypic Classification of Pancreatic Cancer Reveals Three Distinct Subtypes with Prognostic/Predictive Significance. Clin Cancer Res. 2018;24(18):4444\u0026ndash;54.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBhatia R, Bhyravbhatla N, Kisling A, Li X, Batra SK, Kumar S. Cytokines chattering in pancreatic ductal adenocarcinoma tumor microenvironment. Semin Cancer Biol. 2022;86:499\u0026ndash;510.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSuijkerbuijk SJE, van Rheenen J. From good to bad: Intravital imaging of the hijack of physiological processes by cancer cells. Dev Biol. 2017;428(2):328\u0026ndash;37.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTakimoto T, Wakabayashi Y, Sekiya T, Inoue N, Morita R, Ichiyama K, et al. Smad2 and Smad3 Are Redundantly Essential for the TGF-β\u0026ndash;Mediated Regulation of Regulatory T Plasticity and Th1 Development. J Immunol. 2010;185(2):842\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWartenberg M, Zlobec I, Perren A, Koelzer VH, Gloor B, Lugli A, et al. Accumulation of FOXP3\u0026thinsp;+\u0026thinsp;T-cells in the tumor microenvironment is associated with an epithelial-mesenchymal-transition-type tumor budding phenotype and is an independent prognostic factor in surgically resected pancreatic ductal adenocarcinoma. Oncotarget. 2015;6(6):4190\u0026ndash;201.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePalmer DH, Jackson R, Springfeld C, Ghaneh P, Rawcliffe C, Halloran CM et al. Pancreatic Adenocarcinoma: Long-Term Outcomes of Adjuvant Therapy in the ESPAC4 Phase III Trial. J Clin Oncol. 2024;0(0):JCO.24.01118.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJu Y, Xu D, Liao Mmiao, Sun Y, Bao W, dai, Yao F, et al. Barriers and opportunities in pancreatic cancer immunotherapy. Npj Precis Oncol. 2024;8(1):199.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFriess H, Yamanaka Y, B\u0026uuml;chler M, Ebert M, Beger HG, Gold LI, et al. Enhanced expression of transforming growth factor β isoforms in pancreatic cancer correlates with decreased survival. Gastroenterology. 1993;105(6):1846\u0026ndash;56.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZwart ES, van Ee T, Affandi AJ, Boyd LNC, Rodriguez E, den Haan JMM, et al. Spatial immune composition of tumor microenvironment in patients with pancreatic cancer. Cancer Immunol Immunother. 2023;72(12):4385\u0026ndash;97.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaramitopoulou E, Wenning AS, Acharjee A, Zlobec I, Aeschbacher P, Perren A, et al. Spatially restricted tumour-associated and host-associated immune drivers correlate with the recurrence sites of pancreatic cancer. Gut. 2023;72(8):1523\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlver TN, Bergholtz H, Holm MB, Dorg LT, Skrede ML, Kure EH, et al. Spatial Transcriptomics Reveals Cancer and Stromal Cell Heterogeneity Between Center and Invasive Front of Pancreatic Cancer. Mod Pathol. 2025;38(6):100726.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQazi S, Chang WH, Lee C, Trieu V. Abstract B070: TGFB2 mRNA levels prognostically interact with Interferon-alpha receptor activation of IRF9 and IFI27, and makers for tumor-associated macrophages impacting overall survival in PDAC. Cancer Res. 2024;84(17Supplement2):B070.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWagner M, Kleeff J, Friess H, B\u0026uuml;chler MW, Korc M. Enhanced Expression of the Type II Transforming Growth Factor-β Receptor Is Associated with Decreased Survival in Human Pancreatic Cancer. Pancreas. 1999;19(4):370.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJavle M, Li Y, Tan D, Dong X, Chang P, Kar S, et al. Biomarkers of TGF-β Signaling Pathway and Prognosis of Pancreatic Cancer. PLoS ONE. 2014;9(1):e85942.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOstapoff KT, Cenik BK, Wang M, Ye R, Xu X, Nugent D, et al. Neutralizing Murine TGFβR2 Promotes a Differentiated Tumor Cell Phenotype and Inhibits Pancreatic Cancer Metastasis. Cancer Res. 2014;74(18):4996\u0026ndash;5007.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeyell M, Garris CS, Laughney AM. Cancer metastasis as a non-healing wound. Br J Cancer. 2021;124(9):1491\u0026ndash;502.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMasugi Y, Abe T, Ueno A, Fujii-Nishimura Y, Ojima H, Endo Y, et al. Characterization of spatial distribution of tumor-infiltrating CD8\u0026thinsp;+\u0026thinsp;T cells refines their prognostic utility for pancreatic cancer survival. Mod Pathol. 2019;32(10):1495\u0026ndash;507.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoustakas A, L\u0026ouml;hr JM, Heuchel RL. Cellular heterogeneity in pancreatic cancer: the different faces of gremlin action. Signal Transduct Target Ther. 2022;7(1):1\u0026ndash;3.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCalon A, Espinet E, Palomo-Ponce S, Tauriello DVF, Iglesias M, C\u0026eacute;spedes MV, et al. Dependency of Colorectal Cancer on a TGF-β-Driven Program in Stromal Cells for Metastasis Initiation. Cancer Cell. 2012;22(5):571\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMucciolo G, Henr\u0026iacute;quez JA, Jihad M, Teles SP, Manansala JS, Li W, et al. EGFR-activated myofibroblasts promote metastasis of pancreatic cancer. Cancer Cell. 2024;42(1):101\u0026ndash;e11811.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShen W, Tao G, qing, Zhang Y, Cai B, Sun J, Tian Z. qiang. TGF-β in pancreatic cancer initiation and progression: two sides of the same coin. Cell Biosci. 2017;7(1):39.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrincipe DR, Doll JA, Bauer J, Jung B, Munshi HG, Bartholin L, et al. TGF-β: Duality of Function Between Tumor Prevention and Carcinogenesis. JNCI J Natl Cancer Inst. 2014;106(2):djt369.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHosein AN, Dougan SK, Aguirre AJ, Maitra A. Translational advances in pancreatic ductal adenocarcinoma therapy. Nat Cancer. 2022;3(3):272\u0026ndash;86.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeng Z, Fan T, Xiao C, Tian H, Zheng Y, Li C, et al. TGF-β signaling in health, disease and therapeutics. Signal Transduct Target Ther. 2024;9(1):1\u0026ndash;40.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWalters B, Uynuk-Ool T, Rothdiener M, Palm J, Hart ML, Stegemann JP, et al. Engineering the geometrical shape of mesenchymal stromal cells through defined cyclic stretch regimens. Sci Rep. 2017;7(1):6640.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeggett SE, Hruska AM, Guo M, Wong IY. The epithelial-mesenchymal transition and the cytoskeleton in bioengineered systems. Cell Commun Signal. 2021;19(1):32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaiques O, Sallan MC, Laddach R, Pandya P, Varela A, Crosas-Molist E, et al. Matrix mechano-sensing at the invasive front induces a cytoskeletal and transcriptional memory supporting metastasis. Nat Commun. 2025;16(1):1394.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVilleneuve C, Hashmi A, Ylivinkka I, Lawson-Keister E, Miroshnikova YA, P\u0026eacute;rez-Gonz\u0026aacute;lez C, et al. Mechanical forces across compartments coordinate cell shape and fate transitions to generate tissue architecture. Nat Cell Biol. 2024;26(2):207\u0026ndash;18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSt\u0026ouml;berl S, Flommersfeld J, Kreft MM, Benoit M, Broedersz CP, R\u0026auml;dler JO. Nuclear deformation and dynamics of migrating cells in 3D confinement reveal adaptation of pulling and pushing forces. Sci Adv. 2024;10(34):eadm9195.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGeorgouli M, Herraiz C, Crosas-Molist E, Fanshawe B, Maiques O, Perdrix A, et al. Regional Activation of Myosin II in Cancer Cells Drives Tumor Progression via a Secretory Cross-Talk with the Immune Microenvironment. Cell. 2019;176(4):757\u0026ndash;e77423.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSamain R, Maiques O, Monger J, Lam H, Candido J, George S, et al. CD73 controls Myosin II\u0026ndash;driven invasion, metastasis, and immunosuppression in amoeboid pancreatic cancer cells. Sci Adv. 2023;9(42):eadi0244.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGraziani V, Rodriguez-Hernandez I, Maiques O, Sanz-Moreno V. The amoeboid state as part of the epithelial-to-mesenchymal transition programme. Trends Cell Biol. 2022;32(3):228\u0026ndash;42.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRhim AD, Oberstein PE, Thomas DH, Mirek ET, Palermo CF, Sastra SA, et al. Stromal Elements Act to Restrain, Rather Than Support, Pancreatic Ductal Adenocarcinoma. Cancer Cell. 2014;25(6):735\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShields MA, Dangi-Garimella S, Redig AJ, Munshi HG. Biochemical role of the collagen-rich tumour microenvironment in pancreatic cancer progression. Biochem J. 2011;441(2):541\u0026ndash;52.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMascharak S, Guo JL, Foster DS, Khan A, Davitt MF, Nguyen AT, et al. Desmoplastic stromal signatures predict patient outcomes in pancreatic ductal adenocarcinoma. Cell Rep Med. 2023;4(11):101248.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOgawa Y, Masugi Y, Abe T, Yamazaki K, Ueno A, Fujii-Nishimura Y, et al. Three Distinct Stroma Types in Human Pancreatic Cancer Identified by Image Analysis of Fibroblast Subpopulations and Collagen. Clin Cancer Res. 2021;27(1):107\u0026ndash;19.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTan X, Rosin M, Appinger S, Deierl JC, Reichel K, Coolsen M, et al. Stroma and lymphocytes identified by deep learning are independent predictors for survival in pancreatic cancer. Sci Rep. 2025;15(1):9415.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBolm L, Zghurskyi P, Lapshyn H, Petrova E, Zemskov S, Vashist YK, et al. Alignment of stroma fibers, microvessel density and immune cell populations determine overall survival in pancreatic cancer\u0026mdash;An analysis of stromal morphology. PLoS ONE. 2020;15(7):e0234568.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDrifka CR, Loeffler AG, Mathewson K, Keikhosravi A, Eickhoff JC, Liu Y, et al. Highly aligned stromal collagen is a negative prognostic factor following pancreatic ductal adenocarcinoma resection. Oncotarget. 2016;7(46):76197\u0026ndash;213.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eErkan M, Michalski CW, Rieder S, Reiser\u0026ndash;Erkan C, Abiatari I, Kolb A, et al. The Activated Stroma Index Is a Novel and Independent Prognostic Marker in Pancreatic Ductal Adenocarcinoma. Clin Gastroenterol Hepatol. 2008;6(10):1155\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarstens JL, Correa de Sampaio P, Yang D, Barua S, Wang H, Rao A, et al. Spatial computation of intratumoral T cells correlates with survival of patients with pancreatic cancer. Nat Commun. 2017;8(1):15095.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBingham V, Harewood L, McQuaid S, Craig SG, Revolta JF, Kim CS, et al. Topographic analysis of pancreatic cancer by TMA and digital spatial profiling reveals biological complexity with potential therapeutic implications. Sci Rep. 2024;14(1):11361.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Transforming Growth Factors, Bone Morphogenetic Proteins, Stromal Cells, Pancreatic Neoplasms, AI (Artificial Intelligence), GREM1 protein, Inhibitor of Differentiation Protein 1 (ID1), Spatial Analysis","lastPublishedDoi":"10.21203/rs.3.rs-6506146/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6506146/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eTransforming Growth Factor (TGF) and Bone Morphogenetic Protein (BMP) signalling critically influence pancreatic ductal adenocarcinoma (PDAC) progression, with TGF-B paradoxically exerting both tumour-promoting and -suppressive effects. Parallel to this observation, the specific context-dependent, spatial dynamics of these pathways and their interaction with the tumour microenvironment (TME) remain poorly understood.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe performed a spatially resolved analysis of PDAC on a multi-region tissue microarray cohort of 117 curatively resected PDAC specimens consisting of tumour centre (TC), tumour front (TF), and stromal(-predominant) tissue cores each. Protein (ID1, pSMAD2) and mRNA (TGF-A, TGF-B1/2, BMP4, GREM1) expression were assessed in each tissue compartment by immunohistochemistry and in situ hybridization, respectively, quantified by digital image analysis, and correlated with clinicopathologic features.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eID1 was significantly overexpressed in PDAC cells compared to associated stroma (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01), while pSMAD2 was largely absent in PDAC cells, but preserved among associated stroma compartments, particularly in TF cores (p\u0026thinsp;=\u0026thinsp;0.04). Higher stromal GREM1 signal correlated with reduced overall tumoural ID1 protein expression (p\u0026thinsp;=\u0026thinsp;0.02), and TGF-B2\u003csup\u003ehigh\u003c/sup\u003e/TGF-A\u003csup\u003elow\u003c/sup\u003e stroma was significantly associated with worse survival (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01). Intratumoural TGF-B2 was inversely correlated with stromal pSMAD2 expression (p\u0026thinsp;=\u0026thinsp;0.03) and was associated with lymph node involvement (p\u0026thinsp;=\u0026thinsp;0.02). FOXP3\u003csup\u003e+\u003c/sup\u003e regulatory T-cells were significantly reduced in TGF-B2\u003csup\u003ehigh\u003c/sup\u003e tumours (p\u0026thinsp;=\u0026thinsp;0.04), while higher tumoural TGF-B1 exhibited a trend towards increased FOXP3\u003csup\u003e+\u003c/sup\u003e cells (p\u0026thinsp;=\u0026thinsp;0.08).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eOur spatial analysis reveals intratumoural heterogeneity of TGF/BMP signalling and its significance for PDAC progression. Notably, stromal TGF-B2 emerges as a prognostic biomarker, while TGF-B1 and ID1 are implicated in adverse clinical and pathologic features. These findings highlight the importance of TGF/BMP signalling niches in the TME with implications for PDAC biology and can inform the development of future therapeutic strategies.\u003c/p\u003e","manuscriptTitle":"Spatially resolved analysis of TGF/BMP signalling in pancreatic ductal adenocarcinoma by digital pathology identifies patient subgroups with adverse outcome","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-29 15:03:12","doi":"10.21203/rs.3.rs-6506146/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-07-16T07:37:43+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-15T08:07:48+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"104599241957904592939273788773476713390","date":"2025-07-10T07:16:16+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-10T05:24:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"11827727733186739112750757682171554347","date":"2025-07-09T10:04:46+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-08T06:25:40+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-06-18T22:32:56+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cancer","date":"2025-06-18T15:32:36+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"b6d34a98-62e6-4389-83d7-9bb62f19e2a1","owner":[],"postedDate":"June 29th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-08-25T16:30:57+00:00","versionOfRecord":{"articleIdentity":"rs-6506146","link":"https://doi.org/10.1186/s12885-025-14751-3","journal":{"identity":"bmc-cancer","isVorOnly":false,"title":"BMC Cancer"},"publishedOn":"2025-08-18 15:57:00","publishedOnDateReadable":"August 18th, 2025"},"versionCreatedAt":"2025-06-29 15:03:12","video":"","vorDoi":"10.1186/s12885-025-14751-3","vorDoiUrl":"https://doi.org/10.1186/s12885-025-14751-3","workflowStages":[]},"version":"v1","identity":"rs-6506146","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6506146","identity":"rs-6506146","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}