Microbiota-dependent activation of the myeloid calcineurin-NFAT pathway inhibits B7H3- and B7H4-dependent anti-tumor immunity in colorectal cancer.

Bacterial sensing by intestinal tumor cells contributes to tumor growth through cell-intrinsic activation of the calcineurin-NFAT axis, but the role of this pathway in other intestinal cells remains unclear. Here, we found that myeloid-specific deletion of calcineurin in mice activated protective CD8+ T cell responses and inhibited colorectal cancer (CRC) growth. Microbial sensing by myeloid cells promoted calcineurin- and NFAT-dependent interleukin 6 (IL-6) release, expression of the co-inhibitory molecules B7H3 and B7H4 by tumor cells, and inhibition of CD8+ T cell-dependent anti-tumor immunity. Accordingly, targeting members of this pathway activated protective CD8+ T cell responses and inhibited primary and metastatic CRC growth. B7H3 and B7H4 were expressed by the majority of human primary CRCs and metastases, which was associated with low numbers of tumor-infiltrating CD8+ T cells and poor survival. Therefore, a microbiota-, calcineurin-, and B7H3/B7H4-dependent pathway controls anti-tumor immunity, revealing additional targets for immune checkpoint inhibition in microsatellite-stable CRC.

Intratumoral heterogeneity affects tumor regression and Ki67 proliferation index in perioperatively treated gastric carcinoma.

BACKGROUND: Intratumoral heterogeneity (ITH) is a major problem in gastric cancer (GC). We tested Ki67 and tumor regression for ITH after neoadjuvant/perioperative chemotherapy. METHODS: 429 paraffin blocks were obtained from 106 neoadjuvantly/perioperatively treated GCs (one to five blocks per case). Serial sections were stained with Masson's trichrome, antibodies directed against cytokeratin and Ki67, and finally digitalized. Tumor regression and three different Ki67 proliferation indices (PI), i.e., maximum PI (KiH), minimum PI (KiL), and the difference between KiH/KiL (KiD) were obtained per block. Statistics were performed in a block-wise (all blocks irrespective of their case-origin) and case-wise manner. RESULTS: Ki67 and tumor regression showed extensive ITH in our series (maximum ITH within a case: 31% to 85% for KiH; 4.5% to 95.6% for tumor regression). In addition, Ki67 was significantly associated with tumor regression (p < 0.001). Responders (<10% residual tumor, p = 0.016) exhibited prolonged survival. However, there was no significant survival benefit after cut-off values were increased ≥20% residual tumor mass. Ki67 remained without prognostic value. CONCLUSIONS: Digital image analysis in tumor regression evaluation might help overcome inter- and intraobserver variability and validate classification systems. Ki67 may serve as a sensitivity predictor for chemotherapy and an indicator of ITH.

Exploration of the tumor biological significance of PCLO in gastric cancer - Results from a large Central European cohort.

INTRODUCTION: A recent multiregional whole exome sequencing of 48 tumour samples from 9 gastric adenocarcinomas discovered PCLO mutations in 23 (47.9%) tumour samples. Based on that unexpected high prevalence of PCLO mutations, we hypothesized a tumour biological significance of PCLO in gastric cancer (GC). METHODS: Tumour samples (whole tissue sections) obtained from 466 patients resected for therapy-naive GC were stained with an anti-PCLO antibody. The Histoscore for tumour cells and the presence of immunostaining of stromal cells and tumour vessels was docu-mented for each case. An algorithm for PCLO immunopositivity was formed and correlat-ed with clinicopathological patient characteristics. RESULTS: 175 GCs were classified as PCLO-positive within tumour cells, and 291 as negative. Stromal cells were positive for PCLO in 106 cases and tumour vessels in 84. PCLO positive GCs more often showed an intestinal phenotype, a lower T-category and were more commonly associated with Helicobacter pylori-infection. A separate analysis of PCLO ex-pression in intestinal and diffuse type GCs, respectively, showed no significant correla-tions. Patients with PCLO negative/low tumour cells showed a shortened overall (14.0±1.4 vs. 16.0±1.8 months) and tumour specific survival (15.0±1.6 months vs. 17.9±3.6). Compar-ison of PCLO's genotype with its phenotype in 48 tumour samples obtained from nine cases showed no direct correlations with missense mutations. DISCUSSION/CONCLUSION: Our data provide evidence that PCLO is differentially expressed in GC and might delay tumour progression.

The putative pleiotropic functions of meprin ß in gastric cancer.

BACKGROUND: The gastric microbiome and inflammation play a key role in gastric cancer (GC) by regulating the immune response in a complex manner and by inflammatory events supporting carcinogenesis. Meprin ß is a zinc endopeptidase and participates in tissue homeostasis, intestinal barrier function and immunological processes. It influences local inflammatory processes, dysbiosis and the microbiome. Here, we tested the hypothesis that meprin ß is expressed in GC and of tumor biological significance. PATIENTS AND METHODS: Four hundred forty whole mount tissue sections of patients with therapy-naive GC were stained with an anti-meprin ß antibody. The histoscore and staining pattern were analyzed for each case. Following dichotomization at the median histoscore into a "low" and "high" group, the expression was correlated with numerous clinicopathological patient characteristics. RESULTS: Meprin ß was found intracellularly and at the cell membrane of GC. Cytoplasmic expression correlated with the phenotype according to Lauren, microsatellite instability and PD-L1 status. Membranous expression correlated with intestinal phenotype, mucin-1-, E-cadherin-, ß-catenin status, mucin typus, microsatellite instability, KRAS mutation and PD-L1-positivity. Patients with cytoplasmic expression of meprin ß showed a better overall and tumor-specific survival. CONCLUSIONS: Meprin ß is differentially expressed in GC and has potential tumor biological relevance. It might function as a tumor suppressor or promotor depending on histoanatomical site and context.

The cancer-associated meprin ß variant G32R provides an additional activation site and promotes cancer cell invasion.

The extracellular metalloprotease meprin ß is expressed as a homodimer and is primarily membrane bound. Meprin ß can be released from the cell surface by its known sheddases ADAM10 and ADAM17. Activation of pro-meprin ß at the cell surface prevents its shedding, thereby stabilizing its proteolytic activity at the plasma membrane. We show that a single amino acid exchange variant (G32R) of meprin ß, identified in endometrium cancer, is more active against a peptide substrate and the IL-6 receptor than wild-type meprin ß. We demonstrate that the change to an arginine residue at position 32 represents an additional activation site used by furin-like proteases in the Golgi, which consequently leads to reduced shedding by ADAM17. We investigated this meprin ß G32R variant to assess cell proliferation, invasion through a collagen IV matrix and outgrowth from tumor spheroids. We found that increased meprin ß G32R activity at the cell surface reduces cell proliferation, but increases cell invasion.

Gastric Carcinomas with Stromal B7-H3 Expression Have Lower Intratumoural CD8+ T Cell Density.

CD8+ T cells are the main effector cells of anti-cancer immune response that can be regulated by various costimulatory and coinhibitory molecules, including members of the B7 family. B7 homolog 3 (B7-H3) appears as a promising marker for immunotherapy; however, its significance in gastric cancer (GC) is unclear yet. We evaluated the spatial distribution of CD8+ T cells in relation to the expression of B7-H3 by double immunohistochemical staining. The level of B7-H3 intensity was scored manually (0-3) and dichotomized into B7-H3-low and B7-H3-high groups. The distribution and density of CD8+ T cells was analysed using whole slide digital imaging. B7-H3 was expressed mainly in the stromal compartment of GC (n = 73, 76% of all cases). Tumours with high expression of B7-H3 showed larger spatial differences of CD8+ T cells (86.4/mm2 in tumour centre vs. 414.9/mm2 in invasive front) when compared to B7-H3-low group (157.7/mm2 vs. 218.7/mm2, respectively) (p < 0.001). This study provides insight into the expression pattern of B7-H3 in GC of Western origin. In GCs with higher level of B7-H3 expression, CD8+ T cells were spatially suppressed in the tumour centre suggesting that B7-H3 might be involved in tumour escape mechanisms from the immune response.

Prognostic significance of tumour budding assessed in gastric carcinoma according to the criteria of the International Tumour Budding Consensus Conference.

AIMS: In this study, we aimed to independently evaluate the utility and prognostic value of tumour budding (TB) according to the International Tumour Budding Consensus Conference (ITBCC) criteria in a large and a well-characterised European gastric cancer (GC) cohort. METHODS AND RESULTS: In 456 consecutive, surgically treated GCs, TB was assessed according to the ITBCC criteria and scored as Bd0 (no buds), Bd1 (one to four buds), Bd2 (five to nine buds) or Bd3 (≥10 buds). Cases with TB present were divided into low- (Bd1/Bd2) and high-budding (Bd3) groups. The TB score was analysed in relation to the clinicopathological parameters, overall survival (OS) and tumour-specific survival (TSS); 115 (25.2%) cases had no, 104 (22.8%) had low and 237 (52.0%) had high TB. The TB score correlated significantly with sex, Laurén phenotype, pT-, pN- and M categories, histological grade, R status; and lymph node ratio, lymphatic invasion, perineural invasion and HER2-, MET- and MSI status. In both total and intestinal-type early invasive GC (n = 57 and n = 41, respectively), significant associations between the presence and extent of TB and presence of lymph node metastasis were detected. Significant differences in OS and TSS between the TB groups were found; however, TB did not retain significance in multivariate models. CONCLUSIONS: Our data show that the ITBCC criteria can be applied to GC. The data correlated significantly with the diverse clinicopathological characteristics, including patient outcome, and can help to standardise diagnostics and research into special histological features of malignant tumours in general and GC in particular.

Questioning the IGF1 receptor's assigned role in CRC - a case for rehabilitation?

BACKGROUND: The insulin-like growth factor 1 receptor (IGF1R) is suspected to be involved in colorectal carcinogenesis and has been associated with worse survival in colorectal cancer (CRC). We hypothesized that the alleged suspect might be in truth beyond any suspicion. We investigated if the expression of the IGF1R in CRC correlates with (1) clinicopathological patient characteristics, including survival, and hence is involved in colon cancer biology; (2) the expression of the IGF1R in CRC is linked to the expression of the insulin receptor (IR). METHODS: We evaluated 4497 CRC samples from 1499 patients for the expression of IGF1R in tumor cells by immunohistochemistry. Cytoplasmic (cCC-IGF1R) and membranous (mCC-IGF1R) immunostaining was evaluated by employing a modified HistoScore (HScore), which was dichotomized into low or high IGF1R expressions. The IGF1R status was correlated with clinicopathological patient characteristics, survival and the IR expression status. RESULTS: cCC-IGF1R and mCC-IGF1R (HScore> 0) were found in 85.4 and 60.8% of all CRCs. After dichotomization of the HScores, 54.9 and 48.6% were classified as cCC-IGF1R-high and mCC-IGF1R-high, respectively. IGF1R was associated with tumor localization, local tumor growth, lymphatic vessel invasion, grading, mismatch repair protein expression status and IR-expression. We found no significant association with overall or tumor-specific survival, with a tendency for an even improved overall survival for cCC-IGF1R. CONCLUSIONS: IGF1R expression is frequent and biologically relevant in CRC, but does not correlate with patient survival. The IGF1R might be beyond suspicion in CRC after all.

decompTumor2Sig: identification of mutational signatures active in individual tumors.

BACKGROUND: The somatic mutations found in a tumor have in most cases been caused by multiple mutational processes such as those related to extrinsic carcinogens like cigarette smoke, and those related to intrinsic processes like age-related spontaneous deamination of 5-methylcytosine. The effect of such mutational processes can be modeled by mutational signatures, of which two different conceptualizations exist: the model introduced by Alexandrov et al., Nature 500:415-421, 2013, and the model introduced by Shiraishi et al., PLoS Genetics 11(12):e1005657, 2015. The initial identification and definition of mutational signatures requires large sets of tumor samples. RESULTS: Here, we present decompTumor2Sig, an easy to use R package that can decompose an individual tumor genome into a given set of Alexandrov-type or Shiraishi-type signatures, thus quantifying the contribution of the corresponding mutational processes to the somatic mutations identified in the tumor. Until now, such tools were available only for Alexandrov signatures. We demonstrate the correctness and usefulness of our approach with three test cases, using somatic mutations from 21 breast cancer genomes, from 435 tumor genomes of ten different tumor entities, and from simulated tumor genomes, respectively. CONCLUSIONS: The decompTumor2Sig package is freely available and has been accepted for inclusion in Bioconductor.

The expression of the insulin receptor in gastric cancer correlates with the HER2 status and may have putative therapeutic implications.

BACKGROUND: Metabolic reprogramming in gastric cancer (GC) involves not only an alteration of glucose metabolism, but also of insulin receptor (IR) expression. We investigated if (1) GCs express the IR in cancer cells (CC-IR) and vasculature (VIR), (2) IR expression is clinically relevant and may be a novel target of GC treatment. METHODS: 467 primary GCs were studied by immunohistochemistry using an IR-specific antibody. CD31-immunostaining ensured the presence of representative intratumoral microvasculature. VIR, and membranous and cytoplasmic CC-IR (mCC-IR, cCC-IR) were evaluated using a modified HistoScore (HScore) and subsequently dichotomized into low or high IR expressions. The IR status was correlated with clinico-pathological patient characteristics, including survival and HER2 status. RESULTS: VIR, mCC-IR, and cCC-IR (HScore > 0) were found in 97.0%, 87.6%, and 95.7% of all GCs. After dichotomization of the HScores, 50.7, 48.8, and 50.3% were classified as VIR-high, mCC-IR-high, and cCC-IR-high, respectively. IR was associated with the Laurén phenotype, tumor localization, local tumor growth, vascular invasion, perineural invasion, tumor budding, mucin phenotype, UICC stage, worse survival, and the HER2 status. On multivariate analysis, VIR status was an independent prognosticator of overall (p = 0.010) and tumor-specific (p = 0.006) patient survival. CONCLUSIONS: VIR and CC-IR expressions are frequent in GC, biologically significant and even correlate with the HER2 status, opening avenues for novel putative therapeutic interventions in GC.

Interpreting whole genome and exome sequencing data of individual gastric cancer samples.

BACKGROUND: Gastric cancer is the fourth most common cancer and the second leading cause of cancer death worldwide. In order to understand the genetic background, we sequenced the whole exome and the whole genome of one microsatellite stable as well as one microsatellite unstable tumor and the matched healthy tissue on two different NGS platforms. We here aimed to provide a comparative approach for individual clinical tumor sequencing and annotation using different sequencing technologies and mutation calling algorithms. RESULTS: We applied a population-based whole genome resource as a novel pathway-based filter for interpretation of genomic alterations from single nucleotide variations (SNV), indels, and large structural variations. In addition to a comparison with tumor genome database resources and a filtering approach using data from the 1000 Genomes Project, we performed pyrosequencing analysis and immunohistochemistry in a large cohort of 428 independent gastric cancer cases. CONCLUSION: We here provide an example comparing the usefulness and potential pitfalls of different technologies for a clinical interpretation of genomic sequence data of individual gastric cancer samples. Using different filtering approaches, we identified a multitude of novel potentially damaging mutations and could show a validated association between a mutation in GNAS and gastric cancer.

MET in gastric cancer--discarding a 10% cutoff rule.

AIMS: We aimed to develop a putative predictive biomarker score for future hepatocyte growth factor receptor (MET)-targeted therapy of gastric cancer (GC). METHODS AND RESULTS: MET expression and MET amplification were analysed by immunohistochemistry (IHC) and chromogenic in-situ hybridization (CISH) in 470 GC patients. Immunostaining was documented with the HistoScore. The percentage area of MET-amplified tumour cell clones was assessed by virtual microscopy. The expression of MET was heterogeneous in primary and metastatic GC. Immunostaining intensity (MET-IHC 2+/3+) correlated with MET amplification and a positive MET status was defined by a combination of MET-IHC 2+ or 3+ with MET amplification, or MET-IHC 3+ without MET amplification. The prognostic significance of the MET status was independent from the percentage area of positive tumour cells (e.g. <10 versus ≥10%). MET-positive GCs were microsatellite stable and of KRAS/PIK3CA wild-type. MET-positive GCs had a very poor prognosis, with a median survival of 5.4 months and a hazard ratio of 2.126. CONCLUSIONS: A combination of immunohistochemistry and CISH is suitable to assess MET status. If MET status is used as a predictive biomarker, prospective studies should pay specific attention to adequate tissue sampling, should ignore cutoff values for tumour areas, may consider the KRAS and PIK3CA genotype as negative predictive markers and should carry out the analysis expeditiously.

Multicellular ovarian cancer spheroids: novel 3D model to mimic tumour complexity.

In vitro, spheroid models have become well established in cancer research because they can better mimic certain characteristics of in vivo tumours. However, interaction with the tumour microenvironment, such as cancer-associated fibroblasts, plays a key role in tumour progression. We initially focused on the interaction of tumour cells with fibroblasts. To model this interaction, we developed a spheroid model of ovarian cancer and fibroblasts. To this end, ovarian cancer cell lines and ex vivo primary cells were simultaneously and sequentially seeded with fibroblasts in a scaffold-free system at different ratios and subsequently characterized with respect to changes in morphology, proliferation, and viability. We demonstrated that co-cultures are able to form by far more compact spheroids, especially in cells that form aggregates in mono-culture. In addition, the co-cultures were able to increase proliferation and sensitivity to cisplatin. Simultaneous seeding led fibroblasts invade the core in both cell lines and primary cells. These results show differences in formation, firmness, and size between co-culture and mono-culture. Our model is designed to better represent and characterize the mutual influencing factors of fibroblasts and tumour cells. Fibroblast-supplemented multicellular spheroids are a valuable tool for tumour microenvironment interaction and new drug discovery.

MDM2 amplification is rare in gastric cancer.

The MDM2 proto-oncogene (MDM2) is a primary negative regulator of p53. The latter is frequently mutated in gastric cancer (GC). In the present study, we aimed to validate gene amplification, protein expression, and the putative tumor biological function of MDM2 in a well-characterized Western GC cohort. MDM2 amplification and protein expression were studied in a cohort of 327 GCs by fluorescence in situ hybridization (FISH) and immunohistochemistry. Gene amplification and protein expression were correlated with diverse clinicopathological patient characteristics including patient outcome. Immunohistochemically, 97 GCs (29.7%) were categorized as MDM2 positive and 230 GCs (70.3%) as negative. An amplification of MDM2 was found in 11 (3.4%) cases without evidence of intratumoral heterogeneity. Nine of these eleven (81.8%) cases showed MDM2 protein expression. MDM2 amplification correlated significantly with MDM2 protein expression (p < 0.001). On a case-by-case analysis, MDM2-amplified cases showed varied histological phenotypes and were most commonly microsatellite stable; EBV, HER2, and MET negative; and FGFR2 positive. A single case harbored both, MDM2 amplification and TP53 mutation. MDM2 amplification and MDM2 expression, respectively, did not correlate with overall or tumor-specific survival. Our targeted analysis of MDM2 in a well-characterized cohort of GC patients showed that MDM2 amplification is rare, of no specific histological phenotype, and may not be always mutually exclusive with TP53 mutations. Given the low number of cases, currently, no diagnostic or therapeutic recommendation related to MDM2 amplification can be given for GC of Western origin.

The tumor biological significance of RNF43 and LRP1B in gastric cancer is complex and context-dependent.

Gastric cancer (GC) is the fifth most common cancer in the world with a poor prognosis. Both RNF43 and LRP1B function as tumor suppressors in the Wnt signaling pathway and have been described to be frequently mutated in GC. In this study of a large and well characterized cohort of 446 GCs we explored the significance of expression of RNF43 and LRP1B and their correlations with clinicopathological patient characteristics. Immunostaining of whole mount tissue sections was documented with the histoscore. Dichotomized at the median, we separated the cohort into a low/negative and a high/positive group of RNF43 and LRP1B expression, respectively. Apart from the entire cohort, we also examined the intestinal and diffuse type GCs separately. Regarding the entire cohort, the expression of RNF43 and LRP1B correlated significantly with the Lauren phenotype and with each other. Interestingly, differences were noted regarding RNF43 between the intestinal and diffuse type GCs. Survival analysis of the intestinal type GCs showed that RNF43 low/negative GCs tended to have a better outcome compared with RNF43 high/positive GCs [24.5 months overall survival (OS) and 25.0 months tumor-specific survival (TSS) vs. 14.1 months OS and 17.9 months TSS, respectively]. To the contrary, diffuse type GCs with RNF43 low/negative had a worse outcome compared with RNF43 high/positive GCs (12.9 months OS and 18.2 months TSS vs. 17.1 months OS and 21.5 months TSS, respectively). On multivariate analysis, RNF43 low/negative versus high/positive was an independent prognosticator of survival in diffuse type GC (hazard ratio 2.393 for OS and 2.398 for TSS). These data support the contention that the expression and biological effect of RNF43 and LRP1B in GC is context-dependent.

SMARCA4 and SMARCE1 in gastric cancer: Correlation with ARID1A, and microsatellite stability, and SMARCE1/ERBB2 co-amplification.

BACKGROUND: Recent studies have shown an association between certain subunits of the SWI/SNF complex with specific tumor characteristics in gastric cancer (GC). In an earlier study, we applied multiregional whole exome sequencing on multiple primary tumor samples and found alterations of the SWI/SNF complex in 78% of the cases. ERBB2, which encodes for Her2/neu, is a well-known predictive biomarker used to guide the treatment of GC in the palliative setting. SMARCE1, which encodes for a subunit of the SWI/SNF complex, is localized in close genetic proximity to ERBB2. AIM: As little is known about the significance of the SWI/SNF complex in GC biology and the potential relationship between ERBB2 and SMARCE1 upregulation, we examined the expression patterns of SMARCA4 and SMARCE1, two mutually exclusive catalytic ATPase subunits of the SWI/SNF complex, in a well characterized GC cohort. MATERIALS AND METHODS: The expression of SMARCA4 and SMARCE1 was studied by immunohistochemistry in connection with clinicopathological patient characteristics in a cohort of 468 GCs. Digital droplet polymerase chain reaction was performed for amplification analysis on ERBB2 and SMARCE1. RESULTS: Immunohistochemical staining of whole-mount tissue sections found a diffusely "gray scale" expression of SMARCA4 in 446 (95.2%) GCs and of SMARCE1 in 463 (98.8%) GCs. The expression of SMARCA4 and SMARCE1 correlated significantly with ARID1A, p53, and microsatellite status. No correlation was found with the patient prognosis. The amplification analysis of SMARCE1 showed amplification in 4 of 34 cases. In 3 of 34 cases, SMARCE1 was co-amplified with ERBB2. We also found a co-expression of SMARCE1 and Her2/neu in a subset of patients. CONCLUSION: While the effect of a co-amplification is currently unknown, synergistic effects of SMARCE1 and Her2/neu overexpression should be explored in future studies, holding potential for an improved treatment of GC.

LAG3 in gastric cancer: it's complicated.

PURPOSE: Lymphocyte activation gene 3 (LAG3) is thought to contribute to T cell exhaustion within the tumor microenvironment of solid tumors. This study aimed to analyze the spatial distribution of LAG3 + cells in relation to clinicopathological and survival data in a large set of 580 primary resected and neoadjuvantly treated gastric cancers (GC). METHODS: LAG3 expression was evaluated in tumor center and invasive margin using immunohistochemistry and whole-slide digital image analysis. Cases were divided into LAG3-low and LAG3-high expression groups based on (1) median LAG3 + cell density, (2) cut-off values adapted to cancer-specific survival using Cutoff Finder application. RESULTS: Significant differences in spatial distribution of LAG3 + cells were observed in primarily resected GC, but not in neoadjuvantly treated GC. LAG3 + cell density showed evident prognostic value at following cut-offs: in primarily resected GC, 21.45 cells/mm2 in tumor center (17.9 vs. 10.1 months, p = 0.008) and 208.50 cells/mm2 in invasive margin (33.8 vs. 14.7 months, p = 0.006); and in neoadjuvantly treated GC, 12.62 cells/mm2 (27.3 vs. 13.2 months, p = 0.003) and 123.00 cells/mm2 (28.0 vs. 22.4 months, p = 0.136), respectively. Significant associations were found between LAG3 + cell distribution patterns and various clinicopathological factors in both cohorts. In neoadjuvantly treated GC, LAG3 + immune cell density was found to be an independent prognostic factor of survival (HR = 0.312, 95% CI 0.162-0.599, p < 0.001). CONCLUSION: In this study, a higher density of LAG3 + cells was associated with favorable prognosis. Current results support the need for extended analysis of LAG3. Differences in the distribution of LAG3 + cells should be considered, as they could influence clinical outcomes and treatment responses.

Hepatic myofibroblasts exert immunosuppressive effects independent of the immune checkpoint regulator PD-L1 in liver metastasis of pancreatic ductal adenocarcinoma.

Introduction: Pancreatic ductal adenocarcinoma (PDAC) represents the 4th most common cause of cancer-related deaths in Western countries. Most patients are diagnosed at advanced stages, often already with metastases. The main site of metastasis is the liver and hepatic myofibroblasts (HMF) play a pivotal role in metastatic outgrowth. Immune checkpoint inhibitors (ICI) targeting programmed death ligand 1 (PD-L1) or programmed cell death protein 1 (PD-1) improved treatment of several cancers but not of PDAC. Therefore, this study aimed to better understand the impact of HMF on PD-L1 expression and immune evasion of PDAC cells during liver metastasis. Methods: Formalin-fixed and paraffin embedded biopsy samples or diagnostic resection specimens from liver metastases of 15 PDAC patients were used for immunohistochemical analyses. Serial sections were stained with antibodies directed against Pan-Cytokeratin, αSMA, CD8, and PD-L1. To investigate whether the PD-1/PD-L1 axis and HMF contribute to immune escape of PDAC liver metastases, a stroma enriched 3D spheroid coculture model was established in vitro, using two different PDAC cell lines, HMF, and CD8+ T cells. Here, functional and flow cytometry analyses were conducted. Results: Immunohistochemical analysis of liver tissue sections of PDAC patients revealed that HMF represent an abundant stroma population in liver metastases, with clear differences in the spatial distribution in small (1500 µm) and large (> 1500 µm) metastases. In the latter, PD-L1 expression was mainly located at the invasion front or evenly distributed, while small metastases either lacked PD-L1 expression or showed mostly weak expression in the center. Double stainings revealed that PD-L1 is predominantly expressed by stromal cells, especially HMF. Small liver metastases with no or low PD-L1 expression comprised more CD8+ T cells in the tumor center, while large metastases exhibiting stronger PD-L1 expression comprised less CD8+ T cells being mostly located at the invasion front. HMF-enriched spheroid cocultures with different ratios of PDAC cells and HMF well mimicking conditions of hepatic metastases in situ. Here, HMF impaired the release of effector molecules by CD8+ T cells and the induction of PDAC cell death, an effect that was dependent on the amount of HMF but also of PDAC cells. ICI treatment led to elevated secretion of distinct CD8+ T cell effector molecules but did not increase PDAC cell death under either spheroid condition. Conclusion: Our findings indicate a spatial reorganization of HMF, CD8+ T cells, and PD-L1 expression during progression of PDAC liver metastases. Furthermore, HMF potently impair the effector phenotype of CD8+ T cells but the PD-L1/PD-1 axis apparently plays a minor role in this scenario suggesting that immune evasion of PDAC liver metastases relies on other immunosuppressive mechanisms.

Expression and role of the immune checkpoint regulator PD-L1 in the tumor-stroma interplay of pancreatic ductal adenocarcinoma.

Introduction: Immune checkpoint inhibitors (ICI), e.g., targeting programmed cell death protein 1-ligand 1 (PD-L1) or its receptor PD-1, have markedly improved the therapy of many cancers but so far failed in pancreatic ductal adenocarcinoma (PDAC). Macrophages represent one of the most abundant immune cell populations within the tumor microenvironment (TME) of PDAC being able to either support or restrain tumor progression depending on their phenotype. To better understand treatment failure of PD-L1/PD-1 inhibitors in PDAC, this study examined PD-L1 expression in the context of a dynamic TME in PDAC with a particular focus on the impact of macrophages. Methods: Formalin-fixed and paraffin embedded tissue samples of primary PDAC tissues and corresponding liver metastases were used for immunohistochemical analyses. Serial sections were stained with antibodies detecting Pan-Cytokeratin, CD68, CD163, CD8, and PD-L1.To investigate whether the PD-1/PD-L1 axis and macrophages contribute to immune escape of PDAC cells, a stroma enriched 3D spheroid coculture model was established in vitro, using different PDAC cell lines and macrophages subtypes as well as CD8+ T cells. Functional and flow cytometry analyses were conducted to characterize cell populations. Results: Immunohistochemical analyses revealed that PD-L1 is mainly expressed by stroma cells, including macrophages and not PDAC cells in primary PDAC tissues and corresponding liver metastases. Notably, high local abundance of macrophages and strong PD-L1 staining were commonly found at invasion fronts of tumoral lesions between CD8+ T cells and tumor cells. In order to investigate whether PD-L1 expressing macrophages impact the response of PDAC cells to treatment with PD-L1/PD-1 inhibitors, we developed a spheroid model comprising two different PDAC cell lines and different ratios of in vitro differentiated primary M1- or M2-like polarized macrophages. In line with our in situ findings, high PD-L1 expression was observed in macrophages rather than PDAC cells, which was further increased by the presence of PDAC cells. The effector phenotype of co-cultured CD8+ T cells exemplified by expression of activation markers and release of effector molecules was rather enhanced by PDAC macrophage spheroids, particularly with M1-like macrophages compared to mono-culture spheroids. However, this was not associated with enhanced PDAC cell death. ICI treatment with either Durvalumab or Pembrolizumab alone or in combination with Gemcitabine hardly affected the effector phenotype of CD8+ T cells along with PDAC cell death. Thus, despite strong PD-L1 expression in macrophages, ICI treatment did not result in an enhanced activation and cytotoxic phenotype of CD8+ T cells. Conclusion: Overall, our study revealed novel insights into the interplay of PDAC cells and macrophages in the presence of ICI.