Changes in the immune landscape of TNBC after neoadjuvant chemotherapy: correlation with relapse.

Introduction: Patients with high-risk, triple negative breast cancer (TNBC) often receive neoadjuvant chemotherapy (NAC) alone or with immunotherapy. Various single-cell and spatially resolved techniques have demonstrated heterogeneity in the phenotype and distribution of macrophages and T cells in this form of breast cancer. Furthermore, recent studies in mice have implicated immune cells in perivascular (PV) areas of tumors in the regulation of metastasis and anti-tumor immunity. However, little is known of how the latter change during NAC in human TNBC or their impact on subsequent relapse, or the likely efficacy of immunotherapy given with or after NAC. Methods: We have used multiplex immunofluorescence and AI-based image analysis to compare the immune landscape in untreated and NAC-treated human TNBCs. We quantified changes in the phenotype, distribution and intercellular contacts of subsets of tumor-associated macrophages (TAMs), CD4+ and CD8+ T cells, and regulatory T cells (Tregs) in PV and non-PV various areas of the stroma and tumor cell islands. These were compared in tumors from patients who had either developed metastases or were disease-free (DF) after a three-year follow up period. Results: In tumors from patients who remained DF after NAC, there was a marked increase in stromal CD163+ TAMs, especially those expressing the negative checkpoint regulator, T-cell immunoglobulin and mucin domain 3 (TIM-3). Whereas CD4+ T cells preferentially located to PV areas in the stroma of both untreated and NAC-treated tumors, specific subsets of TAMs and Tregs only did so only after NAC. Distinct subsets of CD4+ and CD8+ T cells formed PV clusters with CD163+ TAMs and Tregs. These were retained after NAC. Discussion: Quantification of stromal TIM-3+CD163+ TAMs in tumor residues after NAC may represent a new way of identifying patients at high risk of relapse. PV clustering of immune cells is highly likely to regulate the activation and function of T cells, and thus the efficacy of T cell-based immunotherapies administered with or after NAC.

An integrated gene-to-outcome multimodal database for metabolic dysfunction-associated steatotic liver disease.

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the commonest cause of chronic liver disease worldwide and represents an unmet precision medicine challenge. We established a retrospective national cohort of 940 histologically defined patients (55.4% men, 44.6% women; median body mass index 31.3; 32% with type 2 diabetes) covering the complete MASLD severity spectrum, and created a secure, searchable, open resource (SteatoSITE). In 668 cases and 39 controls, we generated hepatic bulk RNA sequencing data and performed differential gene expression and pathway analysis, including exploration of gender-specific differences. A web-based gene browser was also developed. We integrated histopathological assessments, transcriptomic data and 5.67 million days of time-stamped longitudinal electronic health record data to define disease-stage-specific gene expression signatures, pathogenic hepatic cell subpopulations and master regulator networks associated with adverse outcomes in MASLD. We constructed a 15-gene transcriptional risk score to predict future hepatic decompensation events (area under the receiver operating characteristic curve 0.86, 0.81 and 0.83 for 1-, 3- and 5-year risk, respectively). Additionally, thyroid hormone receptor beta regulon activity was identified as a critical suppressor of disease progression. SteatoSITE supports rational biomarker and drug development and facilitates precision medicine approaches for patients with MASLD.

Host obesity alters the ovarian tumor immune microenvironment and impacts response to standard of care chemotherapy.

BACKGROUND: The majority of women with epithelial ovarian cancer (OvCa) are diagnosed with metastatic disease, resulting in a poor 5-year survival of 31%. Obesity is a recognized non-infectious pandemic that increases OvCa incidence, enhances metastatic success and reduces survival. We have previously demonstrated a link between obesity and OvCa metastatic success in a diet-induced obesity mouse model wherein a significantly enhanced tumor burden was associated with a decreased M1/M2 tumor-associated macrophage ratio (Liu Y et al. Can, Res. 2015; 75:5046-57). METHODS: The objective of this study was to use pre-clinical murine models of diet-induced obesity to evaluate the effect of a high fat diet (HFD) on response to standard of care chemotherapy and to assess obesity-associated changes in the tumor microenvironment. Archived tumor tissues from ovarian cancer patients of defined body mass index (BMI) were also evaluated using multiplexed immunofluorescence analysis of immune markers. RESULTS: We observed a significantly diminished response to standard of care paclitaxel/carboplatin chemotherapy in HFD mice relative to low fat diet (LFD) controls. A corresponding decrease in the M1/M2 macrophage ratio and enhanced tumor fibrosis were observed both in murine DIO studies and in human tumors from women with BMI > 30. CONCLUSIONS: Our data suggest that the reported negative impact of obesity on OvCa patient survival may be due in part to the effect of the altered M1/M2 tumor-associated macrophage ratio and enhanced fibrosis on chemosensitivity. These data demonstrate a contribution of host obesity to ovarian tumor progression and therapeutic response and support future combination strategies targeting macrophage polarization and/or fibrosis in the obese host.

CCF-GNN: A Unified Model Aggregating Appearance, Microenvironment, and Topology for Pathology Image Classification.

Pathology images contain rich information of cell appearance, microenvironment, and topology features for cancer analysis and diagnosis. Among such features, topology becomes increasingly important in analysis for cancer immunotherapy. By analyzing geometric and hierarchically structured cell distribution topology, oncologists can identify densely-packed and cancer-relevant cell communities (CCs) for making decisions. Compared to commonly-used pixel-level Convolution Neural Network (CNN) features and cell-instance-level Graph Neural Network (GNN) features, CC topology features are at a higher level of granularity and geometry. However, topological features have not been well exploited by recent deep learning (DL) methods for pathology image classification due to lack of effective topological descriptors for cell distribution and gathering patterns. In this paper, inspired by clinical practice, we analyze and classify pathology images by comprehensively learning cell appearance, microenvironment, and topology in a fine-to-coarse manner. To describe and exploit topology, we design Cell Community Forest (CCF), a novel graph that represents the hierarchical formulation process of big-sparse CCs from small-dense CCs. Using CCF as a new geometric topological descriptor of tumor cells in pathology images, we propose CCF-GNN, a GNN model that successively aggregates heterogeneous features (e.g., appearance, microenvironment) from cell-instance-level, cell-community-level, into image-level for pathology image classification. Extensive cross-validation experiments show that our method significantly outperforms alternative methods on H&E-stained and immunofluorescence images for disease grading tasks with multiple cancer types. Our proposed CCF-GNN establishes a new topological data analysis (TDA) based method, which facilitates integrating multi-level heterogeneous features of point clouds (e.g., for cells) into a unified DL framework.

Tissue-infiltrating neutrophils constitute the major in vivo source of angiogenesis-inducing MMP-9 in the tumor microenvironment.

According to established notion, one of the major angiogenesis-inducing factors, pro-matrix metalloproteinase-9 (proMMP-9), is supplied to the tumor microenvironment by tumor-associated macrophages (TAMs). Accumulated evidence, however, indicates that tumor-associated neutrophils (TANs) are also critically important for proMMP-9 delivery, especially at early stages of tumor development. To clarify how much angiogenic proMMP-9 is actually contributed by TAMs and TANs, we quantitatively evaluated TAMs and TANs from different tumor types, including human xenografts and syngeneic murine tumors grown in wild-type and Mmp9-knockout mice. Whereas host MMP-9 competence was required for full angiogenic potential of both normal and tumor-associated leukocytes, direct comparisons of neutrophils versus macrophages and TANs versus TAMs demonstrated that macrophages and TAMs secrete 40- to 50-fold less proMMP-9 than the same numbers of neutrophils or TANs. Correspondingly, the levels of MMP-9-mediated in vivo angiogenesis induced by neutrophils and TANs substantially exceeded those induced by macrophages and TAMs. MMP-9-delivering TANs were also required for development of metastasis-supporting intratumoral vasculature, characterized by ≥ 11-µm size lumens and partial coverage with stabilizing pericytes. Importantly, MMP-9-producing TAMs exhibit M2-skewed phenotype but do not express tissue inhibitor of metalloproteinases-1 (TIMP-1), a novel characteristic allowing them to secrete TIMP-1-free, neutrophil-like MMP-9 zymogen unencumbered by its natural inhibitor. Together, our findings support the notion whereby TANs, capable of immediate release of their pre-stored cargo, are the major contributors of highly angiogenic MMP-9, whereas tumor-influxing precursors of macrophages require time to differentiate, polarize into M2-skewed TAMs, shut down their TIMP-1 expression, and only then, initiate relatively low-level production of TIMP-free MMP-9 zymogen.

Angiogenic capacity of M1- and M2-polarized macrophages is determined by the levels of TIMP-1 complexed with their secreted proMMP-9.

A proangiogenic function of tissue-infiltrating monocytes/macrophages has long been attributed to their matrix metalloproteinase-9 zymogen (proMMP-9). Herein, we evaluated the capacity of human monocytes, mature M0 macrophages, and M1- and M2-polarized macrophages to induce proMMP-9-mediated angiogenesis. Only M2 macrophages induced angiogenesis at levels comparable with highly angiogenic neutrophils previously shown to release their proMMP-9 in a unique form, free of tissue inhibitor of metalloproteinases-1 (TIMP-1). Macrophage differentiation was accompanied by induction of low-angiogenic, TIMP-1-encumbered proMMP-9. However, polarization toward the M2, but not the M1 phenotype, caused a substantial downregulation of TIMP-1 expression, resulting in production of angiogenic, TIMP-deficient proMMP-9. Correspondingly, the angiogenic potency of M2 proMMP-9 was lost after its complexing with TIMP-1, whereas TIMP-1 silencing in M0/M1 macrophages rendered them both angiogenic. Similar to human cells, murine bone marrow-derived M2 macrophages also shut down their TIMP-1 expression and produced proMMP-9 unencumbered by TIMP-1. Providing proof that angiogenic capacity of murine M2 macrophages depended on their TIMP-free proMMP-9, Mmp9-null M2 macrophages were nonangiogenic, although their TIMP-1 was severely downregulated. Our study provides a unifying molecular mechanism for high angiogenic capacity of TIMP-free proMMP-9 that would be uniquely produced in a pathophysiological microenvironment by influxing neutrophils and/or M2 polarized macrophages.

Tumor MMP-1 activates endothelial PAR1 to facilitate vascular intravasation and metastatic dissemination.

Intravasation, the active entry of primary tumor cells into the vasculature, remains the least studied step in the metastatic cascade. Protease-mediated escape and stromal invasion of tumor cells represent widely accepted processes leading up to the intravasation step. However, molecular factors that contribute directly to tumor cell vascular penetration have not been identified. In this study, the in vivo role of the collagenolytic protease, MMP-1, in cancer cell intravasation and metastasis was analyzed by using a highly disseminating variant of human HEp3 epidermoid carcinoma, HEp3-hi/diss. Although naturally acquired or experimentally induced MMP-1 deficiency substantially suppressed HEp3-hi/diss intravasation, supplementation of recombinant MMP-1 to MMP-1-silenced primary tumors restored their impaired vascular dissemination. Surprisingly, abrogation of MMP-1 production and activity did not significantly affect HEp3-hi/diss migration or matrix invasion, suggesting noncollagenolytic mechanisms underlying MMP-1-dependent cell intravasation. In support of such noncollagenolytic mechanisms, MMP-1 silencing in HEp3-hi/diss cells modulated the microarchitecture and integrity of the angiogenic vasculature in a novel microtumor model. Concomitantly, MMP-1 deficiency led to decreased levels of intratumoral vascular permeability, tumor cell intravasation, and metastatic dissemination. Taking advantage of PAR1 deficiency of HEp3-hi/diss cells, we further show that endothelial PAR1 is a putative nontumor-cell/nonmatrix target, activation of which by carcinoma-produced MMP-1 regulates endothelial permeability and transendothelial migration. The inhibitory effects of specific PAR1 antagonists in live animals have also indicated that the mechanisms of MMP-1-dependent vascular permeability in tumors involve endothelial PAR1 activation. Together, our findings mechanistically underscore the contribution of a tumor MMP-1/endothelial PAR1 axis to actual intravasation events manifested by aggressive carcinoma cells.

Spontaneous metastasis in congenic mice with transgenic breast cancer is unaffected by plasminogen gene ablation.

Plasminogen (Plg) plays a central role in tissue remodeling during ontogeny, development, and in pathological tissue remodeling following physical injury, inflammation and cancer. Plg/plasmin is, however, not critical for these processes, as they all occur to a varying extent in its absence, suggesting that there is a functional redundancy with other proteases. To explore this functional overlap in the transgenic MMTV-PyMT breast cancer metastasis model, we have combined Plg deficiency and a pharmacological metalloprotease inhibitor, which is known to reduce metastasis in this model, and has been shown to synergistically inhibit other tissue remodeling events in Plg-deficient mice. While metalloprotease inhibition dramatically reduced metastasis, we found no effect of Plg deficiency on metastasis, either independently or in combination with metalloprotease inhibition. We further show that Plg gene deficiency is of no significant consequence in this metastasis model, when analyzed in two different congenic strains: the FVB strain, and a F1 hybrid of the FVB and C57BL/6J strains. We suggest that the extensive backcrossing performed prior to our studies has eliminated the confounding effect of a known polymorphic metastasis modifier gene region located adjacent to the Plg gene.

No effect of NGAL/lipocalin-2 on aggressiveness of cancer in the MMTV-PyMT/FVB/N mouse model for breast cancer.

NGAL/lipocalin-2 is a siderophore-binding protein that is highly expressed in several cancers. It is suggested to confer a proliferative advantage to cancer cells. Its expression has been correlated with aggressiveness of breast cancer as determined both in patients and in mouse breast cancer models. This was recently confirmed in two mouse models of spontaneous breast cancer in wild-type and lipocalin-2-deficient mice. We used a similar strategy using a different mouse strain. Lipocalin-2-deficient mice and mouse mammary tumor virus-polyoma middle T antigen (MMTV-PyMT) mice were crossed into the same FVB/N background. All mice developed tumors by week 8. The mice were sacrificed on week 13 and tissue was processed for biochemical and histological analysis. The total tumor volume and number of metastases were quantitated in 26 lipocalin-2-deficient mice and 34 wild-type controls. Lipocalin-2 expression in tumors of MMTV-PyMT-positive and wild-type mice was assessed by quantitative real-time PCR and by immunohistochemistry. The expression of the lipocalin-2 receptors 24p3R and megalin and of Mmp-9, transferrin receptor, and Bdh2 (a producer of a mammalian siderophore) were quantitated by real-time PCR. No significant difference was observed between wild-type and lipocalin-2-deficient mice. Lipocalin-2 was highly expressed in tumors from wild-type mice, but the expression did not correlate with tumor size. No effect of lipocalin-2 was observed with respect to time to tumor appearance, total tumor volume, or to the number of metastases. Histology and gelatinolytic activity of the mammary tumors did not differ between wild-type and lipocalin-2-deficient mice. We conclude that NGAL/lipocalin-2 does not invariably affect the aggressiveness of breast cancers as assessed in mouse models, thus questioning the role of lipocalin-2 in cancer development.

Phenotypic overlap between MMP-13 and the plasminogen activation system during wound healing in mice.

BACKGROUND: Proteolytic degradation of extracellular matrix is a crucial step in the healing of incisional skin wounds. Thus, healing of skin wounds is delayed by either plasminogen-deficiency or by treatment with the broad-spectrum metalloproteinase (MP) inhibitor Galardin alone, while the two perturbations combined completely prevent wound healing. Both urokinase-type plasminogen activator and several matrix metallo proteinases (MMPs), such as MMP-3, -9 and -13, are expressed in the leading-edge keratinocytes of skin wounds, which may account for this phenotypic overlap between these classes of proteases. METHODOLOGY: To further test that hypothesis we generated Mmp13;Plau and Mmp13;Plg double-deficient mice in a cross between Mmp13- and Plau-deficient mice as well as Mmp13- and Plg-deficient mice. These mice were examined for normal physiology in a large cohort study and in a well-characterized skin wound healing model, in which we made incisional 20 mm-long full-thickness skin wounds. PRINCIPAL FINDINGS: While mice that are deficient in Mmp13 have a mean healing time indistinguishable to wild-type mice, wound healing in both Plau- and Plg-deficient mice is significantly delayed. Histological analysis of healed wounds revealed a significant increase in keratin 10/14 immunoreactive layers of kerationcytes in the skin surface in Mmp13;Plau double-deficient mice. Furthermore, we observe, by immunohistological analysis, an aberrant angiogenic pattern during wound healing induced by Plau-deficiency, which has not previously been described. CONCLUSIONS: We demonstrate a phenotypic overlap, defined as an additional delay in wound healing in the double-deficient mice compared to the individual single-deficient mice, between MMP-13 and the plasminogen activation system in the process of wound healing, but not during gestation and in postnatal development. Thus, a dual targeting of uPA and MMP-13 might be a possible future strategy in designing therapies aimed at tissue repair or other pathological processes, such as cancer invasion, where proteolytic degradation is a hallmark.

Spontaneous metastasis in matrix metalloproteinase 3-deficient mice.

Matrix metalloproteinases (MMPs) have been linked to the metastatic potential of tumor cells due to their ability to degrade the extracellular matrix. MMP-3 (stromelysin-1) is upregulated in a wide variety of human tumors. We used the MMTV-PyMT breast cancer model to determine if MMP-3 is involved in tumorigenesis and metastatic growth. In this model the stromal expression of MMP-3 mRNA resembles the predominant MMP-3 expression pattern observed in human ductal breast carcinomas. We studied a cohort of 63 PyMT transgenic mice, either deficient for MMP-3 or wild-type controls. The degree of metastasis did not differ significantly between the two groups of mice, although the median lung metastasis volume was more than threefold increased in MMTV-PyMT mice deficient in MMP-3. Likewise, primary tumor growth rate and lymph node metastasis were not significantly affected by MMP-3-deficiency. By comparing mRNA levels in MMP-3-deficient PyMT tumors with PyMT wild-type tumors we excluded compensatory transcriptional changes of other MMPs or their specific inhibitors. Thus, we conclude that genetic ablation of MMP-3 does not significantly affect tumor growth and metastasis in the MMTV-PyMT model.

Metastasis is strongly reduced by the matrix metalloproteinase inhibitor Galardin in the MMTV-PymT transgenic breast cancer model.

Matrix metalloproteinases (MMP) have several roles that influence cancer progression and dissemination. However, low molecular weight metalloproteinase inhibitors (MPI) have not yet been tested in transgenic/spontaneous metastasis models. We have tested Galardin/GM6001, a potent MPI that reacts with most MMPs, in the MMTV-PymT transgenic breast cancer model. We followed a cohort of 81 MMTV-PymT transgenic mice that received Galardin, placebo, or no treatment. Galardin treatment was started at age 6 weeks with 100 mg/kg/d, and all mice were killed at age 13.5 weeks. Galardin treatment significantly reduced primary tumor growth. Final tumor burden in Galardin-treated mice was 1.69 cm3 compared with 3.29 cm3 in placebo-treated mice (t test, P = 0.0014). We quantified the total lung metastasis volume in the same cohort of mice. The median metastasis volume was 0.003 mm(3) in Galardin-treated mice compared with 0.56 mm(3) in placebo-treated mice (t test, P < 0.0001). Thus, metastasis burden was reduced more than 100-fold, whereas primary tumor size was reduced only 2-fold. We also found that primary tumors from Galardin-treated mice exhibited a lower histopathologic tumor grade, increased collagen deposition, and increased MMP-2 activity. MMPs are known to have tumor-promoting and tumor-inhibitory effects, and several clinical trials of broad-spectrum MPIs have failed to show promising effects. The very potent antimetastatic effect of Galardin in the MMTV-PymT model does, however, show that it may be possible to find broad-spectrum MPIs with favorable inhibition profiles, or perhaps combinations of monospecific MPIs, for future clinical application.

Extracellular proteolysis in transgenic mouse models of breast cancer.

Growth and invasion of breast cancer require extracellular proteolysis in order to physically restructure the tissue microenvironment of the mammary gland. This pathological tissue remodeling process depends on a collaboration of epithelial and stromal cells. In fact, the majority of extracellular proteases are provided by stromal cells rather than cancer cells. This distinct expression pattern is seen in human breast cancers and also in transgenic mouse models of breast cancer. The similar expression patterns suggest that transgenic mouse models are ideally suited to study the role of extracellular proteases in cancer progression. Here we give a status report on protease intervention studies in transgenic models. These studies demonstrate that proteases are involved in all stages of breast cancer progression from carcinogenesis to metastasis. Transgenic models are now beginning to provide vital mechanistic insight that will allow us to combat breast cancer invasion and metastasis with new protease-targeted drugs.