|Home | About | Journals | Submit | Contact Us | Français|
The association between malignancy and development of a paraneoplastic leukocytosis, the so-called leukemoid reaction, has long been appreciated. Although a leukemoid reaction has conventionally been defined as a peripheral blood leukocytosis composed of both mature and immature granulocytes that exceeds 50,000/μL, a less profound leukocytosis may be appreciated in many patients harboring a malignant disease. More recent insights have shed new light on this long-recognized association, because research performed in both murine models and cancer patients has uncovered multiple mechanisms by which tumors both drive myelopoiesis, sometimes leading to a clinically apparent leukocytosis, and inhibit the differentiation of myeloid cells, resulting in a qualitative change in myelopoiesis. This qualitative change leads to the accumulation of immature myeloid cells, which due to their immune suppressive effects have been collectively called myeloid-derived suppressor cells. More recently, myeloid cells have been shown to promote tumor angiogenesis. Cancer-associated myeloproliferation is not merely a paraneoplastic phenomenon of questionable importance but leads to the suppression of host immunity and promotion of tumor angiogenesis, both of which play an integral part in tumorigenesis and metastasis. Therefore, cancer-associated myeloproliferation represents a novel therapeutic target in cancer that, decades after its recognition, is only now being translated into clinical practice.
ARG = arginase; ATRA = all-trans-retinoic acid; CSF = colony-stimulating factor; EG = endocrine gland; G-CSF = granulocyte-CSF; IL = interleukin; iNOS = inducible nitric oxide synthase; JAK = Janus kinase; MDSC = myeloid-derived suppressor cell; PKR = prokineticin; STAT = signal transducers and activators of transcription; VEGF = vascular endothelial growth factor
Approximately 20% of patients with advanced non–small cell lung cancer, the leading cause of cancer deaths worldwide, may have an associated leukocytosis.1,2 Conversely, the finding of a relative neutrophilia (or lymphopenia) may help to distinguish a malignant lesion from a benign lesion in patients presenting with undiagnosed lung masses.3 More importantly, a large, pooled analysis of pretreatment prognostic factors that used data collected from patients with advanced non–small cell lung cancer treated in several North Central Cancer Treatment Group clinical trials demonstrated, on both univariate and multivariate analyses, that leukocytosis at diagnosis is associated with inferior overall survival.2 Similarly, tumor-associated neutrophilia or monocytosis is an adverse prognostic feature in metastatic melanoma,4,5 small cell lung cancer,6 colorectal cancer,7,8 head and neck cancer,9 non-Hodgkin lymphoma,10 and renal cell cancer11,12 with tumor infiltration by these myeloid-derived cells portending an equally poor prognosis.11,13-17
Work performed using a murine lung cancer model in the 1980s revealed that the spleens of tumor-bearing mice became hypercellular within a few weeks of tumor engraftment.18 This was associated with an elevated peripheral blood leukocyte count and a relative monocytosis secondary to a substantial increase in hematopoiesis in these animals. Furthermore, supernatants collected from cultured tumor cells stimulated myelopoiesis in vitro, with more than 50% of the cells generated belonging to either the granulocyte or monocyte/macrophage lineage. This expanded population of myeloid-lineage cells was subsequently shown to inhibit T-cell proliferation. Contemporaneous work describing the production of hematopoietic cytokines by non–small cell lung carcinomas seemed to corroborate these findings.19-25 Collectively, these clinical observations and corroborative laboratory studies linked carcinogenesis and tumor progression with myelopoiesis. Although it is doubtful that the importance of these observations was fully appreciated at the time,26 the observation that neutrophil depletion in tumor-bearing animals could restrain tumor growth provided important clues that would not be fully appreciated until the identification of myeloid-derived suppressor cells (MDSCs), the focus of this review, many years later.27
To identify relevant literature for this review, PubMed was searched using the following keywords: leukemoid reaction, leukocytosis, neutrophilia, monocytosis, immature myeloid cell, myeloid-derived suppressor cell, lymphocyte, angiogenesis, prognosis, treatment, review, cytokine, tumor, and cancer (including specific histologies, eg, non–small cell lung cancer, melanoma, renal cell cancer, non-Hodgkin lymphoma). Studies deemed to be the most relevant were included.
Myeloid-derived suppressor cells are a heterogeneous population of immature myeloid cells that are functionally defined by their ability to suppress host anti-tumor immunity.28,29 The expansion of these cells is mediated by tumor- and stromal-derived factors that drive myelopoiesis and inhibit myeloid differentiation, including hematopoietic (eg, colony-stimulating factors [CSF]) and inflammatory (eg, interleukin [IL] 1, IL-6) cytokines, arachidonic acid metabolites, damage-associated molecular pattern proteins, and vascular endothelial growth factor (VEGF).30-43 These cells comprise less than 1% of peripheral blood mononuclear cells in healthy humans; however, they greatly expand in the tumor-bearing host and accumulate in the bone marrow, peripheral blood, secondary lymphoid organs, and tumors where they may comprise up to approximately 50% of peripheral blood mononuclear cells in some patients with metastatic disease.44-51
Although MDSCs are not a discrete subset of myeloid-derived cells, they are generally defined as CD11b+HLA-DR−/low cells that express the common myeloid antigen CD33 and often fail to express other lineage-specific markers.28,44,52,53 Murine studies demonstrate that MDSCs may be subdivided into granulocytic and monocytic MDSCs, both of which have unique properties.54,55 Like their murine counterparts, human MDSCs may variously express granulocyte or monocyte markers.45-47,49,56
Myeloid-derived suppressor cells suppress host antitumor immunity by a variety of mechanisms, many of which become fully activated within the tumor microenvironment in response to inflammatory cytokines.57-59 Suppression mediated by these cells has classically been associated with metabolism of the amino acid L-arginine by either inducible nitric oxide synthase (iNOS) or arginase (ARG).60 Activation of these enzymes results in depletion of L-arginine and production of reactive oxygen and nitrogen-oxide species that culminate in the down-regulation of CD3 (required for T-cell activation after antigenic stimulation), posttranslational modification of the T-cell receptor (resulting in diminished antigen recognition), disruption of cell-cycle progression, impaired IL-2 production, and induction of T-cell apoptosis.61-70 Myeloid-derived suppressor cell–mediated suppression of T-cell immunity may occur independently of iNOS and ARG because multiple other suppressive mechanisms have been described.55,71-74 Also, MDSCs inhibit the antitumor immune response indirectly by promoting induction of regulatory T cells.46,75 Not surprisingly then, the expansion of MDSCs is associated with a poor response to immunotherapeutic strategies, but their depletion may restore immunologic competence.28,45,76-78 Therefore, MDSCs represent an attractive therapeutic target.79 However, depletion of MDSCs and other myeloid elements has implications that extend beyond the host anti-tumor immune response, as recent work so elegantly illustrates.
Tumor- and stromal-derived proangiogenic factors promote formation of the tumor vasculature that is required for tumor growth and progression.80 Vascular endothelial growth factor A (VEGF-A) is a well-characterized factor that promotes tumor angiogenesis.81 Vascular endothelial growth factor A neutralization mediated by the monoclonal antibody bevacizumab inhibits tumor growth and angiogenesis in tumor xenografts and is now widely used in various human malignancies.80,81 Tumor-infiltrating myeloid cells, including neutrophils, are not only a source of VEGF-A; through the production of matrix metalloproteinase 9, they also facilitate liberation of VEGF from the extracellular matrix.82-84 Mice that transgenically express the SV-40 T-antigen under the control of a rat insulin promoter spontaneously develop islet cell tumors at 3 months. In these mice, neutrophil infiltration promotes the angiogenic switch and tumor progression.85,86 Similarly, tumorinfiltrating monocytes may closely associate with the tumor vasculature and promote angiogenesis.87-89 Although controversial, experimental evidence suggests that myeloid cells may differentiate into endothelial cells or be directly incorporated into the tumor vasculature, thus promoting tumor angiogenesis.84,90
In a landmark article, Shojaei et al91 demonstrated that CD11b+Gr1+ myeloid cells not only promote tumor angiogenesis but also confer resistance to VEGF neutralization. They showed that, although some murine tumors were sensitive to VEGF-A blockade, other tumors were largely refractory to VEGF-directed therapy. In contrast to tumors sensitive to VEGF neutralization, “refractory” tumors recruited CD11b+Gr1+ myeloid cells to the tumor microenvironment, and the recruitment of these cells was associated with a less pronounced reduction in the tumor vasculature after VEGF neutralization. Because myeloid cells may be directly incorporated into the growing tumor vasculature, the authors excluded this possibility and concluded that myeloid cells may directly promote tumor angiogenesis in a VEGF-independent fashion. This was supported by the finding that myeloid cells isolated from refractory tumors when mixed with “sensitive” tumor cells before inoculation were able to promote tumor growth and mediated refractoriness to anti-VEGF treatment. Collectively, these data suggest that depletion of myeloid cells may confer increased sensitivity to VEGF-directed therapies in otherwise refractory tumors. Indeed, combined treatment with anti-VEGF and an antibody (anti-Gr1) capable of depleting CD11b+Gr1+ myeloid cells significantly reduced tumor growth in tumors refractory to anti-VEGF treatment alone. In similarly performed experiments, neutralization of tumor-derived granulocyte-CSF (G-CSF) reduced the expansion of CD11b+Gr1+ cells, reduced tumor angiogenesis, and slowed tumor growth.92 Conversely, administration of recombinant G-CSF, which is frequently given to prevent chemotherapy-induced neutropenia, promoted the expansion of these cells and conferred partial resistance to VEGF-directed therapy. In subsequent work, tumor-derived G-CSF was shown to up-regulate the expression of the angiogenic protein Bv8 (or prokineticin [PKR] 2) in myeloid cells.93 Bv8 and the closely related protein endocrine gland VEGF (EG-VEGF) on binding their G-protein–coupled receptors (PKR-1 or EG-VEGFR1 and PKR-2 or EG-VEGFR2) are endothelial cell mitogens.94,95 Furthermore, Bv8 itself promotes myelopoiesis and tumor angiogenesis and thus represents a novel therapeutic target.93,96
Because tumor-associated myeloid cells, including MDSCs, suppress host anti-tumor immunity and promote tumor angiogenesis, targeting these cells represents a novel therapeutic approach in human cancers. The list of tumor-associated factors that drive myelopoiesis in the tumor-bearing host and promote activation, recruitment, and effector function of these myeloid cells continues to expand and presents many opportunities for therapeutic targeting of these cells. In contrast, the outright depletion of myeloid cells represents an equally plausible approach.
The sheer number of soluble factors involved in driving myelopoiesis, leading to the accumulation of MDSCs and other myeloid cells in the tumor-bearing host, renders the neutralization of each of these factors a daunting task. Furthermore, because many of these cytokines and tumor-derived factors may be redundant, targeting any one of them may be of little benefit. For example, a reduction in MDSCs was not observed in a small series of patients treated with a fusion protein designed to neutralize VEGF, (ie, VEGF-trap).97 In contrast, the VEGF-neutralizing antibody bevacizumab was shown to reduce accumulation of MDSCs in tumor-bearing animals.98 Similarly, stem cell factor promotes accumulation of MDSCs and induction of T-cell unresponsiveness on engagement of its receptor c-kit.33 In a murine model, c-kit blockade impaired accumulation of MDSCs, prevented induction of T-cell unresponsiveness, decreased tumor angiogenesis, and improved the anti-tumor response in conjunction with an immunotherapeutic treatment strategy.33 This is certainly a clinically relevant approach because multiple tyrosine kinase inhibitors that inhibit c-kit (among other targets, including VEGF receptors) are available (eg, imatinib, sunitinib, sorafenib). In fact, in patients with metastatic renal cell carcinoma treated with sunitinib, the frequency of MDSCs returned to near normal after 2 cycles of treatment and was associated with increased T-cell responsiveness.99 The improvement in T-cell activity noted in these patients may not be solely explained by the observed reduction in MDSCs because sunitinib improved T-cell activity in vitro, suggesting that sunitinib may directly impair the suppressive capacity of these cells. Whether these observed effects are due to inhibition of c-kit, VEGF, or CSF-1 receptors is unknown. In a similar study, administration of sunitinib in the usual fashion (ie, 4 weeks of treatment followed by 2 weeks off treatment with each treatment cycle) was associated with a reduction in the frequency of both neutrophils and monocytes.100 After the 2 weeks off treatment, a substantial increase in the frequency of dendritic cells was observed, particularly in patients who had responded to treatment. Collectively, these data suggest that sunitinib may not only inhibit the accumulation of MDSCs but also may restore normal myeloid differentiation.
Although a diverse array of tumor- and stromal-derived factors promote myelopoiesis in the tumor-bearing host, many of these factors converge on a shared intracellular signaling pathway.101,102 The Janus kinases (JAKs) mediate many downstream signaling events after cytokine receptor engagement by the appropriate ligand, culminating in receptor and signal transducers and activators of transcription (STAT) phosphorylation. Phosphorylation of STAT proteins and their subsequent dimerization result in nuclear translocation and transcriptional regulation of genes involved in cell proliferation and differentiation. STAT3 has emerged as a central regulator of MDSC expansion and function.103-107 The utility of STAT3 inhibition was demonstrated by the finding that its pharmacological inhibition both depleted MDSCs and promoted their differentiation in tumor-bearing animals.108-110 Whether newly described JAK2 inhibitors will target MDSC is largely unexplored.111,112 In addition, interferon γ and IL-4/IL-13 contribute to the activation of MDSCs within the tumor microenvironment via activation of STAT1 and STAT6, respectively, and represent equally rational therapeutic targets.55,57-59
Cancer-associated myelopoiesis is characterized not only by enhanced myelopoiesis but also by impaired myeloid differentiation.113,114 Retinoids, including all-trans-retinoic acid (ATRA), have long been implicated in myeloid biology and are clinically used in the treatment of acute promyelocytic leukemia given their ability to promote myeloid differentiation.115 More recently, ATRA was shown to promote MDSC differentiation both in vitro and in vivo in murine models, and in patients with metastatic renal cell carcinoma in whom sufficient ATRA levels were achieved.116-118
Immature myeloid cells, although commonly measured in the peripheral blood, influence host immunity and tumor angiogenesis within secondary lymphoid organs and the tumor microenvironment. Not surprisingly then, multiple chemokines and other chemoattractants have been implicated in their recruitment, each of which may be therapeutically targeted. The complement system, although typically associated with innate immunity to various pathogens, also contributes to the recruitment of myeloid cells to the tumor microenvironment.119 Components of the complement cascade localize to the tumor endothelium, and transplanted tumors grow more slowly in mice deficient in the complement component C3 compared with wild-type mice.120 Myeloid-derived suppressor cells express the C5a receptor and are recruited to the tumor site by C5a deposited within the tumor vasculature. Remarkably, a peptide antagonist of the C5a receptor was as effective as systemic chemotherapy in slowing tumor growth. The C5-blocking antibody eculizumab is currently in clinical use; whether it can similarly impair recruitment of MDSCs is unknown.121 Conventional chemokines, including SDF-1 (stromal cell-derived factor 1), CXCL5, CCL2, and CCL5, have likewise been implicated in recruitment of MDSCs and represent novel therapeutic targets.122-124
The functional inhibition of MDSC-mediated immune suppression with pharmacological inhibitors of iNOS and ARG is commonly used for in vitro studies, but many of these agents are not clinically feasible. These studies adequately demonstrate that functional inhibition of myeloid cells is a reasonable goal.28 However, more recent work has provided clinically feasible means to achieve that end. For example, the phosphodiesterase 5 inhibitor sildenafil has been shown to down-regulate iNOS and ARG expression in MDSCs, increase T-cell activation, and slow tumor growth in a murine model.125 The clinical relevance of these findings was supported by the demonstration that sildenafil also promoted T-cell expansion in peripheral blood mononuclear cells obtained from patients with cancer.
Although initially developed to decrease the gastrointestinal adverse effects of classical nonsteroidal antiinflammatory drugs, the addition of a nitric oxide moiety to these compounds (eg, nitroaspirin) results in suppression of iNOS and impaired generation of reactive oxygen species. On administration to tumor-bearing mice, inhibition of both iNOS and ARG was observed and was associated with an increase in tumor antigen-specific T cells.126-128 Furthermore, efficacy was improved when nitroaspirin was combined with a cancer vaccine.
B7-H1 (PD-L1, CD274), a homologue of the T-cell costimulatory ligands B7-1 (CD80) and B7-2 (CD86), is widely expressed by both tumor cells and myeloid-derived cells within the tumor microenvironment, including MDSCs, and inhibits T-cell activation and effector function.129-133 Furthermore, many of the factors implicated in the activation of MDSCs regulate B7-H1 expression.46,47,55,73,74,130 In an ovarian cancer model, B7-H1–expressing MDSCs suppressed T-cell immunity in a B7-H1–dependent fashion.73
Although B7-H1 is directly suppressive, it was also shown to regulate ARG expression in MDSCs. In addition, B7-H1 expression by myeloid cells may indirectly suppress antitumor immunity by promoting induction of suppressive regulatory T cells.46,134 Monoclonal antibodies that target both B7-H1 and its counterreceptors, PD-1 and B7-1, are in various stages of clinical development.135-138 Although specific targeting of factors that promote expansion, migration, and effector functions of myeloid cells is certainly rational, an alternative approach simply seeks to deplete these cells altogether. Cytotoxic chemotherapeutic agents, increasingly appreciated for their immunomodulatory properties, effectively deplete MDSCs in tumor-bearing hosts. Gemcitabine, a pyrimidine antimetabolite that impairs DNA synthesis, may provide the best example; 2 studies demonstrated that gemcitabine reduced the frequency of MDSCs without inducing concomitant lymphopenia in multiple tumor models and improved both T-cell and natural killer cell immunity.139,140 Additional agents, including 5-fluorouracil and taxane-based therapies, may likewise specifically deplete myeloid elements in the tumor-bearing host.47,91
Many of these studies, whether performed in murine models or in patients with cancer, include clinical or immunologic end points, even though myeloid cells have been implicated in promoting tumor angiogenesis. In a murine model that assessed the contribution of myeloid cells in angiogenesis, antibody-mediated neutralization of the angiogenic protein Bv8, produced by myeloid cells, resulted in a significant reduction in circulating MDSCs, whereas VEGF neutralization had no effect.93 Impressively, anti-Bv8 administration resulted in a profound reduction in both tumor growth and tumor vascular volume and, when combined with either cytotoxic chemotherapy or anti-VEGF, resulted in a further reduction in tumor growth. This work provides yet further evidence supporting the therapeutic targeting of cancer-associated myelopoiesis combined with conventional chemotherapeutic or biologic agents.
Chronic inflammation has been implicated in the development of neoplastic disease for almost 150 years.141 Since then, many epidemiological and animal studies support this association.142 In the modern era, the association between malignant diseases, particularly when metastatic or locally advanced, and development of a leukocytosis has been appreciated, even though the importance of this association remained obscure. However, a growing body of work demonstrates that carcinogenesis and tumor progression result in both quantitative and qualitative changes in myelopoiesis. Consequently, cancer-associated myelopoiesis is characterized by accumulation of myeloid cells at various stages of differentiation, resulting in suppression of host anti-tumor immunity and promotion of tumor angiogenesis. Therefore, two seemingly unrelated fields—tumor immunology and angiogenesis—have converged, such that therapeutic strategies that target tumor angiogenesis appear to have immunologic consequences. Conversely, immunotherapeutic approaches incorporating strategies that target myeloid cells may impair tumor angiogenesis. Despite a long-standing understanding of the association between tumor progression and myeloproliferation, rudimentary as that understanding may have been, the therapeutic implications of this association are only now becoming clearer. One suspects that Rudolf Virchow, who first implicated chronic inflammation in carcinogenesis more than a century ago, might conclude that “what has been will be again, what has been done will be done again; there is nothing new under the sun” (Ecclesiastes 1:9).