Cancer and Pregnancy: Parallels in Growth, Invasion, and Immune Modulation and Implications for Cancer Therapeutic Agents

Five days after fertilization, the human zygote forms into a structure consisting of 2 primary cell lines: the inner cellmass (or embryoblast) and the trophoblast. Trophoblast cells constitute the outer layer of the blastocyst, rapidly proliferating and invading the maternal endometrial decidua around day 7. A monolayer of cytotrophoblast cells surrounds the embryonic disc as the embryo completely embeds beneath the uterine decidua. By day 9, cytotrophoblast cells have differentiated into 2 distinct cell types: the syncytiotrophoblast and the extravillous trophoblast (EVT). The multinucleated syncytiotrophoblast cells form the external layer and are terminally differentiated. These cells are involved in fetomaternal nutrient exchanges and endocrine functions (such as β-human chorionic gonadotropic production). In contrast, EVT cells have a proliferative and invasive phenotype, migrating through the syncytiotrophoblast into the uterine wall to anchor the placenta beginning around day 14 after implantation. These EVT cells display a phenotype strikingly similar to cancer cells with their capacity for proliferation, migration, and establishment of a blood supply, making them a compelling model for oncologic comparison (Figure). This review highlights several shared characteristics of trophoblast and tumor cells and discusses them in the context of existing or developmental targeted cancer therapeutics (Table 1).

Like tumor cells, trophoblast cells have a very high proliferative capacity and exhibit molecular characteristics found in rapidly dividing cancer cells. For example, increased telomerase activity, typically not observed to a substantial degree in normal somatic cells, is detectable in 85% of human cancers. In fact, the intracellular concentration of telomerase is exponentially related to the proliferative capacity of a cell. In human pregnancy, telomerase activity is highest during the first trimester and decreases with maturation of the placenta. Survivin, a protein that promotes proliferation and inhibits apoptosis, is overexpressed in many cancers and is also up-regulated by trophoblast cells. Inhibition of survivin by knockdown with small interfering RNA leads to a marked decrease in proliferation introphoblast cell lines. A similar decrease in proliferation is seen with survivin in small interfering RNA treatment of prostate, glioma, non-Hodgkin lymphoma, cervical cancer cells, and breast cancer cells. Both survivin and telomerase levels are dramatically higher in hydatidiform moles than in normal placentas, providing insight into the potential involvement of these 2 different mechanisms in neoplastic transformation.

Another pathway supportive of both trophoblast and cancer cell proliferation is the IGF pathway (for expansion of all gene symbols, see Glossary of Genetics Terminology at the end of the article). By binding to the IGF1R on cytotrophoblast cells, IGF stimulates proliferation through the MAPK pathway and survival via activation of the PI3K pathway. Normally, levels of IGF are tightly regulated by IGF-binding proteins and protease pregnancy-associated plasma protein A, a binding protein. Loss of binding protein regulation may contribute to the malignant phenotype. In cancer cells, the IGF1R pathway is not only mitogenic and antiapoptotic but is involved in protecting cancer cells from damaging effects of chemotherapy and radiation, potentially as a result of its effects on downstream signaling pathways. Additionally, the fetal form of the insulin receptor IR-A, which is highly expressed in fetal tissues and responsive to IGF2, is also a member of the IGF-signaling system. In many cancers, including those of the breast and ovary, dysregulation of this fetal form of the insulin receptor becomes the predominant isoform leading to IGF2-stimulated proliferation and survival.^, 

The sine qua non of both a successful pregnancy and the growth of cancer is the establishment of a blood and nutrient supply, and invasion through normal tissues is required for this process. However, whereas cancer cells spread throughout the host and then engage in local proliferation, trophoblasts follow an organized pattern of differentiation from proliferation to invasion without distant metastasis. Some of the molecular switches involved in this differentiation pattern and their relevance for cancer therapeutic agents are discussed in the sections that follow.

As EVT cells migrate down the cytotrophoblast cell columns into the maternal decidua (Figure), they encircle and erode into the maternal spiral arteries and differentiate from a proliferative phenotype into an invasive phenotype. This differentiation occurs at about 10 to 12 weeks of gestation and is associated with opening of the intervillous space and exposure to maternal blood. Many parallels can be observed between invasive EVT cells and cancer cells. Some of these similarities are highlighted in the sections to follow; for a more in-depth discussion, readers should refer to excellent reviews by Soundararajan and Rao and Ferretti et al.

Requirements for cellular invasion include changes in cell adhesion molecules, secretion of proteases, and availability of growth factors. An example of a cellular program used by both cancer cells and trophoblast cells to promote invasion is epithelial-mesenchymal transition, which results in loss of cell-to-cell contact inhibition. Associated with this program are changes in integrin expression and loss of E cadherin, allowing loss of polarity and enhanced motility.^,  Both trophoblast and cancer cells secrete proteases to degrade extracellular matrix proteins required for dispersal through tissues. The cytoplasm of migratory EVT cells express HSP27, which is correlated with MMP2 activity. Basal HSP27 levels are unusually high in cancer cells, protecting them from apoptotic stimuli, and are associated with metastatic potential. Finally, growth factors such as epidermal growth factor stimulate motility of EVT cells through phosphorylation of p42 and p44 MAPKs and the PI3K-dependent proteins, AKT and p38. Epidermal growth factor is associated with tumor cell invasiveness through expression of MMPs.

Switches involved in triggering trophoblast and cancer cell molecular programs for invasion are not completely understood. The Wnt pathway, a system highly conserved across species involved in cellular proliferation and motility, has recently been implicated in switching trophoblast cells from a proliferative to an invasive phenotype. Activation of the Wnt pathway is aberrant in many cancers, resulting in escape of β-catenin from proteosomal degradation, with subsequent β-catenin translocation into the cell nucleus and activation of multiple target genes. Although direct activation of β-catenin alone has shown no effect on motility of EVT cells, inhibition of the Wnt—β-catenin pathway can block blastocyst implantation. In EVT cells, activation of PAR1 (also known as the thrombin receptor) also stabilizes β-catenin and is associated with a proliferative and invasive capacity, whereas application of PAR1-silencing RNA inhibits EVT invasion. Consistent with the need for tight regulation of invasive trophoblast cells, PAR1 is expressed in EVT cells between the 7th and 10th gestational week but is abruptly shut off by the 12th week. Constitutive increased expression of PAR1 can be seen in cancer cells, especially in cells lacking normal p53 activity. In vitro assays have shown PAR1 antagonism to inhibit MMP1-induced endothelial cell activation in tumor—endothelial cell communication. Whether this system could successfully be targeted for cancer therapy is under investigation. Other signal transduction pathways common in both trophoblast and cancer cell invasion include the JAK-STAT pathway, FAKs, G proteins, Rho-associated kinase, MAPKs, PI3K, and SMAD family proteins. All of these pathways represent areas of current anticancer therapeutic development.

As EVTs acquire an invasive phenotype during placental development, they become polyploid (4N-8N) by switching from mitotic division to endoreduplication, a process in which G2 or M phase (4N) cells replicate DNA without undergoing mitosis. In trophoblast cell lines, polyploid trophoblast giant cells are relatively resistant to the DNA-damaging effects of radiation, illustrating a mechanism by which survival is promoted in invasive trophoblast cells. This process can also be observed in cancer cells treated with DNA-damaging agents. Endoreduplication can be induced in tumor cells on exposure to genotoxic agents such as paclitaxel and cisplatin; a nonproliferative, senescent state in a small population of cells is induced in the latter case. The polyploid tumor cells can undergo depolyploidization to form diploid, cisplatin-resistant escape cells. In cells with an impaired p53 system, treatment with the Aurora kinase inhibitor VX-680 leads to endoreduplication followed by apoptosis. However, in 2 wild-type p53 cancer cell lines, stabilization of p53 by Nutlin-3a, an inhibitor of the p53-binding protein MDM2, leads to initial endoreduplication followed by the emergence of stable radiation- and cisplatin-resistant tetraploid clones. A better understanding of the EVT endoreduplication process may lead to the development of targeted drugs to maintain tumor cell chemotherapeutic sensitivity.

As trophoblasts invade maternal spiral arteries, they further differentiate to display a vascular phenotype in a process termed vasculogenic mimicry, in which cells other than endothelial cells form vascular structures. Vasculogenic mimicry can also be observed in aggressive cancers, and the genes and signaling pathways involved with the process of vasculogenic mimicry may be shared between EVT and cancer cells. For example, the matrix glycoprotein—binding galectin 3 is highly expressed in EVT cells. Galectin-3 also appears to be a key factor in the development of an endothelial phenotype and the tube formation well described in aggressive melanomas. Galectin inhibitors are in preclinical testing as cancer therapeutic agents. Mig-7 was found in circulating tumor cells and tumor tissue (regardless of tissue of origin) from more than 200 patients with cancer; notably, it was absent from healthy controls. Mig-7 expression is associated with invasion and vasculogenic mimicry in cancer cells and also has recently been demonstrated in invasive embryonic cytotrophoblasts, peaking when EVT cells invade maternal decidua and remodel the vasculature during early placental development. This finding represents the only known expression of Mig-7 in noncancerous cells. Cancers with an endothelial phenotype have not been shown to be responsive to antiangiogenic therapies. Because cancer therapy aimed at proliferating cells is less likely to be effective in invading cells, galactin-3, Mig-7, and other pathways involved in vasculogenic mimicry may also be important targets for cancer therapy.

Molecular circuits involved in neoangiogenesis separate from vasculogenic mimicry are also likely shared between EVT and tumor cells. Angiopoietins and VEGF family members are extremely important in both spiral artery remodeling in placentation and the growth of many tumor types. Inhibition of VEGF has become an important therapeutic strategy in many cancers, although resistance can develop, resulting from the induction of an angiogenic rescue program characterized by the up-regulation of multiple angiogenic genes in hypoxic tumor cells and supporting stroma.^,  Another member of the VEGF family, PGF, is a part of the VEGF blockade—associated rescue program that is involved in the response to pathologic conditions, such as wounds, ischemia, inflammation, or cancer. Both VEGF and PGF are highly expressed in trophoblast cells. It is interesting that serum levels of PGF increase after treatment of patients with cancer with the anti-VEGF monoclonal antibody bevacizumab. Preclinical studies indicate that PGF blockade reduces neoangiogenesis and lymphangiogenesis, hampers recruitment of intratumoral macrophages, and is not associated with the typical anti-VEGF adverse effects (thrombosis, hypertension, proteinuria, and microvascular pruning) in healthy mice.

Also important for angiogenesis is the oxygen-sensitive MTOR pathway. Central to controlling trophoblast cell proliferation in response to nutrients and growth factors, MTOR is expressed on the transporting epithelium of intact human placenta. It is downstream of the PI3K/AKT pathway; controls cell cycle progression and cell size and mass; is involved in angiogenesis via the VEGF, IGF, and HIF-1α—signaling pathways; and is constitutively activated in many malignancies.^,  The MTOR inhibitor everolimus has antiangiogenic properties. A better understanding of the PI3K/AKT/MTOR pathway and other molecular circuits used by trophoblast cells in proliferation, invasion, and endothelial interactions may lead to the development of targeted therapies for cancer. Overall, we are in our infancy of understanding the complexity, redundancy, and interrelatedness of these molecular pathways in both placentation and neoplasia.

In addition to sharing many proliferative and invasive features, the cells of the trophoblast, like cancer cells, actively modulate the host immune response to develop and sustain a nutrient supply. Historically, the placenta was considered an inert, mechanical barrier protecting the semiallogeneic fetus from maternal immunologic attack. Current evidence, however, supports just the opposite—many maternal and placental immunomodulatory factors are required for adequate placental invasion. Around 40% of decidual cells are cells of the innate immune system (eg, NK cells, macrophages, and DCs), a substantial proportion considering that the uterus is a nonlymphoid organ. Likewise, although cancer previously has been considered immunologically invisible to the host, many recent studies support the notion that cancer cells actively engage immune cells; for example, the presence of tumor-infiltrating lymphocytes has been well described in the literature. The main components of the maternal immune response at the fetomaternal interface and the similarities to the tumor microenvironment are discussed in the sections that follow.

The most abundant immune cell present at the fetomaternal interface is the uterine NK (uNK) cell, which constitutes approximately 70% of all immune cells found in this tissue. Uterine NK cells are thought to be recruited from peripheral blood when interleukin (IL)-15 is secreted by endometrial stromal cells. They are distinct from peripheral blood NK cells in that they do not express CD16, the FcRγIIIA receptor required for antibody-dependent cell-mediated cytotoxicity. The mechanisms associated with this loss of CD16 are unclear but may be related to high levels of TGF-β within the microenvironment. Also, in contrast to peripheral blood NK cells, uNK cells are more immunomodulatory than cytotoxic, secreting galectin 1 to induce tolerogenic DCs as well as angiogenic factors VEGF and PGF that are important for decidual remodeling. An improper balance of cytotoxic to regulatory NK cells could contribute to recurrent miscarriage and pre-eclampsia. Expression of IL-15 and NK cell infiltration have been reported in many different malignancies, including renal cell carcinoma, with variable prognostic implications. Recently, tumor-infiltrating CD16-NK cells have also been characterized and appear to behave similarly to uNK cells with respect to cytokine production and reduced cytotoxic activity. A closer look at factors that determine the balance of killer and regulatory NK cells during pregnancy may help identify mechanisms that shift immunity toward NK cytotoxic activity in patients with cancer.

Also infiltrating the decidua, albeit in smaller numbers than uNK cells, are macrophages, T_reg, and DCs. Macrophages phagocytose apoptotic EVT cells and secrete IL-10 and IDO, contributing to the tolerogenic T_H2 milieu. Gene expression profiling of decidual macrophages supports an immunosuppressive/anti-inflammatory phenotype with higher expression of CCL18, IGF1, IDO, neuropilin 1, and other genes associated with M2-polarized macrophages. Tumor-associated macrophages can be both inflammatory and immunosuppressive, and T_H1/T_H2 polarization is effected through the activation of NF-κB (also known as NFKB1). In fact, in vitro studies suggest that tumor-associated macrophages may be re-educated to display a classically activated rather than an M2 phenotype by inhibition of inhibitory kappa B kinase β, the major activator of NF-κB.

Regulatory T cells are additional important mediators of tolerance in both pregnancy and cancer. Immunophenotypically, these cells express surface CD4, CD25, and FOXP3, and they expand in both decidua and peripheral blood during normal pregnancies. This expansion is antigen-specific and is induced by paternal/fetal alloantigens and not simply by hormonal changes in pregnancy. A decrease in this lymphocyte subset is associated with spontaneous abortion and pre-eclampsia. Regulatory T cells are also expanded in cancer and are implicated in impaired antitumor immunity, suppression of effector T lymphocyte proliferation, and increased tumor blood vessel density, suggesting an important link between immunity and angiogenesis. Regulatory T cells in patients with cancer also recognize tumor-specific antigens and proliferate in response to antigenic stimulation. Targeting the T_reg population to boost antitumor immunity is under investigation with agents such as denileukin diftitox (IL2/diphtheria fusion protein) or LMB-2 (Fv fragment of CD25 antibody/Pseudomonas endotoxin A fusion protein) and CTLA-4 inhibitors.^,  Some of the benefit of cytotoxic chemotherapy may be derived from concomitant impairment of the immunosuppressive T_reg proliferation driven by the cancer.

Antigen-presenting CD83⁺ DCs are involved in the maintenance of the T_H2-predominant state in decidual tissues, as well as at other mucosal surfaces. However, the role of the DC is likely more complex than antigen presentation and secretion of immunosuppressive cytokines. Ablation of uterine DCs leads to decidualization failure and embryo resorption in mice; this occurs even with syngeneic pregnancy in mice in which alloantigens are absent. Dendritic cells also represent another link between immunity and angiogenesis, secreting soluble FLT1 (also known as VEGFR1) and TGF-β1 required for endothelial cell survival and vascular maturation. In the absence of DCs, angiogenesis is severely impaired. In cancer, DCs also play a role that is more than immunoregulatory through their production of potent angiogenic growth factors. Moreover, cancer cells can secrete substances that suppress maturation of DCs, including VEGF, TGF-β, hepatocyte growth factor, and osteopontin, thereby maintaining a proangiogenic, immature DC phenotype.

Expression of certain cell surface molecules on both trophoblast and cancer cells can also confer immunologic protection. Among the most important of these molecules is the nonpolymorphic, highly conserved class I human leukocyte antigen (HLA) molecules such as HLAG; in contrast, the highly diverse classical HLA class I proteins A, B, and C are essential in cell-mediated immune responses. In fact, in trophoblast cells, interferon-γ fails to stimulate classical HLA class I expression. A similar property of down-regulated or absent classical HLA class I expression can cloak cancer cells from the host's immune system. Cancer treatment modalities including gamma irradiation, radiopharmaceutical samarium-153-ethylenediaminetetramethylenephosphonate, and chemotherapeutic agents such as 5-fluorouracil and hypomethylating agents increase HLA class I expression.

Expression of HLAG on trophoblast cells and cancer cells has important immunomodulatory effects. In the placenta, HLAG expression is most evident on EVTs at the fetomaternal interface, with lower expression at the proliferative area of the villous column and increased expression with invasive, interstitial, and endovascular EVT cells. On the basis of sequence homologies, HLAG has been proposed as the ancestral MHC class I gene and has only a few known sequence variations in humans, in sharp contrast to the profound allelic diversity (measured in the hundreds of allelic variants) of classical MHC class I genes. Human leukocyte antigen-G interacts with NK cells via inhibitory receptors, such as CD94/NKG2A, ILT2, and killer cell immunoglobulin-like receptor KIR2DL4. The role of HLAG is to suppress cytolytic killing by both NK and cytotoxic T cells, induce apoptosis of immune cells, regulate cytokine production in blood mononuclear cells, and reduce stimulatory capacity and impair maturation of DCs (reviewed in Hunt et al). Within the tumor microenvironment, the generation of HLAG⁺—suppressive NK cells occurs by trogocytosis (ie, the rapid cell-to-cell contact-dependent transfer of membranes and associated molecules from one cell to another), leading to the inhibition of other HLAG⁺ (cross-inhibition) or HLAG⁻ NK cells through HLAG and ILT2 cross-linking. Expression of HLAG is associated with a poor prognosis in patients with lymphoproliferative disorders, melanoma, mesothelioma, breast carcinoma, ovarian carcinoma, renal cell carcinoma, squamous esophageal cancer, gastric carcinoma, cervical cancer, non—small cell lung cancer, bladder cancer, prostate cancer, endometrial cancer, colorectal cancer, and myeloid malignancies, including acute myeloid leukemia.^,  However, relatively little is known about the regulation of the expression of this important immunomodulatory molecule. Regulation of HLAG expression may be at the epigenetic level, with transcription of HLAG being detectable in acute myeloid leukemia cell lines after treatment with 5-aza-2′-deoxycytidine. Some preliminary evidence also supports a micro-RNA regulatory mechanism. Clearly, HLAG represents an attractive target for immune-based cancer therapies given its preferential expression in many malignancies as well as limited expression in normal tissues. Targeting HLAG with a peptide-based vaccine strategy to develop a cytotoxic T-cell response against tumor cells bearing the molecule has proved feasible, although much work remains before other methods of HLAG inhibition could lead to restoration of antitumor immunity.

Other cell surface tolerance signals common between trophoblasts and cancer cells include CD200 (OX-2) and CEACAM-1. Trophoblast cells expressing CD200 can inhibit CD8⁺ T cytotoxic lymphocyte (CTL) generation and shift the cytokine balance toward T_H2 in vitro. Expression of CD200 is a negative prognostic factor in patients with multiple myeloma and acute myeloid leukemia, and it has been shown to down-regulate T_H1 cytokines in vitro in solid tumors, including melanomas, ovarian carcinomas, and renal cell carcinomas. As a potential cancer stem cell marker, CD200 may be a promising target for these cells that survive conventional chemotherapy. CEACAM-1 (CD66a), expressed on both trophoblasts and IL-2-activated decidual leukocytes, plays a role in inhibiting NK-mediated cytolysis. Colocalization of osteopontin on EVT cells is associated with an invasive phenotype important for successful placentation. CEACAM-1 expression in cancer is associated with increased angiogenesis in non-small cell lung cancer; in melanoma, it has been shown to be predictive of the development of metastatic disease. Expression of other immunomodulatory molecules, including components of the extrinsic apoptotic pathway such as FAS, TNF superfamily receptors, TRAIL, and B7 family members such as B7H1 (or programmed death ligand 1, PDL-1), is also common between trophoblast and cancer cells (Table 1).

Chemokines and cytokines also play a role in promoting a tolerogenic environment in placentation and the tumor microenvironment. Implantation of the blastocyst occurs in a T_H1-predominant (inflammatory) milieu, but the fetomaternal interface must transition to a T_H2-polarized (immunologically tolerant) state for pregnancy to continue (for an excellent review, refer to van Mourik et al). However, before implantation can occur, the endometrial lining must be receptive in the so-called window of implantation, in which many immunomodulatory genes are up-regulated monthly during the midsecretory phase of the menstrual cycle. Under the influence of progesterone, the endometrial epithelium up-regulates decay-accelerating factor and osteoponin expression, and the endometrial stroma increases IL-15 expression.^,  Expression of complement regulatory proteins (eg, decay-accelerating factor) is a well-established immunomodulatory mechanism used by many cancers to escape complement-mediated cell death and evade an immune response by inhibiting T-cell proliferation. Osteopontin has T_H1 cytokine functions and is chemotactic for macrophages, T cells, and DCs, the last of which it induces to secrete IL-12 and tumor necrosis factor α (TNF-α).^,  Osteopontin is overexpressed in many cancers and is associated with metastatic potential. Additionally, tissues that physiologically express high levels of osteopontin, such as bone, lung, and liver, may create a receptive microenvironment for metastasis via interaction with osteopontin receptor CD44 on the surface of cancer cells.

RANTES (CCL5) is a chemokine produced by trophoblasts that may play a role in apoptosis of potentially harmful maternal CD3⁺ cells. Melanoma cells can induce tumor-infiltrating lymphocytes to secrete RANTES and subsequently undergo apoptosis as another mechanism to evade an immune response. Trophoblast cells also secrete chemoattractant cytokines, such as GRO-α, MCP1, and IL-8, to actively recruit the CD14⁺ monocytes to the fetomaternal interface. GRO-α is an oncogenic and angiogenic cytokine driven by RAS, which is inappropriately activated in most cancers. Capable of inducing vascular permeability along with mononuclear cell recruitment, MCP1 is associated with angiogenesis and malignant pleural effusions. Inhibition of MCP1 can lead to reduced malignant angiogenesis and recruitment of tumor-associated macrophages in a mouse model of melanoma. Finally, the IL-8 pathway is well known to be a central immune and angiogenic factor within the tumor microenvironment and is important in stress-induced chemotherapeutic resistance.

A tryptophan-catabolizing enzyme, IDO is important in promoting tolerance by inhibiting proliferation of lymphocytes both at the fetomaternal interface and tumor microenvironment. Tryptophan levels have been observed to decrease in pregnancy with a return to normal, nonpregnant levels in the puerperium, possibly a result of tryptophan degradation by IDO-expressing trophoblast cells. Expression of HLAG on DCs can be induced by IDO, indicating potential cooperation in immune suppression between these 2 molecules. Tumor-derived PGE2 secretion can increase IDO expression in local DCs. Antigen-expressing cells and IDO-expressing tumor cells might also contribute to local immunosuppression in tumor-draining lymph nodes. Pharmacologic inhibitors of IDO are under development and in early-stage clinical trials as anticancer agents. Induction of IDO can also be blocked in vitro by cyclooxygenase 2 inhibitors. When murine breast cancer vaccine recipients received the oral cyclooxygenase 2 inhibitor celecoxib, an increase in tumor-specific CTLs was observed.

Trophoblast invasion and spiral artery remodeling are tightly controlled processes, likely kept in check both by molecular programming of trophoblast cells and by paracrine immune factors. We have much to gain in terms of developing novel immunologic interventions for our patients with cancer by closely examining both the similarities and differences of the intimate cross-talk that occurs within the tumor and placental microenvironments.

Similar to the increasing antigenic burden of progressive cancer, fetal DNA can be found circulating in maternal blood by the second trimester in the height of the tolerogenic cytokine milieu. Although its immunologic consequences have not been fully elucidated, this circulating DNA likely contributes to tolerance and eventual exhaustion of antigen-specific CTLs. This phenomenon is well described for the human immunodeficiency virus, chronic infection with which leads to progressive HIV-specific T cell dysfunction. In addition to circulating nucleic acids, cellular fragments, known as microparticles or exosomes, can be detected in the peripheral blood of pregnant women in the third trimester. Trophoblast-derived microparticles are proinflammatory, activate the coagulation system, can cause endothelial dysfunction, and are circulating at higher levels in pre-eclamptic vs normal pregnancies. These microparticles are also involved in down-regulation of T-cell activity and deletion of activated T cells through interactions with FAS or TRAIL on the microparticle surface. A similar phenomenon of cancer cell-derived microparticles contributing to the hypercoagulable state and impaired antitumor immunity of patients with cancer has been described (reviewed in Amin et al). Microparticles derived from melanoma cells have been shown to express HLAG, likely contributing to their immunomodulatory properties.

Just as circulating tumor cells have been identified in patients with early-stage malignancies, intact trophoblast cells are also known to circulate in the maternal peripheral blood as early as the ninth week of pregnancy. These fetally derived cells can engraft in the mother irrespective of HLA disparity and establish a long-term microchimerism that persists for decades after parturition. Rates of fetal microchimerism are decreased in female patients with cancer (34%) compared with healthy controls (57%), and the immunomodulatory implications of this decrease are unclear. An increased number of fetal microchimeric cells in aggressive breast carcinoma and melanoma during pregnancy have been observed. Whether these cells were recruited to the tumor microenvironment by inflammation and behave as innocent bystanders or whether they participate in tumor progression by providing angiogenic or tolerogenic signals is unclear at this time.

Many additional immunomodulatory proteins are secreted by trophoblast cells and can be found circulating in maternal peripheral blood. Among these molecules, soluble HLAG may be the most extensively studied. Soluble HLAG impairs NK/DC cross-talk, promotes proinflammatory cytokine secretion from both uterine and peripheral blood mononuclear cells, and induces apoptosis of CD8⁺ cells through CD8 ligation and FAS-FASL interaction. Soluble HLAG has been well documented in malignancies, including acute leukemia, multiple myeloma, lymphoproliferative disorders, breast and ovarian carcinoma, renal cell carcinoma,^,  lung cancer, gliomas, and melanoma. Cancer cells can also trigger monocytes to release HLAG, further down-regulating antitumor immunity. Whether HLAG can be targeted to break cancer-specific tolerance remains to be investigated.