Immunology of Isolated and Recurrent Spontaneous Pregnancy Loss

Spontaneous pregnancy loss is the most common complication of pregnancy: approximately 70% of human conceptions fail to achieve viability, and an estimated 50% are lost before the first missed menstrual period.¹ Recurrent pregnancy loss (RPL) or recurrent spontaneous abortion is much less common, occurring in about 1 in 100 pregnant women.²

Historically, recurrent abortion was defined as three or more clinically recognized pregnancy losses before 20 weeks of gestation. Using this definition, RPL occurs in approximately 1 in 300 pregnancies.³ However, many recommend that clinical investigation and intervention be initiated after two consecutive spontaneous abortions, especially if any of the following are coexistent: fetal heart activity was identified before any of the pregnancy losses, fetal karyotyping of pregnancy tissues revealed normal chromosome content, the woman is older than 35 years of age, or the couple also shows subfertility. 

The etiologies of RPL and their respective prevalences are controversial. The only undisputed causes of RPL are parental chromosomal abnormalities and thrombotic complications of the antiphospholipid syndrome (APS). Most investigators consider the following conditions to be associated with RPL: uterine anatomic abnormalities (congenital or acquired); endocrine abnormalities; infections; and immunologic factors, including those associated with APS.^{4, 5, 6} Other factors, such as environmental exposures, have been implicated but probably account for a small fraction of cases. The exact proportion of patients affected by each of these factors differs depending on the population being studied. Etiologic proportions may also be affected by referral bias and therefore may differ based on the interests of the investigator conducting a particular study. 

After a thorough evaluation, the potential cause for RPL remains unexplained in approximately 45–50% of patients.^{4, 5, 6} Investigators have frequently suggested that a substantial proportion of these otherwise unexplained RPL cases may have an immunologic component. Many cases of either isolated or recurrent spontaneous pregnancy loss may have more than one potential etiology, and immunologic mechanisms have been frequently invoked as contributory. This is certainly proposed when infection is associated with pregnancy loss.

Reproductive tract infections with bacterial, viral, parasitic, zoonotic, and fungal organisms have been theoretically linked to both isolated and recurrent pregnancy loss. However, the etiologic mechanisms linking specific organisms to either isolated or recurrent pregnancy loss remain unclear and must certainly differ among infectious organisms.^{7, 8, 9} For instance, some viral organisms, such as herpes simplex virus¹⁰ and human cytomegalovirus,¹¹ have been shown to infect the placenta/fetus directly, and pregnancy loss related to these infections may be the result of direct damage to the developing fetus by the infectious organism. Alternatively, pregnancy disruption might be an untoward effect of the immune response to the infectious agent. This mechanism may be responsible for some adverse infection-associated events later in gestation, such as intrauterine growth restriction,¹² premature rupture of membranes, and preterm parturition.¹³ Conversely, there is evidence that some of the same mechanisms that potentially protect the fetus from autoimmune rejection may also protect virally infected placental cells from recognition and clearance. Some of the fetoprotective immunoregulatory events that occur at the maternal–fetal interface could thereby make the implanting fetus particularly vulnerable to certain infectious agents, including herpes, cytomegalovirus, and HIV.¹⁴

The number of investigations involving reproductive immunology and its relationship to both isolated and recurrent pregnancy loss has expanded exponentially during the past few decades. Many of these studies, however, have been poorly designed or poorly controlled or have involved insufficient numbers of patients to warrant their conclusions. The clinical study of the immunology of early pregnancy is also hampered by a variety of intrinsic factors. First, a diversity of distinct immune alterations may result in isolated or recurrent pregnancy loss.¹⁵ Each of these may act alone or in combination with additional immune or non-immune factors to result in pregnancy loss. Second, many RPL patients present after their current pregnancy has expired but before its expulsion. Diagnosis of an immunologic cause for that index pregnancy loss is therefore difficult because the expected physiologic immune reaction to the presence of nonviable tissue may mask any alternative underlying immune causes for the demise itself. Prior studies are also compromised because of their failure to correct for embryo and fetal chromosomal abnormalities, which are independent causes of pregnancy loss. Perhaps the most important intrinsic factor is the reassuring fact that for most patients with RPL, the possibility that their next pregnancy will result in a term delivery is quite high. Therefore, very large studies are needed to detect therapeutic effects, including those related to immunomodulatory interventions.

A few basic immunologic concepts should be re-examined before a more directed discussion of the involvement of the immune system in pregnancy maintenance.¹⁶ Those concepts include the following immunologic subcategorizations: innate versus acquired immune responses; cellular versus humoral immune responses; antigen presentation by major histocompatibility complex (MHC) class I versus MHC class II molecules; and peripheral versus mucosal immune systems.  Each of these subdivisions is related to the others, and each is useful in understanding particular aspects of investigations on the immune interactions at the maternal–fetal interface. 

Innate Versus Acquired Immune Responses 

Immune responses are classically divided into innate and acquired. Innate responses represent the body's first line of defense against pathogenic invasion. Innate responses are rapid and are not antigen-specific. Cell types and mechanisms typically considered vital to innate immunity include complement activation, phagocytosis by macrophages, and lysis by natural killer (NK) cells and possibly by T cells expressing the γδ T-cell receptor (TCRγδ) (see below). Acquired immune responses, in contrast, are antigen-specific and are largely mediated by classical T and B cells. Acquired responses can be classified as either primary (a response associated with initial antigen contact) or secondary (a rapid and powerful amnestic response associated with repeated contact to the same antigen). 

Cellular Versus Humoral Immune Responses 

Further subdivision of acquired immunity into either cellular or humoral responses is useful, even though it grossly oversimplifies the complexity of immune interactions. By definition, the ability to mount a cellular immune response to a particular antigen can be transferred to a naïve (nonimmunized) individual via the lymphocytes (but not plasma or serum) from a separate immunized subject. Conversely, humoral immune responsiveness to a particular antigen can be transferred to a naïve subject using only the plasma or serum from an immunized individual. No cells need be transferred, and the response is known to be dependent on the presence of antibodies in the immunizing sera. Put simply, cellular responses require cell-to-cell interactions, whereas humoral responses are antibody-mediated. Like innate and acquired immunity, these cellular and humoral responses are intricately intertwined. 

Antigen Presentation by MHC Class I Versus MHC Class II Molecules 

Cellular and humoral immune responses are largely dependent on the presence of two sets of genes in the MHC that play major roles in determining antigen specificity. These genetic loci encode the MHC class I and MHC class II products, as well as many of the supporting effector molecules involved in antigen presentation. Both MHC class I and MHC class II molecules help to alert the immune system to alterations that would require an immune response. MHC class I molecules (HLA-A, -B, and -C) are present on the surface of nearly every cell in the human body and are important in defense against intracellular pathogens, such as viruses, and against oncogenic transformation. MHC class I molecules act as important ligands for the T-cell receptor on CD8+ cytotoxic/suppressor T cells and for a variety of receptors on NK cells.¹⁷ MHC class I-mediated intercellular interactions generally activate the cellular portion of the immune response and result in killing of the antigen-presenting (i.e. virally infected or cancerous) cell expressing the class I molecule. 

In contrast, MHC class II molecules (HLA-DR, HLA-DP, and HLA-DQ) are present on the surface of a limited number of 'professional' antigen-presenting cells. These include dendritic cells, macrophages, monocytes, B cells, and tissue-specific antigen presenters (e.g. Langerhans cells in the skin). MHC class II molecules are important in defense against extracellular pathogens, such as bacterial invaders. The major ligand for MHC class II is the T cell receptor on CD4+ T-helper cells. MHC class II-mediated interactions generally result in the modulation of humoral immune responses. 

Peripheral Versus Mucosal Immune Systems 

Perhaps the immunologic categorization most relevant to the discussion of pregnancy maintenance is that which divides the immune system into two distinct compartments: peripheral and mucosal. In general, basic immunologic principles have been well-tested and -described for the immune effector cells populating only one of these compartments—the peripheral immune system. Until recently, very little was known about the mucosal immune compartment, and its description continues to lag far behind that of the periphery. 

The peripheral immune system consists of the spleen and peripheral blood, and it is generally responsible for protection against blood-borne pathogens. In contrast, the mucosal immune system is typically the first to encounter those pathogens that enter via the extensive surface areas of the lacrimal ducts, respiratory and gastrointestinal tracts, mammary ducts, and genitourinary tracts. Some effector cells within mucosal sites such as the gastrointestinal tract and reproductive tract appear to have been educated outside of the thymus—a process referred to as extrathymic education.¹⁸ Thus, the mucosal immune system appears to be responsible for providing at least initial immunologic protection against the majority of exogenous pathogens and may play a role in its own effector cell education.

Immune interactions at the maternal–fetal interface differ from those of the peripheral immune system and from those ar other mucosal immune sites.¹⁹ The first of the mucosal immune sites to be extensively studied was the gastrointestinal tract, and investigations revealed its immunology to be quite distinct from that of the periphery.²⁰ Studies extending to the reproductive tract documented immunologic distinctions from both the periphery and the gastrointestinal tract.¹⁹ Further application to immunologic interactions at the site of embryo implantation have revealed additional unique immunologic characteristics. For instance, descriptions of the immune cells that populate both the female reproductive tract and the peri-implantation decidua implicate a particularly important role for 'innate-like' immune interactions during implantation and early pregnancy.^{21, 22}

Investigations on the immune cells of the female reproductive tract, particularly those populating the decidua at the time of implantation and during early pregnancy, have identified at least four potentially important lines of investigation and four unusual immune cell types: decidual granular lymphocytes, NKT cells, TCRγδ+ cells, and dicidual macrophages. 

Decidual Granular Lymphocytes 

Whereas the human endometrium is normally populated by T cells, macrophages, NK-like cells, and a very limited number of B cells,^{21, 22, 23} the relative proportions of these resident cells show menstrual cyclicity. The most dramatic changes are noted during the late luteal phase and in early pregnancy, when the proportion of one unusual type of immune cell rises to nearly 70–90% of the total endometrial lymphocyte populations.^{21, 22, 23} These unusual cells have been variably called decidual granular lymphocytes (DGLs), large granular lymphocytes (LGLs), and decidual NK cells.²⁴ The origin, functional capabilities, and physiologic purpose of these cells remain enigmatic, but their abundance at sites of implantation compels further study.^{25, 26} Although the effector functions of peripheral NK cells are known to be regulated by interactions between target-cell expressed classical MHC class I molecules (HLA-A and B) and activating and inhibitory receptors on the peripheral NK cell surface, the NK-like cells of the decidua exhibit receptors encoded by the killer immunoglobulin-like receptor (KIR) gene family that have been shown to react preferentially with trophoblast HLA-C molecules and possibly with nonclassical MHC class I products.^{27, 28} Decidual NK cells come into direct contact with the allogeneic fetally derived placental cells responsible for uterine artery remodelling during the first trimester of pregnancy. Here, they may serve to support trophoblast invasion, vascular remodelling, and vascular growth in the decidua through the production of a variety of cytokines and other angiogenic factors rather than serving the cytotoxic functions more typical of their peripheral counterparts.^{27, 28, 29, 30}  Although this particular cell type differs from similar cells isolated from the periphery, most believe it to be an NK cell variant. If so, the implantation site represents the largest accumulation of NK cells in any state of human health or disease. Investigations have shown that, rather than being derivatives of peripheral NK cells, decidual NK cells likely derive from endometrial NK cells that undergo proliferation during pregnancy.³¹ Given this relationship, it is unlikely that measures of peripheral NK cells would be good predictors of decidual NK cell competency or their effects on pregnancy outcomes.

NKT Cells 

Peripheral immune cells have been described that have characteristics of both NK cells and T cells.³² These NKT cells and their ligands (e.g. CD1) have been shown to be present in the decidua of animals^{33, 34, 35} and have been implicated in some forms of pregnancy loss.³³ In addition, the presence of NKT cells at the implantation sites of murine pregnancies has been shown to be mediated by interactions with fetally expressed MHC class I or class I-like products.³⁴ Initial investigations in humans have also demonstrated the presence of NKT cells within the endometrium and decidua.³⁶ The NK-like phenotype of these cells suggests innate immune function. 

TCR γ δ+ Cells and Decidual Macrophages

In the peripheral immune compartment, the vast majority of T cells express a T-cell receptor composed of an αβ heterodimer (TCRαβ+). In addition to TCRαβ+ T cells, the human reproductive tract is also populated by a subset of T cells with a distinctive T-cell receptor, the γδ heterodimer (TCRγδ+ cells). The number of these cells in the decidua increases in early pregnancy.^{37, 38, 39} TCRγδ+ T cells appear to fulfill functions quite distinct from their TCRαβ+ counterparts; these functions may include direct, non-MHC-restricted recognition of antigens within tissues.⁴⁰ TCRγδ+ T cells may fill a protective niche missed or poorly addressed by B cells and TCRαβ+ T cells. By secreting Il (interleukin)-10 and TGF (transforming growth factor)-β, TCRγδ+ T cells support trophoblast invasion and survival and may therefore play a very important role in the maintenance of early pregnancy.⁴¹ Similar to the vast array of functions that macrophages exert in the periphery, decidual macrophages are responsible for a variety of activities at the maternal–fetal interface, many of which are linked to pregnancy maintenance. Through their production of IL-10 and indoleamine 2,3-dioxygenase (IDO) activity, decidual macrophages have been shown to possess a predominately immunosuppressive phenotype, deemed M2 polarization. They aid in maternal immune tolerance to the semiallogenic fetus as well as transformation of uterine vasculature.⁴²

In conclusion, very characteristic immune effector cells populate the human decidua. To date, insufficient patient numbers have largely hampered investigations into whether alterations in these cellular populations (including T cells, decidual NK cells, and NKT cells) determine pregnancy outcome. Efforts made to target NK cells to treat infertility and pregnancy loss have been ineffective.⁴³ Still, the general consensus of these studies is that these populations are altered in RPL patients.^{44, 45, 46, 47, 48, 49}

Immune Cell Education and Homing to the Female Reproductive Tract

One very important concept in immunology that has particular application to pregnancy is that of immune tolerance. The concept is very well described for the immune cells of the peripheral immune system.¹⁶ Those immune cells arising in the bone marrow and destined to become circulating T lymphocytes typically pass through the fetal thymus, where they undergo a process termed thymic education. During thymic education, T cells 'mature' from CD4/CD8 double-positive cells to single-positive (CD4+ or CD8+) cells. This maturation occurs by a two-step process that spares only those T cells capable of recognizing one's own MHC class I or class II molecules but incapable of recognizing other self peptides. This education promotes T-cell tolerance to self by allowing selection and survival only of T cells that will not react against the organism's own peptides. 

If one considers the implanting blastocyst as the most common example of a successful allograft, one must question how the concepts of tolerance apply at the maternal–fetal interface against tissue that is most certainly not self. A few concepts require attention. First, are there characteristics other than those phenotypic differences already described that promote tolerance to fetal antigens among the resident immune cells in the maternal decidua? Both animal and human data suggest that the immune cells at the maternal–fetal interface are selected and maintained in ways that differ both from peripheral immune cells and from cells populating other mucosal sites, such as the intestine.^{19, 50} Second, if the cells populating the maternal–fetal interface have distinctive phenotypes and antifetal reactivity, how are these particular cells recruited to the site of implantation? Cellular recruitment (homing) to the intestine has been shown to rely on the presence of receptors (integrins) on the surface of particular immune cell subpopulations that specifically interact with ligands on the surface of the endothelial cells of blood vessels within the intestine (e.g. selectins, VCAM, MECAs).^{51, 52} Similar homing characteristics also appear to define immune cell types and vascular structures within the reproductive tract.^{51, 52, 53, 54} 

Control of Decidual Immune Cell Immunoreactivity 

Because safeguarding the fetus from maternal immune rejection is obviously crucial for successful reproduction, it is not surprising that evolution has selected a wide variety of overlapping and partially redundant immune adaptations at the maternal–fetal interface. In addition to distinctions in phenotype, homing, and antigenic education, the immune cells that populate the maternal decidua also appear to be affected by local immunomodulation. We know from both human and animal studies that immune responses to fetal antigens, including proinflammatory responses and those involved in transformation of the uterine vasculature, can be detected.^{55, 56, 57, 58} Thus, the regulation of these responses at the maternal–fetal interface may be critical for pregnancy maintenance. The concept that successful pregnancy requires some form of suppression of the maternal immune response is supported by reports that failure to downregulate maternal responses to recall antigens, such as tetanus toxoid and influenza, is associated with poor pregnancy outcome among RPL patients.⁵⁷ Some of the mechanisms hypothesized to be important in this localized immune regulation include alterations in localized cytokine profiles, the immune effects of pregnancy-associated hormones, and the in situ tolerance-promoting characteristics of tryptophan metabolism.

Cytokine Dysregulation

An essential role for isolated cytokines in reproductive success or failure is not well supported by animal studies. To date, although most cytokines have been gene-deleted in separate animal models, few of these factors appears to be essential to pregnancy maintenance. Two cytokines of the IL-6 family, leukemia inhibitory factor (LIF) and IL11, are required for blastocyst implantation,⁵⁹ but not for subsequent embryogenesis.^{60, 61} The production of these two cytokines is distinct in both space and time; LIF is produced by the endometrial epithelium during the blastocyst attachment/adhesion phase of implantation, whereas IL11 levels reflect decidualized stromal cell function postimplantation.⁶² Although this does not argue that other cytokines and soluble immunoregulatory factors are inconsequential, there certainly seems to be significant redundancy among these substances at the site of implantation.

It may not be particularly relevant to examine these factors in isolation. Rather, groups of cytokines may best reflect immune responses.^{63, 64} The division of antigen-stimulated immune responses involving CD4+ T cells into T helper 1 (Th1) responses and T helper 2 (Th2) responses appears to be useful, albeit somewhat oversimplified. This subclassification is defined both by the characteristics of the CD4+ cells present and by the cytokines these cells secrete. In general, the characteristics of the cytokine response depends on the environment in which naïve CD4+ (Th0) cells mature and differentiate: Th0 cells exposed to interferon (IFN)-γ become Th1-type cells, and those exposed to interleukin (IL)-4 become Th2-type cells.^{63, 64} Th1 responses are associated with inflammation and are typically characterized by the release of IFN-γ, IL-12, IL-2, and tumor necrosis factor (TNF)-β. Th2 cell responses are associated with antibody production and the cytokines IL-10, IL-4 and IL-5. ^{65, 66, 67} Although TNF-α can be secreted by both Th1 and Th2 cells, it is most often associated with a Th1 response.^{67, 68} Th1 and Th2 responses are intimately interdigitated, with each type of response promoting itself while limiting the alternative response.^{69, 70, 71, 72}

The human endometrium and decidua are replete with immune and inflammatory cells capable of cytokine secretion.^{73, 74, 75, 76, 77} Extension of the Th1/Th2 paradigm to pregnancy has generated provocative hypotheses. Most women with normal pregnancies appear to have a predominant Th2 immune response to undefined trophoblast antigens.⁷⁸ In contrast, some patients with recurrent pregnancy loss exhibit a dysregulation of their T-helper response to antigens at the site of implantation, with shifts toward Th1 inflammatory responses.^{78, 79, 80, 81} Further, Th1-type cytokines have been shown to be harmful to an implanting embryo.^{82, 83, 84} Depending on the individual series, 15–20% of nonpregnant women with a history of otherwise unexplained recurrent spontaneous abortion have been found to have evidence of abnormal in vitro Th1 cellular immune responses to trophoblast antigens. Fewer than 3% of women with normal reproductive histories have these responses.^{78, 79, 83, 85} 

Yet there is much more to the story. For instance, two additional T helper cell subsets have garnered significant attention in the field of pregnancy maintenance: Th17 and T regulatory (Treg) cells. Treg cells have an important role in the development of pregnancy-specific immune responses in mouse models.⁸⁶ In humans, Tregs purified from pregnant women suppress autologous peripheral blood mononuclear cell (PMBC) secretion of IFN-γ in response to paternal or unrelated PMBCs in mixed culture. These cells also suppress IL-4 secretion against paternal but not unrelated alloantigens during pregnancy.⁸⁷ Furthermore, studies have shown that Treg cells make up about 14% of the CD4+ T cells in the decidua and increases in the frequencies of Treg cells in the peripheral blood of pregnant women are seen during the first and second trimesters when compared to non-pregnant women.⁸⁸ An absence of Treg expansion has been found in decidual tissues from women suffering miscarriage, as well preeclampsia and preterm labor, suggesting that Treg cell function is important in the maintenance of pregnancy and in the development of a healthy placenta.^{89, 90} Some propose that peripheral Treg levels may even be used to predict the risk of pregnancy loss in newly pregnant women with a history of miscarriage.⁹¹ This has not been studied fully and should not be included in standard testing at this time.

Many of the autoimmune effects attributed to Th1 cells in the past are now being attributed to Th17 cells.⁹² The recent identification of these cells has implicated yet another set of mechanisms that may contribute to recurrent pregnancy loss. In normal pregnancy, Th17 cell proportions in the peripheral blood do not change.⁹³ In contrast, Th17 cell proportions increase in the peripheral blood and decidua of women with recurrent pregnancy loss. Furthermore, the levels of interleukins related to Th17 cells, such as pro-inflammatory IL17 and IL23, were also found to be significantly increased in both the peripheral blood and decidua of women with histories of RPL when compared to normal, early pregnant women.⁹⁴ Later reports examining the relationship between Th17 and Treg cells found an inverse correlation between the populations of these two cells types in the peripheral blood and decidua of women with pregnancy loss and demonstrated an elevated ratio of Th17/Treg cells in these women compared to normal pregnant and non-pregnant women.^{95, 96} Serum levels of IL17 have also been positively correlated with increased levels of Th17 cells and an increased Th17/Treg ratio. This suggests that Treg cells may function to inhibit IL17 production, and this suppressive effect may be impaired in women with recurrent pregnancy loss.^{96, 97}

The literature on reproductive immunology is often frustratingly inconsistent. Part of the frustration that arises when one attempts to synthesize the data surrounding cytokine dysregulation and poor reproductive outcome lies in the variations among study designs. Methods for the documentation of cytokine dysregulation among RPL patients have included confirmation of the abnormality within the endometrium^{98, 99, 100} and the decidua.¹⁰¹ Other investigators use peripheral lymphocytes from women with a history of RPL and stimulate them in vitro with trophoblast antigens.^{78, 102} Can these two types of studies be compared? Whether peripheral cytokine levels reflect T-helper cell dysregulation at the maternal–fetal interface and whether this localized dysregulation affects peripheral as well as local immune responses during pregnancy remain controversial.^{103, 104, 105} 

Reproductive Hormones

Notable gender differences in immune responsiveness^{106, 107, 108} likely reflect the fact that reproductive hormones have dramatic effects on peripheral cell-mediated immunity. Human chorionic gonadotropin (hCG), estrogen and progesterone have been shown to be potent immunomodulators, although the latter has received the most attention in terms of implications for the maintenance of the semiallogeneic implanting conceptus.¹⁰⁹ It is known that elevations in both estrogen and progesterone are seen early in pregnancy and these high levels are maintained throughout the course of the pregnancy.¹¹⁰In vitro evidence supports progesterone's immunosuppressive attributes by demonstrating an inhibition of mitogen-induced proliferation of and cytokine secretion by CD8+ T cells.¹¹¹ By playing a role in the modulation of inflammatory cytokine expression and Treg cell homing, hCG has also been implicated in immune tolerance towards the fetus in early pregnancy.

The significant effects of reproductive hormones at the maternal-fetal interface may be due to their increased concentration at this location when compared to concentrations found in the periphery.¹¹² Progesterone may promote the development of a cytokine microenvironment favoring pregnancy maintenance. Progesterone-mediated changes in T cell gene expression have been associated with Th2-type T helper cell responses and with increased expression of leukemia inhibitory factor.^{101, 113} High levels of progesterone stimulate the production of a protein called progesterone-induced binding factor (PIBF) by lymphocytes; the levels of PIBF continue to rise throughout gestation and drop after delivery.¹¹⁴ These elevated concentrations of PIBF have been shown to not only promote the differentiation of CD4+ T cells into Th2 cells, but also to increase Th1 type cytokine production. Furthermore, decreased levels of PIBF have been linked to preterm labor, miscarriage, and preeclampsia.^{115, 116} Because a shift in the intrauterine immune environment from Th2 to Th1 has been linked with early pregnancy loss,^{101, 117} the elevated concentrations of progesterone and PIBF characteristic of early pregnancy may promote an immune environment favoring pregnancy maintenance. Of course, progesterone is not present in isolation at the maternal-fetal interface, so in vitro studies addressing isolated effects on immunity present an incomplete picture of what actually occurs in vivo.

Although not specifically addressing pregnancy, several reports have focused attention on the general immunomodulatory effects of estrogens. Estrogens have been shown to improve immune responses in men after significant trauma/hemorrhage,¹¹⁸ to suppress cell-mediated immunity after thermal injury,¹¹⁹ and to protect against chronic renal allograft rejection,¹²⁰ all in animal models. In vitro, estrogens downregulate delayed-type hypersensitivity reactions and promote the development of Th2-type immune responses, particularly when present in high, and often supraphysiologic, concentrations.^{121, 122, 123} T reg cells numbers are known to increase during human pregnancy¹²⁴ and their levels are lower in women with spontaneous pregnancy loss.¹²⁵ This is thought to be the result of exposure to pregnancy hormones. Estrogen exposure potentiates suppressive function and proliferation of T reg cells in humans.¹²⁶  

Of course, one obvious question poses itself when hypotheses on progesterone/estrogen immunomodulation and their role in pregnancy maintenance are examined. If elevated levels of progesterone and estrogen are sufficient to promote tolerance to the implanting fetus, are pregnant women also systemically immunosuppressed? The answer to this question is complex. Some forms of immunoreactivity are indeed systemically suppressed in pregnant women. Examples of these effects include the symptomatic remission of such Th1-mediated autoimmune disorders as rheumatoid arthritis and multiple sclerosis among pregnant women with disease resurgence postpartum.^{127, 128} In contrast, Th2-mediated autoimmune disorders such as systemic lupus erythematosus tend to worsen during pregnancy.¹⁰⁹ Further, parasitic diseases and some viral diseases, including varicella, are particularly aggressive when first encountered during pregnancy.^{129, 130} Finally, although the levels of potentially immunosuppressive hormones are systemically elevated in pregnant women, their concentrations at the maternal–fetal interface appear to be significantly higher than in the maternal circulation.¹¹² This may help to delimit many immune effects to the site of implantation. To this point, it is important to note that pregnant women still have protection against the vast majority of infectious diseases and are subject to the consequences of the vast majority of autoimmune disorders.

Tryptophan Metabolism

A novel immunoregulatory pathway involving the amino acid tryptophan and its catabolizing enzyme, indoleamine 2,3 dioxygenase (IDO), has been proposed as a mechanism for maternal tolerance to the fetal allograft. T cells require tryptophan for activation and proliferation.^{131, 132} Local alterations in tryptophan supply or in tryptophan metabolism at the maternal–fetal interface would therefore be predicted to modulate immune cell function at this site. Evidence implicating tryptophan metabolism in maternal–fetal immune interactions arose initially in animal models. The inhibition of IDO in mice has been shown to promote loss of allogeneic but not syngeneic fetuses, an effect mediated by lymphocytes.¹³³ IDO activity has also been shown to aid in the suppressive role of Treg cells in murine studies.¹³⁴ Further, hamsters fed diets high in tryptophan have increased rates of fetal wastage.¹³⁵ IDO has been reported to be expressed in human uterine decidua¹³⁶ and alterations in serum tryptophan levels with increasing gestational age during human pregnancy have also been noted.¹³⁷

MHC Antigens in the Placenta 

Using available reagents and investigative techniques, early investigators failed to detect MHC-encoded transplantation antigens in the placenta. It was already known that these MHC class I and class II antigens promoted the rejection of transplanted allogeneic tissue in nonreproductive systems. Therefore, their proposed absence on the fetal semiallograft fostered the hypothesis that the maternal immune cells were tolerant of an implanting pregnancy because the invading tissues simply were not recognized as foreign. This hypothesis, and the data on which it was based, are partially correct. 

Indeed, MHC class II molecules are not expressed on the surface of placental trophoblast cells,^{138, 139} nor are the classical MHC class I transplantation antigens HLA-A and -B. However, a group of unique MHC class I products—HLA-C, HLA-E, and HLA-G—are expressed on a subpopulation of placental cells called extravillous cytotrophoblast cells.^{139, 140, 141, 142, 143, 144, 145} Extravillous cytotrophoblast cells are derived from the villous core of the placental cotyledon. During the course of early placental maturation, these cells take on invasive qualities. Invasion of extravillous cytotrophoblast cells may be regulated by a number of factors, including MHC expression patterns, integrin expression patterns/integrin switching,¹⁴⁶ and in situ oxygen tension. Invasion is characterized initially by movement of the extravillous cytotrophoblast cells from the tips of the placental anchoring villae toward positions deep within the maternal decidua. When they are present in the decidual vasculature, they can replace cells within the walls of arterial decidual vessels.^{147, 148} At this site, these fetally derived extravillous cytotrophoblast cells intimately contact maternal immune effector cells, thereby exposing the fetus to potential MHC-restricted recognition as nonself.

The MHC class I molecules on the extravillous cytotrophoblast cell are represented by the classical MHC class I product HLA-C and the nonclassical HLA-E and HLA-G products. Several lines of investigation have been pursued to determine why all placental cells downregulate expression of the classical transplantation antigens (HLA-A and HLA-B), whereas invasive extravillous cytotrophoblasts express HLA-C, HLA-E, and HLA-G.^{21, 25, 132, 148} There may be functional similarities among HLA-C, HLA-E, and HLA-G that promote recognition by immune cells at the maternal–fetal interface and downregulation of antifetal responses. Because NK cells recognize and kill cells lacking MHC I products, interactions of these class I products with the abundant NK-like immune cells at the maternal–fetal interface may avert NK cell receptor-mediated killing of extravillous cytotrophoblast.² Interactions of trophoblast HLA-C, HLA-E, and/or HLA-G with surface receptors on either NK cells or other maternal immune effector cells may modulate immune cell cytokine expression profiles.^{25, 26} Supporting the latter hypothesis, aberrant cytokine secretion has been documented when peripheral blood lymphocytes from RPL patients are stimulated by HLA-G-bearing cells in vitro.¹⁴⁹ Further, both decidual and peripheral immune cells have been shown to alter their cytokine secretion patterns when exposed to HLA-G.¹⁵⁰ The expression by fetus-derived cells of any or all of the trophoblast MHC class I products could also promote essential decidual and vascular invasion, spiral artery remodelling, and angiogenesis necessary for reproductive success.^{58, 151, 152} Altered trophoblast expression of HLA-G has been linked to disorders of placental invasion, such as pre-eclampsia.^{151, 153} Finally, the effects of MHC class I products on maternal reactivity toward the implanting pregnancy could be indirect. Secretion of soluble forms of HLA-C, HLA-E, and/or HLA-G into the maternal circulation may have pro-tolerance immune inductive effects at sites distant from the maternal–fetal interface.¹⁵⁴

Although associations between HLA typing and recurrent pregnancy loss have been suggested (see below), few investigations have specifically addressed the role of trophoblast MHC class I products (HLA-C, HLA-E, and HLA-G) in pregnancy loss and these studies have provided conflicting results. Two small studies investigating a possible association between polymorphisms in these MHC genes and the occurrence of RPL were negative.^{155, 156}  Newer studies have suggested that isolated polymorphisms of the HLA class Ib, particulary HLA-G, and KIR/classical HLA combinations, are indeed associated with reduced fertility and early pregnancy loss.^{157, 158,} Furthermore, when investigating the interplay between inhibitory and activating MHC receptor functions, some have proposed that a balance shift towards the activating state may predispose to recurrent miscarriage.^{159, 160} Others have addressed the possibility that because trophoblast cells do not spontaneously express either classical HLA-A and HLA-B products nor MHC class II products, an aberrant expression of these molecules might result in an adverse maternal immune response to the implanting fetus. Data supporting this hypothesis remain limited.^{161, 162, 163}

Humoral Immune Mechanisms 

The literature concerning the study of humoral/antibody-mediated immune responses and isolated or recurrent spontaneous pregnancy loss is controversial. In fact, much of this literature has now become obsolete. Responses to pregnancy-specific antigens certainly occur, and patients with RPL can display altered humoral responses to endometrial antigens.¹⁶⁴ Still, investigators continue to debate whether recognition of fetal antigens is essential/protective or harmful. 

Immunoprotection: Blocking Antibody Deficiency

One hypothesis promoting maternal recognition of the implanting pregnancy as essential rests on the supposition that humoral immune-mediated pregnancy loss results from a deficiency in the maternal production of protective (blocking) factors. The production of these protective factors (presumably antibodies) was thought to be the result of immune recognition of fetus-derived (paternal) alloantigens. It was suggested that these protective factors, termed 'blocking antibodies', would prevent the maternal, cell-mediated, antifetal immune response that was believed to occur in all pregnancies. The production of 'blocking antibodies' was therefore deemed crucial; in the absence of these factors, abortion would occur.¹⁶⁵

A deficiency in 'blocking antibodies' was demonstrated in vitro by maternal hyporesponsiveness in mixed lymphocyte culture with paternal stimulator cells.¹⁶⁵ Hyporesponsiveness in these assays was associated with RPL;^{57, 166, 167} however, this association is now thought to represent the immune effects of repeated losses rather than the cause of pregnancy loss. Further evidence refuting a role for the blocking antibody hypothesis in immune-mediated pregnancy loss comes from reports of successful pregnancies among women who do not produce serum factors capable of mixed lymphocyte culture inhibition¹⁶⁵ and among women who do not produce antipaternal cytotoxic antibodies.¹⁶⁶ 

Trophoblast-Lymphocyte Cross-Reactive Antigens

Out of the blocking antibody theory and its associated investigations came a number of related propositions that are no longer invoked. One of these involved the description of what was thought to be a novel HLA-linked alloantigen system called trophoblast-lymphocyte cross-reactive antigens, or TLX. The presence of TLX was first suggested by reports that polyclonal rabbit antisera could recognize both lymphocytes and trophoblast cells.¹⁶⁸ A number of initial reports then linked TLX to maternal blocking antibody deficiency and recurrent pregnancy loss.^{169, 170} TLX, however, was subsequently found to be identical to CD46, a complement receptor that is thought to protect the placenta from complement-mediated attack.¹⁷¹ CD46 is not a novel alloantigen; it is not even an alloantigen. CD46 can be found on a wide variety of cells and has no significant role in distinguishing self from nonself. 

HLA Sharing

Linked to the hypothesis that maternal recognition of the fetal antigens was essential to pregnancy maintenance were studies suggesting that parental HLA (classical MHC class I) sharing predisposed a couple with RPL to blocking antibody deficiency.^{172, 173} According to this theory, if maternal and paternal HLA products were too similar, fetally expressed paternal alloantigens would not elicit a maternal immune response to produce 'blocking antibodies'. The two largest studies addressing this possibility used a small religious sect, the Hutterite community, as their study population. The first study was prospective, with population-based controls, and demonstrated that HLA heterogeneity was not essential for successful pregnancy among the Hutterites.¹⁷⁴ A follow-up study, however, reported contrasting results. In fact, among the Hutterite population studied in this more recent, 10-year, prospective trial, complete sharing of the entire HLA region was associated with an increase of spontaneous pregnancy loss.¹⁷⁵ The conclusions drawn by these authors should be considered carefully, because the incidence of complete sharing of the HLA region between sexual partners in outbred populations is exceedingly rare; only isolated and significantly inbred populations have a chance for such HLA homogeneity. Therefore, the authors conclude that HLA typing is of no clinical utility in outbred populations. A similar link between MHC class II typing and recurrent pregnancy loss has also been reported,^{176, 177}  but its significance is unclear.

Bystander Effect: Autoimmunity and Pregnancy Loss

Antibody-mediated mechanisms for recurrent abortion in which humoral immune reactivity might directly result in adverse antifetal effects have been proposed. These include reports on associations between RPL and the presence of antisperm and antitrophoblast antibodies; however, each has been subsequently shown to have minimal clinical relevance.^{7, 162} In contrast, demonstrated associations between pregnancy loss and autoantibodies directed against nonreproductive tissues provide indirect evidence for humoral immune-mediated pregnancy loss. For instance, antithyroid antibodies have been linked to adverse pregnancy outcomes. One large, retrospective study reported an increased prevalence of these antibodies among women with a history of RPL, even in the absence of thyroid endocrinologic abnormalities.¹⁷⁸ Others have reported similar findings.¹⁷⁹ In contrast, two studies addressing pregnancy outcome and the presence of serum antithyroid antibodies in maternal blood have demonstrated neither an association with RPL nor a need for antithyroid antibody testing among RPL patients.^{180, 181} The differences between these findings are most likely related to differences in study designs. For example, older studies used a TSH value of 4.5 or 5.0 mlU/ml to define the upper limits of normal, but the American Endocrine Society now defines hypothyroidism as a TSH greater than 2.5 mlU/ml. As a result, it is possible that many of the “euthyroid” women examined in older studies were actually hypothyroid; therefore, it was clinical hypothyroidism, and not anti-thyroid antibodies, that caused the high prevalence of recurrent pregnancy loss.¹⁸² This is further supported by studies showing that the presence of antithyroid antibodies in euthyroid women does not affect pregnancy outcomes.^{179, 183}

Antiphospholipid Antibodies 

More consistent, and more therapeutically relevant, have been reports linking organ-nonspecific autoantibodies associated with APS to adverse pregnancy outcomes. Clinical and laboratory evidence for antiphospholipid antibodies were originally based in vivo on thrombosis and in vitro on prolongation in one of the phospholipid-dependent coagulation tests such as the activated partial thromboplastin time or the Russell viper venom time. With time, patterns among the thrombotic complications associated with antiphospholipid antibodies were noted and criteria proposed that defined APS. Although many of the complications of APS are systemic, some are pregnancy-specific, including spontaneous abortion, premature labor, premature rupture of membranes, stillbirth, intrauterine growth restriction, and pre-eclampsia.¹⁸¹ The importance of pregnancy-specific complications in the diagnosis of APS gained renewed prominence when the defining criteria for the disorder were reassessed in 1998 (the Sapporo criteria) and again in 2006 (Table 1, see below).^{184, 185, 186} 

Table 1. Criteria for diagnosing the antiphospholipid antibody syndrome 

(Adapted from Miyakis et al.¹⁸⁶) 

Clinical data have linked antiphospholipid antibodies such as anticardiolipin or lupus anticoagulant to adverse pregnancy outcome.^{187, 188} In a large series of couples with recurrent abortion, the incidence of APS was 3–5%.⁷ Pathologic data, however, do not consistently demonstrate causal involvement of APS in pregnancy loss. For instance, although placental infarction, abruption, and hemorrhage are considered to be defining placental lesions in APS, they are often missing in women with antiphospholipid antibodies.¹⁸⁹ Conversely, women with recurrent abortion but no biochemical evidence of antiphospholipid antibodies often demonstrate placental findings consistent with the diagnosis of APS.¹⁹⁰ 

Antiphospholipid antibodies, however, have mechanistic links to placental thrombotic changes and subsequent adverse pregnancy outcome. Antiphospholipid antibodies may indirectly promote thrombosis. Associations between antiphospholipid antibodies and alterations in prostacyclin/thromboxane metabolism have been demonstrated. Local alterations in these pathways at the level of the maternal–fetal interface could promote vascular constriction, platelet adhesion, and placental infarction.^{191, 192, 193, 194} Similarly, antiphospholipid antibody positivity has been linked to reductions in the levels of placental antithrombotic products, such as annexin V, among women with RPL.¹⁹⁵ Antiphospholipid antibodies may promote localized atherosclerosis. Supporting this concept, atherosclerosis has been shown to develop rapidly in the spiral arteries of patients positive for antiphospholipid antibodies.¹⁹⁶ 

In contrast to the focus on the thrombophilic qualities of antiphospholipid antibodies, newer studies have supported a primarily autoimmune effect of these antibodies. In fact, low thrombotic event rates can be demonstrated in women with APS and past histories of early loss. Rather than the finding of blood clots in placentas following spontaneous miscarriage, evidence has been found supporting defective endovascular trophoblastic invasion, proliferation, chemotaxis, maturation, and differentiation.¹⁹⁷ To illustrate, IgM against phosphatidylserine inhibits the trophoblast syncytialization¹⁹⁸ in cell cultures and sera from antibody-positive RPL patients inhibits trophoblast adhesion to endothelial cells in vitro.¹⁹⁹ Other studies have linked antiphospholipid antibodies to decreases in the release of hCG from human placental explants and the induction of an inflammatory response secondary to complement activation on the trophoblast surface.²⁰⁰

Attention in the field of antiphospholipid antibodies and pregnancy loss has turned partially away from the antigens cardiolipin and phosphatidylserine but toward a protein cofactor called β2 glycoprotein 1.^{201, 202} This glycoprotein acts as an essential cofactor in antiphospholipid antibody reactivity against membrane phospholipid and, in many instances, may provide the true antigenic determinant for a given antiphospholipid antibody.^{201, 202} Because both extravillous cytotrophoblast and syncytiotrophoblast cells have been demonstrated to synthesize β2 glycoprotein 1,²⁰³ the placental microenvironment has the potential for robust antigenicity. Anti- β2 glycoprotein 1 antibodies are an independent risk factor for thrombotic events and for pregnancy loss in APS patients.^{204, 205, 206} They are the sole antibody detected in 3–10% of APS patients.^{207, 208, 209}

Although much is now known about the immunology of the human genital tract, many questions remain. Immunologic interactions at the maternal–fetal interface are particularly intriguing, but in most cases, basic immunologic findings in the laboratory have yet to be fully applied toward diagnostic or therapeutic use at the bedside. Much progress will undoubtedly be made during the next decade; however, in this era of evidence-based medicine, very few immunologic tests or therapies are presently warranted in the diagnosis and treatment of patients with isolated or recurrent spontaneous pregnancy loss. Most of these tests remain experimental at best and demand further investigation before they can be clinically applied. Still others are of historical interest only. 

Antibody Testing 

As mentioned above, the 1998 reassessment of the criteria used for the diagnosis of APS made complications of pregnancy defining clinical criteria.¹⁸⁴ A brief review of the 2006 revised Sapporo criteria (see Table 1) demonstrates that although most patients with an isolated spontaneous pregnancy loss do not warrant testing for APS, many patients with recurrent losses do require further laboratory assessment. Screening for the lupus anticoagulant (activated partial thromboplastin time or Russell viper venom testing) and assessment for the presence of anticardiolipin antibodies and anti- β2 glycoprotein 1 antibodies is indicated for most patients with RPL but few with isolated losses. 

Although investigators have suggested testing for a large number of isolated organ-specific and nonspecific autoantibodies among patients with RPL, testing for antibodies other than the lupus anticoagulant, anticardiolipin antibodies and anti- β2 glycoprotein 1 antibodies remains controversial.^{186, 210, 211} For example, although some have shown the prevalence of the organ-specific antithyroid antibodies to be increased among patients with a history of RPL,^{178, 180} others have found no association between the presence of antithyroid antibodies and recurrent loss.¹⁷⁹ Even if the prevalence of antithyroid antibodies is increased among these patients, their clinical and mechanistic significance remains unclear.²¹² 

There is no consistent evidence supporting the use of extensive panels of serum or site-specific autoantibodies or alloantibodies (including antinuclear antibodies and antipaternal cytotoxic antibodies) in the evaluation of either isolated or recurrent pregnancy loss. Use of these panels often serves only to verify the statistical tenet that if the number of tests performed reaches a critical limit, at least one will be positive in every patient. 

Cytokine Testing 

The hypothesis that cytokine dysregulation (Th1/Th2 imbalance) at the maternal–fetal interface may affect pregnancy maintenance is gaining momentum, although the number of cytokines involved and the complexity of their local interactions makes interpretable study in humans a daunting undertaking. Although there exists evidence that these effects may be clinically important in some cases of pregnancy loss.^{21, 213} Our ability to diagnose such dysregulation, however, has been hampered by a number of variables, many of which are difficult to circumvent. We have herein described that the immune cells populating the maternal–fetal interface are unique and characteristic. Therefore, although Th1/Th2 dysregulation may be occurring within the immune microenvironment at the site of implantation, peripheral immune cells may be spared. Safe methods to assess the immune characteristics at the site of an ongoing and desired pregnancy do not presently exist. Can significant alterations in cytokine profiles that are isolated within the uterus be detected using peripheral serum testing? This remains unclear. Some small studies have documented peripheral shifts toward Th1 cytokine profiles in RPL patients who subsequently lose their pregnancy, but not in similar patients who have a successful pregnancy outcome.¹⁰⁵ Other larger studies have failed to show any association between peripheral cytokine alterations and pregnancy outcome among patients with a history of recurrent losses.²¹⁴ One additional caveat is that pregnancy demise is itself accompanied by dramatic immunologic alterations within the uterus directed toward the presence of necrotic tissues. Therefore, studies on cytokine dysregulation, including those reporting a peripheral shift toward Th1 profiles at the time of fetal demise among patients with RPL,¹⁰⁴ may not be able to differentiate cause from effect.

Immune Cell Evaluation 

Efforts have been made to evaluate both the proportions and activity of immune cells isolated from patients with a history of spontaneous pregnancy loss. These studies are also limited both by the inability to sample sites of implantation during desired pregnancies and by the tenuous association between most decidual and peripheral immune characteristics. Diagnostic use of mixed lymphocyte cultures, originally proposed to evaluate the presence of 'blocking antibody' activity, has not been consistently validated. The identification of large numbers of NK-like cells at sites of implantation led to the hypothesis that alterations in the prevalence and activity of these cells might significantly affect pregnancy maintenance. Although the rationale may be questioned, this hypothesis has been extended to promote testing of the prevalence and activity of peripheral NK cells among patients with RPL.^{215, 216} Again, because decidual and peripheral NK cells are markedly different, one must resolve whether intrauterine events are reflected using NK cells isolated from the periphery of these women. Two small studies reported that the evaluation of these peripheral NK cell characteristics predicts prognosis and assists in the counseling of patients with RPL.^{215, 216} These studies have not been adequately substantiated and, in fact, are contradicted by a 2011 comprehensive review of over 780 publications. In this review, the authors did not find consistent evidence linking peripheral NK cell testing results with pregnancy outcomes among women with a history of RPL.²¹⁷ Their conclusions are further supported by recent flow cytometric analyses demonstrating distinct differences between the phenotypes of peripheral and endometrial NK cells in women with RPL.²¹⁸ In summary, the weight of evidence does not support testing for peripheral NK cell activity or basing treatments on the results of peripheral NK assays in women with recurrent pregnancy loss.²¹⁹

HLA Screening

Testing for parental HLA profiles is not indicated in outbred populations. Those reports demonstrating that HLA sharing is associated with poor pregnancy outcomes were strictly limited to the specific populations studied.^{174, 175} In the most recent and definitive of these investigations,¹⁷⁵ only patients with complete sharing of the entire HLA locus demonstrated an increased risk for adverse pregnancy outcomes. As discussed, the chance of discovering complete sharing of the entire HLA region among partners in outbred populations is infinitesimally small. Therefore, generalized HLA screening of patients with either isolated or recurrent spontaneous pregnancy loss cannot be rationalized.

Like immunologic testing, 'immunologic' therapies for patients with presumed immunologic pregnancy losses have been best substantiated among patients with concomitant hypercoagulability. Treatment regimens for patients with adverse pregnancy outcomes and APS is appropriately discussed alongside other proposed therapies for immune-mediated pregnancy loss, even though the treatments are themselves best subcategorized as antithrombotic rather than immunomodulating. That said, the antithrombotic agent heparin has been shown to bind to antiphospholipid antibodies in vitro, an effect that suggests that it possesses distinct immunoregulatory properties distinct from its anthithrombotic effects.²²⁰ In addition, heparin promotes trophoblast invasion,²²¹ inhibits trophoblast apoptosis,²²² restores placental hCG production^{221, 223} and directly alters cellular immune responses.²²⁴

Antithrombotic Therapy 

Aspirin and Unfractionated Heparin

The safety and efficacy of the combined use of aspirin (ASA) and heparin has been best studied among pregnant women who carry the diagnosis of APS. Although variations have been described,^{225, 226, 227} a typical medical regimen for the antithrombotic management of these patients is shown in Table 2.

Table 2. Antithrombotic treatment of antiphospholipid antibody syndrome during pregnancy

Therapy with combinations of ASA and unfractionated heparin in this patient population is not without associated risks. Increases in the incidence of pregnancy-specific complications, including premature labor, premature rupture of the membranes, intrauterine growth restriction, intrauterine fetal demise, pre-eclampsia, and abruptio placenta have all been reported.^{226, 227, 228}  Specific maternal risks include gastric bleeding and osteopenia. Pregnant patients with APS should be considered high risk and managed in conjunction with a perinatologist. Despite these risks, a recent systematic literature review and meta-analysis examining different aspirin and heparin treatment regimens confirmed the finding of improved live birth rates in women with antiphospholipid syndrome treated with unfractionated heparin plus aspirin compared with aspirin monotherapy.²²⁹ This treatment combination, however, has not been found to be beneficial in women with recurrent miscarriage but without APS.^{230, 231}

Low-Molecular-Weight Heparin

The inconvenient dosing schedules and unacceptable side effect profiles associated with unfractionated heparin use led to the development of low-molecular-weight heparin (LMWH). When compared with unfractionated heparin, LMWH has an increased antithrombotic ratio, allowing treatment of inappropriate clotting with fewer bleeding side effects.²³² The incidence of thrombocytopenia and osteoporosis are similarly lower using LMWH. The prolonged half-life of LMWH compared with its unfractionated counterpart permits less frequent dosing and less frequent thrombotic monitoring. Both promote patient acceptance and compliance. 

Whereas LMWH has been shown to be safe, effective, and convenient in the treatment of many clotting disorders unrelated to reproduction,²³² their use in pregnancy is only now being comprehensively evaluated. Still, specific use of LMWH (in combination with low-dose ASA) among patients with RPL and thrombotic disorders has been quickly adopted clinically. To date, the safety and efficacy of LMWH use have been suggested by interventional trials conducted among RPL patients with APS^{226, 233} and among those with activated protein C resistance associated with factor V Leiden.^{234, 235, 236} The same meta-analysis that supported combined therapy with aspirin and unfractionated heparin, however, did not find significant efficacy for a combination of low molecular weight heparin and aspirin, but supported the need for larger controlled trials.^229,

Aspirin Prophylaxis

Patients seeking therapy for reproductive disorders, including those experiencing isolated and recurrent spontaneous pregnancy losses, tend to be well-informed. They are therefore quick to apply the well-publicized cardiovascular benefits and infrequent side effects of daily prophylactic low-dose ASA to their own medical condition. Many patients with histories of recurrent loss are now either self-prescribing this therapy or inquiring about its efficacy. To summarize the data addressing prophylactic ASA use among patients with RPL, current data does not support its clinical application either among patients with unexplained RPL with losses at less than 10 weeks or in those with coagulopathies.. Although ASA/heparin has been demonstrated to be effective in the prevention and treatment of adverse pregnancy outcomes among patients with APS (see above), the sole use of low-dose ASA among patients with RPL and APS appears ineffective.^{225, 237} Other studies have reached similar conclusions against the efficacy of aspirin in preventing future pregnancy loss. In one large, prospective interventional trial, the use of prophylactic low-dose ASA could not be recommended for the treatment of patients with unexplained recurrent early pregnancy losses.²³⁸ Its use among women with unexplained late pregnancy losses (later than 13 weeks gestation) did, however, appear promising.²³⁸ Most recently, a multicenter, double-blind, placebo-controlled trial was conducted to more adequately assess the use of aspirin in preventing pregnancy loss in women with one loss at less than 20 weeks’ gestation during the previous year and in women with one to two previous losses with no restriction in gestational age at the time of loss. In the group of women with one to two previous losses, preconception-initiated low-dose aspirin did not increase the rate of live births or reduce rates of pregnancy loss. The group of women with a single loss at less than 20 weeks’ gestation in the previous year, however, did experience a higher pregnancy rate and live birth rate. Overall the authors concluded from these somewhat confusing results that aspirin could not be recommended to decrease the rate of pregnancy loss.²³⁹

Immunomodulating Therapies 

Other than the antithrombotic approaches, all alternative therapies for immune-mediated spontaneous pregnancy loss can be classified as immunoregulatory. Historical controversy over whether an immune response to the implanting pregnancy was essential or detrimental has led to the development and use of both immunosuppressive and immunostimulating therapies for the treatment of these patients. Although a large number of immune interventions have been proposed, these therapies remain experimental, and many have been proven ineffective; some even appear to be unsafe. 

Although immune interactions seem to be involved in some instances of isolated spontaneous pregnancy loss, most of the published interventional studies involving immunotherapy for the prevention of subsequent pregnancy loss have been limited to patients with recurrent losses. Even these studies, however, have been limited by poor study design, small sample size, lack of patient prestratification by maternal age and number of prior losses before randomization, and a variety of methodologic and statistical inaccuracies. Another glaring omission in most studies has been the inconsistent analysis of aborted tissues for chromosomal abnormalities, because approximately 60% of individual losses are undoubtedly due to numeric chromosomal aberrations.²⁴⁰  Immunotherapy for RPL has been addressed in a number of excellent reviews^{6, 241, 242} and will be briefly summarized here. 

Immunosuppressive Therapies 

The hypothesis that inappropriate cellular or humoral immune responses to the implanting fetus causes pregnancy loss fostered the development of several immunosuppressive therapeutic approaches to the treatment of patients with immune-mediated RPL. These suppressive therapies have included the use of standard immunosuppressive drugs, immunosuppressive reproductive hormones, and intravenous immunoglobulins. 

Standard Immunosuppressive Therapies

Several immunoregulatory regimens, including the use of cyclosporine, pentoxifylline, and nifedipine, have been suggested based on related efficacy trials in the autoimmunity and transplantation literature. Extension of most of these interventions to patients with RPL, however, has been hampered by known or unknown maternal and fetal risks. For instance, plasmapheresis has been used for the prevention of thrombotic complications associated with the presence of antiphospholipid antibodies, and only one isolated study has examined its efficacy among patients with recurrent abortion and antiphospholipid antibodies.²⁴³ The best-studied of the standard immunosuppressive approaches has involved the use of corticosteroid therapy. Although there exists evidence that prednisone is useful in the treatment of patients with recurrent pregnancy loss and APS,²⁴⁴ antithrombotic therapy with heparin and aspirin has been shown to be as or more effective and safer in this same patient population. A large, randomized, placebo-controlled trial addressing the use of prednisone in combination with low-dose ASA among women with autoantibodies and RPL may relegate this therapy to historical interest only. In this investigation, pregnancy outcomes for treated and control patients were similar, but the incidence of maternal diabetes and hypertension and the risk of premature delivery were all increased among those treated with prednisone and ASA.²⁴⁵ 

Immunosuppressive Reproductive Hormones

Both estrogen and progesterone have immunomodulating effects.^{106, 109, 111, 246, 118, 119, 120, 121, 122, 123, 127, 247} The well-described and marked elevations of these reproductive hormones during pregnancy make them attractive candidates for use as immunosuppressive agents among patients with presumed immune-mediated pregnancy losses. As mentioned previously, in vitro evidence suggests that progesterone alters cytokine responses,^{111, 246, 248, 249} which might promote its appropriate use in patients with RPL and cytokine dysregulation at the maternal–fetal interface. Documenting such local dysregulation, however, is problematic. Small studies on empiric progesterone use in patients with unexplained RPL have presented conflicting results. However, a meta-analysis has advocated its use in patients with recurrent, but not isolated pregnancy losses.²⁵⁰ Vaginal administration of progesterone may increase local, intrauterine concentrations of the hormone better than systemic administration and may therefore provide improved local immunosuppression while averting systemic side effects.  

Intravenous Immunoglobulin

The pooling of the immunoglobulin proteins isolated from the serum of a large number of volunteer blood donors results in a blood product known as intravenous immunoglobulin (IVIg). The use of IVIg may be accurately categorized as immunosuppressive therapy, although the mechanistic basis for IVIg immunoregulation is poorly understood. Still, the list of potential immunoregulatory mechanisms is long and includes: decreased autoantibody production and increased autoantibody clearance, T-cell and Fc receptor regulation,²⁵¹ complement inactivation,²⁵² enhanced T-cell suppressor function, decreased T-cell adhesion to the extracellular matrix,²⁵³ and downregulation of Th1 cytokine synthesis.²⁵⁴ Because IVIg is a blood product, its use is associated with a variety of real and potential risks. Like other blood products, IVIg is screened for transmissible agents such as HIV and hepatitis. However, up to 200 different donors are required to manufacture one vial of IVIg, and the transmission of prion-related disorders (e.g. 'mad cow disease', Jakob-Creutzfeldt disease) and other disorders that are not readily detected in routine screening should be considered a potential risk of therapy. Side effects of IVIg therapy include nausea, headache, myalgias, hypotension, and in some instances anaphylaxis.²⁵⁵ 

Therapy with IVIg for recurrent pregnancy loss is expensive, invasive, and time-consuming, requiring multiple intravenous infusions over the course of pregnancy. Although the list of financial and medical costs appears daunting, does treatment outcome using IVIg among patients with RPL justify the expense, side effects, and risks of this immunointervention? The data addressing the use of IVIg in patients with RPL is fairly extensive and, until recently, the answer to this question remained unclear. Although a number of relatively small studies using a variety of treatment protocols produced conflicting results, larger and more consistent studies have been unable to conclusively justify the use of IVIg in patients with unexplained (and presumed immunologic) RPL.^{256, 257, 258, 259, 260, 261, 262, 263, 264} Still, some may argue that the use of IVIg may be indicated among those specific patients with autoimmune-mediated pregnancy loss associated with APS.^{265, 266} Although they would be useful, no safety and efficacy studies have been performed comparing the use of IVIg with that of unfractionated heparin/ASA or LMWH/ASA in these patients.  The results of a large, randomized, multicentered placebo-controlled trial evaluated treatment with IVIg to saline placebo in women with idiopathic secondary recurrent miscarriage, defined as “a history of at least one prior ongoing pregnancy followed by three or more consecutive unexplained miscarriages” and concluded that no significant effect was found in patients treated with IVIg; this is the largest randomized controlled trial on this topic to date.²⁶⁷

Immunostimulating Therapies 

The development of immunostimulating therapies for the prevention of spontaneous pregnancy losses was based on the premise that an appropriate immune recognition and reaction to the implanting fetus was essential to subsequent pregnancy maintenance. There may be significant problems with this underling premise. First, although we know that pregnancies are specifically recognized and anti-paternal immune reactions mounted by the pregnant woman,¹⁶⁴ a requirement for these responses has never been definitively demonstrated. Further, the fetal/paternal alloantigens recognized by the maternal immune system have never been appropriately isolated and identified. This explains the variety of antigen preparations and routes of administration that have been promoted during the development of immunostimulating 'therapies' for RPL. For instance, syncytiotrophoblast microvillus plasma membrane vesicles have been prepared and administered intravenously to mimic the fetal cell contact with maternal blood that normally occurs in pregnancy.²⁶⁸ The erroneous belief that TLX was part of an idiotype/anti-idiotype control system²⁶⁹ led to the use of third-party seminal plasma suppositories for immunostimulation in women with histories thought to be consistent with immune-mediated RPL.²⁷⁰ Neither therapy was proven effective. 

Leukocyte Immunization 

The best-studied immunostimulation therapy to be promoted for use among patients with immune RPL involved the intravenous administration of either paternal or pooled donor leukocytes. Because leukocytes from either source supply innumerable alloantigens, it is not surprising that maternal immune responses would be stimulated. Immune mechanisms responsible for the transfer of these systemic responses to potential fetoprotective effects were suggested^{271, 272, 273, 274} but never proven. Like IVIg, paternal or donor leukocytes represent blood products, and their administration to a third party carries the same risks as with other blood products. Reported risks specific to the use of leukocyte immunization in pregnant women encompass a variety of adverse fetal and maternal effects, including graft-versus-host disease, severe intrauterine growth restriction, and potentially fatal fetal thrombocytopenia.^{7, 261, 275, 276, 277, 278, 279} Again, as with the use of IVIg in RPL patients, the question arises of whether the treatment outcome with the use of leukocyte immunization among patients with RPL justifies the expense, side effects, and risks of this immunointervention. The data addressing this question are similarly heterogeneous and similarly confusing.^{275, 276, 280, 281, 282, 283} However, the most recent and largest trial evaluating the efficacy of leukocyte immunization in patients with unexplained RPL was a part of the Recurrent Miscarriage Study (REMIS).²⁸³ This investigation involved more than 90 patients per treatment arm and was prospective, placebo-controlled, randomized, and double-blinded. The study indicated that paternal leukocyte immunization was not efficacious in couples with unexplained RPL; indeed, women receiving leukocyte immunization were more likely to experience a repeat loss than women receiving the placebo.

Immunologic interactions are complex. To those studying the immune interactions surrounding implantation, it is sobering to note that pregnancy does not appear to require an intact maternal immune system. To this point, women and animals who lack immunoglobulins (agammaglobulinemic) successfully reproduce.²⁸⁴ Women with severe immune deficiencies, mice that lack T and B cells (SCID mice), and mice with congenital absence of their thymus (nude mice) also carry pregnancies to term deliveries.²⁸⁵ Defining the role of a particular immune factor in pregnancy maintenance is therefore likely to be neither simple nor expedient. 

The prognosis for a patient with RPL is good. Most women with a history of RPL is more likely to carry their next pregnancy to term than to miscarry. In fact, for patients with a history of RPL, the risk of subsequent pregnancy loss is estimated to be 24% after two clinically recognized losses, 30% after three losses, and 40–50% after four losses.²⁸⁶ These prognostic data, while helpful, still frustrate attempts to document the efficacy of interventional therapy. Unless interventions are remarkably effective, treatment groups must be almost prohibitively large to demonstrate alterations in outcome. 

Clinical studies on immune-mediated pregnancy loss have been regrettably difficult to interpret. The disease itself is hard to accurately diagnose, typically relying solely on the exclusion of other etiologies. RPL represents a number of different specific disorders. With inconsistent clinical definition, trials involving these patients are difficult to compare and evaluate. Trial design is frequently substandard, with lack of rationale, lack of appropriate controls, and poor statistical analysis limiting the ability to draw rational conclusions from reported results. 

The future, however, is not bleak. The rapid expansion in our understanding of the molecular and cellular immune environment at the maternal–fetal interface will surely stimulate clinical applications, and clinical insight will translate into therapeutic advances. The next several decades will undoubtedly witness significant inroads into the diagnosis and treatment of immune-mediated isolated and recurrent spontaneous pregnancy loss.