Recurrent spontaneous abortion (RSA) is defined as ≥3 pregnancy losses within the first 20 weeks of gestation, whereas an increasing number of researchers hold the opinion that ≥2 sequential pregnancy losses are sufficient for defining RSA. Approximately 5% of females of childbearing age suffer from RSA and the cause is unknown in over half of the cases, which makes it difficult to perform any evidence-based diagnosis and treatment (1). The etiology of RSA involves multiple factors, including immune disorders, chromosomal abnormalities, endocrine disorders and uterine abnormalities, among which immune disorders are a key factor, and efforts have been made in recent years to control RSA through agents targeting the immune system (2,3). Currently available treatments for RSA, including aspirin, anticoagulants and hormonal support, have been indicated to exert immunomodulatory effects, suggesting that targeting immunological disorders may be of value in the treatment of RSA (4).

Tumor necrosis factor (TNF)-α is a pro-inflammatory T helper (Th)1 cell cytokine, which regulates the inflammatory mechanism in various pathologies, including RSA (5–7). Regarding the immunological background of RSA, the imbalance between Th1 cytokines, particularly TNF-α, and Th2 cytokines, such as IL-10, is profound; therefore, TNF inhibitors, which are commonly used treatments for inflammatory and immune diseases, have been applied for the treatment of females with RSA (8–16). The focus of the present review was to outline the mechanistic involvement of TNF-α in RSA immune disorder and discuss clinical studies that have attempted to improve pregnancy outcomes in patients with RSA by using TNF inhibitors.

The immunological mechanism is largely altered during pregnancy in response to the development of the fetomaternal relationship. Embryos are considered as allografts to the mother, as the antigens expressed by the embryos at the fetomaternal interface are paternal and foreign, and these antigens may subsequently induce alloimmune responses in the mother (3,17); this means that the maternal immune tolerance may be broken down by the implantation of the embryo and, thus, pro-inflammatory cytokines (such as the TNF-α) may be released (Fig. 1). In addition, certain autoimmune disorders, such as the antiphospholipid syndrome (APS) and positivity for other antibodies such as anti-citrulline protein antibody, are also characterized by the overexpression of pro-inflammatory cytokines in peripheral blood mononuclear cells (PBMCs), and the overproduced TNF-α may be released into the local circulation of the implantation site, further contributing to embryonic/fetal morbidity (Fig. 1) (18,19). Approximately 5–15% of females with RSA present with APS, which refers to the presence of antiphospholipid antibodies (aPLs) and their association with venous/arterial thrombosis and hypercoagulation. The presence of aPLs has been indicated to coexist with activated CD4⁺ T cells and disrupted Th1/Th2 cytokine homeostasis (18). The dysregulated immune responses damage the placental villi and embryonic tissues, interrupting the implantation of the fertilized oocyte and embryonic development, resulting in pregnancy failure. For instance, the Th17 cell and regulatory T cell (Treg) imbalance (specifically, increased numbers of Th17 cells and decreased numbers of Tregs) may be an important immune factor contributing to the occurrence of RSA (20). In addition, in the case of RSA, it has been hypothesized that the adaptive immune system recognizes the alloantigen from the father, so that any further attempts at fertilization with the same partner would result in failure (3).

As a critical part of the innate immune system, natural killer (NK) cells are considered as a strong risk factor for RSA and they may be classified as peripheral and uterine NK cells (1). Uterine NK cells were initially referred to as large granular lymphocytes due to the presence of granules in their cytoplasm and shared properties (CD56⁺) with the peripheral NK cells. However, uterine NK cells (CD16⁻/CD3⁻) are different from peripheral NK cells (CD16⁺/CD3⁺) regarding the expression of the CD16 and CD3 antigens (21). Peripheral NK cells are antiviral and antineoplastic due to their cytotoxic nature, while uterine NK cells are less cytotoxic and they produce receptors responsible for the recognition of antigens on the extravillous trophoblast surface and the secretion of cytokines (21).

Uterine NK cells are specifically reported to be associated with various reproductive disorders, including RSA, uterine fibroids and fetal growth restriction (22). Several studies using immunohistochemistry as the detection tool have reported that uterine NK cell numbers are increased in the endometrium of females with RSA at the mid-secretory phase. By contrast, studies using flow cytometry have indicated no change in uterine NK cell numbers in females with RSA (23–25). Due to the discrepancies in uterine NK cell numbers among different patients, it remains to be determined whether uterine NK cells are functional in RSA. It was previously suggested that mRNA expression is altered in endometrial uterine NK cells in early pregnancy compared with non-pregnancy NK cells (26); however, other studies indicated that endometrial uterine NK cells are non-functional in females with RSA (27,28). In addition, the adverse effects of uterine NK cells on trophoblasts, such as invasion of the trophoblast and development of trophoblast abnormalities, still require confirmation (29–31). It has been indicated that uterine NK cells are capable of promoting angiogenesis, since they also produce important pro-angiogenic factors, including vascular endothelial growth factors, placental growth factors and angiopoietin 2, and uterine NK cell deficiency caused poor spiral artery development in a mouse model (21). However, whether the involvement of uterine NK cells in angiogenesis contributes to the pathogenesis of RSA remains to be fully elucidated.

In immune responses, a Th1/Th2 cell ratio <10.3 is considered a safe range for successful pregnancy (32). The Th1 cells, along with the subsequent immune activation, increase the secretion of IFN-γ, IL-2 and TNF-α, which induce toxicity against the trophoblast and trigger placental injury via activating immune cells, such as NK cells. With regard to Th2 cells, they produce cytokines such as IL-4, IL-13 and IL-10, which promote trophoblast growth and favor pregnancy maintenance. The Th2 cells protect the embryo by inhibiting immune rejection by the maternal immune system (33). The evaluation of Th1/Th2 cytokines suggests that, in the first trimester of pregnancy, the concentrations of Th1 cytokines were higher in PBMCs of females with RSA compared with those with normal pregnancies (34). In addition, females who are prone to RSA but have a successful pregnancy tend to express lower Th2 cytokine levels compared with those who are not prone to RSA and suffer a pregnancy loss (35). Detailed flow cytometry data also revealed that females with RSA have significantly lower IL-10-producing CD3⁺/CD8⁺ T-cell counts but higher TNF-α-producing CD3⁺/CD4⁺ T-cell counts compared with non-pregnant fertile controls (36). The aforementioned evidence suggests the implication of Th1/Th2 imbalance in the pathogenesis of RSA (Fig. 1).

Tregs suppress the alloimmune response towards the fetus (37). In cases of allogeneic organ grafts, Tregs prevent autoimmunity, whereas in RSA, the number of available studies is limited (38). In a mouse model of spontaneous abortion, Tregs (CD4⁺/CD25⁺) from normal pregnant/non-pregnant mice inhibited lymphocyte proliferation and IFN-γ secretion in vitro, whereas in vivo, the adoptive transfer of Tregs from normal pregnant mice to maternal mice with abortion prevented fetal rejection, indicating the function of Tregs in promoting maternal tolerance of the fetus (39).

TNF-α is a Th1 pro-inflammatory cytokine that is located on chromosome 6p21.3 and acts via binding to TNF receptor (TNFR)I and TNFRII on cells. It was previously reported that TNF-α is abundant in the body fluids of patients with autoimmune diseases and that it is closely associated with neurodegenerative diseases (40). For instance, patients with RA or systemic lupus erythematosus frequently exhibit high levels of TNF-α and TNFRI (41). Elevated TNF-α levels were also reported in patients with multiple sclerosis (42). In addition, the levels of TNF-α along with those of other pro-inflammatory cytokines are increased during the progression of Alzheimer's disease and may be involved in neurological deterioration through nitric oxide pathways (43).

TNF-α is produced by multiple cell types, including immune cells (such as Th1 cells, NK cells, neutrophils, monocytes and macrophages) and non-immune cells (such as neuronal cells) (44). These cells are also key regulators of placentation, the dysregulation of which is a critical contributor to the pathogenesis of RSA. It has been reported that TNF-α may cause unsuccessful pregnancy through several mechanisms, including the following: i) induction of inflammation and disruption of the Th1/Th2 balance; ii) activation of the complement, causing trophoblast cell death; and iii) upregulation of IL-10 levels, leading to fetal loss (Fig. 1). In detail, TNF-α is considered to be responsible for inducing inflammation during placentation and implantation, and the balance between Th1 cytokines (predominantly TNF-α) and Th2 cytokines (predominantly IL-10) is critical for successful pregnancy (7). TNF-α may also stimulate the production of IFN-γ though promoting Treg cell differentiation to Th1 cells, while excessive expression of TNF-α and IFN-γ may activate the complement, thus leading to trophoblast cell death (45,46). In addition, TNF-α has been reported to upregulate programmed cell death-1 levels in monocytes and induce IL-10 production, which underlies part of the immunological mechanisms involved in normal pregnancy (47), whereas in RSA, a tendency for a disequilibrium between TNF-α and IL-10 is present and the TNF-α/IL-10 ratio appears to be elevated in females with implantation failure, recurrent fetal loss and other complications, such as hypertensive syndrome (Fig. 1) (48,49). Initially, studies based on animals revealed that injecting TNF-α causes embryonic death in pregnant mice and TNF-α was indicated to be upregulated in a mouse model with a high incidence of RSA (50). In vivo, TNF-α was observed to be detrimental to mouse and cattle blastocysts due to its toxic effect, and it was also demonstrated to be involved in α-galactosylceramide (ligand expressed by Va14 NK T cells)-induced embryonic death (51).

In pregnant female patients, TNF-α is primarily produced by macrophages in the placenta during the first trimester and a high TNF-α level is a pivotal factor for adverse pregnancy conditions, including RSA, gestational hypertension and gestational diabetes mellitus (6). In addition, TNF-α gene polymorphisms may be associated with increased RSA risk. In Saudi females, the −308 G/A polymorphism in the TNF-α gene promoter was reported to be correlated with the occurrence of unexplained RSA (52). Another study reported that the −863C/A and −238G/A TNF-α polymorphisms, which are at the promoter region, are also risk factors for RSA (53). One study attempted to recreate the increased maternal TNF-α level by exogenously applying TNF-α to the culture media of first-trimester villous placental explants. The inflammatory antibody arrays and ELISA suggested that granulocyte/macrophage-colony-stimulating factor (CSF), C-C motif chemokine ligand 5 and IL-10 were upregulated, whereas IL-4 and macrophage CSF levels were decreased in the presence of TNF-α (6). Therefore, targeting TNF-α may be a key strategy in immunological treatments for RSA.

TNF inhibitors suppress TNF-α binding to TNFRI and TNFRII, thus inhibiting the immune response, dendritic cells and the differentiation of Tregs, effects that are considered to be anti-abortive, and are thus applied in RSA (54–59). For instance, TNF inhibitors may suppress the expression of transcription factors, proteases or protein kinases (including NF-κB, caspases and MAPK) and inhibit the release of pro-inflammatory cytokines, chemokines and adhesion molecules (including IL-1, IL-6, IL-8, MMPs and intercellular adhesion molecule 1) through blocking TNF-α binding to TNFRI and TNFRII, and further inhibiting the activation of TNF-α that would lead to the inflammatory and immune response (54–56). On the other hand, TNF inhibitors may also restrain the activation of immune cells (including macrophages, T-cells and B-cells), inhibit the differentiation from CD4⁺ T cells to Th1 and Th17 cells and suppress the maturation of dendritic cells, which further lowers the levels of pro-inflammatory cytokines (including IL-1, IL-17 or IFN-α), thus suppressing the inflammatory reaction (57–59). In rats with abnormal maternal inflammation-induced RSA, administration of IL-10 or TNF inhibitor (etanercept) prevented pregnancy loss (13). Another study reported that treatment with etanercept blocked TNF-α activity, reduced fetal loss and restored the junctional zone ratio, placental invasion and thinning of spiral arterial walls in a syngeneic mouse model of placental insufficiency (14). It was also reported that etanercept achieved a 62% reduction of inflammation-associated coagulopathies in a rat model, which may reduce pregnancy-related disorders, including RSA (60).

In addition, since common treatments, including prednisone, heparin, aspirin, cyclosporine A or intravenous immunoglobulin (Ig), have been indicated to not be efficacious in reducing TNF-α and the cytotoxicity of NK cells in a number of patients with RSA, TNF inhibitors, which function in the innate immune system, are increasingly being investigated for their efficacy in reducing RSA. As shown in Table I, A randomized controlled trial enrolling patients with RSA with innate immune disorders reported that etanercept significantly reduced TNF-α and NK cell activity at weeks 4-10 of gestation and female patients treated with etanercept had higher live birth rates compared with those treated with placebo (9). Among females with RSA who are treated with anticoagulants, the addition of TNF inhibitors (etanercept or adalimumab) combined with intravenous immunoglobulin increased the live birth rate; however, whether the addition of TNF inhibitors alone is able to improve the live birth rate remains to be determined (10). In a study of 30 females with a history of RSA or failure of in vitro fertilization, 4 doses (25 mg) of etanercept twice weekly prior to conception significantly reduced NK cell activity, particularly among females with RSA (8). Other studies comparing adalimumab + intravenous immunoglobulin with intravenous immunoglobulin- or adalimumab-alone (16) and evaluating the efficacy of etanercept-alone or in combination with methotrexate and rofecoxib in patients with RSA are shown in Table I (8–12,16).

For the subsequent analysis of the currently available data on the application of TNF inhibitors for RSA, the use of TNF inhibitors, the treatment regimens, complications and clinical follow-up were compared among the available clinical studies on RSA (8–12,16) and various noteworthy points became apparent: i) Among all of the TNF inhibitors, etanercept was the most frequently used in clinical practice for the treatment of RSA. In detail, 5 out of the 6 studies used etanercept for the treatment of RSA, while only two studies used adalimumab (adalimumab and etanercept were both used in one study) (10)); in addition, the use of etanercept achieved a higher rate of healthy deliveries compared with adalimumab; ii) TNF inhibitor monotherapy was more commonly used than combined therapy, although the results suggested that combined therapy may achieve a higher live birth rate; iii) in terms of complications, only a small number of studies performed to date have evaluated the complications among patients with RSA who received TNF inhibitor treatment and, thus, no relevant data are provided in the present review; and iv) regarding the clinical follow-up, most studies terminated the follow-up at live birth or the occurrence of abortion, with the exception of one study that continued the follow-up until 3 months after the delivery. The most attractive endpoint is the live birth rate, which ranged from 50 to 89.47% in patients with RSA after treatment with TNF inhibitors.

The safety of TNF inhibitors during pregnancy has long been studied, since they are widely applied drugs for rheumatoid arthritis (RA), including patients with RA during pregnancy. Successful ovulation induction, conception and normal delivery are reported among females receiving long-term TNF inhibitors as anti-cytokine treatment for RA (11,12). Regarding pregnant females with inflammatory bowel disease, infliximab and other TNF inhibitors have been reported to be safe, as there was no reported increased incidence of spontaneous abortion or other adverse pregnancy outcomes (61). However, according to the official guidelines of the German Society of Gynecology and Obstetrics, the Austrian Society of Gynecology and Obstetrics and the Swiss Society of Gynecology and Obstetrics, the use of TNF inhibitors should be reserved for specific conditions, such as Crohn's disease and other autoimmune diseases, due to possible side effects such as skin irritation and drug-induced lupus (15).

As one of the initially approved TNF inhibitors, etanercept, a fusion protein comprised of the extracellular domain of human TNFRII receptor and the Fc region of human IgG1, was approved for RA treatment in the US and Europe in 1998 (62,63). In the same year, infliximab, a chimeric monoclonal antibody that contains a murine antigen-binding region joined to the human IgG1 constant region, was approved for the treatment of Crohn's disease in the US and soon after (in 1999) in Europe, launching the new era of TNF inhibitors in bowel diseases (64). Both etanercept and infliximab were later applied and approved for other inflammatory and autoimmune diseases, such as ankylosing spondylitis and ulcerative colitis. Etanercept is administered as a weekly subcutaneous injection, while infliximab is delivered intravenously in-hospital (63,64). Following the success of the aforementioned treatments, the fully human antibodies adalimumab and golimumab were produced by immunizing genetically engineered mice with human TNF-α, which solved the problem of the immunogenic nature of infliximab (a chimeric monoclonal antibody), as well as its relatively severe adverse effects such as the production of anti-antibody or the occurrence of allergy (62,63,65–71). The release of TNF inhibitors to the market has revolutionized the treatment of RA, ankylosing spondylitis, psoriasis, Chron's disease and ulcerative colitis, particularly RA and ankylosing spondylitis, for which TNF inhibitors have become a standard treatment, and they are generally well-tolerated, with long-term efficacy (72). In addition, the open-label extensions of clinical trials of TNF inhibitors and observations post-market release also support their efficacy and safety.

TNF inhibitors are among the best-selling drugs worldwide (73). However, the cost of these biologicals is high, with an annual cost per patient of >20,000 USD. Therefore, in addition to the TNF inhibitors already launched, several other biologicals are available, including biosimilars and novel drugs under investigation. For instance, a humanized anti-TNF monoclonal antibody, SSS-07, is currently under investigation in a clinical trial for RA in China [National Clinical Trial (NCT) no. NCT02460393] (74). In addition, the approved biosimilars of infliximab include CT-P13 (66) and SB2 (67); the approved biosimilars of etanercept include GP2015 (68) and SB4 (62); and the approved biosimilars of adalimumab include ABP501, BI695501, GP2017 and SB5 (69). So far, >20 newly developed anti-TNF biologicals were in the pipeline, as well as new indications of the already approved drugs (73).

With research efforts focused on drug development, the dosing of TNF inhibitors may become more practicable and cost-effective in clinical practice and their application may be expanded to immunomodulatory conditions other than the commonly approved indications, e.g., the immune disorder underlying RSA.

There are currently numerous studies on the application of TNF inhibitors for the treatment of autoimmune and inflammatory diseases. Considering the contribution of immunological disorders to the mechanism underlying RSA, the application of TNF inhibitors in patients with RSA aims to target the immune aspect of this condition. In pre-clinical studies, TNF inhibitors have been indicated to successfully reduce NK cell numbers and Th1-related cytokine levels, which are risk factors for RSA, and clinical attempts have already achieved an improvement in the live birth rate. However, therapies for RSA using TNF inhibitors still lack sufficient supportive data. In addition, the development of TNF inhibitors in the pharmaceutical industry is still growing, which may be promising for the future clinical application of TNF inhibitors in RSA.