During the first trimester of pregnancy, dNK cells accumulate and become the dominant leukocyte population, then decline and only 40–50% dNK cells remain by late gestation [15, 16]. dNK cells can secrete interferon γ (IFN-γ) to activate angiogenesis-related gene expression, promote helical artery remodeling, and also stimulate vascular growth directly through the production of vascular endothelial growth factor (VEGF) and placental growth factor in decidua tissue. Besides, dNK cells can also exhibit low cytotoxicity relative to peripheral NK cells [2]. All of the above suggest that dNK cells are essential for the maintenance of early pregnancy.

In the trophoblast-independent stage of uterine artery remodeling, dNK cells surround the spiral artery and synthesize a series of soluble factors to induce extravillous trophoblast invasion and cause vascular structural instability, including interleukin-8 (IL-8), IFN-γ induced protein-10 (IP-10, also known as C-X-C motif chemokine ligand 10), granulocyte-macrophage colony-stimulating factor, PLGF, IFN-γ, tumor necrosis factor (TNF)-α, VEGF, angiopoietin-1/2 and matrix metalloproteinases-2/9 [17]. Among them, IFN-γ regulates trophoblast invasion and maintains the integrity of the decidua [18]. IFN-γ produced by dNK cells induces upregulation of long non-coding RNA maternally expressed gene 3 expressions in vascular smooth muscle cell (VSMC), and both the treatment of dNK cells or IFN-γ and the overexpression of long non-coding RNA maternally expressed gene 3 can inhibit VSMC proliferation and stimulate its migration and apoptosis, leading to impaired uterine spiral artery remodeling [19]. Moreover, dNK cells can regulate trophoblast invasion by binding chemokine IP-10 and IL-8 with C-X-C motif chemokine receptor 1 and C-X-C motif chemokine receptor 3 expressed by invasive extravillous trophoblast [20]. The decreased expression of CD276 in the trophoblast of URSA patients impairs the inhibition function of IL-8 and IP-10 secretion by dNK cells and maintains the microenvironment of excessive inflammation [21]. A study comparing dNK cells from age-matched healthy controls and URSA patients found lower levels of CD49a and higher expression of perforin, granzyme B, and IFN-γ in the URSA patients [22]. From the above-mentioned studies, it is clear that dNK cell dysfunction is closely related to uterine spiral artery reconstruction (Figure 2).

According to the production of cytokines, human NK cells are divided into four types, including NK1, NK2, NK3, and NKr1 subgroups. Among them, NK1 generates IFN-γ and TNF-α, NK2 excretes IL-4, IL-5 and IL-13, NK3 cells produce TGF-β, and NKr1 cells secrete IL-10 [23]. In URSA patients, the NK1/NK2 ratio of dNK cells (producing IFN-γ/IL-4 NK cell ratio) increased, and it was independent of whether the karyotype was normal. NKp46 is a receptor on NK cells, which is involved in cytotoxicity and cytokine production [24]. A study showed that in women with low %NKp46⁺ dNK, the ratio of NK1/NK2 was significantly higher than that of women with high %NKp46⁺ dNK, and the threshold of %NKp46⁺ dNK cells was 86.52% [25]. On the whole, the above research suggests that URSA is associated with the interference effect of NK cells on the uterine spiral artery remodeling and the imbalanced proportion of NK cells in the decidua, which may result in the inflammatory state of the micro-environment of the maternal-fetal interface. The expression of NKp46 may help predict this type of URSA, providing a possible detection indicator for its diagnosis.

Among leukocytes at the maternal-fetal interface, CD4⁺T cells can be activated and then differentiate into Th1, Th2, and Th17 cells to produce the corresponding types of cytokines. Th1 cells are differentiated by IL-12-driven, and Th2 cells are differentiated by IL-4-driven, and factors other than differences in cytokine environments may also selectively affect Th1/Th2 development [26]. Th17 is a distinct Th cell population that primarily produces IL-17A, and activated Th17 cells also secrete IL-6, IL-17F, IL-21, IL-22, TNF-α, and granulocyte-macrophage colony-stimulating factor which induce inflammatory responses by promoting the recruitment of inflammatory cells [27]. Tregs are a subset of CD4⁺T cells expressing high levels of CD25 and the transcription forkhead box protein P3 (Foxp3), and Tregs are essential for maintaining immune homeostasis and preventing autoimmunity by suppressing auto-reactive T cells [28]. T cells play a central role in regulating the immune response. In early gestation, the maternal-fetal interface is a Th1 cell environment that promotes blastocyst implantation. In mid-gestation, the maternal-fetal interface converts to a Th2 cell environment, as this period is the main stage of fetal development and an anti-inflammatory immune microenvironment is necessary for fetal growth. Finally, in late gestation, the maternal-fetal interface switches back to a Th1 cell-dominated immune state, which is vital for delivery [29]. The T cell homeostasis at the maternal-fetal interface of URSA is shown in Figure 3.

Macrophages, as antigen-presenting cells, account for 20–30% of leukocytes at the maternal-fetal interface [50]. Macrophages are involved in the complex regulation of the maternal-fetal interface through the regulation of secreted cytokines, phagocytosis, and immune homeostasis. In fact, they not only affect vascular remodeling but also are actively associated with trophoblast invasion and pregnancy maintenance [51]. Macrophages can be classified into classically activated (M1) and alternatively activated (M2) subtypes according to their cytokine production and function [52]. In early pregnancy, uterine macrophage populations with an alternatively activated steady-state phenotype are involved in embryo implantation and placental development as well as in the prevention of placental infection in both mice and humans [53]. During the implantation phase, the M1 type predominates, while later it changes to a mixed population of M1 and M2 types, which takes part in trophoblastic invasion, infection resistance, and tissue vascular remodeling [54]. The mixed M1/M2 population persists until mid-pregnancy, and then decidua macrophages shift to an M2-dominant phenotype after completion of placental development, which promotes maternal immune tolerance to the semi-carcinogenic fetus and protects fetal growth until delivery [55]. As labor approaches, decidua M2-like macrophages become more abundant, and they may have an immunomodulatory role before term pregnancy [56]. Research shows that CD206⁺M2-like MΦs may be essential for embryo implantation through the regulation of endometrial proliferation via Wnt/β-catenin signaling [57]. Consequently, an appropriate proportion of M1/M2 has a vital impact on successful pregnancies. Altered or disturbed M1/M2 balance and related functions during pregnancy may lead to complications and imbalance in the M1/M2 ratio is strongly associated with RSA [58].

A study by Tsao et al. [58] found that M1 macrophages were abundant in the decidua of spontaneous abortion and URSA, while M2 macrophages were significantly more frequent in the endometrium of luteal phase and normal pregnancy, suggesting that M2 polarization is important for successful early human pregnancy. Fetal-maternal tolerance is mediated, at least in part, by the a2 isoform of vacuolar ATPase, an important molecule expressed at the fetal-maternal interface. In the placenta of a mouse model prone to abortion, reduced expression of the a2 isoform of vacuolar ATPase may bias the polarization pattern toward the M1 phenotype and induce inflammation [59]. In placental tissues from URSA patients, peroxisome proliferator-activated receptor γ (PPARγ) expression is significantly downregulated. As it has been confirmed that PPARγ is essential as a nuclear receptor for the maturation of alternatively activated M2 macrophages, the reduced level of PPARγ results in a decreased level of M2 differentiation [60]. Besides, reduced PD-1 expression on decidual macrophages accompanied by reduced programmed cell death-ligand 1 (PD-L1) expression on placental villi was observed in URSA permutations, in which PD-1/PD-L1 axis disorders induce M1 differentiation [61]. The M1/M2 balance in URSA patients is shifted towards the M1 type, which secretes a large amount of pro-inflammatory cytokines (e.g., TNF-α, IL-6, and IL-1β) [62]. URSA patients are found to have abnormally high levels of TNF-α, which leads to the downregulation of stathmin-1, and ultimately by regulating E-calmodulin/ β linked protein pathway promotes the development of URSA [63]. Disturbance of the PD-1/PD-L1 axis at the maternal-fetal interface in URSA patients promotes macrophage polarization toward the M1 type, which may further promote inflammatory cytokine secretion and increase the risk of URSA development (Figure 4).

In addition, current studies suggest that the pathogenic process of URSA may be associated with changes in microRNA (miRNA) expression profiles in decidua and villi [64, 65]. Zhu et al. [66] showed that downregulated miR103 enhances M1 polarization via the signal transduction and activator of transcription 1 (STAT1)/interferon regulatory factor 1 (IRF1) pathway. Also, Ding et al. [67] found that miR-146a-5p and miR-146b-5p of extracellular vesicles secreted by M1 macrophages inhibited epithelial-mesenchymal transition in trophoblast cells by directly suppressing TNF receptor-associated factor 6 expression after transcription, thereby increasing the risk of URSA. The above studies reveal that related miRNA targets may be effective diagnostic markers for URSA and can possibly serve as novel therapeutic targets by regulating M1/M2 homeostasis.

DCs are important players in linking innate and adaptive immunity [68]. DCs have a dual role as they can differentiate into potent antigen-presenting cells to further activate effector T cells and in an immature state they can enhance immune tolerance by inducing the production of Tregs [69]. In vivo, most DCs are in an immature state [immature DCs (imDCs)]. ImDCs process antigens through DEC⁺205 receptors and promote CD8⁺T cell proliferation while inhibiting cytotoxic IFN γ production, which means that imDCs mainly maintain their immune tolerance. In contrast, mature DCs (mDCs) can upregulate class 1 and class 2 major histocompatibility complex (MHC) molecules with binding peptides on their surface. Through presenting binding peptides to T cell receptors, mDCs ultimately induce specific T cell activation and secretion of IL-12, leading to Th1 differentiation of CD4⁺T cells which may promote URSA formation [70]. Moreover, it was found that mDCs increased and imDCs decreased in URSA patients, suggesting that the immature state of DCs facilitates pregnancy while mDCs may play an important role in the pathogenesis of URSA [71]. Lai et al. [72] showed that DCs in RSA patients and RSA mice were characterized by less expression of CD274, significantly higher serum levels of IL-12 and IL-6 than in the normal pregnancy group, and decreased expression of TGF-β and IL-10, thus allowing T cells to balance Th1/Th2/Th17/Treg subsets to Th1/Th17. Changes in the expression of these functional molecules and cytokine secretion by DCs may disrupt the balance of maternal and fetal immune tolerance status and lead to URSA, so the mechanism of DCs’ role in URSA patients is closely related to T cell differentiation.

BAs, also known as anti-husband lymphocyte antibodies, are a class of immunoglobulin G (IgG) antibodies produced mainly against the embryonic HLA Il and trophoblast lymphocyte cross-antigens. One view is that early URSA may be related to the patient’s low response to embryonic and semi-allogeneic antigens and inability to produce appropriate BAs, which ultimately leads to embryonic rejection [73]. BAs mainly include anti-HLA-D/DR antibodies, MLR-Bf, and anti-paternal cytotoxic antibodies, etc. [74, 75].

Pandey et al. [76] found that MLR-Bf is essentially IgG-3. It is present in pregnant women and women with URSA who successfully became pregnant after immunotherapy treatment. A clinical study by Nonaka et al. [77] observed that after paternal lymphocyte immunotherapy in URSA patients (BA negative), the pregnancy success rate in BA negative patients was lower (30%) than that in BA positive patients (75%). Khonina et al. [78] also proved that MLR-Bf-negative URSA women had increased proliferative cell responses to paternal alloantigens and enhanced production of soluble inhibitory activity factor (MLR-Bf) after they received paternal lymphocytes immunotherapy, which contributes to improved pregnancy outcomes. Most women with alloimmune causes of URSA have increased sharing of HLA and significantly lower titers of MLR-Bf, which may be associated with preventing the mother from making anti-paternal cytotoxic antibodies, while paternal lymphocyte immunotherapy can increase blocking antibody expression [10]. In conclusion, the above-mentioned studies suggest an increased risk of URSA in women with negative closed antibodies, and the conversion of closed antibodies from negative to positive in URSA patients by lymphocyte immunotherapy would decrease the incidence of URSA. Therefore, closed antibody testing could also be used as a criterion for determining URSA.

Cytokines are cell signaling molecules with a vital role as immune regulators in pregnancy, regulating gametogenesis, implantation, and fetal development [2]. After normal pregnancy implantation, Th1/Th2 is essential for maintaining pregnancy at the maternal-fetal interface, in which Th1 cells play a pro-inflammatory role by secreting TNF-α, IFN-γ, IL-17, and IL1-β while Th2 cells promote anti-inflammatory effects by producing IL-4, IL-6, IL-10, and IL-1RA [80, 81].

In URSA patients, TNF-α gene polymorphism is studied most widely. TNF-α-308G/A and/or -238G/A are two common sites in research. A meta-analysis by Zhang et al. [82] showed that TNF-α-308G/A and -238G/A polymorphisms were not significantly associated with the risk of RSA in the whole population. However, the studies of Zhao X and Liu C indicated that TNF-α-238G/A is significantly associated with URSA risk in Asian populations [83, 84]. Lee et al. [85] reported similar findings in the Korean population and proposed that TNF-α-1031t>C may also be a risk factor for URSA.

Among all these related genes, the IL-6 gene is the most extensively studied anti-inflammatory gene polymorphism in URSA patients, and IL-6-174G/C is the most common research site. However, all studies have failed to prove the association between this specific polymorphism and RSA [79]. One study investigated the relationship between RSA and the single nucleotide polymorphism of -634C→G in the promoter region of the IL-6 gene in the Japanese population. The results showed that there was a significant difference in the -634C→G genotype frequency (CC and CG/GG) between URSA patients and the control group, and the risk of URSA in G allele carriers was lower than that in wild-type (CC) women [86]. Ma et al. [87] also reported that IL-6-634C/G polymorphism may be a genetic protective factor of URSA in the Chinese population.

Located on chromosome 6, HLA is characterized by a wide range of polymorphisms [88]. HLA coding genes are human MHC, which have become the focuses of reproductive immunity [89]. HLA can be divided into HLA class 1 and HLA class 2. Classical HLA-A, -B, and -C genes have a high degree of polymorphism, while nonclassical HLA-E, -F, and -G genes show decreased polymorphism. Class 2 genes encode five different molecules, among which HLA-DP, -DQ, and -DR are polymorphic, while HLA-DM and HLA-DO exhibit reduced polymorphism and function in regulating other HLA class 2 molecules [90].

At present, most of the studies on this aspect focus on the relationship between HLA-G, HLA-D, and URSA, while few pieces of literature study the relationship between HLA-I (A, B, C) and URSA.

Angiogenesis is an important step in chorionic villus formation. Thrombin-activatable fibrinolysis inhibitor (TAFI) regulates fibrinolysis and inflammation. VEGF, endothelial nitric oxide synthase (eNOS), and thrombin can activate genetic polymorphisms of TAFI, leading to the occurrence of URSA [124].

VEGF-1154G/A (rs1570360), +936C/T (rs3025039), -634G/C (rs2010963), and -583T/C (rs3025020) polymorphisms are found to be associated with URSA susceptibility [125]. Interestingly, a study conducted in a specific population suggested that VEGF-1154G/A polymorphism was irrelevant to the URSA susceptibility of Chinese Han women [126]. However, the GA genotype of VEGF-1154G>A polymorphism was found to have a remarkable difference between URSA patients and the control group in the Korean population [127].

The 27 bp variable number tandem repeat (VNTR) polymorphism in intron 4 and Gly298Asp polymorphism in exon 7 of eNOS seem to affect nitric oxide production and are relevant to different pregnancy diseases [128]. There was a strong relationship between eNOS Glu298Asp polymorphism and URSA in East Asians, while in Caucasians and East Asians, no correlation was observed between eNOS intron 4 VNTR polymorphism and unexplained RPL [129]. In contrast, the 4 VNTR polymorphism in Northern Indian women was found to be associated with URSA [130].

Pruner et al. [131] studied the TAFI+1040C/T allele and found that compared with the T allele, women who carried the C allele had a 9% lower risk of URSA, though the results were not statistically significant. In addition, two studies pointed out that TAFI+1040C/T could not be used as a predictor of URSA in Egyptians [124, 132]. However, the data of Fang et al. [133] implied that the TAFI+1040C allele has a protective effect against URSA and may be related to its genetic susceptibility.

As the most important steroid hormone in the female reproductive system, estrogen plays a variety of roles in the uterus to ensure successful pregnancy and conception, including cervical maturation, embryo implantation, and placental formation. The physiological function of estrogen is mediated by two estrogen receptors, namely estrogen receptor 1 (ESR1) and ESR2, which are expressed in various tissues of the female reproductive system. Studies have shown that estrogen receptors may induce immune tolerance at the maternal-infant interface, indicating that ESR is a key factor in the pathogenesis of URSA [134, 135].

Among research on ESR1, Yin et al. [136] reported that ESR1 rs9340799 (-351A/G) polymorphism may have diverse roles in RSA in different ethnic populations. For Asians, ESR1 rs9340799(-351A/G) polymorphism may reduce the risk of RSA, while for non-Asians it may increase the risk. Tang et al. [135] found that ESR1 gene polymorphisms rs9340799, rs2234693, and rs3798759 were not related to URSA in Chinese Han and Hui populations. Another study conducted on Tunisian and Egyptian women showed there was a significant association between the ESR1 variant and the increased risk of RSA [137, 138]. However, no significant association was found between ESR2 gene polymorphism and URSA [138–140].

It is well known that genetic research is complex and needs further confirmation in different environments. Also, ethnic differences may interfere with genotype frequency, thus affecting the research results. The above results show that the relationship between gene polymorphism in different media and the risk of URSA is closely related to different ethnic groups and different environments. It is worth noting that most of the above studies are limited by regions and races, and the sample size is also insufficient. In this regard, a large number of sample studies need to be conducted in different regions and populations to further explore the relationship between gene polymorphism in different media and URSA.