Background
Recurrent loss of pregnancy is an important reproductive health issue, affecting 2–5% of couples [
1]. Almost half of the cases remain unexplained and are treated empirically, using progesterone supplementation, anticoagulation, and/or immunomodulatory approaches [
1]. Immunological causes, such as immunological rejection or development of an intrauterine micromilieu that is harmful for fetus and pregnancy, are suspected reasons for idiopathic recurrent miscarriage (iRM) [
2,
3]. In previous studies, we were able to show highly elevated uterine NK cells in iRM compared to non-iRM patients [
4]. Furthermore, compared with female healthy controls (HC), iRM patients showed higher counts of circulating activated CD4+ and CD8+ T cells [
5] and cytotoxic NK cells [
6] and lower levels of circulating presumably immunoregulatory IL10 + CD56bright NK cells [
6]. In contrast, renal transplant recipients with functioning grafts late post-transplant exhibited lower cytotoxic and higher IL10 + CD56bright NK cell counts than HC [
6]. Moreover, in transplant recipients with good long-term graft outcome, high cytotoxic NK cells were associated with impaired graft outcome whereas high immunoregulatory IL10 + CD56bright NK cells were associated with good long-term graft function. We speculate that these findings on cytotoxic and immunoregulatory IL10 + CD56bright NK cells might have relevance for the fetal allograft in pregnant women. Because women with idiopathic RM showed high cytotoxic and low immunoregulatory IL10 + CD56bright NK cells in the circulation, we concluded that the immune system of iRM patients is dysbalanced and that this might favor immunologically-induced abortion. Mor et al. postulated that the first trimester of pregnancy is associated with inflammation, which is required for blastocyst implantation [
3]. The second trimester is characterized by an anti-inflammatory and T helper 2 (TH2)-type immune microenvironment that is necessary for fetal growth, and in the third trimester there is a switch to an inflammatory and TH1-type immune state, which is necessary for labor and delivery [
3]. We hypothesize that iRM patients exhibit a continued inflammatory immune response during the second trimester that cannot be counter-regulated by increasing inhibitory immune mechanisms, which eventually results in an unfavorable intrauterine micromilieu that is harmful for the fetus and results in the loss of pregnancy.
We compared iRM patients with female HC as well as a series of HC including males to identify abnormally increased or decreased cell subsets and cytokine levels in-vivo and in-vitro. Additionally, iRM patients were compared with dialysis patients and transplant recipients late post-transplant. All transplant recipients received low dose immunosuppression and were free of infection or rejection. As shown in our previous publication, this group consists of patients with down-regulated cytotoxic effector mechanisms and up-regulated counter-regulating immunoregulatory mechanisms forming an immune system in balance [
7]. While immunosuppressive drugs inhibit the formation of cytokines and cytokine receptors, cells that already express certain cytokines and cytokine receptors act similar in patients with or without immunosuppressive drug treatment. We hypothesize that the comparison of iRM patients with kidney recipients late post-transplant represents a comparison of patients with an unbalanced to patients with a balanced immune system. Immune mechanisms that are associated with good or poor graft outcome in transplant recipients might also be associated with successful or unsuccessful pregnancy of the semi-allogeneic fetus.
In the present study, we looked for indications for a dysregulation of IL4, IL10 and TGFß induced counter-regulating inhibitory immune mechanisms. We investigated IL4+, IL10+, TGFß+ and IFNy+ NK, NKT and T cells in the blood of HC, iRM and transplant patients and compared intracellular cytokine production with cytokine levels measured in plasma. In addition, we stimulated NK cells in PBMC cultures using the tumor cell line K562 as stimulator and measured cytokine expression in NK cells before and after stimulation and in the supernatants of the in-vitro stimulated cells. We speculated that cells with T cell receptor interact with HLA on allogeneic cells whereas NK cells without T cell receptor might react with cells missing or expressing particular HLA class I antigens (inhibiting or activating KIRs) or cells expressing stress-induced MICA and MICB (ligand NKG2D) and/or HLA-E (ligand NKG2A) on the cell surface [
8].
Discussion
In a previous study we observed that, compared with female healthy controls, iRM patients possessed abnormally high cytotoxic and abnormally low IL10 + CD56bright NK cells in the peripheral blood, whereas renal transplant recipients on low-dose immunosuppression late post-transplant exhibited abnormally low cytotoxic and abnormally high IL10 + CD56bright NK cell counts [
6]. Moreover, iRM patients showed higher absolute numbers of NK cells with predominantly inhibitory killer-cell immunoglobulin-like receptors (KIR) and CD94/NKG2 receptors than female HC [
8]. They exhibited significantly increased levels of NK cells with inhibitory CD158a, CD158b, CD158e, NKG2A and stimulatory NKG2D receptors, whereas female ESRD patients had normal and female transplant recipients lower or normal numbers of these cells compared to female HC [
8]. Interestingly, the phenotype CD158a+/CD158b+/CD158e+/NKG2A+/NKG2D+ was more common in iRM patients than in female HC and female transplant recipients late post-transplant, supporting the hypothesis that iRM patients have an abnormal up-regulation of inhibitory KIR and NKG2A receptors as well as of stimulatory NKG2D determinants [
8]. We concluded that the dysbalance of cytotoxic and potentially immunoregulatory NK cells in iRM patients might contribute to the pathogenesis of iRM and that the balance of low cytotoxic and high IL10 + CD56bright NK cells in renal transplant recipients might be favorable for good long-term graft acceptance. In the present study, we investigated additional NK cell subsets with presumable immunoregulatory function in the blood of iRM and transplant patients as well as HC, and we analyzed whether these cells might affect the cytokine levels in plasma and contribute to a systemic immunosuppressive cytokine pattern.
Our data show that iRM patients have abnormally high absolute cell counts of IL4+ and TGFß1+ NK, NKT and T lymphocytes in the blood that show low IL4R and TGFßR expression. These cells do not need paracrine secreted IL4 and TGFß1 because of their own strong production of these cytokines. Transplant recipients, in contrast, have normal cell counts of IL4+ and TGFß1+ NK, NKT and T lymphocytes and abnormally high IL4 and TGFß1 plasma levels. Although iRM patients show the highest numbers of IL4+ and TGFß1+ cell subsets in the blood, they have the lowest levels of these cytokines in plasma compared to HC and all patient groups. The data suggest that NK, NKT and T lymphocyte counts do not strongly correlate with plasma levels of IL4 and TGFß1 and do not induce a systemic immunosuppressive milieu in the circulation. Presumably, the peripherally increased IL4+ and TGFß+ NK, NKT and T cells in iRM patients originate from the uterus, however, they are unable to establish a micromilieu in the uterus that is protective for the fetus [
9,
10].
The opposite holds true for IL10. In the present study, summarizing IL10 + CD56bright and IL10 + CD56dimCD16+ NK cells as IL10+ NK cells, there was no difference in IL10+ NK cell numbers among iRM patients, female HC and female transplant recipients. In a previous study, we reported on abnormally high IL10 + CD56bright NK cells in transplant recipients and abnormally low IL10 + CD56bright NK cells in iRM patients [
6] and these observations are paralleled by abnormally low IL10 plasma levels in iRM patients and abnormally high IL10 plasma levels in transplant recipients in the present study. The data support the hypothesis that plasma IL10 originates in part from IL10 + CD56bright NK cells and that IL10 + CD56bright NK cells might strongly contribute to systemic immunosuppression in the circulation. Moreover, in the previous study we described a striking association of IL10 + CD56bright NK cell counts in the blood with graft function in renal transplant recipients [
6] suggesting that this NK cell subset might be strongly immunoregulatory. In the same study, CD8+ perforin+granzymeB+ NK cells, which can be thought of as the opponent of IL10 + CD56bright NK cells, were shown to be associated with impaired graft outcome [
6]. The combination of high IL10 + CD56bright NK cell counts and high IL10 plasma levels in transplant recipients with good long-term graft function supports the hypothesis that IL10+ NK cell subsets might have a strong immunoregulatory role in transplant recipients. IL10+ NK cells that suppressed antigen-specific T-cell responses were reported by Deniz et al. [
11] and others as summarized by Vivier et al. [
12]. iRM patients, in contrast, show a deficit of this presumably immunoregulatory subset in the circulation and we assume that this deficit might contribute to the pathogenesis of idiopathic RM. As summarized by Mor et al. [
3], a pro-inflammatory immune response supports implantation and placentation of the blastocyte and initiates pregnancy, followed by an anti-inflammatory stage of fetal growth and a final pro-inflammatory switch for labor and delivery. Our data suggest that patients with idiopathic RM do not switch to an anti-inflammatory stage.
In a previous study, we reported on abnormally high activated CD3 + DR+, CD4 + DR+ and CD8 + DR+ T lymphocytes in the circulation of patients with idiopathic RM [
5]. When these pre-activated T lymphocytes were stimulated in-vitro using phytohaemagglutinin, pokeweed mitogen or anti-CD3 monoclonal antibody, they showed decreased in-vitro proliferation compared to lymphocytes of healthy female controls tested in parallel [
5]. Similar results were obtained in the present study for the production of TGFß1 in NK cells. Separated peripheral blood lymphocytes of iRM patients showed the highest spontaneous TGFß1 production of all study groups as determined in cell culture supernatants. When NK cells were stimulated using tumor cell line K562, TGFß1 levels in supernatants were lower after stimulation than prior to stimulation, suggesting strong spontaneous production of TGFß1 in circulating NK cells that could not be increased by additional stimulation in-vitro and decreased due to consumption by activated cells.
After abortion, patients with idiopathic RM have a combination of abnormally high proportions of circulating pre-activated T lymphocytes, high cytotoxic NK cells as well as high IL4+, TGFß1+ and IFNy+ NK cells and, in addition, low proportions of circulating IL10 + CD56bright NK cells, low plasma levels of IL4, IL10, TGFß1 and TNFα and low proliferation of T lymphocytes stimulated polyclonally in-vitro. These findings suggest the persistence of an inflammatory immune response in patients with idiopathic RM that cannot be efficiently counter-regulated by up-regulated NK, NKT and T cell subsets with an immunoregulatory phenotype. The deficit of proliferation of T cells and TGFß1 production of NK cells after in-vitro stimulation might be a consequence of up-regulated immunosuppressive components of the immune system, increased consumption of cytokines in an autocrine manner by the activated immune system of the patients, or beginning exhaustion of pre-activated immune cells due to permanent activation of these particular cell subsets. These factors might partially explain the discrepancy between increased IL4+ and TGFß+ NK/NKT/T cells and decreased IL4 and TGFß levels in the plasma of iRM patients.
Although stable transplant recipients have normal counts of circulating IL4+ and TGFß+ NK, NKT and T cells, they have abnormally high IL4 and TGFß1 plasma levels suggesting that tissue resident NK, NKT and T cells and/or other cell subsets produced those cytokines. The higher plasma levels of these immunosuppressive cytokines might contribute to immunoregulatory immune mechanisms that allow the commonly practiced stepwise tapering of immunosuppressive drug doses in transplant recipients, eventually resulting in good long-term graft function in the presence of minimal doses of immunosuppressants. In contrast, iRM patients showed normal or slightly decreased IL4 and TGFß1 plasma levels. Yue et al. reported on significantly higher serum TGFß in the first trimester as compared to non-pregnant women whereas serum IL10 was similar in pregnant and non-pregnant women [
13]. The same authors observed increased serum levels of immunosuppressive IL35 in normal early pregnancy and decreased levels in iRM patients [
13]. Burns et al. reported on higher IL1β, IL6, IL8, TNFα, IFNγ, IL10, and IL1 receptor antagonist (IL-1RA) levels during late pregnancy in amniotic fluid than in cord blood or maternal plasma [
14]. Decreased CCL5/RANTES plasma levels observed by us and others [
15] may contribute to iRM pathogenesis because it was shown in-vitro that RANTES suppresses alloantigen specific T-cell responses [
15] and promotes a pro-implantatory microenvironment that influences trophoblast cell survival and modulates the balance of maternal Treg/T effector lymphocytes in favor of maternal tolerance [
16,
17]. Our data are in line with this observation and show low IL10, TGFß1, TNFα, IL1ß, IL1RA, IL5, IL6, IL12, GM-CSF, VEGF and CCL5 in the plasma of iRM patients, in agreement with the observations of Burns et al. [
14]. A limitation of our study might be that a potential systemic increase of immunosuppressive cytokines was undetectable in our assays, possibly because the iRM patients were at least 2 months after the last pregnancy at the time of testing. The only cytokine that was highest in iRM patients compared to all other examined groups was G-CSF. Because G-CSF plays an effective role in pregnancy success [
18] the strongly increased plasma levels may reflect a counter-regulation to overcome the pathomechanisms of iRM. Interestingly, G-CSF therapy is effective in reducing pregnancy loss [
19].
It should be further mentioned that iRM patients showed higher Treg than female transplant recipients (
p = 0.037) and similar Treg numbers as female healthy individuals (
p = 0.933), as also shown previously [
6]. IL10+ T cells were similar in all groups (iRM vs female transplant recipients
p = 0.923 and vs female HC
p = 0.486) [
6].
Down-regulation of the stimulated immune system might be an option for balancing the immune system of iRM patients. In a small clinical trial, 40 patients with increased NK cells were treated with intralipid infusions that should exert an anti-inflammatory effect on the up-regulated immune system [
20]. Our own preliminary data showed that iRM patients with very high pre-pregnancy peripheral NK lymphocyte counts did not benefit from lipid infusions and probably need additional (pre-pregnancy) medication [
20]. In a recently published study, Liu et al. reported results of low-dose immunotherapy in iRM patients and showed lower proportions of TH1 cells and higher proportions of TH2 cells and Treg as well as higher TGFß1 serum levels after than before treatment [
21]. Patients with immunotherapy achieved significantly more frequently pregnancy than untreated patients [
21]. The role of immunotherapy in clinical practice is discussed controversially [
22]. In consideration of live birth and abortion rates, treatment protocols using combinations of corticosteroid, low dose aspirin, unfractionated heparin, G-CSF, low molecular weight heparin and/or intravenous immunoglobulin appear to be effective treatment strategies [
19]. Treatment with immunomodulatory substances should downregulate cytotoxic alloresponses and should induce immune mechanisms that establish a favorable micromilieu for the fetus.
Finally, from the perspective of a transplantation immunologist, which some of the authors are, female long-term transplant recipients and healthy females constitute interesting control groups for comparison with iRM patients. As detailed by Mor et al. [
3], intervals with inflammatory and intervals with anti-inflammatory immune responses alternate during pregnancy. The first pregnancy trimester is shaped by an inflammatory, the second by an anti-inflammatory and the third again by an inflammatory immune response. These alterations are dynamic and there are phases in between showing both types of immune response. A control group of females with normal pregnancy would be rather heterogeneous with respect to the dominating immune response in the periphery depending on the time point of blood donation during pregnancy.
There is another aspect. According to our hypothesis, iRM patients had rejected the allogeneic fetus immunologically more than 3 times according to the definition of iRM. All iRM patients were investigated at least 3 months and maximally 12 months after the last abortion. One would think that the immune system of these female patients should be more comparable to the immune system of a non-pregnant than to that of a pregnant female. After abortion and antigen loss, the immune system of iRM patients should have returned to a resting state. However, as shown in this study and published in a previous manuscript [
6], iRM patients showed an abnormally high cytotoxic NK cell response. We interpret this extremely high cytotoxic response as a long-term immune response against a persisting antigen and speculate that this cytotoxic immune response is directed towards fetal cells that persist in the iRM patients. Physiologically, immune responses stop when the triggering antigen is eliminated. Transplant recipients are facing a similar immunological challenge. They are continuously stimulated by an allograft. In contrast to the iRM patients, they show quiescence parameters of the immune system that are similar to female healthy individuals, as reported in our publications. Patients with good long-term graft function show normal cytotoxic and normal immunoregulatory responses in the circulation, which agrees with the notion that their immune system is in balance. The low maintenance doses of immunosuppressive drugs would not be sufficient to prevent acute graft rejection immediately post-transplant. Obviously, transplant recipients develop post-transplant immune mechanisms that facilitate the reduction of immunosuppressive drugs to a low maintenance level. The low maintenance dose of immunosuppressants is able to prevent rejection triggered by unspecific immune stimulation, induced for example by infections. Infections are able to trigger rejections by IFN increases and up-regulation of donor MHC antigens on cells of the transplant.
Female long-term transplant recipients and healthy females constitute interesting control groups for comparison with iRM patients. These control groups did not show the strong cytotoxic and counter-regulating immunoregulatory responses found in iRM patients, which supports the hypothesis that iRM patients, despite evidence for ongoing immunoregulatory activity, are not able to establish a low level balance of immunostimulatory (IFNy+ cells, plasma IFNy, TNFα, G-CSF, etc) and immunosuppressive (IL4+ and TGFß+ cells, plasma IL4, TGFß, IL10, etc.) mechanisms. We speculate that the missing balance prevents successful pregnancy and that immunosuppressive therapy might be an option to facilitate such a balance. Our conclusion is based on the observations obtained in female healthy controls and female long-term transplant recipients with good graft function. We feel that the observations in these individuals offer an interesting view on iRM patients that would not be obtained when iRM patients would be compared with normal pregnant women who are known to show heterogeneous immune reactions that depend on the duration of pregnancy, as described by Mor et al. [
3].