REVIEW ARTICLE: Clinical Implication of Natural Killer Cells and Reproduction


  • Joanne Kwak-Kim,

    1. Department of Obstetrics and Gynecology, Rosalind Franklin University of Medicine and Science/The Chicago Medical School, North Chicago, IL, USA
    2. Department of Microbiology and Immunology, Rosalind Franklin University of Medicine and Science/The Chicago Medical School, North Chicago, IL, USA
    Search for more papers by this author
  • Alice Gilman-Sachs

    1. Department of Microbiology and Immunology, Rosalind Franklin University of Medicine and Science/The Chicago Medical School, North Chicago, IL, USA
    Search for more papers by this author

Joanne Kwak-Kim, Department of Obstetrics and Gynecology, Department of Microbiology and Immunology, Rosalind Franklin University of Medicine and Science/The Chicago Medical School, 3333 Green Bay Road, North Chicago, IL 60064, USA.


The regulation of natural killer (NK) cells in the peripheral blood and endometrial layers has been associated with reproductive immunopathology such as recurrent spontaneous abortions (RSA), infertility of implantation failures, or pre-eclampsia. The placenta has a complex anatomical structure and different subsets of NK cells with various functional roles can directly interact with trophoblasts. NK cell subpopulations and their functions, putative roles of NK cells in peripheral blood and endometrium are reviewed in relation to RSA and infertility. An increase in NK cell numbers and /or activity in pre- or post-conceptional period in women with RSA or infertility with multiple implantation failures are a significant clinical concern. In addition, immuno-phenotypic characteristics of NK cells in these women support the changes for their increased activity status. Further studies are needed to explore underlying mechanism of NK cells in RSA, infertility, and other reproductive immunopathologies. Possible neurological and hormonal control of NK cells and NK cell interaction with various leukocyte populations need further investigation in women with reproductive failures.


Natural killer (NK) cells constitute the predominant leukocyte population in uterine mucosa and considerable effort has been made to investigate NK cells and their roles at the time of implantation and in pregnancy. Dysregulation of NK cells has been associated with reproductive pathologies, such as recurrent spontaneous abortions (RSA),1 infertility,2,3 and pre-eclampsia.4 In women with these conditions, NK cell parameters, either in absolute numbers or in proportion (%), subsets, functional activity such as cytotoxicity or secretory cytokine profile, receptor or gene expression, have been extensively investigated in peripheral, endometrial or decidual NK cells.

Natural killer cells are innate immune effectors. There are two distinct subsets of human NK cells identified by cell surface density of CD56.5 A majority (approximately 90%) of human NK cells are CD56dim and express high levels of FcγRIII (CD16) as well as perforin. CD56dim cells produce low levels of NK-derived cytokines but they are potent mediators of antibody dependent cellular cytotoxicity (ADCC), LAK activity or natural cytotoxicity. These cells are effective killers and more cytotoxic than other NK subsets. Even resting CD56dim cells are more cytotoxic than CD56bright cells.6,7 A majority of circulating maternal NK cells in blood are CD56dim cells and they have direct contact with chorionic villi at the intervillous space. Therefore, how the developing fetus evades NK cytotoxicity from these cells may determine the reproductive outcome.

In contrast, a minority (approximately 10%) of NK cells are peripheral blood CD56bright and CD16dim/neg NK cells. These cells are the primary source of NK cell-derived cytokines and thought to be an important inflammatory or regulatory subset.8,9 CD56bright cells produce in excess of 25-fold more IL-10 than CD56dim cells. CD56bright and CD56dim cells are functionally and phenotypically different subsets of NK cells. Most endometrial NK cells are CD56bright and CD16dim/neg. Therefore, endometrial NK cells have mainly inflammatory or regulatory roles. These cells are potent secretors of an array of angiogenic factors and induce vascular growth in the decidua.10 Therefore, their major function has been speculated to assist fetal development.11

Here, we review the clinical implications and putative role of human NK cells in RSA and infertility. NK cells in peripheral blood and uterine mucosa prior to and during pregnancy are reviewed in relation to reproductive outcome. Further, we discuss receptor expression and the genetic control of NK cells in these women and define some questions that need to be addressed in future research.

NK cell and HLA-G

Both syncytiotrophoblast and villous cytotrophoblast cells do not express classical class I human leukocyte (HLA-A or -B) or class II (HLA-DP, -DQ or -DR) alloantigens.12 However, the selective expression of HLA-C, HLA-E or HLA-G was reported on extravillous cytotrophoblast.4,13 Lack of HLA antigen expression on villous cytotrophoblast and syncytiotrophoblast protects these cells from cytotoxic T cells [which could conceivably recognize peptide major histocompatibility complex (MHC) if they were expressed on these cells]. However, the lack of expression of classical MHC class I alloantigens may render villous trophoblasts vulnerable to NK-cell (CD56dim/16+ cytotoxic) mediated lysis. Therefore, the notion that HLA-G may protect cytotrophoblast from NK lysis does not explain evasion mechanism of villous cytotrophoblast and syncytiotrophoblasts in intervillous space, which are directly in contact with peripheral blood NK cells. Consequently, regulation of peripheral blood NK cell cytotoxicity during a pregnancy seems crucial for the success of pregnancy.

The HLA-G was thought to act as a cell surface class I MHC molecule and protects cytotrophoblasts from maternal MHC non-restricted NK cell attack. The expression of HLA-G by HLA-null cell transfectants has been shown to decrease their sensitivity to NK cell-mediated cytolysis.14 The cytolytic activities of uterine blood mononuclear cells as effectors against cytotrophoblasts were inhibited in both semiallogenic and allogenic combinations. This inhibition was reversed when cytotrophoblast cells were incubated with the pan-class I monoclonal antibody (mAb) (W6/32).15 Despite these evidences, a notion that non-classical class I HLA-G expression may provide the protective effect has been challenged.16,17 Trophoblast cell lines, both JAR (HLA class I-negative) and JEG-3 (HLA-G and HLA-Cw4-positive) cells were reported to be resistant to peripheral blood NK cell-mediated lysis and equally lysed by IL-2-stimulated peripheral blood NK cells. In addition, down regulation or masking HLA molecules on HLA-G positive JEG-3 did not affect the resistance to lysis.16 Polyclonal decidual NK cells were unable to kill freshly isolated extravillous trophoblast cells even when MHC class I molecules were blocked by mAb or down-regulated by acid treatment.18 Fetal development was not obviously affected in a human with a homozygous HLA-G-null mutation.19 Therefore, during pregnancy, an HLA class I-independent mechanism of resistance to NK lysis analogous to that observed with choriocarcinoma cell line might be involved at the interface between the syncytiotrophoblast and peripheral blood lymphocytes, in addition to a possible HLA-G-dependent mechanism that could take place at the interface between extravillous trophoblast and decidual lymphocytes.16 It is noteworthy that some of the previous experiments utilized uterine blood mononuclear cells or decidual lymphocytes for the investigation of interaction between extravillous cytotrophoblasts and uterine NK cells instead of using isolated uterine NK cells.15 Therefore, these studies might overlook possible intercellular reaction among various leukocyte populations in decidua. In addition, the studies of isolated uterine NK cells often utilized extensive manipulation of the tissue samples, such as enzymatic lysis, which may cause changes in cellular characteristics. Hence, possibilities of uterine NK cytotoxicity against extravillous cytotrophoblast and other mechanisms for inhibition of uterine NK cell cytotoxicity exist and this possibility should be carefully explored.

The MHC class I chain related proteins A (MIC-A) is a non-classical MHC class I homolog genetically linked to the classical MHC. It is a non-classical ligand for NKG2D, a C-type lectin-like activating immunoreceptor.20,21 Recently, a novel mechanism has been reported that soluble MIC-A and B down-regulate the activating NK cell receptor NKG2D on peripheral blood mononuclear cells (PBMC), with the consequent inhibition of the NKG2D-dependent cytotoxic response. The expression of MIC is restricted to the syncytiotrophoblast both on the apical and basal cell membrane and in cytoplasmic vacuoles. Soluble MIC molecules were present at elevated levels in maternal blood throughout normal pregnancy and released by placental explants in vitro.22 Similarly, sHLA-G1 has been reported to inhibit NK cell cytotoxicity by controlling NK cell signal transducer STAT3, which has downstream effects including perforin expression, proliferation and cytotoxicity, but not IL-2R alpha expression.23

Another mechanism, which might induce NK-mediated killing, is ADCC. NK-resistant choriocarcinoma cells became sensitive to lysis by unstimulated PBMC in the presence of phytohemagglutin (PHA), to ADCC in presence of anti-Tja antibodies, and to mAb redirected killing using anti-TR antibodies, anti-CD16 and anti-CD244/2B4.24 Therefore, certain conditions to promote cross linkage and activation of NK cells, or autoimmune conditions, which promote trophoblast surface antibody binding might cause destruction of trophoblasts by peripheral blood NK (pNK) cell lysis. Clinical evidence, which supports ADCC as a mechanism of trophoblast destruction, can be seen in women with fetomaternal blood group P incompatibility (i.e. lack blood group P) and anti-Tja antibodies. Anti-Tja serum contains a mixture of anti-P, anti-Pk, and anti-P1 antibodies, principally of the IgG class.25 Women with these antibodies have trophoblast cells, which express paternal P and Pk antigens.25 Women with anti-P1PPk antibodies were reported to have a very high incidence of spontaneous abortions and elimination of anti-P1PPk antibodies resulted in normal childbirth.26 Antiphospholipid antibody (APA) recognizes both anionic phospholipids and/or beta2GPI on trophoblast cell structures, the highest binding being found when cells displayed the greatest amount of syncytium formation.27 Therefore, autoimmune conditions such as the APA syndrome or alloimmune conditions such as fetomaternal P incompatibility may induce ADCC against trophoblasts and consequently result in trophoblast destruction. Previously we reported elevated NK cell levels (%) in RSA women with APA1 and recently a similar study has been reported.28 This suggests that in addition to complement-mediated lysis, ADCC might be a potential trophoblast-destruction mechanism leading to abortion in group p women24 and women with autoimmune conditions.

In contrast to cytotoxicity or lysis, it has been suggested that decidual NK cells may have a beneficial effect on reproductive outcome.10 Uterine NK cells and trophoblasts may have a ‘peaceful’ and constructive liaison rather than a cytotoxic and confrontational relationship.10 However, decidual NK cells are cytotoxic and evasion mechanism of extravillous cytotrophoblast from NK cytotoxicity needs to be further investigated as decidual NK cells are unable to kill trophoblast even in the presence of mAb to MHC class I molecules and NK cell receptors.18

Peripheral blood NK cell levels

The systemic regulation of NK cells is essential for achievement of a successful reproductive outcome. Peripheral blood NK cell levels were reported not to be significantly changed post-conception as compared to pre-conception level, both in women with RSA and in normal controls.2,29 When NK cell subpopulations are studied, CD56+/16+ NK cell levels were significantly down regulated in pregnant women as compared to non-pregnant women.2 In our study of normal pregnant women without any history of RSA or infertility and confirmed delivery of index pregnancy, CD56+ NK cell levels were not significantly changed during the first, second and third trimester (Fig. 1).

Figure 1.

 Peripheral blood CD56+ NK cell levels (%) in normal pregnant women without a history of infertility or spontaneous abortion, who delivered a live born infant for index pregnancy.

An abnormal increase in peripheral blood NK cell parameters [either in NK cells absolute value or in proportion (%) prior to conception or during early pregnancy] is associated with RSA (Table I) and infertility with multiple implantation failures.1,29–32 37.3% of women with RSA manifest mild to moderate increase (12–18%) and 14.7% had a marked increase (>18%) in peripheral blood CD56+ NK cells.1 NK cell numbers of <12% were strongly associated with a subsequent pregnancy carried to term.2,33 The numbers and proportions of NK cells are significantly higher in patients with RSA with anti-phospholipid antibody syndrome (APS) than in APS without RSA.1,28 Increased numbers of NK cells correlate with reduced gestational age at abortion in RSA patients with APS.28 Most importantly, down regulation of NK cells in women with RSA is associated with a favorable pregnancy outcome.29,34

Table I.   Peripheral Blood NK Cells and NK Cytotoxicity Studies in Women with Recurrent Spontaneous Abortions
StudyStudy populationNK CTXNK cell levels
Aoki et al.4068 women with RSA
47 controls
Pre-conceptional elevation is predictive of the risk of abortion (RR 3.5)
Kwak et al.1116 women with RSA (>3): 35 pregnant and 81 non-pregnant
Controls are 17 non-pregnant and 22 pregnant women
Increase in CD56+, CD56+/16+
Coulam et al.3070 women with infertility
44 women with RSA (>2)
Increase in CD56+. Positive predictive value of 78%
Emmer et al.3343 women with RSA (>2)
37 healthy controls
No difference preconception. Increase during early pregnancyNo difference in CD56+, CD56+/16+. Increased CD56+/16+ cells during early pregnancy. Women with <12% CD56 have higher success rate
Morikawa et al.42113 women with RSAElevated NK CTX in women who aborted index pregnancy as compared to women who delivered%CD56 not related with cause and number of RSA
Souza et al.869 women with RSA (>2)No differenceNo difference
Michou et al.3525 women with RSA
11 normal controls
30 women with sporadic abortion
33 women with infertility
Significantly increased CD56+/16 cells but not CD56+
Yamada et al.39113 RSA (>2) with normal karyotypeIncreaseIncrease
Perricone et al.29394 non-pregnant women with RSA
187 pregnant women with RSA
42 non-pregnant healthy controls
36 pregnant healthy women
Increase in %CD56 and absolute number

Emmer et al., reported no differences in numbers of CD56+ and CD56+/16+ cells and NK cytotoxicity between RSA and normal controls prior to conception.33 However, a longitudinal study revealed that compared with controls, in RSA women higher numbers of CD56+/16+ NK cells were present during early pregnancy, paralleled by an increase in cytotoxic NK cell reactivity.33 Others reported significantly increased CD56+/16 cells but not CD56+ in women with RSA as compared to normal controls or those with sporadic abortion or infertility.35

The percentages of CD56+ and CD56+/16+ cells in the peripheral blood on the day of embryo transfer were significantly higher in women with IVF failure than in women with implantation.31 A significant increase in CD69 expression on CD56+ NK cells was demonstrated in women with RSA and infertility as compared with that of normal controls.36 Later, simple enumerations of peripheral blood NK cells (including total CD56+, CD56dim and CD56bright NK cells with IVF treatment outcome and pregnancy outcome) were reported not to be different and an increase in the absolute count of activated NK cells (CD56dim/16+/69+) in the peripheral blood is associated with a reduced rate of embryo implantation in IVF treatment.37,38 High levels of peripheral blood NK cells were found to predict biochemical pregnancy and spontaneous abortion with normal karyotype in the next pregnancy.30,39 Elevated CD56+ cells in pregnant women (who are not donor egg recipients, or who have not undergone an ICSI procedure, not receiving IVIg treatment, and not carrying multiple gestation) predict loss of a karyotypically normal conceptus with a specificity of 87% and positive predictive value of 78%. While the specificity value of this test is high in both infertile and RSA populations, the sensitivity is 86% in RSA and only 54% in infertility women suggesting that enumeration of peripheral blood CD56+ NK cells does not identify all losses among infertility women.30 Yamada et al., also reported peripheral blood CD56% NK cells as predictors of biochemical pregnancy and spontaneous abortion with normal chromosome karyotype (CD56+ >16.4%).39 Relative risk of biochemical pregnancy losses or spontaneous abortion with normal karyotype was reported to be 4.9.39 In women with infertility, specificity for %CD56 NK cells tends to be higher than sensitivity, and negative predictive value was higher than positive predictive value.30,39 This may reflect heterogeneity in infertility population and multifactorial etiologies for infertility.

Thus, %CD56 NK cell levels have clinical value for monitoring of pregnant women with a history of RSA or infertility, and changes in NK cell levels may predict pathological obstetrical outcome. A study is needed to improve sensitivity and positive predictive value of the test further. Patient selection criteria, evaluation of NK cell subsets, new cell markers, timing of the study and analysis technique should be investigated further.

NK cell cytotoxicity

Many studies following NK cytotoxicity and pregnancy outcome have been published (Table I). High pre-conceptional peripheral NK cytotoxicity has been found to be associated with abortions in the next pregnancy in women with RSA.40,41 NK cell activity in non-pregnant women who later experienced subsequent abortion with normal chromosomes, but not abnormal chromosome, was significantly higher than that in women with subsequent live birth.39,42 Primary aborters had the highest proportion and concentration of NK cells and the highest NK activity using the standard assay. Secondary aborters had an intermediate level of these NK cell indices, whereas the control patients had the lowest levels.43

In women with subsequent live birth, NK cell activity at early pregnancy (4–5 gestational weeks) significantly decreased later (at 6–7 and at 8–9 gestational weeks). NK cell activity in women with subsequent abortion with normal chromosomes at 6–7 gestational weeks was significantly higher than that in women with subsequent live birth, while NK cell activity at 6–7 gestational weeks in women with subsequent abortion with abnormal chromosomes was the same as the level in women with subsequent live birth.44 High values of pre-conceptional NK cell activity (>46%; relative risk 3.6, 95% CI 1.6–8.0) and percentage (>16.4%; 4.9, 1.7–13.8) were found to predict biochemical pregnancy and spontaneous abortion with normal karyotype in the next pregnancy.39 Low levels of NK cell activity in the peripheral blood was reported to be present in most successful cases in women with immunotherapy for RSA.29,45

The same trend was also noticed in women with infertility.46 It has been reported that elevated peripheral NK activity during a follicular phase in patients with unexplained infertility is a risk factor for attaining pregnancy success.47 In infertile women, none of the successfully pregnant women of that cycle had elevated levels of NK cytotoxicity whereas 50% of those experiencing a chemical pregnancy loss and those not becoming pregnant had elevated levels of NK cytotoxicity.3,47 Elevated NK cell cytotoxicity was reported to be associated with increased IL-2 levels, but not with IL-10 or TGFβ1, and elevation of both NK cell cytotoxicity and IL-2 was reported to be a risk factor for RSA.41 a2V-ATPase has been reported to be expressed on subsets of PBMC and regulates the extracellular environment, which facilitates NK cytotoxicity or cytokine secretion.48 In women with RSA or implantation failures, expression of NKp46, NKp44, NKp30, and a2V-ATPase on CD56bright NK cells was significantly up-regulated compared with those of CD56dim NK cells.49 This study suggests dysregulation of NK cytotoxicity and cytokine production in women with RSA and infertility. The systemic levels of tumour necrosis factor-α (TNF-α) and interferon-γ (IFN-γ) have no association with implantation or miscarriage rate in women undergoing IVF treatment. However, high levels of TNF-α and IFN-γ are associated with elevated levels of activated NK cells and this may subsequently exert a negative impact on reproduction.50

The flow cytometric assay for NK cytotoxicity is reproducible and precise, and can be successfully used to evaluate patient samples.51 A lack of correlation was reported between peripheral blood NK cell cytotoxicity and percentages or absolute counts of CD56+/16+, CD56+/16 or CD3+/56+ lymphocyte subsets. Therefore, NK cell cytotoxicity and peripheral blood NK cell values measure different aspects of NK cells and do not correlate.51 The analysis of peripheral NK cell characteristics appears a suitable diagnostic tool for RSA and infertility. Immunomodulation aimed at NK cell function appears a promising therapeutic measure.33,52

NK cell receptor expression; evidence of activation

The activation of an NK cell to kill a target cell is controlled by a complex interaction between activating and inhibitory receptor signals and can be modulated by cytokines. NK cells express C-type lectin receptors such as CD94 (an inhibitory receptor) and CD69 (early activation marker).53 In addition, CD128, the receptor for IL-8 chemokine54 and CD122 (IL-2R) and CD25, the receptor for IL-2,55 are constitutively expressed on NK cells. Other receptors such as CD30 and CD154 (CD40L) are also expressed on NK cells.56,57 Previously we reported that significantly decreased CD94 and increased CD69 expression on peripheral blood NK cells in women with RSA and infertility in comparison with those of normal controls.36 Consequently, unbalanced CD69 and CD94 expression results in higher proportion of activated NK cells in vivo in these women. The expression of other receptors such as CD25, 30, 122, 128, or 154 was not different. An imbalance between inhibitory and activating receptor expression was also found in women with implantation failures.58 CD158a and CD158b inhibitory receptor expression by CD56dim/CD16+ and CD56bright/CD16 NK cells was significantly decreased, and CD161 activating receptor expression by CD56+/CD3+ NKT cells was significantly increased in women with implantation failures when compared with those of normal controls.58 Infertile women had a significantly higher expression of NK cell activation markers of CD69+ and CD161+ than fertile women had. NK cytotoxicity correlated inversely with expression of NK cells bearing the inhibition marker of CD94+.3

Resistance of trophoblast to NK-mediated cytotoxicity was reported to be the result of insufficient activating interactions between the various triggering NK receptors and their target cell ligands such as CD48.24,59 Unbalanced activating and inhibitory receptors in women with reproductive failures disrupt this mechanism and result in NK-mediated cytotoxicity against trophoblasts. This imbalance may explain the adverse reproductive outcome.

Genetic control for NK receptors

Natural killer receptors have killer cell immunoglobulin like-receptors (KIR) that recognize specific alleles of the HLA-C locus when expressed on targets. These KIR are known to be very polymorphic. A genetic variation at the KIR locus may influence the susceptibility to RSA. Phenotypic differences in KIR expression between decidual and blood NK cells in the same woman were reported.60 Therefore, KIR expression on pNK and dNK is reviewed separately. A limited repertoire of inhibitory receptors of the KIR family (inhKIR) was reported in dNK cells from women with RSA. Many dNK cells lack inhKIRs specific for the fetal HLA-Cw antigens.61 Consequently, in women with RSA, a limited maternal inhKIR repertoire and a lack of maternal inhKIR-fetal HLA-C epitope matching fail to deliver signals to inhibit NK cell activation and protect the embryo.62,63 Decreased ligands for inhKIRs could potentially lower the threshold for NK cell activation, mediated through activating receptors, thereby contributing to pathogenesis of RSA.63

Significantly higher gene frequency of activating KIR2DS1 and increased numbers of activating KIR genes in peripheral blood NK cells were noticed in women with RSA compared with normal controls. In addition, women who possessed more than two activating KIR genes were found more frequently in patients than those in controls,63 although, contradictory results were also reported.64 Decreased number of KIR was reported in women with RSA, in particular KIR2DL2.65 From a cohort of husband and wife couples, the women with a KIR2DS1 gene, and with decreased group 2 HLA-C alleles for the homologous inhibitory receptor KIR2DL1, had a tendency to have RSA.63 This suggests that a genetic variation at the KIR locus influences the susceptibility to unexplained RSA and decreased ligands for inhibitory KIRs could potentially lower the threshold for NK cell activation, mediated through activating receptors, thereby contributing to pathogenesis of RSA.63

Type1/NK1 shift

Peritt et al., reported that human NK cells could be differentiated into NK1 and NK2 subsets. NK cells grown in IL-12 (NK1) produce IL-10 and IFN-γ, whereas NK cells grown in IL-4 (NK2) produce IL-5 and IL-13.53 The presence of NK cell subsets that are analogous to T-cell subsets suggests a new role for NK cells in innate inflammatory responses and in their effect on adaptive immunity. Other studies have defined other subsets of NK cells such as NK3 or NKr1 cells. An NK1 shift in peripheral blood or endometrial NK cells which represent innate inflammatory responses has been reported in women with reproductive failures.66,67 In early pregnancy, peripheral blood CD56bright and CD56dim NK cells are significantly increased compared with those in non-pregnant women. In the early pregnancy decidua, the main constituents of the population of CD56bright and CD56dim NK cells are TGF-beta-producing NK3 type cells and NK1 type cells are rare.67 In women who miscarry, IL-10 producing NKr1 cells and its subset in the peripheral blood were significantly decreased. NK3 type cells in decidua were significantly decreased compared with those in normal pregnant subjects.67 Recently, we reported that NK-1 shift in peripheral blood NK cells was identified in non-pregnant women with RSA and implantation failures. Proportion (%) of CD56bright/interferon-γ (+)/TNF-α (+) cells was significantly higher in women with RSA and implantation failures compared with that of healthy controls. Proportion of CD56bright/IL-4(+)/IL-10(+) cells was very low (<2%) in all groups, but was significantly lower in women with RSA than that of controls. The TNF-α/GM-CSF expressing CD56bright cell ratio was significantly higher in women with RSA and implantation failures than in controls.66 Other evidence of NK1 was reported in a woman who received specific immune therapy for anti-wasp venom immunotherapy. Massive infiltration of CD56+ cells (72.7%) and decreasing CD3:CD56 ratio in decidual vessels blood was reported in a woman who received specific immunotherapy and had immature birth. Thus NK1 shift induced by specific immunotherapy may trigger abortions or immature birth.68

Endometrial/decidual NK cells

The origin of decidual CD56bright/16 NK cells remains unknown, but because they closely resemble the minor agranular CD56bright NK cell population in the blood, one possibility was suggested that this blood NK cell population might move into the uterus to proliferate, differentiate, enlarge, and acquire cytoplasmic granules in the hormone-rich mucosal microenvironment. Recently, uterine segment transplantation in mice has shown that uterine NK precursor cells do not self-renew in the uterus. Transplantable uterine NK precursor cells are found in spleen, lymph nodes, bone marrow, thymus, and liver.69 The recruitment and/or increase in uterine NK cells (CD56+) into the uterus is not dependent on the physical presence of an implanting embryo, but is instead controlled hormonally.70 In decidua, human first-trimester trophoblast cells produce CXCL12, which in turn chemo-attracts decidual CD56bright/16 NK cells and at a local level to modulate the immune milieu at the materno–fetal interface.71

Studies suggest that CD56+ uNK cells play an important role in implantation and the increase in cytotoxic NK cells in the peripheral blood and that the endometrium affects the therapeutic results of IVF-ET.31 In the endometrial tissue obtained at a previous phase of the IVF cycle, the increase in CD56dim/16+ NK cells (%) and the decrease in CD56bright/16 cells in the aborted group were significant when compared with those of the delivered group in women with IVF-ET. It was suggested that modifications of NK cytotoxicity or NK subpopulations might contribute to improvements in IVF outcomes.31 When endometrial CD56+ NK cells were analyzed, the number of NK cells in women with a history of RSA (mean 11.2% range 1.1–41.4%) was significantly higher than in the control women (mean 6.2% range 2.2–13.9%). However, the number of endometrial NK cells was reported not to be a good predictor as no significant difference in uNK numbers was observed between women who miscarried and women who had a live birth in a subsequent pregnancy.72

We have reported that significantly higher proportion of women with RSA demonstrated abnormal placental lesions at the implantation site compared with women with elective abortion. Immunopathologic evaluation of the placental implantation site that terminated in a spontaneous abortion revealed a significantly elevated CD57+ NK cells at the implantation site (29.6%) (P = 0.030), inadequate cytotrophoblast invasion depth (54.1%) (P = 0.000), inadequate syncytium formation (44.1%) (P = 0.004), and presence of thromboembolism in decidual vessels (33.9%) (P = 0.025).73 Analysis of decidual cells from miscarriage by scanning electron microscope (SEM) and a transmission electron microscope (TEM) revealed a moderate degree of degeneration, a decreased number of junctional complexes and an increased number of NK cells. This finding suggests a lack of intercellular communication, a phenomenon which obviously alters proper implantation and leads to the induction of embryonic disgenesis and miscarriage.74

There is some evidence that uNK cells may interact with other cells in the decidua. Significantly more activated leukocytes were detected in the decidua of women with unexplained RSA who had a normal male karyotype compared to the other groups.75 The perforin expression on CD56+/3/8 NK cells and CD3+/8+ T cells in miscarriage with normal karyotype was significantly increased compared with those in miscarriage with abnormal karyotype and induced abortion.76 This supports the notion that cellular immunity in general is involved in unexplained RSA.

The NK cell receptor expression in decidual NK cells was reported to be similar to that of peripheral blood NK cells. The CD158a and CD94 expression in miscarriage with normal karyotype were significantly decreased compared with those in miscarriage with abnormal karyotype and those in induced abortion.76 A positive correlation between CD94 and CD158a expression on NK cells, negative correlations between CD94 on NK cells and perforin on NK cells/T cells, and between CD158a on NK cells and perforin on T cells were found in the decidua. A divergence of NK cell repertoire in the decidua might be related to etiology of sporadic miscarriage with normal karyotype.76

Table II lists immunohistochemistry studies of endometrium and decidua of women with RSA and discrepant reports exist regarding the proportion of NK cells. It is interesting to notice that study populations, controls, and NK markers vary in each study. Studies using CD57 mAb demonstrated elevated NK cell populations in decidua73,77versus, studies using CD56 did not see any change.75,78 These differences may be related to patient selection criteria, timing of sample collection, presence of LPD,79 NK subpopulations and NK cell markers. Studies using flow cytometry provide a powerful analysis of NK populations.80,81 However, histological evaluation of endometrial or decidual NK cells is limited. In addition, contamination of tissue samples with peripheral blood NK cells or other CD56+ cells is highly likely, and often it requires intense technical manipulation for cell preparation. Therefore, interpretation of data should be made carefully even considering technical aspect of NK cell studies.

Table II.   Immunohistochemical Studies of Endometrial and Decidual Tissues in Women with RSA
Endometrial study
StudyStudy populationMid luteal endometrial biopsy
Clifford et al.8729 women with RSA (>3)
Controls are 10 parous women
Increased endometrial CD56+ NK cells
Quenby et al.8822 women with RSA (>3)
Controls are 9 fertile women
8/22 had a few CD57+ cells versus none in the controls. Women with miscarriages had significantly more CD4+, CD8+, CD14+, CD16+, and CD56+ leukocytes than either those who had live births or women with proven fertility
Michimata et al.8917 women with RSA (>2 with normal karyotype)
Controls are 15 women with male factor infertility
No difference in CD56+ cells
Tuckerman et al.7287 women with RSA
10 normal controls
Increased endometrial CD56+ NK cells. No predictive value for the index pregnancy
Decidual study
StudyStudy populationPlacental tissue
Hill et al.7820 women with RSA (>4)
Controls are 20 elective terminations
No difference in CD56+ cells
Vassiliadou and Bulmer7740 women with SA
Controls are 19 elective terminations
50% had significantly increased numbers of ‘classic’ CD57(+) NK cells when compared with normal human pregnancy
Kwak et al.7371 women with RSA (>3)
Controls are 20 elective terminations
29.6% (P = 0.03) demonstrated elevated CD57+ NK cells at the implantation site
Quack et al.7517 women with RSA of normal male pregnancy
Controls are 20 elective terminations and 21 unexplained RSA with trisomy 16
No difference in CD56+ NK cells. Decreased CD56:CD45 ratio in women with RSA of normal male pregnancy compared to normal elective abortion


The RSA and unexplained infertility with multiple implantation failures are associated with TH1 immune response, in which NK cell is partly attributing to the immunopathology. It is important to understand that NK indices are not reflecting specific immune responses to pregnancy and NK cell numbers and activity can fluctuate according to different variables, such as hormonal effect, exercise, time of day and sympathetic response to stressors.82–85 Study results may vary depending on laboratory techniques, sample management, or selection of study population. Therefore, the so called ‘NK studies’ may reflect a wide scope of NK characteristics in different NK subsets. In addition, peripheral blood NK cell numbers do not necessarily correlate with NK cytotoxicity.51 Considering these caveats, an increase in NK cell numbers and /or activity in pre- or post-conception period when assayed in (each individual woman) women with recurrent pregnancy losses or infertility with multiple implantation failures is of significant clinical utility and concern. In addition, immuno-phenotypic characteristics of NK cells in these women support the changes for their increased NK activity status. Further studies are needed to explore the underlying mechanism of NK cells in RSA, infertility, and other reproductive immunopathologies. Possible neurological and hormonal control of NK and T cells in women with reproductive failures needs further investigation.