Pre-implantation endometrial leukocytes in women with recurrent miscarriage

S. Quenby1,4, M. Bates2, T. Doig2, J. Brewster3, D.I. Lewis-Jones3, P.M. Johnson2 and G. Vince2

1 Division of Obstetrics and Gynaecology, City Hospital, Hucknall Road, Nottingham, 2 Department of Immunology and 3 Department of Obstetrics and Gynaecology, University of Liverpool, Liverpool, L69 3BX, UK


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Immunohistochemistry was used to investigate the leukocyte populations in the endometrium of women suffering recurrent miscarriage. Mid-luteal phase endometrial biopsies were taken from 22 patients with idiopathic recurrent miscarriage and from nine women with normal obstetric histories. The samples were dated histologically and stained with a panel of monoclonal antibodies to identify leukocytes. The outcome of any pregnancy in subsequent cycles following the biopsy was determined. Similar numbers of cluster designation (CD)3+ and CD8+ cells were seen in both groups. However, CD4+, CD14+, CD16+, CD56+ and MHC class II+ cells were significantly higher in the recurrent miscarriage group than in the controls. Two patients had B cells (CD22+) in their endometrium. No CD57+ cells were seen in the controls; however, eight of the patients had a few CD57+ cells present. Only two patients, both from the recurrent miscarriage group, had CD69+ leukocytes in their endometrium. Patients who had miscarriages following endometrial biopsy had significantly more CD4+, CD8+, CD14+, CD16+, and CD56+ leukocytes in their endometrium than either those who had live births or women with proven fertility. A different population of leukocytes was found in the pre-implantation endometrium from recurrent miscarriage patients as compared to those from fertile controls. These differences were accentuated in women who had a miscarriage subsequent to the biopsy compared with those who subsequently had a live birth.

Key words: endometrial leukocytes/immunohistochemistry/recurrent miscarriage


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Leukocytes are an important constituent of human endometrium, accounting for 10% of stromal cells in the proliferative phase. Immediately prior to implantation in the secretory phase, 20% of endometrial cells are leukocytes and, in early pregnancy, leukocytes comprise 30% of decidual cells (Bulmer et al., 1991Go). The leukocyte population in the endometrium consists of T cells, macrophages, and large granular lymphocytes. T cells account for 45% of leukocytes in the proliferative phase of the cycle. Their numbers remain constant throughout the menstrual cycle, although they decrease in proportion relative to other types of leukocyte by the secretory phase of the cycle (Bulmer, 1996Go). Macrophages account for a substantial proportion of the leukocyte population in human endometrium throughout the menstrual cycle. Endometrial macrophages have been characterized by their consistent expression of CD14, CD68 and MHC class II (Bulmer, 1996Go). The largest leukocyte population in the human endometrium is that of large granulated lymphocytes (LGL) which express the natural killer (NK) cell antigen CD56. In contrast to peripheral blood NK cells, LGL do not express the low affinity Fc{gamma} receptor CD16, or the CD57 molecule. Around the time of implantation, LGL comprise 70–80% of the leukocyte population in the endometrium and numbers increase still further if conception occurs (King et al., 1989Go; Starkey, 1992Go).

It has been suggested that women with recurrent early pregnancy loss (RPL) may have an impaired CD56+ leukocyte response in the decidua (Hill et al., 1995aGo), whilst women with spontaneous losses (not RPL) have increased numbers of CD57 `classical' NK cells (Vassiliadou and Bulmer, 1996aGo). However, it is difficult to characterize leukocyte populations after recurrent miscarriage, as it is difficult to distinguish causes of miscarriage from post-miscarriage inflammatory responses (Bulmer, 1996Go). Lachapelle and co-workers used flow cytometry of dispersed non-pregnant endometrium to study leukocytes in women with recurrent miscarriage compared with normal obstetric histories (Lachapelle et al., 1996Go). They found an increased CD4:CD8 ratio and an increase in CD20+ B cells. The proportion of NK cells was identical in both groups; however, those women with recurrent miscarriage had an increased proportion of CD16+ CD56dim LGL compared to the normal population where CD16 CD56bright LGL were more prevalent. Here, we have used immunohistochemistry to investigate the immunophenotypic profile of the endometrium of women suffering recurrent pregnancy loss.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Local ethical committee approval was obtained for this study. Women attending for their first visit at the Recurrent Miscarriage Clinic at the Liverpool Women's Hospital were asked if they would give informed consent to take part in the study. In accordance with the Miscarriage Clinic protocol, they were advised not to try to conceive for 6 weeks until the results of their investigations were available (Quenby and Farquharson, 1993Go; Li, 1998Go). If they consented to the study, a mid-luteal phase endometrial biopsy was taken with a Wallace endometrial sampler. The sample was taken 21 days from their last menstrual period in women with regular 28 day cycles. Each patient had a serum oestrogen and progesterone test taken at the time of the biopsy to ensure that these values were consistent with the mid-luteal phase. Patients were excluded if there was a known cause for the miscarriages, e.g. anti-phospholipid syndrome (APS) (as described in Drakeley et al., 1998Go), endocrine cause [as defined by oligomenorrhoea (Quenby and Farquharson, 1993Go)] or balanced translocation.

Biopsies were obtained from 25 women. One biopsy was excluded because the patient subsequently had two positive tests for lupus anticoagulant. All biopsies were examined histologically by two observers to date the endometrium according to Noyes criteria (Noyes et al., 1950Go); patients were included in the study if their biopsy was consistent with days 19–22 of the cycle. The biopsies of two women were excluded from the study; one woman did not ovulate in the cycle studied, as her progesterone level was <10 nmol/l, and the biopsy of the other woman showed endometrial differentiation consistent with early secretory phase (<day 18), perhaps because of luteal phase insufficiency [biopsies accurately timed from the luteinizing hormone (LH) surge in two consecutive cycles would be necessary to make this diagnosis]. The remaining 22 biopsies comprised the patient group (Table IGo). These women had had at least three consecutive miscarriages; one had second trimester miscarriages as well and one had very early biochemical losses [raised serum human chorionic gonadotrophin (HCG) concentrations at 4 weeks gestation as the only evidence of pregnancy]. Six patients had a previous live birth and one had a history of an ectopic pregnancy. The outcomes of the pregnancies following endometrial biopsy were: 11 patients had live births at term, four patients had first trimester miscarriages and one patient had an ectopic pregnancy. Of the remaining patients, one patient was lost to follow-up and five patients have yet to conceive.


View this table:
[in this window]
[in a new window]
 
Table I. Characteristics of the women used in this study
 
The control population consisted of patients admitted to the day care gynaecology unit for laparoscopic tubal ligation, who were using only barrier methods of contraception and were in the mid-luteal phase of their menstrual cycle, as estimated from their last menstrual period. They were included in the study provided they had two or more normal pregnancies and excluded from the study if they had more than one miscarriage. This group of women also had serum oestradiol and progesterone estimations to ensure they were in the mid-luteal phase of the cycle, and histological examination of the endometrial biopsy confirmed that they were between days 19 and 22 of the cycle. The two groups were of similar age (Table IGo).

The endometrial samples were immediately frozen in liquid nitrogen and stored at –70°C. Cryostat sections 5 µm in thickness were cut and mounted on glass slides. Sections were stained with haemalum to check that the biopsies contained endometrial tissue. After drying overnight at room temperature, slides were wrapped in aluminium foil and frozen at –20°C until immunostaining. Sections were stained using the alkaline phosphatase anti-(alkaline phosphatase) (APAAP) system using a panel of 14 monoclonal antibodies (Table IIGo).


View this table:
[in this window]
[in a new window]
 
Table II. Monoclonal antibodies used in this study
 
Sections were fixed for 10 min in acetone and, after washing in Tris-buffered saline (TBS 0.05 mol/l, pH 7.6), were incubated with the appropriate diluted monoclonal antibody for 30 min (Table IIGo). Mouse IgG was used in place of the first antibody as a negative control. After 2x5 min washes in TBS, bound antibodies were detected by the APAAP method using rabbit anti-(mouse IgG) IgG (diluted 1:25; Dako Ltd, High Wycombe, UK) for 30 min, washed in TBS and then incubated with pre-formed complex of calf intestinal alkaline phosphatase and mouse monoclonal anti-(alkaline phosphatase) (APAAP, diluted 1:50; Dako Ltd.) for a further 30 min. Staining was developed with Napthol AS-MX phosphate and Fast Red (Sigma, Poole, Dorset, UK) with the inclusion of 1 mmol/l levamisole to block any endogenous alkaline phosphatase. Slides were counterstained with haemalum and mounted in Aquamount (BDH, Poole, Dorset, UK).

The number of positive cells for each monoclonal antibody used was counted in 10 random fields at x400 magnification. The observer was blinded to the identity of the slide. On sections stained for CD45+ cells, a count was made of both positive and negative cells to allow calculation of the total number of cells per high-power field. Glandular epithelial cells were not included in this count. The numbers of positive cells for each antibody were expressed as a percentage of the total cells for that patient to allow for inter-patient variability.

The statistics were calculated on the Arcus (Cambridge, UK) software package for personal computers. The data was tested with the Shapiro–Wilks test and found to be non-normally distributed. Therefore the non-parametric Mann–Whitney U-test was used to assess differences between the control and recurrent miscarriage endometrium using two-sided P-values. The difference between the endometrium of women that had a live birth subsequent to the biopsy was compared to that of those who miscarried following the biopsy, using the Mann–Whitney U-test. 95% confidence intervals were calculated and the conventional level of P < 0.05 was taken as the limit of significance.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Cytokeratin and vimentin were used as positive controls staining glandular epithelium and stroma respectively. Those sections stained with mouse IgG alone showed no positive labelling (Figure 1cGo). Endometrial biopsies were dated by histology and were equally distributed over days 19–22 of the cycle in both the RPL and control groups.



View larger version (120K):
[in this window]
[in a new window]
 
Figure 1. Immunohistochemical labelling of endometrial tissue. Visualization by alkaline phosphatase/anti-alkaline phosphatase with Fast Red. CD45 antibody labelling similar numbers of cells in endometrium from a patient who had suffered recurrent miscarriages (a) and a patient with a normal obstetric history (b). Negative control stained with mouse IgG alone (c). CD56 antibody labelling more cells in the endometrium from a woman who had suffered recurrent miscarriages then miscarried again in the pregnancy after the biopsy (d), than a woman who had suffered recurrent miscarriages then had a live birth in the pregnancy after the biopsy (e). (f) Endometrium from control with a normal obstetric history with fewer CD56+ cells than in (d). Bar = 10 µm (all panels).

 
The number of CD45+ cells as a proportion of the total number of stromal cells was greater in the patients (Figure 1aGo) compared to the controls (Figure 1bGo) but this difference was not statistically different (Table IIIGo). CD45+ cells constituted 23% of all cells in the patients and 17% in the controls (Table IIIGo). There were similar proportions of CD3+ and CD8+ T cells in both groups (Table IIIGo) but there were significantly higher numbers of CD4+ leukocytes in patients compared with controls (Table IIIGo). CD14+, CD16+ and CD56+ cells were also significantly higher in the patients than the controls (Table IIIGo). No CD57+ cells were seen in the controls; however, eight of the patients had a few CD57+ cells present. MHC class II positive cells were more common in the endometrium of patients compared with controls (Table IIIGo).


View this table:
[in this window]
[in a new window]
 
Table III. Endometrial leukocytes expressed as a percentage of total cells per 10 high power fields.
 
The recurrent miscarriage patients were then analysed on the basis of the outcome of their subsequent pregnancy. Patients who had miscarriages in the pregnancy after the endometrial biopsy had significantly more CD4+, CD8+, CD14+, CD16+, and CD56+ (Figure 1dGo) leukocytes than either those who had live births (all P < 0.05; Figure 1eGo) or control endometrium (Figure 1fGo; Table IVGo). Two patients had more CD22+ B cells than any other patient or control; one of these went on to have a live birth and one a miscarriage. Only two patients had significant numbers of CD69+ leukocytes in their endometrium and both these miscarried.


View this table:
[in this window]
[in a new window]
 
Table IV. Endometrial leukocytes expressed as a percentage of total cells per 10 high power fields
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The percentage of leukocytes as a proportion of endometrial cells in both the controls and the patients was ~20%, as expected from the published literature (Bulmer et al., 1991Go). However, there appear to be intrinsic differences in the population of leukocytes in the endometrium of women with recurrent miscarriage and women with a successful obstetric history.

CD3, a marker for all T lymphocytes (CD4 and CD8), labelled a similar proportion of leukocytes in the pre-implantation endometrium of patients and controls. Other studies have shown reduced numbers of T cells in normal first trimester decidua compared to normal endometrium (Vassiliadou and Bulmer, 1996bGo), although similar numbers of T cells were found in normal first trimester decidua compared to decidua obtained after spontaneous abortion (Vassiliadou and Bulmer, 1998Go). However, the majority of spontaneous abortions are due to fetal chromosomal abnormalities and hence extrapolation of data from spontaneous abortions may not be applicable to cases of recurrent miscarriage.

In the present study, more CD4+ than CD3+ cells were observed in the endometrium. Macrophages can also express the CD4 antigen (Woods et al., 1983Go) and macrophages (labelled with CD14) were significantly more numerous in the patients than the controls (Table IIIGo). Therefore, the greater proportions of CD4+ cells may be due to an increase in macrophages (co-expressing CD14 and CD4) in the patients. Further experiments with double labelling are necessary to confirm this. Women who miscarried after the biopsy had more CD4+ and CD14+ leukocytes than women who had live births following the biopsy (Table IVGo), suggesting that these cells have a role in miscarriage. The exact function of these endometrial macrophages is unknown, although several roles have been suggested. Endometrial macrophages increase in number during the luteal phase of the cycle and increase further in early pregnancy decidua, suggesting that they may have a role in placentation (Bulmer et al., 1991Go; Hunt and Robertson, 1996Go). Macrophages strongly express MHC class II antigens necessary for antigen presentation (Bulmer, 1996Go) and are therefore able to activate T cells. Decidual macrophages may also have a phagocytic role, as trophoblast invasion into the decidua in normal pregnancy could produce debris requiring removal by phagocytosis (Bulmer, 1996Go). There is also evidence of a role for macrophages in miscarriage from a murine model of miscarriage, where embryo loss was found to be associated with activated macrophages (Gendron et al., 1990Go). In humans, a small increase in the number of CD68 labelled macrophages was found in the decidua from spontaneous abortions compared to normal controls (Vassiliadou and Bulmer, 1996aGo); however, this could have been a post-abortion inflammatory response.

Macrophages have also been proposed as immunosuppressive cells in human early pregnancy decidua. This immunoregulatory activity has been reported to be mediated by the secretion of prostaglandin E2 by decidual macrophages and may block the function of lymphoid cells in the decidua, with potential lytic activity against the placental trophoblast (Parhar et al., 1989Go). However, the importance of local immunoregulatory cells in pregnancy has yet to be established. Macrophages are also capable of producing a range of cytokines including macrophage colony stimulating factor (M-CSF), granulocyte colony stimulating factor (G-CSF) and tumour necrosis factor (TNF)-{alpha}. In common with T cells and LGL, decidual macrophage cytokine production in pregnancy could play a role in the control of placental growth (Bulmer, 1996Go). CD56+, CD16 cells have been found to act synergistically with endometrial macrophages to enhance the release of interferon (IFN)-{gamma} (Marzusch et al., 1997Go). NK cells can be activated by the cytokines interleukin (IL)-2, TNF-{alpha} and IFN-{gamma}; thus, they could act together with other activated decidual cell populations to attack trophoblast populations, resulting in subsequent pregnancy loss (Vassiliadou and Bulmer, 1996aGo). Recent attention has focused on elucidating the immunobiological roles of cytokines in normal human pregnancy following the accumulated reports of complex cytokine activity with in-utero placental tissues (reviewed in Robertson et al., 1994Go; Lim et al., 1996Go, 1998Go). It has been suggested that recurrent miscarriage may be associated with a Th1 cytokine (IFN-{gamma} IL-2, TNF-{alpha}) response, whereas successful pregnancy may be associated with a Th2 (IL-4, IL-5, IL-10) cytokine response (Hill et al., 1995bGo).

In the present study, only two patients were found to have increased numbers of B cells in their endometrium; one subsequently had a live birth and one a miscarriage. This is in contrast with the study by Lachapelle et al. who found that recurrent miscarriers with increased B cells in their endometrium were more likely to miscarry (Lachapelle et al., 1996Go). However, these authors used flow cytometric analysis of dispersed cell populations, which may not be an accurate reflection of the in-vivo situation, since tissue digestion may result in a skewed population of cells.

The present study found more CD56+ LGL in the pre-implantation endometrium of the recurrent miscarriage patients than in the controls. Furthermore, there were more CD56+ leukocytes in the endometrium of patients that went on to have miscarriages than in those who had live births (Table IVGo). This is in contrast to previous findings (Lachapelle et al., 1996Go), which found similar numbers of CD56+ cells in recurrent miscarriage patients and controls, but those women with recurrent miscarriage had an increased proportion of CD16+ CD56dim LGL compared to controls where CD16 CD56bright LGL were more prevalent. Although higher numbers of LGL were also found in decidua from spontaneous miscarriages than from normal pregnancy decidua (Vassiliadou and Bulmer, 1996aGo), this may be due to post-miscarriage inflammatory changes. Christiansen put forward three hypothetical models for the pathogenesis of recurrent miscarriage (Christiansen, 1996Go); one of these (model B), predicted that an increase in CD56+ cells might indicate a poor prognosis in recurrent miscarriage patients, and our results are in accordance with his prediction. However, unexplained infertility patients were found to have fewer CD56+ cells than fertile controls (Klentzeris et al., 1994Go). Recently King et al. highlighted the importance of maternal LGL in human implantation, suggesting LGL are directly involved in maternal allogeneic recognition of the placenta via their expression of receptors for HLA-G and HLA-C, present on some fetal trophoblast populations (King et al., 1998Go).

The present study demonstrated more CD16+ leukocytes in the pre-implantation endometrium of the patients than in the controls, and these were also more prevalent in those women who subsequently miscarried compared to those who had a successful pregnancy (Tables III and IVGoGo). This may be indicative of a chronic or latent infection; further studies would be needed to confirm this.

In the present study of pre-implantation endometrium, CD57+ cells were observed in only eight out of 22 miscarriers and were not seen in the controls. The CD57+ cells were also seen more frequently in women who subsequently miscarried than in those who had live births (Table IVGo). Vassiliadou and Bulmer (1996a) also found more CD57+ leukocytes in decidua of spontaneous miscarriages compared to normal pregnancy Therefore CD57+ or `classical' NK cells could be hostile to invading trophoblast.

The only two patients in this study with significant numbers of CD69+ cells miscarried, implying a deleterious role for activated leukocytes in implantation. Other evidence for this role comes from a study showing that CD56+ cells that were also CD69+ were found more commonly in the decidua from spontaneous abortions than from normal pregnancy (Kodama et al., 1998Go), but again, this may be due to post-miscarriage inflammatory changes.

In conclusion, a different population of leukocytes was found in the pre-implantation endometrium from recurrent miscarriage patients as compared to those from fertile controls. Furthermore, when the endometrium of women who subsequently miscarried was compared to those who had live births, the differences seen in the original analysis were accentuated. The endometrium of recurrent miscarriers could be hostile to the invading trophoblast. Another possible explanation for our data would be that the endometrium of recurrent miscarriers is more receptive, leading to the implantation of abnormal fertilized ova which subsequently miscarry.


    Acknowledgments
 
The work was supported by the NHS Executive Northwest, Research and Development Directorate. T.Doig was funded by a Wellcome Trust Vacation Scholarship.


    Notes
 
4 To whom correspondence should be addressed Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Bulmer, J.N. (1996) Cellular constituents of human endometrium in the menstrual cycle and early pregnancy. In Bronson R.A., Alexander N.J., Anderson D. et al. (eds), Reproductive Immunology. Blackwell Science, Oxford. pp. 212–239

Bulmer, J.N., Longfellow, M. and Ritson, A. (1991) Leukocytes and resident blood cells in endometrium. Ann. N. Y. Acad. Sci., 622, 57–68.[Medline]

Christiansen, O.B. (1996) A fresh look at the causes and treatments of recurrent miscarriage, especially its immunological aspects. Hum. Reprod. Update, 2, 271–279.[Abstract]

Drakeley, A.J., Quenby, S. and Farquharson, R.G. (1998) Mid-trimester loss-appraisal of a screening protocol. Hum. Reprod., 13, 1975–1980.[Abstract]

Gendron, R.L., Nestel, F.P., Lapp, W.S. and Baines, M.G. (1990) LPS-induced fetal resorption in mice is associated with the intrauterine production of TNF-{alpha}. J. Reprod. Fertil., 90, 395–402.[Abstract]

Hill, J.A., Melling, G.C. and Johnson P.M. (1995a) Immunohistochemical studies of human uteroplacental tissues from first trimester spontaneous abortion. Am. J. Obstet. Gynecol., 173, 90–96.[ISI][Medline]

Hill, J.A., Polgar, K. and Anderson, D.J. (1995b) T-helper 1-type immunity to trophoblast in women with recurrent spontaneous abortion. JAMA, 273, 1933–1936.[Abstract]

Hunt, J.S. and Robertson, S.A. (1996) Uterine macrophages and environmental programming for pregnancy success. J. Reprod. Immunol., 32, 1–25.[ISI][Medline]

King, A., Wellings, V., Gardner, L. and Loke, Y.W. (1989) Immunohistochemical distribution of the unusual large granular lymphocytes in human endometrium throughout the menstrual cycle. Hum. Immunol., 24, 66–73.

King, A., Burrows, T., Verma, S. et al. (1998) Human uterine lymphocytes. Hum. Reprod. Update, 4, 480–485.[Abstract/Free Full Text]

Klentzeris, L.D., Bulmer, J.N., Warren, M.A. et al. (1994) Lymphoid tissue in the endometrium of women with unexplained infertility: morphometric and immunohistochemical aspects. Hum. Reprod., 9, 646–652.[Abstract]

Kodama, T., Hara, T., Okamoto, E. et al. (1998) Characteristic changes of large granular lymphocytes that strongly express CD56 in endometrium during the menstrual cycle and early pregnancy. Hum. Reprod., 13, 1036–1043.[Abstract]

Lachapelle, M.-H., Mirion, P., Hemmings, R. and Roy, D.C. (1996) Endometrial T, B, and NK cells in patients with recurrent spontaneous abortion. J. Immunol., 156, 4027–4034.[Abstract]

Li, T.C. (1998) Guides for practitioners. Recurrent miscarriage: principles of management. Hum. Reprod., 13, 478–482.[ISI][Medline]

Lim, K.J.H., Odukoya, O.A., Li, T.C. et al. (1996) Cytokines and immuno-endocrine factors in recurrent miscarriage. Hum. Reprod. Update, 2, 469–468[Abstract/Free Full Text]

Lim, K.J.H., Odukoya, O.A., Ajjan, R.A. et al. (1998) Profile of cytokine mRNA expression in peri-implantation human endometrium. Mol. Hum. Reprod., 4, 77–81.[Abstract]

Marzusch, K., Buchholz, F., Ruck, P. et al. (1997) Interleukin-12 and interleukin-2-stimulated release of interferon-{gamma} by uterine CD56++ large granular lymphocytes is amplified by decidual macrophages. Hum. Reprod., 12, 921–924.[ISI][Medline]

Noyes, R., Hertig, A. and Rock, J. (1950) Dating the endometrial biopsy. Fertil. Steril., 1, 3–25.[ISI][Medline]

Parhar, R.S., Yagel, S. and Lala, P.K. (1989) PGE2-mediated immunosuppression by first trimester human decidual cells blocks activation of maternal leukocytes in decidua with potential anti-trophoblast activity. Cell. Immunol., 120, 61–74.[ISI][Medline]

Quenby, S. and Farquharson, R. (1993) Predicting recurring miscarriage: what is important? Obstet. Gynecol., 82, 132–38.[Abstract]

Robertson, S.A., Seamark, R.F., Guilbert, L.J. and Wegmann, T.G. (1994) The role of cytokines in gestation. Crit. Rev. Immunol., 14, 239–292.[ISI][Medline]

Starkey P.M. (1992) Natural killer cells/large granular lymphocytes in pregnancy. In Lewis, C.E. and McGee, J.O'D. (eds), The Natural Killer Cell. IRL Press, Oxford, pp. 205–240.

Vassiliadou, N. and Bulmer, J.N. (1996a) Immunohistochemical evidence for increased numbers of classical CD57+ natural killer cells in the endometrium of women suffering spontaneous early-pregnancy loss. Hum. Reprod., 11, 1569–1574.[Abstract/Free Full Text]

Vassiliadou, N. and Bulmer, J.N. (1996b) Quantitative analysis of T lymphocytes subsets in pregnant and non-pregnant human endometrium. Biol. Reprod., 55, 1017–1922.[Abstract]

Vassiliadou, N. and Bulmer, J.N. (1998) Characterization of endometrial T lymphocyte subpopulations in spontaneous early loss. Hum. Reprod., 13, 44–47.[Abstract]

Woods, G.S., Warner, N.L. and Warnke, R.A. (1983) Anti-Leu-3/T4 antibodies react with cells of monocyte/macrophage and Langerhans lineage. J. Immunol., 131, 212–216.[Abstract/Free Full Text]

Submitted on February 5, 1999; accepted on June 9, 1999.