Arch Gynecol Obstet (2009) 279:493–497

DOI 10.1007/s00404-008-0736-y

ORIGINAL ARTICLE

Are CD57+ Natural Killer cells really important in early pregnancy failure?

Emel Ebru Ozcimen · Halil Kiyici · Ayla Uckuyu · Filiz Fatma Yanik

Received: 27 May 2008 / Accepted: 14 July 2008 / Published online: 8 August 2008© Springer-Verlag 2008

AbstractObjective The aim of this study was to compare theCD57+ Natural Killer (NK) cell counts in normal pregnan-cies and in cases grouped according to diVerent types ofearly pregnancy failure.Materials and methods A prospective case control studywhich was set in Baskent University Faculty of Medicine,Obstetrics and Gynecology Department. A total of 119women whose pregnancies ended in the Wrst trimester weredivided into elective pregnancy termination, incompletemiscarriage, intrauterine demise, ectopic pregnancy andrecurrent pregnancy loss groups. CD57+ NK cells werestained and counted in the histologic preparations of thedecidua in all of these groups.Results CD57+ NK cell counts were 2.14 § 1.42 in con-trol, 2.24 § 1.92 in incomplete miscarriage, 1.82 § 1.34 inintrauterine demise, 2.54 § 1.80 in ectopic pregnancy and3.42 § 2.15 in recurrent pregnancy failure group. Therewere no statistically signiWcant diVerences between thecontrol group and the other four groups with respect to theCD57+ NK cell counts.

Conclusion This study suggests that CD57+ NK cellcount is not associated with early pregnancy failure.

Keywords CD57 · Natural Killer cells · Early pregnancy failure · Recurrent pregnancy loss

Introduction

Early pregnancy loss or failure can be considered as themost common complication of pregnancy. Approximately15% of clinically diagnosed pregnancies end up with mis-carriage. Abnormal embryonic karyotype, abnormal hor-monal milieu or several other environmental factors mightplay role in the development of an unsuitable environmentfor the implantation of the blastocyst [1].

Human endometrial and decidual tissues containimmune and inXammatory cell populations during diVerentphases of the reproduction, which are very dynamic [2]. Inthe early proliferative phase of the menstrual cycle, T cellsare the major leukocytes in the endometrium, but theydecrease in number as pregnancy ensues [3]. Macrophagesare also present within the endometrial tissue during allphases of the menstrual cycle, although they are most abun-dant in the late secretory endometrium and in the decidua ofearly pregnancy [4]. However, the major component ofstromal leukocytes in the late secretory endometrium aswell as in the Wrst trimester decidua is the ‘Natural Killer(NK)’ cell population [5].

Natural Killer cells, which are large granular lympho-cytes, recognize abnormal cells such as virus-infected ortumor cells via their cell-surface receptors, which controltheir activation, proliferation and eVects [5].

Uterine NK cells resemble the dominant CD56 popula-tion of peripheral blood NK cells in some phenotypic

This study was supported by Baskent University Grant KA06/238 after Institutional Review Board Approval.

E. E. Ozcimen · A. Uckuyu · F. F. YanikDepartment of Obstetrics and Gynecology, Baskent University Faculty of Medicine, Ankara, Turkey

H. KiyiciDepartment of Pathology, Baskent University Faculty of Medicine, Ankara, Turkey

E. E. Ozcimen (&)Feritpasa mah, Malazgirt sok, No: 4/18, Selcuklu,42060 Konya, Turkeye-mail: eparlakyigit@yahoo.com

123

494 Arch Gynecol Obstet (2009) 279:493–497

characteristics. They express CD56 as the killer activatoryand inhibitory receptors. They lack expression of othertypical NK cell markers like CD16 or CD57 [5].

Natural Killer cells express the cell-surface antigensCD16 and CD56. CD16 is a low-aYnity receptor for IgGcomplexes and it is responsible for antibody-dependent cel-lular cytotoxicity. NK cells may be divided into two popu-lations according to CD56 expression: CD56 dim andCD56 bright. CD56 dim cells are cytotoxic in vitro. CD56bright cells produce immunoregulatory cytokines, such asinterferon-gamma and tumor necrosis factor-alpha (TNF-alpha) and have little cytotoxic ability [6]. NK cells alsoexpress several other cell-surface antigens, namely, CD57,CD44, CD25, CD4 etc. [5].

Abo et al. [7] deWned a group of lymphocytes, whichthey called human natural killer-1 (HNK-1). Today they areknown as CD57+ NK cells. In recent years, it has been sug-gested that quantitative alterations of decidual NK cell pop-ulation may be correlated with recurrent spontaneouspregnancy loss. In this prospective case-control study, weaimed to analyze decidual CD57+ NK cells in women withnormal pregnancies and in cases grouped according to thetypes of early pregnancy failure. Our hypothesis wasCD57+ NK cells are important in early pregnancy failures.

Materials and methods

Patients’ characteristics

A total of 119 women, whose pregnancies ended in the Wrsttrimester between 6th and 12th gestational weeks in theDepartment of Obstetrics and Gynecology of Baskent Uni-versity Hospital were divided into Wve groups. These were(1) elective pregnancy termination, (2) incomplete miscar-riage, (3) intrauterine demise, (4) ectopic pregnancy and (5)recurrent pregnancy loss groups. Elective pregnancy termi-nation group was accepted as the control group.

Exclusion criteria were: patients older than 35 years, whohad systemic diseases (diabetes mellitus, hypertension, thy-roid disease), who used progesterone preparations or drugsother than vitamins and the pregnancies following assistedreproductive technologies. TORCH group of infections wereexcluded by performing serological tests in all the patients.

Recurrent pregnancy loss was deWned as three or moreWrst trimester pregnancy losses. In the other groups, thepatients did not have any previous early pregnancy loss. Inthe recurrent pregnancy loss group, there were no endocri-nologic, anatomic or thrombophilic reasons to explain theetiology of the losses.

Intrauterine demise was diagnosed when the cardiacactivity of the embryo could no longer be observed in a nor-mally ongoing pregnancy.

Elective pregnancy termination group was the controlgroup and the other four groups were the study groups, asmentioned before. Initially 119 women were included sothat each group consisted of 23 cases. However, 4 of thesecases were excluded later on because of the problems inhistologic preparations and 4 other cases were includedinstead of them, making up a total of 119 women includedin the study, but 115 of those were evaluated. All womensigned two informed consent forms, one for the surgicalevacuation and the other for the study. Baskent Universityethic committee gave permission for this study.

Tissue collection

Tissue samples were obtained from the patients whose char-acteristics are described above, via endometrial curettage.

Surgical evacuation was performed within an hour fol-lowing the conWrmation of the diagnosis of incomplete mis-carriage, intrauterine demise or ectopic pregnancy byultrasound. Carman aspiration and Novak curette were usedto evacuate the uterus. Carman aspiration was performedinitially which removed the fetal and/or placental tissues inthe control group as well as in the incomplete miscarriage,intrauterine demise, and recurrent pregnancy loss groups.Novak curette was used primarily in the ectopic pregnancygroup and following the Carman aspiration in the other fourgroups to obtain the decidual tissue. Decidual tissues wereaseptically rinsed by an isotonic salt solution and immedi-ately transferred to the pathology laboratory. The samepathologist examined all the specimens. The pathologistwas blinded to the type of pregnancy.

In the ectopic pregnancy group, the decidua was takenfrom the endometrium. We included the ectopic pregnan-cies which were evacuated surgically.

Immunohistochemistry

Three-�m thick sections were rested at 56°C for 8–12 h.DeparaYnization and rehydration procedures were madeby xylene and alcohol solutions. Antigen unmasking proce-dure was performed by boiling in citrate buVer solution (pH6). Slides were cooled to room temperature in 30 min.Endogenous peroxidase was blocked by the application of0.35% Hydrogen Peroxide solution for 5–10 min. Follow-ing this, CD57 Ab-1 (Clone NK-1 Neo Markers) wasapplied and the slides were incubated at 33°C for 90 min.Biotinylated goat anti-polyvalent and large volume strepta-vidin peroxidase and AEC chromogen (Neo Karkers kit)were then applied, respectively. Sections were counter-stained with hematoxylin, dehydrated in alcohol andmounted with Faramount Aqueous Mounting medium.Tris-buVer solution was used for inter-reagent washes.Non-neoplastic tonsil tissue was used as positive control.

123

Arch Gynecol Obstet (2009) 279:493–497 495

QuantiWcation

CD57 positive cells were counted within an area of 1 cm²for each slide. One slide per patient was examined. CD57positive cells in areas of microscopic hemorrhage anderythrocyte extravasations were not counted (Fig. 1). Fourcases with inXamed decidua or necrotic areas wereexcluded because these factors could aVect the CD57 cellnumbers in decidual tissue. Four other patients wereincluded instead to the suitable groups, to achieve a totalnumber of 23 cases in each group.

Statistics

Values in the data are expressed as the mean § SD of themean. SPSS 11 was used to analyze the data. Mann–Whit-ney U test was used to compare the results of one groupwith the other. One-way Anova test was used to show thediVerences in the descriptive features between groups.P values were calculated on all occasions and accepted tobe signiWcant if <0.05. Power analysis was performed toobtain the enough patients’ numbers.

Results

The mean age of the women included in the study was28.96 § 4.25 years. The average gravidity and parity were3.15 § 1.22 and 1.84 § 1.05, respectively. The mean gesta-tional age calculated according to the last menstrual period,was 7.25 § 1.69 weeks on evacuation.

Mean CD57+ NK cell count was 2.38 § 2.01 per cm2 inthe whole study population. Maternal age and gestationalage did not show any signiWcant diVerences in the groups(P > 0.05) (Table 1).

There were no signiWcant diVerences in CD57+ NK cellcounts between the control group and the incomplete mis-carriage, intrauterine demise, ectopic pregnancy and recur-rent pregnancy loss groups (P > 0.05). When the studygroups were analyzed among each other, CD57+ NK cellcounts were observed to be higher in the recurrent preg-nancy loss group, but the diVerences were not statisticallysigniWcant (P > 0.05) (Table 2).

Fig. 1 CD57+ NK cells in decidua. The one on upper left corner maybe located in the lumen of a small capillary. £400

Table 1 Features of the control and the study groups, Konya 2007

Data are expressed as mean § standard error. One-way Anova test was used to compare the groups

NS not statistically signiWcant

Control Incomplete miscarriage

Intrauterine demise

Ectopic pregnancy

Recurrent pregnancy loss

P value

Maternal age (years) 29.11 § 4.30 28.64 § 6.30 29.42 § 3.17 29.67 § 4.0 29.8 § 2.19 NS

Gestational age (weeks) 6.44 § 1.09 7.90 § 1.80 7.80 § 1.62 7.29 § 0.65 7.27 § 1.80 NS

Gravidity 1.56 § 0.72 2.32 § 1.24 1.65 § 0.88 2.68 § 1.84 2.95 § 0.36 <0.05

Parity 0.55 § 0.30 1.32 § 1.25 1.59 § 0.66 1.54 § 1.23 1.70 § 0.62 NS

Table 2 Comparison of the Wve groups according to CD57+ NK cell counts, Konya 2007

Data are expressed as mean § standard error. One-way Anova test was used to compare the groups

NK Natural Killer; NS not statistically signiWcant

P > 0.05 = NS

Controln = 23

Incomplete miscarriagen = 23

Intrauterine demisen = 23

Ectopic pregnancyn = 23

Recurrent pregnancy failuren = 23

P value

CD57+ NK cell counts (number/cm2)

2.14 § 1.42 2.24 § 1.92 1.82 § 1.34 2.54 § 1.80 3.42 § 2.15 NS

123

496 Arch Gynecol Obstet (2009) 279:493–497

When evaluated within the whole study population,CD57+ NK cell counts did not have any correlation withage, gravidity or gestational age (Table 3) (P > 0.05; r =0.164, 0.184 and 0.152, respectively).

Discussion

In recent years, it has been reported that alterations in theNK cell populations in the decidua may be responsible forspontaneous abortions [8, 9]. Studies of NK cell popula-tions in women suVering early pregnancy loss may castlight on the etiology. CD56 is the most commonly usedmarker for NK cells. There are various reports about theimportance of CD56+ NK cells in early pregnancy loss. Butthe role of CD57+ NK cells in early pregnancy loss is notexactly known. As there are no enough reports to evaluatethe importance of CD57+ NK cells in early pregnancy loss,we preferred to investigate the presence of CD57+ NK cellsin the decidua of complicated pregnancies instead ofCD56+ NK cells.

Vassiliadou et al. reported an increase in the number ofCD57+ NK cells in the endometrium of the patients whohad early pregnancy loss, compared to those with uncom-plicated pregnancies. Out of 40 spontaneous abortion cases,20 had increased numbers of CD57+ NK cells in theirdecidua. The authors proposed that CD57+ NK cells, whenabundant in human decidua, might be activated by localcytokines to attack trophoblasts. But in the same studyabout 10% of women with a normal pregnancy hadincreased numbers of CD57+ NK cells in the decidua.According to the authors’ opinion, these normal pregnan-cies with increased numbers of CD57+ NK cells, mighthave destination to abort [10].

Ericsson et al. [11] showed that in normal circumstancesendometrial NK cells do not express the classic NK cellmarker, CD57. Otherwise, the CD57+ NK cells when pres-ent in the endometrium, might cause problems due to theirdamaging characteristics.

In fact it is diYcult to explain why our results were notin accordance with the similar studies in the literature. Ourmethod of evaluation of CD57+ NK cells in the histopa-thological specimen might be one of the possible factorswhich might have aVected the results of the study. CD57positive lymphocytes naturally occur in the blood circula-tion and vascular component is normally prominent in thedecidual tissue. In addition, microscopic hemorrhage anderythrocyte extravasations are commonly observed in thedecidua. Minute capillaries may be left unnoticed in animmunostained slide. All these facts might cause diYcultyin diVerentiating a true inWltration of inXammatory cellsfrom an artiWcial overlapping. In this study, we tried toeliminate circulating CD57+ cells as possible as we could.But the ones that we have counted as ‘decidual’, actuallycould be the ‘circulating’ cells which were overlappingwith the decidual tissue. CD57+ cells within minute capil-laries might have been mistakenly recognized as if theywere inWltrating cells. Thus we can say that, for an accuratecalculation, it is quite important to exclude circulating cellsfrom the count. As we mentioned before we excluded fourcases from the study as the decidua appeared to be necroticand inXamed.

Perhaps it might be more useful to know the numbers ofthe samples comprised an implantation site. CD57+ NKcells in the implantation site might have a diVerent role inthe pathogenesis of the pregnancy loss. But to obtain a sam-ple from the implantation site is not so easy practically.

We could not karyotype the miscarriages. But it wouldhave been better if the miscarriages had been karyotyped.Perhaps the number of CD57+ NK cells might changeaccording to normal and abnormal chromosome numbers.

Quenby et al. [12] have also reported that CD57+ NKcells were not observed in endometrial tissue of normalwomen. Their results were in accordance with the explana-tion that, these NK cells were hostile to the invadingtrophoblasts and in cases with recurrent miscarriages, endo-metrium revealed an increased reactivity to the implanta-tion of abnormal fertilized ova.

Quenby et al. had performed their study in mid-lutealphase of the menstrual cycle in ‘non-pregnant’ patients. Thismight perhaps explain the diVerent results observed in theirstudy and ours. One other reason may be that we workedwith Ab-1 (Clone NK-1) epitope of the CD57 marker, notthe same epitope used in their study. Thus, this methodo-logic diVerence might have aVected the results as well.

The immunological mechanisms that provide the main-tenance of normal human pregnancy and determine the lossof a pathological pregnancy remain poorly understood.There are some reports which indicate that CD57+ NK cellsplay role in pregnancy losses. But in our study we could notWnd any diVerences in the CD57+ NK cell populationbetween the control cases and the cases with early

Table 3 Correlation analysis of CD57+ NK cell counts with age, gra-vidity and gestational age, within the whole study population, Konya2007

Spearman’s correlation was used. Correlation is signiWcant at the 0.05level (2-tailed)

NK Natural Killer; R correlation value; P level of signiWcance

Age (years)n = 115

Gravidityn = 115

Gestational weeksn = 115

CD57+ NK cell counts

R 0.164 0.184 0.152

P >0.05 >0.05 >0.05

123

Arch Gynecol Obstet (2009) 279:493–497 497

pregnancy failure, as we have mentioned before. So ourresults raised this question in our minds: Are CD57+ NKcells really important in early pregnancy failure?

In our recurrent pregnancy loss group, the number ofCD57+ NK cells appeared to be higher compared to theother groups, but the diVerences were not statistically sig-niWcant (P > 0.05). Perhaps this diVerence is going to besigniWcant in a larger study population.

Finally, we can say that this study is the only one in theliterature which compares diVerent types of early preg-nancy failure case groups with each other with respect tothe presence and abundance CD57+ NK cells in thedecidua. Most of the previous studies compared normalpregnancies with recurrent pregnancy loss or spontaneousabortion cases. In our opinion, grouping the cases accord-ing to the types of early pregnancy failure might provideclues for the possible role of various NK cell subtypes inthese diVerent pathologic conditions. Of course larger studypopulations are needed to obtain more accurate results andto explain the role of CD57+ NK cells in several early preg-nancy failure case groups, if any.

ConXicts of interest No author of the current work has a personal orWnancial relationship that could inappropriately inXuence (bias) his orher actions or the writing of this manuscript, and no Wnancial or otherpotential conXicts of interest exist (includes involvement with anyorganization with a direct Wnancial, intellectual, or other interest in thesubject of the manuscript) regarding the manuscript. There are nogrants or sources of Wnancial support related to the topic or topics ofthe manuscript to be reported.

References

1. Lobo SC, Huang SJ, Germeyer A et al (2004) The immune envi-ronment in the human endometrium during the window of implan-tation. Am J Reprod Immunol 52:244–251

2. Michel MZ, Khong TY, Clark DA, Beard RW (1990) A morpho-logical and immunological study of human placental bed biopsiesin miscarriage. Br J Obstet Gynecol 97:984–988

3. Quack KC, Vassiliadou N, Pudney J, Anderson DJ, Hill AJ (2001)Leukocyte activation in the decidua of chromosomally normal andabnormal fetuses from women with recurrent abortion. HumReprod 16(5):949–955

4. Sargent IL, Borzychowski AM, Redman CWG (2006) NK cellsand human pregnancy—an inXammatory view. Trends Immunol27(9):399–404

5. Dossiou C, Giudice LC (2005) Natural Killer cells in pregnancyand recurrent pregnancy loss: endocrine and immunologicperspectives. Endocr Rev 26(1):44–62

6. Rai R, Sacks G, Trew G (2005) Natural Killer cells and repro-ductive failure-theory, practice and prejudice. Hum Reprod20(5):1123–1126

7. Abo T, Balch CM (1981) A diVerentiation antigen of human NKand K cells identiWed by a monoclonal antibody (HNK-1).J Immunol 127:1024–1029

8. Winger EE (2007) CD57+ cells and recurrent spontaneous abor-tion. Am J Reprod Immunol 58:311–314

9. Saito S (2000) Cytokine network at the feto-maternal interface.J Reprod Immunol 47:87–103

10. Vassiliadou N, Bulmer JN (1996) Immunohistochemical evidencefor increased numbers of ‘classic’ CD57+ natural killer cells in theendometrium of women suVering spontaneous early pregnancyloss. Hum Reprod 11:1569–1574

11. Eriksson M, Meadows SK, Wira CR, Sentman CL (2004) Uniquephenotype of human uterine NK cells and their regulation byendogenous TGF-�. J Leukoc Biol 76:1–9

12. Quenby S, Bates M, Doig T, Brewster J, Lewis-Jones DI, JohnsonPM, Vince G (1999) Preimplantation endometrial leukocytes inwomen with recurrent miscarriage. Hum Reprod 14:2386–2391

123