Women with endometriosis show a defect in natural killer activity resulting in a decreased cytotoxicity to autologous endometrium*

Transcription

1 FERTILITY AND STERILITY Copyright ~ 1991 The American Fertility Society Printed on acid-free paper in U.S.A. Women with endometriosis show a defect in natural killer activity resulting in a decreased cytotoxicity to autologous endometrium* Didier J. Oosterlynck, M.D.t:j: Freddy J. Cornillie, Ph.D.:j: Mark Waer, M.D., Ph.D.t Michel Vandeputte, M.D., Ph.D.t Philippe R. Koninckx, M.D., Ph.D.:j: Rega Institute for Medical Research, and University of Leuven, Leuven, Belgium Objective: The role of natural killer (NK) cells in the decreased cellular immunity of women with endometriosis was investigated. Design, Setting, Patients: Thirty-four women were investigated prospectively before a CO2- laser laparoscopy for infertility and/or pain at the University Hospital Gasthuisberg. Endometriosis was scored blindly. Main Outcome Measure: The cytotoxicity, directed against the endometrium, was mediated by NK cells because this cytotoxicity could be removed by treating the effector cells with the NKspecific anti-leu -llb monoclonal antibody. Consequently, we evaluated prospectively in those women the lymphocyte-mediated cytotoxicity toward NK sensitive (K562-assay) and autologous endometrial target cells. Results: The NK activity (K562-assay) and the cytotoxicity against autologous endometrial cells were similarly decreased in women with endometriosis and correlated with the severity of the disease. Using heterologous effector cells, the decreased chromium release in women with endometriosis was less pronounced but still present. Conclusion: The decreased cytotoxicity to endometrial cells in women with endometriosis is mainly because of a defect in NK activity but is also partially because ora resistance of the endometrium to NK cytotoxicity. Fertil Steril 56:45, 1991 Endometriosis is a frequent pathological condition in women during their reproductive years and is known to be associated with infertility and/or pelvic pain. The reported prevalence of endometriosis varies with the surgeon's awareness of all types of lesions 1 and/or the indication for laparoscopy. In a recent study concerning 179 consecutive laparoscopies performed at our center, endometriosis was Received July 17, 1990; revised and accepted March 15, * Supported by the National Fund for Scientific Research, Brussels, Belgium. t Rega Institute for Medical Research. :\: Department of Obstetrics and Gynecology, University Hospital Gasthuisberg, University of Leuven. Reprint requests: Didier Oosterlynck, M.D., Rega Institute for Medical Research, Minderbroedersstraat 10, B-3000, Leuven, Belgium. diagnosed in 77% of patients with infertility and in 82% of patients with pelvic pain. 2 In contrast, endometriosis was found in only 1.3% to 5% of women undergoing tubal sterilization. 3-5 The cause of endometriosis remains largely unknown, although numerous hypotheses have been suggested. Sampson 6 formulated in 1927 a by now well-established hypothesis. He proposed that viable endometrial cells, by way of retrograde menstruation through the fallopian tubes, could reach the peritoneal cavity and form endometriotic lesions. Because retrograde menstruation is seen in the majority of women with patent tubes 7 and because viable endometrial cells are found in >50% of the peritoneal fluids of women whether endometriosis is present or not,s and because those cells are able to implant on ectopic sites,9 the question remains why all women do not develop endometriosis. Oosterlynck et al. NK cell activity and endometriosis 45

2 Therefore, a reliable hypothesis could be a deficiency in the rejection mechanisms of these autologous endometrial cells in the peritoneal cavity. In rhesus monkeys with endometriosis, Dmowski and co-workers1o showed a decreased blastogenesis response of peripheral lymphocytes to autologous endometrial antigens. Similar experiments performed in women revealed the same trend, but the results were not significant. ll The cell-mediated cytotoxicity toward autologous endometrial cells, however, was significantly reduced in women, suggesting an immunological basis for the development of endometriosis. ll The present study was designed to further investigate the defect in cytotoxicity toward endometrial cells found in women with endometriosis and to specify whether the defect was caused by an effector cell deficiency or by an increased resistance to cytolysis of the target cells. Patients MATERIALS AND METHODS Women (n = 34) who were scheduled for a CO 2- laser laparoscopy, because of infertility (n = 18), pain (n = 11), or both problems (n = 5) were investigated. During laparoscopy, the presence and severity of endometriosis was scored blindly by one surgeon (P.K.), who at that time was not informed of any results of the immunological assays. Endometriosis was staged according to the revised American Fertility Society (AFS) ClassificationP The patients with endometriosis were scored as follows: AFS I (n = 9), AFS II (n = 9), and AFS III, and IV (n = 6). In 10 women no endometriosis was found; we classified them as AFS o. Four of them had a normal pelvis, whereas 6 had adhesions. Cytotoxicity Assay Using Endometrial Cells As Targets Target Cells Isolation. An endometrial biopsy was taken with a pipelle de Cornier (Prodimed, Neuilly-en-Thelle, France) during the luteal phase of the cycle (day 21 ± 3) or the cycle before the laparoscopy. Part ofthe endometrial tissue was used for dating by routine histology. The remaining tissue was washed free from all contaminating red blood cells and digested in a collagenase-deoxyribonuclease (DNase) solution (10.5 U collagenase A and 1 X 105 U DNase I/100 ml phosphate-buffered saline [PBS], Boehringer Mannheim, Mannheim, Germany) for 1 hour at 37 C with continuous stirring. The single cell suspension was centrifuged (200 X g for 7 minutes), washed once with minimal essential medium (MEM) (Gibco, Grand Island, NY) and resuspended in Roswell Park Memorial Institute (RPMI) medium (Gibco) supplemented with 20% fetal calf serum (FCS, Gibco). Routinely, 20 to 50 X 10 6 cells were harvested and with trypan blue exclusion viability was >90%. Labeling. In all assays, 1 X 10 6 cells were chromated with 200 #LCi (51Cr) sodium chromate (Amersham International, Amersham, United Kingdom; specific activity, 350 to 600 mci/mg) for 2 hours in 0.2 ml RPMI with 20% FCS at 37 C with 5% CO 2 We did previous experiments to identify the optimal labeling method. With a labeling time of 1, 2, and 4 hours the labeling index (maximal release/spontaneous release) was 3.7, 4.2, and 3.8 in a first experiment and 4.6, 5.4, and 5.0 in a second experiment, respectively. After labeling, the cells were washed four times (400 X g for 10 minutes) with 7 ml MEM and resuspended in RPMI + 20% FCS at a concentration of 2 X 105 cells/ml. Effector Cells From heparinized blood, peripheral blood lymphocytes were separated by Ficoll (Lymphoprep TM; Nyegaard, Oslo, Norway) centrifugation. Isolated cells were washed twice and resuspended in RPMI with 20% FCS at a final concentration of 5 X 10 6 lymphocytes/ml (viability> 95%). Each assay was duplicated with heterologous control lymphocytes. These effector cells were always from the same male control person (D.O.). Comparable results with other, both male (n = 4) and female (n = 9), allogenic effector cells were found in 7 and 14 assays, respectively. Cytotoxicity Assay One hundred microliters (containing 2 X 10 4 or 1 X 10 4 cells) 51Cr labeled endometrial cells and 100 #LL effector cells (5 X 105 cells) were added to each well of a round bottomed microtiter plate in quadruplate. Thus an effector/target cell ratio of 25/1 and of 50/1 was assayed in each patient. Because the same statistically significant results were obtained using both effector/target cell ratios, only the results for the 25/1 ratio will be given. Spontaneous release was determined in control cultures containing labeled endometrial cells and 100 #LL medium (RPMI 46 Oosterlynck et al. NK cell activity and endometriosis Fertility and Sterility

3 + 20% FCS); maximal release of 51Cr was measured by addition of 100 JLL saponine 10% (a detergent) to 100 JLL target cells. The plates were incubated at 37 C in 5% CO 2 for 20 hours, and the assay was terminated by centrifuging the plates for 2 minutes at 400 X g. This was the optimal duration for the assay because after 4 hours the specific immune release (SIR) was only 10% of the SIR obtained after 20 hours. One hundred microliters of supernatant was collected from each well, and the amount of 51Cr release was measured during 1 minute (counts per minute [cpm]) in a gamma counter. Percentage of specific lysis was calculated using the following formula: cpm experimental release 07 SIR - cpm spontaneous release /0 = X 100 cpm maximal release - cpm spontaneous release Cytotoxicity Assay Using K562 Cells As Targets The K562-cellline is a natural killer (NK)-sensitive erythromyeloid leukemic human cell line that is used as a routine assay to measure NK activity. One million K562 cells were resuspended in 0.2 ml RPMI with 20% FCS and 200 JLCi 51Cr during 1 hour at 37 C in 5% CO 2 Thereafter, target cells were washed three times (400 X g for 10 minutes) and resuspended in a final concentration of 1 X 105 cells/ml RPM I with 20% FCS. The chromium release assays were performed by incubating 1 X 10 4 K562 cells with 5 X 105 effector cells in a final volume of 200 JLL in round bottomed microtiter plates in quadruplate. Control cultures contained medium plus target cells, and maximal release was evaluated by lysis of the K562 cells with saponine 10%. The plates were incubated for 4 hours in 5% CO2 at 37 C. After this period, the plates were centrifuged at 400 X g for 2 minutes and 100 JLL supernatant of each well was collected and counted for 51Cr radioactivity. The percentage of specific lysis was determinated as described above for the assay with endometrial target cells. Flow Cytometry Analysis of Lymphocytes Subsets of lymphocytes were evaluated using a FACStar Plus (Becton Dickinson, Mountain View, CA). The following phycoerythrin or fluorescein isothiocyanate (FITC)-conjugated monoclonals were used: anti-ieu-l1c (CD16), a marker for NK cells; anti-ieu-7 (CD57), a marker for NK cells; anti-ieu- 4 (CD3), a marker for T cells; anti-ieu-3a (CD4), a marker for T helper cells; anti-ieu-2a (CD8), a marker for T cytotoxic/suppressor cells; anti-ieu-12 (CD19), a marker for B cells; and anti-ieu-16 (CD20), a marker for B cells (Becton Dickinson). To each 200 JLL of blood, in a 1/2 dilution with PBS, 20 JLL of an FITC-conjugated monoclonal and 20 JLL of another phycoerythrin-conjugated monoclonal antibody was added simultaneously, and staining was performed at 4 C for 20 minutes. Subsequently, cells were incubated with 2 ml NH 4CI (9.3 gil) to lyse red blood cells (this was done twice for 5 minutes). Cells were centrifuged, and fixation was done with 0.5% to 1 % (1 ml) paraformaldehyde. Results are given as percentage of lymphocytes, positive after staining with a given monoclonal antibody. Depletion of the NK Activity by an Anti-leu-lIb Peripheral blood lymphocytes, separated by Ficoll centrifugation, were incubated with anti-ieu-l1b (Becton Dickinson) for 45 minutes at room temperature (10 JLL monoclonal antibody/5 X 10 6 lymphocytes in 1 ml). Cells were centrifuged (twice at 2,000 X g for 1 minute) and resuspended in 1 ml rabbit complement (diluted 1/2, Gibco) and incubated for 60 minutes at 37 C. Cells were centrifuged and resuspended for functional NK assays. At the same time, viability was assessed (trypan blue staining revealed a viability of >90%). Histology To ascertain that all biopsies were taken during the midluteal phase of the cycle, part of the endometrial tissue was processed for routine histology and dating. Patients whose endometria were proliferative or showed a chronic endometritis were excluded from the study. No biopsies of the endometriotic lesions were taken in 3 patients with only subtle endometriosis (these lesions were vaporized) and, obviously, also not in the 10 patients without endometriosis. At laparoscopy, biopsies were taken from endometriotic lesions in 21 patients, and endometriosis was confirmed by histology (F.C.) in 20 patients (95%). Statistical evaluation ofthe data was done with Statistical Analysis System, using Student's t-test, ANOVA, and Spearman's correlation where appropriate. RESULTS Lymphocyte-Mediated Cytotoxicity to NK-Sensitive (K562-Assay) and Endometrial Target Cells The results of the cytotoxicity assays are shown in Table 1. The SIR in the assay with K562 target Oosterlynck et al. NK cell activity and endometriosis 47

4 Table 1 Lymphocyte-Mediated Cytotoxicity in 34 Women" o I (n = 10) (n = 9) Revised AFS 12 II (n = 9) III and IV (n = 6) Probability K562-assayb Autologous cytotoxicity assayt Heterologous cytotoxicity assay" Ratio' 62.2 (±S.5) 21.0 (±S.7) ls.7 (±7.0) 52.4 (±1I.7) 12.4 (±4.5) 22.3 (±7.S) 0.94 (±0.29) 2.0 (±1.1) 44.9 (±10.6) 6.1 (±3.3) 11.2 (±6.0) 2.3 (±1.0) 39.S (±11.2) 5.2 (±3.3) 11.7 (±5.3) 3.15 (±2.0) 0.002< < "Values are percent means ± SD of specific immune release; ANOV A used for P values. b Short (4 hours) cytotoxicity assay against K562 target cells to measure NK activity, effector/target cell ratio = 50/1. < P (Spearman) between K562-assay and autologous cytotoxicity assay = d Cytotoxicity assay (20 hours) against autologous endometrial cells, effector/target cell ratio = 25/1. cells was 62.2%, 52.4%, 44.9%, and 39.8% for the patients staged as class 0, I, II, and III to IV, respectively, according to the revised AFS classification. The NK activity was significantly decreased in relation to the severity of the disease (ANOV A, P = 0.002). The SIR in the cytotoxicity assay performed against autologous endometrial cells was 21.0%, 12.4%, 6.1%, and 5.2%, respectively. The greater the severity of endometriosis, the more the SIR decreased (ANOV A, P = ). These assays thus reveal that women with endometriosis have a defect in NK function and are unable to lyse their own endometrial cells as well as patients without endometriosis. The heterologous lymphocyte assay was used to discriminate whether the latter defect was because of an impaired function of the cytotoxic cells or because of a resistance of the endometrial target cells. In those assays with heterologous controllymphocytes, we found a SIR of 18.7%,22.3%, 11.2%, and 11.7%, respectively (ANOVA, P = 0.01). This demonstrates that when severe endometriosis (AFS II, III, and IV) was present, heterologous lyme Cytotoxicity assay (20 hours) of heterologous control lymphocytes against the endometrial cells of the study patients, effector/target cell ratio = 25/1., Ratio of the heterologous cytotoxicity assay to the autologous cytotoxicity assay. g This analysis was done using the logarithm of the ratio. phocytes also had difficulties to lyse those endometrial cells. The ratio of the SIR with the heterologous lymphocytes to the SIR with the autologous lymphocytes was 0.94, 2.0, 2.3, and 3.15, respectively (ANOV A, P = 0.002; this analysis was done using the logarithm of the ratio). This demonstrates that when no endometriosis is present, the specific release caused by the autologous lymphocytes reaches the specific 51Cr release caused by the heterologous effector cells. White Blood Cell Count and Leucocyte Subpopulation in Patients With or Without Endometriosis The decreased SIR in the previous assays could not be explained by quantitative differences in total white blood cell count being 7.16 X 10 9 /L (±1.46); 7.19 (±2.39); 7.28 (±1.04), and 6.57 (±0.67) in women without and with endometriosis classes I, II, and III to IV, respectively. Moreover, flow cytometry analysis showed no significant alteration in the per- Table 2 Cells in the Different Lymphocyte Subsets in Women With and Without Endometriosis, Measured With Flow Cytometry Monoclonal antibody Cells labeled Without endometriosis (n = S) With endometriosis (n = 15) Anti-leu-lIc (CD16) Anti-leu-7 (CD57) Anti-leu-4 (CD3) Anti-leu-3a (CD4) Anti-leu-2a (CDS) Anti-leu-12 (CD19) NK cell NK cell Tcell T helper cell T cytotoxic/suppressor cell B cell (±S.4)" 10.S7 (±2.53) 65.5 (±6.5) 44.1 (±6.4) 22.5 (±5.4) 9.6 (±2.5) ls.06 (±1I.9) 12.2 (±2.6) 70.5 (±9.5) 44.2 (±7.7) 23.5 (±9.9) 9.0 (±2.S) "Values are percents ± SD. 48 Oosterlynck et at. NK cell activity and endometriosis Fertility and Sterility

5 Table 3 Cytotoxicity Assays After Depletion With an Anti-Ieu-11b Untreated Iymphocytes b Lymphocytes after NK cell depletion C Probability<! K562-assay Autologous cytotoxicity assay Heterologous cytotoxicity assay 61.4 (±12.2) 18.4 (±12.8) 25.0 (±6.2) 6.9 (±4.1) 3.1 (±1.6) 2.2 (±2.3) Values are percent means ± SD in five study patients. b Normal untreated lymphocytes. C Lymphocytes treated with anti-ieu-11b and with rabbit complement. d Significance was measured with a Student's t-test. centage of lymphocytes from the various lymphocyte subsets (Table 2). These data reveal that the defect in NK activity in patients having endometriosis is not because of a quantitative defect. NK Cells Are Responsible for the Lysis of the Endometrial Cells The next question was if the NK cell could also be the effector cell in the assay with endometrial target cells and if the defect in both cytotoxicity assays was caused by the same cell population. Regardless of whether they had endometriosis or not (Table 3), in vitro depletion of the NK activity was performed with the NK-specific anti-ieu-llb monoclonal antibody in five study patients. The NK activity, measured with the K562-assay, decreased from 61.4 % to 6.9% for these five patients after treating the effector cells with the anti -leu -11 b mab. For the same patients, the specific chromium release in the assay with autologous endometrial cells decreased from 18.4% to 3.1 %, and in the assay with heterologous lymphocytes from 25.0% to 2.2%. Flow cytometry analyses with the F ACStar Plus showed no alteration in cell count for the other lymphocyte subsets (Table 4), thus excluding damage to the other lymphocyte subsets (by, e.g., toxic complement factors). Because we were able to reduce significantly the specific 51Cr release in those assays with endometrial target cells after depletion of the NK cells, this lymphocyte subpopulation appears to be also the effector cell in these assays. DISCUSSION It has been suggested that endometrial cells, arriving in the peritoneal cavity through the fallopian tubes, could implant and proliferate in some patients because of defects in the cell-mediated immune system. Dmowski and co-workers10,1l studied both lymphocyte proliferation and cytotoxicity toward endometrial cells. They tested the blastogenesis response of peripheral lymphocytes to autologous endometrial antigen in vitro, in women as well as in rhesus monkeys. In monkeys with endometriosis, there was a significant decreased thymidine uptake showing that the specific cell-mediated immune response to that antigen was not properly activated. A similar trend was found in women. Neither Steele et alp nor ourselves (data not shown) found a difference in the lymphocyte stimulation response to the mitogen phytohemagglutinin between women Table 4 Cells in the Different Lymphocyte Subsets After NK Depletion With an Anti-Ieu-11b Untreated Lymphocytes treated Lymphocytes after NK Monoclonal antibody Cells labeled lymphocytes with complement b cell depletion C Anti-Ieu-11c (CDI6) Anti-Ieu-7 (CD57) Anti-Ieu-4 (CD3) Anti-Ieu-3a (CD4) Anti-Ieu-2a (CD8) Anti-Ieu-12 (CDI9) Anti-Ieu-16 (CD20) NKcell NK cell T cell T helper cell T cytotoxic/suppressor cell B cell B cell % % % Normal untreated lymphocytes. b Lymphocytes treated with rabbit complement but without anti-ieu-11b. C Lymphocytes treated with anti-ieu-11b and with rabbit complement. Oosterlynck et at. NK cell activity and endometriosis 49

6 with or without endometriosis. This confirms that no general depression of the immune function in endometriosis patients is found. In their cytotoxicity assays toward 51Cr-Iabeled endometrial target cells, they found significant differences between patients and controls. In 26 patients without endometriosis, a SIR of 14.6% ± 3% and in 27 patients with endometriosis, a SIR of 6% ± 1.56% were shown (P < 0.01).11 These results were even more significant when the severity of the disease was taken into account. Our experiments confirm the defect in cytotoxicity of autologous lymphocytes to 51Cr-Iabeled endometrial target cells in patients with endometriosis. The decrease in 51Cr release strongly correlated with the severity of the disease, according to the revised AFS classification. The first question to be answered was which immunological mechanism could be responsible for the in vitro lysis of the endometrial cells? The strong correlation between the K562-assay and the autologous assay for each patient (Spearman, P = 0.001) pointed toward a defect in the NK cell function. We were indeed able to decrease the SIR in the autologous and heterologous assay (6 and 11.3 times, respectively) by depletion of the NK activity, after treating the lymphocytes in vitro with an NK-specific anti-ieu-11b monoclonal antibody. The latter experiment thus confirms the importance this lymphocyte subpopulation in the in vitro lysis of endometrial cells. Heterologous lymphocytes from a male control person to all endometria were used to elucidate the question about the importance of the target cells. However, control lymphocytes also manifested difficulties to lyse the endometrial cells of patients with severe endometriosis (AFS II, III, and IV). There must be some resistance to NK-mediated lysis in these endometrial cells. Natural killer cells are present in the peripheral blood of all normal persons and resemble morphologically large granular lymphocytes (LGLs). These cells are capable of spontaneous cytotoxic activity against a wide variety of both autologous and allogenic target cells without requirement of prior sensitization. Unlike T cells, they are not restricted by the major histocompatibility complex. Natural killer cells represent a third distinct functional population of lymphocytes in addition to T and B cells. They can be identified in peripheral blood by their expression of different surface antigens.13 First, the CD56 (NKHl) surface antigen, present in >90% of LGLs, has been reported to be present on all functional NK cells. The CD16 (leu 11) surface antigen is present on the NK cells, which is responsible for the major part of the NK activity. Apart from this antigen, many NK cells can coexpress CD57 (leu 7), although CD57 negative cells also display NK activity. In conclusion, the predominant population of cells exhibiting NK activity are CD3-, CD56+, CDI6+ LGLs. With the use of these antigen markers, NK cells have been found to represent 10% to 15% of the mononuclear population in peripheral blood. The major in vivo role of NK cells is to exercise immune surveillance and to defend against virally infected or tumor cells.14 It is tempting to speculate that the same cells are involved in the protection against outgrowth of endometrial cells on ectopic sites. Therefore, it must be established that the NK cell defect in endometriosis is primary and not secondary to the inflammation provoked by endometriosis. To answer this question, the women with endometriosis included in this study will be reinvestigated 3 to 4 months after CO 2-laser excision of the endometriotic lesions. However, the fact that endometrial cells from endometriotic patients are also more resistant to cytotoxicity by heterologous lymphocytes suggests that other factors than NK cell defects are involved. Weare presently investigating which mechanism provokes the increased cytotoxicity resistance of the endometrial cells in women with endometriosis. Acknowledgments. The authors thank Halina Sobis, M.D., Ph.D., Rega Institute for Medical Research, Leuven, Belgium, for useful discussions; Etienne Clerinx for his excellent technical assistance; Luc Willebords and Greet Pottie for performing the F ACS analysis; J ozef Goebels and Eric Fonteyn for culturing the tumor cell line; and Diane W olput for typing this manuscript. REFERENCES 1. Stripling MC, Martin DC, Chatman DL, Vander Zwaag R, Poston WM: Subtle appearance of pelvic endometriosis. Fertil Steril 49:427, Cornillie FJ, Oosterlynck D, Lauweryns JM, Koninckx PR: Deeply infiltrating pelvic endometriosis: histology and clinical significance. Fertil Steril 53:978, Hasson HM: Incidence of endometriosis in diagnostic laparoscopy. J Reprod Med 16:135, Strathy JH, Molgaard CA, Coulam CB, Melton LJ III: Endometriosis and infertility: a laparoscopic study of endometriosis among fertile and infertile women. Fertil Steril 38: 667, Drake TS, Grunert GM: The unsuspected pelvic factor in the infertility investigation. Fertil Steril 34:27, Sampson JA: Peritoneal endometriosis due to menstrual dissemination of endometrial tissue into the peritoneal cavity. Am J Obstet Gynecol 14:422, Oosterlynck et al. NK cell activity and endometriosis Fertility and Sterility

7 7. Liu DTY, Hitchcock A: Endometriosis: its association with retrograde menstruation, dysmenorrhoea and tubal pathology. Br J Obstet Gynaecol 93:859, Koninckx PR, Ide P, Vandenbroucke W, Brosens IA: New aspects of the pathophysiology of endometriosis and associated infertility. J Reprod Med 24:257, Ridley JH, Edwards IK: Experimental endometriosis in the human. Am J Obstet Gynecol 76:783, Dmowski WP, Steele RW, Baker GF: Deficient cellular immunity in endometriosis. Am J Obstet Gynecol 141:377, Steele RW, Dmowski WP, Marmer DJ: Immunologic aspects of human endometriosis. Am J Reprod Immunol 6:33, The American Fertility Society: Revised American Fertility Society Classification of Endometriosis: Fertil Steril 43:351, Ritz J, Schmidt RE, Michon J, Hercend T, Schlossman SF: Characterization of functional surface structures on human natural killer cells. Adv Immunol 42:181, Trinchieri G, Perussia B: Biology of disease. Human natural killer cells: biologic and pathologic aspects. Lab Invest 50: 489, 1984 Oosterlynck et al. NK cell activity and endometriosis 51