Superovulation with human menopausal gonadotropins is associated with endometrial gland-stroma dyssynchrony*

aes FERTILITY AND STERILITY Vol. 61, No.4, April 1994 Copyright ee) 1994 The American Fertility Society Printed on acid-free paper in U. S. A. r I Superovulation with human menopausal gonadotropins is associated with endometrial gland-stroma dyssynchrony* Claudio A. Benadiva, M.D.t Deborah A. Metzger, M.D., Ph.D.:\: Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, University of Connecticut Health Center, Farmington, Connecticut Objective: To investigate the prevalence of endometrial inadequacy in endometrial biopsies from women undergoing superovulation with hmg and to correlate these findings with the hormonal milieu. Design: Controlled, retrospective analysis. Setting: University-based, tertiary referral, outpatient infertility clinic. Subjects: Endometrial biopsies were performed during the late luteal phase in 89 women undergoing hmg superovulation combined with lui. Results were compared with the initial biopsies obtained as part of their routine infertility evaluation. Main Outcome Measures: Biopsies were dated by two different observers using standard dating criteria. Serum samples obtained during the midluteal phase were assayed in duplicate for E2 and P levels using commercially available RIAs. Results: Fifty-seven percent of the endometrial biopsies showed differences in the dating of the glandular epithelium that differed by >2 days when compared with the stroma. In contrast, only 13% of endometrial biopsies obtained during a nonstimulated cycle showed gland-stroma dyssynchrony. When cycles associated with gland-stroma dyssynchrony were compared with cycles associated with coordinated development of the glands and stroma, no significant differences were observed in E2 level on the day of hcg administration, midluteal serum P, midluteal E2 level, or P:E 2 ratios. Conclusions: This study demonstrates that when endometrial biopsies are obtained during the late luteal phase in patients undergoing ovarian hyperstimulation there is a significant dyssynchrony in the maturation of the glandular epithelium and the stroma. This may reflect the degree of responsiveness of an individual woman's endometrium rather than a result of the hormonal milieu. Fertil Steril 1994;61:700-4 Key Words: Infertility, endometrium, implantation, luteal phase, superovulation Pregnancy rates with IVF-ET have been disappointing particularly when compared with the results Received August 25, 1993; revised and accepted December 9, 1993. * Presented at the 48th Annual Meeting of The American Fertility Society, New Orleans, Louisiana, November 2 to 5, 1992. t Present address: The Center for Reproductive Medicine and Infertility, Cornell University Medical College, Department of Obstetrics and Gynecology, New York, New York. * Reprint requests: Deborah A. Metzger, M.D., Ph.D., Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, University of Connecticut Health Center, Farmington, Connecticut 06030-1230 (FAX: 203-679- 1436). obtained when donated eggs are transferred to agonadal women, suggesting a detrimental effect of ovarian hyperstimulation on endometrial receptivity (1). Moreover, clinical pregnancy wastage rates associated with hmg-treated cycles are much higher than those observed during natural cycles (2). The use of hmg in ovulatory women is associated with a high rate of abnormal endometrial development (3-6). However, this observation has been documented mostly on biopsies obtained during the early luteal phase, at the time of an unsuccessful ET, and before the implantation "window." As a result, very little is known about the full developmental potential of the endometrium during 700 Benadiva and Metzger Endometrial dyssynchrony with hmg Fertility and Sterility

superovulation. To study what effects superovulation may have on endometrial development, endometrial biopsies were obtained in the late luteal phase from women who failed to c-onceive after undergoing hmg superovulation with lui. Morphological findings were correlated with serum E 2 and P levels measured during the midluteal phase. Results were compared with endometrial biopsies obtained during the luteal phase of the same women as part of their initial infertility workup. hcg as day 14 in stimulated cycles and day of menses as day 28 in natural cycles. Serum samples were assayed in duplicate for E 2 and P levels using commercially available RIAs (Estradiol MAlA; Serono-Baker Diagnostics, Inc., Allentown, P A and lmmuchem Direct Progesterone Kit; ICN Biomedical, Costa Mesa, CA). Results are expressed as means ± SD and were statistically analyzed using Student's t-test and x2 RESULTS MATERIALS AND METHODS The endometrial biopsies of 89 women who had been treated with hmg superovulation with lui during 1989 to 1991 at the University of Connecticut Health Center were reviewed retrospectively. Before treatment, all couples underwent a complete infertility evaluation that included endometrial biopsy, postcoital test, hysterosalpingogram, semen analysis, and, in the majority of cases, diagnostic laparoscopy. Mean patient's age was 33.2 ± 3.4 (mean ± SD) years. Infertility factors included endometriosis (43%), male factor (25%), unexplained (18%), anovulation (6%), pelvic adhesions (6%), and immunologic (2%). Endometrial biopsies were performed during the late luteal phase (mean cycle day 25.4 ± 1.2, normalized to day ofhcg as day 14). The endometrial biopsies from nonstimulated cycles were obtained on the same women (cycle day 25.2 ± 1.5, normalized to day of menses as day 28; P = 0.4) as part of the initial infertility evaluation. All women undergoing ovarian stimulation received 5,000 to 10,000 IU IM hcg when at least two follicles reached an average diameter of 16 mm on vaginal ultrasound. Half of the patients received 50 mg P suppositories twice a day. A quantitative {JhCG was obtained on day 25, before biopsy. If the {J-hCG level was ;;:::5.0 miu/ml (conversion factor to SI units, 1.00), the biopsy was postponed until a repeat {J-hCG level was determined. Endometrial biopsies were performed with a Unimar Pipelle (Prodimed, Neuilly-en-Thelle, France) endometrial suction curette. Biopsies were dated by two different observers (C.B. and D.M.) who were blinded to the patients' treatment, using the standard dating criteria of Noyes et al. (7). Glands and stroma were separately evaluated; gland and stroma dyssynchrony was defined as > 2 days discrepancy between the dating of each individual component. Luteal phase defect (LPD) was defined as an endometrial biopsy having synchronous glands and stroma> 2 days out-of-phase when compared with the cycle day normalized to day of Vol. 61, No.4, April1994 The mean endometrial dating for all samples in stimulated cycles was day 24.5 ± 1.2 for the stroma and day 21.8 ± 2.4 for the glands (P < 0.0001). Examples of synchronous and asynchronous endometrial biopsies are illustrated in Figures 1 and 2. In 51 biopsies (57%), the dating of the stroma differed by >2 days when compared with the glands (day 24.6 ± 0.1 versus 20.2 ± 1.5; P < 0.0001). In contrast, in the nonstimulated cycles, the dating of the stroma did not differ when compared with the glands (day 25 ± 1.1 versus 23.9 ± 2.4, respectively; P = 0.1). Only 13% of biopsies showed gland-stroma dyssynchrony (P = 0.004 when compared with stimulated cycles). At the time of the endometrial biopsy, the patients had completed a mean of 3.3 ± 1.2 (range, 1 to 7 cycles) treatment cycles with superovulation and lui without conception. No difference was noted in the number of cycles undertaken by patients with or without gland-stroma dyssynchrony (3.2 ± 1.1 versus 3.4 ± 1.5, respectively; P = 0.6). Similarly, no Figure 1 Endometrial biopsy showing coordinated development of the glands and stroma (glands, day 26; stroma, day 26). Benadiva and Metzger Endometrial dyssynchrony with hmg 701

Figure 2 Endometrial biopsy showing dyssynchronous endometrium (glands, day 19; stroma, day 26) association was noted between the number of treatment cycles and the proportion of gland-stroma dyssynchrony (75%, 64%, and 66% for cycles 2, 3, and 4, respectively). When stimulated cycles associated with glandstroma dyssynchrony were compared with stimulated cycles associated with coordinated development of the glands and stroma, no significant differences were observed in E 2 level on the day of hcg, midluteal serum P, midfuteal E 2 level, or P:E 2 ratios (Table 1). In addition, the rate of luteal phase support with P was similar in cycles with and without gland-stroma dyssynchrony (45% versus 57%, respectively; P = 0.3). In stimulated cycles, women receiving luteal support experienced gland-stroma dyssynchrony as often as women without support (51% versus 67%; P = 0.1). When only the stromal dating was considered, eight women (8.9%) showed out-of-phase biopsies. In seven of eight biopsies, the stroma revealed retarded maturation, whereas only one showed a more advanced endometrial development. In nonstimulated cycles, all biopsies were in-phase. DISCUSSION The development of a receptive endometrium is a major factor determining the outcome of assisted reproductive technology treatments. The primary function of the endometrium is to provide a substrate for the implantation and maintenance of blastocyst development. During the follicular phase, the glands and stroma grow rapidly because of estrogen-stimulated proliferation. After ovulation, P produced by the corpus luteum ( CL) promotes secretory transformation and decidualization of the endometrium, which in turn provides essential support for implantation and maintenance of early pregnancy. The embryo arrives in the endometrial cavity on day 18 or 19, but implantation does not occur until days 20 to 21 at the height of glandular secretory activity (8). The study presented here investigated late luteal phase endometrial biopsies obtained in hyperstimulated cycles around the time of implantation. Our data demonstrate that in a large proportion of patients the glandular epithelium and stroma do not develop synchronously. These changes did not correlate with differences in serum preovulatory or midluteal E 2 and P levels or with the administration of exogenous P and occur significantly more often in stimulated cycles when compared with spontaneous cycles in the same patients. These findings suggest that the dyssynchrony observed in the development of the glands and stroma may represent individual variation in endometrial sensitivity to high circulating steroid hormone concentrations. An increasing amount of evidence suggests that the relative relationship between E 2 and P, rather than the serum P levels alone, determines the developmental potential of the endometrium. In a series of experiments in the castrated monkey model subjected to an artificial menstrual cycle and serial endometrial biopsies, Good and Moyer (9) were able to demonstrate that E 2 and P have different but complementary effects on the degree of gland and stroma differentiation and individually affect the temporal sequence of luteal phase events in the Table 1 Hormonal Parameters and Proportion of Luteal Phase Support in Stimulated Cycles With and Without Gland -Stroma Dyssynchrony* E 2 level on day of hcg (pg/ml)t Midluteal P (ng/ml) Midluteal E 2 level (pg/ml)t P:E 2 ratio Proportion with luteal phase support (%) Gland -stroma dyssynchrony (n = 51) 961.0 ± 535.0 22.2 ± 18.7 557.0 ± 594.0 89.0 ± 607.0 45 *Values are means± SD. t Conversion factor to SI unit. t, not significant. Conversion factor to SI unit. No dyssynchrony (n = 38) 956.0 ± 697.0 30.1 ± 32.5 526.0 ± 406.0 6.5 ± 49.0 57 p value t ~s 702 Benadiva and Metzger Endometrial dyssynchrony with hmg Fertility and Sterility

separate cell types. Luteal phase E2 primarily affects gland development, whereas P primarily affects the stroma. A normal secretory endometrium results when the effects of each of these hormones are balanced against the other. We have observed in our study that differences in gland-stroma development occurred in more than half of the women subjected to high levels of E2 and P resulting from superovulation. We hypothesize that cycles with gland-stroma dyssynchrony are associated with unbalanced sex steroid effect on the individual components of the endometrium. Abnormalities in endometrial development during spontaneous cycles are associated with infertility and early and repetitive miscarriage. The fact that the use of hmg in ovulatory women is associated with a high rate of abnormal endometrial development raises concern that this treatment may have detrimental effects on endometrial receptivity. Garcia et al. (3) examined endometrial biopsies from 21 patients, who did not undergo ET, 1 to 3 days after hcg administration. Eleven of the biopsies were "advanced," whereas the remainder were developed appropriately for the cycle day. Ben-Nun et al. (10) separately evaluated gland and stroma development in endometrial biopsies obtained 48 hours after ovum pick-up from women undergoing IVF. Whereas stromal development was consistently advanced, glandular dating lagged 2 to 3 days behind. The authors concluded that the relatively underdeveloped glands allowed the endometrium to "catch up" to embryo development, whereas the advanced stromal maturity may be of benefit in embryo nidation and development. However, most of these studies report on endometrial development during the early luteal phase. Our study demonstrates that when endometrial biopsies are obtained during the late luteal phase in patients undergoing ovarian hyperstimulation a significant proportion show dyssynchrony in the maturation of the glands relative to the stroma. These biopsies were obtained when patients failed to conceive after three to six treatment cycles. Although our study does not provide information on the characteristics of the endometrium in conception cycles, it is conceivable that the observed endometrial dyssynchrony may be associated with higher rates of unsuccessful implantation. The combination of retarded gland and advanced stroma development is similar to the type of endometrial morphology observed with exogenous hormonal replacement, which is associated with a high conception rate in women undergoing oocyte donation (11). However, the steroid hormone milieu in successful egg donation cycles differs substantially from that observed during ovarian hyperstimulation. We suggest that if gland development is retarded sufficiently, the implantation window may be shifted late enough in the luteal phase to make CL rescue by embryonic hcg difficult, resulting in nonconception cycles or very early pregnancy losses. In support of this hypothesis, Liu et al. (2) reported that 12.1 % of patients undergoing IVF experienced transient elevation in serial serum hcg levels late in the luteal phase. Moreover, in 14 of 36 patients with occult pregnancies, implantation occurred between days 28 and 31, based on the day that hcg was detected, thus indicating a delay in implantation. This delay in detection of hcg contrasts with successful pregnancies, in which hcg is first detected between days 22 and 27. Several observations have led to the current recommendations regarding luteal phase support after the administration ofhmg. First, in the absence of pregnancy, short luteal phase lengths are common, often <12 days, implying aberrant luteinization and inadequate CL life span. Ben-Nun et al. (10) demonstrated significantly higher pregnancy rates per ET cycles when P supplementation was started 2 days before follicular aspiration. In our study, approximately half ofthe patients received exogenous P supplementation. Although it is conceivable that this may have biased the results, luteal phase support did not correlate with the presence or absence of gland-stroma dyssynchrony, suggesting an individual response of the endometrium to the supraphysiologic hormone levels rather than a pharmacologic effect. Although the importance of this histologic finding is controversial, it is conceivable that improving the gland-stroma relationship in hyperstimulated cycles could have a positive impact on implantation rates. A better understanding ofthe cellular mechanisms involved in endometrial gland and stromal development is necessary before other alternatives to the widely accepted exogenous P supplementation can be recommended. In a review of the literature, Ben-Nun et al. (6) summarized the data on endometrial biopsies in IVF cycles. Of 198 endometrial biopsies performed around ET time in those studies, 99 (50%) show a delay in endometrial maturation. In our study, when only the stroma was considered, the incidence of LPD was only 8.9%. Nevertheless, if dating of the glands was used for evaluation of endometrial adequacy, the incidence of LPD increased to 59%, similar to the rate noted in their review. These find- Vol. 61, No.4, April 1994 Benadiva and Metzger Endometrial dyssynchrony with hmg 703

ings emphasize the importance of assessing both endometrial components when evaluating endometrial biopsies from stimulated cycles. In conclusion, our study demonstrates that a significant discrepancy between glands and stroma in the late luteal phase is a frequent finding in patients undergoing ovarian hyperstimulation with gonadotropins. The dyssynchrony between the maturation of the glands and the stroma appears to be independent of the midluteal E2 and P levels, the P:E2 ratio, the number of treatment cycles, or the administration of luteal phase support and probably reflects different degrees of responsiveness of an individual's endometrium rather than variations in the hormonal milieu. The potential effects of these findings on embryo implantation and the establishment of a viable pregnancy remain unclear. Efforts should be devoted to the creation of different treatment regimens to match the individual variability of response to hormonal stimulation. Acknowledgments. We acknowledge the valuable contribution of Linda Matonis, M.D., University of Connecticut, Farmington, Connecticut. REFERENCES 1. Paulson RJ, Sauer MV, Lobo RA. Embryo implantation after human in vitro fertilization: importance of endometrial receptivity. Fertil Steril 1990;53:870-4. 2. Liu H-C, Jones HW Jr, Rosenwaks Z. The efficiency ofhuman reproduction after in vitro fertilization and embryo transfer. Fertil Steril 1988;49:649-53. 3. Garcia JE, Acosta AA, Hsiu J-G, Jones HW Jr. Advanced endometrial maturation after ovulation induction with human menopausal gonadotropin/human chorionic gonadotropin for in vitro fertilization. Fertil SteriI1984;41:31-5. 4. Forman RG, Eychenne B, Nessmann C, Frydman R, Robel P. Assessing the early luteal phase in in vitro fertilization cycles: relationships between plasma steroids, endometrial receptors, and endometrial histology. Fertil Steril 1989;51: 310-6. 5. Metzger DA. Luteal phase following ovulation induction. Endometrial pathology and luteal support. Assist Reprod Rev 1992;2:62-77. 6. Ben-Nun I, Jafl'e R, Fejgin MD, Beyth Y. Therapeutic maturation of endometrium in in vitro fertilization and embryo transfer. Fertil Steril 1992;57:953-62. 7. Noyes RW, Hertig AT, Rock J. Dating the endometrial biopsy. Fertil SteriI1950;1:3-25. 8. Rosenwaks Z. Donor eggs: their application in modern reproductive technologies. Fertil Steril 1987;47:895-909. 9. Good RG, Moyer DL. Estrogen-progesterone relationships in the development of secretory endometrium. Fertil Steril 1968;19:37-49. 10. Ben-Nun I, Ghetler Y, Jaffe R, Siegal A, Kaneti H, Fejgin M. Effect of preovulatory progesterone administration on the endometrial maturation and implantation rate after in vitro fertilization and embryo transfer. Fertil Steri! 1990;53:276-81. 11. Navot D, Anderson TL, Droesch K, Scott RT, Kreiner D, Rosenwaks Z. Hormonal manipulation of endometrial maturation. J Clin Endocrinol Metab 1989;68:801-7. 704 Benadiva and Metzger Endometrial dyssynchrony with hmg Fertility and Sterility