An analysis of endometrial biopsies performed for infertility

FERTILITY AND STERILITY Copyright" 1987 The American Fertility Society Vol. 48, No.5, November 1987 Printed in U.S.A. An analysis of endometrial biopsies performed for infertility Bert J. Davidson, M.D., Ph.D.*t Thais V. Thrasher, M.D.:!: Ibrahim M. Seraj, M.D.* Lama Linda University Sclwol of Medicine, Loma Linda, California The authors evaluated 774 endometrial biopsies that were performed for infertility. Complications arose in.6%. Lag of more than 2 days was found in 19%; luteal phase defect (LPD) was diagnosed in 5.7%. Most of the incidence of LPD can be predicted from chance occurrence. There was no association between abnormal biopsies and basal body temperature patterns, or between pathology, pregnancy outcome, and treatment. Exceptions included women with multiple spontaneous abortions and patients treated with clomiphene citrate (CC). An endometrial biopsy was performed in a pregnancy cycle in 4.0%, with an abortion rate not significantly different from the total study group. The authors conclude that an endometrial biopsy is relatively safe; however, the diagnostic and therapeutic consequences are limited. Endometrial biopsies may be useful only if performed in cases of habitual abortion or ovulation induction with CC. Fertil Steril 48:770, 1987 Even before Noyes et au described the criteria currently used to date an endometrial biopsy, investigators encountered women in whom the histology of the endometrium lagged behind the actual date of the menstrual cycle. In 1949, Jones 2 published an article in which she associated this finding with infertility, and the entity of luteal phase defect (LPD) became the topic of several hundreds of reports. Despite the attention given to this condition, its existence is still not fully accepted by some investigators, and there seems to be a wide variety of opinions as to the definition, incidence, pathophysiology, and treatment of LPD. In her infertility population, Jones 4 found LPD in.5%, whereas patients with habitual abortion had an incidence of approximately 5%~ Rosenberg et ap diagnosed Received December 29, 1986; revised and accepted July 15, 1987. * Department of Gynecology and Obstetrics. t Reprint requests: Bert J. Davidson, M.D., Ph.D. Department of Gynecology and Obstetrics, Loma Linda University School of Medicine, Loma Linda, California 9254. * Department of Pathology. 770 Davidson et al. Endometrial biopsies in infertility LPD in 2 of 96 infertility patients (8.1%), and Wentz 6 in 28 of 149 patients (19%). Factors possibly involved in this condition are abnormalities of the pituitary-hypothalamic axis, ovarian defects, hyperprolactinemia, endometrial progesterone (P) receptor defects, and iatrogenic causes, such as clomiphene citrate (CC) treatment and in vitro fertilization (IVF).7 Therapy for LPD is given according to the perceived abnormality, but often includes, among other, P, CC, and bromocriptine. Because there is a wide variety of factors involved and there are discrepancies of findings between different authors and institutions, we set out to evaluate the experience with endometrial biopsies in our infertility population in the last 10 years. We specifically addressed the risks, complications, and diagnostic and therapeutic implications of the procedure. MATERIALS AND METHODS All endometrial biopsies performed for infertility between 1975 and 1985 at Loma Linda University Medical Center were identified by systematic review of the pathology and clinic records. All slides

were reviewed independently by two of the authors who specialize in reproductive pathology. If their diagnosis differed, the slide was evaluated again by the two reviewers together. When necessary, new pathology preparations were made from the original paraffin blocks. The endometrial biopsy histology was read according to the standard criteria as described by Noyes et au The endometrial biopsies were performed by four reproductive endocrinologists. Specimens were obtained from the anterior portion of the fundus of the uterus. The material then was fixed in Bouin's or a 10% buffered formalin solution and stained with hemotoxylin-eosin. All patient histories were evaluated. Age, gravidity, parity, the presence or absence of abortions, cycle dates, complications ofthe biopsies, the completeness of the infertility workup, therapies, and subsequent pregnancies and their outcome were noted. An infertility workup was considered complete if, in addition to the endometrial biopsy, a semen analysis, a postcoital test, and a hysterosalpingogram and/or laparoscopy were performed. When available, basal body temperature (BBT) charts taken during endometrial biopsy cycles were reviewed independently by two of the authors and the results compared. The presence or absence of temperature elevations and the length of the luteal phase were evaluated. The patterns were graded as normal, equivocal, low-temperature elevation, and short luteal phase. Statistical analysis was performed using chisquare analysis or Fisher's exact test. RESULTS A total of 774 biopsies were recorded in 579 patients. Using standard criteria for a 28-day cycle, the median day of the biopsies evaluated was 26, with a range of day 2 to 28. Of these, 451 patients underwent one biopsy, 90 had two, and three or more biopsies were done in 8 patients. The mean and standard error of the mean age in these women was 28.5 ± 2 years, with a range of 18 to 4 years. There were 94 nulligravidas; 88 women had at least one pregnancy but no live births; and 97 patients had at least one child (Table 1). Patients were followed for 2 months to more than 7 years. Of the patients who underwent a complete evaluation, 40% conceived at least once. There were 28 complications (.6%) associated with the endometrial biopsy. They included "difficult biopsy" due to cervical stenosis (12), excessive Table 1 Patient Characteristics No. of patients No. of biopsies One biopsy Two biopsies Three or more biopsies Gravida 0 Gravida ;;0,: 1, para 0 Para;;o,: 1 774 579 451 (77%) 90 (16%) 8 (7%) 94 (68%) 88 (15%) 97 (17%) bleeding (5), fever (4), excessive pain (2), vasovagal reaction (2), uterine perforation (2), and interrupted pregnancy (1). Except for the last complication, none of them was serious or longstanding. One of the patients with fever after biopsy developed clinical signs of pelvic inflammatory disease. She delivered subsequently, as did two other women with fever. The women with fever who did not become pregnant was found to have bacterial cystitis. In the review of the biopsy slides, there was a difference of 2 days or more from the original interpretation in 16%, leading to a possible missed diagnosis of LPD in 18 patients and an unnecessary repeat biopsy in 15 who had an incorrect diagnosis of LPD. There was a lack of a definitive diagnosis in 12 others. The biopsy findings are listed in Table 2. The histologic findings lagged behind the actual date of the menstrual cycle by more than 2 days in 149 biopsies (19.%). In patients (5.7%), there were two who had subsequent lagging biopsies, and they were considered to have LPD. Most patients with this diagnosis who were treated were given CC; a minority received P suppositories or a combination of the two. In the patients with presumed LPD, there were 18 pregnancies. Of these, 12 were delivered at term; 6 ended in spontaneous abortion. There was no difference between the outcome of the pregnancies whether the patients were or were not treated (Table ). Three patients who delivered at term after treatment had another pregnancy, which again ended at term. However, in the second pregnancy they did not receive any treatment. In Table 2 Biopsy Findings Lag of more than 2 days Two lagging biopsies (patients) Uneven maturation Dyssynchrony Endometritis Atypical adenomatous hyperplasia Adenocarcinoma 149 (19.%) (5.7%) (4.%) 9(1.2%) 1 (1.7%) 2 (0.%) 2 (0.%) Vo!' 48, No.5, November 1987 Davidson et al. Endometrial biopsies in infertility 77l

Table Treatment No treatment Outcome of Pregnancy" After Treatment of LPD X 2 = 0.0281; P = NS. Term delivery 5 7 Spontaneous abortion 19 patients who were treated for LPD, a third biopsy was available for review. Less than half of these patients showed resolution of the LPD. There was no difference in the occurrence of pregnancy and its outcome, whether or not the LPD was resolved (Table 4). Other biopsy findings included uneven maturation (), dyssynchrony (9), endometritis (1), atypical adenomatous hyperplasia (2), and adenocarcinoma (2). There were 25 patients (4.%) who had suffered two or more spontaneous miscarriages; LPD was diagnosed in 7 of them. All 7 patients were treated, and 6 of the 7 achieved pregnancies and delivered at term. Because of ovulatory defects, 69 patients were being treated with CC at the time of their initial endometrial biopsy, and 25 of them (6%) had endometrial histology lagging the cycle date for more than 2 days. This was significantly increased as compared with the total study group (P < 0.02). The presence of uneven maturation and dyssynchrony did not correlate with lagging endometrial histology. These two findings also were never found again in a subsequent endometrial biopsy. Generalized endometritis was found in nine patients, five of whom became pregnant, four of whom delivered at term. Localized endometritis was seen in four patients, one of which delivered. All patients were treated with tetracycline. In 22 menstrual cycles in which an endometrial biopsy was performed, a BBT chart could be evaluated. There was no correlation evident between abnormalities such as short luteal phase and low temperature elevation and the biopsy interpretation. An endometrial biopsy was performed in a preg- Table 4 Correction of LPD and Occurrence and Outcome of Pregnancy" LPD corrected 4 LPD not corrected 5 X2 = 0.4598; P = NS. No Term Spontaneous pregnancy delivery abortion 2 2 nancy cycle in 2 patients (4.0%). Excluding the 1 case in which the biopsy site was removed, there was a 17% spontaneous abortion rate. This was not significantly different from the 14% incidence in the total study group. There were 24 pregnancies in the cycle subsequent to the biopsy. This incidence was not different from that in the endometrial biopsy cycles. There were 7 spontaneous miscarriages in these 24 pregnancies (29%). That proportion was significantly increased over the incidence in the total study group, with a P level of <0.05 (Table 5). DISCUSSION From this study it can be concluded that, in general, an endometrial biopsy carries little risk. Most complications are immediate and short-lived, al~ though the implantation site was removed in one instance. In cases of pregnancy, performing an endometrial biopsy does not appear to affect the spontaneous abortion rate. This observation confirms what is reported by several other authors.8-10 The 17% incidence of spontaneous abortion in the patients who had an endometrial biopsy performed in a pregnancy cycle was not different from the 14% in the total study group, and both incidences are within the range reported in a review of spontaneous abortions in infertility patients by JansenY To our knowledge, the question regarding whether performing an endometrial biopsy increases immediate, "silent" abortions or, conversely, increases the chance of implantation by endometrial stimulation has not been addressed in normal menstrual cycles. The latter possibility of the creation of a "bed for implantation" was sug- Table 5 Pregnancies, Spontaneous Abortions, and Endometrial Biopsies' n % Patients pregnant in biopsy cycle 2 4 Spontaneous abortions after biopsy 4/2 17 Spontaneous abortions in total group 28/201 14 Pregnancies in subsequent cycle 24 4 Spontaneous abortions in subsequent cycle 7124 29 P NS NS <0.05 Incidences of pregnancies and spontaneous abortions in the biopsy and the subsequent cycle are compared with the total study group. 772 Davidson et al. Endometrial biopsies in infertility

gested by Karow et al. 9 on a theoretical basis, quoting animal data. We approached this question by reviewing carefully the incidence of pregnancies in the menstrual cycle subsequent to the endometrial biopsy, and found a similar occurence of pregnancies, indicating no direct or indirect positive or negative effect. There was an actual increase of spontaneous abortions in the cycle following the endometrial biopsy, suggesting lasting effects on implantation. These results are in contrast to what was reported by Jacobson and MarshallP They found a similar rate of spontaneous abortions in the patients who underwent biopsy during pregnancy, but those patients had a significantly lower pregnancy rate. All patients studied, however, received induction of ovulation with human menopausal gonadotropin combined with human chorionic gonadotropin, and the decreased pregnancy incidence may be related to this therapeutic regimen. In many institutions, the endometrial biopsy is interpreted by general pathologists, many of whom lack experience or interest in gynecologic pathology. In our hospital, all biopsies are originally read by the pathologist on call for the day. Despite the rigid criteria used to evaluate the biopsy, the interpretation may be fairly difficult because of subtle daily changes, variations in fixation, and quantity of biopsy material. In view of these considerations, all biopsies were re-evaluated. In our study, there was a clinically significant difference in 16% with the original interpretation, leading to a missed diagnosis in 45 patients. Obviously, having an experienced and interested pathologist read all endometrial specimens will add to the accuracy of the diagnoses. The finding of a biopsy lagging the actual day of the menstrual cycle by more than 2 days is a frequent event in our experience and in many other series. 5-7 The 19.% incidence suggests that, on theoretical statistical grounds, one can expect to find two consecutive lagging biopsies in.6% (0.19 X 0.19) of all patients. By definition, this constitutes the diagnosis of LPD. This was statistically different from the 5.7% observed in our study population with a P level of <0.05. Nevertheless, it means that the majority of women with a diagnosis of LPD do not have the abnormality. In 1976, Jones4 found LPD as an etiologic factor in approximately.5% of het infertility cases, exactly the incidence of the statistical chance occurrence, as Driessen et al.a noted in their review. This high incidence of the diagnosis of LPD without clinical significance could explain, in part, that there was no correlation between the diagnosis, presence, absence, or success of treatment and the occurrence and outcome of pregnancy in our series. This gross inaccuracy should be discussed with the infertile patient before suggesting an initial or additional endometrial biopsy and treating a supposed abnormal luteal phase. The exception to this is the woman with two or more spontaneous miscarriages, in which 28% had a diagnosis of LPD. This incidence is within the range quoted by other authors.4 We found a highly significant correlation between the women with LPD in this group and successful outcome of pregnancy after treatment. This finding confirms a recent report by Balasch et al.1a The proportion of women with habitual abortions in a population studied for LPD may have an important effect on the result of such a study. It is suggested that administration of CC, in itself a medication used to treat LPD, can induce this abnormality.7 In this series, the incidence of lagging endometrial histology in patients who underwent their initial biopsy while receiving ovulation induction with CC was almost double that of the total study group. This suggests that an evaluation of the endometrium is an acceptable procedure in the workup of these patients, once appropriate menstrual intervals are achieved. Dyssynchrony and uneven maturation are two intriguing findings, encountered occasionally while evaluating an endometrial biopsy. Dyssynchrony is diagnosed when there is a difference in maturation between the stromal and the glandular components of the endometrium. In uneven maturation, the endometrium exhibits different stages of maturation in different areas of the tissue. There was no correlation between these two findings and LPD, and the conditions were never found at a subsequent biopsy, suggesting that these abnormalities have no clinical importance. In a small percentage of the biopsies, endometritis was found. The severity of this condition varied from focal infiltrates to full-blown general infections. The numbers were too small to draw any conclusion about the importance of endometritis in infertility, but pregnancy and spontaneous abortion rates were in the general range, and the histology improved uniformly after antibiotic therapy. BBT charting has been used in the evaluation of the luteal phase for two purposes. First, the rise in Vol. 48, No.5, November 1987 Davidson et ai. Endometrial biopsies in infertility 77

temperature can be used to predict the appropriate time for an endometrial biopsy, 1 or 2 days before the onset of menses. Second, many practitioners also use the temperature pattern in the diagnosis of LPD. Two patterns are distinguished that are thought to have diagnostic value. The "short" luteal phase is the condition in which menses ensues 11 days or fewer after the temperature rise. The other pattern exhibits a lack of maintenance of the thermogenic shift of at least 0.4 e in the luteal phase. Some physicians consider either of these findings enough evidence that LPD is present and treat patients accordingly. We reviewed all biopsy cycles in which a complete BBT chart was available and found no correlation between the two presumed abnormal patterns and the absence or presence oflpd. This inconsistency also was described by Annos et a1. 14 in a small series of women with a diagnosis of LPD. We conclude that the usefulness of an endometrial biopsy in the evaluation of female infertility is still under discussion and may depend upon the individual institution. Because of the variety of diagnostic and therapeutic consequences and the reportedly high incidence of LPD and the success of treatment in some studies, many practitioners perform endometrial biopsies as a routine, as was done in our institution. A review of one's own practice, as was performed in this study, may define the utility and indications of endometrial biopsies. In our study, the yield and therapeutic consequences appear to be of importance only in women with repeated spontaneous miscarriages or ovulation induction with ee. In other cases, the finding of LPD may not have clinical significance, and lack of conception or unsuccessful pregnancy after treatment mandates the continuation of the infertility workup. REFERENCES 1. Noyes RW, Hertig AT, Rock J: Dating the endometrial biopsy. Fertil Sterill:, 1950 2. Jones GES: Some newer aspects of the management of infertility. JAMA 141:112, 1949. Driessen F, Holwerda PJ, vd Putte SCJ, Kremer J: The significance of dating an endometrial biopsy for the prognosis of the infertile couple. Int J Fertil 25:112, 1980 4. Jones GS: The luteal phase defect. Fertil Steril27:51, 1976 5. Rosenberg SM, Luciano AA, Riddick DH: The luteal phase defect: the relative frequency of, and encouraging response to, treatment with vaginal progesterone. Fertil Steril4:17, 1980 6. Wentz AC: Endometrial biopsy in the evaluation of infertility. Fertil Steril :121, 1980 7. Andrews WC: Luteal phase defects. Fertil Steril 2:501, 1979 8. Buxton CL, Olson LE: Endometrial biopsy inadvertently taken during conception cycle. Am J Obstet Gynecol 105:702, 1969 9. Karow WG, Gentry WC, Skeels RF, Payne SA: Endometrial biopsy in the luteal phase of the cycle of conception. Fertil Steril 22:482, 1971 10. Wentz AC, Herbert CM III, Maxson WS, Hill GA, Pittaway DE: Cycle of conception endometrial biopsy. Fertil Steril 46:196, 1986 11. Jansen RPS: Spontaneous abortion incidence in the treatment of infertility. Am J Obstet GynecolI4:451, 1982 12. Jacobson A, Marshall JR: Detrimental effect of endometrial biopsies on pregnancy rate following human menopausal gonadotropin/human chorionic gonadotropin-induced ovulation. Fertil Steril :602, 1980 1. Balasch J, Creus M, Marquez M, Burzaco I, Vanrell JA: The significance of luteal phase deficiency on fertility: a diagnostic and therapeutic approach. Hum Reprod 1:145,1986 14. Annos T, Thompson IE, Taymor ML: Luteal phase defect: difficulties encountered in diagnosis and treatment. Obstet Gynecol 55:705, 1980 774 Davidson et ai. Endometrial biopsies in infertility