Introduction

Endometrial cancer (EC) and atypical endometrial hyperplasia (AEH) are occasionally detected in childbearing women [1]. However, since the standard treatment for EC and AEH is hysterectomy, patients diagnosed with these are unable to bear children after treatment. In these cases, high-dose medroxyprogesterone acetate (MPA) therapy can be used as a fertility-sparing treatment for young women with EC or AEH in their early stages [2]. Furthermore, repeated MPA therapy is effective for intrauterine recurrence [3]. Although MPA therapy is a fertility-sparing treatment, we have previously reported that many women needed infertility treatment to conceive thereafter, and disease recurrence, endometrial thickness during ovulation, and age of pregnancy permission influenced infertility [4]. Furthermore, perinatal outcomes are important after MPA therapy, but data regarding perinatal complications on these pregnancies remain limited. Since repeated MPA therapy is sufficiently effective for intrauterine recurrence [3], patients often undergo multiple dilatations and curettage (D&C) to evaluate its effects. Considering the association between D&C and perinatal complications, Ota et al. have reported that seven out of ten patients with placenta accreta or increta needed D&C because of artificial or spontaneous abortions [5]. Furthermore, D&C for miscarriage is reportedly a risk factor for abnormally invasive placentation, including placenta accreta, increta, and parcreta [6, 7]. Therefore, the development of placenta accreta in pregnancy after MPA therapy due to D&C has become a concern.

Considering this, this study aimed to investigate perinatal outcomes in pregnancy after MPA therapy for early stage EC and AEH, as well as to determine whether pregnancy after MPA therapy is associated with a higher risk of placenta accreta.

Materials and methods

Patients

Data of 51 pregnancies in 46 women who received MPA therapy for EC or AEH and delivered after 22 weeks of gestation at Keio University Hospital were retrospectively reviewed. Women with twin pregnancies were excluded from the study (n = 2). MPA therapy was administered as described in our previous report, and it was provided if the following indications applied: (1) diagnosis of AEH or endometrioid carcinoma (EM) G1 by D&C; (2) no myometrial invasion, cervical involvement, or extrauterine lesions diagnosed by magnetic resonance imaging (MRI) and computed tomography (CT); and (3) desire for childbearing [3]. However, three women with EM G2, who had a strong desire for childbearing, received MPA therapy in this study, because the ratio of the solid component was < 10% and nuclear atypia was weak. All patients underwent at least two D&Cs and/or hysteroscopic transcervical resection (TCR) before conception. In our hospital, they are not managed with any particular perinatal care because of pregnancy after MPA therapy. Therefore, the mode of delivery was decided based on obstetric conditions. However, we examined placental pathology after delivery to investigate recurrence. In this study, placenta accreta was defined as manual removal of the placenta, massive postpartum hemorrhage (blood loss > 500 g), or a retained placenta [8].

This study was approved by the Ethics Committee of Keio University School of Medicine (Nos. 20150103 and 20120243).

Statistical analysis

We performed a retrospective matched case–control study to determine the risk of placenta accreta in pregnancy after MPA therapy compared with a control group consisting of singleton pregnancies without any history of maternal malignancy treatments (e.g., MPA therapy, conization, and trachelectomy). The control group was recruited from the perinatal database between 2013 and 2019 at Keio University Hospital (n = 3988) by propensity score matching (PSM) using the JMP software (ver. 15, SAS Institute, Cary, NC, USA). The score was estimated using a logistic regression model, and greedy matching (ratio of 1:1 and matching without replacement) with a caliper width with a 0.20 standard deviation of the estimated logit was performed. Factors used for PSM were maternal age, parity, and in vitro fertilization and embryo transfer (IVF-ET). Finally, the balance of each covariate between cases and controls was evaluated using standardized differences (between-group difference/pooled standard deviation). An absolute standardized difference value < 10% was considered a relatively small imbalance.

Data are presented as the median (range) or number of cases (percentage). Continuous data were compared between groups using the Mann–Whitney U test. Categorical variables were analyzed using the Chi-square test or Fisher’s exact test. Odds ratios (ORs) and their 95% confidence intervals (CIs) for the association between the number of D&Cs and placenta accreta were evaluated using logistic regression analysis. All tests were statistically analyzed using JMP software (ver. 15, SAS Institute, Cary, NC), and statistical significance was established at p < 0.05.

Results

Oncologic characteristics in pregnant women after MPA therapy

The oncologic characteristics of the MPA group are presented in Table 1. Of the 46 patients, 21 were diagnosed with AEH, 22 with EM G1, and 3 with EM G2. The median number of MPA therapy sessions was 1 (range 1–5), and D&C was 3 (range 2–11). The median total duration to achieve complete response was 6 months (range 1.5–60). None of the patients experienced recurrence due to placental pathology.

Table 1 Oncologic characteristics in MPA therapy group (n = 46)
Full size table

Comparison of maternal and obstetrics characteristics between the MPA and control groups

A comparison of maternal and perinatal outcomes between the MPA (n = 51) and control (n = 51) groups is presented in Table 2. No placenta increta or percreta was recorded in either group. However, the incidence of placenta accreta was higher in the MPA group than in the control group (15.7 vs. 0%, p = 0.0058), while no differences in the other perinatal outcomes were observed between groups. While gestational weeks at delivery in the MPA group were later than those in the control group (p = 0.0058), no difference in the incidence of preterm delivery was observed between groups. Furthermore, no differences in birth weight, blood loss at delivery, or incidence of cesarean section (C-section) were observed between groups. In the MPA therapy group, six women developed preterm pregnancy due to preterm premature rupture of membrane (PROM) in breech presentation (n = 2), abnormal bleeding from placenta previa or low-lying placenta (n = 1), uterine rupture (n = 1), and hypertensive disorder of pregnancy (HDP) (n = 1). Moreover, in the MPA therapy group, two patients developed uterine rupture (preterm and term); prophylactic cerclage could not be removed in one patient, while she was in labor, and another delivered by vacuum-assisted vaginal delivery due to prolonged labor.

Table 2 Comparison of maternal and perinatal characteristics between MPA group and control group
Full size table

Comparison of maternal and perinatal characteristics in the MPA group between the placenta accreta and non-placenta accreta groups

A comparison of the maternal and perinatal characteristics between the placenta accreta and non-placenta accreta groups is presented in Table 3. No differences in maternal age at delivery, pre-pregnancy body mass index (BMI), incidence of nulliparity, IVF-ET, EC, placenta previa, number of MPA therapies, and D&C were observed between groups. However, the number of patients who received ≥ 6 D&Cs was higher in the placenta accreta group than in the non-placenta accreta group (50.0 vs. 14.0%, p = 0.018). Patients with ≥ 6 D&Cs demonstrated a 6.0-fold increased risk of placenta accreta (p = 0.043, 95% CI 1.05–34.1) than those who received ≤ 3 D&Cs.

Table 3 Comparison of maternal and perinatal characteristics in MPA group between placenta accreta group and non-placenta accreta group
Full size table

Discussion

Pregnancy after MPA therapy is associated with a high risk of placenta accreta. In particular, patients who underwent ≥ 6 D&Cs before conception were at a higher risk of placenta accreta than those who received ≤ 3 D&Cs (OR = 6.0). Since there is a paucity of data on perinatal outcomes of pregnancy after MPA therapy, we considered that our results could provide important information for clinicians and patients after MPA therapy.

When women with EC or AEH receive MPA therapy, some D&Cs should be performed to diagnose and evaluate the effects of treatment. However, an increased number of D&Cs in women with a history of abortion also increases the risk of placenta accreta [6, 7]. In previous reports on pregnancy after MPA therapy, only one small sample size has been reported to have D&Cs and placenta accreta, and no placenta accreta occurred among 10 patients after MPA therapy who underwent 3.5 D&Cs (median, range 3–9) [9]. Because our data were larger than those in this previous report, our information might be more useful. Moreover, because the diagnostic accuracy of each test varies, the risk should be assessed based on clinical information. Therefore, the frequency of D&Cs may be important in predicting the occurrence of placenta accreta in pregnancy after MPA therapy.

Although some clinicians are concerned regarding the recurrence of using hormonal therapy for conception, the safety of IVF-ET has been investigated in a previous report [10]. In our study, 23 patients (45.1%) conceived via IVF-ET, and none of them experienced recurrence. Furthermore, the incidence of preterm delivery after MPA therapy in our data was 11.8%, equal to that of previous data (8.6%) [11]. Although no difference in preterm delivery was identified between the MPA therapy and control groups in this study, the incidence of preterm delivery in the MPA therapy group was higher than that of the general Japanese population (5%). In our cases, preterm delivery with MPA therapy occurred because of preterm PROM in breech presentation (n = 2), abnormal bleeding from placental previa or low-lying placenta (n = 1), uterine rupture (n = 1), and HDP (n = 1). Nevertheless, we believe that no specific reason was attributable to preterm delivery during pregnancy after MPA therapy. Furthermore, Arendas et al. have reported that after MPA therapy, a pregnant woman voluntarily underwent a C-section for fear of placenta accreta and to avoid the risk of uterine rupture, although neither occurred [12]. From our data, uterine rupture was more frequent in the control group than in the MPA therapy group. However, further prospective research should be performed to determine this association when patients have a vaginal delivery after MPA therapy.

In our hospital, TCR has recently been performed to evaluate the effect of MPA therapy, since it has been previously reported as useful [13]. Although TCR for endometrial cancer could spread cancer cells through the fallopian tubes into a patient’s peritoneal cavity [14], according to a previous randomized controlled study, hysteroscopy for endometrial cancer did not increase the risk of the intraperitoneal transport of cancer cells [15]. When compared with D&C, TCR allows for the tumor to be removed quickly and effectively; Nevertheless, both TCR and D&C could minimize damage to the uterine muscle layer and endometrium and contribute to improved perinatal outcomes.

This study had some limitations. First, this was a retrospective study with a small sample size; however, this is the largest study on pregnancy after MPA therapy and PMS was used to evaluate obstetric outcomes. Thus, our results could be meaningful in the management of pregnant patients after MPA therapy. Second, approximately half of the patients who received MPA therapy were delivered at other hospitals. Since we were unable to obtain their detailed information, we evaluated perinatal outcomes in women who received MPA therapy at our hospital.

In conclusion, pregnancy after MPA therapy is associated with a high risk of placenta accreta. In cases with a high frequency of D&C, placenta accreta should be considered.