This retrospective study was carried out based on the clinical data of 144 patients who were reffered to the Shanghai JiAi Genetics & IVF Institute affiliated to Obstetrics and Gynecology Hospital of Fudan University between June 2012 and July 2018 for IVF treatment after fertility-sparing treatment of atypical endometrial hyperplasia (AEH) or early stage endometrioid endometrial cancer (EEC). After exclusion of: (1) patients with extra-endometrial malignant tumors, chromosomal abnormalities, or accompanying male infertility factors (n = 7), (2) patients who underwent fresh embryo transfer cycle or preimplantation genetic testing (PGT) cycle (n = 2) and (3) patients with recurrence of AEH/EEC within 3 months after ovulation induction or 1 month before FET (n = 5), 97 patients who underwent IVF treatment after fertility-sparing treatment of AEH (n = 74) or well-differentiated early-stage EEC (n = 23, International Federation of Gynecology and Obstetrics (FIGO) stage IA, without myometrial invasion and extrauterine metastases) were finally included in this study and their clinical and laboratory data were retrospectively analyzed. The overall design of the current study was shown in Figure 1. Pathological diagnosis of endometrial biopsy under hysteroscopy was confirmed in all patients in Obstetrics and Gynecology Hospital of Fudan University. According to the conventional guidelines in China, patients diagnosed with early-stage endometrioid endometrial cancer (EEC) or atypical endometrial hyperplasia (AEH) should undergo total hysterectomy under informed consent agreement. Therefore, before the start of IVF treatment, ethical approval was obtained from the Ethics Committees of Obstetrics and Gynecology Hospital of Fudan University (number JIAI E 2012-02). All patients were fully aware of the risks of conservative treatment and IVF treatment for AEH and EEC. All patients signed an informed consent to use their clinical and pathological data for research purpose. Details on each treatment approach and different protocols were as described below.

Conservative treatment and evaluation of therapeutic effects were described in a previous study (10). Briefly, megestrol acetate treatment (160 mg po qd) was initiated immediately after AEH or EEC diagnosis by hysteroscopy. Comprehensive evaluation was performed under hysteroscopy every 3 months during treatment till complete response.

Fertility treatment was initiated after confirmation of complete response by two consecutive hysteroscopic evaluations and endometrium biopsies in a 3-month interval. Women with tubal diseases underwent IVF treatment directly. Women who failed to conceive spontaneously within 12 months or after other infertility treatments like ovulation induction for 6 consecutive months or 2 consecutive artificial insemination failures were also offered IVF treatment. Women with extra-endometrial malignant tumors, chromosomal abnormalities, or accompanying male infertility factors were excluded. Fresh embryo transfer cycle and preimplantation genetic testing (PGT) cycle were also excluded. All patients underwent hysteroscopic endometrial biopsy within 3 months after ovulation induction and 1 month before FET to exclude the recurrence of AEH and EEC.

For all patients, endometrial hysteroscopic biopsy was taken prior to FET. Since we planned to follow up the recurrence rate of endometrial lesions after ovulation induction, the patients were subjected to the ovulation induction regimens with shorter treatment period, which included: PPOS, mild stimulation protocols, standard regimen and short agonist regimen. The appropriate regimen for ovarian stimulation was chosen by experienced doctors based on comprehensive consideration of each patient's basic hormone levels, BMI (body mass index), AFC (antral follicle count), drug compliance and financial affordability. The patients in PPOS protocol were administered 20 mg/d oral dydrogesterone (DYG) (Duphaston; Abbott Biologicals B.V., Netherlands) and human menopausal gonadotropin (HMG) (Lizhu Pharmaceutical Trading Co., Zhuhai, China) injection at a dose of 150–300 international unit (IU) daily from day 2/3 of the menstrual cycle to the day of trigger. Compared to medroxyprogesterone acetate (MPA, another progesterone in PPOS regimen), dydrogesterone showed a similar profile to natural progesterone (11); in light of this, dydrogesterone was used for patient treatment. The mild stimulation protocol was performed as follows: starting from day 3 of the menstrual cycle, patients received oral CC (Codal Synto Ltd., Cyprus) at 100 mg/d or LE (FuRui, Hengrui Ltd, China) at 10 mg/d and HMG/r-FSH (Gonal-f) (Merck Serono SA Aubonne Branch) injection at 150–300 IU daily. Standard protocol included short agonist regimen and antagonist regimen. Short agonist regimen was performed from day 2/3 of the menstrual cycle and consisted of an injection of 0.1 mg/day of GnRH-a (Decapeptyl, Ferring, Switzerland) which was followed by the injection of 150–300 IU/day of Gn from the next day. For the antagonistic regimen, the fixed scheme was adopted, and the antagonists-Cetrotide (Merck Serono SA Aubonne Branch) was administered on the 6th day of Gn use. In all protocols, the dosage of Gn was adjusted according to the level of estrogen and follicular growth.

Ovulation was induced with 5,000 to 10,000 IU of human chorionic gonadotropin (hCG) (Livzon/China) according to the patient's body weight and peak E₂ level when the diameter of at least one follicle was >18 mm (12). Oocyte retrieval was carried out 35 h after hCG trigger and conventional IVF was performed for most patients, and intracytoplasmic sperm injection (ICSI) was performed. Following the criteria judged by Cummins et al. (13), the symmetry or regularity of the blastomere, the cytoplasmic quality, and the fragmentary degree of embryo were checked on day 3 after oocyte retrieval, and all good-quality embryos (grade I and II 8-cell embryos) were cryopreserved by vitrification whereas the embryos in grade III and IV needed to be expanded by culture until they enter the blastocyst stage. Then, Gardner scoring was used to select eligible blastocysts (not <3BC grade) for vitrification on day 5/6 (14). Thawing was carried out on the day of the blastocyst transfer. The cryoloop ring was removed from the liquid nitrogen and placed in the air for 5 s before being immersed in recovery solutions. Next, the laser-assisted incubation of fully recovered blastocysts was performed and then the blastocysts were transferred to the blastocyst culture medium (Quinn's, SAGE, USA). A detailed description of FET procedure was described in our previous publications (15). In brief, the FET process was executed in a natural or induced menstrual cycle. The patient underwent routine ultrasound examinations and oral estradiol valerate (Sigma, St. Louis, Missouri, USA) on day 2 to day 4 of menstruation. The endometrium was monitored by ultrasound, and when its thickness reached 7 mm, 40 mg qd progesterone was injected intramuscularly. Embryo transfer was conducted under ultrasonic guidance, with only one or two embryos implanted per patient at a time. After transplantation, progesterone vaginal gel (Crinone, Merck Serono, Germany) was given to support the luteal phase.

Serum β-hCG test was performed on the 14th day after FET using the total β-hCG kit (Beckman Coulter/USA) to confirm the occurrence of pregnancy. Serum β-hCG higher than 100 mIU/mL indicated biochemical pregnancy and ultrasound was used to detect yolk sac 4 weeks after FET was defined as clinical pregnancy (16).

Since most of the pregnancies and recurrences occur in the first 2 years (3), our follow-up time was 24 months, which was initiated after the ovarian stimulation. None of the patients were lost during the 2-year follow-up, therefore; all 97 patients were included in the univariate or multivariate analysis of risk factors for the recurrence of endometrial disease. We retrospectively evaluated the basic information of each patient, clinical and laboratory embryo data, clinical pregnancy outcomes, and pathological report of hysteroscopy biopsy after treatment. The baseline characteristics of patient included age, anti-mullerian hormone (AMH), body mass index (BMI), homeostasis model assessment insulin resistance (HOMA-IR) index (calculated by fasting blood glucose (FBG; mmol/L) × fasting insulin (FINS; μU/mL)/22.5), lesion types, treatment time of endometrial lesions, and IVF interval time. The effectiveness of ovarian stimulation regimens was evaluated by the E₂ level on hCG trigger day, Gn dosage, duration of Gn treatment, the number of oocytes obtained, MII oocytes rate, fertility rate, good-quality embryo rate, FET cycle cancellation rate, thickness of endometrium, and pregnancy rate. The safety of ovarian stimulation regimens was assessed by comparing the recurrence rate of endometrial lesions.

Statistical analysis was performed using SPSS 25.0 software (IBM Corp., Armonk, NY, USA). Quantitative variables were expressed as means ± standard deviation and categorical variables were described using frequencies and proportions. For quantitative variables, the Shapiro-Wilk test and Levene test were used to evaluate their normal distribution and variance homogeneity, respectively. The Chi-squared test was used to analyze categorical variables. For quantitative variables, the Kruskal-Wallis test was used to analyze the differences among three or more groups, and comparison between two groups was executed by the Mann-Whitney-Wilcoxon test or unpaired t-test. Binary logistic regression was applied to explore the risk factors on the recurrence of endometrial diseases, and receiver operating characteristic (ROC) curve was used to evaluate the diagnostic accuracy. P < 0.05 was considered to be statistically significant. Before the retrospective analysis, sample size was calculated using Raosoft sample size calculator (http://www.raosoft.com/samplesize.html) to detect the minimum number of patients to be included. Assuming a confidence level of 90%, a margin of error of 5% and a response distribution of 50%, the minimum sample size was determined to be 89 eligible patients.

A total of 97 patients (23 cases diagnosed with early-stage EEC and 74 with AEH) who had undergone different ovulation stimulation regimens were included in this study. Only the first oocyte retrieval cycle of each patient was included in the retrospective analysis; we collected 37 cycles in PPOS protocol, 32 cycles in mild stimulation protocol (CC/LE+Gn), and 28 cycles in standard protocol. The basic information of patient characteristics in different ovulation stimulation regimens was shown in Table 1. There was no significant difference in age, HOMA-IR, lesion types, treatment time of endometrial lesions, and IVF interval time among the three different ovulation stimulation regimens (all P > 0.05, Table 1). The three groups significantly differed in AMH (P = 0.004) and BMI (P = 0.002). Specifically, the AMH of PPOS (P = 0.027) and standard regimen (P = 0.009) groups was higher than that of CC/LE+Gn regimen group, and the BMI decreased significantly in CC/LE+Gn regimen group when compared with the standard regimen groups (P = 0.018). However, no significant difference in baseline characteristics of patient was observed between the PPOS and standard regimen groups (Table 1).

As shown in Table 2, compared to the standard regimen group, the content of E₂ level on hCG trigger day (P = 0.003) and the number of oocytes retrieved (P < 0.001) were significantly lower in the CC/LE+Gn regimen group. Moreover, the results showed that compared to the standard regimen group (0.72 ± 0.31%), the good-quality embryo rate was significantly increased in PPOS regimen group (0.89 ± 0.26%) (P = 0.034, Table 2). In addition, the pregnancy rate (40.54%) of PPOS regimen group was significantly higher than that of the CC/LE+Gn regimen group (9.38%; P = 0.023), but the difference between PPOS and the standard regimen groups (21.43%) was not statistically significant (P = 0.259, Table 2). However, there was no significant difference in Gn dosage, duration of Gn treatment, MII oocytes rate, fertility rate, FET cycle cancellation rate, and recurrence rate among the different ovulation stimulation regimens (all P > 0.05, Table 2). These results indicated that patients in PPOS protocol had a better response to ovulation induction without increasing the recurrence rate.

As shown in Table 3, 18 cases were diagnosed with the recurrence of endometrial disease, and 79 cases were diagnosed with no recurrence of endometrial disease. Univariate analysis showed significant differences in age, treatment time of endometrial lesions, and duration of Gn treatment between the recurrence and non-recurrence groups (P < 0.05, Table 3). The age (33.94 ± 4.45 years) and treatment time of endometrial lesions (12.94 ± 6.93 months) in the recurrence group were significantly higher than those in the non-recurrence group (31.51 ± 3.68 years and 6.34 ± 4.61 months, respectively), while the duration of Gn treatment (7.94 ± 2.35 days) was significantly lower than that of the non-recurrence group (9.77 ± 2.58 days). No significant difference in other indicators such as AMH, BMI, HOMA-IR, ovulation induction regimens, lesion types, IVF interval time, E₂ level on hCG trigger day and Gn dosage was observed between the recurrence and non-recurrence groups (P > 0.05, Table 3).

We further performed multivariate logistic regression analysis to investigate the risk factors for the recurrence of endometrial disease. The results of unadjusted model showed that the age (aOR: 1.316, 95% CI: 1.06–1.633, and P = 0.013), treatment time of endometrial lesions (aOR: 1.282, 95% CI: 1.09–1.509, and P = 0.003), E₂ level on hCG trigger day (aOR: 1.001, 95% CI: 1–1.001, and P = 0.015) and duration of Gn treatment (aOR: 0.547, 95% CI: 0.319–0.939, and P = 0.029) were significantly associated with the endometrial disease recurrence (P < 0.05, Table 4). Nevertheless, there were no significant correlations between disease recurrence and other indicators such as AMH, BMI, HOMA-IR, lesion types, IVF interval time, ovulation induction regimens and Gn dosage (P > 0.05, Table 4). Furthermore, each parameter was conditionally filtered to obtain an adjusted model, including age, BMI, treatment time of endometrial lesions, E₂ level on hCG trigger day, and duration of Gn treatment. As shown in Table 4, in the adjusted model, the ovulation induction age (aOR: 1.228, 95% CI: 1.043–1.447, and P = 0.014) and treatment time of endometrial lesions (aOR: 1.236, 95% CI: 1.106–1.381, and P < 0.001) were significantly correlated with the recurrence of endometrial disease, and the established model had an area under ROC curve of 0.826 (P < 0.001).