Women with PCOS who underwent IVF/ICSI cycles using PPOS (hMG + MPA) or PPOS cotreatment with letrozole (hMG + MPA + LE) from January 2018 to March 2021 were enrolled. This study was approved by the Ethics Committee (Institutional Review Board) of Shanghai Ninth People’s Hospital. Patients met the following criteria: 1. Between 20 and 40 years of age; 2. Basal follicle-stimulating hormone (FSH) level <10 mIU/ml; and 3. No more than one previous nonviable embryo cycle. The diagnosis of PCOS was made according to the 2003 Rotterdam consensus (Rotterdam, 2004) and the recent guidelines for PCOS 2018 (Teede et al., 2018), whereby at least two out of three of the following criteria were met: 1) oligo- and/or anovulation; 2) biochemical and/or clinical evidence of hyperandrogenism; or 3) polycystic ovarian morphology on ultrasound. Where both oligo- or anovulation and hyperandrogenism were present, ultrasound was not necessary for diagnosis. Once diagnosed, its assessment and management covered reproductive, metabolic, and psychological features. The following diseases were excluded: hyperandrogenism and ovulatory dysfunction of other aetiologies, including congenital adrenal hyperplasia, androgen-secreting tumours, hyperprolactinemia and thyroid disease. Women who were on medications for conditions such as diabetes and hypertension were also excluded. A flowchart of the study is shown in Figure 1.

From menstrual cycle day 3 (MC3) to the trigger day, patients were administered 150–225 IU/d hMG (Anhui Fengyuan Pharmaceutical Co., China) via intramuscular injection and 4 mg/d MPA (Shanghai Xinyi Pharmaceutical Co., China) via oral administration. Letrozole (Jiangsu Hengrui Medicine Co., China) 2.5 mg/day was given from cycle day 3 onwards for only 5 days in the study group. The starting dose of hMG was 150 IU/day for patients with a high antral follicle count (>20) or slightly elevated basal FSH (7–10 mIU/ml), and a daily dose of 225 IU hMG was given to the other patients. For both groups, the hMG doses were adjusted depending on the number and size of the developing follicles on ultrasound as well as serum concentrations of sex hormones, including FSH, LH, oestradiol (E2) and progesterone (P4), administered beginning on MC8 and every 2–4 days thereafter. The final stage of oocyte maturation was triggered by the combination of 0.1 mg triptorelin (Decapeptyl; Ferring Pharmaceuticals, Germany) and 1000 IU hCG (Lizhu Pharmaceutical Trading Co., China) once the leading follicle exceeded 20 mm or at least three follicles exceeded 18 mm in diameter. Oocyte retrieval was performed 34–36 h after the trigger and guided by transvaginal ultrasound (TVS). All follicles with a diameter over 10 mm were aspirated (Kuang et al., 2015).

After oocyte retrieval, fertilization was carried out in vitro by either standard IVF or intracytoplasmic sperm injection (ICSI) according to the semen quality and previous fertilization conditions (Xu et al., 1998). Embryos were examined for the number or regularity of blastomeres and the degree of fragmentation. Top-quality embryos (including grade-1 and grade-2 6-cell embryos and above) were frozen by vitrification within 3 days after oocyte retrieval according to the criteria described by Cummins et al. (Cummins et al., 1986; Kuang et al., 2015). The non-top-quality embryos were placed in extended culture, and blastocysts with good morphology were also frozen on day 5 or 6 (Kuang et al., 2015).

Serum FSH, LH, E2 and P levels were measured on MC3, MC8, MC10-12, the trigger day and the day after trigger. Hormone concentrations were measured by means of chemiluminescence (Abbott Biologicals B.V, the Netherlands). The lower limits of sensitivity were as follows: FSH, 0.06 mIU/ml; LH, 0.09 mIU/ml; E2, 10 pg/ml; and P4 0.1 ng/ml. The upper limit for the E2 measurement was 5,000 pg/ml. If the E2 level on the trigger day or the next day was greater than the upper limit, it was recorded as 5,000 pg/ml without repeating the assay after sample dilution.

Endometrial preparation and FET were arranged for the second cycle after oocyte retrieval as previously described (Kuang et al., 2014; Kuang et al., 2015). Mild stimulation with letrozole was first recommended for FET, while hormone replacement therapy (HRT) was conducted for those who failed to become pregnant with mild stimulation cycles or with a history of a thin endometrium (endometrial thickness ≤6 mm). In the mild stimulation cycle, women were administered letrozole 2.5/5 mg for 5 days beginning at MC3. When the dominant follicle was ≥17 mm in diameter with an endometrial lining >8 mm, an E2 level >150 pg/ml and a P4 level <1 ng/ml, a bolus of urinary hCG (5000 IU) was injected for ovulation triggering. Progesterone commenced 2 and 3 days later, followed by day 3 embryo transfer 4 and 5 days later or blastocyst transfer 6 and 7 days later via abdominal ultrasound guidance.

Women who failed pregnant in the first FET cycle routinely underwent hysteroscopy and endometrial curettage, the uterine growth would be removed if they have and send pathology. For patients with intrauterine adhesions, decomposition of adhesions or placement of intrauterine contraceptive ring (IUD) were given according to the severity. If patients have had hysteroscopy within 6 months, the hysteroscopic procedure would not be repeated.

The primary outcome measure of this study was the hMG dose. The secondary measures included the durations of ovarian stimulation, the implantation rate in FET cycles, the number of oocytes retrieved, the number of viable embryos, profound LH suppression (a serum LH level less than 1 mIU/ml on the trigger day during ovarian stimulation (Liu et al., 2018). The oocyte retrieval rate was calculated as the total number of oocytes retrieved divided by the total number of follicles punctured; the mature oocyte rate was calculated as the total number of mature oocytes divided by the total number of oocytes retrieved; the fertilization rate was calculated as the total number of fertilized oocytes divided by the total number of mature oocytes; and the cleavage rate was calculated as the total number of cleaved embryos divided by the total number of fertilized oocytes. The clinical pregnancy rate, miscarriage rate, and ectopic pregnancy rate was recorded. The viable embryo rate per oocyte was defined as the number of viable embryos divided by the number of oocytes retrieved. The cycle cancellation rate was estimated based on the number of patients who had no viable embryos after complete oocyte retrieval. The implantation rate was calculated according to the number of gestational sacs divided by the number of embryos transferred. Clinical pregnancy was defined as the presence of at least one gestational sac with or without foetal heart activity on ultrasound examination at 4 weeks after FET, and the clinical pregnancy rate was the number of clinical pregnancies divided by the number of FET cycles. The miscarriage rate was the proportion of pregnancies that ended in spontaneous or therapeutic abortion.

A propensity score matching (PSM) model was established to balance differences in baseline characteristics between the two groups. To estimate the propensity score, we selected 10 covariates, namely, age; antral follicle count (AFC); basal levels of FSH, LH, E2 and P4; type of infertility (primary or secondary); infertility duration; previous IVF attempts (0, 1–2 or ≥3); and body mass index (BMI). Patients using MPA + hMG + LE were matched with the MPA + hMG group using the nearest-neighbour random matching algorithm at a ratio of 1:1. PSM was conducted using the R statistical programming language (version 4.0.3; R Foundation for Statistical Computing, Vienna, Austria).

The data were evaluated by Student’s t test for continuous variables with a normal distribution and the Mann–Whitney U test for continuous variables with a nonnormal distribution. These data are presented as the mean (standard deviation, SD). Pearson’s chi-square test or Fisher’s exact test was used for categorical variables (presented as %), as appropriate. Data analysis was performed with the Statistical Package for the Social Sciences (version 24, SPSS Inc.). Two-sided p < 0.05 was considered statistically significant.

Figure 1 shows a profile summary of the study. Briefly, a total of 2453 PCOS women who were candidates for assisted reproductive technology treatment from January 2018 to March 2021 were enrolled in this study, and 401 cycles were excluded as described in the Materials and Methods section. Of the remaining 2052 patients, 224 patients undergoing the MPA + hMG + LE protocol were matched with 224 patients who underwent the MPA + hMG protocol. All these patients completed oocyte retrieval cycles and succeeded in producing oocytes (range, 1–62), while 34 patients had no viable embryos. Among the remaining patients, 371 women completed FET cycles. No significant between-group differences were found in the postmatching analysis with regard to any baseline characteristics (all p > 0.05) (Table 1).

The MPA + hMG + LE and MPA + hMG groups had comparable numbers of oocytes retrieved (17.5 ± 9.16 vs 16.8 ± 10.29, p > 0.05) and viable embryos (5.48 ± 3.25 vs 5.37 ± 4, p > 0.05). The hMG doses and durations of ovarian stimulation ((1949.89 ± 725.03 IU vs 2017.41 ± 653.32 IU and 9.03 ± 1.79 days vs 9.21 ± 2.18 days, respectively, p > 0.05) were similar between the two groups. The number of follicles with diameters larger than 10 or 14 mm on the trigger day was significantly higher in the study group (p < 0.01). To our surprise, there was a significantly lower mature oocyte rate (76.74 vs 83.77%, p < 0.01) and fertilization rate (79.23 vs 81.48%, p < 0.05) in the study group than in the control group. The rates of cycle cancellation for nonviable embryos did not differ between the two groups. No premature LH surge or moderate to severe OHSS was observed during the study (Table 2).

The endocrine dynamics of FSH, LH, E2 and P4 during ovarian stimulation are presented in Figure 2. The FSH level increased dramatically after hMG administration for 5 days and then remained stable until the trigger day. After the dual trigger, the FSH level increased dramatically to nearly 20 mIU/ml. No significant differences were observed in FSH levels at any time point between the two groups (Figure 2A).

During the entire process of controlled ovarian stimulation (COS), LH in the two groups remained at a low level. However, there were differences in the change trend of LH between the two groups. LH in the control group showed a downwards trend. However, in the study group, LH remained basically unchanged in the first 5 days, followed by a downwards trend until the trigger day. The LH values were significantly higher in the study group than in the control group at each observation point during the COS (p < 0.01). In addition, significantly fewer patients experienced profound LH suppression in the study group (7.14%, 16/224) than in the control group (14.29%, 32/224, p < 0.05). None of the patients in either group experienced a premature LH surge (Figure 2B).

Serum E2 increased consistently with the development of multiple follicles and was significantly higher in the control group than in the study group at each observation point (p < 0.01) (Figure 2C).

In addition, the P level increased gradually during ovarian stimulation, especially on the day after trigger, in the two groups. On the trigger day and the day after, the p values in the study group were significantly higher than those in the control group (p < 0.01) (Figure 2D).

In this study, 382 women completed a total of 574 FET cycles: 189 patients completed 280 FET cycles in the MPA + hMG + LE group, and 193 patients completed 294 FET cycles in the MPA + hMG group. By the end of the study, 252 women had completed one FET cycle, 83 women had completed two FET cycles, 35 women had completed three FET cycles, nine women had completed four FET cycles, three women had completed five FET cycles, and the remaining 32 women did not complete FET cycles for personal reasons. A total of 970 embryos were thawed, and the rate of viable frozen-thawed embryos was 100%. The implantation rate in the study group (42.14%, 201/477) was significantly higher than that in the control group (34.69%, 171/493) (p < 0.05). The clinical pregnancy rate, miscarriage rate and ectopic pregnancy rate were comparable between the two groups (p > 0.05). The live birth rate and neonatal status require further statistical analysis (Table 3). The implantation rate in Table 3 represented the overall implantation rate and included all FET cycles associated with the IVF cycle. Similar calculations were used for other pregnancy outcome rates in this study. To minimize the impact of recurrent implantation failure and hysteroscopic surgery, we also analysed the implantation rate of first FET cycle in patients who underwent IVF for the first time. These supplementary data showed that the implantation rate in LE cotreatment group was significantly higher than that in the control group (57.4 vs 43.5%, p < 0.05) (Supplementary Table S1). To elucidate the effect of different endometrial preparation protocols on implantation rates during FET cycles, we performed sub-analyses of implantation rates for different endometrial preparation protocols in both groups. Our result showed nonsignificant difference of implantation rate among three types of endometrial preparation protocols in the two groups (Supplementary Table S2).

In our study, there was a waiting period between oocyte retrieval and embryo transfer, as we used the freeze-all strategy, and 1,462 extra embryos were not transferred; therefore, we cannot show the complete pregnancy rate data. Moreover, some women were less than 3 months pregnant, and some had not yet given birth; thus, we cannot provide data on ongoing pregnancy and live birth rates. Additionally, this retrospective study may have potential unknown or unmeasured covariates, which may lead to incomplete or inexact matching and thus lower the robustness of the study findings. Furthermore, 80% of PCOS patients in China weigh in the normal range (Yang et al., 2022), so our data showed that women with PCOS were thinner than those in other studies, leading our conclusion may not apply to all types of PCOS patients. Prospective randomized controlled studies with larger sample sizes and basic research should be performed in the future.