This retrospective study was conducted at the Department of Assisted Reproduction at the Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine, a tertiary research and education hospital. Women with PCOS who underwent IVF/intracytoplasmic sperm injection (ICSI) cycles with stimulation of PPOS or short protocol from April 2014 to July 2019 were reviewed. Eligible participants were 20–40 years old with a history of infertility of over 1 year. PCOS was diagnosed according to the Rotterdam consensus by meeting two out of the three following criteria: (1) oligo- and/or anovulation; (2) biochemical and/or clinical evidence of hyperandrogenism; and (3) polycystic ovarian morphology on ultrasound (Rotterdam EA-SPCWG, 2004). Patients with other etiologies of ovulatory dysfunction and hyperandrogenism were excluded, such as hyperprolactinemia, congenital adrenal hyperplasia, androgen-secreting tumors, and thyroid disease.

Patients were excluded if they met one of the following criteria: (1) fresh transfer cycles; (2) age >40 years; (3) basal follicle-stimulating hormone (FSH) level ≥10 IU/L; (4) endometriosis grade 3 or higher; (5) history of ovarian surgery; (6) uterine anomalies; (7) history of recurrent spontaneous abortion; (8) abnormal chromosomal karyotype; (9) fetal reduction in the first frozen-thawed embryo transfer (FET) cycles; or (10) lost to follow-up. In cases of patients who underwent more than one IVF/ICSI cycle with the use of the same stimulation program, only the first IVF/ICSI cycle was included in the cohort group for propensity score matching (PSM).

This study was approved by the Institutional Review Board of the Shanghai Ninth People’s Hospital affiliated with Shanghai Jiao Tong University School of Medicine.

All patients were examined by transvaginal ultrasound screen and blood test to confirm the ovarian status and determine the baseline hormone profile on day 3 of the menstrual cycle (MC3). The PPOS protocol was administered as described previously (Wang et al., 2016; Dong et al., 2017). In brief, human menopausal gonadotropin (hMG) at a dose of 150–225 IU/day (Fengyuan Pharmaceutical Co., Maanshan, China) and oral medroxyprogesterone acetate (MPA) 10 mg/day (Xianju Pharmaceutical Co., Taizhou, China) were administered daily from MC3 until the trigger day (Dong et al., 2017). In the short protocol group, patients received 0.1 mg of triptorelin (Ferring International Center SA, Germany) starting on MC2 and 150–225 IU of hMG daily starting on MC3 (Wang et al., 2016; Zhu et al., 2016).

In both groups, serum hormone concentrations and the number and size of developing follicles were measured every 2–4 days by ultrasound and blood examination from MC8–9. The hMG dose was adjusted based on the individual ovarian response and the dynamics of FSH, LH, and other hormones. When three dominant follicles reached 18 mm in diameter, the final stage of oocyte maturation was induced.

For the PPOS group, most patients with PCOS (n = 243) were triggered with both 0.1 mg triptorelin and 1,000 IU human chorionic gonadotropin (hCG, Lizhu Pharmaceutical Trading Co., Shanghai, China). A few patients with PCOS received 0.1 mg triptorelin (n = 42) or 2,000–3,000 IU hCG (n = 19). In the short protocol group, all patients were triggered with 2,000–3,000 IU hCG.

Oocyte retrieval was conducted 34–37 h after the trigger, guided by transvaginal ultrasound. The MPA dose was maintained at the same level and continued up to the trigger day.

The aspirated oocytes were fertilized via IVF or ICSI according to semen parameters (Henkel and Schill, 2003). Oocyte insemination was performed following standard procedures for IVF/ICSI (Salha et al., 1998; Rienzi et al., 2011). All embryos were cultured in separate microdroplets of a continuous single culture medium (Irvine Scientific, Santa Ana, CA, United States). Embryos were scored according to Cummins’s standard on day 3; high-quality embryos, which were defined as grade I or grade II, were selected for vitrification (Cummins et al., 1986). Non-high-quality embryos were subjected to extended culture until the blastocyst stage. Blastocysts with good morphology (grade ≥3 BC) were selected for cryopreservation (Gardner et al., 2000).

The detailed protocol for the endometrial preparation has been previously described (Wang et al., 2016; Huang et al., 2019). The distributions of patients receiving different endometrium preparation methods were similar between the study group and control group after PSM. In our center, viable embryos defined as blastocysts with good morphology or high-quality embryos after cleavage were suitable for embryo transfer. FET was performed by skilled physicians with the guidance of abdominal ultrasound. When pregnancy was achieved, luteal-phase support was continued until 10 weeks of gestation.

The follow-up system at our center was previously described (Zhu et al., 2016; Huang et al., 2019). The primary end point was the implantation rate, and the secondary outcome was the live birth rate. Other measurements included total hMG dose; FSH, LH, estradiol (E2), and progesterone (P) on trigger day; FSH difference ratio; LH difference ratio; normal fertilization rate; cleavage rate; valid embryo rate per oocyte retrieved; and biochemical pregnancy, ectopic pregnancy, clinical pregnancy, and miscarriage events after FET.

The FSH difference ratio was calculated as the difference between FSH on the trigger day and FSH on MC3 divided by FSH on MC3. The LH difference ratio was calculated as the difference between LH on MC3 and LH on the trigger day divided by LH on MC3. Normal fertilization rate was defined as the number of normally fertilized oocytes divided by the number of total retrieved metaphase II stage oocytes. Cleavage rate was defined as the number of zygotes cleaved divided by the number of normally fertilized oocytes. The viable embryo rate per oocyte retrieved was defined as the number of viable embryos divided by the number of oocytes retrieved. The implantation rate was defined as the number of gestational sacs observed in the uterus (excluding gestational sacs in patients with ectopic pregnancy) divided by the number of embryos transferred. Clinical pregnancy was defined as at least one gestational sac with or without fetal heart activity at 7 weeks after FET. Miscarriage was defined as the loss of clinical pregnancy before the gestational week 24. Live birth was identified as the delivery of at least one live baby after at least 24 weeks gestation.

Statistical analyses and PSM were performed using Statistical Package for the Social Sciences version 25.0 software (SPSS Inc., Chicago, IL, United States).

The normality of quantitative variables was tested by the Kolmogorov–Smirnov tests, Shapiro–Wilk tests, histograms, and Q-Q plots. Data are presented as the mean ± SD or medians (first quartile, third quartile) as appropriate. Comparison of between-group differences was performed with Student’s t-test or Mann–Whitney U-test. For qualitative variables, the Chi-square test or Fisher exact test was used to analyze the differences, and data are presented as % (n/N).

A PSM model was established using logistic regression. To balance significant differences between the two groups, 14 covariates were selected into the PSM model to estimate the propensity score, such as age (continuous), body mass index (BMI) (continuous), duration of infertility (continuous), gravidity (continuous), parity (continuous), infertility type (primary or secondary) and diagnosis (PCOS + tubal factor, PCOS + male factor, PCOS + mixed factors, or PCOS only/other factors), basal endocrine profiles (all continuous), antral follicle count (continuous), insemination method (IVF, ICSI, IVF + ICSI), endometrial thickness on FET day (continuous), endometrial preparation (mild stimulation, hormone replacement therapy), and the number and stage of embryos transferred (1 or 2, cleavage or blast, respectively). Patients with PCOS that received short protocol were matched with patients in the PPOS group using the nearest-neighbor random matching algorithm at a ratio of 1:2. All P-values were based on two-sided tests, and P < 0.05 was considered to indicate statistical significance.

A flow chart illustrating the study design with details on patient selection is shown in Figure 1. Briefly, a total of 1,756 female patients with PCOS who underwent IVF/ICSI were screened from our database, such as 1,544 treated with PPOS and 212 treated with the short protocol (Supplementary Table 1). After PSM, 152 patients who underwent the short protocol were matched with 304 patients treated with the PPOS protocol, and these 456 patients represented the final study group. All 456 participants successfully completed at least one FET cycle after oocyte retrieval and freeze-all cycles. No significant differences were found between the PPOS and short protocol groups in post-matching analysis with regard to characteristics of patients and the first FET cycle, such as age, BMI, duration of infertility, gravidity, parity, infertility type and diagnosis, basal endocrine profiles, antral follicle count, insemination method, endometrial thickness on FET day, endometrial preparation and the number and stage of embryos transferred (Supplementary Table 2). The standardized differences before and after matching are shown in Supplementary Figure 1. To decrease the potential bias, the first FET cycles were included after matching the characteristics described above.

We compared the clinical outcomes of the total FET cycles from the 1,756 patients with PCOS (Supplementary Table 3) and found a higher implantation rate of embryos originating from patients receiving the PPOS protocol (42.2% vs. 31.4%, P < 0.05) and higher live birth rate per transfer (47.3% vs. 35.7%, P < 0.05, Supplementary Table 3) compared with patients treated with the short protocol.

In the comparison of the outcomes of the 456 matched patients, the implantation rate (43.4% vs. 31.9%), biochemical pregnancy rate (66.1% vs. 55.9%), and clinical pregnancy rate (61.8% vs. 47.4%) were all higher in the PPOS group than in the short protocol group (all P < 0.05, Table 1). These results indicate that the embryos from the PPOS group exhibited better implantation potential than those from the short protocol group. Notably, the live birth rate was also much better in the PPOS group (48.4% vs. 36.8%, P < 0.05). However, we observed a similar ectopic pregnancy rate (3.2% vs. 2.8%, P > 0.05) and miscarriage rate (18.6% vs. 19.4%, P > 0.05) between the two groups. In addition, the oocyte performance in the early developmental stage, such as the normal fertilization rate, cleavage rate, and valid embryo rate per oocyte retrieved, was similar between the two groups (P > 0.05, Supplementary Figure 2).

To identify the possible factors that may contribute to the difference in the embryo implantation potential between the PPOS and short protocol groups, we examined the hMG dose and hormone profile in stimulation cycles corresponding to the 842 embryos used in FET cycles. Compared with the short protocol group, the PPOS group used a higher dose of hMG (2031.88 ± 552.31 IU vs. 1709.38 ± 979.85 IU, P < 0.001, Figure 2A). Furthermore, the serum concentration of FSH increased markedly on the trigger day in the PPOS group (13.02 ± 4.01 IU/L vs. 9.93 ± 3.17 IU/L, P < 0.001, Figure 2B), and this may be from the higher dose of hMG used in the PPOS group. A higher FSH difference ratio was also observed in the PPOS group (1.66 ± 0.86 vs. 1.01 ± 1.20, P < 0.001, Figure 2F). However, the LH level on the trigger day was decreased significantly in the PPOS group (2.00 ± 1.51 vs. 3.28 ± 2.28, P < 0.001, Figure 2C), and a higher LH difference ratio was also observed (0.53 ± 0.27 vs. 0.15 ± 0.60, P < 0.001, Figure 2F). No significant difference was observed in E2 and P levels on the trigger day between the two groups (P > 0.05, Figures 2D,E).

To determine the relationship between the hMG dose and hormone levels with the implantation rate in the first FET cycle, we performed logistic regression analysis (Table 2). The independent variables included all factors that showed a significant difference between the two groups, such as total hMG dose, FSH on trigger day, LH on trigger day, FSH difference ratio, LH difference ratio, and ovulation trigger method (Figure 2). Logistic regression revealed a significant positive effect of the LH difference ratio on implantation success [P = 0.027, odds ratio (OR) = 1.861, 95% CI: 1.074–3.226]. The other independent variables did not reach statistical differences.

To further study the LH difference ratio, we performed receiver operating curve (ROC) analysis to observe the influence in embryo implantation (Figure 3A, P = 0.049, 95% CI: 0.500–0.609, cutoff value = 0.385), biochemical pregnancy (Figure 3B, P = 0.040, 95% CI: 0.501–0.615, cutoff value = 0.385), clinical pregnancy (Figure 3C, P = 0.016, 95% CI: 0.511–0.620, cutoff value = 0.385), and live birth (Figure 3D, P = 0.005, 95% CI: 0.525–0.631, cutoff value = 0.435).

Based on the cut-off value identified in ROC analysis, we divided the 456 patients with PCOS into two groups according to the LH difference ratio: the low ratio group (LH difference ratio <0.385) and the high ratio group (LH difference ratio ≥0.385). No significant difference in the endometrial thickness on FET day, endometrial preparation or the number or stage of embryos transferred was observed between the two groups (Supplementary Table 4).

We also evaluated the embryo implantation potential and subsequent clinical outcomes in the two groups and found that the implantation rate (43.0% vs. 33.2%), biochemical pregnancy rate (68.1% vs. 52.5%), clinical pregnancy rate (63.1% vs. 45.6%), and live birth rate (50.3% vs. 33.5%) were significantly higher in the high ratio group compared with the low ratio group (all P < 0.05, Table 3), indicating that patients with a high LH difference ratio showed improved outcome. Consistent with the previous analysis, the high ratio group had a higher percentage of patients treated with the PPOS protocol (PPOS and short protocol: 77.5% and 22.5%), and the low ratio group had a lower percentage (PPOS and short protocol: 46.2% vs. 53.8%), suggesting that the difference in the clinical outcome may be from the higher LH difference ratio induced by PPOS.