Contributing factors for pregnancy outcomes in women with PCOS after their first FET treatment: a retrospective cohort study

Abstract

Objective

We aim to explore the contributing factors of clinical pregnancy outcomes in PCOS patients undergoing their first FET treatment.

Methods

A retrospective analysis was conducted on 574 PCOS patients undergoing their first FET treatment at a private fertility center from January 2018 to December 2021.

Results

During the first FET cycle of PCOS patients, progesterone levels (aOR 0.109, 95% CI 0.018-0.670) and endometrial thickness (EMT) (aOR 1.126, 95% CI 1.043-1.419) on the hCG trigger day were associated with the clinical pregnancy rate. Similarly, progesterone levels (aOR 0.055, 95% CI 0.007-0.420) and EMT (aOR 1.179, 95% CI 1.011-1.376) on the hCG trigger day were associated with the live birth rate. In addition, AFC (aOR 1.179, 95% CI 1.011-1.376) was found to be a risk factor for preterm delivery.

Conclusions

In women with PCOS undergoing their first FET, lower progesterone levels and higher EMT on hCG trigger day were associated with clinical pregnancy and live birth, and AFC was a risk factor for preterm delivery. During FET treatment, paying attention to the patient’s endocrine indicators and follicle status may have a positive effect on predicting and improving the pregnancy outcome of PCOS patients.

Keywords:

• Polycystic ovary syndrome
• frozen embryo transfer
• IVF
• pregnancy outcomes
• contributing factor

Introduction

Polycystic ovary syndrome (PCOS) is the most common reproductive and endocrine disease characterized by hyperandrogenism, oligo- or anovulation, and polycystic ovaries [1, 2]. It affects about 6-21% of women of reproductive age and is the main cause of anovulatory infertility, and up to 72% of women with PCOS experience infertility due to anovulation [3–10]. Previous studies showed that women with PCOS were at a higher risk of adverse pregnancy outcomes and pregnancy complications, such as miscarriage, gestational diabetes mellitus, preterm delivery, etc [11–15].

In vitro fertilization (IVF) is an important treatment for infertile patients with PCOS to improve their clinical pregnancy outcomes. However, women with PCOS are prone to develop ovarian hyperstimulation syndrome (OHSS) during controlled ovarian hyperstimulation. It was reported that frozen embryo transfer (FET) can reduce the risk of OHSS and improve implantation rate and live birth rate than fresh-embryo transfer in patients with PCOS [16–18]. Moreover, with the development of vitrification and single embryo transfer, the use of FET for PCOS patients is increasingly accepted by fertility clinicians and patients [19]. It’s important to know which patients with PCOS undergoing FET cycles have a higher risk of adverse pregnancy outcomes and warrant additional treatment. Previous studies have reported the effect of obesity [20, 21], high anti-mullerian hormone (AMH) [22, 23] on pregnancy outcomes in patients with PCOS undergoing FET. However, to our knowledge, the factors influencing pregnancy outcome at the first FET cycle in patients with PCOS are still unclear and require further exploration.

Therefore, this study is a retrospective analysis of pregnancy outcomes in patients with PCOS who underwent their first frozen-embryo treatment, with the aim of identifying the essential factors affecting the pregnancy outcomes in the first FET cycle and providing more detailed consultation information for PCOS patients undergoing assisted reproductive technology (ART).

Methods

Study population

PCOS patients who underwent their first IVF treatment at Shenzhen Zhongshan Urology Hospital (SZUH) from January 2018 to December 2021 were evaluated. The Institutional Review Board of the Reproductive Research Ethics Committee of Shenzhen Zhongshan Urology Hospital approved this study (Approval number: SZZSECHU-20180030). Patient informed consent was waived as this was a retrospective non-interventional study.

The inclusion criteria included: (1) patients with PCOS diagnosed according to the Rotterdam criteria [24]; (2) infertile women aged 20 to 40 years old; (3) fertilization by IVF or intracytoplasmic single sperm injection (ICSI); (4) patients who did not undergo fresh embryo transfer for preventing ovarian hyperstimulation, inappropriate endometrium, high progesterone concentration on the human chorionic gonadotropin (hCG) trigger day or some complications, and underwent frozen-embryo transfer in our hospital; (5) patients who received the single day 5 blastocyst stage embryo transfer; (6) patients undergoing hormone replacement treatment (HRT), and antagonist protocol was used for ovarian stimulation. The exclusion criteria were as follows: (1) patients with Cushing’s syndrome, congenital adrenal hyperplasia, androgen-secreting tumors, atypical adrenal hyperplasia, hyperprolactinemia, thyroid dysfunction and other conditions that cause hyperandrogenism and ovulatory dysfunction; (2) women with metabolic syndrome and cardiovascular disease; (3) patients with congenital or acquired uterine anomalies; (4) chromosomal abnormalities in either member of the couple.

Study procedures

To ensure comparability of patients, we included only patients with PCOS treated with the antagonist protocol, the replacement cycle, had day 5 single blastocyst transferred during FET, and had not received embryo transfers before.

Ovarian stimulation

After assessing the patient’s ovarian reserve and responsiveness, an antagonist protocol was used for ovulation induction. Follicles were monitored by serial transvaginal sonography and sex hormone measurements. Gonadotropin (Gn) was injected on the second day of menstruation, and when follicles were more than 12 mm in diameter, gonadotropin-releasing hormone (GnRH) antagonist was used until the hCG trigger day. When at least two follicles were 17-18 mm in diameter, hCG was injected to promote oocyte maturation, and oocytes were retrieved 34-36 h after hCG injection guided by transvaginal ultrasound.

Fertilization and cultivation

After 4-6 h of oocytes retrieval, oocytes were fertilized by IVF or ICSI, depending on the semen parameters of the male partner, such as sperm concentration, viability and morphology. In all our included patients, the embryos were cultured to the blastocyst stage on day 5 and the quality of the embryos was assessed based on Gardner’s grading system [25]. Embryos with many cells and tightly packed were considered to be of high quality. Fresh transfers were cancelled in all the enrolled patients for some reason, so the embryos were preserved by vitrification for later frozen transfers. Laboratory procedures were performed by 4 trained embryologists, numbered A, B, C, and D respectively for the convenience of subsequent analyses, each with more than 5 years of laboratory experience.

Cryopreservation and thawing

The vitrification method was used for cryopreservation. The blastocysts were first immersed in equilibration medium consisting of 7.5% dimethylsulfoxide (DMSO), 7.5% ethylene glycol (EG), and 5% (m/v) human serum albumin (HSA) (SAGE BioPharma, USA) for 10 min. Next, they were transferred to a vitrification medium for 45-60 s, which contained 0.5 mol/L (M) sucrose, 15% (v/v) DMSO, 15% (v/v) EG and 5% (m/v) HSA. Regarding thawing, the blastocysts were immersed into 1 M, 0.5 M and 0 M sucrose containing 5% (m/v) HSA sequentially for 1 min, 3 min and 6 min, respectively. Only the first step of thawing was performed at 37 °C, the rest were operated at room temperature. Blastocysts were subjected to laser-assisted incubation (ZILOS-tk laser, Hamilton Thorne, US) immediately after thawing and transferred 2-3 h later under ultrasound guidance.

Endometrial preparation

All of the patients we included were treated with hormone replacement cycle, with endometrial preparation starting on day 2-3 of the menstrual cycle with 4 mg of oral estradiol valerate (Progynova; Bayer, Germany) per day, increasing by 2 mg every 5 days. Estrogen therapy was given for 15 days. Before embryo transfer, 60 mg progesterone (ZheJiang XianJu Pharmaceuticals, China) was injected intramuscularly daily for 4/6 days.

Luteal phase support

From the day of embryo transfer, 20 mg progesterone tablets (Duphaston; Abbott, Netherlands) were supplied orally twice daily and 90 mg progesterone gel (Crinone; Merck, Germany) was administered vaginally once daily. β-hCG was tested 11 days after transfer and a value greater than 5 IU/mL was considered a positive biochemical pregnancy, and luteal support continued until 12 weeks of gestation after confirmation of pregnancy.

Outcome measurement

Hormones, such as serum testosterone (T), prolactin (PRL), follicle-stimulating hormone (FSH), luteinizing hormone (LH), estradiol (E2), progesterone (P), AMH and thyroid stimulating hormone (TSH) were carried out on Cobas e601 (Roche Diagnostics, Germany) by chemiluminescence. Fasting plasma glucose (FPG), fasting insulin (FINS) and other biochemical parameters were measured using Cobas c501 autoanalyzer (Roche Diagnostics, Germany). Antral follicle count (AFC) is defined as the number of small follicles with a diameter of about 2-9 mm in both ovaries on the second day of a woman’s menstrual cycle. The insulin sensitivity was evaluated using the quantitative insulin sensitivity check index (QUICKI), QUICKI = 1/(Log FPG + Log FINS). QUICKI combines information of FPG and FINS levels has a good correlation with the gold standard for assessing insulin sensitivity, the glucose clamp technique. Previous studies have shown that the QUICKI index below 0.357 suggests a diagnosis of insulin resistance [26].

Clinical pregnancy was confirmed by the presence of a gestational sac and primitive heart tube pulsations through vaginal ultrasound at 7-10 weeks of gestation. Clinical pregnancy rate referred to the proportion of the number of clinical pregnancy cycles to the number of transplant cycles. Live birth was defined as a baby born alive after 28 weeks of gestation and live birth rate was calculated by the number of live birth cycles divided by the number of embryo transferred cycles. Miscarriage was the termination of a pregnancy at less than 28 weeks’ gestation and miscarriage rate equaled the number of miscarriage cycles divided by the number of clinical pregnancy cycles. Preterm delivery was defined as birth between 28 and 37 weeks of gestation and preterm delivery rate was calculated as the proportion of preterm birth cycles to the number of live births.

Statistical analysis

SPSS software (version 26.0; SPSS Inc.) was used for statistical analysis. Kolmogorov-Smirnov method was applied to test the normality of continuous variables. The data fitting the normal distribution were expressed as the mean ± standard deviation, the data not fitting the normal distribution were expressed as the median (interquartile range, IQR), and the rate or percentage was used to describe categorical variables. Two independent sample t test, Mann-Whitney U test or Pearson chi-square test was used to analyze the differences between the two groups. Binary logistic regression model was used to analyze the related factors affecting the pregnancy outcome of PCOS patients, and the odds ratio (OR) and 95% confidence interval (CI) were calculated. Variables with P value less than 0.20 in univariate analysis and variables with a potential clinical association were included in multivariate logistic regression. A P-value less than 0.05 was considered statistically significant.

Results

A total of 574 patients with PCOS were enrolled in the study, of whom 397 achieved clinical pregnancy, 326 resulted in live births, 71 experienced miscarriages, and 53 had a preterm birth delivery. The baseline characteristics during IVF treatment between the women with and without clinical pregnancy group were present in Table 1. Compared with women with non-pregnancy, women who achieved clinical pregnancy had significantly lower progesterone levels on hCG trigger day (P = 0.014) and significantly higher EMT on hCG trigger day (P = 0.006). A multivariate regression analysis was conducted to identify factors that might affect clinical pregnancy rates in women with PCOS who received their first FET cycle. After adjusting for the effect of maternal age, body mass index (BMI), AFC, type of infertility, embryo quality, duration of infertility, and estradiol on hCG trigger day, EMT on hCG trigger day (adjusted OR [aOR] = 1.216, 95% CI 1.043-1.419, P = 0.013) and progesterone on hCG trigger day (aOR = 0.109, 95% CI 0.018-0.670, P = 0.017) were associated with clinical pregnancy rates (Table 2).

Table 1. The baseline clinical characteristics during IVF treatment between the women with and without clinical pregnancy group.

Characteristics	Clinical pregnancy (n = 397)	Non-clinical pregnancy (n = 177)	P-value
Age (years)	30.00 (27.00-32.00)	30.00 (27.00-32.00)	0.924
BMI (kg/m^2)	22.15 (19.80-25.33)	21.91 (19.81-24.67)	0.431
Duration of infertility (years)	3.10 (1.20-4.10)	3.10 (2.10-4.18)	0.144
Type of infertility			0.353
Primary	267 (67.3%)	112 (63.3%)
Secondary	130 (32.7%)	65 (36.7%)
AMH (ng/mL)	7.71 (5.83-10.50)	7.84 (5.24-10.56)	0.643
AFC	24.00 (20.00-30.00)	25.00 (21.00-30.50)	0.134
Basal FSH (mIU/mL)	5.93 (4.96-7.02)	5.81 (5.06-6.79)	0.580
Basal LH (mIU/mL)	7.14 (4.86-10.99)	8.46 (4.84-11.16)	0.262
Basal E2 (pg/mL)	37.75 (28.73-53.00)	38.44 (29.65-52.05)	0.562
Basal P (ng/mL)	0.23 (0.12-0.42)	0.21 (0.12-0.38)	0.532
Basal T (ng/mL)	0.41 (0.29-0.61)	0.41 (0.28-0.62)	0.800
Basal PRL (ng/mL)	14.50 (10.49-21.22)	14.85 (10.31-22.26)	0.964
Fasting glucose (mmol/L)	5.31 (5.01-5.57)	5.30 (4.94-5.59)	0.600
Fasting insulin (μU/mL)	10.50 (7.10-15.90)	10.75 (7.10-17.28)	0.727
QUICKI	0.33 (0.31-0.35)	0.33 (0.31-0.36)	0.958
TSH (μIU/mL)	2.01 (1.48-2.70)	1.88 (1.36-2.57)	0.281
E2 (pg/mL) on hCG day	194.30 (143.10-265.85)	206.65 (156.95-287.48)	0.217
P (pg/mL) on hCG day	0.11 (0.06-0.18)	0.13 (0.08-0.22)	0.014
EMT (mm) on hCG day	9.00 (8.00-10.00)	9.00 (8.00-10.00)	0.006
No. of oocytes retrieved	21.00 (18.00-27.00)	22.00 (18.00-28.00)	0.454
Embryo quality			0.145
Cycle with high-quality embryos	383 (96.5%)	166 (93.8%)
Cycle without high-quality embryos	14 (3.5%)	11 (6.2%)
Fertilization method			0.165
IVF	341 (85.9%)	144 (81.4%)
ICSI	56 (14.1%)	33 (18.6%)
Clinician for embryo transfer			0.195
A	84 (21.2%)	48 (27.1%)
B	83 (20.9%)	29 (16.4%)
C	133 (33.5%)	65 (36.7%)
D	97 (24.4%)	35 (19.8%)
Duration of oral estrogen administration	22.00 (22.00-22.00)	22.00 (22.00-22.00)	0.603

Table 2. Multivariable logistics regression analysis on the contribution of the potential factors to clinical pregnancy outcomes.

Variables	Adjusted OR	95% CI	P-value
Maternal age (years)	0.987	0.934–1.044	0.654
BMI (kg/m^2)	1.019	0.969–1.072	0.466
AFC	0.980	0.960–1.002	0.070
Secondary infertility	0.847	0.572–1.253	0.406
Embryo quality	2.070	0.902–4.752	0.086
Duration of infertility (years)	0.960	0.879–1.048	0.358
E2 (pg/mL) on hCG day	1.000	1.000–1.001	0.603
P (pg/mL) on hCG day	0.109	0.018–0.670	0.017
EMT (mm) on hCG day	1.216	1.043–1.419	0.013

A regression analysis of the factors influencing the live birth rate of FET in PCOS patients was conducted (Table 3). Thicker endometrium on hCG trigger day was associated with live birth (OR = 1.264, 95% CI 1.107-1.443, P = 0.001) in the univariate regression analysis, while after adjusting for age, BMI, AFC, fasting insulin, E2 on hCG trigger day, and embryo quality, thicker endometrium (aOR = 1.179, 95% CI 1.011-1.376, P = 0.036) and lower progesterone concentration (aOR = 0.055, 95% CI 0.007-0.420, P = 0.005) on hCG trigger day were positively associated with live birth outcomes.

Table 3. Logistics regression analysis on the contribution of the potential factors to live birth outcomes.

	Univariate analysis		Multivariate analysis
Variables	OR (95% CI)	P-value	aOR (95% CI)	P-value
Maternal age (years)	0.971 (0.926–1.017)	0.214	0.950 (0.896–1.008)	0.089
BMI (kg/m^2)	0.986 (0.945–1.030)	0.533	0.995 (0.935–1.059)	0.869
Secondary infertility	0.917 (0.647–1.299)	0.625
Duration of infertility (years)	0.979 (0.908–1.056)	0.585
AMH (ng/mL)	0.994 (0.956–1.033)	0.746
AFC	0.983 (0.964–1.003)	0.090
Basal FSH (mIU/mL)	1.043 (0.947–1.148)	0.397
Basal LH (mIU/mL)	0.997 (0.969–1.026)	0.844
Basal E2 (pg/mL)	0.999 (0.995–1.002)	0.478
Basal P (ng/mL)	1.031 (0.939–1.132)	0.521
Basal T (ng/mL)	1.032 (0.678–1.573)	0.882
Basal PRL (ng/mL)	1.001 (0.999–1.003)	0.435
Fasting glucose (mmol/L)	0.939 (0.651–1.354)	0.735
Fasting insulin (μU/mL)	0.979 (0.956–1.002)	0.070	0.976 (0.948–1.005)	0.108
QUICKI (<0.357)	0.726 (0.440–1.199)	0.212
TSH (μIU/mL)	1.027 (0.879–1.200)	0.74
E2 (pg/mL) on hCG day	1.000 (0.999–1.000)	0.135	1.000 (0.999–1.001)	0.626
P (pg/mL) on hCG day	0.198 (0.037–1.057)	0.058	0.055 (0.007–0.420)	0.005
EMT (mm) on hCG day	1.264 (1.107–1.443)	0.001	1.179 (1.011–1.376)	0.036
No. of oocytes retrieved	1.004 (0.983–1.024)	0.729
Embryo quality	2.034 (0.898–4.609)	0.089	1.894 (0.728–4.924)	0.190
Fertilization method	0.704 (0.447–1.107)	0.129
Clinician for embryo transfer		0.319
Duration of oral estrogen administration	0.987(0.931–1.046)	0.647

We also performed logistics regression analyses of miscarriage and preterm delivery rates in patients with PCOS. In Table 4, The univariate analysis showed that QUICKI less than 0.357 (OR = 2.776, 95% CI 1.121-6.871, P = 0.027) and fewer oocytes retrieved (OR = 0.961, 95% CI 0.927-0.996, P = 0.031) were associated with miscarriage in the first FET cycle of PCOS patients. However, after adjusting for maternal age, BMI, duration of infertility, fasting insulin, E2 on hCG day, EMT on hCG day, and embryo quality, no factors were found to have a significant effect on the outcome of miscarriage. As presented in Table 5, the multivariate logistics analysis indicated that AFC was an independent risk factor for preterm delivery in PCOS patients (aOR = 1.053, 95% CI 1.003-1.106, P = 0.037).

Table 4. Logistics regression analysis on the contribution of the potential factors to miscarriage outcomes.

	Univariate analysis		Multivariate analysis
Variables	OR (95% CI)	P-value	aOR (95% CI)	P-value
Maternal age (years)	1.063 (0.989–1.144)	0.099	1.060 (0.964–1.164)	0.228
BMI (kg/m^2)	1.067 (0.999–1.138)	0.052	1.057 (0.963–1.161)	0.246
Secondary infertility	0.906 (0.521–1.576)	0.727
Duration of infertility (years)	0.998 (0.891–1.117)	0.967	0.967 (0.850–1.099)	0.604
AMH (ng/mL)	1.024 (0.967–1.084)	0.423
AFC	1.010 (0.980–1.041)	0.504
Basal FSH (mIU/mL)	0.998 (0.870–1.145)	0.977
Basal LH (mIU/mL)	0.977 (0.929–1.027)	0.358
Basal E2 (pg/mL)	1.000 (0.995–1.006)	0.874
Basal P (ng/mL)	0.929 (0.712–1.212)	0.587
Basal T (ng/mL)	0.943 (0.484–1.838)	0.862
Basal PRL (ng/mL)	0.997 (0.993–1.002)	0.234
Fasting glucose (mmol/L)	1.195 (0.743–1.924)	0.462
Fasting insulin (μU/mL)	1.029 (0.994–1.065)	0.101	1.011 (0.959–1.066)	0.678
QUICKI (<0.357)	2.776 (1.121–6.871)	0.027	1.967 (0.702–5.513)	0.198
TSH (μIU/mL)	1.008 (0.799–1.272)	0.946
E2 (pg/mL) on hCG day	1.001 (1.000–1.001)	0.051	1.001 (1.000–1.002)	0.270
P (pg/mL) on hCG day	0.873 (0.055–13.772)	0.923
EMT (mm) on hCG day	0.834 (0.680–1.023)	0.081	0.895 (0.704–1.139)	0.369
No. of oocytes retrieved	0.961 (0.927–0.996)	0.031	0.968 (0.926–1.013)	0.161
Embryo quality	0.530 (0.161–1.741)	0.295	0.809 (0.154–4.250)	0.802
Fertilization method	1.304 (0.649–2.618)	0.456
Clinician for embryo transfer		0.315
Duration of oral estrogen administration	1.015 (0.915–1.125)	0.782

Table 5. Logistics regression analysis on the contribution of the potential factors to preterm delivery outcomes.

	Univariate analysis		Multivariate analysis
Variables	OR (95% CI)	P-value	aOR (95% CI)	P-value
Maternal age (years)	0.971 (0.891–1.059)	0.511	0.988 (0.877–1.114)	0.849
BMI (kg/m^2)	0.987 (0.912–1.068)	0.740	0.966 (0.864–1.080)	0.539
Secondary infertility	0.756 (0.395–1.445)	0.397
Duration of infertility (years)	0.985 (0.862–1.127)	0.830
AMH (ng/mL)	1.020 (0.954–1.090)	0.562	1.009 (0.897–1.135)	0.884
AFC	1.044 (1.008–1.080)	0.015	1.053 (1.003–1.106)	0.037
Basal FSH (mIU/mL)	0.969 (0.822–1.142)	0.704
Basal LH (mIU/mL)	0.992 (0.940–1.045)	0.753
Basal E2 (pg/mL)	1.003 (0.997–1.010)	0.330
Basal P (ng/mL)	0.919 (0.672–1.257)	0.596
Basal T (ng/mL)	1.248 (0.612–2.544)	0.542	0.933 (0.272–3.198)	0.911
Basal PRL (ng/mL)	1.000 (0.997–1.004)	0.905
Fasting glucose (mmol/L)	1.809 (0.924–3.541)	0.084	2.476 (0.888–6.908)	0.083
Fasting insulin (μU/mL)	1.023 (0.975–1.073)	0.361
QUICKI (<0.357)	0.954 (0.395–2.300)	0.916
TSH (μIU/mL)	0.860 (0.614–1.205)	0.381
E2 (pg/mL) on hCG day	1.000 (0.998–1.001)	0.776
P (pg/mL) on hCG day	2.097 (0.090–49.043)	0.645
EMT (mm) on hCG day	0.960 (0.770–1.197)	0.717
No. of oocytes retrieved	0.981 (0.943–1.021)	0.342
Embryo quality	1.786 (0.222–14.403)	0.586
Fertilization method	1.245 (0.498–3.115)	0.640
Clinician for embryo transfer		0.194
Duration of oral estrogen administration	0.978 (0.843–1.134)	0.769

Discussion

In this study, we found that higher progesterone concentration and thinner EMT on the day of hCG administration were associated with the decreased clinical pregnancy rate and live birth rate in PCOS patients undergoing their first single blastocyst frozen transfer. QUICKI < 0.357 and fewer oocytes retrieved was associated with miscarriage in univariate analysis but lost statistical significance after adjusting for confounders. In addition, more AFC was a risk factor for preterm birth.

Studies showed that increased serum progesterone on the day of hCG administration had a detrimental effect on pregnancy outcomes in fresh embryo transfer cycles [27–29], and it was also indicated that in patients with PCOS, elevated progesterone on the hCG trigger day was associated with reduced clinical pregnancy rate in fresh IVF cycles [30]. This may be attributed to the fact that abnormally high progesterone values during a fresh cycle can provoke a premature conversion of the endometrium to the secretory phase and an early closure of the implantation window [31, 32]. Additionally, it was reported that in FET cycles, clinical pregnancy rates in patients with high progesterone in fresh cycles were not significantly different from those with normal progesterone concentration [32, 33]. Therefore, physicians often propose frozen embryo transfer to address the embryo-uterine asynchrony in patients with elevated progesterone concentration in the ovarian stimulation cycles. However, it appeared that pregnancy outcomes in frozen cycles were also affected by high progesterone levels. A study from Hong Kong reported that elevated progesterone levels lasting two days or more before the LH surge were associated with reduced clinical pregnancy rates in patients undergoing FET-natural cycles, and women with elevated progesterone levels in ovarian stimulation cycles also tend to have elevated progesterone levels in subsequent natural cycles [34]. Kofinas et al. reported that in patients undergoing HRT-FET cycles, progesterone levels greater than 20 ng/dL on the day of embryo transfer were associated with reduced live birth rate [35]. In the PCOS patients undergoing HRT-FET cycles included in our study, elevated progesterone concentration on the hCG trigger day remained associated with lower clinical pregnancy and live birth rate.

Elevated serum progesterone on the hCG trigger day may derive from the accumulation of progesterone production from multiple follicles during ovarian stimulation for IVF treatment [29, 36]. Generally, the mechanisms of adverse pregnancy outcomes caused by elevated progesterone on hCG trigger day may be two aspects, one is oocyte damage and the change of embryo quality, and the other is the impairment of endometrial receptivity. There is still no consensus on the effect of elevated progesterone on embryo quality. In a donor oocyte programme, no significant difference in the pregnancy rate was found between the recipients who received oocyte donated by patients with and without elevated progesterone on the day of hCG administration [37]. Turgut et al. found that elevated progesterone had no negative effect on embryological parameters of blastocysts [38]. However, two large retrospective studies indicated that elevated serum progesterone on the day of oocyte maturation was associated with a lower rate of high-quality blastocyst formation [39,40]. In GnRH antagonist IVF/ICSI cycles, high serum progesterone levels on the hCG trigger day were associated with reduced embryo utilization and cumulative live birth rates [41]. Currently, limited studies have been conducted on impaired oocyte quality associated with high progesterone in IVF cycles, and the mechanisms involved are unclear. Some animal studies showed that lower follicular progesterone concentrations improve bovine oocyte development in vitro [42], and that oocyte maturation and developmental capacity were regulated by progesterone-responsive genes [39, 43, 44]. In this study, the oocytes collected during stimulation cycles in patients with PCOS may be compromised by high progesterone, which led to impaired embryo quality and reduced high-quality embryos, affecting the pregnancy outcome in frozen embryo cycle.

Most studies supported that elevated progesterone had negative effects on endometrial microenvironment [33, 45, 46]. In antagonist IVF cycles, progesterone receptors on the endometrium were up-regulated on the day of hCG administration, and the effect of progesterone was amplified, leading to premature luteinization and increased endometrial maturation [47]. Liu et al. reported that high levels of progesterone before oocyte retrieval impaired components of the NK cell-mediated cytotoxic pathway in the endometrium [28]. Although the premature luteinization (progesterone elevation) of the endometrium could be resolved in HRT-FET cycles to a large extent [48], endometrial gene expression and epigenetics was abnormally altered by exposure to high progesterone levels on hCG trigger day [45, 46], resulting in impaired endometrial receptivity and may even influence pregnancy outcomes in the subsequent FET cycles. Overall, a combined effect of compromised embryo quality and impaired endometrial receptivity may be responsible for the reduced clinical pregnancy and live birth rates in FET cycles in PCOS patients with high progesterone in this study. Therefore, in view of the detrimental effects of high progesterone on pregnancy in fresh cycles, a freezing strategy is not sufficient, and it was suggested that the total dose of ovarian stimulation should be reduced in order to establish a balance between the low value of progesterone and the high number of oocytes to improve the pregnancy outcomes [41]. Meanwhile, when high progesterone levels are observed in fresh IVF cycles in patients with PCOS, physicians should regularly monitor progesterone levels and inform patients of the potential risk of adverse pregnancy outcomes in FET cycles.

Endometrial thickness is the most commonly used indicator to evaluate endometrial receptivity in clinical practice [49]. Previous studies have shown that thin endometrium has adverse effects on pregnancy outcomes in fresh cycles [50–53]. It was also reported that endometrial thickness was positively correlated with clinical pregnancy rate or live birth rate in frozen embryo transfer cycles [54–56], and similar findings were observed in patients with PCOS [57, 58]. Our study showed that thinner endometrium on hCG trigger day was associated with reduced clinical pregnancy and live birth rates in PCOS patients undergoing their first FET cycle. This is consistent with Basir’s study in which they found that the group with lower endometrial thickness in stimulated cycles also had lower pregnancy rates in frozen and natural cycles, and that there was a strong correlation between endometrial thickness in stimulated and natural cycles, suggesting that thinner endometrium in the first IVF cycle may be difficult to improve in subsequent natural and frozen cycles [59]. This may be attributed to endometrial insensitivity resulting in poor endometrial development and may be associated with some intrinsic uterine pathology [59, 60], as well as defects in endometrial estrogen and progesterone receptors that may exist in impaired developed endometrium [61, 62].

Studies have reported reduced endometrial receptivity in PCOS patients [63, 64], which may be caused by endocrine and metabolic disorders such as progesterone resistance, androgen or insulin elevation [65, 66]. A significant negative correlation between endometrial thickness and serum total testosterone was observed in PCOS patients [67]. The mechanisms for implantation failure or pregnancy loss due to thin endometrium have not been fully elucidated. The thinner functional layer may expose the embryo to higher oxygen concentrations in the basal endometrium, which is unfavorable for fetal growth [68]. Other studies have suggested that abnormal transcriptional changes in thin endometrium were also involved in pregnancy failure [69, 70]. Our study suggested that PCOS patients with thinner endometrium in their first IVF cycle may remain at a higher risk of pregnancy failure in FET cycles compared to patients with thicker endometrium. Active treatments are needed to increase endometrial thickness, such as administration of estrogen [71], low-dose aspirin [72], vaginal sildenafil [73], intrauterine infusion with granulocyte colony-stimulating factor (G-CSF) [74] or platelet-rich plasma (PRP) [75, 76]. However, there is a lack of conclusive evidence to prove the effectiveness of these methods, and PCOS patients with thin endometrium detected in the IVF cycle should be informed that the pregnancy outcome may remain unsatisfactory even in the frozen embryo cycle.

Previous studies have found an increased risk of preterm birth in women with PCOS during natural pregnancy and IVF cycles [11, 77–81]. Our study identified the higher AFC as a risk factor for preterm delivery in patients with PCOS undergoing their first FET, which to our knowledge has not been reported before. A recently published retrospective study of 4266 live birth cycles showed that AFC of more than 24 (OR = 1.378, 95% CI: 1.035-1.836) was an independent risk factor for preterm delivery in patients of IVF cycles [82]. Currently, there are few data on risk factors of preterm delivery in women with PCOS. The mechanism of increased risk of preterm delivery in PCOS women is not very clear, which may be involved with hyperandrogenism, glucose and lipid metabolism disorders [78, 83]. Excess androgen may lead to alterations in cervical remodeling and myometrial function [84], and even result in myometrial relaxation through non-genomic actions [85, 86]. Abnormally increased androgens may also promote preterm birth by increasing the risk of gestational diabetes mellitus [85, 87]. Studies reported a positive association between follicle number and androgen levels in women with PCOS [88], which may explain the association of more AFC with preterm birth. Our findings suggest that AFC may be a more sensitive factor than androgen level in predicting preterm delivery in FET cycles. AFC is a common indicator to evaluate women’s ovarian reserve and the response to ovarian stimulation in the ART cycle [89–91]. It was found that PCOS patients with lower AFC during IVF cycles are more likely to conceive, possibly due to the higher responsiveness to ovarian stimulation in patients with higher AFC, leading to premature luteinization and impaired endometrial receptivity [30]. A study reported that AFC was a risk factor for clinical pregnancy loss in women with PCOS undergoing IVF cycles [92]. These findings suggested that high AFC had a negative effect on pregnancy outcomes in PCOS patients undergoing IVF. Overall, the effect of AFC on the risk of preterm delivery in PCOS patients undergoing FET cycles still needs more data to explore.

QUICKI is a common indicator of insulin sensitivity, and based on previous studies [26, 93], we classified patients with QUICKI values below 0.357 as the insulin resistance group. Our univariate regression analyses showed that QUICKI less than 0.357 and a smaller number of oocytes retrieved were associated with miscarriage. However, after adjusted for relevant variables such as age and BMI, the effects of QUICKI less than 0.357 and number of oocytes on miscarriage were no longer statistically significant. In our additional analyses, we found that PCOS patients with higher BMI were more likely to have QUICKI values below 0.357 (OR = 1.385, 95% CI 1.232-1.556, p < 0.001), suggesting patients with higher BMI would be more susceptible to insulin resistance. Obesity was often accompanied by insulin resistance and they were both common metabolic symptoms in patients with PCOS [94, 95]. Studies reported that higher BMI was a risk factor for pregnancy loss in patients with PCOS [92, 96]. Therefore, our data are insufficient to support that insulin resistance could perform an independent role in miscarriage in PCOS patients, and more data may be required for stratification studies with different ranges of BMI.

Studies showed that the number of oocytes retrieved was associated with increased cumulative live birth rate [97–99], and patients with a higher number of oocytes retrieved tended to have more options for embryo transfer, and were theoretically more likely to have good quality embryos transferred [97]. The quality of the embryo transferred may explain the result of our univariate analysis where patients with fewer oocytes retrieved experienced higher miscarriage rate. We further analyzed the correlation between the number of oocytes retrieved and the rate of high-quality embryo transferred, and a positive association was found between them (OR = 1.128, 95% CI 1.034-1.231, P = 0.007). Consequently, the effect of the number of oocytes retrieved on miscarriage disappeared after adjusted for the embryo quality. In addition, it’s noted that PCOS patients usually had ovarian hyperresponsiveness and may easily obtain more oocytes, which warranted prevention of the risk of OHSS and thrombosis associated with increased oocyte numbers [99]. Therefore, the number of oocytes retrieved was suggested to be controlled to an optimal range in patients with PCOS [99, 100].

This study was concerned with the factors influencing pregnancy outcome in PCOS patients undergoing their first FET cycle. Our inclusion criteria were strict and excluded some patients with underlying diseases, thus improving homogeneity and comparability among patients. For example, we excluded PCOS patients with metabolic syndrome given that PCOS patients with metabolic syndrome was reported to have lower cumulative live birth rates and reduced fertility than PCOS women without metabolic syndrome [101]. Additionally, the conclusions were more reliable after controlling confounders using multivariate logistic regression analysis. However, there are still several limitations in this study. First, the sample size was relatively small, which may account for the fact that some of important factors affecting pregnancy outcomes in women with PCOS, such as age and BMI, were not confirmed in our study. Moreover, the limited sample size also did not allow for further analysis on the specific thresholds and predictive modeling of the contributing factors in our study. Second, only PCOS patients who underwent their first FET were included, and the ovarian stimulation protocol and endometrial preparation protocol adopted by patients were controlled for better homogeneity, which also reduced the generalizability of our findings. In addition, it’s noted that patients’ serum progesterone levels and EMT during the transfer cycle before progesterone support are also important factors in assessing a patient’s pregnancy outcome, but unfortunately the data is not kept at our hospital.

Conclusion

In conclusion, this study analyzed the contributing factors of pregnancy outcomes in PCOS patients undergoing their first single blastocyst frozen-embryo transfer. We found that progesterone and EMT on hCG administration day was associated with the clinical pregnancy rate and live birth rate, and AFC was a risk factor for preterm birth. During the FET cycle, fertility clinicians should be concerned about abnormal changes in these indicators and inform patients about the potential risk of adverse pregnancy outcomes. More data are needed to elucidate the role of these factors in influencing and predicting pregnancy outcomes in PCOS patients.

Authors’ contributions

S.L., T.L.Y. designed the study. S.X., M.L.M., X.C. conducted data collection and supervised the study. X.C., L.H. analyzed the data. X.C. drafted the original manuscript. S.L., T.L.Y., L.H. revised the paper.

Disclosure statement

The authors declare that they have no competing interests or relevant relationships.

Data availability statement

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Additional information

Funding

This work was supported by the Basic Research Program of Shenzhen (JCYJ20210324123412035 and JCYJ20220530172814032).