In vitro maturation without gonadotropins versus in vitro fertilization with hyperstimulation in women with polycystic ovary syndrome: a non-inferiority randomized controlled trial

Abstract

STUDY QUESTION

Does in vitro maturation (IVM) result in non-inferior cumulative live birth rates compared to those after standard in vitro fertilization (IVF) in infertile women with polycystic ovary syndrome (PCOS)?

SUMMARY ANSWER

One cycle of IVM, without any stimulation, was inferior to one cycle of standard IVF in women with PCOS in terms of 6-month cumulative live birth rates, when choosing single vitrified-warmed blastocyst transfer.

WHAT IS KNOWN ALREADY

IVM is an emerging alternative treatment for women with PCOS who need assisted reproductive technology. Since a minimal or even zero dose of gonadotropins are required in the IVM procedure, the occurrence of ovarian hyperstimulation syndrome (OHSS) is eliminated. Only one clinical trial comparing the pregnancy outcome between IVM with FSH priming and IVF has been reported. However, it is still unknown whether IVM treatment without any stimulation can offer a similar live birth outcome in women with PCOS as compared to that in women receiving the standard IVF procedure with ovarian stimulation.

STUDY DESIGN, SIZE, DURATION

This single-centre, open-label randomized controlled non-inferiority trial in an academic infertility centre in China was performed between March 2018 and July 2019.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Women aged 20–38 years with PCOS and infertility scheduled for their first IVF attempt were eligible. In total, 351 women were randomly allocated to receive one cycle of unstimulated IVM (n = 175) or one cycle of standard IVF with a flexible GnRH antagonist protocol and hCG as ovulatory trigger (n = 176). A freeze-all and single blastocyst transfer strategy was used in both groups. The primary outcome was ongoing pregnancy (leading to live birth) within 6 months after randomization. A non-inferiority margin of 15% was considered.

MAIN RESULTS AND THE ROLE OF CHANCE

The IVM procedure without additional gonadotropin resulted in a lower ongoing pregnancy (leading to live birth) within 6 months after randomization compared to standard IVF treatment (22.3% vs. 50.6%; rate difference −28.3%; 95% confidence interval [CI]: −37.9% to −18.7%). Moderate-severe OHSS did not occur in the IVM group, while in the IVF group, ten women (5.7%) had moderate OHSS and one woman (0.6%) had severe OHSS. There was no statistically significant difference in the occurrence of obstetric and perinatal complications.

LIMITATIONS, REASONS FOR CAUTION

The trial was conducted using an IVM protocol without additional stimulation in a single centre, which may limit its generalizability. In addition, a GnRH agonist trigger rather than hCG for IVF stimulation in women with PCOS would be more consistent with current clinical practice.

WIDER IMPLICATIONS OF THE FINDINGS

Although IVM is considered to be a convenient, inexpensive and safe alternative to IVF for women with PCOS, our results indicated that one cycle of IVM without any stimulation was inferior to one cycle of standard IVF in terms of the cumulative live birth rate. The inferiority of IVM without ovarian stimulation could be mainly due to the limitations in the developmental potential of embryos. Further IVM development should be tested and validated in a freeze-only and blastocyst transfer setting. Further RCTs are needed to evaluate the effectiveness and safety of other IVM protocols or multiple cycles of IVM compared to IVF.

STUDY FUNDING/COMPETING INTEREST(S)

This study was supported by the National Key Research and Development Program of China (2016YFC1000201 and 2018YFC1002104) and the National Science Foundation of China (81730038). B.W.M. is supported by a NHMRC Investigator grant (GNT1176437). All other authors declare no competing interests.

TRIAL REGISTRATION NUMBER

Clinicaltrials.gov NCT03463772.

TRIAL REGISTRATION DATE

29 January 2018.

DATE OF FIRST PATIENT’S ENROLMENT

16 March 2018.

Introduction

Polycystic ovary syndrome (PCOS) is the most common reproductive endocrine disease, with an incidence of just over 5% in Chinese women of reproductive age (Li et al., 2013). More than half of women with PCOS have an/oligo-ovulatory infertility and when ovulation induction with oral agents, gonadotropins or laparoscopic ovarian surgery fails, in  vitro fertilization (IVF) is the recommended treatment (Teede et al., 2018). In IVF, the use of ovarian stimulation aims to induce multifollicular development, but this is associated with an increased risk of ovarian hyperstimulation syndrome (OHSS), especially in women with PCOS (Mourad et al., 2017). Severe OHSS is a potentially life-threatening iatrogenic complication. Methods to prevent OHSS include decreasing the hCG dose for the ovulatory trigger, using gonadotropin-releasing hormone (GnRH) agonists for the trigger, and employing a freeze-all strategy. Despite these strategies, OHSS still occurs, albeit less frequently.

In  vitro maturation (IVM), established since 1991 (Cha et al., 1991), has been a procedure in which immature oocytes are retrieved at the germinal vesicle stage and then matured in vitro to reach the metaphase II (MII) stage (Chian et al., 2013). It has been considered an alternative treatment option to standard IVF (Walls et al., 2015). As IVM requires zero or a minimal dose of gonadotropins, women undergoing IVM can avoid exposure to supraphysiological steroid levels, resulting in a low treatment burden, and elimination of the occurrence of OHSS (Huang et al., 2010).

Observational studies on IVM have suggested inconsistent results, with live birth rates ranging from 5.3% to 42.6% (Gremeau et al., 2012; Junk and Yeap, 2012; Ho et al., 2018; Braam et al., 2019). With the improvement of the IVM technique and the adoption of a ‘freeze-all strategy’, IVM has yielded comparable live birth rates compared to those from IVF in some observational studies (Junk and Yeap, 2012; Ortega-Hrepich et al., 2013). Although IVM is no longer considered experimental, given the lack of level 1 evidence (Siristatidis et al., 2018), both the international guideline for PCOS and the latest American Society for Reproductive Medicine (ASRM) practice committee opinion on IVM call for randomized controlled trials (RCTs) comparing clinical outcomes of IVM versus IVF (Practice Committees of the American Society for Reproductive Medicine, 2021; Teede et al., 2018). To date, only one single-centre RCT has compared the effectiveness of IVM and IVF in women with a high antral follicle count, and the study showed that live births approach inferiority for IVM compared with IVF when cumulative outcomes are considered (Vuong et al., 2020). It is worth noting that IVM with FSH priming and a pre-maturation step protocol was used in this RCT. Mild ovarian stimulation in PCOS patients to increase the efficiency of IVM has still been a topic of significant debate. Until now it has remained unknown whether IVM treatment without any stimulation can offer a similar live birth outcome in women with PCOS compared to those receiving the standard IVF procedure with ovarian stimulation.

Therefore, we performed an RCT to assess the effectiveness and safety of one cycle of IVM without the use of gonadotropins or hCG priming versus one cycle of standard IVF with ovarian stimulation and hCG as the ovulatory stimulus, both with a freeze-all and single-blastocyst transfer strategy, in women with PCOS.

Materials and methods

Study design and oversight

This was a single-centre, open-label, non-inferiority RCT performed at the Reproductive Medical Center of Peking University Third Hospital (Beijing, China). The study was approved by the Ethics Committee of Peking University Third Hospital (Zheng et al., 2020) and registered as NCT03463772. All authors assume responsibility for the completeness and accuracy of the data and analyses and the fidelity of the trial to the protocol. A detailed study protocol has been published previously (Zheng et al., 2020).

Study population

Infertile women diagnosed with PCOS according to the revised Rotterdam criteria (Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group, 2004) who were between 20 and 38 years of age and scheduled for their first IVF attempt were eligible to participate in the trial. If a woman wanted to participate, written informed consent was acquired at their next scheduled visit. To initiate the IVM or IVF treatment cycle in anovulatory cases, patients were administered oral dydrogesterone (Duphaston, Abbott, OLST, Netherlands) 20 mg daily for 10–14 days and oral contraceptives Diane-35 (cyproterone acetate 2 mg, ethinylestradiol 35 mg, Bayer and its generics) for 21 days. After withdrawal bleeding, eligible participants were again assessed for the exclusion criteria on day 2 or day 3 following the onset of menstrual bleeding.

Exclusion criteria were: couples scheduled for preimplantation genetic testing (PGT) with abnormal results on parental karyotyping, women who had undergone unilateral ovariectomy or had congenital/acquired uterine malformations or a male partner diagnosed with azoospermia. Potentially eligible women were informed about the trial during their first consultation, and if a woman desired to participate, written informed consent was required at the next scheduled visit.

Randomization

After written informed consent was obtained, participants were randomly assigned in a 1:1 ratio to receive IVM or IVF treatment on day 2/3 of the menstrual cycle according to a computer-generated randomization list, with a variable block size of 4 or 6. Opaque sealed envelopes with randomized assigned groups printed inside were numbered consecutively. The randomization list and envelopes were prepared by a statistician at the Research Centre of Clinical Epidemiology of Peking University Third Hospital, independent of the investigator team. Due to the nature of the interventions, the study was not blinded.

Study procedures

In the IVM group, a protocol without ovarian stimulation was used, i.e. IVM without the use of gonadotropins or human chorionic gonadotropin (hCG) priming. Participants visited the clinic on days 6–8 for a transvaginal ultrasound examination to exclude the development of a dominant follicle. Retrieval of immature oocytes was scheduled once the endometrial thickness reached at least 6 mm (Monis and Tetrokalashvili, 2021; Nalaboff et al., 2001) and the largest follicle was less than 10 mm. Transvaginal ultrasound-guided oocyte retrieval was conducted with a single-lumen 19G aspiration needle (K-OPS-7035-REH-ET; Cook, Queensland, Australia) in 90 mmHg suction pressure with the use of intravenous sedation. All cumulus-oocyte complexes (COCs) were transferred into IVM medium containing 5 mg/ml serum protein substitute (IVM media kit; Sage, CT, USA) supplemented MENOPUR (Menopur; Ferring, Kiel, Germany) in a 5% CO₂ incubator at 37 °C. Each MENOPUR contains 75 IU FSH and 10 IU hCG, which is equivalent to 75 IU of FSH and 75 IU of LH. The final concentration in the IVM culture system was 0.075 IU/ml FSH and 0.010 IU/ml HCG. All cumulus cells were denuded after 28–32 h of culture and the oocyte maturation process was evaluated. All MII oocytes were inseminated using intracytoplasmic sperm injection (ICSI).

In the IVF group, a flexible GnRH-antagonist protocol was used for ovarian stimulation. Recombinant FSH (rFSH) (Gonal-F; Serono, Aubonne, Germany) with a starting dose ranging from 112.5 to 225 IU was administered on day 2/3 of the menstrual cycle. Transvaginal ultrasound, serum luteinizing hormone (LH), serum oestrogen, and progesterone were measured to monitor follicle growth. The doses of rFSH were adjusted according to the ovarian response. GnRH-antagonist (Cetrotide; Serono, Aubonne, Germany) 0.25 mg daily was administered subcutaneously when at least one follicle reached a diameter of 12 mm (usually between Day 5 and Day 8 of ovarian stimulation), until and including the ovulatory trigger day. When two or more follicles reached a diameter of at least 17 mm, 250 µg of rhCG (Ovidrel; Serono, Aubonne, Germany) was administered for triggering. Oocyte retrieval was performed 36 (_±2) hours after triggering with the use of intravenous sedation. IVF or ICSI was used for fertilization based on the semen analysis results.

Fertilization was considered normal when two pronuclei were present between 16 and 18 h after ICSI or IVF. All zygotes were cultured in cleavage medium (G-1plus, Vitrolife, USA) for an additional 48–52 h after fertilization. Cleavage embryonic development was assessed according to the developmental stage and degree of cytoplasmic fragmentation. Good quality embryos were defined as 5–8 cells with less than 30% fragments and even size. A freeze-all plus single-blastocyst embryo transfer strategy was applied in all participants. On Day 3, the cleavage-stage embryo was cultured in blastocyst culture media (G-2 plus, Vitrolife, USA) for another 2–3 days. All transferable blastocysts were vitrified and then thawed in subsequent embryo transfer cycles. A hormone replacement therapy-based protocol was used for endometrial preparation. Oral oestradiol valerate (Progynova, Delpharm Lille, Lys-Lez-Lannoy, France) at a dose of 3 mg twice daily was administered on days 1–3 of the menstrual cycle if the endometrial thickness was less than 6 mm and no ovarian cyst was observed on ultrasound. An ultrasound scan was repeated 10 days later. When the endometrial thickness reached 8 mm, 90 mg vaginal progesterone gel (Crinone, Merck Serono, Watford, UK) daily and oral dydrogesterone (Duphaston, Abbott, OLST, Netherlands) 20 mg twice daily were administered. Only one blastocyst was transferred on day 7 after progesterone administration. Luteal phase support was continued until 12 weeks of gestation.

On Day 12 after blastocyst transfer, serum hCG was measured. In women with a positive pregnancy test, an ultrasound scan was performed 28 days after embryo transfer, and clinical pregnancy was diagnosed if a gestational sac was observed. Pregnant women were followed up and treated based on routine clinical practice. At 12 weeks of gestation, an ultrasound scan was repeated to confirm ongoing pregnancy, which was defined as a viable pregnancy with a foetal heartbeat. Information on pregnancy outcomes was collected as planned (Zheng et al., 2020).

Study outcomes

The primary outcome in the protocol (Zheng et al., 2020) was ongoing pregnancy (leading to live birth, defined as a baby born live at _≥22 weeks of gestation; Zegers-Hochschild et al., 2017) within 6 months of the first oocyte retrieval cycle after randomization. Since some patients did not have their first frozen embryo transfer until 3 and 6 months after randomization, we have included all transfer cycles within 6 months after randomization in the analysis as our primary endpoint so that all women could have at least one embryo transfer cycle. Secondary outcomes included implantation, clinical pregnancy, and time to ongoing pregnancy leading to live birth. Safety outcomes included OHSS (classified as mild, moderate or severe according to the RCOG Guideline; Royal college of Obstetricians and Gynaecologists, 2016), miscarriage, ectopic pregnancy and obstetric and perinatal complications. Full definitions of these terms are provided in Supplementary Table SI.

Statistical analysis

Our primary hypothesis was that the rate of ongoing pregnancy leading to live birth in the IVM group would be non-inferior to that in the IVF group. Assuming a live birth rate of 35% in the IVF group which was based on our clinical data, a sample size of at least 175 women per group was required to demonstrate non-inferiority with a power of 80%, and a non-inferiority margin of 15% for the lower limit of the two-sided 95% CI, including a dropout rate of 10%.

Continuous variables are presented as means with standard deviations (SDs) when normally distributed; otherwise, their medians and inter-quantile ranges (IQRs) are reported. Categorical variables are presented as proportions in each group. For outcome variables, a comparison between groups was performed using the independent sample t-test or Mann-Whitney U test for continuous variables according to the data distributions, or Pearson chi-square test or Fisher’s exact test for categorical variables as appropriate. The absolute rate differences (ARDs), relative risks (RRs) and their 95% confidence interval (CI) between the two groups were calculated. The 95% CIs of the ARDs were used to evaluate whether IVM is non-inferior to IVF. For the time to ongoing pregnancy (leading to live birth), Kaplan–Meier curves were constructed. The primary analysis was performed according to the intention-to-treat principle, with a per-protocol (PP) analysis being performed as an additional test. All statistical analyses were performed using the statistical package SPSS, version 25.0 (SPSS Inc., Chicago, IL, USA).

Results

Participants

Between March 2018 and July 2019, 2008 women with PCOS planning their first IVF cycle were evaluated for eligibility. A total of 231 were not eligible (43 women with a male partner diagnosed with azoospermia; 32 women or their male partner with a known abnormal chromosome karyotype; 27 women scheduled for PGT; 77 women participating in other trials; and 52 women with other diseases, including type 2 diabetes mellitus), while 1426 women declined to participate (Fig. 1).

Figure 1.

Enrolment, randomization and follow-up of participants.

The 351 participating women were randomly assigned to IVM (n = 175) or IVF (n = 176). The baseline characteristics of the women were comparable between the two groups (Table I). In the IVM group, three women had their oocyte retrieval procedure cancelled, one patient was found to have a dominant follicle on D6-8 ultrasound and opted for the IVF treatment in the next cycle, and another two women asked to transfer to IVF treatment group after randomization. The endometrial thickness of all cases reached at least 6 mm when the retrieval of immature oocytes was scheduled. In the IVF group, three oocyte retrieval procedures were cancelled and a flexible GnRH-antagonist protocol was used for ovarian stimulation. Seven patients had unexpectedly poor responses to ovarian stimulation, and the treatment was cancelled. All of these seven cases began a new ovarian stimulation cycle with other protocols (long or short protocols) and were not included in the per-protocol set (Fig. 1). After oocyte retrieval, seven women in the IVF group refused blastocyst culture and had cleavage-stage embryo transfer or freezing. The outcomes of all these women were included in the intention-to-treat analysis. One IVF patient was lost to follow up after 12 weeks of gestation. The PP analysis set was 169 in the IVM group and 158 in the IVF group.

Table I.

Baseline demographic and clinical characteristics of the participants.*

	IVM group (N = 175)	IVF group (N = 176)
Age
Mean (SD)—years	28.9 (2.9)	29.5 (3.2)
Distribution no. (%)
<35 years	164 (93.7)	166 (94.3)
≥ 35 years	11 (6.3)	10 (5.7)
Body-mass index†—kg/m2
Mean (SD)	24.9 (4.0)	24.8 (4.0)
Fertility history
Duration of the attempt to conceive—years
<Median [IQR]	3.0 [2.0, 5.0]	3.0 [2.0, 4.0]
Previous conception—no. (%)	46 (26.3)	47 (26.7)
Indications for IVF—no. (%)
Anovulation only	42 (24.0)	34 (19.3)
Anovulation combined with other factors
Female factors	42 (24.0)	53 (30.1)
Male factors	75 (42.9)	76 (43.2)
Both factors	16 (9.1)	13 (7.4)
Ultrasonographic findings
Antral follicle count in each ovary—no. (%)
12–20	113 (64.6)	128 (72.7)
>20	62 (35.4)	48 (27.3)
Endometrial thickness—mm
<Mean (SD)	5.7 (1.9)	5.8 (1.7)
Laboratory tests
Follicle-stimulating hormone—IU/l
<Mean (SD)	5.67 (1.67)	5.96 (1.93)
Luteinizing hormone—IU/l
<Median [IQR]	7.03 [4.15, 10.50]	6.34 [4.01, 10.15]
Estradiol—pmol/l§
<Median [IQR]	159.00 [123.00, 202.00]	168.00 [126.00, 219.00]
Hyperandrogenism‖
<Distribution-no. (%)	89 (50.9)	72 (41.1)
Anti-Mullerian hormone—IU/l**
<Median [IQR]	9.98 [6.88, 15.08]	8.89 [5.99, 13.74]
Number of ovulation induction cycles before randomization
Median[IQR]	2.00 [0.00, 3.00]	2.00 [0.00, 3.75]
PCOS phenotype #
A—no. (%)	57 (32.6)	50 (28.4)
C—no. (%)	32 (18.3)	22 (12.5)
D—no. (%)	86 (49.1)	103 (58.5)

^*The denominators of patients who were included in each analysis were provided if they differed from the total numbers in the relevant study group. Percentages may not total to 100 because of rounding. SD denotes the standard deviation, and IQR denotes the interquartile range. There were no significant differences between the two groups.

^†The body-mass index is the weight in kilograms divided by the square of the height in METRE.

^§Data regarding oestradiol were missing for 2 patients (1.1%) in the IVM group and 2 patients (1.1%) in the IVF group.

^‖Hyperandrogenism was defined as Total testosterone > 2.53 nmol/l or Androstenedione > 11.5 nmol/l. Data were missing for 1 patient (0.55%) in the IVF group.

^**Data regarding anti-Mullerian hormone were missing for 16 patients (9.1%) in the IVM group and 18 patients (10.2%) in the IVF group.

^#The features of PCOS are clinical and/or biochemical hyperandrogenism (HA), ovulatory dysfunction (OD) and polycystic ovarian morphology (PCOM). There PCOS phenotypes are currently classified as phenotype-A (HA + OD + PCOM), phenotype-B (HA + OD), phenotype-C (HA + PCOM), and phenotype-D (OD + PCOM).

Primary and pregnancy outcomes

At 6 months after randomization, in the IVM group, 107 women had at least one single-blastocyst transfer cycle and 62 women had no blastocyst for transfer. In the IVF group, 153 women had at least one blastocyst transfer cycle and 6 women had no blastocyst for transfer (Fig. 1).

Outcome data for the randomized women are presented in Table II. The primary endpoint, cumulative ongoing pregnancy (leading to live birth) rate at 6 months after randomization in the IVM group was 39/175 (22.3%) versus 89/176 (50.6%) in the standard IVF group (ARD −28.3% (95% CI: −37.9 to −18.7, P < 0.001). The results of per-protocol analysis were consistent with the results of the ITT analysis (Table II). Among women who had no ongoing pregnancy within six months, 16 women in the IVM group and 41 women in the IVF group still had remaining blastocysts.

Table II.

Fertility outcomes of women with PCOS within 6 months of the first oocyte retrieval after randomization.

	IVM group	IVF group	Rate difference (95% CI)	Rate ratio in the IVM group (95% CI)	P value
Intention-to-treat analysis	(N = 175)	(N = 176)
Primary outcome
Live birth—no. (%)‖	39 (22.3)	89 (50.6)	−28.3 (−37.9 to −18.7)	0.44 (0.32 to 0.60)	<0.001
Secondary pregnancy outcomes
Conception—no. (%)†	67 (38.3)	125 (71.0)	−32.7 (−42.6 to −22.9)	0.54 (0.43 to 0.66)	<0.001
Clinical pregnancy—no. (%)‡	56 (32.0)	114 (64.8)	−32.8 (−42.7 to −22.9)	0.49 (0.38 to 0.62)	<0.001
Singleton	55	111
Twins*	1	3
Implantation (per embryo)—  no./total no. (%)§	56/142 (39.4)	116/245 (47.4)	−7.9 (−18.1 to 2.3)	0.83 (0.65 to 1.06)	0.14
Ongoing pregnancy—no. (%)¶	41 (23.4)	94 (53.4)	−30.0 (−39.7 to −20.3)	0.44 (0.32 to 0.59)	<0.001
Pregnancy complication
Ectopic pregnancy—no. (%)	0 (0.0)	0 (0.0)	–	–	–
Pregnancy loss—no./total no. (%)
Among biochemical pregnancies	11/67 (16.4)	11/125 (8.8)	7.62 (−2.55 to 17.78)	1.87 (0.85 to 4.07)	0.11
Among clinical pregnancies	20/56 (35.7)	26/114 (22.8)	12.91 (−1.82 to 27.63)	1.57 (0.96 to 2.55)	0.07
First trimester	15/56 (26.8)	20/114 (17.5)	9.24 (−4.30 to 22.78)	1.53 (0.84 to 2.75)	0.16
Second trimester	5/56 (8.9)**	6/114 (5.3)	3.67 (−4.85 to 12.19)	1.70 (0.54 to 5.31)	0.36
Per-protocol analysis	(N = 169)	(N = 158)
Primary outcome
Live birth—no. (%)‖	38 (22.5)	83 (52.5)	−0.30 (−0.40 to −0.20)	0.43 (0.31 to 0.59)	<0.001
Secondary pregnancy outcomes
Conception—no. (%)†	65 (38.5)	116 (73.4)	−0.35 (−0.45 to −0.25)	0.52 (0.42 to 0.65)	<0.001
Clinical pregnancy—no. (%)‡	55 (32.5)	108 (68.4)	−0.36 (−0.45 to −0.26)	0.48 (0.37 to 0.61)	<0.001
Singleton	55	108
Twin	0	0
Implantation (per embryo)—  no./total no. (%)§	55/135 (40.7)	108/221 (48.9)	−0.08 (−0.19 to 0.02)	0.83 (0.65 to 1.07)	0.135
Ongoing pregnancy—no.(%)¶	40 (23.7)	88 (55.7)	−0.32 (−0.42 to −0.22)	0.43 (0.31to 0.58)	<0.001
Pregnancy complications
Ectopic pregnancy—no/total no.(%)	0 (0.0)	0 (0.0)
Pregnancy loss—no/total no.(%)
Among biochemical pregnancy	10/65 (15.4)	8/116 (6.9)	8.49 (−1.42 to 18.40)	2.23 (0.93 to 2.57)	0.067
Among clinical pregnancy	20/55 (36.4)	26/108 (24.1)	12.29 (−2.77 to 27.34)	1.51 (0.93 to 2.45)	0.099
First trimester	15/55 (27.3)	20/108 (18.5)	8.75 (−5.11 to 22.62)	1.47 (0.82 to 2.65)	0.198
Second trimester	5/55 (9.1)**	6/108 (5.6)	3 (−5.20 to 12.28)	1.64 (0.52 to 5.12)	0.395

^‖‖Live birth was defined as detection of a viable foetuswith a foetal heartbeat at 22 weeks gestation.

^†Conception was defined as serum hCG ≥ 5 mIU/ml.

^‡Clinical pregnancy was defined as the detection of a gestational sac in the uterine cavity.

^*All twins were derived from patients who requested two cleavage embryo transfer.

^§The implantation rate was the number of gestational sacs divided by the number of embryos that were transferred.

^¶Ongoing pregnancy was defined as the detection of a viable foetus with a foetal heartbeat at 12 weeks gestation.

^**Three cases of pregnancy loss occurred between 23 and 28 weeks.

The implantation rates of IVM-derived blastocysts and IVF-derived blastocysts were 39.4% and 47.4%, respectively (ARD −7.9%, 95% CI: −18.1 to 2.3). The cumulative incidences of conception, clinical pregnancy and ongoing pregnancy within 6 months was lower in the IVM group than in the IVF group (Table II). The time to ongoing pregnancy leading to live birth was longer in the IVM group as shown in the Kaplan-Meier curve (P < 0.001, Supplementary Fig. S1).

Treatment cost, oocyte retrieval characteristics and safety endpoints

The average costs, treatment duration before oocyte retrieval, and the frequency of visits in the IVM group were less than those in the IVF group (P < 0.001, Table III). The mean number of retrieved oocytes were 14.0 and 18.0 in the IVM and IVF groups, respectively. In the IVM group, 2784 COCs were collected from 5 to 10 mm follicles on the recovery day (Table IV). There were 2407 (86.5%) and 377 (13.5%) COCs with compacted and sparse cumulus cell patterns, respectively. No expanded COC was observed because hCG priming wasn’t used in our IVM protocol. After 28–32 h of culture, 1272 MII (45.7%), 298 MI (10.7%), 767 GV (27.5%) and 447 degenerated (16.1%) oocytes were observed. The maturation rate increased to 57.3% (1595/2784) of oocytes by 48 h of culture. In the IVF group, 36 cycles underwent ICSI for fertilization according to semen quality. The maturation rate of the 36 ICSI cycles (73.6%) in the IVF group was significantly higher than those of the IVM procedure (Table IV). The mean number of good-quality embryos and transferable blastocysts was lower in the IVM group than that in the IVF group (2.0 vs. 8.0 and 1.0 vs. 5.0, respectively, Table IV).

Table III.

Oocyte retrieval characteristics and safety secondary endpoints.

	IVM group (N = 175)	IVF group (N = 176)	Between group difference (95% CI)	P value
Oocyte retrieval characteristics
Duration of the follicular phase—days	8.9 ± 2.0	11.9 ± 2.3	−3.0 (−3.5 to −2.5)	<0.001
Gonadotropin dose—IU	–	1275.0 [1157.5, 2306.3]	–	–
Cost—RMB yuan	11 616 [11 415, 11 840]	16 059.31 [14 600, 18 808]	–	<0.001
Visit—times	3.0 [3.0, 4.0]	6.0 [6.0, 7.0]	–	<0.001
Safety endpoints—no./total no. (%)
Moderate-severe OHSS	0/175 (0.0)	11/176 (6.3)	−6.36 (−9.99 to −2.72)	0.001
Gestational diabetes mellitus cases among clinical pregnancies	5/56(8.9)	8/114 (7.0)	1.91 (−6.91 to 10.73)	0.760
Gestational hypertensive cases among clinical pregnancies	4/56 (7.1)	11/114 (9.6)	−2.51 (−11.16 to 6.15)	0.776
Antepartum haemorrhage cases among all deliveries	0/36	0/88	–	–
Birth weight, g¶	3428.9 ± 476.5	3377.5 ± 609.4	51.32 (−177.06 to 279.72)	0.657
Largeness for gestational age among all deliveries	5/36 (13.9)	22/88 (25.0)	−11.11 (−25.58 to 3.36)	0.174
Smallness for gestational age among all deliveries	1/36 (2.8)	1/88 (1.1)	1.64 (−3.88 to 13.07)	0.498
Preterm birth among all deliveries	5/36 (13.9)	12/88 (13.6)	0.25 (−13.13 to 13.63)	>0.999
Congenital anomalies among all deliveries	0/36 (0.0)	1/88 (1.1)†	−1.14 (−6.16 to 8.55)	>0.999
Perinatal mortality among all deliveries	0/36	0/88	–	–

^†The only case of congenital anomaly was congenital anal atresia.

^¶Data included only patients who delivered singleton. Values are expressed as the mean ± standard deviation.

Table IV.

Embryology outcomes of IVM without stimulation versus IVF treatment.

	IVM group	IVF group	Between group difference (95% CI)	P-value
No of cycles	169	158
Fertilization method‡				<0.001
With IVF	0 (0.0)	121 (76.6)	–
With ICSI	169 (100.0)	36 (22.8)	–
With mixed IVF and ICSI	0 (0.0)	1 (0.6)	–
No. of oocytes retrieved
Total	2784	3109
per patient (median(IQR))	14 (9, 20)	18 (11, 25)	−3 (−5, −1)	0.001
No. of mature oocytes
Total (%)	1272/2784 (45.7)	535/727 (73.6)§
per patient (median (IQR))	6 (4, 9)	13 (8, 20)	−7 (−9, −4)	<0.001
No. of 2PN (pronuclear) zygotes
Total (%)†	751/1272 (59.0)†	1857/3109 (59.7)†
Per patient (median (IQR))	4 (2, 6)	10 (6, 15)	−6 (−8, −5)	<0.001
No. of good—quality embryo
Total (% per 2PN)	517/751 (68.8)	1390/1857 (74.9)
Per patient (median (IQR))	2 (1, 4)	8 (5, 12)	−5 (−6, −4)	<0.001
No. of vitrified blastocysts
Total (% per 2PN)	204/751 (27.2)	864/1857 (46.5)
Per patient (median (IQR))	1 (0, 2)	5 (3, 7)	−3 (−4, −3)	<0.001
Patients with no available blastocyst (%)	62 (36.7)	6 (3.8)	32.9 (4.0, 25.0)	<0.001

^‡IVF: in vitro fertilization. ICSI: intracytoplasmic sperm injection. Mixed IVF and ICSI was performed as half ICSI (50% oocytes were inseminated by IVF and 50% oocytes by ICSI).

^§Data included only 36 patients who were fertilized with ICSI method.

^†All the COCs in IVM group were denuded of cumulus cells after 28–32 h of culture and evaluated for the maturation process. ICSI was selected as insemination method. The 2PN percentage = No. of 2PN/No. of MII. In IVF group, the routine insemination method or ICSI was selected respectively according to the husband’s semen analysis. The 2PN percentage = No. of 2PN/No. of oocytes retrieved.

In the IVF group, 10 women (5.7%) had moderate OHSS and one woman (0.6%) had severe OHSS (Table III). In the IVM group, there were no cases of OHSS. In addition, there was no statistically significant difference in the occurrence of obstetric and perinatal complications, including gestational diabetes mellitus, hypertensive disorders of pregnancy, antepartum haemorrhage and perinatal mortality. The only case of congenital anomaly was a newborn with congenital anal atresia in the IVF group.

Discussion

The present study was an RCT with an adequate sample size to investigate whether IVM treatment without stimulation provides comparable live birth outcomes to those of standard IVF treatment, using a freeze-all and single blastocyst transfer strategy. We found one cycle of IVM without gonadotrophins to be inferior to one cycle of IVF with ovarian stimulation in terms of 6-month cumulative ongoing pregnancy rates.

A strength of this trial was the use of a freeze-all and single blastocyst transfer strategy, reflecting the current practice shift. In addition, there was only one patient lost to follow-up during the trial, partly due to the national mandatory quality standard for clinics performing assisted reproduction in China (Qiao and Feng, 2014). Several limitations should be considered when interpreting the findings: First, the trial was conducted using an IVM protocol with no stimulation in a single centre, which may limit its generalizability, as IVM protocols with stimulation are widely available in some settings. Second, the rate of patients declining participation was high, which might impact the representation of the participants. Third, the timeframe set for the original primary outcome when the trial was designed was too short as 24 women did not have their first transfer until 3–6 months post-randomization. We therefore adjusted primary outcome included all transfers within 6 months, which refers to an outcome change, but it is more clinically relevant. Nevertheless, this change in primary outcome did not affect the conclusion. Finally, a GnRH agonist trigger rather than hCG in PCOS for IVF stimulation would be more consistent with current clinical practice.

It is worth noting that the IVM protocol in this trial did not involve the use of gonadotrophins or hCG priming. Although we acknowledge the ongoing debate on the definition of IVM, strictly speaking, the term IVM only refers to the maturation in vitro of immature COCs collected from antral follicles (De Vos et al., 2016). Therefore, the use of exogenous LH or hCG before oocyte retrieval to initiate the maturation process in vivo may not be considered IVM. Applying short FSH priming in women with PCOS undergoing IVM sometimes appears to improve implantation and pregnancy rates (Mikkelsen et al., 2001; Wynn et al., 1998), however, other studies (Trounson et al., 1998; Son et al., 2006) have not demonstrated any difference in the fertility outcome. Thus the results from these studies of FSH priming were conflicting. We adopted an unstimulated IVM protocol in our trial.

A limited number of retrospective observational studies comparing IVM and IVF have reported similar inferior pregnancy outcomes in the IVM group (Gremeau et al., 2012; Das et al., 2014; Walls et al., 2015), with a live birth rate per started cycle ranging from 16.5% to 33.9% in the IVM group. Compared to IVF with ovarian stimulation and ovulation triggering, the lower success rate of IVM could be explained by decreased maturation and a lower number of MII oocytes. In our trial, we observed a maturation rate of 45.7% after 28–32 h of culture in IVM, consistent with other studies of IVM with no hCG priming (Gremeau et al., 2012; Walls et al., 2015; Braam et al., 2019; Lin et al., 2020), significantly lower than that of IVF cycle with ovarian stimulation. Although fertilization and cleavage rates were comparable, the mean numbers of 2PN, good-quality D3 embryos and usable blastocysts were lower in the IVM group. Extending the culture time may increase the maturation rate. In our study, 323 immature oocytes after denudation reached maturity within 32–48 h of incubation, of which 86 MII oocytes were injected by ICSI, and 40 were fertilized, but none developed into useable blastocysts. This result agreed with Son’s report (Son et al., 2005) that oocytes reaching MII faster in an IVM program have better embryonic developmental competence. Another possible explanation of the lower success was that embryo development potential derived from IVM was poorer than that derived from in  vivo maturation. Asynchronous maturation of the nucleus and cytoplasm in IVM oocytes may be a key factor. Compared with mature oocytes obtained in vivo, there is a significant decline in the rate of blastocyst formation, implantation, pregnancy and live birth of IVM oocytes despite similar fertilization rates (Lin et al., 2003; Söderström-Anttila et al., 2005; Mikkelsen and Lindenberg, 2001; Le Du et al., 2005). A longer time to ongoing pregnancy leading to live birth in the IVM group could be due to more women with no transferable embryos in the IVM group compared to the IVF group (62 and 6, respectively). Other reports have showed increased ratios of abnormal embryo development and stagnation (Suikkari, 2008; Das et al., 2014), altered genes reflecting oocyte competence (Guzman et al., 2013), or increased ratios of embryo arrest (Walls et al., 2015; Roesner et al., 2017), all contributing to a decreased number of blastocysts in IVM cycles. The effect of ovarian stimulation and triggering could also be a potential reason for differences between groups. The use of gonadotropins in IVF cycle induced multiple follicular development and obtained more high-quality matured oocytes; therefore future trials comparing IVM and IVF with the same ovulation triggering agent are needed.

Blastocyst formation has been the key limiting process in embryo-derived IVM technology development. There were 68 patients in our trial who had no embryos or blastocysts acquired for transfer, including 62 cases in the IVM group and 6 cases in the IVF group. However, if IVM-derived embryos developed into useable blastocysts, they had a comparable implantation potential (41.1%) as embryos obtained from the standard IVF procedure (50.7%, Table V). We observed that the rates of ongoing pregnancy leading to live birth were comparable between the IVM group (28.0%) and the IVF group (37.5%) after the first single blastocyst transfer (Table V). IVM protocols with a pre-maturation step have been evaluated in other studies. Ho et al. (2018) demonstrated that IVM has the potential to produce live births comparable to those of conventional IVF but a recent trial from the same group did not demonstrate non-inferiority of IVM (Vuong et al., 2020).

Table V.

Outcomes after only the first single blastocyst transfer.

	IVM group （N = 107）	IVF group （N = 152）	Rate difference (95% CI)	Rate ratio in IVM group (95% CI)	P-value
Primary outcome
Live birth—no. (%)‖	30 (28.0)	57 (37.5)	−0.07 (−0.19 to 0.05)	0.75 (0.52 to 1.03)	0.112
Secondary pregnancy outcomes
Conception—no. (%)†	54 (50.5)	87 (57.2)	−0.07 (−0.19 to 0.06)	0.88 (0.70 to 1.11)	0.281
Clinical pregnancy—no./total no. (%)‡	44 (41.1)	77 (50.7)	−0.10 (−0.21 to 0.02)	0.81 (0.62 to 1.07)	0.130
Singleton	44	77
Twin	0	0
Implantation (per embryo)— no./total no. (%)§	44/107 (41.1)	77/152 (50.7)	−0.10 (−0.21 to 0.02)	0.81 (0.62 to 1.07)	0.130
Ongoing pregnancy—no./total no. (%)¶	32/107 (29.9)	62/152 (40.8)	−0.11 (−0.22 to 0.01)	0.73 (0.52 to 1.04)	0.073
Pregnancy complications
Ectopic pregnancy—no/total no. (%)	0/107 (0.0)	0/152 (0.0)
Pregnancy loss—no/total no. (%)
Among biochemical pregnancy— no./total no. (%)	10/54 (18.5)	10/87 (12.4)	7.02 (−5.32 to 19.36)	1.61 (0.72 to 3.61)	0.245
Among clinical pregnancy— no./total no. (%)	16/44 (36.4)	21/77 (27.2)	9.09 (−8.26 to 26.44)	1.33 (0.78 to 2.27)	0.296
First trimester pregnancy loss	12/44 (27.3)	15/77 (19.5)	7.79 (−8.06 to 23.65)	1.40 (0.72 to 2.72)	0.322
Second trimester pregnancy loss	4/44 (9.1)	6/77 (7.8)	1.30 (−9.09 to 11.69)	1.17 (0.34 to 3.91)	0.803

^‖‖Live birth was defined as the detection of a viable foetus with a foetal heartbeat at 22 weeks’ gestation.

^†Conception was defined as serum hCG ≥ 5 mIU/ml.

^‡Clinical pregnancy was defined as detection of a gestational sac in the uterine cavity.

^§The implantation rate was the number of gestational sacs divided by the number of embryos that were transferred.

^¶Ongoing pregnancy was defined as the detection of a viable foetus with foetal heartbeat at 12 weeks’ gestation.

IVM was introduced into clinical practice due to its safety and low costs. Our results confirmed the safety of IVM in terms of avoiding OHSS. In the IVF group, hCG triggering was with 250 µg of recombinant (r) hCG (Ovidrel), and moderate-severe OHSS occurred in 6.3% of women despite the use of the freeze-all strategy in a standard GnRH-antagonist stimulation protocol. This was partly due to the use of hCG triggering and could have been further reduced if a GnRH-agonist trigger were used. In contrast, no case of moderate-severe OHSS was observed in the IVM group. Moreover, only an average of three outpatient visits per woman were required for monitoring follicular development in the IVM group, which was 50% less than that in the IVF group. The average treatment cost was reduced by nearly one third in the IVM group. Concerning women with no transferable blastocysts in the IVM group, it would be feasible to restart the next IVM cycle immediately as no stimulatory drugs were used. Given that a large proportion of the IVM arm did not reach an embryo transfer, the comparison of IVM and IVF in a single cycle within a fixed duration may be considered suboptimal. Future trials comparing multiple cycles of IVM versus one cycle of IVF, with accompanying cost-effective analyses in different settings would a better insight into clinical practice.

Current evidence on neonatal and long-term outcomes of offspring following IVM is limited. In a large multicentre report involving 1421 IVM babies in 22 countries, 18 babies showed major birth (Qiao and Feng, 2014) defects, similar to the corresponding proportion among babies resulting from spontaneous pregnancy (Chian et al., 2014). A recent single-centre cohort study showed no significant difference in 2-year anthropometry and health outcomes among offspring born following IVM versus those born following ICSI (Belva et al., 2020). Our trial did not show increased neonatal complications in the IVM group, although it was not powered for these outcomes. Further follow-up studies are needed to provide evidence on the long-term safety of IVM.

IVM is limited in clinical application possibly due to the low maturity rate of immature oocytes in vitro and subsequent low cumulative pregnancy outcomes. It has been reported that the addition of cytokines, hormones and other small molecular substances to the IVM system may improve the developmental potential of immature oocytes (Yu et al., 2012; Li et al., 2015; Gilchrist et al., 2016; Stocker et al., 2020). Additionally, human cumulus-enclosed germinal vesicle oocytes from early antral follicles reveal heterogeneous cellular features, and immature oocytes derived from smaller follicles (<6 mm) have lower maturation rates (Guzman et al., 2012; Sanchez et al., 2015), which could be compensated with a prematuration culture system (Sanchez et al., 2017, 2019). Future research should explore the optimal criteria for screening PCOS patients, the performance of short FSH priming, the timing of immature oocyte retrieval and embryo transfer, and even the type of endometrial preparation that is warranted in the IVM strategy in order to improve outcomes.

Conclusions

In summary, we found IVM without ovarian stimulation to be inferior to conventional IVF with ovarian stimulation for women with infertility and PCOS. Given the convenience, inexpensiveness and safety of IVM, further development followed by further RCTs are needed to evaluate the effectiveness and safety of other IVM protocols or multiple cycles of IVM compared to IVF.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Authors’ roles

J.Q. and R.L. conceived the study idea. J.Q., R.L., R.W., B.W.M., X.Z. and W.G. participated in the design of the study. X.Z., W.G., S.Y., Y.X. and L.W. participated in the recruitment of participants and the assessment of clinical outcomes. L.Z. and D.Z. coordinated the data collection and performed the analysis. X.Z. and W.G. wrote the first draft of this manuscript. L.Z., D.Z., S.Y., Y.X., L.W., R.W., B.W.M., R.L. and J.Q. were involved in the critical revision of the manuscript. All authors approved the final manuscript.

Funding

This study was supported by the National Key Research and Development Program of China (2016YFC1000201 and 2018YFC1002104) and the National Science Foundation of China (81730038). The study funders had no role in the study design, implementation, analysis, manuscript preparation or decision to submit this article for publication.

Competing interests

B.W.M. was supported by a NHMRC Investigator (GNT1176437). All other authors declare no competing interests.