Effect of the oxytocin receptor antagonist nolasiban on pregnancy rates in women undergoing embryo transfer following IVF: analysis of three randomised clinical trials

Abstract

STUDY QUESTION

Does a single oral dose of nolasiban 900 mg administered 4 h before embryo transfer (ET) increase pregnancy rates in women undergoing IVF?

SUMMARY ANSWER

In an individual patient data (IPD) meta-analysis of three clinical trials, a single oral dose of nolasiban 900 mg was associated with an increased ongoing pregnancy rate of an absolute 5% (relative 15%).

WHAT IS KNOWN ALREADY

Several clinical studies have shown that blocking activation of oxytocin receptors by an oxytocin receptor (OTR) antagonist has the potential to decrease uterine contractions, increase endometrial perfusion and enhance endometrial decidualisation and other parameters of endometrial receptivity. It has been hypothesised that antagonism of oxytocin receptors could improve the likelihood of successful embryo implantation and thus increase pregnancy and live birth rates following ET.

STUDY DESIGN, SIZE, DURATION

This is an analysis of three randomised, double-blind, placebo-controlled trials, which randomised 1836 subjects between 2015 and 2019. We describe the results of a meta-analysis of individual participant data (IPD) from all three trials and the pre-specified analyses of each individual trial.

PARTICIPANT/MATERIAL, SETTING, METHODS

Participants were patients undergoing ET following IVF/ICSI in 60 fertility centres in 11 European countries. Study subjects were below 38 years old and had no more than one previously failed cycle. They were randomised to a single oral dose of nolasiban 900 mg (n = 846) or placebo (n = 864). In IMPLANT 1, additional participants were also randomised to nolasiban 100 mg (n = 62) or 300 mg (n = 60). Fresh ET of one good quality embryo (except in IMPLANT 1 where transfer of two embryos was allowed) was performed on Day 3 or Day 5 after oocyte retrieval, approximately 4 h after receiving the study treatment. Serum hCG levels were collected at 14 days post oocyte retrieval (Week 2) and for women with a positive hCG result, ultrasound was performed at Week 6 post-ET (clinical pregnancy) and at Week 10 post-ET (ongoing pregnancy). Pregnant patients were followed for maternal (adverse events), obstetric (live birth, gestational age at delivery, type of delivery, incidence of twins) and neonatal (sex, weight, height, head circumference, Apgar scores, congenital anomalies, breast feeding, admission to intensive care and specific morbidities e.g. jaundice, respiratory distress syndrome) outcomes.

MAIN RESULTS AND THE ROLE OF CHANCE

In an IPD meta-analysis of the clinical trials, a single oral dose of nolasiban 900 mg was associated with an absolute increase of 5.0% (95% CI 0.5, 9.6) in ongoing pregnancy rate and a corresponding increase of 4.4% (95% CI −0.10, 8.93) in live birth rate compared to placebo. Similar magnitude increases were observed for D3 or D5 transfers but were not significantly different from the placebo. Population pharmacokinetics (PK) demonstrated a correlation between higher exposures and pregnancy.

LIMITATIONS, REASON FOR CAUTION

The meta-analysis was not a pre-specified analysis. While the individual trials did not show a consistent significant effect, they were not powered based on an absolute increase of 5% in ongoing pregnancy rate. Only a single dose of up to 900 mg nolasiban was administered in the clinical trials; higher doses or extended regimens have not been tested. Only fresh ET has been assessed in the clinical trials to date.

WIDER IMPLICATIONS OF THE FINDINGS

The finding support the hypothesis that oxytocin receptor antagonism at the time of ET can increase pregnancy rates following IVF. The overall clinical and population PK data support future evaluation of higher doses and/or alternate regimens of nolasiban in women undergoing ET following IVF.

STUDY FUNDING/COMPETING INTERESTS

The trials were designed, conducted and funded by ObsEva SA. A.H., O.P., E.G., E.L. are employees and stockholders of ObsEva SA. E.L. is a board member of ObsEva SA. G.G. reports honoraria and/or non-financial support from ObsEva, Merck, MSD, Ferring, Abbott, Gedeon-Richter, Theramex, Guerbet, Finox, Biosilu, Preglem and ReprodWissen GmbH. C.B. reports grants and honoraria from ObsEva, Ferring, Abbott, Gedeon Richter and MSD. P.P. reports consulting fees from ObsEva. H.T. reports grants and or fees from ObsEva, Research Fund of Flanders, Cook, MSD, Roche, Gedeon Richter, Abbott, Theramex and Ferring. H.V. reports grants from ObsEva and non-financial support from Ferring. P.T. is an employee of Cytel Inc., who provides statistical services to ObsEva. J.D. reports consulting fees and other payments from ObsEva and, Scientific Advisory Board membership of ObsEva.

TRIAL REGISTRATION NUMBERS

ClinicalTrials.gov: NCT02310802, NCT03081208, NCT03758885

TRIAL REGISTRATION DATES

December 2014 (NCT02310802), March 2017 (NCT03081208), November 2018 (NCT03758885)

FIRST PATIENT’s ENROLMENT

January 2015 (NCT02310802), March 2017 (NCT03081208), November 2018 (NCT03758885).

IVF, embryo transfer, ongoing pregnancy, live birth, nolasiban, oxytocin receptor antagonist

Introduction

Infertility and subfertility are public health issues affecting more than 10% of women worldwide. Assisted reproductive technologies (ARTs) such as IVF have given individuals and couples with infertility access to invaluable treatment options, with over 7 million children born following IVF (Berntsen et al., 2019). Despite many important advances over time, IVF success rates remain modest, yielding live birth rates of approximately 25–30% per initiated cycle (Luke et al., 2012; CDC, 2018; Toftager et al., 2017). The likelihood of live birth following IVF is increased with transfer of multiple embryos; however, widespread application of multiple embryo transfer (ET) led to high rates of multiple gestation, which is associated with increased risk of poor outcomes mainly due to short- and long-term complications associated with preterm delivery (Berntsen et al., 2019). For this reason, consensus has increasingly shifted in favour of single ET (SET) in many countries (Berntsen et al., 2019). However, according to the US CDC National Summary Report for 2016, about 60% of all fresh transfers from non-donor oocytes still involved the transfer of multiple embryos resulting in a 20% rate of multiple infant live births (CDC, 2018). Improvement in pregnancy outcomes following single ET is therefore important.

Key factors known to influence the success of ET after IVF include maternal age, embryo quality, ET technique and operator experience, the number of embryos transferred and, ultimately, uterine receptivity to implantation (Decleer et al., 2012). Many interventions aimed at improving the likelihood of successful implantation have been attempted, including adherence compounds (EmbryoGlue), Intralipid infusion, endometrial scratching and endometrial receptivity arrays. A recent review found that there was no robust evidence to support an increase in live birth rate for any intervention except potentially for endometrial scratching (Heneghan et al., 2016). However, the benefit of endometrial scratching was not confirmed in a recent large randomised trial (Lensen et al., 2019).

The rate of uterine contractions at the time of ET is negatively correlated with pregnancy rates after IVF (Fanchin et al., 1998; Zhu et al., 2014), and endometrial blood flow on the day of ET has been shown to be an important predictive parameter for endometrial receptivity (Kim et al., 2014). Oxytocin receptors are expressed during the human peri-implantation phase in the myometrium, endometrium and uterine blood vessels, and several clinical studies have shown that blocking activation of oxytocin receptors by an oxytocin receptor (OTR) antagonist has the potential to decrease uterine contractions and increase endometrial perfusion (Pierzynski et al., 2007; Pierzynski 2011; Kalmantis et al., 2012). Furthermore, Sztachelska showed evidence of improvement of enhanced endometrial decidualisation and other parameters of endometrial receptivity via oxytocin antagonism (Sztachelska et al., 2019). As such, it has been hypothesised that antagonism of oxytocin receptors could improve the chances of successful implantation and thus increase pregnancy rates following ET. However, currently there is no proven treatment available for administration at the time of ET to increase the likelihood of implantation and subsequent live birth following IVF.

Nolasiban is a novel investigational, competitive and reversible OTR. Together with being orally active and having higher specificity for the oxytocin receptor than the intravenously administered OTR antagonist, atosiban (approved in Europe for the treatment of preterm labor), nolasiban’s half-life of approximately 12 h allows for maintained exposure for up to 2 days (versus atosiban which has a half-life of 13 min) following a single oral dose (Kim et al., 2017). Supporting its potential impact on the uterine environment at the time of ET, in-vitro preclinical models using strips of human myometrium have shown that nolasiban suppresses spontaneous contractions (Kim et al., 2017). In Phase 1 studies, nolasiban was shown to have linear pharmacokinetics (PK), and was well tolerated up to a 2400 mg single dose (Data on file).

Three randomised, placebo-controlled trials, a single Phase 2 dose-ranging trial (IMPLANT 1) (Tournaye et al., 2017) and two large double-blind, randomised, placebo-controlled Phase 3 trials (IMPLANT 2 and IMPLANT 4) were conducted to investigate whether treatment with nolasiban increased the probability of ongoing pregnancy in women undergoing fresh SET. A meta-analysis of data from the three clinical trials was performed to estimate the overall efficacy of nolasiban. To assess for potential subgroups in which nolasiban performed better than in the overall study population, subgroup identification analyses were conducted on the Phase 3 datasets (Lipkovich et al., 2011). Finally, using data from Phase 1 studies and IMPLANT 4 participants, a preliminary PK model was developed to assess for potential interaction between nolasiban exposure and likelihood of pregnancy.

This report provides an overview of the clinical development of nolasiban to date and the totality of the evidence of its potential for use in the improvement of IVF outcomes following ET.

Materials and methods

Trial designs and populations

Clinical trials

IMPLANT 1 was a randomised, double-blind, parallel group, dose-ranging trial comparing single oral doses of 100, 300 and 900 mg nolasiban to placebo in women undergoing ET following IVF or ICSI. IMPLANT 2 and IMPLANT 4 were randomised, double-blind, parallel group, Phase 3 trials comparing a single oral dose of 900 mg nolasiban to placebo. All treatments were administered 4 h before ET. An additional trial, IMPLANT 3, similar to IMPLANT 4 had been planned in the USA, but was cancelled prior to enrolment.

The trials were conducted in 60 fertility centres in 11 European countries and Canada between 2015 and 2019 (Supplementary Table SI). All clinical protocols were approved by an ethics committee at each participating centre and applied in accordance with International Conference on Harmonization guidelines, along with the applicable regulations and ethical principles of the Declaration of Helsinki. All the women provided written informed consent. The clinical trials were registered on ClinicalTrials.gov as NCT02310802 (IMPLANT 1), NCT03081208 (IMPLANT 2) and NCT03758885 (IMPLANT 4). The data underlying this article will be shared on reasonable request to the corresponding author.

The study populations of all three trials were similar. Women were eligible if they were scheduled for IVF or ICSI, followed by fresh ET. They had to be 37 years of age or younger, have undergone no more than one previous failed stimulation cycle, and have at least one good quality embryo available for transfer. In IMPLANT 1 only, subjects had to have evidence of at least 1.5 uterine contractions/min on transvaginal ultrasound video captured at baseline on the day of ET, prior to nolasiban/placebo administration. Uterine contractions were assessed by manual reading of an accelerated (×20) 6-min video of the transvaginal ultrasound of the sagittal plane of the uterus.

In IMPLANT 2 and IMPLANT 4, women were excluded if they had serum progesterone (P4) levels >1.5 ng/ml (4.7 nmol/l) on the day of ovulation trigger. Women were also excluded if they produced more than 20 oocytes, had known severe endometriosis (stage III or IV according to the revised ASRM classification) (American Society for Reproductive Medicine, 1997) and/or adenomyosis or a history of recurrent pregnancy loss, or required their partner to undergo testicular extraction to obtain sperm. Uterine contractions were not assessed and therefore not used as an eligibility criterion. A comparison of inclusion/exclusion criteria across the trials is presented in Supplementary Table SII.

IMPLANT 1 was conducted in women undergoing ET on D3 post oocyte retrieval and single or double ET was allowed. IMPLANT 2 was conducted in women undergoing SET on either D3 or D5 post oocyte retieval; enrolment was capped for D3 and D5 to ensure equal numbers in each group. IMPLANT 4 was conducted in women undergoing SET on D5 post oocyte retrieval only.

In all three trials, patients were subjected to ovarian stimulation using human gonadotropins and a gonadotropin-releasing hormone antagonist, followed by a single injection of hCG to trigger ovulation. Luteal support was administered vaginally using 200 mg natural micronised P4 three times a day, starting within 24 h after oocyte retrieval until at least 6 weeks after ET, unless the pregnancy test was negative after 14 days. One (or two for some IMPLANT 1 patients) good-quality embryo was transferred approximately 4 h after nolasiban treatment. Embryo quality was graded according to the Istanbul Consensus (Alpha Scientists, 2011) for D3 embryos and the Gardner grading scale (>3BB) for D5 embryos (Gardner, 2015). ET was performed by a trained physician using a soft catheter and ultrasound guidance. Difficult transfers, for example those requiring a switch to a hard catheter, were recorded. Serum hCG levels were collected at 14 days post oocyte retrieval (Week 2) and for women with a positive hCG result, ultrasound was performed at Week 6 post-ET (to confirm clinical pregnancy) and at Week 10 post-ET (to confirm ongoing pregnancy). Pregnant patients were followed for pregnancy outcomes. The IMPLANT 4 trial is ongoing; the primary endpoint of ongoing pregnancy as well as the live birth and maternal, obstetric and neonatal follow-up are reported here.

The treatment allocation process was managed via an interactive web response system (IWRS). Prior to the start of the study, a randomisation list was generated by a designated statistician then transmitted to the assigned clinical packaging organisation for labelling and to the integrated IWRS. The randomisation list was secured in an electronic file with access restricted to only the designated personnel directly responsible for labelling and handling the study medication and the randomisation list provider until the study database was locked and ready to be unblinded. Blinded treatment kits were sent to the site and kept in controlled conditions. Once a patient was confirmed as eligible with at least one good quality embryo to be transferred, the eligibility was confirmed in the IWRS, which then allocated a treatment kit number with the correct randomised treatment for the next patient at that site. This treatment kit number corresponded to one of the blinded treatment kits kept at the site. Placebo treatments were identical in appearance to the active treatments. Treatment allocation was blinded to patients, site staff, investigators, study management and the Sponsor.

In all three trials, randomisation was done after eligibility was confirmed on the day of ET and subjects were randomised in blocks of pre-determined length. In IMPLANT 1, subjects were randomised to one of four treatment groups (100, 300, 900 mg nolasiban or placebo) in a 1:1:1:1 ratio, stratified by site. In IMPLANT 2, patients were randomised in a 1:1 ratio to 900 mg nolasiban or placebo, stratified by site and enrolment was capped for D3 and for D5 to ensure there were equal proportions of D3 or D5 transfers recruited overall. In IMPLANT 4, patients were randomised in a 1:1 ratio to 900 mg nolasiban or placebo, stratified by site, age (<35, >35 years) and weight (<70, >70 kg).

Efficacy outcomes

Pregnancy outcomes included positive pregnancy test (hCG ≥10 IU 14 days after oocyte retrieval), clinical and ongoing pregnancy (intrauterine gestational sac with a heartbeat on ultrasound (US) 6 and 10 weeks after ET, respectively), live birth (live delivery after 24 weeks of gestation), and pregnancy loss (any positive pregnancy test that was not followed by a live birth). The primary endpoint in IMPLANT 1 was clinical pregnancy. Ongoing pregnancy was selected as the primary endpoint for IMPLANT 2 and IMPLANT 4 as it is objective and strongly predicts live birth. Post-treatment uterine contraction frequency was an outcome in the IMPLANT 1 study only.

Statistical analyses

IMPLANT 1

A target sample size of 240 (60 in each treatment group) was calculated to provide 80% power to detect a linear trend in clinical pregnancy rates, assuming 20% in the placebo group and up to 40% at the highest dose. The primary analysis was a test for a linear trend of clinical pregnancy rates across increasing levels of dosage using a Cochran-Armitage test. Secondary analyses were conducted by fitting a logistic regression model with treatment to assess whether the slope was equal to zero. Individual dose versus placebo comparisons were tested via Fisher’s exact test and as contrasts within the logistic regression model. Secondary endpoints included ongoing pregnancy rate at 10 weeks post-oocyte retrieval, LBR and uterine contractions just prior to ET. Additional analyses exploring the dose response were done using a subset of subjects excluding all patients with serum P4 in the upper quartile on the day of ET and on the subset of subjects who had SET.

IMPLANT 2

A target sample size of 760 (380 per treatment group) was calculated to provide approximately 90% power to detect a common odds ratio (OR) of 1.63 or greater for 900 mg to placebo. The assumptions included a Day 3 ongoing pregnancy rate of 29.3% for placebo (based on IMPLANT 1 results) and 40.3% for nolasiban 900 mg and a Day 5 rate of 37.2% for placebo (Glujovsky et al., 2016) and 49.1% for 900 mg. The analyses of the primary efficacy endpoint of ongoing pregnancy and the key secondary endpoint of live birth were performed on the overall population by applying an exact logistic regression model, with treatment as the independent variable, the day of ET (D3 or D5) as a covariate, and the clinical sites as strata. Sites that had all positive or all negative pregnancy outcomes were pooled prior to analysis in order to include all patients in the analysis. Separate exact logistic regression models were prospectively planned for the two (D3 and D5) transfer sub-groups. In all analyses, a 2-sided type I error rate of 0.05 was applied to ascertain whether the difference between the treatment groups was statistically significant.

IMPLANT 4

A target sample size of 820 subjects for IMPLANT 4 was calculated, based on the IMPLANT 2 D5 results, to provide approximately 90% power to detect a common OR of 1.59 or greater for nolasiban 900 mg compared to placebo, with an assumed placebo pregnancy rate of 35%. The analyses of the IMPLANT 4 primary efficacy endpoint (ongoing pregnancy) was performed by fitting a stratified logistic regression model with treatment as the independent variable, age (<35, ≥35 years) and weight (<70, ≥70 kg) as covariates, and the clinical sites as strata. Sites that only had all positive or all negative pregnancy outcomes were pooled prior to analysis in order to include all patients in the analysis. In addition, common risk differences between nolasiban and placebo were calculated using a Cochran-Mantel-Haenszel (CMH) test, with age and weight categories as stratification factors.

Meta-analysis of ongoing pregnancy and live birth

The meta-analysis of 10-week ongoing pregnancy and live birth comparing patients who received 900 mg nolasiban versus placebo across IMPLANT 1, IMPLANT 2 and IMPLANT 4 was performed via a one-stage approach using individual patient data (IPD) based on a Cochran-Mantel-Haenszel test and using the variables study and day of ET as the stratification factor (four strata: IMPLANT 1 Day 3, IMPLANT 2 Day 3, IMPLANT 2 Day 5 and IMPLANT 4 Day 5). Mantel-Haenszel estimates of the odds ratio and the common risk difference between nolasiban and placebo were calculated, along with stratified Newcombe confidence limits. A Breslow-Day test was conducted to assess for homogeneity of the four strata.

Additional Cochran-Mantel-Haenszel tests using the variable day of ET were performed separately on the pooled individual subject Day 3 (from IMPLANT 1 and IMPLANT 2) and Day 5 data (from IMPLANT 2 and IMPLANT 4) using site as stratification factor. A Breslow-Day test was conducted to test for homogeneity of the studies at each day of transfer (D3 and D5).

SIDES analysis

In an effort to identify potential subgroups in which nolasiban performed better than in the overall study population with regard to ongoing pregnancy at week 10, a statistical method known as Subgroup Identification Based on Differential Effect Search (SIDES) analysis was conducted on IMPLANT 2 and 4 data (Lipkovich et al., 2011) as an exploratory analysis, knowing that such analyses can inflate type I error rates. SIDES was applied to data from IMPLANT 4 to identify potential subgroups with larger nolasiban treatment effect compared to placebo. As a form of cross-validation, any identified subgroups were tested against IMPLANT 2 data. The same process was repeated for IMPLANT 2 data, with testing of any identified subgroups against IMPLANT 4 data. Input parameters included, among others, baseline characteristics such as subject age, weight, BMD, primary versus secondary infertility, type of infertility, medical history (e.g. endometriosis), method of fertilisation (IVF versus ICSI), number of good quality embryos, embryo grade, E2 and P4 levels and total FSH dose.

Population PK-(IMPLANT 4)

A PK exposure model was developed based on two Phase 1 studies in healthy volunteers, in which PK samples were frequently obtained: Study 18-OBE001-003 was an open label, single-dose, two part study designed to assess the mass balance recovery, metabolite profile, metabolite identification and absolute bioavailability of [¹⁴C]-Nolasiban (EudraCT: 2018-002645-10) and study 18-OBE001-004 was a Phase 1, double-blind, randomised, placebo-controlled, parallel study to assess the safety and PK/PD effects of nolasiban (EudraCT: 2018-003702-36). The IMPLANT 4 trial collected sparse pharmacokinetic samples approximately at 3.5 h and, for the majority of subjects, also at 5 and ≥7 h (up to 72 h) after nolasiban administration. The data from sparse sampling was fitted to the PK model, in order to estimate individual peak and overall exposures for IMPLANT 4 patients. The estimated values were then used to assess for relationships between nolasiban exposure and pregnancy rates in IMPLANT 4.

Safety outcomes

Ongoing pregnancies were followed until delivery and neonates for 28 days after birth. Follow-up included maternal (adverse events), obstetric (live birth, gestational age at delivery, type of delivery, incidence of twins) and neonatal (sex, weight, height, head circumference, Apgar scores, congenital anomalies, breast feeding, admission to intensive care and specific morbidities e.g. jaundice, respiratory distress syndrome) outcomes.

Results

Baseline characteristics

In IMPLANT 1, 430 patients were screened (from January to June 2015), 247 were randomised, received trial medication and were included in the full analysis set (FAS) for efficacy. In IMPLANT 2, 1,103 patients were screened (from March to September 2017), 779 were randomised and 778 received trial treatment and were included in the FAS for efficacy. In IMPLANT 4, 1264 patients were screened (from November 2018 to June 2019), 810 were randomised and 807 received trial treatment and were included in the FAS for efficacy. The actual numbers randomised in each trial were slightly different to the sample size targets (IMPLANT 1: 247 vs. 240; IMPLANT 2: 779 vs. 780; IMPLANT 4: 810 vs. 820) due to the very high rate of recruitment and the difficulty in predicting exactly how many patients entering into screening would ultimately be randomised. Details of screening, randomisation and inclusion in the analysis sets are provided in Fig. 1. The numbers of patients randomised at each of the participating sites are shown in Supplementary Table SI.

Figure 1.

Study enrolment and disposition.

Demographics and baseline characteristics of participants in each trial were similar (Table I). The mean age of women in all three studies was 31 years and mean BMI was ∼24 kg/m². The mean number of oocytes retrieved was between 9 and 11, and the mean number of good quality embryos ranged from 2.5 to 3.8. In IMPLANT 1, single ET was performed in 61% of subjects and two embryos were transferred in 39% of subjects; these proportions were consistent across the treatment groups. Overall, 2.7% of transfers were classed as difficult transfers.

Table I

Demographic and clinical characteristics of the patients at baseline.^*

	IMPLANT 1		IMPLANT 2		IMPLANT 4
Characteristic	Placebo	Nolasiban 900 mg	Placebo	Nolasiban 900 mg	Placebo	Nolasiban 900 mg
	n = 65	n = 60	n = 390	n = 388	n = 409	n = 398
Age (y)	31.5 ± 3.3	31.1 ± 3.3	31.4 ± 3.2	31.1 ± 3.3	31.5 ± 3.4	31.4 ± 3.5
Race (% White)	95.4%	93.3%	99.5%	99.0%	99.3%	99.2%
Body mass index (kg/m2)	23.4 ± 4.1	23.7 ± 4.0	23.8 ± 4.3	24.0 ± 4.4	23.7 ± 4.0	23.7 ± 3.7
Primary Infertility, n (%)	NC	NC	263 (67.4)	267 (68.8)	294 (71.9)	282 (70.9)
Type of infertility, n (%)†	NC	NC
Unexplained			86 (22.1)	90 (23.2)	108 (26.4)	120 (30.2)
Male only			122 (31.3)	131 (33.8)	153 (37.4)	133 (33.4)
Female			182 (46.7)	167 (43.0)	148 (36.2)	145 (36.4)
Tubal factor			73 (18.7)	78 (20.1)	71 (17.4)	77 (19.3)
Ovulation dysfunction			53 (13.6)	42 (10.8)	30 (7.3)	33 (8.3)
Endometriosis			35 (9.0)	40 (10.3)	50 (12.2)	35 (8.8)
Other			75 (19.2)	57 (14.7)	23 (15.5)	14 (9.5)
Duration of stimulation (d)	9.0 (1.9)	8.9 (1.4)	9.4 ± 1.8	9.4 ± 1.8	9.4 ± 1.7	9.9 ± 3.7
Gonadotrophin total dose (IU)	1768 ± 814	1703 ± 540	1846 ± 645	1870 ± 650	1724 ± 721	1671 ± 646
Fertilisation method, n (%)
IVF	6 (9.2)	6 (10.0)	35 (9.0)	44 (11.3)	58 (14.2)	63 (15.8)
ICSI	59 (90.8)	54 (90.0)	355 (91.0)	344 (88.7)	351 (85.8)	335 (84.2)
Estradiol level on day of hCG trigger (nmol/L)	NC	NC	6653 ± 4183	7383 ± 4205	6610 ± 2937	6622 ± 3026
Progesterone level on day of hCG trigger (nmol/L)	NC	NC	1.97 ± 1.15	1.88 ± 0.99	2.23 ± 3.17	2.01 ± 1.05
No. of oocytes retrieved (n)	11.0 ± 5.2	10.2 ± 4.1	9.4 ± 4.4	9.7 ± 4.3	10.0 ± 4.5	9.8 ± 4.5
No. good quality embryos (n)†	3.7 ± 2.3	3.5 ± 2.9	2.5 ± 2.0	2.6 ± 2.0	2.4 ± 1.8	2.5 ± 1.8

*

Plus–minus values are mean ± SD. Percentages may not total 100 because of rounding. NC denotes not collected.

^†

More than one reason for female infertility could be reported for each patient therefore percentages do not add up to 100. “Other” included oligo-amenorrhoea, diminishing ovarian reserve, normal menstrual cycle with luteal defect, uterine factor, secondary amenorrhoea, uni- or bilateral hydrosalpinx and primary amenorrhoea.

‡

Good quality embryos on D3 were defined as having 6 to 10 blastomeres with sizes consistent with the stage of division, absence of multinucleation, and less than 10% granularity in the cytoplasm of the blastomeres. Good quality embryos on D5 were defined as reaching Gardner development Grade 3 and have inner cell mass and trophectoderm scores of A or B and there had to have been no multinucleated blastomeres observed at the 4-cell stage.

Efficacy endpoints by trial

Ongoing pregnancy, live birth, clinical pregnancy, positive pregnancy test and pregnancy loss rates in the individual clinical trials for nolasiban 900 mg and placebo are shown in Table II.

Table II

Pregnancy, live birth and pregnancy loss outcomes per trial for nolasiban 900 mg and placebo.^*

Variable	Placebo	Nolasiban 900 mg	Difference (95% CI)†	Odds ratio (95% CI) ‡	P-value‡
IMPLANT 1, n (%)	N = 65	N = 60
Positive pregnancy test	33 (50.8)	32 (53.3)	2.6 (−12.3, 17.4)	1.11 (0.61, 2.00)	0.774
Clinical pregnancy	22 (33.8)	28 (46.7)	12.8 (−2.1, 27.3)	1.71 (0.93, 3.13)	0.145
Ongoing pregnancy	19 (29.2)	27 (45.0)	15.8 (1.0, 30.2)	1.98 (1.07, 3.68)	0.069
Live birth delivery	19 (29.2)	26 (43.3)	14.1 (−0.7, 28.6)	1.85 (1.00, 3.44)	0.102
Pregnancy loss§	14 (42.4)	6 (18.8)	−23.7 (−46.7, −0.4)	0.32 (0.08, 1.08)	0.059
IMPLANT 2
Overall population, n (%)	N = 390	N = 388
Positive pregnancy test	155 (39.7)	175 (45.1)	5.4 (−1.8, 12.3)	1.26 (0.93, 1.71)	0.134
Clinical pregnancy	114 (29.2)	145 (37.4)	8.1 (1.1, 15.1)	1.50 (1.08; 2.08)	0.014
Ongoing pregnancy	111 (28.5)	138 (35.6)	7.1 (0.04, 14.1)	1.43 (1.03, 1.99)	0.031
Live birth delivery	108 (27.7)	135 (34.8)	7.1 (0.04, 14.1)	1.44 (1.03; 2.00)	0.025
Pregnancy loss§	47 (30.3)	40 (22.9)	−7.5 (−18.2, 3.4)	0.69 (0.41, 1.16)	0.167
D3 ET only, n (%)	N = 194	N = 194
Positive pregnancy test	65 (33.5)	69 (35.6)	2.1 (−8.1, 12.2	1.10 (0.71, 1.73)	0.666
Clinical pregnancy	44 (22.7)	53 (27.3)	4.6 (−5.6, 14.8)	1.29 (0.79. 2.12)	0.290
Ongoing pregnancy	43 (22.2)	49 (25.3)	3.1 (−7.1, 13.2)	1.19 (0.73, 1.96)	0.477
Live birth delivery	43 (22.2)	48 (24.7)	2.6 (−7.6, 12.7)	1.16 (0.71, 1.91)	0.552
Pregnancy loss§	22 (33.8)	21 (30.4)	−3.4 (−20.4, 13.4)	0.85 (0.39, 1.89)	0.714
D5 ET only, n (%)	N = 196	N = 194
Positive pregnancy test	90 (45.9)	106 (54.6)	8.7 (−1.4, 18.7)	1.42 (0.92, 2.21)	0.112
Clinical pregnancy	70 (35.7)	92 (47.4)	11.7 (1.6, 21.4)	1.66 (1.05, 2.62)	0.022
Ongoing pregnancy	68 (34.7)	89 (45.9)	11.2 (1.1, 20.8)	1.59 (1.02, 2.47)	0.034
Live birth delivery	65 (33.2)	87 (44.8)	11.7 (1.6, 21.3)	1.64 (1.05, 2.56)	0.025
Pregnancy loss	25 (27.8)	19 (17.9)	−9.9 (−23.6, 4.2)	0.57 (0.27, 1.18)	0.122
IMPLANT 4 (D5 ET only), n (%)	N = 409	N = 398
Positive pregnancy test	207 (50.6)	204 (51.3)	0.68 (−6.20, 7.54)	1.01 (0.76, 1.35)	0.931
Clinical pregnancy	162 (39.6)	164 (41.2)	1.66 (−5.09, 8.39)	1.06 (0.79, 1.41)	0.700
Ongoing pregnancy	160 (39.1)	161 (40.5)	1.40 (−5.33, 8.12)	1.05 (0.79, 1.40)	0.745
Live birth delivery	158 (38.6)	155 (38.9)	0.39 (−6.31, 7.08)	1.00 (0.75, 1.34)	0.980
Pregnancy loss§	49 (23.7)	49 (24.0)	0.31 (−7.92, 8.56)	1.01 (0.64, 1.59)	0.965

*

D3/D5 ET denotes ET on day 3/day 5 after oocyte pick-up. This table is based on the full analysis sets.

†

Difference between percentages with exact 95% CI, not accounting for site or ET day (IMPLANT 1 and 2). Common risk (percentage) difference with age and weight categories as stratification factors, and stratified Newcombe 95% CI (IMPLANT 4).

‡

Odds ratio, 95% CI and P-value estimated using: a logistic regression model with fixed effect for treatment group (IMPLANT 1), a stratified exact logistic regression model with fixed effect for treatment group and ET day as a covariate (IMPLANT 2), a stratified logistic regression model with fixed effect for treatment group and age and weight as covariates (IMPLANT 4). Site was used as a stratification factor except for the analysis of pregnancy loss.

§

Pregnancy loss was defined as any positive pregnancy test at 14 days post-OPU not resulting in a live birth. The % denominator is the number of subjects with positive pregnancy test at 14 days post-OPU. It was analysed using a discrete time-to-event model (via exact logistic regression), modelling separately non-pregnancy up to and post Day 14.

IMPLANT 1

Efficacy results for all treatment groups (placebo, 100, 300 and 900 mg) are shown in Supplementary Table SIII. There was an overall absolute 9.0% (95% CI −3, 21) increase in the 6-week clinical pregnancy rate in nolasiban-treated subjects (at all doses) compared to placebo (42.9% vs. 33.8%). By group, the clinical pregnancy rates were 46.8%, 35.0% and 46.7% in the 100, 300 and 900 mg group, respectively. When each of the treatment groups were compared to the placebo group individually, no statistically significant differences were observed, although the 900 mg group approached statistical significance in further analyses via logistic regression with treatment, site, and ET rate as covariates. Similarly, the primary analysis of the primary efficacy endpoint for the FAS (Cochran-Armitage test of a linear trend in proportions) did not demonstrate an increase in the clinical pregnancy rate with increasing dose level (P = 0.33). The pregnancy loss rate was 18.8% in the nolasiban 900 mg arm, compared to 42.4% in the placebo arm (OR 0.32, 95% CI 0.08, 1.08, P = 0.059).

Further analyses were conducted to identify potential factors predictive of pregnancy outcomes and these suggested that the elevated P4 serum level on the day of ET was a negative predictor for pregnancy. With exclusion of these subjects from the efficacy analyses, significant dose-related increases for the pregnancy rates and LBR were observed, with the largest increases in ongoing pregnancy and LBR compared to placebo seen in the 900 mg group (51.0% vs. 30.6% for 900 mg vs. placebo for both ongoing pregnancy and LBR). Analyses of pregnancy endpoints assessed via the Cochran-Armitage test of a linear trend in proportions provided evidence of an increase in pregnancy rates with increasing dose levels (P = 0.035 and P = 0.025 for 10-week ongoing pregnancy and LBR, respectively) (Supplementary Table SIII).

The ongoing pregnancy rates for the subgroup of subjects undergoing fresh single ET (the target population for subsequent clinical trials) were 28.2%, 35.9%, 33.3% and 42.9% for placebo, 100, 300 and 900 mg respectively, and were similar to the LBR (Supplementary Table SIII). Ongoing pregnancy rate was thus chosen as the primary efficacy endpoint for subsequent clinical studies.

There was a decrease from baseline in the median number of uterine contractions per minute for the 900 mg group (13.3%) compared to the placebo group (0.0%) (P = 0.051). No such evidence was seen for the 100 and 300 mg doses. This endpoint was not measured in later clinical studies.

IMPLANT 2

In IMPLANT 2, the overall ongoing pregnancy rates for the D3 and D5 SET combined were 35.6% and 28.5% in the nolasiban and placebo arms (OR 1.43 95% CI 1.03, 1.99, P = 0.031). Considered separately, the increase was statistically significant (OR 1.59 95% CI 1.02, 2.47, P = 0.034) in the D5 subgroup, but not significant (OR 1.19 95% CI 0.73, 1.96, P = 0.477) in the D3 subgroup.

Similarly, the overall LBRs for the D3 and D5 SET combined were 34.8% in the nolasiban group compared to 27.7% in the placebo group (OR 1.44 95% CI 1.03, 2.00, P = 0.025). Again, the increase was significant (OR 1.64 95% CI 1.05, 2.56, P = 0.025) in the D5 subgroup and not significant (OR 1.16 95% CI 0.71, 1.91, P = 0.552) in the D3 subgroup. The clinical pregnancy rates were 37.4% and 29.2% for nolasiban versus placebo (OR 1.50 95% CI 1.08, 2.08, P = 0.014), and rates for the positive 2-week pregnancy test were 45.1% and 39.7% (OR 1.26 95% 0.93, 1.71, P = 0.134) for nolasiban and placebo, respectively. The pregnancy loss rate was 22.9% in the nolasiban arm, compared to 30.3% in the placebo arm (OR 0.69 95% CI 0.41, 1.16, P = 0.167).

In D3 ET subjects, ongoing pregnancy was achieved in 49 of 194 women (25.3%) in the 900 mg nolasiban treatment groups compared to 43 of 194 women (22.7%) in the placebo group (Table II), with an OR for nolasiban versus placebo of 1.19 (P = 0.477, 95% CI 0.74, 1.90) and a difference of 3.16% (95% CI −5.4, 11.6).

In D5 ET only, ongoing pregnancy was achieved in 89 of 194 women (45.9%) in the 900 mg nolasiban treatment groups compared to 68 of 196 subjects (34.7%) in the placebo groups (Table II). The difference between the two treatment groups was statistically significant (P = 0.029), with an odds ratio for nolasiban versus placebo of 1.59 (95% CI 1.04, 2.41), an absolute increase of 10.6% (95% CI 0.84, 20.1). Live birth rates were also improved with nolasiban, with 44.8% and 33.2% of women having a live birth (OR 1.64 95% 1.05, 2.56, P = 0.025), showing substantial absolute and relative increases of 11.7% (95% CI 1.6, 21.3) and 35%, respectively.

IMPLANT 4

In IMPLANT 4, the ongoing pregnancy rate with nolasiban 900 mg was 40.5% compared to 39.1% with placebo (OR 1.05 95% CI 0.79, 1.40, P = 0.745); thus, the trial did not confirm the significant effects observed in IMPLANT 2.

IPD meta-analysis of ongoing pregnancy and live birth

As described above, a meta-analysis using individual participant data (IPD) from the three clinical trials was conducted to assess the overall treatment effect of the 900 mg dose of nolasiban.

In the total population (D3 and D5 combined), ongoing pregnancy was achieved in 326 of 846 subjects (38.5%) in the 900 mg nolasiban treatment groups compared to 290 of 864 subjects (33.6%) in the placebo groups (Fig. 2). The overall analysis, with the four categories (IMPLANT 1 Day 3, IMPLANT 2 Day 3, IMPLANT 2 Day 5 and IMPLANT 4 Day 5) as the stratification factor, showed a common risk difference between the nolasiban and placebo treatment groups of 5.03% (95% CI 0.48, 9.56), with an odds ratio for nolasiban versus placebo of 1.25 (95% CI 1.02, 1.52, P = 0.029). The Breslow–Day test was not significant, supporting equal odds ratios across the four strata (P = 0.232).

Figure 2.

IPD meta-analysis of ongoing pregnancy and live birth across IMPLANT 1, IMPLANT 2 and IMPLANT 4. Ongoing pregnancy was defined as a foetus with heartbeat at 10 weeks post-ET confirmed by ultrasound. D3/D5 ET denotes embryo transfer on Day 3/Day 5 after oocyte retrieval. Live birth was defined as a live delivery after 24 weeks of gestation. †Common risk (percentage) difference between Nolasiban and Placebo, along with stratified Newcombe confidence limits. ‡Cochran-Mantel-Haenszel test: Site used as a stratification factor for the by study and by ET day results. Study used as a stratification factor for the overall analysis. IMPLANT 2 D3 and IMPLANT 2 D5 were treated as separate studies for this meta-analysis.

A similar pattern was observed for live birth. Live birth was achieved in 316 of 846 subjects (37.4%) in the 900 mg nolasiban treatment groups compared to 285 of 864 subjects (33.0%) in the placebo groups (Fig. 2). The overall analysis, with the four categories (IMPLANT 1 Day 3, IMPLANT 2 Day 3, IMPLANT 2 Day 5 and IMPLANT 4 Day 5) as the stratification factor, showed a common risk difference between the nolasiban and placebo treatment groups of 4.43% (95% CI −0.10, 8.93), with an odds ratio for nolasiban versus placebo of 1.22 (95% CI 1.00, 1.49, P = 0.053). The Breslow-Day test was again not significant, supporting equal odds ratios across the four strata (P = 0.178).

In the pooled D3 population, ongoing pregnancy was achieved in 76 of 254 subjects (29.9%) in the 900 mg nolasiban treatment groups compared to 62 of 259 subjects (23.9%) in the placebo groups. Live birth was achieved in 74 of 254 subjects (29.1%) in the 900 mg nolasiban treatment groups compared to 62 of 259 subjects (23.9%) in the placebo groups (Fig. 2).

In the pooled D5 population, ongoing pregnancy was achieved in 250 of 592 subjects (42.2%) in the 900 mg nolasiban treatment groups compared to 228 of 605 subjects (37.7%) in the placebo groups. Live birth was achieved in 242 of 592 subjects (40.9%) in the 900 mg nolasiban treatment groups compared to 223 of 605 subjects (36.9%) in the placebo groups (Fig. 2).

The common risk differences and CMH test by day of ET (pooled D3: IMPLANT1 + IMPLANT2; pooled D5: IMPLANT2 + IMPLANT4) showed slightly higher risk differences for the pooled D3 (Ongoing pregnancy: 6.25; 95% CI −1.46, 13.90. Live birth: 5.47; 95% CI −2.21, 13.09) compared to the pooled D5 ET data (Ongoing pregnancy: 4.09 95% CI −1.48, 9.63. Live birth: 3.58; 95% CI −1.97, 9.09). However, overall, the risk differences and corresponding odds ratios were similar for D3 and D5 ET and the 95% CI for the odd-ratios included 1.

SIDES analysis

The exploratory SIDES analysis applied to IMPLANT 2 data identified potential subgroups with better response to nolasiban compared to placebo based on higher weight, larger total FSH dose, and ET Day 5. SIDES applied to IMPLANT 4 data identified subgroups based on lower total FSH dose, lower E2 on day of hCG trigger and lower BMI. However, the subgroups identified in one study were not confirmed in the other study, and it was concluded that there was no evidence to suggest any relevancy to the efficacy of nolasiban.

Population PK and PK/PD

A correlation between exposure and pregnancy outcomes was performed. For Cmax (maximum serum concentration), a generally steady, non-linear increase in the likelihood of being pregnant was observed with increasing concentrations. Area-under-the-curve (AUC) also showed an increased likelihood of pregnancy at high exposures. An increase was also at very low exposures, although this is likely to be an artefact due to few subjects at these exposure levels. Overall, the results suggested a trend towards a greater benefit of nolasiban at upper ranges of exposure for both Cmax and AUC (Fig. 3).

Figure 3.

Response (pregnant/non-pregnant) by simulated Cmax (ng/mL) and simulated AUC (ng/mL*hr). Cmax signifies Maximum concentration, AUC signifies area under curve (of plasma concentration vs time). A response of 1 is a positive response (pregnancy at 14 days post-implantation) and 0 is a negative response (no pregnancy). Each dot represents 1 patient and dots are jittered for readability around 1 (pregnancy) or 0 (no pregnancy). The blue line is a loess spline through the data indicating a correlation or a lack of correlation between exposure (x axes) and response (y axes); i.e, the loess spline going towards 1 at higher exposures indicates an association between higher exposures and pregnancy. The model fit is presented in Supplemental Fig. S1.

Safety

Maternal, obstetric and neonatal outcomes for the three trials are shown in Table III. For the IMPLANT 2 and IMPLANT 4 differences including 95% CI were calculated between nolasiban 900 mg and placebo treated subjects.

Table III

Maternal, obstetric and neonatal outcomes for IMPLANTS 1, 2 and 4.^*

	IMPLANT 1				IMPLANT 2			IMPLANT 4
Variable	Placebo	Nolasiban 100 mg	Nolasiban 300 mg	Nolasiban 900 mg	Placebo	Nolasiban 900 mg	Difference (95% CI)†	Placebo	Nolasiban 900 mg	Difference (95% CI)†
Events up to Week 10 post-ET (n)	65	62	60	60	391	387		408	399
Miscarriage, n (%)	13 (20.0)	5 (8.1)	8 (13.3)	5 (8.3)	44 (11.3)	38 (9.8)	−1.4 (−8.4, 5.6)	46 (11.3)	40 (10.0)	−1.3 (−8.2, 5.7)
Headache, n (%)	2 (3.1)	3 (4.8)	1 (1.7)	3 (5.0)	11 (2.8)	11 (2.8)	0.0 (−7.0, 7.1)	6 (1.5)	5 (1.3)	−0.2 (−7.2, 6.7)
Vaginal hemorrhage, n (%)	8 (12.3)	2 (3.2)	2 (3.3)	3 (5.0)	11 (2.8)	8 (2.1)	−0.8 (−7.8, 6.4)	12 (2.9)	9 (2.3)	−0.7 (−7.6, 6.2)
Nausea, n (%)	2 (3.1)	0 (0)	1 (1.7)	1 (1.7)	3 (0.8)	6 (1.6)	0.8 (−6.3, 7.9)	5 (1.2)	8 (2.0)	0.8 (−6.2, 7.7)
OHSS, n (%)	0 (0)	2 (3)	3 (5)	2 (3)	4 (1.0)	4 (1.0)	0.0 (−7.1, 7.1)	9 (2.2)	6 (1.5)	−0.7 (−7.6, 6.2)
Ectopic pregnancy, n (%)	1 (2)	1 (2)	0 (0)	1 (2)	4 (1.0)	1 (0.3)	−0.8 (−7.9, 6.3)	1 (0.2)	3 (0.8)	0.5 (−6.4, 7.4)
Obstetric outcomes
Delivery, n (%)	19	25	21	26	108	131§		158	155
Singleton	15 (79)	21 (84)	17 (81)	20 (77)	107 (99.1)	126 (96.2)	−2.9 (−15.6, 9.8)	155 (98.1)	152 (98.1)	0.0 (−11.1, 11.1)
Twins	4 (21)	4 (16)	4 (19)	6 (23)	1 (0.9)	5 (3.8)	2.9 (−9.8, 15.6)	3 (1.9)	3 (1.9)	0.0 (−11.1, 11.1)
Type of delivery, n (%)
Spontaneous	10 (53)	8 (32)	4 (19)	15 (60)	47 (43.5)	58 (44.3)	0.8 (−11.9, 13.4)	74 (46.8)	81 (52.3)	5.4 (−5.8, 16.6)
Induced	1 (5)	5 (20)	3 (14)	0 (0)	5 (4.6)	10 (7.6)	3.0 (−9.7, 15.7)	13 (8.2)	12 (7.7)	−0.5 (−11.7, 10.4)
Caesarian	8 (42)	11 (44)	14 (67)	11 (42)	56 (51.9)	63 (48.1)	−3.8 (−16.4, 9.0)	71 (44.9)	62 (40.0)	−4.9 (−16.1, 6.1)
Term (≥37 wks)	15 (65.2)	20 (69.0)	18 (72.0)	20 (62.5)	97 (89.8)	112 (85.5)	−4.3 (−16.9, 8.4)	142 (89.9)	141 (91.0)	1.1 (−9.9, 12.3)
Preterm (32 to 366/7 ws)	NR	NR	NR	NR	11 (10.2)	15 (11.5)	1.3 (−11.4, 14.0)	14 (8.9)	11 (7.1)	−1.7 (−13.0, 9.2)
Very preterm (≤316/7 wks)	NR	NR	NR	NR	0 (0)	4 (3.1)	3.1 (−9.7, 15.7)	2 (1.3%)	3 (1.9%)	6.7 (−10.4, 11.7)
Neonatal outcomes at delivery
Neonates (n)	23	29	25	32	109	136		161	158
Male, n (%)	15 (65.2)	11 (37.9)	10 (40.0)	21 (65.6)	49 (45.0)	67 (49.3)	4.3 (−8.4, 16.8)	94 (58.4)	80 (50.6)	−7.8 (−18.7, 3.4)
Female, n (%)	8 (34.8)	17 (58.6)	15 (60.0)	11 (34.4)	60 (55.0)	69 (50.7)	−4.3 (−16.8, 8.4)	67 (41.6)	78 (49.4)	7.8 (−3.4, 18.7)
Gestational age (wks)	37.6 ± 2.3	37.1 ± 4.0	36.1 ± 4.7	36.4 ± 4.0	38.7 ± 1.9	38.2 ± 2.8	−0.55 (−1.18, 0.07)	38.5 ± 2.1	38.7 ± 2.0	0.20 (−0.25, 0.66)
Weight (g)	3016.7 ± 536.3	2748.0 ± 858.5	2626.8 ± 1011.6	2793.8 ± 955.8	3174 ± 517	3137 ± 690	−37 (−194, 120)	3195.8 ± 592.6	3213.7 ± 578.9	18 (−111, 147)
Height (cm)	49.20 ± 3.10	48.17 ± 5.41	47.16 ± 7.91	48.09 ± 6.70	50.9 ± 3.7	50.4 ± 4.6	−0.4 (−1.5, 0.6)	50.87 ± 3.51	51.20 ± 3.72	0.33 (−0.47, 1.13)
Head circumference (cm)	33.47 ± 1.76	32.86 ± 3.24	32.06 ± 4.54	32.35 ± 2.60	34.3 ± 1.6	33.7 ± 2.2	−0.6 (−1.1, −0.04)	33.81 ± 3.51	33.88 ± 1.84	0.08 (−0.62, 0.77)
Apgar score 1 min	9.0 ± 0.9	8.9 ± 1.5	7.6 ± 3.0	8.6 ± 1.9	9.09 ± 1.28	9.01 ± 1.50	−0.08 (−0.44, 0.28)	9.06 ± 1.48	9.08 ± 1.22	0.02 (−0.28, 0.32)
Apgar score 5 min	9.9 ± 0.5	9.5 ± 0.9	8.8 ± 1.7	9.3 ± 1.4	9.65 ± 0.76	9.61 ± 0.84	−0.04 (−0.25, 0.17)	9.68 ± 0.78	9.68 ± 0.74	0.0 (−0.17, 0.17)
Congenital anomaly, n (%)	2 (9)	4 (14)	0 (0)	2 (6)	4 (3.7)	5 (3.7)	0.0 (−12.6, 12.6)	6 (3.7)	6 (3.8)	0.0 (−10.9, 11.2)
Neonatal outcomes at 28 days after delivery
Weight (g)	4187.3 ± 1250.7	3802.0 ± 815.7	3659.6 ± 784.2	3989.4 ± 1105.5	4259 ± 747	4234 ± 946	−25 (−246, 196)	4226.7 (859.4)	4212.0 (730.4)	−15 (−193, 163)
Height (cm)	54.45 ± 4.95	53.8 ± 4.30	52.47 ± 4.55	52.49 ± 7.90	55.1 ± 3.5	54.8 ± 4.8	−0.4 (−1.5, 0.8)	54.65 (3.54)	54.99 (3.47)	0.34 (−0.50, 1.18)
Head circumference (cm)	36.64 ± 2.34	36.67 ± 2.28	36.71 ± 1.53	35.74 ± 3.08	37.2 ± 1.6	36.7 ± 2.5	−0.5 (−1.1, 0.1)	36.64 (3.65)	37.17 (1.99)	0.52 (−0.21, 1.25)
Breast-feeding, n (%)	NR	NR	NR	NR	91 (83.5)	111 (81.6)	−1.9 (−14.4, 10.7)	127 (78.9)	121 (76.6)	−2.3 (−13.4, 8.7)
Admission to intensive care, n (%)	NR	NR	NR	NR	9 (8.3)	9 (6.6)	−1.6 (−14.2, 10.9)	6 (3.7)	4 (2.5)	−1.2 (−12.2, 9.7)
Neonatal morbidity, n (%)‡
Jaundice	9 (39)	4 (14)	4 (16)	5 (16)	20 (18.3)	18 (13.2)	−5.1 (−17.6, 7.5)	11 (6.8)	13 (8.2)	1.4 (−9.7, 12.2)
Respiratory distress syndrome	0 (0)	6 (21)	3 (12)	2 (6)	10 (9.2)	8 (5.9)	−3.3 (−15.8, 9.3)	3 (1.9)	4 (2.5)	0.7 (−10.4, 11.6)
Difficulty in feeding	1 (1)	3 (10)	2 (8)	0 (0)	2 (1.8)	11 (8.1)	6.3 (−6.4, 18.7)	11 (6.8)	6 (3.8)	−3.0 (−14.1, 7.7)
Difficulty in thermoregulation	0 (0)	1 (3)	0 (0)	1 (3)	0 (0)	6 (4.4)	4.4 (−8.2, 16.9)	0 (0)	0 (0)	0
Bronchopulmonary dysplasia	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	3 (2.2)	2.2 (−10.4, 14.7)	1 (0.6)	0 (0)	−0.6 (−11.6, 10.4)

*

Plus–minus values are mean ± SD. Percentages are calculated as a percentage of each group of outcomes (Events up to week 10, obstetric outcomes, and neonatal outcomes). Percentages may not total 100 because of rounding. This table is based on the safety analysis set. NR denotes not reported.

§

Specific details on obstetric and neonatal outcomes were missing for 4 subjects.

†

Difference between percentages with corresponding exact 95% confidence interval/difference between means with corresponding 95% confidence interval.

‡

The occurrence of the following neonatal morbidities was recorded: Asphyxia, bronchopulmonary dysplasia, cerebellar haemorrhage, hypoxic-ischaemic encephalopathy, intra-ventricular haemorrhage, jaundice, meningitis, necrotising enterocolitis, neonatal acidosis, periventricular leukomalacia, respiratory distress syndrome, retinopathy of prematurity, anaemia, white matter injury—hypotension, difficulty in thermoregulation, difficulty in feeding. Those that were reported in more than 2% of neonates in either group are listed.

The most commonly reported adverse events, in ≥1% of the mothers, were miscarriage (or therapeutic abortion), headache, vaginal haemorrhage, OHSS and ectopic pregnancy. The incidences were generally similar between placebo and nolasiban treated groups and the 95% CI for the differences included zero. There were no notable differences in obstetric outcomes.

The percentage of infants admitted to neonatal intensive care was also similar in the placebo group and nolasiban groups. The most commonly observed neonatal morbidities were jaundice and respiratory distress syndrome.

Discussion

Despite significant advances in the field of IVF and ET over the past 40 years, success rates remain relatively low (Luke et al., 2012; CDC, 2018; Toftager et al., 2017). Of the various tested measures to improve pregnancy rates following ET, none except luteal support with P4, have been shown to boost pregnancy and live birth outcomes (van der Linden et al., 2015; Mol and Barnhart 2019). A recent review of common add-ons suggested to improve endometrial receptivity, including immune therapies, endometrial scratching, endometrial receptivity array, uterine artery vasodilation and human chorionic gonadotropin instillation, concluded that there is no robust evidence that any are effective or safe (Lensen et al., 2019).

While the mechanisms of implantation failure are not fully understood, it is widely believed that oxytocin-modulated uterine contractions, uterine blood flow and endometrial receptivity may contribute. Since uterine contractile activity can adversely affect implantation, pharmacological tocolytics might be expected to improve pregnancy rates following ET. In a meta-analysis including six studies (Huang et al., 2017) reported that the mixed oxytocin/vasopressin 1a receptor antagonist, atosiban, administered intravenously before, during and after ET, enhanced pregnancy outcomes in the general population of women undergoing IVF. However, there was no conclusive evidence of a significant improvement in LBR.

The clinical program for the novel oral OTR antagonist nolasiban has provided substantial evidence of the potential for its use in the improvement of pregnancy outcomes following fresh ET. The IMPLANT 1 dose ranging results supported the investigation of 900 mg of nolasiban in Phase 3 trials. Statistically significant and clinically relevant increases in ongoing pregnancy and LBRs were demonstrated in IMPLANT 2, with the D5 subgroup showing greater improvement. However, the IMPLANT 4 trial did not confirm a significant effect of nolasiban in women undergoing fresh D5 SET. The results of the IMPLANT 4 trials meant that no further funding was available to start enrolment of the IMPLANT 3 trial that had been planned in USA.

In the meta-analysis of ongoing pregnancy across all three completed trials, a single dose of nolasiban 900 mg showed a similar magnitude of effect in D3 and D5 (common risk differences of 6.25% and 4.09% for ongoing pregnancy and 5.47% and 3.58% for live birth, respectively), although statistical significance was not achieved for either ET day when considered separately. Overall, for D3 and D5 ET combined, nolasiban increased ongoing pregnancy in women undergoing fresh ET by an absolute 5.0% compared to placebo (i.e., a 15% relative increase), with an odds ratio of 1.25 (95% CI 0.48, 9.56, P = 0.029), and live birth by an absolute 4.4%, with an odds ratio of 1.22 (95% CI 1.00, 1.49, P = 0.053), suggesting that nolasiban has the potential to improve IVF outcomes after ET.

In the IMPLANT 1 and 2 trials, the pregnancy loss rate was numerically lower with nolasiban 900 mg compared to placebo. This was not the case in IMPLANT 4. A pregnancy loss was defined as any positive pregnancy test that did not result in a live birth. This observation suggests that the potential improvement in pregnancy rates by nolasiban may in part be due to reduced pregnancy loss.

The exploratory SIDES analyses did not identify any potential subgroups in which nolasiban performed better than in the overall population. However, the IMPLANT 4 population PK results demonstrated trends towards higher AUC and Cmax in women who became pregnant compared to women who did not, suggesting a larger potential for benefit of nolasiban in patients within the upper range of exposure.

Maternal, obstetric and neonatal safety outcomes were very similar between the nolasiban and placebo treatment groups.

The limitations of these trials include absence of randomisation to D3 or D5 and minor differences across the protocols for selecting patients for D3 or D5 transfer in IMPLANT 2, which may have resulted in lower baseline likelihood of successful IVF outcomes in the D3 transfer group and thus may have affected the relative outcomes in the D5 and D3 populations. Ongoing pregnancy has been recommended as the best primary endpoint for trials in reproductive medicine (Braakhekke et al., 2014) and was chosen as the primary endpoint IMPLANT 2 and 4. Live birth rates were also measured in each trial and were very similar to the ongoing pregnancy rates in each trial. The ovarian stimulation protocol used among clinical sites was not specifically stipulated in the protocols, and together with differences in operator experience with ET may have affected results. Furthermore, some of the sites recruited few patients, requiring pooling of sites for statistical analysis.

Patients were highly selected and therefore potentially more likely to have a higher baseline chance of pregnancy. It is also possible that patients who may have the greatest potential for clinical impact from OTR antagonism, e.g. patients with certain pre-existing conditions such as moderate-to-severe endometriosis, were excluded from the trials. For these reasons, larger sample sizes for hypothesis testing may be required to demonstrate differences between nolasiban and placebo. It is also possible that some of the variability between studies is due to chance, again highlighting the need for large, well-powered clinical trials. Lastly, the meta-analysis was not a prospectively designed hypothesis-driven analysis. These limitations should thus be considered when drawing conclusions from the data.

Conclusions

In an IPD meta-analysis of the clinical trials a single oral dose of nolasiban 900 mg administered 4 h before ET was associated with an absolute increase of 5.0% (95% CI 0.5, 9.5) in ongoing pregnancy rate and an absolute increase of 4.4% (95% CI −0.10, 8.93) in live birth rate compared to placebo. SIDES analysis did not identify subgroups that consistently showed a larger treatment effect than in the full dataset. Nolasiban was well tolerated at a dose of 900 mg and there were few differences in maternal, obstetric and neonatal outcomes between the nolasiban and placebo treatment groups. The IMPLANT 4 population PK data suggest that higher doses and extended regimens may provide superior efficacy compared to a single dose of nolasiban prior to ET, supporting additional trials to investigate its potential to improve IVF outcomes.

Data availability

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

Acknowledgements

The participating centres and corresponding Principal Investigators are listed in Supplementary Table SI. In particular, the authors would like to acknowledge Attila Torok MD, PhD, Aleš Sobek MD, PhD, Mariusz Kiecka MD, PhD, Fernando Sanchez Martin MD, PhD, Sven Olaf Skouby MD, DMSci, Andrei Soritsa MD, PhD, Marcos Ferrando Serrano MD. The authors would also like to acknowledge Dr Renato Fanchin for his contribution to analysis of uterine contractions in the IMPLANT 1 trial. The authors would also like to thank the patients who consented to participate in the trials.

Authors’ roles

C.B., P.P., H.T., H.V., A.H., P.T. and E.L. designed and implemented the clinical trials. G.G., C.B., P.P., H.T. and H.V. recruited subjects. P.T. performed the statistical analyses. O.P. performed the PK analyses. G.G., J.D., A.H., E.G., P.T., O.P. and E.L. drafted the manuscript. All authors were involved in the analysis of the results, reviewed the manuscript and approved the final version.

Funding

All work described in this manuscript was funded and sponsored by ObsEva SA.

Conflicts of interest

A.H., O.P., E.G., and E.L. are employees and stockholders of ObsEva SA. E.L. is a board member of ObsEva SA. G.G. reports honoraria and/or non-financial support from ObsEva, Merck, MSD, Ferring, Abbott, Gedeon-Richter, Theramex, Guerbet, Finox, Biosilu, Preglem and ReprodWissen GmbH. CB reports grants and honoraria from ObsEva, Ferring, Abbott, Gedeon Richter and MSD. P.P. reports consulting fees from ObsEva. H.T. reports grants and or fees from ObsEva, Research Fund of Flanders, Cook, MSD, Roche, Gedeon Richter, Abbott, Theramex and Ferring. H.V. reports grants from ObsEva and non-financial support from Ferring. P.T. is an employee of Cytel Inc., who provides statistical services to ObsEva. J.D. reports consulting fees and other payments from ObsEva and, Scientific Advisory Board membership of ObsEva.