A 5-year multicentre randomized controlled trial comparing personalized, frozen and fresh blastocyst transfer in IVF

402 RBMO VOLUME 41 ISSUE 3 2020 1 Department of Pediatrics, Obstetrics and Gynecology, University of Valencia, Valencia 46010, Spain 2 Igenomix Foundation-INCLIVA, Parque Tecnológico de Paterna, Ronda Narciso Monturiol Estarriol 11B, Paterna Valencia 46980, Spain 3 Igenomix SL, Parque Tecnológico de Paterna, Ronda Narciso Monturiol Estarriol 11B, Paterna Valencia 46980, Spain 4 IVI-RMA Valencia, Plaza de la Policía Local 3, Valencia 46015, Spain 5 SBALAGRM-Sofia, 5 Baku St, Sofia 1756, Bulgaria 6 IVI-RMA Barcelona, Ronda General Mitre 14 Barcelona 08017, Spain 7 Bahçeci Health Group-Fulya IVF Centre, Hakki Yeten Cad, 11 Kat 3, Terrace Fulya Istanbul 34365, Turkey 8 IVI-RMA Madrid, Av Del Talgo 68, Aravaca Madrid 28023, Spain 9 Oak Clinic Japan, 2-7-9 Tamade-Nishi, Nishinari-ku Osaka 557-0045, Japan 10 ProcreaTec, Calle Manuel de Falla 6-8, Madrid 28036, Spain 11 IVI-RMA Vigo, Plaza Francisco Fernández del Riego 7, Vigo 36203, Spain 12 Nilo Frantz Reproductive Medicine, Av Dr Nilo Peçanha 1221 – 10° andar, Boa Vista Porto Alegre, Brazil 13 Centro de Infertilidade e Medicina Fetal do Norte Fluminense, R Barão da Lagoa Dourada 409 – Centro, Campos dos Goytacazes Rio de Janeiro 28035-210, Brazil 14 Department of Reproductive Medicine, Universitair Ziekenhuis Brussel, 101 Laarbeeklaan, Jette Brussels 1090, Belgium 15 IVI-RMA Panama, Calle 50 & Calle 57 Este, Panama City Panamá 07185 16 IVI-RMA Sevilla, Av República Argentina 58 Sevilla 41011, Spain 17 Departamento de Cirugía, Universidad de Sevilla, Avda. Sánchez Pizjuan S/N Sevilla 41009, Spain 18 Departamento de Biología Molecular e Ingeniería Bioquímica, Universidad Pablo de Olavide Sevilla 41013, Spain 19 IVI-RMA Bilbao, Leioa Paseo Landabarri 1Vizcaya 48940, Spain 20 Department of Obstetrics and Gynaecology, Monash University, Monash Medical Centre, 246 Clayton Road, Clayton Victoria 3168, Australia †Joint first authors. © 2020 The Authors. Published by Elsevier Ltd on behalf of Reproductive Healthcare Ltd. This is an open access article under the CC BY-NC-ND license. (http://creativecommons.org/licenses/by-nc-nd/4.0/) *Corresponding author. E-mail address: [email protected] (C.Simón). https://doi.org/10.1016/j.rbmo.2020.06.002 1472- 6483/© 2020 The Authors. Published by Elsevier Ltd on behalf of Reproductive Healthcare Ltd. This is an open access article under the CC BY-NC-ND license. (http://creativecommons.org/licenses/by-nc-nd/4.0/) Declaration: CS is co-inventor of the patent for gene expression profile (ERA) issued to Igenomix and Head of the Scientific Advisory Board of Igenomix. EL reports grants from Finox, personal fees from Ferring, MSD, Angelini and Ovascience, and other non-financial support from Ferring, outside the submitted work. JA G-V reports grants from Merck, MSD, Ferring, Gedeon Richter, and Theramex, outside the submitted work. BM reports grant (GNT1082548) from the Australian National Health and Medical Research Council (NHMRC) related to the submitted work and personal fees from ObsEva and Merck KGaA, Geurbert and Merck, outside of the submitted work. CG, XS, M and DV are employees of Igenomix SL. Other authors declare no conflict of interest. KEYWORDS Endometrial receptivity Endometrial receptivity analysis (ERA) Fresh embryo transfer (ET) Frozen embryo transfer (FET) Personalized embryo transfer (PET) Window of implantation (WOI) ARTICLE A 5-year multicentre randomized controlled trial comparing personalized, frozen and fresh blastocyst transfer in IVF BIOGRAPHY Professor Carlos Simón is Board Certified and Professor of Obstetrics and Gynaecology at the University of Valencia, Spain, and Senior Lecturer, Beth Israel Deaconess Medical Center Harvard University, Boston, USA. His main interest is the human embryonic implantation process, including the embryo, maternal endometrium, and the crosscommunication between them. CarlosSimón1,2,4,*,†, CarlosGómez3,1, SergioCabanillas4, IavorVladimirov5, GemmaCastillón6, JuanGiles4, KubraBoynukalin7, NecatiFindikli7, MustafaBahçeci7, IsraelOrtega8, CarminaVidal4, MiyakoFunabiki9, AlexandraIzquierdo10, LourdesLópez10, SusanaPortela11, NiloFrantz12, MarcosKulmann12, SagiriTaguchi9, ElenaLabarta4, FranciscoColucci13, ShariMackens14, XavierSantamaría6, ElkinMuñoz11, SaúlBarrera15, Juan AntonioGarcía- Velasco8, ManuelFernández16,1 7,18, MarcosFerrando19, MaríaRuiz3, Ben WMol20, DianaValbuena2,3for the ERA-RCT Study Consortium Group RBMO VOLUME 41 ISSUE 3 2020 403 KEY MESSAGE This study demonstrates the clinical potential of personalized embryo transfer guided by the endometrial receptivity analysis test at the first appointment, which should be confirmed in larger randomized controlled trials. ABSTRACT Research question: Does clinical performance of personalized embryo transfer (PET) guided by endometrial receptivity analysis (ERA) differ from frozen embryo transfer (FET) or fresh embryo transfer in infertile patients undergoing IVF? Design: Multicentre, open-label randomized controlled trial; 458 patients aged 37 years or younger undergoing IVF with blastocyst transfer at first appointment were randomized to PET guided by ERA, FET or fresh embryo transfer in 16 reproductive clinics. Results: Clinical outcomes by intention-to-treat analysis were comparable, but cumulative pregnancy rate was significantly higher in the PET (93.6%) compared with FET (79.7%) (P = 0.0005) and fresh embryo transfer groups (80.7%) (P = 0.0013). Analysis per protocol demonstrates that live birth rates at first embryo transfer were 56.2% in PET versus 42.4% in FET (P = 0.09), and 45.7% in fresh embryo transfer groups (P = 0.17). Cumulative live birth rates after 12 months were 71.2% in PET versus 55.4% in FET (P = 0.04), and 48.9% in fresh embryo transfer (P = 0.003). Pregnancy rates at the first embryo transfer in PET, FET and fresh embryo transfer arms were 72.5% versus 54.3% (P = 0.01) and 58.5% (P = 0.05), respectively. Implantation rates at first embryo transfer were 57.3% versus 43.2% (P = 0.03), and 38.6% (P = 0.004), respectively. Obstetrical outcomes, type of delivery and neonatal outcomes were similar in all groups. Conclusions: Despite 50% of patients dropping out compared with 30% initially planned, per protocol analysis demonstrates statistically significant improvement in pregnancy, implantation and cumulative live birth rates in PET compared with FET and fresh embryo transfer arms, indicating the potential utility of PET guided by the ERA test at the first appointment. INTRODUCTION When reproductive success or failure relies on collaboration between two partners, the functionality and synchronization of both are necessary. One reason why IVF treatments are not successful when good-quality embryos or even euploid embryos are transferred into the endometrial cavity (Rubio etal., 2017) may be due to the endometrial factor. Globally, IVF remains inefficient, with current live birth rates of only 25–30% per started cycle (Adamson etal., 2018); for comparison, rodents or rabbits have more than 95% implantation rate. The main difference lies in the endometrial and decidual control in human implantation events, versus embryo control in non-menstruating species (Simon and Giudice, 2017). The endometrial mucosa is the anatomic and functional cellular mediator for embryonic implantation. Women menstruate in an endless resetting process to synchronize the uterus for the imminent arrival of a blastocyst. Progesterone is the single most important hormone controlling the initiation and course of pregnancy inducing the acquisition of a 12–48-h transient modification of the endometrial epithelium known as the window of implantation (WOI), demonstrated clinically (Navot etal., 1991), epidemiologically (Wilcox etal., 1999), and morphologically (Murphy etal., 2004). This limited period can be found from LH + 6 to LH + 9 in a natural cycle or from progesterone + 4 to progesterone + 7 in a hormonal replacement therapy (HRT) cycle (Ruiz- Alonso etal., 2013; 2014). The transition from anatomical to molecular medicine has demonstrated the feasibility of an objective molecular classification of the human endometrium using transcriptomic profiling throughout the menstrual cycle (Ponnampalam etal., 2004; Talbi etal., 2006) as well as during the WOI (Riesewijk etal., 2003). Precision medicine is focused on targeted individualized treatment. On this basis, in 2011, we reported the transcriptomic signature of human endometrial receptivity from 238 genes with an artificial intelligence platform and algorithm able to identify the WOI of a given patient, leading to the creation of the endometrial receptivity analysis (ERA) (Diaz-Gimeno etal., 2011; 2013). The clinical utility of ERA is in guiding a personalized embryo transfer (PET) by synchronizing the embryo with the WOI to the patient in a personalized manner for the initial indication of implantation failure of endometrial origin (Ruiz-Alonso etal., 2013; 2014; Mahajan etal., 2015; Hashimoto etal., 2017; Tan etal., 2018). The goal is to improve clinical implantation by personalizing, diagnosing and synchronizing the endometrial factor. In this multicentre, international, randomized controlled trial (RCT), infertile patients undergoing clinical evaluation for IVF were also assessed whether PET guided by the ERA test increased the probability of live birth, pregnancy and implantation rates versus frozen embryo transfer (FET) or fresh embryo transfer. MATERIALS AND METHODS Study design An RCT evaluating PET guided by the ERA test versus FET or fresh embryo transfer. Participants were recruited from 16 reproductive clinics in Europe, USA and Asia, and randomized to one of the three arms indicated from 25 November 2013 to 30 April 2017. All additional embryo transfers of the remaining cryopreserved embryos in the PET and FET arms as well as new fresh embryo transfers in the embryo transfer arm carried out within 1 year after the first embryo transfer were also included in the estimation of cumulative rates. Subsequently, all participants who conceived were followed through pregnancy up to delivery; obstetric and neonatal outcomes were also reported. 404 RBMO VOLUME 41 ISSUE 3 2020 Data on pregnancy outcome were prospectively collected from the clinical sites involved and entered them into a secure electronic database. Ethical approval Ethical approval was given by Instituto Valenciano de Infertilidad Ethics Committee (2 July 2013), Spain; Local Ethics Commission of SBALAGRM – Sofia (29 January 2014), Bulgaria; UZ Brussels Medical Ethics Committee, Belgium (26 February 2014 with code 2014/006); Instituto Conmemorativo Gorgas de Estudios de la Salud Ethics Committee Panama 26 February 2014 with code 214/ CBI/ICGES/14); Istanbul Bilim Üniversitesi Klinik Arastirmalari Ethics Committee, Turkey (28 May 2014 with code 28.05.2014/20-145); Hospital das Clínicas da Faculdade de Medicina da USP Ethics Committee Brazil 3 December 2015 with code 1.349.189; Oak Clinic Group's Ethics Committee, Japan (28 February 2016). Each participating site provided a site-specific approval, and all participants provided written informed consent. Data were monitored by a clinical research associate. The protocol was first registered on 7 October 2013, and the first patient recruited on 25 November 2013 (ClinicalTrials.gov ID NCT01954758). Participants Infertile patients undergoing IVF were recruited at their first appointment. Women scheduled for embryo transfer at the blastocyst stage (day 5 or 6) were included. Inclusion criteria were age 37 years or younger, body mass index (BMI) of 18.5–30 and normal ovarian reserve (antral follicle count ≥8 and FSH <8 IU/ml). Ovarian stimulation protocol was decided by the enrolling physician, and the blastocyst vitrification system was the one used by the clinic. Couples who had experienced recurrent miscarriage (more than two previous biochemical pregnancies or spontaneous abortions), implantation failure (more than three failed IVF cycles with goodquality embryos transferred) or who had severe male factor infertility (fewer than 2 million spermatozoa/ml) were not eligible. Pathology affecting the endometrial cavity, including polyps and submucosal myomas, intramural myomas over 4 cm or hydrosalpinx previously operated on before embryo transfer, was not an exclusion criterion. Couples using preimplantation genetic diagnosis for aneuploidy (PGT-A) could also participate. Post-randomization exclusion criteria were progesterone levels of over 1.5 ng/ ml on the day of HCG administration in all groups, absence of blastocysts (day 5 or 6) for embryo transfer, or risk of ovarian hyperstimulation syndrome (OHSS) in the fresh embryo transfer arm. For all these cases, patients did not enter the per protocol analysis but were included in the intention-to-treat analysis. During the first appointment in each participating site and after providing informed consent, suitable patients were randomized into either PET, FET or fresh embryo transfer groups. In the FET group, patients received transfer during a hormonal replacement therapy (HRT) cycle after embryo thawing, following the protocol and timing used in each clinic. Ovarian stimulation, embryo culture and vitrification In all groups, ovarian stimulation was carried out using standard protocols in each of the participant sites according to female age, basal hormone levels, basal ovarian reserve and BMI. After a normal basal ultrasound, gonadotrophins were administered from day 2–3 after menstruation. Serial transvaginal ultrasound examinations and serum oestradiol determination were started on day 5 of ovarian stimulation and repeated every 48 h to monitor the ovarian response. Ovulation triggering was carried out mainly with HCG administration or gonadotrophin releasing hormone agonist, depending on the gonadotrophin protocol used, when at least two follicles reached 18 mm. Oocyte retrieval was scheduled 36 h later. Intracytoplasmic sperm injection (ICSI) or IVF was carried out according to the protocols of the participating sites, and fertilization was assessed 17–20 h after insemination or microinjection. Embryos were cultured using different approaches and culture media depending on the IVF laboratory. Embryo morphology and quality were evaluated according to Gardner's criteria (Gardner etal., 2007). Properly developed blastocysts were transferred on day 5 or 6 in the fresh embryo transfer arm, or vitrified in the PET and FET arms, using different protocols depending of the IVF laboratory. Subsequent blastocyst transfer was carried out at each participant site with different types of transfer catheters. Personalized embryo transfer guided by endometrial receptivity analysis In the PET group, patients underwent one or two endometrial biopsies (the timing of the second biopsy depended on the first result), and embryo transfer was carried out in an HRT cycle at the timing indicated by the ERA test. Endometrial biopsies were collected from the uterine fundus using a Pipelle catheter from Cornier® devices (CCD Laboratories, Paris, France) or similar, under sterile conditions. After the biopsy, the endometrial tissue was transferred to a cryotube containing 1.5 ml of RNALater (QIAGEN, Barcelona, Spain), vigorously shaken for a few seconds, and kept at 4°C or in ice for at least 4 h. The samples were then shipped at room temperature for ERA test. Endometrial preparation for frozen embryo transfer and personalized embryo transfer The endometrium was prepared for the cycle to carry out the ERA test, and subsequent PET and FET were carried out using an HRT cycle, as previously described (Ruiz-Alonso etal., 2013;2014). Briefly, after menstruation, ovarian quiescence was confirmed by vaginal ultrasound evaluation and oestradiol administration started from the second or third day onwards. Sonographic evaluation and oestradiol and progesterone assessment were carried out between 7 and 10 days after endometrial oestradiol preparation treatment. When a trilaminar endometrium measuring 6 mm or more with an endogenous progesterone serum level close to zero was observed, progesterone was administered at dosage and route of the participant physician or clinic for a period of 5 days (progesterone + 5 or 120 h, approximately). Fresh embryo transfer Embryo transfer in fresh cycles was carried out 5 or 6 days after oocyte retrieval according to blastocyst timing. Luteal phase supplementation route and dosage was the one used by the participant physician or clinic. Sample labelling, hybridization and personalized embryo transfer recommendation Total RNA was extracted using the TRIzol method according to the manufacturer's protocol (Life Technologies, Inc., RBMO VOLUME 41 ISSUE 3 2020 405 Carlsbad, CA, USA). About 1–2 µg of total RNA were obtained per mg of endometrial tissue. RNA quality was assessed by loading 300 ng of total RNA onto an RNA Labchip and analysing it in an A2100 Bioanalyzer according to manufacturer's protocol (Agilent Technologies, Inc., Santa Clara, CA, USA). Good-quality RNA samples, with RNA integrity number greater than 7, was a prerequisite for ERA. Sample preparation and hybridization was adapted from the Agilent technical manual (one-color). In short, first strand cDNA was transcribed from 200 ng of total RNA using T7-Oligo(dT) Promoter Primers. Samples were transcribed in vitro, and Cy-3 labelled, all with a Low Input Quick Amp Labeling kit (Agilent Technologies, Inc., Santa Clara, CA, USA). The labelling reaction typically yielded between 4 and 5 µg of complementary RNA with a specific activity greater than 6. Fragmented complementary RNA samples were hybridized onto the customized ERA microarray (Diaz-Gimeno etal., 2011), by incubation at 65°C for 17 h with constant rotation. The microarray was then washed with two 1-min steps in two different washing buffers (Agilent Technologies, Inc., Santa Clara, CA, USA). Hybridized microarrays were scanned in an Axon 4100A scanner (Molecular Devices, Sunnyvale, CA, USA), and data were extracted with the GenePix Pro 6.0 software (Molecular Devices, Sunnyvale, CA, USA). Endometrial receptivity analysis gene expression values were pre-processed and normalized, and the endometrial receptivity status was diagnosed by the ERA computational predictor (Díaz- Gimeno etal., 2011). The ERA test diagnoses an endometrial sample as receptive, pre-receptive or post-receptive with an associated diagnostic probability. This provides a recommendation for PET in a particular patient at P + 4, P + 4.5, P + 5, P + 5.5, P + 6, or P + 7. As described previously (Díaz-Gimeno etal., 2011), the customized microarray for ERA was designed after selecting 238 genes involved in human endometrial receptivity based on different gene expression profiles across the receptive, pre-receptive and post-receptive stages using raw expression data generated for a previous publication from our group. Genes showing an absolute fold-change greater than 3 and a false discovery rate less than 0.05 were selected. Three statistical approaches were used to generate the prediction: the union of the T-Rex gene list (GEPAS) (http:// gepas.bioinfo.cipf.es/) and SAM gene list (http://www.stat.stanford.edu/_tibs/ SAM/), intersected with the multtest gene list (http://www.bioconductor. org/). Selected genes were included in a customized gene expression microarray with a 15Kx8 format by using 4.5 earray (https://earray.chem.agilent.com/earray/). All the probes included in ERA were selected from Agilent's catalogue to avoid crosslinking phenomena. The ERA predictor was based on support vector machine prediction with a specificity and sensitivity of 0.886 and 0.998, respectively (Diaz-Gimeno etal., 2011). Outcome measures Primary outcome was live birth rate (LBR) after the first embryo transfer. Secondary outcomes were cumulative LBR (CLBR), pregnancy rate and implantation rate, both at the first embryo transfer and cumulative rates after 1-year follow-up. Percentages of biochemical pregnancy, clinical miscarriage and ectopic pregnancy from the total number of beta-HCG-positive patients as well as obstetric, delivery and neonatal outcomes were also reported. According to the International Glossary on Infertility and Fertility Care (Zegers- Hochschild, etal., 2017), the following standardized definitions were considered. Pregnancy rate Pregnancy rate is the number of patients with positive serum level of beta-HCG (≥25 mIU/ml) per embryo transfer. Live birth rate The LBR is the number of deliveries that resulted in at least one live birth per embryo transfer. Live birth is defined as the complete expulsion or extraction from a woman of a product of conception after 22 weeks of gestation, which, after such separation, breathes or shows any other evidence of life, such as heart beat, umbilical cord pulsation or definite movement of voluntary muscles, irrespective of whether the umbilical cord has been cut or the placenta is attached. Implantation rate The implantation rate is the number of gestational sacs observed by vaginal ultrasound at the fifth gestational week divided by the number of embryos transferred. Clinical miscarriage rate The clinical miscarriage rate is the number of spontaneous pregnancy losses in which a gestational sac or sacs was previously observed, per number of pregnancies. Biochemical pregnancy rate The biochemical pregnancy rate is the number of pregnancies diagnosed only by beta-HCG detection without a gestational sac visualized by vaginal ultrasound at the fifth week of pregnancy, per number of pregnancies. Ectopic pregnancy rate The ectopic pregnancy rate is the number of pregnancies outside the uterine cavity, diagnosed by ultrasound, surgical visualization or histopathology, per number of pregnancies. For cumulative outcomes, we considered the clinical results obtained from all the embryo transfers performed in the same arm of the study up to 12 months follow-up. Cumulative pregnancy rate The cumulative pregnancy rate (CPR) is the number of patients with positive serum level of beta HCG 25 mIU/ml or above divided by the total number of patients receiving embryo transfer after the same type of transfer arm into which the patient was randomized for up to 12 months’ follow-up. Cumulative live birth rate The CLBR is the number of deliveries that resulted in at least one live birth (as previously defined), per total number of patients receiving embryo transfer following the same type of transfer arm into which the patient was randomized for up to 12 months’ follow-up. All clinical outcomes are presented as percentages. Patients with positive pregnancy tests were followed to delivery. Patients without pregnancy, clinical or biochemical miscarriages, or both, or ectopic pregnancies, were followed up to 12 months after the first embryo transfer to calculate the cumulative outcomes. Sample size calculation and statistical analysis Sample size calculation was based on the hypothesis of 15-percentage point absolute difference in primary and secondary endpoints (alpha = 5%; beta = 20%) in the study group (PET) versus FET and fresh embryo transfer. 406 RBMO VOLUME 41 ISSUE 3 2020 This analysis was carried out using a SPSS Macro for one-sided significance level analysis, resulting in a sample size of at least 130 patients per group. Because of the multicentre and international nature of the study, a 30% drop-out rate was estimated owing to possible loss to follow-up, no blastocyst for transfer or no protocol compliance. Therefore, a plan was made to recruit a total of 546 patients, corresponding to 182 patients per group. This was a non-blinded study; all participants were assigned to each of the three arms using a simple equalprobability randomization method. A software-based application was used to allocate intervention (1:1:1) with randomization stratified by site. Analyses on an intention-to-treat and per protocol basis were conducted for all data. Repeated measures analysis of variance was used to compare noncategorical variables among the three groups, multiple comparisons and post-hoc tests (Bonferroni and twosided Dunnett) were applied for each one to one group comparisons. Mean differences and standard deviation or median and interquartile ranges were used when the variables were not homogeneous, as well as the mean difference with 95% CI values. Chisquared test and two-sided Fisher's exact test were used to compare the study groups with respect to percentages. Differences were estimated as relative risks with 95% CI. We conducted a multivariable regression analysis with a binomial endpoint (live birth Yes/No) to demonstrate the homogeneity of the key baseline variables and the absence of bias towards our final endpoint. We included all baseline variables presented in TABLE 1 as follows: age, body mass index, site, FSH, anti-Müllerian hormone, ethnicity and fertility background. P < 0.05 was considered to be statistically significant. SPSS 25 software (IBM, MD, USA) was used for statistical analysis. RESULTS Between 25 November 2013 and 30 April 2017 569 patients were assessed for eligibility (FIGURE 1). One hundred and eleven patients were excluded from randomization because they did not meet the inclusion criteria (n = 51), declined to participate (n = 43) or were double-randomized by mistake (n =8), or an inclusion error was detected (n = 9). The remaining 458 women were randomly assigned to either PET guided by the ERA (n = 148), FET (n = 154) or fresh embryo transfer) (n = 156). Loss to follow-up occurred in 24 patients, 27 did not receive blastocyst transfer (16 had no blastocyst, six spontaneous pregnancies occurred, two were cancelled owing to OHSS risk, and, in three cases, no embryo transfer data were found), and 139 did not fully comply with the protocol (seven had no PET in the PET group, six had PET in the non- PET groups, 43 had high progesterone, 21 had fresh embryo transfer on day 2, 3 or 4, 10 and seven had fresh and frozen embryo transfer in the PET group, 22 had fresh embryo transfer in the FET group, six had frozen embryo transfer in the embryo transfer group, 13 had frozen embryo transfer owing to OHSS risk in the fresh embryo transfer group, four protocol deviations for other reasons). The intention to treat (ITT) analysis was conducted in 434 patients, PET (n = 141), FET (n = 148) or fresh embryo transfer (n = 145). Per protocol analysis was conducted in 266 patients (PET [n = 80], FET [n = 92] or fresh embryo transfer [n = 94]) (FIGURE 1). Baseline demographics and clinical characteristics were comparable among groups, both at the ITT and per protocol analysis (TABLE 1). The FSH values were similar between PET and FET groups and higher compared with fresh embryo transfer group. About 71% of the patients recruited at their first appointment had no previous IVF attempts. Intention-to-treat analysis of cycle characteristics and embryological data were also broadly similar in all groups (TABLE 2). A statistically significant lower percentage of day-5 blastocysts, however, was transferred in the PET group compared with fresh embryo transfer (69.5% versus 82.1%, respectively; P = 0.01). Similar features were observed by per protocol analysis, where a significantly lower percentage of day-5 blastocysts was transferred in the PET and FET group compared with fresh embryo transfer (85.0% and 84.8% versus 96.8%, respectively; P = 0.01 (Supplementary Table 1), together with a significantly lower number of transferred embryos in the PET versus embryo transfer arms (1.38 versus 1.6; P = 0.01). No morphological differences were recorded in embryo quality in terms of inner cell mass or trophectoderm grading. Clinical outcomes at the first embryo transfer and cumulative rates up to 12 months by ITT analysis are shown in TABLE 3. In the PET group, CPR was significantly higher (132/141 [93.6%]) compared with FET (118/148 [79.7%]) and fresh embryo transfer (117/145 [80.7%]) groups (PET versus FET RR 1.17, 95% CI 1.07 to 1.29, P = 0.0005; PET versus fresh embryo transfer RR 1.16, 95% CI 1.06 to 1.27, P = 0.0013). Significantly more miscarriages took place in the PET than the fresh embryo transfer group at the first attempt (20.5% versus 5.9% P = 0.006) and cumulative miscarriage rate (18.2% versus 4.3%, P = 0.0006) (TABLE 3). Per protocol analysis is presented in TABLE 4. At first embryo transfer, LBR was 45 out of 80 (56.2%) in the PET group, 39 out of 92 (42.4%) in the FET group, and 43 out of 94 (45.7%) in the fresh embryo transfer group (PET versus FET RR 1.33, 95% CI 0.98 to 1.80, P = 0.09; PET versus fresh embryo transfer RR 1.23, 95% CI 0.92 to 1.65, P = 0.17). The number of patients needed to treat (NNT) by ERA (PET) to achieve the estimated LBR versus the control groups (FET and fresh embryo transfer) was calculated according to NNT= 100/Io-Ie (where Io is the LBR in the control group and Ie is the LBR in the study group). NNT of PET versus FET was 7.2 patients (Io = 42.4; Ie = 56.2), and NNT of PET versus fresh embryo transfer was 9.5 patients (Io = 45.7; Ie = 56.2); thus, these were the numbers of patients needed to treat by ERA (PET) to achieve one additional delivery with at least one live birth versus control groups. The CLBR was significantly higher in the PET group (71.2%) compared with FET (55.4%) and fresh embryo transfer (48.9%) groups (PET versus FET, RR 1.28, 95% CI 1.02 to 1.62, P = 0.04; PET versus fresh embryo transfer RR 1.46, 95% CI 1.13 to 1.87, P = 0.003). Pregnancy rate at the first embryo transfer was statistically significantly higher in the PET group versus FET. Pregnancy rate was 72.5% in the PET group, 54.3% in the FET group and 58.5% in the fresh embryo transfer group (PET versus FET, RR 1.33, 95% CI 1.06 to 1.68, P = 0.01; PET versus fresh embryo RBMO VOLUME 41 ISSUE 3 2020 407 transfer, RR 1.24, CI 1 to 1.54, P = 0.057). Statistically significant differences were also found in implantation rate in PET (57.3%) versus FET (43.2%) and fresh embryo transfer (38.6%) groups (PET versus FET, RR 1.33, 95% CI 1.03 to 1.70, P = 0.03; PET versus fresh embryo transfer, RR 1.48, 95% CI 1.14 to 1.92, P = 0.004). Cumulative pregnancy rate in the PET group (95%) was significantly higher than FET (70.6%) and fresh embryo transfer (62.8%) groups (PET TABLE 1 DEMOGRAPHIC AND CLINICAL CHARACTERISTICS OF THE PATIENTS AT BASELINE PET (n = 148) FET (n = 154) Fresh embryo transfer (n = 156) Age, years 33 ± 3.1 32.8 ± 3.4 32.7 ± 3.3 Body mass indexa22.8 ± 2.9 22.9 ± 2.9 22.6 ± 2.8 Ethnicity, n % White 122 (82.4) 127 (82.5) 129 (82.7) Asian 12 (8.1) 12 (7.8) 11 (7.1) Latin American 13 (8.8) 11 (7.1) 13 (8.3) African 0 (0.0) 4 (2.6) 1 (0.6) Other or unknow 1 (0.6) 0 (0.0) 2 (1.2) Current smoker 15 (10.1) 12 (7.8) 15 (9.6) Fertility history Duration of infertility, years 3.1 ± 1.9 3.2 ± 2.1 2.9 ± 2.2 Previous IVF failed, n (%) 0 109 (73.6) 104 (67.5) 112 (71.8) 1 20 (13.5) 23 (14.9) 22 (14.1) 2 10 (6.7) 10 (6.5) 12 (7.7) 3 6 (4.0) 11 (7.1) 6 (3.8) Unknown 3 (2.0) 6 (3.9) 4 (2.6) Previous deliveries, n (%) 1 11 (7.4) 16 (10.3) 17 (10.9) ≥2 3 (2.0) 4 (2.6) 3 (1.9) Spontaneous clinical miscarriages, n (%) 1 23 (15.5) 26 (16.9) 24 (15.4) ≥2 6 (4.0) 3 (1.9) 0 (0.0) Elective abortions, n (%) 3 (2.0) 9 (5.8) 8 (5.1) Previous curettages (1 or 2), n (%) 12 (8.1) 11 (7.1) 10 (6.4) Ectopic pregnancies, n (%) 8 (5.4) 3 (1.9) 4 (2.6) IVF indication, n % Male factor 65 (43.9) 78 (50.6) 50 (32.1) Tubal factor 20 (13.5) 31 (20.1) 33 (21.1) PCOS 27 (18.2) 20 (12.9) 14 (9.0) Ovarian disorders 4 (2.7) 5 (3.2) 7 (4.5) Endometriosis 21 (14.2) 9 (5.8) 13 (8.3) Unexplained 33 (22.3) 33 (21.4) 47 (30.1) Other or unknownb2 (1.3) 4 (2.6) 10 (6.4) Laboratory tests FSH (mU/ml) 5.9 ± 1.9c6.6 ± 2.1 6.9 ± 2.0d AMH (ng/nl) 4.4 ± 3.6 3.7 ± 2.7 3.5 ± 2.9 Data are expressed as mean ± SD unless indicated otherwise. a Body-mass index is the weight in kilograms divided by the square of the height in metres. b Other indications included same sex couples, single women, cervical factor, anatomical factor and male genetic disease. c,dStatistically significant differences between groups at <0.05 level. No significant differences were found in any of the baseline characteristics except for FSH between PET and fresh embryo transfer (P = 0.003). Analysis of variance, Bonferroni and two-sided Dunnett tests were used to compare numerical variables in the three groups. Chi-Squared and Fisher's exact tests were used to compare categorical variables. AMH, anti-Müllerian hormone; FET, frozen embryo transfer; PCOS, polycystic ovary syndrome.; PET, personalized embryo transfer. 408 RBMO VOLUME 41 ISSUE 3 2020 versus FET, RR 1.34, 95% CI 1.17 to 1.55, P < 0.0001; PET versus fresh embryo transfer, RR 1.51, 95% CI 1.28 to 1.78, P < 0.0001) (TABLE 4). For cumulative clinical outcomes (after the first attempt), most of the pregnancies and live births occurred in the second and third embryo transfer (Supplementary Table 2). No differences were found in miscarriages, biochemical or ectopic pregnancies among the three groups investigated (TABLE 4). Obstetrical outcomes, type of delivery and neonatal outcomes were not statistically significant different in all groups analysed (TABLE 5). The binomial multivariable regression analysis did not show any statistical relation between LBR and the rest of variables included in the study (data not shown). Subgroup analysis of clinical outcome was investigated according to the number of previous failed IVF cycles. The LBR and pregnancy rate in the PET group were consistently 10 or more percentage points higher in all categories regardless of the previous failed cycles but did not reach significance owing to the final sample size (Supplementary Table 3). Clinical results were also analysed according to time from endometrial biopsy carried out for the ERA test to transfer for a period of 9 months. Pregnancy rate and LBR were not statistically different throughout this period, ruling out any beneficial or detrimental effect of the injury carried out at endometrial biopsy (Supplementary Table 4). The ERA biopsy was obtained after 118.3 ± 4.6 h of progesterone administration (range 108 to 128 h). Results showed that 50 patients (62.50%) had the WOI at P + 5, whereas 30 out of 80 (37.5%) had a displaced WOI (early receptive [n = 5], late receptive [n = 18] and pre-receptive [n = 7]). In the PET arm, patients underwent embryo transfer according to ERA results; primary and secondary clinical outcomes were similar irrespective of whether PET was carried out in non-displaced (56% LBR [n = 50]) or displaced (56.7% LBR [n = 30]) WOI (Supplementary Table 5). Cost-effectiveness per baby born in Europe and the USA was also calculated in each arm of the study, using as reference the mean cost of procedures in these two geographical locations. This assessment considered the cost of the different procedures involved in each arm of the study. In all arms, the cost of monitoring, retrieval, transfer, drugs and IVF laboratory was included. In FET and PET, the cost of embryo vitrification and HRT monitoring with laboratory work was added. Finally, in the PET arm, the cost of the ERA test plus the additional HRT was incorporated. The cost per baby in the PET arm considering the technology used in the present study was lower than the FET arm (17% in Europe and 15% in USA, approximate) and higher compared with fresh embryo transfer (25% in Europe and 16% in USA, approximate) (TABLE 6). DISCUSSION This international, multicentre RCT was designed to test the effect of the endometrial factor assessed by the ERA test and clinically translated as PET. The first inclusion criterion was patients aged 37 years or younger to avoid advanced maternal age as a possible cause of embryo aneuploidies that could bias the results because of the embryonic factor; the same rationale applies for antral follicle count 8 or over and FSH less than 8 IU/ml to avoid the inclusion of poor responders. In these overall goodprognosis patients, clinical outcomes at the first embryo transfer and cumulative 569 assessed for eligibility 458 randomized 111 were excluded 50 did not meet selection criteria 43 declined to participate 8 double randomization by mistake 9 inclusion errors Enrolment 154 allocated to FET 148 allocated to PET Analysis Follow-up Allocation 156 allocated to fresh ET 141 followed up 81 proceeded with PET 1 pregnancy termination for medical reasons 80 followed up after the first embryo transfer 51 no protocol compliance 1 embryo day-3 embryo transfer 5 embryo day-4 embryo transfer 10 had fresh embryo transfer 7 had FET 18 high progesterol in ovarian stimulation 7 no PET 2 other treatment (INVO) 1 protocol deviation 11 did not receive embryo transfer 2 no to embryo transfer data 6 no blastocyst for embryo transfer 3 spontaneous pregnancy 7 lost to follow-up 148 followed up 92 proceeded with FET 92 followed up after the first embryo transfer 148 ITT analysis 92 per protocol analysis 145 ITT analysis 94 per protocol analysis 45 no protocol compliance 3 embryo day-3 embryo transfer 1 embryo day-4 embryo transfer 22 had fresh embryo transfer 5 had PET 13 high progesterone in ovarian stimulation 1 ovum donation 145 followed up 11 lost to follow-up 95 proceeded with fresh embryo transfer 94 followed up after the first embryo transfer 1 pregnancy termination for medical reasnons 43 no protocol compliance 2 embryo day-2 embryo transfer 8 embryo day-3 embryo transfer 1 embryo day-4 embryo transfer and high progesterone ovarian stimulation 6 had FET 13 had FET owing OHSS risk 1 had PET 12 high progesterone in ovarian stimulation 7 did not received embryo transfer 1 no embryo transfer data 3 no blastocyst for embryo transfer 2 cancelled owing to OHSS risk 1 spontaneous pregnancy 138 proceeded with embryo transfer 137 proceeded wth embryo transfer 6 lost to follow up 9 did not receive embryo transfer 7 no blastocyst for embryo transfer 2 spontaneous pregnancy 132 proceed with embryo transfer 141 ITT analysis 80 per protocol analysis FIGURE 1 CONSORT flow diagram: endometrial receptivity analysis randomized controlled trial. FET, frozen embryo transfer; INVO, intravaginal culture of oocytes; ITT, intention-to-treat; OHSS, ovarian hyperstimulation syndrome; PET, personalized embryo transfer. RBMO VOLUME 41 ISSUE 3 2020 409 TABLE 2 CYCLE CHARACTERISTICS AND EMBRYOLOGICAL DATA: INTENTION-TO-TREAT ANALYSIS PET (n = 141) FET (n = 148) Fresh embryo transfer (n = 145) Antral follicle count 14.8 ± 6.3 14.9 ± 6.6 13.1 ± 5.9 Antagonist protocol, n (%) 124 (87.9) 120 (81.1) 122 (84.1) Agonist protocol, n (%) 10 (7.1) 13 (8.8) 12 (8.3) Unknown, n (%) 7 (5.0) 15 (10.1) 11 (7.6) Total dose of FSH administered, IU 1696.9 ± 687.8 1540.2 ± 635.2 1666.1 ± 669.8 Total dose of HMG administered, IU 1167.03 ± 936 1202.3 ± 987 1165.1 ± 1042.5 Progesterone level at the day of ovulation triggering 1.02 ± 0.7 0.93 ± 0.6 0.92 ± 0.8 Ovulation triggering, n (%) HCG 62 (44.0)c57 (38.5)c110 (75.9)d Agonist 62 (44.0)c67 (45.3)c15 (10.3)d Double triggering 7 (5.0) 7 (4.7) 7 (4.8) Oocytes retrieved 12.4 ± 7.6 11.6 ± 6.0 10.5 ± 6.6 Fertilization technique, n (%) ICSI 106 (75.2) 114 (77.0) 111 (76.6) IVF 5 (3.5) 6 (4.1) 9 (6.2) IVF/ICSI 21 (14.9) 13 (8.8) 21 (14.5) INVO 2 (1.4) 0 (0.0) 0 (0.0) Unknown 7 (5.0) 15 (10.1) 4 (2.8) Fertilization rate, n (%) 1244/1633 (76.2) 1197/1531 (78.2) 1067/1379 (77.4) Embryo stage, n (%) Cleavage 1/181 (0.6) 0 (0.0)c7/211 (3.3)d Morula 2/181 (1.1) 1/208 (0.5) 1/211 (0.5) Early blastocyst 12/181 (6.6) 11/208 (5.3) 5/211 (2.4) Cavitated blastocyst 40/181 (22.1) 47/208 (22.6) 48/211 (22.7) Expanded blastocyst 93/181 (51.4) 100/208 (48.1) 109/211 (51.7) Hatching blastocyst 33/181 (18.2) 49/208 (23.6) 41/211 (19.4) Blastocyst development rate, n (%) 648/1248 (51.9) 636/1239 (51.3) 561/1093 (51.3) Day of embryo development at transfer, n (%)a 2 0 (0.0) 0 (0.0) 2 (1.4) 3 10 (7.1) 4 (2.7) 10 (6.9) 4 7 (5.0)c4 (2.7) 0 (0.0)d 5 98 (69.5)c112 (75.7) 119 (82.1)d 6 16 (11.3) 17 (11.5)c6 (4.1)d Unknown, n (%) 10 (7.1) 11 (7.4) 8 (5.5) Embryo quality (known/total embryos), n149/201 183/220 183/225 Inner cell mass, n (%) A grade 48/149 (32.2) 70/183 (38.3) 56/183 (30.6) B grade 84/149 (56.4) 92/183 (50.3) 110/183 (60.1) C grade 17/149 (11.4) 21/183 (11.5) 17/183 (9.3) Trophectoderm, n (%) A grade 36/149 (24.2) 56/183 (30.6) 46/183 (25.1) B grade 85/149 (57) 95/183 (51.9) 96/183 (52.5) C grade 28/149 (18.8) 32/183 (17.5) 40/183 (21.9) PGT-A cases 6 (4.3) 4 (2.7) 3 (2.1) Number of embryos per transfer 1.52 ± 0.5 1.61 ± 0.5 1.63 ± 0.5 (continued on next page) 410 RBMO VOLUME 41 ISSUE 3 2020 rates up to 12 months by ITT analysis were not statistically significant in the three arms of the study, except that CPR that was significantly higher in the PET group compared with FET and fresh embryo transfer groups, and clinical miscarriages were significantly higher in PET versus fresh embryo transfer. When the analysis was carried out per protocol, LBR significance in PET versus FET and fresh embryo transfer was not achieved at the first embryo transfer (56.2%, 42.4%, 45.7%, respectively; PET versus FET, P = 0.09; PET versus fresh embryo transfer, P = 0.17). Significantly higher CLBR after 12 months’ follow-up from PET versus FET and fresh embryo transfer, however, were observed (71.2%, 55.4%, 48.9%, respectively; PET versus FET, P = 0.04; PET versus fresh embryo transfer; P = 0.003). Implantation rate at first attempt was significantly higher in PET compared with FET and fresh embryo transfer. In addition, pregnancy rate after PET was significantly higher than FET at the first attempt and significantly higher than FET and fresh embryo transfer at cumulative rate (TABLE 4). The delivery of a healthy baby may take several treatment cycles; therefore, the CLBR per patient is a key outcome variable. To ensure that results are presented without bias, a statistical comparison was made of CLBR for each of the four cycles carried out in each arm of the study regardless of the follow-up time (Supplementary Table 2). The significant difference in CLBR in PET can be observed at the second embryo transfer. In the PET arm, outcomes were similar in the non-displaced (n = 50; LBR 56%) and the displaced group (n = 30; LBR 56.7%), indicating the importance of a timely PET. More than 45,500 data and variables corresponding to the 569 patients enrolled in the 16 active sites across continents were recorded in the study Case Report Form. These data were verified according to each patient source document during the continuous on-site and remote monitoring visits. This exhaustive source data verification allowed us to detect and correct mistakes and protocol deviations as well as to resolve queries, ensuring the study data integrity and accuracy. The study was powered to detect statistical differences for a 15-percentage point increase in the primary and secondary outcomes in the PET group guided by ERA versus FET or fresh embryo transfer. This unexpected dilution was the main reason why LBR significance in PET versus FET and fresh embryo transfer was not achieved at the first embryo transfer (56.2%, 42.4%, 45.7%, respectively; PET versus FET, P = 0.09; PET versus fresh embryo transfer; P = 0.17) (TABLE 4). The significantly lower number of blastocysts transferred in the PET arm versus fresh embryo transfer (1.38 versus 1.6; P = 0.01), as well as the lower percentage of day-5 blastocysts (85% versus 96.8%; P = 0.01), reinforces the concept of the positive effect of the personalization of endometrial factor even with embryonic ‘inferiority’ (Supplementary Table 1). Personalized medicine is still in its infancy in reproductive medicine, but most specialists in the field will agree that ‘one size does not fit all’. In fact, clinicians and embryologists have been evaluating precision medicine through assisted reproductive technology treatment for decades (Mol etal., 2018), from the starting dose and dose adjustment of gonadotrophins in ovarian stimulation according to ovarian reserve, age and BMI; determination of oocyte retrieval timing based on ultrasound findings and oestradiol and progesterone levels; selection of the insemination and fertilization technique (ICSI, IVF, or both) according to sperm features and clinical background; or embryo selection based on morphological grading, chromosomal analysis of the cleavage embryos, or both, or blastocyst. Interestingly, endometrial status at the time of embryo transfer was guided in recent decades by endometrial thickness assessed through transvaginal ultrasound together with administration or supplementation of progesterone or HCG as luteal phase support. Endometrial thickness, however, is not a diagnostic test for endometrial receptivity. Indeed, this approach has limited capacity to identify patients with low possibility to conceive after embryo transfer, except for atrophic or hyperplasic endometrium (Kasius etal., 2014). Therefore, the use of TABLE 2 – (continued) PET (n = 141) FET (n = 148) Fresh embryo transfer (n = 145) Warmed HRT embryo transfer data Days of oestradiol, n15.5 ± 3.8c16.6 ± 3.8cNA Endogenous progesterone levelsb0.2 (0.03–1.4) 0.29 (0.05–11.03) NA Exogeneous progesterone administration, h 120 ± 14.4 117.8 ± 9.7 NA Exogeneous progesterone administration, h, range 65.2–163.4 (98.2) 66.4–151.2 (84.8) NA Time between ovarian stimulation and embryo transfer, months 3.2 ± 2.4c2.1 ± 1.4dNA Data are expressed as mean ± SD unless otherwise specified. a Total number of indicated cases may differ from the excluded number of cases indicated in FIGURE 1 because some patients had more than one cause for exclusion. b Progesterone value (ng/ml) before its administration is represented as median (interquartile range). c,dStatistically significant differences between groups at the <0.05 level. Significant differences were found among groups in ovulation triggering (P = 0), embryo stage (cleavage embryo, P = 0.005), day of embryo development (P = 0.006), number of days of oestradiol (P = 0.03) and time between ovarian stimulation and embryo transfer (months) (P = 0.002). No significant differences were found in the other variables. Analysis of variance, Bonferroni and two-sided Dunnett tests were used to compare numerical variables in the three groups, whereas Independent samples Student's t-test was used to compare quantity variables between group PET and FET regarding warmed HRT embryo transfer data. Chi-squared and Fisher's exact tests were used to compare categorical variables. FET, frozen embryo transfer; HMG, human menopausal gonadotrophin; HRT, hormone replacement therapy; ICSI, intracytoplasmic sperm injection; INVO, intravaginal culture of oocytes; PET, personalized embryo transfer; PGT-A, preimplantation genetic diagnosis for aneuploidy.