No difference in cumulative live birth rates between cleavage versus blastocyst transfer in patients with four or fewer zygotes: results from a retrospective study.

After IVF, extending the duration of the embryo culture up to the blastocyst stage enables a better selection of embryos with a superior developmental capacity and consequently a higher implantation potential ( Papanikolaou et al. , 2006 ; Wang and Sun, 2014 ; Glujovsky et al. , 2016 ; Martins et al. , 2017 ). Embryo transfer at the blastocyst stage increases the pregnancy rate per embryo transferred, and this is especially important in the context of single embryo transfer (SET) policies intended to reduce multiple gestations ( Papanikolaou et al. , 2006 ). Although a blastocyst transfer policy does not appear to increase the cumulative live birth rate (CLBR), it optimizes live birth chances following the first embryo transfer and has the additional benefit of shortening the time to pregnancy ( Cameron et al. , 2020 ). Intriguingly, these benefits have not resulted in the widespread substitution of the cleavage-stage transfer policy. In good-prognosis patients, there is a clear consensus favoring a blastocyst- versus cleavage-stage embryo transfer ( Cameron et al. , 2020 ). Opinions are less clear with respect to unselected patient cohorts given the conflicting results published ( Glujovsky et al. , 2016 ; De Croo et al. , 2019 ). Moreover, little is known about the advantage of blastocyst culture in patients with a low yield of available embryos. Two retrospective studies have demonstrated that extended culture of embryos does not alter implantation potential when fewer than three embryos are available ( Vlaisavljević et al. , 2001 ; Kovačič et al. , 2002 ). Conversely, other authors maintain that as in vitro survival of embryos does not relate to in vivo survival, transferring embryos at the blastocyst stage could lead to a loss of viable embryos as a result of them not surviving the prolonged culture ( Maheshwari et al. , 2016 ). Xiao et al. (2019) confirmed this hypothesis in women with only one embryo available on D3, in whom pregnancy rates were higher when the embryo was transferred on D3 than on D5/6. However, these studies are limited by the fact that the only outcome is fresh embryo transfer. Data on subsequent cryopreservation are mostly lacking, and CLBRs are not provided. As a result, the benefits of blastocyst-stage transfer in poor-prognosis IVF patients after cryopreservation remain unknown. In our center, the embryo transfer policy was changed stepwise from cleavage-state to blastocyst-stage in 2012. This modification was accompanied by the introduction of blastocyst vitrification for all supernumerary embryos regardless of the day of fresh transfer. With the information gathered since then, we conducted a study aimed at comparing the CLBR and other relevant IVF outcomes of the two embryo transfer policies, both combined with blastocyst vitrification on D5, in poor-prognosis patients (defined as four or fewer zygotes on D1), after controlling for confounding factors.

Histogram was used to compare PS visually before and after matching. On the matched dataset, the PS for D3 and D5 showed a very similar distribution. Based on the Box-Tidwell method, the variables ‘cycle rank’ and ‘number of oocytes retrieved’ were transformed to obtain linearity in the PS matching logistic regression model. Density plots of the distance after matching were almost identical in both groups. After PS matching with the D3 group (see Materials and methods section), the D5 group consisted of 571 OCCs. The demographic and baseline cycle characteristics and ovarian stimulation data of the study groups after PS matching are shown in Table I . The women’s age, cycle rank and type of GnRH analog were comparable between the D3 group and D5 group. The laboratory parameters, including the number of oocytes retrieved, number of zygotes, number of 6- to 7- and 8-cell embryos on D3, were similar between both groups. Demographic and baseline IVF characteristics of the study groups after propensity score matching. Data are presented for oocyte collection cycles (OCCs); IQR, interquartile range; OHSS, ovarian hyperstimulation syndrome; GEE, generalized estimating equation. Selected for matching in the propensity score model. Women’s age, gonadotrophin dose: P -value based on GEE linear regression; cycle rank, number of 6- to 7- and 8-cells embryos on Day 3, number of oocytes retrieved, number of zygotes: P -value based on GEE Poisson regression; Cause of infertility, type of GnRH analog, fresh transfer, freeze-all cycles and OCC with no transfer owing to poor embryo quality: P -value based on GEE logistic regression. Statistically significant values are indicated in bold. The percentage of OCCs leading to fresh transfers was significantly higher in the D3 group compared to the D5 group (89.7 versus 59.2%; P  < 0.001). The percentage of freeze-all cycles was significantly higher in the D5 group than in the D3 group (5.6% versus 0.0%; P  < 0.001). A similar percentage of OCCs with or without cryopreservation was observed in both groups after PS matching ( P  = 0.299). A small proportion of unsuccessful cycles still had cryopreserved blastocysts in cryostorage (8/571 in the D3 group and 10/571 in the D5 group; Table I ). The percentage of DET was significantly higher in the D3 group (23.8% versus 7.0%; P  < 0.001). Positive hCG rates (26.3% versus 32.8%, P  = 0.296) and twin pregnancy positive hCG (2.6% versus 4.5%, P  = 0.736) rates did not differ between the D3 and D5 groups, respectively. After PS matching, the ongoing pregnancy rate ( P  = 0.603) and LBR per fresh transfer ( P  = 0.171) were similar in the two groups ( Table II ). Clinical outcomes per fresh embryo transfer after propensity score matching. Data are presented for fresh transfers unless indicated. Statistically significant values are indicated in bold, P -value based on generalized estimating equation logistic regression; † Per positive hCG. No differences were observed between the two groups in the percentages of SET ( P  = 1.000). Positive hCG rates ( P  = 0.862) and twin pregnancies positive hCG rates per vitrified-warmed transfer ( P  = 0.488) were similar in the two groups. Ongoing pregnancy and LBR per vitrified-warmed transfer were similar in the two groups (both P  = 0.118; Table III ). Clinical outcomes per cryotransfer after propensity score matching. P -value based on generalized estimating equation logistic regression; *Per positive hCG. No significant difference was observed for the CLBR (17.7% in the D3 group versus 16.8% in the PS-matched D5 group; P  = 0.600). The LBR from fresh transfers was also similar (15.2% in the D3 group versus 12.4% in the PS-matched D5 group; P  = 0.160). No differences were observed in the cumulative twin LBRs per OCC retrieved ( P  = 0.999; Table IV ). Cumulative outcomes after propensity score matching. Data are presented for oocyte collection cycles. P -value based on generalized estimating equation logistic regression.

This was a retrospective, observational, single-center, cohort study conducted at the University Hospital of Ghent (Belgium) between January 2014 and December 2018. Embryos from both fresh IVF and ICSI oocyte collection cycles (OCCs) were included. The sperm samples used were either fresh or frozen partner ejaculates, fresh or frozen surgically retrieved spermatozoa, or frozen donor ejaculates. OCCs were excluded in case of preimplantation genetic testing, oocyte donation cycles, cycles with no oocytes retrieved, no sperm available on the day of oocyte collection and no zygotes or only abnormal zygotes available on Day 1. Poor prognosis was based on our experience and defined as four or fewer zygotes on Day 1. OCCs performed between January 2014 and December 2015 were scheduled for transfer on Day D3 (n = 571), while those performed between January 2016 and December 2018 were scheduled for transfer on D5 (n = 1788; Fig. 1 ). Supernumerary embryos were cultured until D5 and vitrified as blastocysts. The reasons for freeze-all cycles included risk for ovarian hyperstimulation syndrome, elevated progesterone on the day of hCG trigger, or medical conditions (hydrosalpinx, suboptimal endometrium, fever). OCCs were also excluded if the transfer policy protocol in use was not followed. There was no restriction on the woman’s age. Flowchart of the study groups in a retrospective assessment of cumulative live birth rates after cleavage versus blastocyst transfer in patients with four or fewer zygotes. OCC, oocyte collection cycle; PS, propensity score; cryo, cryopreservation. The relevant data on cycles were extracted from electronic patient records (Ideas V6, Mellowood Medical) and stored in a database. Propensity score (PS) matching was performed to reduce the bias between study groups (D3 versus D5 transfer) resulting from certain baseline demographic, clinical and embryologic confounding factors. For pituitary downregulation, short- and long-GnRH agonist and GnRH antagonist protocols were used. The short-agonist protocol was started after at least 14 days of ethinylestradiol 50/levonorgestrel 150 (M50) treatment (Microgynon 50 ® ; Bayer Pharma AG, Berlin, Germany). After stopping M50 (‘Day 0’ of the IVF cycle), a GnRH agonist (GnRH-a) (Triptorelin; Decapeptyl ® ; Ferring, Hoofddorp, The Netherlands) was started on Day 3 until LH or hCG administration. Gonadotrophins (FSH: Gonal-F ® , Serono Benelux, London, UK; or Puregon R; MSD, Oss, the Netherlands; or human menopausal gonadotropin (HMG): Menopur ® ; Ferring, Hoofddorp, The Netherlands) were added starting on Day 5. The long agonist protocol started using decapeptyl depot on Day 21 of the previous natural menstrual cycle. After at least 14 days of GnRH-a pretreatment, additional gonadotrophin administration was started. In both agonist protocols, controlled ovarian stimulation was achieved using daily doses of between 150 and 300 IU of gonadotrophins. For the antagonist protocol, gonadotrophins were started on Day 3 of the natural menstrual cycle, and a GnRH antagonist (Cetrorelix 0.25 mg; Cetrotide ® , Merck Serono, Geneva, Switzerland) was injected s.c. as a daily dose from Day 6 of the cycle until the day before oocyte retrieval. Follicle aspiration was performed 34–36 h after hCG (Pregnyl 5000 IU ® ; MSD Oss, the Netherlands) or recombinant hCG (Ovitrelle ® 6500 IU, Serono Benelux, London, UK) injection. The women were treated with intravaginal progesterone (Utrogestan ® , Besins Healthcare, Brussels, Belgium) starting on the day of hCG or recombinant LH injection to support the luteal phase. Biochemical pregnancy was defined as positive serum levels of hCG 16 days after oocyte retrieval. As for laboratory procedures, no changes were made to the stimulation protocols used during the study period. Oocytes were fertilized by IVF or ICSI. The embryos were cultured individually in sequential media (Cleavage and Blastocyst medium, Cook, Bloomington, IN, USA) in 25 µl microdroplets under oil (Irvine Scientific, Ireland) in a 6% CO 2, 5% O 2 and 89% N 2 incubator at 37°C (Binder 210, VWR, Belgium). Fertilization was checked 16–19 h after IVF or ICSI. Embryo development was evaluated daily until the transfer day. The quality of embryos on Days 2 and 3 was assessed on the basis of the number of blastomeres, the rate of fragmentation and the presence of multinucleation. On Day 4, the evaluation included an assessment of the compaction stage. Assessment on Day 5 was based on the classification system proposed by Gardner and Schoolcraft (1999) , where the embryo ideally develops to the blastocyst stage. Criteria for embryo transfer on Day 3 was having at least six cells, <30% fragmentation and no signs of multinucleation. Criteria for embryo transfer on Day 5 were having blastocysts with at least an expansion stage 1. From expansion stage 3 onwards, an inner cell mass score of A, B or C and trophectoderm score of A, B or C was required. Embryos in the compaction stage were not transferred or vitrified. All transfers were performed using a Cook embryo replacement catheter (Sydney IVF, Cook, USA). The maximum number of embryos transferred was calculated and performed according to the Belgian legislation ( De Neubourg et al. , 2013 ; Peeraer et al. , 2014 ). This calculation takes into account the women’s age, the embryo quality and the rank of the cycle. Women aged <36 years in their first treatment cycle receive one embryo, regardless of the quality of the embryo. For the same age group, in a second treatment cycle, one embryo is transferred unless it is of insufficient quality (in which case double embryo transfer (DET) is possible). In older women or in subsequent cycles, the number of embryos never exceeds two except in women aged >39 years, in whom there is no maximum quantity of embryos to be transferred. Supernumerary blastocysts with at least an expansion stage 1, inner cell mass score of A, B or C, and trophectoderm score of A, B or C were cryopreserved on Day 5. The vitrification procedure was performed using CBS-VIT High-Security straws (CryoBiosSystem, L’aigle, France) with dimethyl sulfoxide-ethylene glycol-sucrose as the cryoprotectants (Irvine Scientific Vit Kit-Freeze, Ireland). For patients with regular ovulatory cycles, embryos were transferred in a natural cycle. During natural cycles, patients were monitored with transvaginal ultrasound and serum estradiol (E2) and LH concentrations. For patients with irregular ovulatory cycles, endometrial preparation was initiated by oral administration of 6–12 mg estradiol valerate (Progynova ® , Bayer, Belgium) once a day until the endometrial thickness was >6 mm on transvaginal ultrasound (artificial cycles). At that moment, once daily 3 × 200 mg micronized progesterone vaginally (Utrogestan ® , Besins, Belgium) was added to the daily oral estradiol intake. The first day of progesterone application was set as Day 0 for calculating the day of thawing. The transfer of warmed embryos was performed on the sixth day after ovulation. Embryos were warmed 1 day before the transfer day. In accordance with Belgian legislation, a maximum of two embryos were transferred per woman. The primary outcome was CLBR per OCC, which included fresh and vitrified-warmed transfers to account for the first live birth ( Maheshwari et al. , 2015 ). CLBR was assessed for transfers occurring up to 18 months following OCC. The secondary outcome was live birth rate (LBR) per fresh transfer, defined as the number of deliveries that resulted in at least one live birth, expressed per 100 cycle attempts ( Zegers-Hochschild et al. , 2017 ). The R packages ‘MatchIt’ and ‘optmatch’ were used to apply optimal matching in a 1:1 ratio ( Ho et al. , 2011 ). The matchit() function of the R package ‘MatchIt’ was applied to estimate the PS using logistic regression (logit) based on the following variables: women’s age at start cycle, cycle rank, type of GnRH analog, number of oocytes retrieved, number of zygotes on Day 1, number of 6-cell embryos on D3, number of 7-cell embryos on D3 and number of 8-cell embryos on D3. The Box-Tidwell method was used to test the linearity of the continuous predictors versus the log(odds) and to apply the appropriate transformation. PSs were compared using density plots. The model converged after four iterations, matching was applied without replacement, and no caliper was used. The baseline characteristics, ovarian stimulation outcomes and clinical outcomes were evaluated before and after matching. The matched dataset was used to compare primary and secondary outcomes. Generalized estimating equation (GEE) with the patient as subject was applied to adjust the clustering of cycles within women. Descriptive statistics were expressed as mean ± SD for continuous data with a normal distribution, whereas those with skewed distributions were expressed as median with interquartile ranges. Frequencies and percentages were used to present categorical data. Differences were analyzed with GEE linear regression (women’s age, gonadotrophins dose), GEE Poisson regression (cycle rank, number of 6- to 7- and 8-cell embryos on D3, number of oocytes retrieved, number of zygotes) and GEE logistic regression (cause of infertility, type of GnRH analog, fresh transfer, freeze-all cycles, OCC with no transfer owing to poor embryo quality, clinical outcomes per fresh transfer, clinical outcomes per cryo transfer and cumulative outcomes per OCC). A P -value of <0.05 was considered statistically significant. Statistical analysis was performed using R version 3.5.1. (R Foundation for Statistical Computing, Vienna, Austria). This retrospective study was approved by the local ethics committee of the University Hospital of Ghent, Belgium (B 670201731234).

In this study, comparing a cleavage-stage versus a blastocyst-stage transfer policy in populations matched according to relevant demographic, clinical and embryologic characteristics, no differences were observed in the CLBR per OCC, both being close to 17% when combined with blastocyst vitrification on Day 5. The LBR per fresh transfer was also comparable between both groups. The superiority of the cumulative outcomes after blastocyst embryo transfer with respect to cleavage-stage embryo transfer in poor-prognosis IVF patients is currently a matter of debate. Our observations support the findings of the retrospective study conducted by Yin et al. (2017) in matched patients who had only one or two embryos on Day 3, in whom embryos were transferred either on Day 2/3 or 5/6 (n = 217 for each group). In the Yin et al. (2017) study, the CLBR was comparable between the two groups (53.0% for Day 5/6 versus 49.8% for Day 2/3; P = NS). The cumulative pregnancy rate was also similar (57.1% versus 53.5% for Day 5/6 versus Day 2/3, respectively; P = NS). However, the time to live birth was significantly lower in the Day 5/6 group. The authors concluded that a Day 5/6 embryo transfer was a more cost- and time-efficient policy. In another retrospective study in patients with one or two cleavage-stage embryos, Haas et al. (2019) also reported a similar cumulative pregnancy rate per retrieval (22% versus 24.6%; P = NS) and ongoing pregnancy rate per retrieval (20% versus 20.2%; P = NS) in patients with 1–2 cleavage-stage embryos that were transferred either on Day 3 (n = 102) or Day 5 (n = 429), respectively. On Day 3, the number of mature oocytes and cleavage-stage embryos was similar in the two groups. Conversely, in another retrospective study in patients with only one viable embryo on Day 3, Xiao et al. (2019) showed that embryo transfer on this day (cleavage stage) resulted in a higher clinical pregnancy rate (14.7% versus 6.8%) and LBR (9.7% versus. 4.4%) than culturing until the blastocyst stage (Days 4–6), respectively, with these differences persisting after adjusting for significant confounders. The superiority of the cleavage-stage transfer policy in patients with few oocytes was also reported by Freeman et al. (2000) despite both policies achieving similar success rates (clinical and ongoing pregnancy rates) in unselected populations. The advantages of blastocyst transfer in patients with low ovarian response have also been gradually recognized. There have been few studies targeting the best embryo transfer strategy for patients with different ovarian responses. Two studies ( Papanikolaou et al. 2008 ; Alfaraj et al. , 2017 ) indicated that in patients with low ovarian response, the LBR was higher in a blastocyst transfer than in a cleavage-stage transfer. The study of Freeman et al. (2000) showed similar success rates in cleavage- and blastocyst-stage transfer strategies, with the exception of patients with few oocytes. Our definition of poor-prognosis patients is based on less than five zygotes on Day 1, which includes a very heterogeneous population associated with differences in the underlying etiology. The low yield of zygotes may be related to poor ovarian response to stimulation yielding few mature oocytes or intrinsic abnormalities in either the oocyte or the sperm, impacting fertilization. However, since we matched on female age, number of oocytes retrieved and number of zygotes on Day 1, subpopulations of women with low ovarian reserve and/or low fertilization rate were not analyzed. These discrepancies are to be elucidated in the ongoing multicenter, non-inferiority, randomized controlled PRECISE trial (ClinicalTrials.gov Identifier: NCT03764865 ) aimed at comparing blastocyst to cleavage-stage embryo transfer in poor-prognosis patients (defined as having ≤5 zygotes on Day 1 after fertilization), whose protocol has recently been published ( Neuhausser et al. , 2020 ). The primary outcome is LBR per retrieval (time frame: 9 months). The study will enroll 658 eligible women at six IVF centers over the course of 22 months. The estimated completion date is 28 February 2024. As in other studies comparing both transfer policies in unselected populations ( Papanikolaou et al. , 2008 ; Glujovsky et al. , 2016 ; Alfaraj et al. , 2017 ), the stepwise change from a cleavage-stage to a blastocyst-stage transfer policy in our center resulted in a similar CLBR per OCC (25.9% for Day 3 versus 28.4% for Day 5, P = NS; De Croo et al. , 2020 ). However, we observed a dramatic increase in transfer cancelation rates, especially in poor-prognosis women (1–4 zygotes available on Day 1). This change in transfer policy was accompanied by the introduction of blastocyst vitrification for all supernumerary embryos, regardless of the day of fresh transfer. This policy gave us the opportunity to analyze the data retrospectively and see if a hybrid transfer and cryopreservation policy might benefit poor-prognosis patients, i.e. fresh embryo transfer on Day 3 and cryopreservation policy on Day 5. Most studies comparing CLBR with different transfer policies use selected patient cohorts based on age ( Rienzi et al. , 2002 ; De Vos et al. , 2016 ), available zygotes ( Rienzi et al. , 2002 ; De Vos et al. , 2016 ; Levi-Setti et al. , 2018 ), number of available embryos on Day 3 ( De Vos et al. , 2016 ; Yin et al. , 2017 ) or a combination of these factors ( Milki et al. , 2000 ; De Vos et al. , 2016 ; Levi-Setti et al. , 2018 ). Deciding the transfer policy based on the number of oocytes, on the one hand, or the number of embryos on Day 3, on the other hand, has disadvantages, as: the effect of low fertilization is not taken into account when the number of retrieved oocytes is chosen to decide the day of transfer; decision-making on Day 3 based on the number of embryos will have the consequence of additional communications with the patients, who will have to be able to come to the clinic on that same day, and patients also need to be available for a transfer on both Day 3 and Day 5; and allocation of patients for the day of transfer based on the number of available embryos on Day 3 is often a clinical decision ( De Vos et al. , 2016 ) influenced by unknown variables. The possibility of deciding the day of transfer on Day 1, based on the number of zygotes, has the benefit of being able to communicate the day of transfer to patients when the fertilization results are reported. The decision between a cleavage-stage transfer policy and a blastocyst-stage policy, in our study, is the result of a change in transfer policy during the two time periods and is thus independent of any individual clinical decision, which adds to an objective allocation of patients to the transfer policy. The use of a high-quality vitrification blastocyst cryopreservation policy on Day 5 could be an interesting combination with a fresh embryo transfer on Day 3 in patients with few zygotes. The proportion of cryopreserved embryos per OCC was similar for cleavage-stage and blastocyst-stage transfer ( P  = 0.299). Two publications evaluated the added value of the vitrified embryos in a Day 3 transfer policy ( Fernández-Shaw et al. , 2015 ; De Vos et al. , 2016 ). Although survival after warming of cleavage-stage embryos with vitrification is higher than with slow-freezing, these studies failed to confirm a higher cumulative success rate despite a higher number of embryos vitrified on Day 3. These reports show that it is not the higher number of vitrified embryos that make a difference in CLBR. A more selective cryopreservation program on Day 5 can equilibrate the CLBR between fresh transfers on Day 3 and Day 5, as shown in our study. Transfer cancelation rates in blastocyst transfer policies range between 0% and 28% ( Papanikolaou et al. , 2008 ). We observed a cancelation rate of 33.3% in the Day 5 group, which is slightly higher than the 30% OCC cancelation rate with 1–2 oocytes, 14% with 3 oocytes and 10% with 4 oocytes reported in the literature ( Connell et al. , 2019 ). This may be explained by our strict transfer policy, as we only transfer on Day 5, and no compaction or earlier stages are considered. Our data show similar LBRs with a significantly higher percentage of SET in the Day 5 group (93.0% versus 76.2% SET in the Day 3 group; P  < 0001). Single blastocyst-stage transfer not only reduces multiple gestations but also results in a higher pregnancy rate per transfer compared to a single cleavage-stage transfer ( Papanikolaou et al. , 2006 ). It has been reported that the LBR after transfer of a single fresh blastocyst is similar to that of transfer of two fresh cleavage-stage embryos ( Thurin et al. , 2004 ). The main limitation of our study stems from its retrospective nature. Despite being nonrandomized, the PS matching allowed comparability of the study populations with respect to confounding factors in a ‘context-based medicine’ approach ( Macklon and Fauser, 2020 ). PS matching is gaining interest in studies evaluating infertility treatments ( Yin et al. , 2017 ; Gu et al. , 2019 ; Qu et al. , 2019 ; Haviland et al. , 2020 ). No major changes were made to the laboratory practices during the shift from a cleavage-stage to blastocyst-stage transfer policy. In conclusion, our retrospective PS matching study in poor-prognosis patients (four or fewer zygotes) shows a similar CLBR per OCC, and LBR per fresh transfer, regardless of the day of fresh transfer, when combined with blastocyst vitrification on Day 5. The blastocyst-stage transfer policy was accompanied by a higher percentage of SET, and the cumulative outcomes were achieved regardless of a higher transfer cancelation rate in the blastocyst group. Our results may benefit clinics not willing to shift to a blastocyst transfer policy in all patients, as similar cumulative outcomes may be achieved in patients with four or fewer zygotes on Day 1 when transfer on Day 3 is combined with blastocyst vitrification on Day 5.