Influence of post-thaw culture duration on pregnancy outcomes in frozen blastocyst transfer cycles

Abstract In this study, we aimed to evaluate whether post-thaw culture duration affected the clinical outcomes of frozen blastocyst transfer. This retrospective cohort study included 3,901 frozen-thawed blastocyst transfer cycles. The cohorts were divided into two groups based on the developmental stage (day 5 [D5] and day 6 [D6]) and culture duration after thawing (short culture, 2–6 h; long culture, 18–20 h). Women in the short culture group following D6 blastocyst transfer were further divided into three subgroups depending on the post-thaw culture period (2, 4, and 6 h). The main outcomes, namely live birth rate (LBR), implantation rate (IR), clinical pregnancy rate (CPR), and abortion rate (AR), showed no statistical differences within the groups following D5 blastocyst transfer. Patients in the long culture group had significantly lower IR (35.5 vs. 45.8%, p < 0.001), CPR (45.3 vs. 56.6%, p = 0.001), and LBR (35.5 vs. 48.5%, p < 0.001) but a significantly higher AR (21.6 vs. 14.3%, p = 0.049) following D6 blastocyst transfer than those in the short culture group. However, the data failed to present the superiority of any short culture duration over another on the live birth outcome for embryos vitrified on D6 (adjusted odds ratio [aOR]: 0.96, 95% confidence interval [95% CI]: 0.53–1.73, p = 0.881, for the 4-h vs. 2-h subgroup; aOR: 1.01, 95% CI: 0.68–1.49, p = 0.974, for the 6-h vs. 2-h subgroup). Both post-thaw protocols can be applied to patients with D5 blastocysts. To optimize the pregnancy outcomes following D6 blastocyst transfer, a short culture period is recommended. Any of the three short culture durations (2, 4, and 6 h) can be applied, depending on the workflow of the laboratory.


Introduction
With the development of the cryopreservation method of vitrification, the need for dealing with supernumerary embryos, and the objective of decreasing the ovarian hyperstimulation syndrome risk, frozen-thawed embryo transfer (FET) has become a commonly used method in various in vitro fertilization treatments (Roque et al. 2019).
The selection criteria for thawed embryos before transfer must be carefully planned and rigorous to maximize the pregnancy outcomes. Normally, embryos are transferred after a short culture duration, that is, after thawing for 1-6 h (Ahlstrom et al. 2013;Ferreux et al. 2018;Tubbing et al. 2018). In contrast, some fertility centers use a long post-thaw culture interval (16-24 h) (Kang et al. 2013;Haas et al. 2016;Yang et al. 2016). One advantage of having the different protocols is the flexibility of laboratory workflow; embryos can be thawed in the afternoon one day before transfer on a working day or in the morning of a transfer day during weekends and holidays (Fang et al. 2016). Moreover, the uterine cavity is an optimal incubator, and in vitro embryo culture conditions cannot replicate the fallopian tube and uterine environments in vivo. As such, some blastocysts fail to develop in vitro but can show high implantation rates in vivo (Haas et al. 2018). Therefore, a short postthaw culture protocol is more feasible. However, several publications have suggested that a short culture period might be insufficient to evaluate the developmental potential of blastocysts and that a long culture period could increase the degree of blastocoel expansion and provide more valuable information required for the selection of embryos (Guerif et al. 2003;Du et al. 2016;Minasi et al. 2016;Herbemont et al. 2018).
Currently, there is no consensus on a culture regimen that yields the best pregnancy results. Some studies have suggested that the post-thaw culture duration should be considered as a confounding factor of FET outcomes (Rato et al. 2012;Haas et al. 2018;Hwang et al. 2020). One observational study indicated that a long-term incubation of 20 h could increase the implantation rate (IR) three-fold compared to a shortterm incubation of 4 h (Guerif et al. 2003). However, other researchers have reported contradictory results; pregnancy outcomes were comparable for blastocysts thawed and cultured overnight than those thawed and transferred on the same day (Fang et al. 2016;Hwang et al. 2020). Nonetheless, their studies included both day 5 (D5) and day 6 (D6) blastocysts, and conclusions could not be drawn for each embryo stage. Herbemont et al. (2018) conducted a prospective randomized study that enrolled patients younger than 38 years of age with only good-quality D5 blastocysts and demonstrated no difference in IR, regardless of the culture duration. Nevertheless, there are no data on the suitability of D6 blastocysts for transfer, which represent a large proportion of blastocyst transfer cycles.
Therefore, we aimed to evaluate the impact of post-thaw embryo culture on clinical outcomes of embryos developed on D5 and D6 after blastocyst transfer to provide novel insights to clinicians and embryologists regarding the warming protocols. Figure 1 shows the selection process. Of the initial 3,901 FET cycles during the study period, 995 cycles were excluded. Overall, 2,906 blastocyst embryo transfers, including 2,035 D5 blastocyst transfers, 871 D6 blastocyst transfers, 950 short culture cycles, and 1,956 long culture cycles, met the eligibility criteria for analysis. We analyzed FET cycles in two distinct parts: part 1 followed D5 embryo transfers, and part 2 followed D6 embryo transfers.

D5 blastocyst transfers
The short culture group consisted of 406 cycles before propensity score matching (PSM) and 384 cycles after matching. The number of participants in the long culture group before and after PSM was 1,629 and 384, respectively. Unpaired and paired baseline characteristics are presented in Table 1. The proportions of type, duration, cause of infertility, and endometrial preparation protocol, as well as female age, anti-M€ ullerian hormone (AMH) level, and number of transferred embryos, revealed significant differences within groups before PSM analysis (p < 0.05). After PSM, the demographics of the matched groups were comparable (p > 0.05).
The morphological characteristics of the thawed blastocysts before vitrification and transfer are summarized in Table 2. When analyzing the embryo status before vitrification, there were no significant differences in blastocoel grade, the number of good-quality embryos, and the number of cycles with good-quality embryos between the two groups (P > 0.05). Surviving blastocysts were defined as those with partial or complete blastocoel re-expansion and >50% intact cells. The survival rates in the short and long culture groups were similar (97.6 vs. 97.4%, p ¼ 0.832). There was a significant difference in the degree of blastocoel expansion between the short and long culture groups (B5 and B6 stage: 26.2 vs. 95.7%, p < 0.001). Table 3 presents the FET outcomes after PSM. There were no significant differences between the two groups in terms of IR (59.1 vs. 63.4%, p ¼ 0.170), clinical pregnancy rate (CPR; 63.8 vs. 66.9%, p ¼ 0.363), abortion rate (AR; 11.8 vs. 14.4%, p ¼ 0.396), and live birth rate (LBR; 56.3 vs. 57.3%, p ¼ 0.771).
The baseline characteristics of the patients who had a live birth and those who did not are presented in Supplementary Table 1. With post-thaw culture duration as the main exposure of interest, the covariate variables in the logistic regression analysis were patient age, infertility type, serum levels of basal follicle-stimulating hormone (FSH) and AMH, endometrial thickness (EMT), transferred embryo number, good-quality embryo number before transfer, and culture duration (short vs. long). The analysis revealed that there was no statistical difference in LBR in the short culture protocol compared with the long culture protocol (adjusted odds ratio [aOR]: 0.94, 95% confidence interval [CI]: 0.70-1.27; p ¼ 0.687).

D6 blastocyst transfers
In total, 544 FET cycles in the short culture group and 327 cycles in the long culture group were enrolled in the analysis. There were no statistically significant differences in the baseline clinical parameters between the two groups (p > 0.05) ( Table 4).
The morphological characteristics of blastocysts before vitrification and transfer are listed in Table 5. Blastocele and embryo quality were comparable between the groups before vitrification. There was no statistically significant difference in the survival rate between the two groups (97.8 vs. 97.2%, p ¼ 0.540). As expected, after a longer culture period, the proportion of hatching (B5) or hatched (B6) blastocysts was significantly higher in the long culture group than that in the short culture group (grade B5 and B6: 82.1 vs. 16.6%, p < 0.001), whereas the embryo quality was similar between the two groups. Table 6 presents the clinical outcomes of the patients. Cycles with a short post-thaw culture period had a significantly higher IR (45.8 vs. 35.5%, p < 0.001) and CPR (56.6 vs. 45.3%, p ¼ 0.001) than those with a long culture period. Meanwhile, a significantly lower AR was found in the short culture group (14.3 vs. 21.6%, p ¼ 0.049) than in the long culture group. Additionally, 264 patients (48.5%) had live births after a post-thaw culture duration of 2-6 h  Regression analysis demonstrated that the impact of culture duration on LBR remained significant in the prediction model (aOR 1.73, 95% CI 1.29-2.32; p < 0.001 for the short vs. long culture group), which was adjusted for maternal age, infertility cause, number of transferred embryos, number of cycles with good-quality embryos before transfer, and culture duration (Supplementary Table 2).
The short culture protocol had significantly better clinical outcomes than the long culture protocol. We further analyzed the influence of three different short culture durations on LBR (2 h vs. 4 h vs. 6 h). The initial cohort of 544 FET cycles was divided into two groups: the LB group (cycles reached live births) and the non-LB group (cycles failed to end with live  Good-quality blastocysts before vitrification were those with scores of 3BB or above (grade 3-6 AA/AB/BA/BB) and those re-expanded with grades of !3BB in the short culture group and !5BB in the long culture group before transfer. births). As shown in Table 7, there was a statistically significant difference in the proportion of good-quality blastocyst transfers between the two groups (p < 0.001). Meanwhile, no significant difference in culture duration (p ¼ 0.901) was observed. Table 8 shows the logistic regression model used to predict live birth. The summary effects of LBR in 4-h and 6-h subgroups versus 2-h subgroup can be expressed as follows: aOR 0.96 (95% CI 0.53-1.73, p ¼ 0.881) and aOR 1.01 (95% CI 0.68-1.49, p ¼ 0.974), respectively.

Discussion
The results of our study demonstrated that different post-thaw culture durations had similar clinical outcomes in D5 embryo transfer cycles. However, in frozen-vitrified D6 embryo transfer cycles, the IR, CPR, and LBR were statistically higher after a short period of warming (2-6 h) than a long period of warming (18-20 h). In terms of LBR, no difference was observed between the three short culture intervals (2, 4, or 6 h). Currently, the subject of which post-thaw culture duration is best remains debated. In 2003, a retrospective study compared the effects of a short or long period after warming (4 h vs. 20 h) on the clinical outcomes of frozen blastocyst transfer cycles (Guerif et al. 2003). In contrast to our study, a longer culture interval increased the CPR three-fold (27.0 vs. 8.0%) and the IR four-fold (23.4 vs. 6.1%) after a 20-h incubation period compared to a 4-h incubation period.   However, the results of this study should be interpreted with caution because there are some major flaws. In addition, a study that involved frozen embryo transfer cycles of D5, D6, and D7 blastocysts showed that pregnancy outcomes were comparable for blastocysts thawed and cultured overnight one day before transfer and those thawed and transferred on the same day (Fang et al. 2016). In another study, Hwang et al. (2020) enrolled patients with both vitrified D5 and D6 blastocysts for transfer and demonstrated that a 2-to 4-h culture interval yielded similar IR, CPR, AR, and LBR to those with a 20-to 24-h culture interval. These two studies collectively examined blastocysts from two different embryo developmental stages, neglecting the fact that D5 and D6 embryos may have metabolic or epigenetic differences, which can lead to some other conclusions (Irani et al. 2018). Here, we found that D5 blastocysts could be transferred at any culture duration, while a short culture duration in D6 blastocyst transfer cycles is preferred. The discrepancy between the two previously published studies (Fang et al. 2016;Hwang et al. 2020) and our study may be related to the inclusion of embryos at different developmental stages. For instance, Hwang et al. (2020) reported that the majority ($ 80%) of the enrolled cycles were D5 transfer cycles, and the remaining enrolled cycles consisted of D6 embryo transfer cycles. Although the molecular mechanisms associated with our findings, namely that the transfer of D6 blastocysts after short-term culture is associated with better pregnancy outcomes than long-term culture, are unclear, they may be explained as follows. Synchronization of the embryo and endometrium is paramount for successful clinical pregnancy; however, it is important to note that optimized embryo-endometrial synchrony is often vague and obscure. Blastocyst transfers are normally performed on the sixth day after progesterone (P) administration (Mackens et al. 2017). Blastocysts developed on D6 are cultured for an additional 24-h period before vitrification  compared with D5 blastocysts. If embryo thawing is scheduled one day before transfer, another 16-to 24-h culture period will ensue, which might contribute to asynchrony between the embryo and endometrium, leading to a decreased IR and a higher AR. Moreover, the prolonged culture time after warming can increase the degree of blastocoel expansion, indicating that a higher proportion of embryos can develop to stage 5 or 6 (Herbemont et al. 2018;Hwang et al. 2020). Marcos et al. (2015) utilized a time-lapse system and observed that human blastocysts reaching stage 5 could experience one or more collapse-expansions of the blastocoel cavity. Their findings revealed that the collapse of blastocysts adversely affected the embryo implantation potential. The collapse-expansion phenomenon was further confirmed in subsequent publications (Bodri et al. 2016;Sciorio et al. 2020). Based on the available evidence, the incidence of collapse-expansion is expected to be much higher after subjecting slowly developed D6 embryos to a longer post-thaw culture. Therefore, we postulate that poor transfer outcomes for D6 blastocysts might be attributed to the high mechanical stress and excessive energy consumption during the repeated collapse-expansion events. In addition to investigating whether a short postthaw duration results in better pregnancy outcomes for D6 blastocysts, the best specific interval remains a debatable subject that would be meaningful for laboratory practice. There were three short post-thaw culture durations of 2, 4, and 6 h in our study. Multivariable logistic regression analysis revealed that the LBR did not differ, regardless of culture duration. A review of the literature showed no clear agreement on the time of blastocoel re-expansion after warming. In a study by Ahlstrom et al. (2013), most of the blastocysts were assessed for re-expansion 1 to 5 h after warming, and the results indicated that the time made little contribution to the re-expansion degree once an interval of 2 h was reached. Hwang et al. (2020) also reported that 90% of blastocysts completed re-expansion within an average of 1.4 to 3.5 h after warming. A longer in vitro culture protocol might increase metabolic or mechanical stress based on the already compromised D6 blastocyst viability. Compared with the 2-h culture group, the 6-h culture group showed no detrimental effects on the LBR (aOR 1.01, 95% CI 0.68-1.49, p ¼ 0.974), indicating that a culture period of up to 6 h could still reliably ascertain embryo viability and developmental competency. Nonetheless, the sample size of this subgroup was relatively small for effective statistical analysis. Therefore, further investigation in a large-scale clinical trial is warranted.
No statistically significant reduction in LBR after long-term culture was observed for blastocysts vitrified on D5. Since the number of good-quality transferred embryos was equally distributed between the two groups, our results were in line with a recent study that enrolled 162 blastocysts transfer cycles with only good-quality D5 blastocysts (Herbemont et al. 2018). The authors investigated whether postthaw culture duration (1 h vs. 18 h) could influence FET results. Their prospective randomized study demonstrated that IR, CPR, and AR were similar, regardless of the post-thaw culture duration. As such, both warming protocols can be applied to patients with good-quality D5 blastocysts.
To date, several studies have suggested a correlation between blastocyst developmental competence and a higher degree of blastocele re-expansion after thawing (Du et al. 2016;Herbemont et al. 2018;Hwang et al. 2020). It is understandable that compared with the short duration group, the long culture group showed a better ability to promote full expansion of blastocoels. As shown in a prospective study (Herbemont et al. 2018), the group with a post-thaw interval of 16 to 22 h had a higher proportion of B5/ 6 grade at the time of transfer than the group with an interval of 0.5 to 5 h (38.6 vs. 12.7%, p < 0.001). Additionally, B5/B6 grade blastocysts in the long culture protocol demonstrated a significantly higher IR than B4 grade blastocysts in the short culture protocol (52.9 vs. 32.9%, p ¼ 0.048). Consequently, the implantation potential of blastocysts cannot always be evaluated after short post-thaw culture intervals. Maezawa et al. (2014) applied time-lapse imaging and revealed that blastocysts that failed to expand 5 h after thawing could not develop to the hatched stage and 25% of blastocysts remained shrunken after 6 h. Thus, an overnight culture protocol in D5 blastocyst transfer cycles can not only select embryos with reexpansion, but also maintain their developmental ability until hatching. Currently, data have shown the priority of transferring D5 blastocysts to D6 blastocysts in both fresh (Ozgur et al. 2015;Franasiak et al. 2018) and frozen (Ferreux et al. 2018;Bourdon et al. 2019) embryo transfer cycles. The underlying reasons can be partly explained by better embryo quality (Ferreux et al. 2018), decreased abnormal spindle rate (Hashimoto et al. 2013), and low aneuploidy incidence (Hernandez-Nieto et al. 2019;Tiegs et al. 2019). All of these reasons suggest that blastocysts developed on D5 might have higher quality and improved implantation potential than D6 embryos to cope well with extended warming periods. To conclude, the abovementioned reasons indicate that a long post-thaw culture period is harmless to the clinical outcomes of D5 blastocyst transfer cycles compared with a short post-thaw culture period.
To the best of our knowledge, the current study is the first to use a large sample size to examine the effects of post-thaw culture duration on frozen blastocyst transfer outcomes involving both D5 and D6 embryos. A major strength of this study was that we evaluated both vitrified/warmed D5 and D6 blastocyst transfer cycles and observed that the postthaw culture duration played an important role in predicting the LBR of D6 blastocysts. Our study also examined frozen D6 blastocyst transfer outcomes from the viewpoint of three short culture strategies after warming. The data showed that LBR had a similar trend after a shorter culture regimen in D6 blastocyst transfer cycles, regardless of post-thaw duration. Another strength was that PSM analysis was used to control for potential covariates, thereby ensuring that the outcomes were independent of different patient characteristics. Finally, our study was conducted over the past 1.5 years, which means that the laboratory conditions, culture media, and embryo selection criteria remained consistent. Given the undisputed high performance of vitrification compared with slow cooling protocols, there seems no need for a longer culture period to allow for survival and additional morphological evaluation of thawed blastocysts. Some may argue that it is a common procedure to culture D6 blastocysts for a shorter post-thaw interval; however, there is a lack of consensus in the literature, and some laboratories continue to culture thawed D6 blastocysts for many hours for logistic reasons. Thus, we believe that our study is clinically relevant and worthy of interest.
Despite our efforts, there were some shortcomings in this study. Although PSM analysis was applied to match confounders, this was a retrospective study, and selection bias could not be excluded. Our results revealed that transferring D6 blastocysts after a short culture interval after thawing yielded better FET outcomes; however, we could not determine which specific interval was the best. A prospective randomized controlled study with a larger sample size is required to confirm our results. In summary, the outcomes of short and long post-thaw culture duration did not differ between D5 blastocyst transfer cycles. Both embryo culture strategies can be applied, and the decision is based on each center's work schedule. However, for blastocysts vitrified on D6, a long culture period after thawing adversely affected clinical results. To optimize pregnancy outcomes, a short post-thaw culture protocol is recommended for patients with D6 blastocysts. In addition, our data indicated that all three short durations achieved successful and comparable pregnancy outcomes that could be feasibly applied by each laboratory.

Participants
Participants who underwent frozen blastocyst transfer at the Reproductive Center of Women's Hospital of Nanjing Medical University, Nanjing, China (Nanjing Maternity and Child Health Care Hospital) between September 2019 and March 2021 were recruited. Patients younger than 43 years of age with at least one D5 or D6 blastocyst using their own eggs were eligible. The exclusion criteria were as follows: (a) presence of uterine malformation; (b) cycles of oocyte donation, vitrified oocytes, or preimplantation genetic testing; (c) transfers of combined D5 and D6 blastocysts or D7 blastocysts; (d) maximal EMT < 6 mm; and (e) missing cycle or follow-up data.

Stimulation and culture protocols
The ovarian stimulation and embryo culture protocols used at our center have been published previously (Ji et al. 2020). All patients were treated using a flexible gonadotropin-releasing hormone antagonist protocol. Oocyte retrieval was scheduled 35-36 h after the injection of 10,000 IU human chorionic gonadotrophin (hCG, Livzon Pharmaceutical Group Inc., Zhuhai, China) to induce final oocyte maturation. After fresh transfer or freezing of two cleavage-stage embryos with good morphology, all supernumerary embryos were cultured until the blastocyst stage, regardless of their quality.
Vitrification and thawing procedures D5 or D6 blastocysts were scored according to the cavity expansion level, along with inner cell mass (ICM) and trophectoderm (TE) grading before freezing (Gardner and Schoolcraft 1999). Blastocysts were selected for vitrification if they were at stage 3 or above, with at least one grade B for either ICM or TE. The vitrification and thawing protocols followed the manufacturer's instructions (Kitazato BioPharma Co., Shizuoka, Japan) and have been previously described (Ji et al. 2021). Assisted hatching with laser treatment was conducted on thawed blastocysts, except for those with a degree of 5 or 6 blastocoel re-expansion. We evaluated the embryos 30 min before transfer. Blastocyst survival was defined as partial or complete blastocoel re-expansion and >50% intact cells. For blastocysts that ceased to divide, an additional warming was performed.
Before October 2020, all D5 and D6 blastocysts were thawed in the afternoon before ET, indicating an additional culture period of 18-20 h before transfer. Since this time, we have applied a new freeze-thaw protocol in FET cycles. Instead of warming the blastocysts overnight for a long period, embryos were thawed and transferred on the ET day after a culture period of 2-6 h. Participants were divided into two groups based on two different culture durations: the short culture group (a post-thaw culture duration of 2-6 h) and the long culture group (a post-thaw culture duration of 18-20 h). Good-quality blastocysts before vitrification were those with scores of 3BB or above (grade 3-6 AA/AB/BA/BB). Before transfer, reexpanded embryos with grades !3BB in the short culture group and !5BB in the long culture group were considered good-quality blastocysts.

Endometrial preparation
The endometrial preparation regimen was selected based on physician and patient preferences. Participants with regular menstruation were treated using a natural cycle (NC) protocol. Ovulation in the NC protocol was determined by monitoring follicular growth with transvaginal ultrasound and measuring the serum P levels. When the leading follicle was ! 18 mm and the P level was 1.5 ng/ml, 10,000 IU hCG was administered. Upon confirmation of ovulation by ultrasound, luteal phase support (LPS) was commenced, and patients were administered 10 mg of oral dydrogesterone (Abbott Biologicals B.V., Weesp, the Netherlands) three times daily. For the artificial cycles (ACs), patients were administered 4-6 mg of oral estrogen (estradiol valerate, Progynova, DELPHARM Lille SAS., Lille, France) once daily starting on the second or third day of the menstrual cycle for one week, which was then adjusted to 6-10 mg according to the EMT and serum estradiol (E 2 ) level. After adequate endometrial proliferation and a serum E 2 concentration ! 200 pg/ml, patients were administered 90 mg of vaginal P (Crinone 8% gel, Fleet Laboratories Ltd., Watford, United Kingdom) once daily and 10 mg of oral dydrogesterone three times daily. All embryo transfers were performed on day 6 of P exposure. If pregnancy was achieved, LPS was continued until gestational week 10-12.

Pregnancy outcome
Clinical pregnancy was confirmed by visualization of the gestational sac by ultrasound 4 weeks after FET. Abortion was termed as fetal loss before 28 weeks of gestation. Live birth was defined as the delivery of a live-born infant beyond gestational week 28. IR was calculated as the number of intrauterine gestational sacs per number of transferred blastocysts. The primary outcome was LBR. The secondary endpoints were IR, CPR, and AR.

Statistical analysis
All data were analyzed using the SPSS software (version 24.0; IBM, NY, USA). Continuous data with normal distribution were presented as mean ± standard deviation and compared by the independent samples t-test, whereas continuous data with non-normal distribution were presented as median with interquartile range and compared using the Mann-Whitney U test. Categorical data were presented as counts (percentage; %) and compared by Pearson's chi-square test or Fisher's exact test.
To ensure comparability between patients with a short culture period and those with a long culture period, we implemented a PSM analysis in D5 embryo transfer cycles. Patients were matched according to the propensity score with a 0.1 SD caliper and a ratio of 1:1. The following variables were included in the matching procedures: age, infertility type, infertility duration, infertility cause, body mass index, basal FSH level, AMH level, EMT, endometrial preparation protocol, number of transferred embryos, and number of good-quality embryos before transfer. Multivariable logistic regression analysis was performed to explore the effect of the post-thaw culture duration on live birth outcomes. Potential confounding factors showing clinical relevance or a p-value < 0.1 in univariate analysis were included in the regression models. Statistical significance was set at p < 0.05.

Ethical approval
This retrospective cohort study was conducted according to the Declaration of Helsinki and approved by the Ethics Committee of Nanjing Maternity and Child Health Care Hospital (NJFY-2020-KY-070). Informed patient consent was not required as the study was retrospective in nature and analyzed patient data anonymously.