Clinical outcome of different embryo transfer strategies after late rescue ICSI procedure: a 10-year total fertilisation failure cohort study

Data of the participants using r-ICSI were extracted from a large dataset based on the medical records of couples who underwent ART procedures between 2009 and 2019 at the Centre for Reproductive Medicine, Peking University Third Hospital. As one of the largest reproductive health centres in China, this centre has recorded information on ovarian stimulation (OS) cycles, followed by 149,054 fresh embryo transfer (fresh ET) and 79,874 FET cycles. Notably, we excluded participants who accepted the donor sperm or implementation of pre-implantation genetic testing (PGT) procedures.

Out of all ART processes, r-ICSI cycles were included, whilst cycles mixed with embryos acquired through other insemination methods such as IVF and/or ICSI were excluded. Finally, 2,270 r-ICSI cycles encountered unexpected TFF across 149,054 cycles of IVF and subsequently adopted a late r-ICSI procedure. Of these patients, 263 women did not have cleavage-stage embryos available for evaluation. The remaining patients were categorized into three groups based on their different embryo transfer (ET) strategies. Specifically, Group 1 consisted of 926 women who underwent fresh ET; Group 2 included 365 women who underwent a freeze-all ET strategy, and Group 3 contained 716 women who experienced blastulation failure (Fig. 1).

We considered the following variables when regarding OS cycles: Essential characteristics of participants [age, infertility years, body mass index (BMI), etc.], causes of infertility [female and male], the level of based serum reproductive hormone [follicle-stimulating hormone (FSH), oestradiol (E2), luteinising hormone (LH), and prolactin (PRL)], medication use, and other variables related to the treatment process.

This study has obtained approval from the Ethics Committee of the Peking University Third Hospital for data handling (No. IRB00006761-M2020007). All procedures conformed to the ethical standards of the responsible committee for human experimentation, and the Helsinki Declaration of 1964 and its later amendments. All patients included in this study gave informed consent.

OS was mainly performed using either a short antagonist protocol, or long agonist protocol coupled with recombinant FSH (Gonal-F alfa, Merck Serono, Germany), recombinant follitropin beta (Puregon, MSD, USA), urofollitropin (Livzon Pharmaceutical Group Inc., China), or menotrophins (Livzon Pharmaceutical Group Inc., China) administered on days 2 or 3 of the menstrual cycle. When the follicles reached 10–12 mm, a GnRH antagonist (Cetrotide, Merck Serono, Germany; Ganirelix, MSD, USA) was applied. Otherwise, GnRH agonists (Triptorelin, Ipsen Pharma Biotech, France) were administered in the last follicle or luteal phase in the ultra-long and long protocols, or one day before gonadotrophin application in the short protocol. When no less than two dominant follicles measuring > 18 mm had been testified by ultrasound, final maturation and ovulation were triggered with human chorionic gonadotropin (hCG, 250 µg) (Choriogonadotropin alfa, Merck Serono, Germany) administration. On other occasions, hCG (1000-2000IU of 250 ug) and GnRH agonist (0.2 mg) were administered in dual trigger cycles, with 0.2 mg GnRH agonist being used as a trigger, according to clinical conditions for high response cycles in the antagonist protocol. Oocyte retrieval was subsequently performed after 36–38 h.

In our Centre for Reproductive Medicine, ICSI is incorporated as a standard procedure under specific conditions including severe male factor infertility, prior fertilization failure. In absence of these, conventional IVF is generally utilized.

The semen analyses performed were in accordance with the World Health Organization (WHO) guidelines. Specifically, we used the criteria outlined in the 4th and 5th editions of the WHO Laboratory Manual for the Examination and Processing of Human Semen. Normozoospermic samples were selected by analysis of semen concentration, motility, and morphology. Oocytes were then examined for the existence of a second polar body after exposure to spermatozoa for 16–19 h. Those without an evident second polar body in all oocytes after IVF insemination, also referred to as TFF, were subjected to r-ICSI within one hour using 1-day semen sample, which was performed in accordance with the standard ICSI procedure [20] And subsequently checked for fertilisation the next day. The two-pronuclear (2PN) fertilisation rate was calculated as the number of oocytes with two pronuclei acquired using the r-ICSI method/number of all retrieved oocytes. One to three embryos were then transferred from the same fresh cycle, accompanied by luteal phase progesterone support (Progesterone Vaginal Gel, Merck Serono, Germany). After this, embryos were cryopreserved using vitrification technology for future FET, carried out in natural or hormone-artificial cycles according to routine clinical procedures, which have been detailed elsewhere [21]. In the case of fresh r-ICSI cycles, cleavage-stage embryos were transferred on Day 3 corresponding to the embryo’s development stage of Day 2. For blastocysts in fresh cycles, the transfer was performed on Day 6 reflecting the blastocyst’s development stage of Day 5. In FET cycles, the transfers were conducted conventionally on either Day 3 or Day 5, depending on the embryo’s developmental stage.

Crucial indicators for outcomes followed the Chinese Society of Reproductive Medicine (CSRM) consensus for consistency in clinical operations [22]. A positive hCG result was determined through blood β-hCG testing 12 to 14 days post-embryo transfer, with a threshold level established at greater than 30 mIU/ml to signify rising hCG. Women who tested positive received an ultrasound examination at least 28–30 days after transfer to detect the existence of a gestational sac, which was then confirmed as a clinical pregnancy. The embryo implantation rate (IR) was subsequently calculated as the number of gestational sacs/the number of embryos transferred. Miscarriage was defined as pregnancy loss before 28 weeks, and the miscarriage rate (MR) was also calculated for pregnant women. The primary objectives for the study were the delivery of a living foetus (or living foetuses) during the first cycle and all cycles after one OS procedure, which were separately associated with the live birth rate (LBR) and cumulative live birth rate (cLBR) [23]. The cLBR follow-up period ended in December 2019.

To test the robustness of this study, we performed several sensitivity analyses. Firstly, to minimise the potential confounding effects of maternal age, we performed stratified analysis by dividing the population into two subgroups based on maternal age (< 35 or ≥ 35 years) before testing the associations between ET strategies and IVF outcomes for the two respective subgroups. We also performed propensity score matching (PSM) [24] to generate matched pairs in a 1:1 ratio balanced with age.

In addition, among our participants, 716 cycles adopting the blastocyst culture strategy in the stimulation cycle suffered from embryo loss and failed. Frozen blastocyst transfer has been reported to result in a higher LBR [17] and has also been adopted as a routine practice in the centre. Therefore, we performed an intention-to-treat (ITT) analysis assuming that all 716 cycles would choose to perform the FET strategy if embryos were available, and then included these in the Full Analysis Set (FAS), from which conservative cLBR was calculated [25].

Baseline characteristics, including demographic variables, hormone parameters, and obstetric outcomes, were compared among the groups. In this study, considering that most of the continuous variables were not normally distributed, these were represented as medians and interquartile ranges [IQRs] and subsequently compared using the Mann–Whitney U test. Categorical variables were described as numbers or percentages and then compared using Pearson’s chi-square test. Binary logistic regression was also used to adjust for potential maternal confounding factors, which appeared to be statistically significant in the univariate analysis or were deemed critical. Additionally, it should be noted that the gonadotropin dose was log-transformed before the adjustment. All statistical analyses were conducted using IBM SPSS 26.0 (IBM Corp. Armonk, NY), with two-tailed tests being performed, and statistical significance set at p < 0.05.

After routine IVF procedures, 7.4% (6006/81,221) of the individuals experienced TFF. A total of 2,270 OS cycles adopting the r-ICSI strategy were included among the women aged between 20 and 50 years who received treatment from January 2009 to December 2019. Of all individuals, 40.8% (926/2270, Group 1) of patients opted for the fresh-ET strategy, whilst 16.1% (365/2270, Group 2) selected the freeze-all ET strategy. Notably, for up to 43.1% of women (979/2270), researchers failed to obtain embryos eligible for transfer after oocyte retrieval, among which 73.4% of women (716/979, Group 3) suffered from transfer cancellation during blastocyst culturing (Fig. 1). Furthermore, the proportion of women without eligible embryos among all r-ICSI patients attributable to blastocyst culturing was 49.4%.

The baseline characteristics of the women during OS cycles are displayed in Table 1. Comparisons between Group 1 and Group 2 revealed significant differences, including maternal age [33.00 (30.00–37.00) vs. 31.00 (29.00–35.00), P < 0.001], infertile years [5.00 (3.00–8.00) vs. 3.00 (2.00–6.00), P < 0.001], levels of FSH [7.10 (5.71–8.69) vs. 6.61 (5.25–7.95) mIU/mL, P < 0.001], and dose of gonadotropins [2850.00 (1950.00–4125.00) vs. 2325.00 (1800.00–3150.00) IU, P < 0.001]. Moreover, Group 1 had a higher tendency towards secondary infertility. Specifically, a higher proportion of women in Group 1 had experienced one (20.73 vs. 19.18%), two (12.63 vs.6.03%), or even more than three pregnancies (9.72 vs. 3.56%) compared with Group 2. Women undergoing fresh ET were exhibited lower oocyte retrieval after stimulation [10.00 (6.00–15.00) vs. 13.00 (9.00–19.00), P < 0.001]; thus, fewer oocytes were retrieved for r-ICSI use [8.00 (5.00–12.00) vs. 10.00 (7.00–14.00), P < 0.001]. Additional comparisons were performed between groups 2 and 3 (Table S4). The latter turned out to be older [33.00 (30.00–37.00) vs. 31.00 (29.00–34.00), P < 0.001], with lower levels of LH [3.07 (1.81–4.51) vs. 3.52 (2.08–5.05), P = 0.008], received higher doses of gonadotropin [2775.00 (1881.25–3750.00) vs. 2250.00 (1650.00–3150.00), P < 0.001], and had lower oocyte retrieval after stimulation [9.00 (7.00–13.00) vs. 13.00 (9.00–18.00), P < 0.001]. The rates of 2PN fertilization following r-ICSI were recorded as 0.50 (0.25–0.71) in Group 1 and 0.54 (0.33–0.72) in Group 2, respectively. Moreover, the number of cleavage-stage embryos showed significant differences between the two groups [6.00 (3.75–9.00) vs. 8.00 (6.00–11.00), P < 0.001].

Compared with women who underwent fresh ET in the stimulation cycle, a higher proportion of individuals within the freeze-all-embryo group exhibited a positive hCG rate (54.79 vs. 10.91%) and positive ultrasound examination results (47.39 vs. 6.59%), higher IR (42.49 vs. 3.86%), and similar MR (19.65 vs. 27.87%); thus, achieving considerably better clinical outcomes in both the LBR of the first ET cycle (37.53 vs. 4.64%) and cLBR (52.60 vs. 8.21%) (Table 1). Notably, women in Group 2 showed a remarkable improvement in LBR compared to others with fresh ET, whilst having more embryos available for extra transfer cycles (Fig. 2). Furthermore, no malformations were observed among the newborns.

Further inspection of the data showed that the fresh ET group had a more advanced maternal age (38.00 vs. 37.00) in the 35 years and up cohort, even after stratification of analysis according to maternal age. Women under 35 years of age had higher clinical pregnancy rates (8.14 vs. 4.29% in the fresh ET group and 48.96 vs. 41.55% in the freeze-all group), LBR (5.97 vs. 2.68% in the fresh ET group and 38.89 vs. 32.47% in the freeze-all group), and cLBR (10.13 vs. 5.36% in the fresh ET group and 53.82 vs. 48.05% in the freeze-all group), along with fewer infertile years, lower FSH levels, lower gonadotropin dose, and fewer oocytes used for r-ICSI. Although the freeze-all group stood out for both the younger and older cohorts, age stratification seemingly failed to mitigate the effects of confounding variables (Table S1). Among the 450 women who were 35 years or older, the eldest participant was 46 years old. A further stratification between women aged ≥ 35 and < 40 years and those aged ≥ 40 and ≤ 46 years revealed consistent trends in LBR and cLBR favouring the freeze-all ET group (Table S5).

Another approach involved producing an equal number of cohorts from the fresh ET and freeze-all groups by propensity score matching. While the fresh ET group was still inclined to have more infertile years [5.00 (3.00–7.00) vs. 4.00 (2.00–6.00), P < 0.001] and higher FSH levels [7.00 (5.58–8.34) vs. 6.09 (4.63–7.71), P < 0.001], there were no marked differences in oocytes remained for r-ICSI [10.00 (6.00–14.00) vs. 10.00 (7.00–14.00), P = 0.149]. Despite this, women who received fresh ET and freeze-all ET demonstrated significant differences in positive hCG rate (31.36 vs. 54.94%), clinical pregnancy (16.76 vs. 47.53%), IR (9.93 vs. 42.59%), LBR (11.81 vs. 37.64%), and cLBR (11.81 vs. 43.96%) (P < 0.001) (Table S2).

To confirm the efficacy of freeze-all ET, the most conserved estimates obtained using the ITT method were applied. As shown in Table S3, gaps between the two groups in the conservative cumulative clinical pregnancy (8.21 vs. 17.76%) and cLBR (6.16 vs. 14.62%) shrunk, although remained significant (P < 0.001).

After adjusting for covariates, including age, BMI, years of infertility, gravity, LH and FSH levels, Gn dose, number of oocytes, and 2PN fertilisation rate, using binary logistic regression analysis, the r-ICSI freeze-all ET method showed a higher likelihood of better obstetric outcomes, compared with those of the fresh ET method for LBR [OR:11.77, 95% CI (8.42–16.45)], cLBR [OR:11.29, 95% CI (7.84–16.27)] and conservative cLBR [OR:2.55, 95% CI (1.83–3.55)] (Table 2).