A retrospective analysis was performed on patients who underwent frozen-thawed embryo transfer (FET) cycle in the Xiamen University Affiliated Chenggong Hospital in the period from January 2018 to December 2022, with 11627 cases in total. Institutional Review Board approval for this retrospective study was obtained from the Ethics Committee of the Xiamen University affiliated Chenggong Hospital. Informed consent was not necessary, because the research was based on nonidentifiable records as approved by the Ethics Committee.

The exclusion criteria included cycles canceled for any reason (n=32), cycles with missing data concerning the cycle characteristics (n=613), and cycles with twice frozen-thawed embryos (n=50). The patients receiving other add-ons or adjuvants during the FET cycles, such as endometrial receptivity test (ERT) or gynecological immunological treatment were also excluded. Patients with luteinized unruptured follicle (LUF) syndrome during the cycles were excluded due to possible disruption of luteal phase endocrine dynamics in these patients. Finally, rare and specific cases, such as carcinoma or vaginal bleeding on the day of transfer were excluded. The inclusion/exclusion criteria were detailed in a flowchart ( Supplementary Figure S1 ).

All patients received an agonist or antagonist protocol with the use of FSH or HMG for ovarian stimulation as previously described (9). Oocytes were recovered 34–36 hours after administration under transvaginal ultrasound guidance and were inseminated using either conventional IVF or intracytoplasmic sperm injection(ICSI). Zygotes and embryos were cultured in traditional incubators (C200, Labotect, G¨ottingen, Germany) at 37 °C, 6%CO2, and, 5%O2 with Cook IVF media (COOK MEDICAL, Bloomington, IN) and oil overlay (Vitrolife, G¨oteborg, Sweden).

The morphological criteria of embryo scoring at the cleavage stage and blastocyst stage adhered to the Istanbul consensus (10) and the Gardner scoring criteria (11). Good quality embryos were defined as Grade 1 cleavage stage embryos according to the Istanbul consensus or blastocysts with a ≧BB Gardener score.

For all cycles, a vitrification protocol, employing 15% dimethylsulphoxide, 15% ethylene glycol, and 0.6 M sucrose as cryoprotectants, was used. Embryos obtained from the patient’s fresh egg retrieval cycle are vitrified and frozen as day 3 cleavage stage embryos or day 5 or 6 blastocysts.

Four major types of endometrial protocols were used for endometrial preparation, including natural cycles (NC), ovulation induction cycles (OI), hormone replacement cycles (HRT), and down-regulated hormone replacement cycles (GnRHa). The NC or OI cycle is based on vaginal ultrasound monitoring, with the day of ovulation as D0, D3 as the day of cleavage embryo thawing, and D5 as the day of blastocyst thawing and transferring; the hormone replacement cycle starts with estrogenic medication for 14–20 days from the second day of menstruation or 28 days after down-regulation of the hormone replacement cycle, and the day of addition of progesterone as D0, D4 as the day of cleavage embryo thawing, and D6 as the day of blastocyst thawing and transferring. The patients who had an unexpected inadequate follicle growth in NC protocol and chose to receive additional stimulation were assigned as “other protocol”.

All patients were given luteal support therapy after embryo transfer. In NC cycles, dexamethasone tablets 20 mg twice daily were given. In the OI cycle, 40 mg/day of progesterone was given intramuscularly or 90 mg/day of xylenol vaginally and 20 mg of oral dextroprogesterone tablets twice a day following ovulation. HRT or GnRHa cycles were supported with estrogen continuation and addition of intramuscular progesterone 40mg/day or vaginally administered chenodexone 90mg/day and orally administered dextroprogesterone tablets 20mg twice daily after ovulation cycle with luteal support.

Serum β-HCG test was performed at D14 after embryo transfer, if β-HCG>5 U/L, it was defined as biochemical pregnancy, and if the gestational sac was seen by ultrasound at D28, it was defined as clinical pregnancy.

For patients receiving atosiban, a total dose of 37.5mg of atosiban acetate (atosiban acetate, 37.5 mg/5 ml, Nanjing Haina Pharmaceutical) was administrated as an intravenous infusion which began half an hour before embryo transfer and lasted for 3 hours. For patients receiving GH, GH (recombinant human GH for injection, 4.0 IU/1.33 mg/1.0 ml/vial, Changchun Jinsai Pharmaceutical Co., Ltd.) was administered 4 IU/day intramuscularly during the preparation of the endometrium before embryo transfer.

Because the supposed mechanism of atosiban is facilitating embryo implantation, we evaluated clinical pregnancy as our primary outcome of interest. The interactions were estimated by comparing the odds ratio (OR) for pregnancy comparing patients with or without atosiban adjuvant in cohorts stratified according to the presence of GH use. More formally, an interaction term (atosiban × GH) was introduced to a multivariate model to calculate the ratio of OR (ORR).

To further evaluate the potential impact of the indications associated with adjuvant usage, we also screen for the potential interactions between atosiban and a series of indications, including age, previous ET attempts, endometriosis, and endometrial thickness in the multivariate models.

To minimize the confounding, the analyses were carried out in both the overall cohort and a propensity score (PS) matched cohort. In the PS-matched cohort, patients were matched according to atosiban use, and GH treatment was the primary covariate for matching. The matching was carried out in 1:1 ratio with the allowance of discarded cases.

Other confounders and covariates used for the matching and multivariate adjustment were selected based on our experience and previous knowledge with the assistance of a direct acyclic graph (DAG). The DAG was created by dagitty software (https://dagitty.net/dags.html) and shown as a supplementary figure ( Supplementary Figure S2 ). The covariates include patient characteristics (age, BMI, previous ART attempts, duration of infertility, endometriosis, PCOS, female endocrine profile, and male same parameters), ovarian stimulation (protocol and starting dosage), laboratory procedures (insemination protocols, blastocyst culture, and cryopreservation), and embryo availability (the number, stage, and quality of embryos transferred). A MatchIt package in R software was used for the PS matching (12). The cobalt package (13) was used to test the balance. Standard differences (D) were calculated to evaluate the balance of the distribution of the baseline characteristics between the groups before and after PS matching. D < 0.1 was used as the threshold to indicate a negligible difference in the mean or prevalence of a covariate (14). The balance of covariates was also examined by the distribution of propensity score (distance) between matched groups ( Supplementary Figure S3 ).

Subgroup analyses for patients with a thin endometrium (endometrial thickness<8mm) and patients with previous ET failure (ET order≧3) were also carried out to evaluate the effects of adjuvants and interactions in patients with specific indications.

For descriptive analyses, continuous variables were analyzed using the Wilcoxon test, and categorical variables were analyzed using the chi-square test or Fisher’s exact test, P < 0.05 was considered to be significant. All analyses were performed by using R statistic software 4.12 (15).

Our study suggested a possible combined effect of atosiban and GH as adjuvants which might benefit the clinical pregnancy. On the other hand, however, atosiban or GH showed no significant effect on pregnancy in the cohort. The finding may support the concept of “combination therapy” (16) to enhance the pregnancy outcome in FET cycles.

To the best of our knowledge, the study is the first to demonstrate the interaction between atosiban and GH in FET cycles. The study design is also fortified by the sample size and a control cohort matched for indications, such as previous ET attempts and endometriosis diagnosis. Nevertheless, the study is still limited by its retrospective nature, which may include unknown or unmeasured confounding or biases. In addition, the euploidy of the embryos transferred was not known in the study, which may significantly affect the measure of the effect.

The measurement of uterine contraction is also absent in the study. It led to an obscure indication and thus might affect the evaluation of the effect of atosiban. However, among the few studies that reported uterine contraction in cycles with atosiban treatment, the uterine peristalsis frequency may not always be relevant to the outcome (17).

In the present study, we used a total dose of 37.5mg atosiban in the study. According to the published data, there have been various types of doses and modes of administration (bolus versus infusion) of atosiban as an adjuvant in ART and they may contribute to the heterogeneity of the studies (6). Nevertheless, the reviews of previous data (6, 18) also suggested that the association between atosiban and ET outcomes is not necessarily linked to the dose of atosiban administration and the effect of different doses/modes of atosiban administration remained less clear.

Although the use of adjuvants in ART appears to be common, the combination of various adjuvants is not always promising. For instance, Motteram et al. reported a combined adjuvant strategy of aspirin, doxycycline, and prednisolone, suggesting no benefit in fresh IVF cycles, and possible harm when used in frozen cycles (19). Low molecular weight heparin, aspirin, and prednisolone, another combination of adjuvants to enhance the outcomes in patients with previous implantation failure is also reported to have a neutral effect (20). Since the data concerning the effect of combined adjuvants to support embryo implantation remained limited, and a neutral even suboptimal outcome would be expected, a careful consultation may be needed. Our data may contribute to future consultation on the use of combined adjuvants.

A possible interpretation of the interactions between atosiban and GH is based on the cross-talk of the signals they mediate. Atosiban is an inhibitor of oxytocin used to reduce the frequency and amplitude of uterine contractions during ET. However, oxytocin has also been shown to have physiologic, metabolic, and anabolic effects targeting the GH-IGF1 axis apart from its role in smooth muscle contractions (21). The oxytocin receptor (OXTR) also interacts with the GH secretagogue (ghrelin) receptor, which consequently was able to attenuate OXTR-mediated Gaq signaling (22). Intravenous oxytocin administration is reported to reduce the circulating levels of ghrelin (23). Growth hormone, in turn, also stimulates the release of oxytocin in a dose-dependent manner (24). When GH is administrated during endometrial preparation with the hope of improving endometrial receptivity, the enhanced GH signaling might also enhance the oxytocin release, which might increase the risks of unwanted uterine contraction. Additional atosiban treatment, therefore might be helpful in these patients.

We did not find a significant association between atosiban treatment alone and pregnancy outcomes in either the overall population or in patients without GH treatment. It appeared to contract with the recent Cochrane review conclusion that intravenous atosiban may increase clinical pregnancy rate (RR 1.50, 95% CI 1.18 to 1.89) (18). However, the largest trial in the review (25) which enrolled 800 individuals suggested a neutral effect of atosiban on patients without specific indications. On the other hand, the smaller RCTs showed significant effects of atosiban may have aggregated indications, such as endometriosis (26) or difficulty in transfer (27). In addition, several more recent studies based on patients with RIF found no significant effect of atosiban on pregnancy rates (20, 28, 29). The evidence as a whole may be against the routine use of atosiban in unselected patients.

Since our atosiban patients were matched for both GH treatment and endometrial thickness, another interpretation for the significant effect of atosiban in GH treated cohort may be the aggregation of patients that have similar indications. It is reported that GH treatment may benefit patients with a thin endometrial thickness (30) and therefore clinicians may tend to use GH in patients with a thin endometrium. Our data also showed a lower mean endometrial thickness in GH-treated patients. On the other hand, however, a thin endometrial thickness is thought to be linked with abnormal uterine peristalsis (31). Atosiban might be more effective in those patients due to a potential bias of selection. Nevertheless, we screened the potential interactions between the atosiban and patients’ indications, including age, BMI, etiologies, and endometrial thickness without a significant finding ( Figure 1 ). Therefore, it is yet to be concluded that endometrial thickness is a potential marker for atosiban use.

A summary of previous studies (30) suggested that GH as an adjuvant could improve ET outcomes by enhancing endometrial receptivity. However, the majority of evidence included in that review was focused on fresh cycles and biased toward poor responders. The potential effects of GH on oocytes and embryos could not be excluded. In FET cycles, where the quantity and quality of the embryos had been determined previously, a few reports also supported the role of GH in improving the endometrial thickness, blood perfusion, and receptivity markers (32, 33). However, the data may not justify the routine use of GH in FET cycles, as they were small and few in number, limited in patients with a thin endometrium. Our data demonstrated in a larger cohort with matched characteristics that GH may not significantly improve pregnancy rates in unselected patients. In body systems, GH is a pleiotropic hormone affecting multiple physiological systems, interacting with numerous signaling pathways including oxytocin. Moreover, evidence also suggested its potential pathological role (34). The complexity of the physiological role of the hormone suggests that the administration of GH during endometrial preparation simultaneously affects multiple tissues and potentially interacts with various internal or external factors. It may warrant further studies regarding the interactions between GH and physiological/therapeutic factors.