2.1 Baseline characteristics comparison between the groups

In the comparison of the cryopreservation (Cryo) group and the recryopreservation (Recryo) group, there were 360 cycles in each group after propensity score matching (Figure 1), and the baseline characteristics between the two groups were similar (Table 1), in terms of age at oocyte retrieval, endometrial preparation protocol, endometrium thickness, blastocyst developmental stage, blastocyst grade, blastocyst expansion level, inner cellular mass (ICM) score, and trophectoderm (TE) score. The density of propensity scores of the two groups almost overlapped. (Figure S1A,B). Similarly, the distribution of standardized differences after matching was more concentrated (Figure S1C,D). Moreover, univariate standardized differences after matching basically concentrated around 0 (Figure S1E). In addition, individual differences after matching decreased significantly (Figure S1F). All the above results indicated the well-balanced distribution between the comparing cohorts after matching.

2.2 Recryo procedure in blastocysts results in impaired blastocyst developmental potential

Clinical outcomes of the FET cycles between the two groups after matching were presented in Table 2. Compared to once cryo, recryo procedure in blastocysts resulted in impaired blastocyst developmental potential, including decreased implantation rate (40.0% vs. 47.8%, p = 0.035), reduced biochemical pregnancy rate (45.3% vs. 58.6%, p < 0.001), declined clinical pregnancy rate (40.0% vs. 47.8%, p = 0.035), higher early miscarriage rate (31.3% vs. 20.3%, p = 0.026), and lower live birth rate (24.7% vs. 34.4%, p = 0.004). However, there were no significant differences regarding the neonatal outcomes of the transferred blastocysts after matching (Table 3). The gender, birthweight, pregnancy complications, and neonatal complications were similar between the groups, despite a higher proportion of cesarean sections in the Recryo group (93.3% vs. 83.1%, p = 0.027).

2.3 Recryo procedure impairs TE function of blastocysts

As Figure 2A shown, blastocysts experiencing recryo procedure had impaired developmental potential, including a decreased implantation rate. Since human chorionic gonadotropin (HCG) level 12 days after blastocyst transfer is associated with implantation competence of transferred blastocysts, the HCG levels between the two groups were compared. There were 172 blastocysts implanted in the Cryo group and 144 in the Recryo group. It was shown that the median HCG level was lower in the Recryo group (657 [385–1000] vs. 798 [469–1000], p = 0.023, Figure 2B), and the distribution of HCG level in the Cryo group was more concentrated around the detected upper limit (Figure 2C). To explore the association between HCG level and blastocyst grade, correlation analyses were performed. In Figure 2D, it was found that HCG level was positively correlated with TE quality (p = 0.002) and it increased with a higher blastocyst grade (p < 0.001). While interestingly, after recryo procedure, the correlation between HCG level and blastocyst grade showed no significance, so did the association with TE score. These results prompted that recryo procedure impairs TE function of blastocysts and HCG secretion.

2.4 Recryo procedure alters the transcriptome of human blastocysts

To explore the effect of recryo procedure on the gene expression characteristics of human blastocysts, single-cell RNA sequencing was performed in recryo and cryo blastocysts. There were 291 differentially expressed genes (DEGs) with 196 upregulated and 95 downregulated (Figure 3A,B). Gene ontology (GO) analyses showed that the DEGs mainly participated in the process of cell adhesion, embryo development, and apoptosis (Figure 3C). To be more specific, the upregulated DEGs associated with embryo implantation were involved in cell–cell communication/cell junction organization (SDK1, SRC, FLNC, CLDN6, CDH5, PLEC, and CLDN4) and degradation of the extracellular matrix (COL1A1, CTSD, COL4A3, ADAMTS1, FURIN, COL12A1, and BMP1; Figure 3D). Moreover, the downregulated DEGs were related to Integrin cell surface interactions (FN1 and FGG; Figure 3E). Several abovementioned genes exhibiting significantly different expression, including FURIN, FLNC, COL1A1, PLEC, BMP1, CLDN6, FN1, and FGG, were further confirmed by quantitative real-time polymerase chain reaction (RT-PCR), and the mRNA expression was in accordance with the results of RNA sequencing, confirming the accuracy of the RNA sequencing (Figure 3F). In addition, the localization of adhesion-related E-cadherin was detected (Figure 3G). E-cadherin was evenly distributed on the cell membrane and arranged neatly, clearly, and continuously in fresh blastocysts in the control group. In the Cryo group, the arrangement of E-cadherin is slightly disordered, unclear, and partially discontinuous. In the Recryo group, the E-cadherin arrangement was chaotic and disorganized, with no distribution on the surface of some cells.

2.5 Recryo procedure activates endoplasmic reticulum stress pathway and induces apoptosis

GSEA analysis showed that the regulation of response to endoplasmic reticulum (ER) stress was activated after recryo (Figure 4A). In Figure 4B, altered genes in ER stress were visualized by heatmap. The expression of several important proteins involved in ER stress, including GRP78, XBP1s, and CHOP, was detected by immunofluorescence staining (Figure 4C–E). After the recryo procedure, the expression of GRP78 increased, which activated the downstream protein expression of XBP1s and CHOP. In addition, it was found that there was a positive regulation of the apoptotic signaling pathway (Figure 4F). Compared with blastocysts in the Cryo group, the apoptotic cells in the Recryo group were significantly increased by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assay (Figure 4G). Moreover, the expression of both Caspase-3 (Figure 4H) and cleaved Caspase-3 (Figure 4I) was upregulated after recryo, indicating the activation of apoptosis-associated caspase cascade reaction. In addition, the number of fragmented nuclei, which were indicated by white arrows, was significantly increased in blastocysts from the Recryo group.

4.1 Study design and ethical approval

In this retrospective study, we enrolled all FET cycles of single blastocyst transfer between January 2014 and December 2022 in the Reproductive Medicine Center of Tongji Hospital. According to the number of times blastocysts experienced cryopreservation, the FET cycles were divided into the Cryo group and the Recryo group. In the Cryo group, embryos were vitrified in the blastocyst stage, then cultured and transferred. In the Recryo group, the vitrification and revitrification procedures were both performed in the blastocyst stage. Couples who underwent PGT were excluded. The indications of recryo procedure and recryo blastocysts consumption were well described as previously reported.¹¹ The baseline characteristics, clinical outcomes, and neonatal outcomes were compared between the Cryo and Recryo groups.

Furthermore, in order to elucidate the molecular mechanisms of the effect of recryo on embryo potential, human embryos were collected and detected. All the embryos investigated in this research were donated by the patients with written informed consent. Fresh blastocysts were from discarded polypronuclei originated embryos. Cryo blastocysts were donated from successful delivery patients, and recryo blastocysts were available from warmed cryo blastocysts experiencing repeated vitrification and warm. This study was approved by the Ethical Committee of Tongji Hospital (TJ-IRB20211280).

4.2 Controlled ovarian stimulation and embryo culture

Controlled ovarian stimulation was performed based on standard gonadotropin-releasing hormone agonist or antagonist protocols.³² Oocytes were retrieved under the guidance of transvaginal ultrasound 36–38 h after HCG administration. The retrieved oocytes were fertilized in either conventional IVF or intracytoplasmic sperm injection (ICSI) when appropriate. The presence of 2 pronuclei (2PN) indicated successful fertilization after insemination. The embryos were cultured in G1-plus medium (Vitrolife) until Day 3 and then cultured in G2-plus medium (Vitrolife) to the blastocyst stage at 37°C with 5% O₂ and 6% CO₂. The blastocyst grade was based on the Gardner scoring system, and the expansion level, ICM score, and TE score were recorded. The morphology assessment was conducted by two experienced embryologists independently, with a third consulted in cases of inconsistency.

4.3 Vitrification and warming procedures

A commercial vitrification kit (Kitazato) was used for blastocyst cryopreservation as previously described.³³ Artificial shrinkage using a laser was performed before vitrification. Blastocysts were transferred into equilibration solution (ES) for 5–8 min and then infiltrated into vitrification solution (VS) for 40–60 s. The blastocysts were immediately loaded on the tip of the Cryotop (Kitazato) and submerged in liquid nitrogen. For blastocyst warming, the Cryotop was immersed in preheated thawing solution (TS) (Kitazato) for 1 min. The blastocysts were then transferred to dilution solution (DS) for 3 min, 5 min in washing solution 1 (WS1) (Kitazato), and another 5 min in washing solution 2 (WS2). After 2 h culture, blastocyst viability was assessed and survival blastocysts were transferred.

4.4 Laboratory and clinical outcomes

Blastocyst transfer was conducted after natural cycle or hormone replacement treatment in our IVF center. Twelve days later, serum HCG was measured using the direct chemiluminescence ADVIA Centaur ThCG kit (Cat:00643953, Siemens). A positive result of HCG measurement was regarded as biochemical pregnancy. The presence of intrauterine gestational sac with active fetal heart using ultrasound 5 weeks after FET was defined as clinical pregnancy. Pregnancy loss within 20 weeks was regarded as early miscarriage. Birth weight lower than 2500 g was defined as low birthweight, and those above 4000 g were defined as macrosomia. Pregnancy complications included several diseases, including gestational diabetes mellitus, hypertensive disorders complicating pregnancy, placenta previa, and placenta abruption. The details of the computing formula of the laboratory and clinical outcomes can be found in previous studies.³⁴

4.5 RNA sequencing and bioinformatics analysis

To investigate the effect of recryo procedure on blastocyst transcriptome, RNA sequencing was performed on cryo and recryo human blastocysts. To eliminate the effect of different sample sources on RNA transcriptome, three vitrified blastocysts (Cryo group) and three revitrified blastocysts (Recryo group) from the same couple were collected for RNA sequencing. The raw data have been deposited in the Genome Sequence Archive (GSA-Human: HRA007913).^{35, 36} The warming blastocysts were cultured in G2-plus medium for another day to partial or complete hatching (expansion level of 5 or 6) and then were collected for RNA sequencing. The commercial Smart-Seq2 based on an Illumina NovaSeq 6000 platform was performed as previously reported by Annoroad Gene Technology. The DEGs between the two groups were analyzed using DESeq2 in R. Genes with adjusted p value <0.05 and absolute log2 (fold changes) >1.5 were considered statistically significant. DEGs expression was visualized using volcano plot and heatmap by the “ggplot2” package. GO analysis and enriched pathway analysis for DEGs were conducted by the “clusterProliler” package. In addition, gene set enrichment analysis (GSEA) was also performed using the “clusterProliler” package.

4.6 Quantitative real-time polymerase chain reaction

RT-PCR was performed to validate the results of RNA sequencing. Cryo and recryo blastocysts were warmed and cultured for another day and then four to five blastocysts were collected in one tube. Total RNA extraction and amplification were performed using Single Cell Sequence Specific Amplification Kit (Cat: P621-01, Vazyme). Quantitative RT-PCR was performed using ChamQ Universal SYBR qPCR Master Mix (Cat: Q711-02, Vazyme) on the LightCycler 480 system (Roche). The relative mRNA expression normalized to GAPDH expression. The primer sequences are listed in Table S1.

4.7 Immunofluorescence and TUNEL assay

The immunofluorescence staining procedure was previously described and performed on fresh, cryo, and recryo blastocysts.^{37, 38} Fully expanded blastocysts after warming were fixed in 4% paraformaldehyde. Anti-E-cadherin antibody (Cat: A20798) was purchased from Abclonal, and anti-Caspase-3 (Cat: 19677-1-AP) was from Proteintech. Anti-GRP78 (Cat: C50B12), anti-XBP1s (Cat: 83418S), and anti-CHOP (Cat: L63F7) antibodies were from Cell Signaling Technology. Primary antibodies were used at 1:100 dilution unless otherwise indicated, and Cy3 conjugated secondary antibodies (Cat: GB21301 for anti-mouse and GB21303 for anti-rabbit, Servicebio) were used at 1:100 dilution. The nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI, Cat: G1012, Servicebio). Caspase-3 activity detection kit for live cell (Cat: C1073M, Beyotime) was utilized for visualization of cleaved Caspase-3. The TUNEL (Cat:C1089, Beyotime) assay was performed for apoptosis detection according to the manufacturer's instructions. Images were captured using a confocal microscope (LSM900, Leica).

4.8 Statistical analysis

Statistical analysis in the current study was performed using SPSS Statistics software (v26.0, IBM). Continuous variables are shown as medians (first quartile, third quartile), the comparison of which was achieved using Mann–Whitney U rank-sum test. The categorical variables are shown as % (n/N), which were compared with chi-squared test or Fisher's exact test as appropriate. A p value less than 0.05 (two-tailed) was considered of statistical significance.

For the convenience of statistics, the “1,” “2,” and “3” scores of ICM or TE corresponded to A, B, and C in the Gardner scoring system. The transferrable blastocysts were divided into three groups based on the ICM and TE scores: good (AA, AB, and BA), average (BB), poor (AC and BC), and others (blastocysts with expansion levels of 1 or 2). Considering that several blastocysts were with expansion levels of 1 or 2, “4” score of ICM or TE was assigned to these blastocysts.

When comparing the two groups, to eliminate the imbalance of the sample size between the Cryo group and the Recryo group, propensity score matching was performed. The demographic characteristics were matched in a 1:1 ratio, including age at oocyte retrieval (year), endometrial preparation protocol (hormone replacement treatment and nature cycle treatment), endometrium thickness on the day of progesterone administration (mm), blastocyst developmental stage (D5, D6, and D7), blastocyst grade (good, average, poor, and others), blastocyst expansion level (1, 2, 3, 4, 5, and 6), ICM score (1, 2, 3, and 4), and TE score (1, 2, 3, and 4). Nearest neighbor matching strategy in random order without replacement (caliper = 0.1) was performed.

Owing to the detection limitation of the kit, the exact HCG values that were higher than 1000 mIU/mL will not be reported, and these values were set as 1000 when analyzing. The distribution of HCG level at 12 days after FET between the groups was visualized as violin plot and density plot using R. Correlation analysis of HCG level and blastocyst grade was performed using ggplot2 in R.