Reassessing the feasibility of the zygote score for predicting embryo viability in IVF/ICSI using the GnRH antagonist protocol compared to the long protocol

Background Many factors from the oocyte/sperm or the process of fertilization may affect the zygote formation. The zygote score (Z-score) describes the quality of a human zygote based on its pronuclear morphology, nucleolar precursor bodies, and alignment of polar bodies, and it can be used in the selection process at the zygote stage for embryo transfer or cryopreservation. Objective The aim of this retrospective cohort study was to investigate the relationship between different controlled ovarian stimulation (COS) protocols and the zygote score (Z-score) and to assess the feasibility of the Z-score for predicting embryo survival in the GnRH-antagonist (GnRH-ant) protocol. Methods It is a retrospective, single-center cohort study. A total of 3,826 zygotes with normal fertilization were analyzed from 744 in vitro fertilization /intra-cytoplasmic sperm injection (IVF/ICSI) cycles (long protocol n = 392; GnRH-ant n = 352) between Jan 2010 and April 2014 in the IVF unit of Chang-Gung Memorial Hospital Kaohsiung Medical Center. Results The Z-score distribution differed significantly between these two protocols. The overall Z-score was poorer for zygotes from GnRH-ant cycles (p<0.05). Univariate and multivariate analyses indicated the type of COS protocol is one of the main determinants of Z-score grading. Our study found good-quality day 3 embryo/blastocyst formation and the cumulative embryo survival rate were correlated with the Z-score but not the COS protocol. With the GnRH-ant protocol, the number of Z1 in the transferred cohort embryos was significantly correlated with the clinical pregnancy rate (r = 0.976; p = 0.024) and live birth rate (r = 0.971; p = 0.029). This correlation was not seen with the long protocol. Conclusions The Z-score distribution for the GnRH antagonist cycles was poorer than that of the long protocol, but the Z-score system is a valuable parameter for predicting embryo viability in the GnRH-ant protocol, providing a strong correlation with the clinical pregnancy rate and live birth rate.


Objective
The aim of this retrospective cohort study was to investigate the relationship between different controlled ovarian stimulation (COS) protocols and the zygote score (Z-score) and to assess the feasibility of the Z-score for predicting embryo survival in the GnRH-antagonist (GnRH-ant) protocol.

Methods
It is a retrospective, single-center cohort study. A total of 3,826 zygotes with normal fertilization were analyzed from 744 in vitro fertilization /intra-cytoplasmic sperm injection (IVF/ICSI) cycles (long protocol n = 392; GnRH-ant n = 352) between Jan 2010 and April 2014 in the IVF unit of Chang-Gung Memorial Hospital Kaohsiung Medical Center.

Results
The Z-score distribution differed significantly between these two protocols. The overall Zscore was poorer for zygotes from GnRH-ant cycles (p<0.05). Univariate and multivariate analyses indicated the type of COS protocol is one of the main determinants of Z-score grading. Our study found good-quality day 3 embryo/blastocyst formation and the cumulative embryo survival rate were correlated with the Z-score but not the COS protocol. With the GnRH-ant protocol, the number of Z1 in the transferred cohort embryos was significantly PLOS

Introduction
Institutional Review Board waived the need for consent. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Controlled ovarian stimulation, oocyte retrieval, oocyte preparation, embryo culture, assessment of fertilization, zygote score, and embryo grading COS, oocyte retrieval, embryo culture, and embryo transfer were performed according to our previously described protocols [12,19]. The long protocol and GnRH-ant protocol for COS were individualized and depended on ovarian reserve, age, baseline serum follicle stimulating hormone (FSH) concentration, and previous response to COS. Briefly, women received the long protocol with pituitary down-regulation using leuprolide acetate (Lupron1; Takeda [20] Standard IVF/ICSI procedures were used for oocyte fertilization, as previously described [12]. Fertilization was evaluated 16-18 h after IVF or ICSI. A zygote with two pronuclei (2PN) was defined as normal fertilization. All zygotes were scored according to the modified Scott scoring system [21]. A single team of embryologists coordinated all procedures of IVF laboratory to ensure that the culture processes and the embryo assessments were standardized. For PN scoring, zygotes were divided into 4 categories (Z1-Z4) based on 3 major features: size and location of the nuclei, appearance of the cytoplasm, and numbers, sizes, and distribution patterns of nucleolar precursor bodies within the nuclei. (S1 Fig)) Z1 zygotes have an equal number of nucleoli aligned at the PN junctions. Z2 zygotes have an equal number and size of nucleoli (3 to 7) that are equally scattered in the 2 PNs, but the nucleoli have not yet aligned at the PN junction. Z3 zygotes are characterized by inequality of the nuclei (unequal size, unequal numbers, or unequal alignment at the PN junction). Z4 zygotes have PN that are separated or different in size and small nucleoli that are partially aligned or scattered [21].
Embryos were cultured on days 1 to 3 in G1 TM medium (Scandinavian IVF Science) and on days 3 to 5 in G2 TM medium (Scandinavian IVF Science). Veeck's morphological grading system [22] was adopted for day-3 embryo scoring. A "good" 3-day embryo was defined as one that had a Veeck's grade of 1, with 8 cells, blastomeres of equal size, and no cytoplasmic fragments. Embryos were assessed by survival, morphology, and rate of cleavage. Embryos that had the same number of blastomeres at two sequential observations and zygotes that remained blocked at the pronuclear stage were classified as "developmentally arrested". Embryos were transferred on day 2, 3, 4, 5 or 6 after oocyte retrieval, as appropriate for each individual. The risk of embryo arrest was defined as the time to a first event. The start point for determining the duration of embryo survival was the date when PN appeared after 16-18 h of incubation, and the end point was after 5 days of extended culture. The study endpoints were the duration of overall survival and event-free survival (i.e. embryo arrest). In the analysis of embryo arrest, data from patients who received a day-2, day-3, day-4, day-5 or day-6 embryo transfer were censored after the time of transfer (loss due to transfer) or cryopreservation. Luteal phase support continued until the day pregnancy was confirmed by detection of hCG in the urine. If conception occurred, micronized progesterone supplementation was provided for an additional 4 weeks. Clinical pregnancy was defined as the presence of an intrauterine gestational sac with positive cardiac movement on ultrasound [23]. The live birth rate per transfer was defined as the proportion of IVF cycles reaching embryo transfer that resulted in the birth of at least one live-born child.

Statistical analysis
Continuous data are given as the mean ± standard deviation (SD). The Mann-Whitney rank sum test was used to compare continuous data, and the χ 2 test or Fisher's exact test was used to compare categorical variables. A Pearson correlation coefficient (r) was calculated for correlations. All statistical analyses were performed with SPSS, version 17.0 (SPSS, Inc., Chicago, IL, USA). All p values were two-sided, and a p value less than 0.05 was considered statistically significant. Logistic regression analysis was used to determine the significance of factors associated with a Z-score of 1 (Z1). Multivariate logistic regression analysis was then performed to further examine factors associated with Z1, considering the following variables: female age, male age, ICSI or IVF, etiology of infertility, body mass index, COS protocol, days of stimulation, FSH dosage, E2 level on the day of hCG detection, E2 level per oocyte, P4 level, maturity of oocyte, and number of same-cohort oocytes. Cumulative survival rates were calculated by life-table analysis using the Kaplan-Meier product limit procedure at each day. The differences between groups with different Z-scores were assessed using the log-rank test. A p-value less than 0.05 was considered significant [24].

General characteristics
During the period, 744 fresh cycles from 643 women were included, which consisted of 352 long protocol cycles and 392 GnRH-ant protocol cycles. Table 1 shows the characteristics of these cycles.
A total of 3,826 zygotes with normal fertilization produced from 744 fresh cycles following IVF/ICSI were examined. We performed fertilization check using the Z-score, a system that has been used to evaluate fertilization in our center since 2001. Factors that affect the Z-score In this retrospective and observational study, we did not perform statistical comparisons between cycles from the different protocols due to the risk of selection bias. Instead, we observed the cohort zygotes from the two protocols and analyzed potential factors that may influence the zygote score. Our previous study indicated good (Z-1) zygote had higher implantation potential. [12]We further analyzed the clinical and laboratory factors associated with a good Z-score of Z1. Univariate analyses (Table 2) showed that a score of Z1 correlated significantly with the COS protocol used (p < 0.001), oocyte maturity (p < 0.001), progesterone on the day of hCG detection (p < 0.001), E2 concentration per oocyte (p = 0.028), total FSH dosage (p = 0.025), and duration of FSH stimulation (p = 0.028).
We then used multivariable logistic regression analysis to identify factors associated with a score of Z1 (Table 3). These results indicate that a score of Z1 was significantly and independently associated with oocyte maturity (p < 0.001), number of cohort 2PN (p < 0.001), and COS protocol used (p < 0.001). The observation of significantly fewer Z1 zygotes from the GnRH-ant protocol is compatible with our clinical observations.
Thus, our study found that good-quality day 3 embryo/blastocyst formation and the cumulative survival rate were correlated with the Z-score but not the COS protocol.  Fig 5A and 5B displays the relationships between the number of good zygotes (Z1) transferred and pregnancy outcomes in the long (A) and GnRH-ant (B) protocols. In our study, a mean of 2.19 ± 0.86 embryos were transferred for each woman; therefore, it is difficult to compare the pregnancy outcomes according to Z-score based on single embryo transfer. Instead of single embryo transfer, we compared the number of good zygotes (Z1) transferred and pregnancy outcomes. For the long protocol, the number of Z1 in the transferred cohort embryos was not significantly correlated with the clinical pregnancy rate (r = 0.976; p = 0.088) or live birth rate (r = 0.944; p = 0.056). For the GnRH-ant protocol, the number of Z1 in the transferred cohort  2. Relationship between the Z-score with the formation of good quality embryos on day 3 (A) and blastocysts on day 5 or 6 (B) following the long protocol and the GnRH-ant protocol.

Discussion
To our knowledge, the current study is the first to report the effect of different COS protocols on zygote morphology and the feasibility of using the Z-score to predict embryo survival in the GnRH-ant protocol. Zygote formation follows the dramatic reorganization of sperm chromatin by many factors stored within the oocyte [25,26]. The morphological characteristics of the zygote indicate gamete quality and the potential for subsequent embryo implantation [27,28]. Zygotes with unequal numbers or sizes of nucleoli likely display asynchrony between male and female pronucleus development [29,30].
Many factors from the oocyte/sperm or the process of fertilization may affect the Z-score. Our IVF laboratory adopted a modified Scott scoring system beginning in 2001 to monitor fertilization. Our previous report (Lan, 2003) indicated that the Z-score is an additional criterion that can be used to select embryos for extended culture and is useful when selecting embryos for transfer. However, the distribution of Z-scores changes significantly when the GnRH-ant protocol is used for COS. In particular, we found that the Z-scores from the GnRH-ant protocol were generally poorer than those from the long protocol. Significantly fewer Z1 zygotes resulted from the GnRH-ant protocol, which is compatible with our clinical observations. Some possible explanation for the differences in the Z-score between the COS protocols involves the different endocrine profiles during follicular growth. First, the long protocol induces profound suppression and leads to simultaneous maturation of antral follicles [8,31,32]. Therefore, more mature follicles are recruited, and typically more Z1 zygotes are obtained. In contrast, the GnRH-ant protocol generates a more natural pattern of follicular recruitment, with uneven follicle sizes and fewer mature oocytes. Second, previous studies have suggested that the COS protocols have differing effects on ovarian E2 metabolism [33,34] and may affect follicular growth and/or luteal function. More specifically, Khalaf et al. [32] found that the protein kinase C (PKC) pathway was desensitized in GnRH-ag-treated granulosa cells, and this increased the levels of FSH and cAMP-mediated steroidogenesis. Third, in addition to the different endocrine profiles between the two protocols, the effect of the GnRH receptor on the oocyte/zygote should also be considered. GnRH receptors are expressed in the follicles at the gonadotropin-sensitive stage and in luteal cells. The correlation between the expression of GnRH receptors and follicular stage suggests that GnRH receptors directly influence folliculogenesis and oocyte development [35,36], but the mechanism by which GnRH affects the ovary is not completely understood. Thus, when GnRH-ant is used in IVF cycles, it has an unclear effect on the morphology and quality of oocytes. Animal studies indicate that GnRH-ant may have an adverse effect on primordial follicle survival in some species [37], but only a few studies examined the effect of GnRH-ant on human oocytes and embryos, with inconsistent results [38][39][40]. The results of our study also indirectly support the hypothesis that GnRH receptors play a role in oocyte maturation and affect zygote formation.
Although more mature oocytes/Z-1 zygotes were produced from long protocol than GnRH-ant possibly because of better endocrine profiles and no effect of GnRH antagonist; however, the overall Z-1 zygotes from long protocol had poorer survival and not correlated with clinical pregnancy. Our results suggested that long protocol only improved morphological maturity of oocytes and zygotes' morphology without changing the survival potential of oocyte. Some Z1 zygotes from long protocol may originated from immature oocytes with poorer potential and therefore had poorer pregnancy outcomes than Z1 zygotes from GnRHant protocol. The findings may also explain why Z-score could not be consistently verified in previous studies which used long protocol. There has been debate regarding the use of the Zscore for predicting outcomes following assisted reproduction [9,25,[41][42][43][44][45][46][47][48][49]. A recent systematic review [18] reported no conclusive data on the clinical efficacy of the Z-score in fresh cycles, even though biological results showed a good relationship with embryo viability and suggest a role in cycles with day-1 transfer/freezing. These inconsistent results may have occurred because almost all of these studies used the long protocol rather than GnRH-ant. With GnRH-ant, the number of Z1 in the transferred cohort of embryos was significantly correlated with the clinical pregnancy rate and live birth rate in our study (r = 0.971; p = 0.029). However, this correlation was not observed for the long protocol. A possible explanation for the Z-score having better predictive value in the GnRH-ant protocol is that zygotes from the GnRH-ant protocol were taken from a more natural pattern of follicle recruitment, while the long protocol only improved the zygotes' morphology without changing the survival potential.
The present study has some limitations. First, this retrospective study has risk of selection bias. A matched control group might improve the quality of the findings. Second, the relationship between pregnancy outcomes and Z-score for single embryo transfer requires further investigation. A mean of approximately two embryos were transferred to most patients in this study, so single-embryo outcomes could not be compared with the Z-score. Third, the timing of assessment is critical, as pronuclear development is a dynamic process. Thus, determination of the Z-score from a single light microscopy observation should be used with caution and only in conjunction with other methods of evaluation [27].
In conclusion, the Z-score distribution was poorer for the GnRH-ant protocol than the long protocol. This may be because these protocols have different effects on the endocrine profile or GnRH receptors. Additionally, the Z-score is a more feasible parameter for predicting embryo viability in IVF/ICSI with the GnRH-ant protocol than the long protocol.