Suboptimal trophectoderm mitochondrial DNA level is associated with delayed blastocyst development

Abstract

Purpose

To provide a comprehensive analysis of mtDNA quantity in D5 and D6 blastocysts, as well as a further insight to the origin of delayed blastocyst development.

Methods

A retrospective cohort analysis of 829 D5 and 472 D6 blastocysts from 460 patients who underwent in vitro fertilization (IVF) with next-generation sequencing (NGS)–based preimplantation genetic testing for aneuploidy (PGT-A). The quantity of trophectoderm mtDNA was extrapolated from the NGS data, followed by the analysis of mean mtDNA levels between D5 and D6 blastocysts of the same ploidy (aneuploid/euploid) and transfer outcomes (positive/negative clinical pregnancy).

Results

D5 blastocysts had significantly higher euploidy rate and clinical pregnancy rate when compared with D6 blastocysts. The proportion of blastocysts derived from patients ≧ 40 years old were similar between the D5 and D6 cohorts. When blastocysts with identical ploidy were analyzed, the D5 cohorts all had significantly higher mean mtDNA levels than their D6 counterparts. Similarly, when embryo transfers with identical outcome were analyzed, the D5 cohorts also had significantly higher mean mtDNA levels than the D6 cohorts. Trophectoderm mtDNA level was independent of maternal age and blastocyst morphology grades.

Conclusions

Our data provided further evidence D5 blastocysts contained significantly greater mtDNA quantity than D6 blastocysts, and mtDNA quantity could be a key factor that affects the development rate of blastocysts. Furthermore, one must avoid using an arbitrary threshold when incorporating mtDNA quantity into the embryo selection criteria, as the observed value may have vastly different clinical implication when blastulation rate is also considered.

Introduction

Although delayed blastulation is not an uncommon event during embryo culture, the precise mechanism responsible remains obscure. On the other hand, the clinical outcomes of day 5 (D5) versus day 6 (D6) blastocyst transfers (BTs) have been extensively investigated in literatures. Highlighted in a recent meta-analysis which included 8110 D5 and 4727 D6 BTs from 23 studies, the overall implantation rate, clinical pregnancy rate, live birth rate, and miscarriage rate were all found to be significantly in favor of D5 BTs [1]. Similar results were also observed when fresh and vitrified BTs were analyzed separately, thereby mitigating the possible effect of embryo-endometrial asynchrony in fresh D6 BTs [1]. Furthermore, in studies that considered only the euploid BTs, the D5 cohorts still yielded significantly better outcomes than their D6 counterparts [2, 3]. These findings indicated, in addition to endometrial and chromosomal factors, there are other critical elements affecting the reproductive potential and the development of D6 blastocysts. In the study by Hashimoto et al., a significantly higher incidence of abnormal spindle formation and lower implantation rate were observed in D6 vitrified-thawed blastocysts, providing another possible explanation for the inferior outcomes of the slower developing embryos [4].

Recently, the quantity of mitochondrial DNA (mtDNA) in trophectoderm cells was regarded as another potential marker for embryo selection. Since each mitochondria organelle has its own unique genome, the mtDNA will also be simultaneously sequenced along with the nuclear DNA (nDNA) during preimplantation genetic testing for aneuploidy (PGT-A). Studies have shown trophectoderm cells from aneuploid blastocysts contained significantly greater mtDNA quantity than euploid blastocysts [5–9]. Furthermore, euploid blastocysts with mtDNA above a threshold level were also found to have reduced implantation potential [5–9]. Therefore, increased mtDNA quantity in trophectoderm cells was regarded as a sign of inferior blastocyst viability. These findings were nevertheless challenged by others. In an analysis by Victor et al. [10], no significant difference in trophctoderm mtDNA quantity was found after applying a set of “correction factors” tailored for respective chromosomal size and composition, regardless of blastocyst ploidy or transfer outcomes. In another interesting study by Treff et al., under the setting of double euploid embryo transfers that resulted in singleton-only pregnancies, no significant difference in the trophectoderm mtDNA quantity was found between the two transferred embryos [11]. Similar findings were also observed in another study by Klimczak et al., as mtDNA quantity was found not to be associated with blastocyst implantation or ongoing pregnancy [12]. The contradictory findings propelled a series of debates centered around the methods of mtDNA quantification used in respective studies [13, 14] and limited the routine use of mtDNA quantity in embryo selection.

Interestingly, in studies which D5 and D6 blastocysts were analyzed separately, significantly higher mtDNA quantity was also noted in the faster growing cohorts [9, 12]. This finding was intriguing since it potentially contradicted the concept of “the less the better” put forth for mtDNA quantity in blastocysts. Nevertheless, the significance of the observed mtDNA disparities could not be further examined as the respective transfer outcomes and various confounding factors were not analyzed. Therefore, the objective of our study is to provide a comprehensive analysis of mtDNA quantity in D5 and D6 blastocysts, as well as a further insight to the origin of delayed blastocyst development.

Material and methods

Study design

This was a retrospective cohort analysis consisted of de-identified next-generation sequencing (NGS) data from patients who underwent IVF with PGT-A at IHMED Fertility Center, Taiwan, from November 2017 to April 2019. The study protocol was approved by the institutional review board of the National Taiwan University Hospital (IRB #201905053RIND).

Study participants and embryo samples

A total of 1301 blastocysts from 460 subjects between the age of 36 and 46 years (mean age = 37.79 years old) were enrolled for analysis. The blastocysts were further divided into four groups according to their ploidy and development rates: euploid D5 blastocyst (n = 341), aneuploid D5 blastocysts (n = 488), euploid D6 blastocyst (n = 163), and aneuploid D6 blastocysts (n = 309). Blastocysts that failed to meet the biopsy criteria on D6 were excluded for analysis. Other exclusion criteria were cases with uterine structural anomaly, untreated hydrosalpinx, and autoimmune disorders such as systematic lupus erythematous and antiphospholipid syndrome. Lastly, all mosaic blastocysts were also excluded.

Ovarian stimulation and embryo processing protocols

All cases underwent flexible gonadotropin-releasing hormone antagonist protocol with the starting dose of recombinant FSH (Gonal-F; Merck) or human menopausal gonadotropin (Menopur; Ferring) from 175 to 225 IU. All oocytes were inseminated by intracytoplasmic sperm injection (ICSI), and the derived zygotes were cultured with a continuous single culture medium (Global Total; Life Global). Five-to-eight-cell trophectoderm biopsies were performed with laser (LYKOS; Hamilton-Thorne) on either day 5 or day 6 post fertilization, depending on the morphological grade of individual blastocysts. Specifically, Gardner’s system was used to score each blastocyst [15], and only the ones graded 3CC or higher were biopsied. All embryos were then cryopreserved by vitrification (Cryotop; Kitazato). Each biopsied sample was then suspended in 1× phosphate-buffered saline (HyClone; GE Healthcare Life Sciences) and placed in a tube containing 2.5 μl of 0.5% polyvinylpyrrolidone.

NGS and PGT-A workflow

Whole genome amplification (WGA) of biopsied samples was performed using the SurePlex WGA Kit (Illumina). Libraries for the WGA samples were prepared with standard Veriseq protocol (Illumina). Single-end, 36 base pairs reads dual-index sequencing was performed with a MiSeq sequencer (Illumina). Phred quality score (Q-score) was used to assess the probability of errors in base calling during sequencing. In the present study, greater than 90% base calls had scores higher than Q30, which was equivalent to base call accuracy of 99.9%. Samtools (Genome Research Limited) was used to remove unmapped, duplicated, and low scoring reads. The filtered reads were then mapped to the reference human genome (hg19) using Burrows-Wheeler Alignment tool [16] and analyzed with BlueFuse Multi Software (Illumina). Embryo ploidy was determined by comparing the reads number from bins of the tested sample to the reads number from the corresponding region of the reference euploid sample.

mtDNA quantification

The mtDNA level of each biopsy sample was calculated from the NGS data using the MitoCalc analyzer [17]. According to the software’s algorithm, the average sequencing coverage for autosomal DNA and mtDNA are proportional, under the assumption that both genomes are sequenced with equal opportunities. Therefore, the mtDNA copy number per cell can be inferred by the ratios of sequence coverages between mtDNA and autosomal DNA with a multiplication factor of two, taken into consideration the presence of two copies of autosomal DNA in a cell. Furthermore, only the reads uniquely mapped to the reference mtDNA genome were included in our analysis, thus minimizing the confounding presence of the reads from the nuclear mitochondrial DNA (NUMTs). The mitochondrial genome coverage depth in our study was between 30× and 50× per analyzed sample.

Embryo transfer and determination of pregnancy

Single thawed euploid embryo transfers with hormone replacement cycle were performed in all cases. Clinical pregnancy was defined as the ultrasound presence of a gestational sac between the 3rd and 4th week post transfer. All biochemical pregnancies, defined as transient elevations of serum beta human chorionic gonadotropin without visualization of a gestational sac, were allocated to the non-pregnant group.

Outcome measures

The study’s primary outcome was the quantity of mtDNA in D5 and D6 blastocysts of matched ploidy and transfer results. The secondary outcomes were the relationship between mtDNA quantity and two key factors known to impact IVF outcome: the maternal age and the morphological grades of blastocysts.

Statistical analysis

For continuous variables, Student’s t test was used to compare the means between two groups, and one-way analysis of variance (ANOVA) followed by post hoc test was used to compare means between three or more groups. Chi-square test with Yates correction was used to compare the observed frequencies of categorical variables. A result was considered statistically significant if the P value is less than 0.05. All statistical analyses were performed with GraphPad Prism version 8.3.0 and Microsoft Excel 2016.

Results

Clinical outcomes of D5 versus D6 blastocysts

D5 blastocysts had significantly higher euploidy rate (41.1% vs. 34.5%; P = .02) and clinical pregnancy rate (69.7% vs. 57.2%; P = .03) compared to D6 blastocysts. Taken into consideration that an individual could produce both D5 and D6 blastocysts in a single IVF cycle, the respective proportion of D5 and D6 blastocysts from patients of very advanced maternal age (vAMA), defined as age ≥ 40 years old in literatures [18], was also calculated. The difference in the proportion of D5 and D6 blastocysts derived from patients with vAMA was not statistically significant. The details of the above results are summarized in Table 1.

Table 1.

The clinical outcomes and age distribution between D5 and D6 blastocysts

	Day 5 blastocysts	Day 6 blastocysts	P value
Euploid rate	41.1% [341/829]	34.5% [163/472]	.02
Clinical pregnancy rate	69.7% [159/228]	57.2% [63/110]	.03
vAMA rate	33.3% [276/829]	38.8% [183/472]	NS

NS, non-significant; vAMA, very advanced maternal age (≥ 40 years old)

mtDNA levels in D5 and D6 blastocysts

When examining blastocysts with the same ploidy, all D5 groups were found to contain significantly greater quantity of mtDNA than the D6 groups. Similarly, when examining transferred blastocysts with the same outcomes, the D5 groups also all had significantly greater mean mtDNA quantity than their D6 counterparts. The details of the above results are summarized in Table 2. In the intra-group analyses of blastocysts with the same development rate, the aneuploid cohorts were found to have significantly higher mean mtDNA levels than euploid cohorts. Likewise, the un-implanted cohorts were found to contain significantly higher mean mtDNA levels than the implanted cohorts (data not shown).

Table 2.

The analysis of mitochondria DNA levels in D5 and D6 blastocysts with identical ploidy and transfer outcomes

	Biopsy day	Number of embryos	mtDNA level [95% CI]	P value
Aneuploid blastocysts	D5	488	0.001239 [0.001109–0.001369]	.003
	D6	309	0.000963 [0.000862–0.001063]
Euploid blastocysts	D5	341	0.001072 [0.000995–0.001149]	.0002
	D6	163	0.000819 [0.000711–0.000927]
ET with positive CP	D5	159	0.000758 [0.000698–0.000818]	.0001
	D6	63	0.000572 [0.000516–0.000628]
ET with negative CP	D5	69	0.000958 [0.000832–0.001082]	.04
	D6	47	0.000756 [0.000661–0.000851]

CI, confidence interval; CP, clinical pregnancy; D5, day 5; D6, day 6; ET, embryo transfer; mtDNA, mitochondria DNA

Relationship between mtDNA quantity and maternal age

When blastocysts with similar blastulation rate and ploidy were analyzed, there was no significant difference in the mean mtDNA levels between the vAMA group and the younger age group. The results from the above analysis are summarized in Table 3.

Table 3.

The analysis of mitochondria DNA levels in vAMA and younger aged patients

	Age	Number of embryos	mtDNA level [95% CI]	P value
Euploid D5 blastocysts	< 40	273	0.001081 [0.000993–0.001169]	NS
	≥ 40	68	0.001033 [0.000883–0.001184]
Aneuploid D5 blastocysts	< 40	280	0.001273 [0.001090–0.001456]	NS
	≥ 40	208	0.001173 [0.001032–0.001313]
Euploid D6 blastocysts	< 40	134	0.000828 [0.000701–0.000956]	NS
	≥ 40	29	0.000779 [0.000641–0.000918]
Aneuploid D6 blastocysts	< 40	155	0.000945 [0.000778–0.001112]	NS
	≥ 40	154	0.000981 [0.000867–0.001094]

CI, confidence interval; D5, day 5; D6, day 6; mtDNA, mitochondria DNA; NS, non-significant; vAMA, very advanced maternal age (≥ 40 years old)

Relationship between mtDNA quantity and blastocyst morphology

The blastocysts were sub-divided into three groups according to Gardner’s scoring system: group 1 (3 to 6 AA; 4 to 6 AB), group 2 (3AB; any BA and BB), and group 3 (Any CA, BC, CB, and CC). When blastocysts with similar blastulation rate and ploidy were analyzed, no significant inter-group difference in the mean mtDNA levels was found among the three morphology groups. The results from the above analysis are summarized in Table 4.

Table 4.

The analysis of mitochondria DNA levels in blastocysts with different morphological grades

	Comparison of mtDNA level between morphology groups	Mean difference [upper CI–lower CI]	q	P value
Euploid D5 blastocysts (Group 1: n = 104) (Group 2: n = 151) (Group 3: n = 86)	Group 1 vs. group 2	0.000347 [0.000701–− 0.000005]	3.273016	NS
	Group 1 vs. group 3	0.000163 [0.000589–− 0.000264]	1.272549
	Group 2 vs. group 3	0.000183 [0.000592–− 0.000224]	1.496341
Aneuploid D5 blastocysts (Group 1: n = 161) (Group 2: n = 221) (Group 3: n = 106)	Group 1 vs. group 2	0.000092 [0.000306–− 0.000121]	1.448164	NS
	Group 1 vs. group 3	0.000238 [0.000488–− 0.000012]	3.160048
	Group 2 vs. group 3	0.000145 [0.000378–− 0.000088]	2.071130
Euploid 6 blastocysts (Group 1: n = 43) (Group 2: n = 59) (Group 3: n = 61)	Group 1 vs. group 2	0.000041 [0.000318–− 0.000235]	0.497093	NS
	Group 1 vs. group 3	0.000127 0.000438–− 0.000184]	1.362088
	Group 2 vs. group 3	0.000168 [0.000483–− 0.000145]	1.785783
Aneuploid D6 blastocysts (Group 1: n = 75) (Group 2: n = 114) (Group 3: n = 120)	Group 1 vs. group 2	0.000028 [0.000417–− 0.000361]	0.241683	NS
	Group 1 vs. group 3	0.000171 [0.000506–− 0.000162]	1.717214
	Group 2 vs. group 3	0.000143 [0.000458–− 0.000171]	1.524540

CI, confidence interval; D5, day 5; D6, day 6; mtDNA, mitochondria DNA; NS, non-significant

Discussion

This study was the first comprehensive analysis of the relationship between mtDNA quantity and blastulation timing. Our results further confirmed that D5 blastocysts contained significantly greater overall mtDNA quantity than D6 blastocysts, even when variables such as maternal age, morphological grade, embryo ploidy, and transfer outcomes were considered. As the reproductive outcomes of D5 blastocysts have long been found to be superior to those of D6 blastocysts, these findings further implied that the relationship between mtDNA quantity and blastocyst viability was not dichotomous but more complex than the “less is better” theory suggested previously.

The control of mtDNA replication in mammalian reproduction is a dynamic process. During the early phase of female embryogenesis, the mtDNA in primordial germ cells would first undergo significant reduction in quantity, then increase gradually during oogenesis and peaks when the oocytes reach metaphase II stage [19]. This process of initial reduction followed by amplification of mtDNA at the end of folliculogenesis constitutes a genetic bottleneck that ensures the preservation of only the healthiest mitochondria genome [20]. With the downregulation of replication factors POLGA, POLGB, and TFAM after fertilization, no net gain of mtDNA quantity would occur up until the morula stage [21–23]. Consequently, the number of mtDNA copies in individual blastomere would progressively decrease after each cell division. Upon reaching the blastocyst stage, the mtDNA replication factors would once again be upregulated, which led to increased overall mtDNA copy number [24–26].

Although the disparity of mtDNA quantity between D5 and D6 blastocysts has been observed [9, 12], no conclusive explanation was given. In their report, Lledo et al. proposed a “dilutional theory” for the phenomenon, as they claimed D6 blastocysts were more likely to contain greater number of cells than D5 blastocysts since they had undergone one more day of cell growth and division [9]. As gainful replication of mtDNA would not occur until only after the blastocyst stage [24–26], the “fixed” amount of mtDNA would be “diluted” among greater number of cells in D6 blastocysts. Nevertheless, the above theory had failed to acknowledge the primary reason for performing D6 cultures was to enable slower growing embryos additional time to develop, as they had not met the morphological criteria of mature blastocysts on the preceding day. Therefore, the overall cellularity between D5 and D6 blastocysts should approximate each other, thereby negating the proposed dilutional effect.

We hypothesized that the D6 blastocysts were derived from oocytes containing suboptimal mtDNA quantity initially; therefore, they were energetically challenged and required additional time to complete the blastulation process. Studies have demonstrated that oocyte quality is intrinsically linked to the quantity of mtDNA they contained. Oocytes from women of diminished ovarian reserve or advanced reproductive age were found to contain significantly lower mtDNA quantity than those of normal ovarian reserve or younger age, respectively [27–29]. Additionally, unfertilized oocytes also have significantly lower mtDNA copy numbers than the fertilized cohort [30–32], with degenerated oocytes containing the least amount of mtDNA among all investigated subgroups [32]. Furthermore, agents that promoted mitochondrial biogenesis (rapamycin, resveratrol) [33–36] or functional capacity (coenzyme Q10) [37, 38] have been found to restore the developmental capacity of aged mammalian oocytes. In a recent meta-analysis that investigated the adjuvant treatments for IVF poor responders, high dose coenzyme Q10 was the only one that was found to increase the likelihood of clinical pregnancy while decrease the chance of IVF cycle cancelation at the same time [39]. These findings clearly demonstrated the importance of adequate mtDNA level in oocytes for successful fertilization and embryogenesis.

One may wonder if mitochondria have such important functions in oocytes, how can increase mtDNA level in blastocysts be linked to aneuploidy and reduced implantation potential? The possible explanation for such paradoxical relationship may be viewed in context of suboptimal mitochondria quality per se, or the overall dysfunctional embryonic development. In the first scenario, when an embryo contained excessive functionally deficient mitochondria, a compensatory mechanism of mitochondrial pool expansion would be activated in effort to restore normal ATP production. However, if most of the newly replicated mitochondria still harbor the same defect, the defunct energy state of the embryo would still be unrectified. Indeed, mtDNA has been found to be highly susceptible to mutation due to its proximity to reactive oxygen species generated by the electron transport chain during oxidative phosphorylation, as well as the lack of a complete DNA repair mechanisms like the nuclear genome [40]. Another plausible theory considered the overall quality of embryo rather than the functional status of mitochondria organelle alone. In the “quiet embryo hypothesis” proposed by Leese et al. [41], embryos with increased metabolic activity were thought to be experiencing certain extrinsic stress (i.e., culture condition) or intrinsic structural defects (i.e., aneuploidy). Greater energy expenditure thus was required in effort to restore proper embryonic development, leading to compensatory mitochondrial replication and a net gain of mtDNA. The excessive ATP production could also lead to increase free radical production that led damage to DNA and other organelles [42, 43].

The significance of mtDNA quantity observed is two-sided. We believe one of the key mechanisms that dictates blastulation rate is the “initial” mtDNA quantity that the oocytes contained. Those with adequate mtDNA would likely blastulate “on time,” while ones with suboptimal quantity would require additional time to complete the blastulation process. On the other hand, blastocyst implantation potential is influenced by multiple factors, including mitochondrial gene pool, embryonic factors, and culture conditions. Disruption of any of the above could lead to the event of premature mtDNA replication before the blastocyst stage. Therefore, although changes in mtDNA quantity are involved in both scenarios, the respective clinical implications are fundamentally different and should be regarded as distinct events. One prime example is illustrated in our data (Table 2), as the D5 euploid blastocysts with positive CP had almost identical mean mtDNA quantity as D6 euploid blastocysts with negative CP. We postulate that the D5 euploid blastocysts with positive CP were derived from oocytes with sufficient quantity of mtDNA and have developed without the perturbance of the above-mentioned conditions. While D6 euploid blastocysts with negative CP were likely derived from oocytes with suboptimal quantity of mtDNA initially (thus requiring an extra day to complete blastulation), they also had experienced critical events that triggered premature replication of mtDNA before reaching the blastocysts stage. Therefore, even though the two cohorts had similar observed mtDNA quantity, the changes in mtDNA quantity in D5 euploid blastocysts with positive CP followed a physiologic pathway while D6 euploid blastocysts with negative CP cohort followed a pathological pathway.

In line with previous reports, our results also showed that mtDNA quantity did not correlate with either maternal age [5–7, 44] or embryo morphology [5, 7, 11]. The possible rationale could be explained by the sample selection method. Since only the D5/D6 blastocysts with 3CC Gardner’s grading or above were included for final analysis, all embryos that exhibited suboptimal growth rate or cellular components were excluded. Therefore, only the mtDNA content from the cohort that demonstrated adequate degree of developmental competence was analyzed. Indeed, similar phenomenon has been observed in studies involving PGT-A, as the reported implantation rates and live birth rates after euploid BTs did not differ significantly among the investigated age groups, irrespective of maternal age [45] or morphological grading [46].

Study limitations

The main limitation of this study was its retrospective design. Therefore, the mtDNA content could only be inferred quantitatively from the original NGS data for PGT-A, while other mitochondrial functional markers such as oxygen consumptions rate and ATP synthesis could not be evaluated. As outlined in a recent publication by Morimoto et al. [44], the increase in OCR in morulae was associated with shortened time required for the embryos to progress to mid-stage blastocyst. The same study also demonstrated that decreased in OCR, but not the mtDNA copy number, was associated increased maternal age and retarded blastocyst growth. These findings have placed further emphasis on the importance of mitochondrial functional status in proper embryonic development. Furthermore, as outlined in the review by Viotti et al., a consensus for the optimal quantification methods for mtDNA is still lacking, with NGS and qPCR both being adopted by numerous groups with conflicting results [47]. To complicate the matter even more, the phenomenon of elevated mtDNA levels in blastocysts was not a universal but rather a center-specific event. In the two of the largest multi-center studies that reported elevated mtDNA levels in euploid blastocysts, only half of the participating centers (17 out of 34) actually observed the phenomenon when the data was reported per individual center instead of the summation from all of the centers together [7, 8]. Lastly, the ongoing debate regarding whether a single trophectoderm biopsy can adequately represent the entire nuclear and mitochondrial genome in an embryo still warrants careful consideration.

Conclusions

This study was a first ever comprehensive analysis of trophectoderm mtDNA quantity in D5 and D6 blastocysts. Our data provided further evidence D5 blastocysts contained significantly greater mtDNA quantity than D6 blastocysts, and mtDNA quantity could be a key factor that affects the development rate of blastocysts. Moreover, one must avoid using an arbitrary threshold when incorporating mtDNA quantity into the embryo selection criteria, as the observed value may have the vastly different clinical implications when blastulation rate is also considered. Further prospective studies that investigate both the quantitative and qualitative aspects of mtDNA are still required to fully elucidate the complex relationship between mitochondria genetics and human reproduction.

Authors’ contributions

Study design and coordination: Frank Shao-Ying Wu and Ni-Chung Lee

Subject recruitment: Shao-Peng Weng, Meng-Shun Shen, and Pei-Chun Ma

Data collection: Shao-Peng Weng, Meng-Shun Shen, and Po-Kuan Wu

Data analysis: Frank Shao-Ying Wu, Ni-Chung Lee, Po-Kuan Wu, and Pei-Chun Ma

Drafting of manuscript: Frank Shao-Ying Wu

Study supervision and manuscript revision: Ni-Chung Lee

Data availability

All data and material are available upon request.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflicts of interest.

Ethics approval

The study protocol was approved by the institutional review board of the National Taiwan University Hospital (IRB #201905053RIND).

Consent to participate

Not applicable due to the retrospective nature of the study.

Consent for publication

The authors transfer to Springer the publication rights and give full consent for all of the information about to be published in Journal of Assisted Reproduction and Genetics.

Code availability

Not applicable.