Evaluating the application value of NGS-based PGT-A by screening cryopreserved MDA products of embryos from PGT-M cycles with known transfer outcomes

Abstract

Purpose

To determine the application value of next-generation sequencing (NGS)-based preimplantation genetic testing for aneuploidies (PGT-A).

Methods

We conducted a retrospective case–control study on a cohort of frozen-thawed embryo transfer (FET) cycles following preimplantation genetic testing for monogenic disorders (PGT-M) between 2014 and 2017. Cycles that produced live births or early miscarriages were divided into live birth group (n = 76) or miscarriage group (n = 19), respectively. The NGS-based aneuploidy screening was performed on the multiple displacement amplification (MDA) products of the embryonic trophectoderm biopsy samples that were cryopreserved following PGT-M.

Results

In the live birth group, 75% (57/76) embryos were euploid and 14.5% (11/76) were aneuploid. The remaining 10.5% (8/76) embryos were NGS-classified mosaic with the high- (≥ 50%) and low-level (< 50%) mosaicism rates at 7.9% (6/76) and 2.6% (2/76), respectively. In the miscarriage group, only 23.5% (4/17) embryos were aneuploid, while 58.8% (10/17) were euploid and 17.6% (3/17) were NGS-classified mosaic with the high- and low-level mosaicism rates at 11.8% (2/17) and 5.9% (1/17), respectively. For live birth and miscarriage groups, the transferable rate was 82.9% (63/76) and 70.6% (12/17), respectively, whereas the untransferable rate was 17.1% (13/76) and 29.4% (5/17), respectively.

Conclusion

The application of NGS-based PGT-A remains questionable, as it may cause at least one in six embryos with reproductive potential to be discarded and prevent miscarriage in less than one in three embryos in single-gene disease carriers.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-022-02447-7.

Introduction

Preimplantation genetic testing (PGT) for aneuploidies (PGT-A) aims to help women with repeated in vitro fertilization (IVF) implantation failures, repeated spontaneous abortions, or advanced maternal age to avoid aneuploid embryo transfer, and reduce the miscarriage rate and time to achieve a live birth. Since the birth of the first baby following PGT-A treatment in 1995 [1], it has been used worldwide for more than 20 years, but its clinical value has always been controversial. The initial controversy focused on PGT-A (PGT-A 1.0) using blastomere biopsy combined with fluorescence in situ hybridization, which did not improve pregnancy outcomes in older women [2–5]. A series of randomized clinical trials (RCTs) with class I evidence has proven that PGT-A 2.0 can improve clinical outcomes [6–9]. PGT-A 2.0 is based on trophectoderm (TE) biopsy and comprehensive chromosome screening, such as array comparative genomic hybridization (aCGH) and next-generation sequencing (NGS).

Although great progress has been made in PGT-A in recent years, its accuracy remains fraught with multiple challenges, including the following: (1) Embryo mosaicism: the existence of both euploid and aneuploid cells within the same embryo. Although embryos detected with aneuploid cells are not transferred, there are reports of healthy live births after the transfer of mosaic embryos [10–13], which has caused some confusion on how to interpret the results of PGT-A. (2) The consistency of TE and the inner cell mass (ICM): the ICM is the tissue that eventually develops into the fetus. The consistency of TE and ICM reported in current research reports is quite different (62.10–98.04%). Whether TE can accurately reflect the karyotype of ICM remains controversial [14–17]. (3) Reliability of single-point biopsy: Gleicher et al. [18] indicated that 50% (5/10) of embryos yield different results from biopsies in different parts of the tissue, suggesting that a single-point biopsy cannot accurately reflect the true karyotype of the embryo. A single-point TE biopsy of six cells is also mathematically unable to accurately reflect the ploidy of a blastocyst with approximately 300 cells in the TE [19]. (4) Bias due to whole-genome amplification (WGA): WGA is an important prerequisite for genome-wide analysis in a limited number of cells; however, this step may cause distortions to the initial DNA template and affect the accuracy of the subsequent analysis [20, 21]. (5) Reliability of the PGT-A detection platform: Currently, the most commonly used PGT-A detection platform is NGS. Compared to aCGH, NGS can detect a low level of mosaicism (> 20% vs. > 40–50%) [22]. Of the TE samples initially diagnosed as euploid by aCGH, 31.6% were found to be mosaic, and 5.2% were found to be polyploid when reanalyzed using NGS [23], which can provide higher resolution and detect deletions and duplications as small as 10 Mb [24].

Hence, the clinical value of PGT-A 2.0 questions what percentage of embryos with reproductive potential would be wasted by NGS-PGT-A diagnosis as aneuploid and what percentage of miscarriages would be prevented by NGS-PGT-A. Since December 2014, our center has been using multiple displacement amplification (MDA) to perform WGA for the cycles of PGT for monogenic diseases (PGT-M), and the remaining MDA products have been properly cryopreserved after PGT-M. To date, more than 200 healthy babies have been born through PGT-M treatment without PGT-A screening. To evaluate the application value of NGS-based PGT-A, we performed NGS-based aneuploidy screening on the stored MDA products of embryos with known embryo transfer outcomes.

Material and methods

Study design and population

This study was conducted in a cohort of frozen-thawed embryo transfer (FET) cycles following PGT-M treatment in the First Affiliated Hospital of Sun Yat-sen University between December 19, 2014, and December 31, 2017. FET cycles that met the following criteria were included: TE biopsy and single blastocyst transfer were performed; embryonic MDA products were well-cryopreserved; clinical outcomes were known (including early miscarriage and healthy live birth). Excluded cycles were PGT-M plus PGT-A cycles (the transferred embryos were screened for aneuploidy); from patients with known chromosomal abnormalities; from couples with a history of recurrent pregnancy loss (RPL); and if MDA products corresponding to the involved embryos could not be found. The FET cycles resulting in early miscarriages (< 12 weeks) were marked as the cases (miscarriage group), and those that reached live birth were marked as the controls (live birth group).

Ethical approval

The study was approved by the Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University. Informed consent was received from all enrolled patients.

Data collection

The characteristics of the included cycles included age, chromosome karyotype, body mass index, basal follicle-stimulating hormone level, pregnancy history, endometrium preparation protocol of FET cycle, and morphological scores of transferred embryos (Table 1). The patients were followed up by telephone calls until December 31, 2020, to collect data on the pregnancy outcomes of the included cycles. For the cycles with live birth, information on prenatal diagnosis, mode of delivery, health status, and growth and development of the infants was collected. Patients who aborted were asked to update the information on aneuploidy detection and results if performed.

Table 1.

Basic characteristics of the included cases

	Live birth group N = 76	Miscarriage group N = 17	P-value
Maternal age (years)	30.38 ± 3.56	31.65 ± 3.97	0.198
Paternal age (years)	32.67 ± 4.20	33.71 ± 4.01	0.357
BMIa (kg/m2)	21.14 ± 2.60	21.35 ± 2.87	0.759
Basal FSHb level (mIU/mL)	5.33 ± 1.69	5.63 ± 1.34	0.502
Gc	1.93 ± 1.57	1.53 ± 1.63	0.342
Pc	0.28 ± 0.53	0.29 ± 0.47	0.899
Ac	1.66 ± 1.30	1.24 ± 1.35	0.232
Number of previous FETd cycles	0.47 ± 0.76	0.47 ± 0.62	0.988
Cycles with good quality embryo transfer	96.10% (73/76)	88.20% (15/17)	0.196
Endometrial preparation protocol			0.493
NCe	22.40% (17/76)	35.30% (6/17)
HRTf	76.30% (58/76)	64.70% (11/17)
OIg	1.30% (1/76)	0.00% (0/17)
Endometrial thickness on the day of progesterone administration (cm)	9.74 ± 2.19	9.29 ± 1.57	0.426
Time from biopsy to aneuploidy detection (years)	3.38 ± 13.36	1.99 ± 0.67	0.667

^aBMI, body mass index; ^bFSH, follicle-stimulating hormone; ^cG, P, A, gestations, productions, abortions; ^dFET, frozen-thawed embryo transfer; ^eNC, natural cycle; ^fHRT, hormone replacement therapy cycle; ^gOI, cycle with ovulation induction

Acquisition and aneuploidy screening of the embryonic MDA products

In the PGT-M cycle, all embryos underwent a TE biopsy on days 5 or 6 after fertilization. Biopsy samples were amplified using Repli-g MiDi kit (Qiagen, Germany) or Repli-g Single-cell kit (Qiagen) for MDA-based WGA. Part of the obtained MDA product was used as a template for α- and β-thalassemia mutation detection performed using fluorescence gap polymerase chain reaction (PCR) and reverse dot blot + singleplex fluorescence PCR, respectively. The detailed procedures of TE biopsy, MDA-based WGA, and PCR-based thalassemia mutation detection are described in our previous study [25–27]. The remaining MDA product was frozen at − 80 °C.

The MDA samples corresponding to the embryos transferred in the inclusion cycles were thawed and screened for aneuploidy using NGS by NextSeq® 550 platform (Illumina, San Diego, CA, USA). We sequenced the amplified genome of each sample at approximately 0.1 × genome depth and resolution of chromosomal abnormality detection at 4 Mb with the NextSeq® 550 platform, formerly known as NexCCS, using Ion Torrent NGS for PGT-A as previously described [28, 29]. We carried out quality control analysis on the test results of the samples and found that when the standard deviation (SD) value was > 5, the background noise was too large and might lead to errors in judgment. An example is shown in Online Resource 1, in which a newborn in the live birth group was diagnosed with euploidy by a karyotype test, whereas the corresponding frozen MDA product was diagnosed as monosomy on chromosome 18 by NGS-PGT-A, with an SD value of 9.248. Therefore, cases with an SD value > 5 were excluded to avoid a diagnostic error. The analysis was undertaken blindly, and a second independent individual decoded the results.

Statistical analyses

The continuous variables were expressed as mean ± SD, whereas categorical variables were expressed as case number (n) and percentage (%). Student’s t-test was used to analyze the normally distributed continuous variables, and the Mann–Whitney U test was used to analyze non-normal data. The Pearson chi-square test was used to analyze the classification variables, and Fisher’s exact test was used when the theoretical frequency was < 5. All tests were conducted bilaterally, and P < 0.05 was considered statistically significant. Data were analyzed using the IBM SPSS Statistics, version 23.0 (IBM Corp., Armonk, NY, USA).

Outcomes

The NextSeq® 550 platform can detect segmental chromosomal abnormalities larger than 4 Mb and detect samples with 30% to 70% mosaicism (Online Resource 2). Based on this platform, embryos with 30 to 70% mosaicism were defined as mosaic embryos, those with < 30% mosaicism were euploid embryos, and those with > 70% mosaicism were defined as true aneuploid embryos. MDA products with two chromosomal abnormalities (regardless of mosaicism or aneuploidy) were classified as double chromosomal abnormalities. Those with three or more chromosomal abnormalities were classified as complex chromosomal abnormalities.

Results

Demographics

A total of 124 FET cycles with complete information were collected. All the couples involved were treated with PGT-M in our center because they were carriers of thalassemia with no history of RPL.

In the live birth group, a total of 103 cases were included. Among them, 12 experienced premature birth (11.7%), and 91 had full-term pregnancies (88.3%). Vaginal delivery occurred in 61 cases (59.2%), and cesarean delivery occurred in 42 cases (40.8%). One of the newborns had an atrial septal defect after birth. The B-ultrasound at the age of 2 years showed that it had healed. All the infants were followed up by telephone until the end of 2020. They are presently 1.71–5.23 years old, and there are no obvious abnormalities in growth or intellectual development. A total of 103 MDA products were tested for the live birth group, of which 11 were excluded for amplification failure, and 16 were excluded because of SD values > 5. The average SD value of the remaining 76 cases was 2.622 (range, 1.377–4.3). The NGS analyses produced approximately 4,486,949–9,412,642 DNA sequence reads per sample, with an average of 7,254,967 reads successfully aligned to the human genome.

Twenty-one embryo transfer cycles in the miscarriage group were included. Of the 21 corresponding MDA products detected, four were not evaluated owing to amplification failure or poor test data. In the remaining 17 cases, the average SD value of NGS data was 2.599 (range, 1.614–4.994). The NGS analyses produced approximately 6,179,326–8,564,805 DNA sequence reads per sample, with an average of 7,297,114 reads successfully aligned to the human genome. There was no significant difference in the basic characteristics of the two groups (Table 1).

Aneuploidy screening of the embryonic MDA products

Compared to the miscarriage group, the live birth group had a higher euploidy rate (75.0% vs. 58.8%, P = 0.296) and lower mosaicism rate (10.5% vs. 17.6%, P = 0.684) and aneuploidy (14.5% vs. 23.5, P = 0.580) (Table 2). Moreover, both the high-level (≥ 50%) and low-level (50%) mosaicism rates were higher in the miscarriage group (11.8% vs. 7.9%, P = 0.971; 5.9% vs. 2.6%, P = 1.000). As is evident, embryos with high-level mosaicism have lower implantation potential than euploid embryos, whereas embryos with low-level mosaicism are reproductively competent as euploid embryos [13, 30]. Our center defined embryos with ≥ 50% mosaicism and aneuploid embryos as untransferable embryos, whereas the euploid embryo and embryos with < 50% mosaicism were considered transferable embryos. A higher transferrable embryo rate was observed in the live birth group (82.9% vs. 70.6%), and a higher untransferable embryo rate was observed in the miscarriage group (29.4% vs. 17.1%). However, the differences were not statistically significant (Table 2).

Table 2.

Results of reanalysis of embryonic multiple displacement amplification samples with next-generation sequencing

	Live birth group N = 76	Miscarriage group N = 17	P-value
Euploid	75.0% (57/76)	58.8% (10/17)	0.296
Aneuploid	14.5% (11/76)	23.5% (4/17)	0.580
Mosaicism	10.5% (8/76)	17.6% (3/17)	0.684
≤ 50%	7.9% (6/76)	11.8% (2/17)	0.971
> 50%	2.6% (2/76)	5.9% (1/17)	1.000
Theoretically transferable			0.411
Yes	82.9% (63/76)	70.6% (12/17)
No	17.1% (13/76)	29.4% (5/17)

In the live birth group, among the embryos with detected chromosomal abnormalities (including aneuploidy or mosaicism), only 5.3% (1/19) showed a single whole-chromosomal abnormality, and all of them were monosomic (Table 3). Moreover, 57.9% (11/19) of cases were observed with a single segmental chromosomal abnormality, and the single segmental duplication rate (47.4%, 9/19) was higher than the single segmental deletion rate (10.5%, 2/19) (Table 3). In the miscarriage group, 57.1% (4/7) of the embryos with detected chromosomal abnormalities were a single whole-chromosomal abnormality, and the trisomy and monosomy rates were 42.9% (3/7) and 14.3% (1/7), respectively. Only two embryos with a single segmental chromosome were detected in the miscarriage group: one segmental duplication (14.3%, 1/7) and one segmental deletion (14.3%, 1/7). All six embryos with double chromosomal abnormalities were observed in the live birth group (all were double-segmental chromosomal abnormalities) (31.5% vs. 0%). One embryo, each with complex abnormalities, was detected in the live birth and miscarriage groups (5.3% vs. 14.3%). Owing to the extremely small sample size, we did not compare the rates of various chromosomal abnormalities between groups.

Table 3.

Detailed information of embryos diagnosed with chromosomal abnormalities (including mosaicism and aneuploidy)

	Live birth group N = 19	Miscarriage group N = 7
Single whole-chromosomal abnormality	5.3% (1/19)	57.1% (4/7)
Trisomy	0% (0/19)	42.9% (3/7)
Monosomy	5.3% (1/19)	14.3% (1/7)
Single segmental chromosomal abnormality	57.9% (11/19)	28.6% (2/7)
Segmental deletion	10.5% (2/19)	14.3% (1/7)
Segmental duplication	47.4% (9/19)	14.3% (1/7)
Double chromosomal abnormalitya	31.5% (6/19)c	0% (0/7)
Complex chromosomal abnormalityb	5.3% (1/19)	14.3% (1/7)

^aDouble chromosomal abnormality represents the detection of two chromosomal abnormalities (regardless of mosaic or aneuploidy)

^bComplex chromosomal abnormality represents the detection of three or more chromosomal abnormalities (regardless of mosaic or aneuploidy)

^cAll six embryos were diagnosed with double-segmental chromosomal abnormalities

Prenatal diagnosis, karyotyping of the newborns, and aneuploidy detection of the miscarriage tissue

In the live birth group, prenatal diagnosis or karyotype detection of the newborns was performed in 28 cases, and all were euploid. At the same time, aneuploidy screening on the corresponding embryonic MDA samples showed 25 euploids and three aneuploids (Online Resource 3). In the miscarriage group, karyotype analysis was performed for six abortuses. The aneuploidy screening results on the embryonic MDA samples of three cases were normal, consistent with the karyotype of abortuses. However, the prenatal diagnosis results of the other three cases did not match the results of the embryos: one was trisomy 22 in the embryo, while the karyotype of aborted tissues was 47, XN, + (mosaic) (22) (66%); two cases showed aneuploidy mosaicism, 46, XN, + (mosaic) (19) (30%) and 46, XN, − (mosaic) (6) (44%) in the embryo, while the karyotypes of these abortuses were normal (Online Resource 3).

The concordance rate of the aneuploidy screening of the embryonic MDA products with prenatal diagnosis, karyotyping of the newborns, and aneuploidy detection of miscarriage tissue (i.e., aneuploidy detection of conception products) was 100% (28/28) in euploid embryos and 0% (0/6) in embryos with chromosomal abnormalities (including mosaicism or aneuploidy).

Discussion

To evaluate the clinical efficacy of PGT-A without compromising the interests of patients, we performed NGS-based aneuploidy screening of the stored MDA products of embryos with known embryo transfer outcomes. Our study suggests that the application of NGS-based PGT-A remains questionable, as it may result in at least one in six embryos with reproductive potential to be discarded and prevent miscarriage in less than one in three embryos.

At present, whether NGS-based PGT-A can improve pregnancy outcomes remains controversial. In recent years, many studies have supported the use of NGS for PGT-A to significantly improve pregnancy outcomes compared with SNP array or ACGH [31–33]. In addition, Maxwell et al. [23] used NGS to reanalyze embryonic biopsy samples detected as euploidy by aCGH and found that 31.6% of these embryos were mosaic and 5.2% were polyploid. These results supported that NGS was more sensitive than other comprehensive chromosome screening (CCS) platforms in detecting chromosomal abnormalities and could eliminate more embryos with high miscarriage risk to improve pregnancy outcomes. However, an RCT performed by Ozgur et al. [34] showed no significant difference in the live birth rate between the transfer of a single best euploid blastocyst selected by NGS and the transfer of a single best morphological blastocyst in young infertile patients. Besides, in a study based on the oocyte donation cycles, compared with the non-PGT-A cycles, the live birth rate of the embryo transfer cycle of the NGS-based PGT-A cycles had no significant improvement (51.2% vs. 47.2, P = 0.42) [35]. These suggested the imperfect predictive value of NGS-based PGT-A in embryo transfer outcomes, which may be due to misdiagnosis caused by PGT-A procedures (such as TE biopsy, WGA, detection platform) or the embryonic self-correction mechanism [11, 36].

Tiegs et al. [29] performed a prospective, blinded, non-selection study on 484 blastocysts, in which the NGS-based PGT-A results of biopsied embryos were not revealed until the outcomes of the transfers were known. The authors reported a 0% aneuploid call clinical error rate, supporting that NGS-based PGT-A would not lead to the waste of reproductively competent embryos. However, we performed NGS-based aneuploidy screening on cryopreserved MDA products of 76 embryos, which resulted in live birth after transferring into the uterus, showing that if these patients had PGT-A during their PGT-M cycles, 17.1% of the embryos would have been wasted owing to being diagnosed as untransferable.

It is worth noting although the same NGS platform (Nextseq 550) was used in both studies, our findings contradict those of Tiegs et al. [29]. The apparent conflicts between these two studies could be explained by several aspects as follows: (1) MDA was used in this study because the primary samples were already amplified by MDA using PGT-M and were not accessible. However, MDA is not the ideal choice for PGT-A. In addition, the embryo MDA products used in our study have been cryopreserved for a long time (the interval between TE biopsy to NGS detection was about 2–3 years) and might have incurred DNA degradation. Therefore, the accuracy of PGT-A results might have been compromised to a certain extent. (2) The extremely limited sample size of our study might have magnified the potential sampling error. (3) The classification criteria used in our study differ from those used by Tiegs et al. [29]—they classified “neither euploidy nor whole-chromosome aneuploidy” as “fragment abnormality,” including “whole-chromosome mosaic or segmental PGT-A diagnoses,” and excluded such “fragment anomalies” from the analysis because of their low incidence. However, these “fragment anomalies” would include some embryos with live birth potential; therefore, the study of Tiegs et al. [29] might not be adequately powered to generate conclusions. In contrast, the results of our study include the effects of segmental aneuploidy and mosaicism. (4) The age distribution and clinical indications of the two studies are different—the study of Tiegs et al. included infertile patients who sought IVF/PGT-A treatment, and 54.5% of the women were of advanced age. In contrast, the clinical indications of our study were PGT-M, and 83.9% (78/93) women included were young. The baseline risk of pregnancy failure may be inherently different in the two study populations.

Furthermore, wasting such a large proportion of embryos deserves our vigilance because it may result in some patients with no transferable embryos and reduce the benefit of PGT-A in improving pregnancy outcomes. Deng et al. [37] found that NGS-based PGT-A could not improve the live birth rate of patients with poor ovarian reserve. As NGS-based PGT-A is still an expensive and invasive procedure whose predictive value on embryonic reproductive potential remains unclear, its routine application to all patients undergoing IVF may be unreasonable. More high-quality studies are still warranted to verify the cost-effectiveness of NGS-based PGT-A. Moreover, it should be performed in patients with suitable indications and sufficient genetic counseling.

Our data suggest that chromosomal abnormalities in embryos account for less than one-third of miscarriages (29.4%). Moreover, the untransferable rate was similar between the groups, suggesting the limitation of NGS-based PGT-A in preventing miscarriages. However, the sample size of this study was too small to draw a firm conclusion; larger prospective studies are warranted to further evaluate the efficiency of NGS-based PGT-A. Other known causes for miscarriage include advanced maternal age, hormones, metabolic factors, autoimmune diseases, abnormal uterine anatomy, infection, male-associated factors, and lifestyle variables [38]. Encouragingly, genetic testing of miscarriage tissues and evidence-based evaluation of RPL can highlight the definite cause of more than 90% of miscarriages [39]. Therefore, clinicians should help patients rule out other possible causes of miscarriage to increase the value of PGT-A and avoid possible waste of time and cost for patients.

The concordance rate of the results of aneuploidy screening of embryonic MDA products with the aneuploidy detection of conception products was 100% (28/28) in euploid embryos, whereas it was 0% (0/6) in embryos with chromosomal abnormalities (mosaicism or aneuploidy). These data suggest that euploidy results seem to be more reliable, whereas the reliability of the results of segmental chromosomal abnormalities and mosaicism is questionable. Victor et al. [40] dissected the embryos diagnosed as aneuploid by NGS-PGT-A and found a 96.8% agreement between the chromosome karyotypes of the ICM and TE; when embryos with segmental chromosomal abnormalities were analyzed separately, the concordance rate declined to 42.9%. Sachdev et al. [41] observed that the concordance rate between TE and ICM was over 97% in euploid and aneuploid embryos diagnosed using NGS-PGT-A; however, it was only 35.2% in mosaic ones, possibly owing to the uneven distribution of mitotic non-disjunction events in the mosaic blastocyst. This warrants further investigations as it may have an important effect on the interpretation of PGT-A results.

In this study, we observed that embryos with different chromosomal abnormalities might have different reproductive potentials. For example, in the live birth group, there were more embryos with single segmental chromosomal abnormalities than embryos with single whole-chromosomal abnormalities (57.9% vs. 5.3%), whereas, in the miscarriage group, the opposite (28.6% vs. 57.1%) was observed. In addition, in the live birth group, the percentage of single segmental duplication in embryos with chromosomal abnormalities was greater than that of single segmental deletion (81.8% vs. 18.2%), whereas previous studies have shown a similar percentage of segmental deletion or duplication in blastocysts regardless of aCGH or NGS detection (45, 46). Unfortunately, due to the limited sample size, we could not perform statistical analysis to conclude whether the differences between the groups were significant. Moreover, the differences observed could be artifacts caused by the WGA procedure as the prevalence of segmental imbalances in this study is higher compared to the study of Girardi et al. [42] (5.6%), which used a larger sample size. Therefore, these interesting observations should be confirmed in larger prospective studies, as it is important to increase the chance of pregnancy for patients who opt for transferring embryos that are diagnosed as untransferable.

It should be noted that the samples used in our study are from patients carrying thalassemia mutations with no history of RPL, and 83.87% of the women were < 35 years old. Therefore, caution should be exercised in extrapolating the results of this study to other populations such as the patients with advanced maternal age (AMA), RPL, or recurrent implantation failure (RIF), as the clinical value of PGT-A is strongly associated with the clinical conditions of patients. However, from the data shown in Online Resource 4 that summarizes the conclusions of PGT-A2.0 on different patients, such as young patients, patients with AMA, RPL, or RIF, we can realize that within a single FET cycle, the clinical benefit of PGT-A 2.0 in young and well-prognosis patients remains controversial, whereas the clinical benefit of PGT-A 2.0 in patients with AMA, RPL, and RIF has been repeatedly demonstrated. At present, it is elusive whether embryos with chromosomal abnormalities are more likely to self-correct in young or good-prognostic populations than in those with other PGT-A indications, such as patients with AMA and RPL, and whether this is one of the reasons for the less significant clinical benefit of PGT-A. These interesting questions deserve to be confirmed in the future. This study provides a more ethical methodological reference for future research: aneuploidy testing of cryopreserved MDA products of embryos with known FET outcomes in these populations.

The advantage of the current study is that we evaluated the clinical value of PGT-A without additional biopsy and caused no harm to the interests of the patients. However, the biggest limitation of this study is that due to the extremely small sample size, especially for the miscarriage group, it is impossible to draw strong conclusions. In addition, in our study, the no result rate was 26.2% (27/103) and 19.0% (4/21) in the live birth group and miscarriage group, respectively, which was significantly higher than the 4–14% no result rate reported by ESHRE PGT Consortium [43]. This may be due to the bias caused by the small sample size in this study or the DNA degradation caused by long-term sample storage, which also threatens the reliability of our conclusions. Besides, due to the small sample size, when analyzing the reproductive potential of embryos with different chromosomal abnormalities, we used all embryos with chromosomal abnormalities as a whole, instead of separately analyzing the reproductive potential in aneuploid and mosaic embryos, which might have affected the accuracy of the results. Furthermore, because only a small number of patients underwent prenatal diagnosis, neonatal karyotype examinations, or genetic testing of miscarriage tissues, we could not verify the clinical accuracy of NGS-based PGT-A. Moreover, due to the limited sample size, we could not confirm whether abnormalities on different chromosomes had different effects on the pregnancy outcomes. Finally, our conclusions are based on the MDA + NGS platform; therefore, it is not appropriate to extrapolate our findings to PGT-A 2.0 based on other CCS platforms such as SNP array and aCGH.

Conclusions

In the present study, we did not observe a clear clinical benefit of NGS-based PGT-A, as it could cause at least one in six embryos with reproductive potential to be discarded and only prevent miscarriage in less than one in three embryos. The data presented here, together with other investigations [34], and the damages done to the embryo during the biopsy, showed that PGT-A might cause more harm than benefit for some patients opting for IVF. However, whether this holds for the most recent PGT-A assay, as well as for a large cohort prospective randomized study, remains to be investigated.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

X.T.S. and D.J.C. contributed to the interpretation of data and drafted the article. C.H.D., Y.X., Y.F., and B.C. were responsible for the analysis of data and revising the manuscript critically for important intellectual content. Y.L.W., J.W., R.L., J.G., J.F.P., H.Z., and Y.H.Z. participated in the acquisition of data. C.Q.Z. contributed to the conception and design and critical review of the article. All authors have approved the final version of the manuscript to be published.

Declarations

Competing interests

The authors declare no competing interests.