Temporal Evaluation of a Minimally Invasive Method of Preimplantation Genetic Testing for Aneuploidy (mi-PGT-A) in Human Embryos

Abstract

Preimplantation genetic testing for aneuploidy (PGT-A) has become a useful approach for embryo selection following IVF and ICSI. However, the biopsy process associated with PGT-A is expensive, prone to errors in embryo ploidy determination, and potentially damaging, impacting competence and implantation potential. Therefore, a less invasive method of PGT-A would be desirable and more cost-effective. Noninvasive methods for PGT-A (ni-PGT-A) have been well-studied but present limitations in terms of cf-DNA origin and diagnostic accuracy. Minimally invasive pre-implantation genetic testing (mi-PGT-A) for frozen-thawed embryo transfer is a promising, less studied approach that utilizes a combination of spent culture media (SCM) and blastocoelic fluid (BF)-derived cell-free (CF)-DNA for genetic testing. This study aimed to optimize the effectiveness of mi-PGT-A for aneuploidy diagnosis by investigating the optimal temporal sequence for this protocol. SCM+BF was collected at either 48 or 72 h of culture after thawing day 3 preimplantation embryos. cf-DNA in the SCM+BF was amplified, analyzed by next-generation sequencing (NGS) and compared with results from the corresponding whole embryos (WEs) obtained from human embryos donated for research. Fifty-three (42 expanded blastocysts, 9 early blastocysts, and 2 morula) WE and SCM+BF samples were analyzed and compared. The overall concordance rate between SCM+BF and WE was 60%. Gender and ploidy concordance improved with extended culture time from 48 h (73% and 45%) to 72 h (100% and 64%), respectively. These results demonstrate that SCM+BF-derived cf-DNA can be successfully used for mi-PGT-A. Our findings indicate that longer embryo culture time prior to SCM+BF-derived cf-DNA analysis improves DNA detection rate and concordance with WEs and decreases the proportion of false positive results.

1. Introduction

Preimplantation genetic testing (PGT) is an important clinical tool for the diagnosis of monogenic disorders and treatment of infertility [1]. Aneuploidy rate increases exponentially with maternal age [2,3] and is a leading cause of pregnancy failure, miscarriage, and congenital abnormalities following natural and artificial reproductive technology (ART)-assisted pregnancies [4,5,6]. With rising global rates of female infertility [7] and uniform aneuploidies appearing in approximately half of IVF-generated embryos (primarily through du novo processes) [8,9], PGT for aneuploidy (PGT-A) detection using next-generation sequencing (NGS) is more frequently being recommended for women of advanced reproductive age as a diagnostic tool to preferentially select euploid embryo(s) for transfer [2].

Current PGT-A procedures involve aspirating 5–10 trophectoderm (TE) cells from the blastocyst-stage embryo followed by DNA extraction, whole genome amplification (WGA), and ploidy determination using NGS of the embryonic genome [10]. The controversial clinical utility of PGT-A stems from concerns regarding diagnostic accuracy, harms associated with embryo biopsy, and costs of the procedure. PGT-A relies on the genetic constitution of the biopsied cells being consistent with the whole embryo, meaning mosaicism may lead to an incorrect classification of embryo ploidy. Rates of embryonic mosaicism vary widely between studies and have been linked to many factors such as age and cellular proliferation rates [11,12].

TE biopsy is an invasive procedure, raising concerns that removing growing cells may impede embryo development and compromise neonatal and long-term outcomes [10,13]. Current meta-analyses investigating PGT-A safety demonstrate conflicting results [14,15,16,17]. Some suggest that compared to non-biopsied embryos, TE biopsies may increase the risk of obstetric, neonatal, and child health complications, including low birthweight and small for gestational age babies [17], and increased odds of preterm birth [17] and hypertensive disorders [14,15,17], while others report no associations [16] or that TE biopsies lowered the risk of very preterm birth and low birthweight [14,15]. The most recent randomized control trials (RCTs) evaluating PGT-A as a means of improving pregnancy outcomes also demonstrate mixed results [10,18,19,20,21]. The proposed benefits of PGT-A, including lower miscarriage rates and higher rates of implantation, ongoing pregnancy, and live births, are predominantly seen in older women (35+ years old) with higher baseline rates of aneuploidy [18,22,23] or recurrent pregnancy failure [24]. In contrast, PGT-A does not appear to improve pregnancy outcomes per embryo transfer or per intention to treat in younger (<35 years old), infertile patients [18,25], which may be due to the higher rates of mosaicism reported in this population [11]. The largest multicenter RCT to date, which included 1212 women (20–37 years old), found that PGT-A via NGS did not improve cumulative live birth rate (LBR) but did significantly lower cumulative clinical pregnancy loss [20]. Furthermore, the invasive nature of the biopsy requires time-consuming embryo manipulation by experienced embryologists, making this a costly procedure for patients. Costs will vary depending on the number of embryos being biopsied and the company that performs the genomic analysis.

Together, these undesirable technical factors and patient costs, both emotionally and financially, have created an interest in less invasive options for chromosome screening in preimplantation embryos. A reliable, less invasive method would increase the safety of the procedure and be a more affordable option for patients.

The recent discovery of cell-free DNA (cf-DNA) in spent human IVF culture media (SCM) [26] and human blastocoelic fluid (BF) [27] presents an exciting avenue for exploring PGT-A techniques that require minimal specialized training and impose negligible risk to the embryo [28]. SCM-derived cf-DNA permits noninvasive PGT-A (ni-PGTA) of embryos and shows promise for improving pregnancy outcomes compared to traditional morphology assessments [29,30,31], with two double-blinded randomized control trials (RCTs) currently underway [28,32]. However, there are numerous technical factors that require optimization prior to the clinical application of ni-PGT-A.

One of the greatest barriers to the clinical implementation of ni-PGT-A is the ability to produce reliable and accurate results from SCM-derived cf-DNA that are representative of whole embryo ploidy [13]. General and complete genetic concordance rates between the embryo and PGT-A result using SCM- or BF-derived cf-DNA varies greatly between studies, ranging from 33.3% to 89.1% for SCM and 37% to 97% for BF [33]. Alternatively, a less studied approach, referred to as minimally invasive PGT-A (mi-PGT-A), involves the combined analysis of SCM and BF, which has been shown to increase cf-DNA quality and quantity compared to SCM or BF alone, improving the power for analysis [13,34]. mi-PGT-A with SCM+BF-derived cf-DNA has yielded promising concordance rates of up to 97.8% with TE biopsies [35] but requires further investigation to determine feasibility. Given that cf-DNA in the SCM and BF may originate from cells of both the TE and inner cell mass (ICM), certain groups have proposed that ni/mi-PGT-A may be a more accurate representation of embryo ploidy and less prone to errors associated with mosaicism compared to TE biopsy [34,36,37]. Another source of discrepancy across studies is the lack of standardized approaches to cf-DNA collection and amplification for ni/mi-PGT-A. Embryonic cf-DNA in the SCM can be fragmented and present in low concentrations (around 8%), requiring technical changes in the protocol to support reliable extraction [35]. A key determinant of optimal cf-DNA extraction from SCM is embryo culture time, which has been generally established as D6 for fresh embryos [33]. However, there is a lack of research to support an optimal culture time for cf-DNA collection from frozen-thawed embryos, particularly for mi-PGT-A procedures [33].

The objective of this study was to evaluate the effectiveness of mi-PGT-A using SCM+BF-derived cf-DNA compared to whole embryos obtained from frozen-thawed supernumerary human embryos donated for research. Our aim was to determine the optimal temporal sequence for sufficient DNA recovery to permit conclusive ploidy testing using SCM+BF from frozen-thawed embryos. We also sought to determine whether SCM+BF-derived cf-DNA can be reliably amplified employing Multiple Annealing and Looping Based Amplification Cycles (MALBAC; Yikon Genomics, Suzhou, Jiangsu, China) WGA technology. Finally, we aimed to determine whether genetic results from SCM+BF are concordant with the genetic composition of the whole embryo.

2. Materials and Methods

2.1. Participant Enrollment and Ethics

This study received approval from the Hamilton Integrated Research and Ethics Board of McMaster University before initiation (HiREB Protocol #7528). Seventy-eight eligible individuals/couples (mean age: 33 years old; range: 20–44 years old) with a total of 268 donated research embryos frozen on day 3 were contacted to obtain project-specific consent for the study. Sixty-two individuals/couples consented to participate in the study. Six declined study participation citing objection to the nature of the study (3/6), spiritual/religious reasons (1/6), or wishing to change the disposition of their surplus embryos (2/6). Seven couples did not return their consents after multiple follow-up attempts, 2 couples did not respond to the clinic’s request for communication, and 1 patient could not be reached due to outdated contact information. Two individuals/couples who consented to the study but used donor gametes to create embryos were excluded prior to embryo thaw due to changes in Canadian donor ova and sperm research guidelines.

2.2. Embryo Selection, Thaw, and Culture

Our clinic’s research embryo database was queried for donated research embryos frozen on day 3. Consent to use these embryos for research was given by both gamete providers (where applicable) at a previous discussion regarding disposition of the surplus embryos; project-specific informed consent was subsequently obtained following HiREB approval.

After receiving project-specific consent for research, a total of 214 day 3 donated research embryos were thawed according to the manufacturer’s recommended instructions for slow cooled (GyneMed, Sierksdorf, Germany) embryos in conjunction with their respective freezing protocol. Thawed embryos were rinsed 3 times in pre-equilibrated Global Total media (LifeGlobal, Cooper, Trumbull, CT, USA) before being placed individually into a 20 uL drop (Global Total) in uGPS dishes (LifeGlobal, Cooper, Trumbull, CT, USA) and covered with mineral oil (LiteOIL, Cooper, Trumbull, CT, USA). Embryos were cultured in a humidified tri-gas incubator (Panasonic, Newark, NJ, USA) set to 5.5% C0₂ and 5.0% O₂ to achieve an optimal media pH = 7.3 ± 0.05. Embryos were inspected to ensure that there were no overt residual cumulus cells attached to the zona pellucida. After 48 h of culture, embryos were assessed for blastocyst development. Embryos not expanded sufficiently were cultured and assessed the next day (72 h culture). This generated 42 expanded blastocysts (Gardner scoring scale “3” or higher). These were analyzed along with 9 early blastocyst embryos and 2 morulae to determine if there was viable DNA in these samples. As a result, 53 WE and SCM+BF samples were analyzed and compared. Embryos that did not result in a blastocyst were discarded as per standard operation procedures for embryo disposal.

2.3. DNA Collection from WE and SCM+BF

Adequately expanded blastocyst(s) (“3” or higher based on the Gardner scoring scale) were collapsed by using a Saturn laser (Somagen, Edmonton, AB, Canada). A single laser pulse (between 300 and 400 ms) at a cell junction in an area of trophectoderm cells away from the ICM was used to breach the zona pellucida. Embryos were left in the drop for 20–25 min to allow the blastocoelic fluid to drain from the blastocyst before collecting the whole embryo (WE) and SCM+BF. Collapsed WEs were removed (Figure 1) with a unique pipette and rinsed through 3 successive drops of PBS (Cooper, USA) before being placed into a 0.2 mL PCR tube containing 5 uL of lysis buffer solution (Yikon, Suzhou, Jiangsu, China—cat#: XK-043). Using a separate unique transfer pipette, each corresponding SCM+BF was harvested from the micro drop with an elongated pipette tip (Mandel, Guelph, ON, Canada). Media was aspirated up and down carefully 3 times to avoid oil contamination before transferring it into a 0.2 mL PCR tube with 5 uL Lysis Buffer (Figure 1). PCR tubes were labeled with a unique ID number from a random number table. Samples were stored at −20 °C according to the manufacturer’s recommendations until shipping for analysis.

2.4. DNA Analysis

All samples were analyzed at LifeLabs Genetics Laboratory (Toronto, ON, Canada). Samples were run alongside 14 control wells including 2 blank control wells to ensure no artifacts were introduced during sample analysis, 2 uncultured and 2 cultured media control wells to create a baseline relative to culture conditions, and 8 wells with known levels of DNA to ensure assay sensitivity and accuracy.

Yikon’s ChromInst™ assay workflow was used to screen chromosomal aneuploidies in WE samples. For the SCM+BF samples, the NICSInst™ assay protocol was used. Whole genome amplification of individual cells was achieved through Multiple Annealing and Looping Based Amplification Cycles (MALBAC) technology, which is specifically designed to amplify small amounts of genomic DNA with reduced amplification bias. Amplified DNA libraries were pooled and sequenced using next-generation sequencing technology (Illumina, San Diego, CA, USA). Sequences were analyzed with ChromGo™ analysis pipeline (Yikon, Suzhou, Jiangsu, China), and copy number evaluation was performed on a chromosome arm and deletion level. ID codes were then broken to compare genetic analysis results between WE and SCM+BF samples.

2.5. Statistical Analysis:

A sample size calculation was performed to determine noninferiority of SCM+BF testing as compared to the gold standard of whole embryo testing. A total of 40 expanded blastocysts were required for testing to be 95% certain that the upper limit of a one-sided 95% confidence interval (or equivalently a 90% two-sided confidence interval) would exclude a difference in favor of the standard group of more than 25%. Our sample size included 40 blastocysts for testing as well as an additional 2 samples to account for any challenges in the testing process. Based on an estimated blastulation rate of 20%, a minimum of 210 donated research embryos frozen on day 3 were sought out for this study.

3. Results

3.1. DNA Recovery and Amplification

The DNA detection rate was 100% (52/52) for whole embryos and 88% (46/52) for all SCM+BF samples (Table 1). When further analyzed by culture time post thaw, the DNA detection rate for SCM+BF samples was 69% (11/16) for those collected at 48-h post thaw and 97% (35/36) for samples collected at 72-h post thaw (Table 1). DNA was detected in 86% (18/21) of our SCM+BF samples collected from low-quality embryos (possessing a C grade) at 72-h post-thaw but could not be detected in 100% (3/3) of SCM+BF samples collected from low-quality embryos at 48-h post-thaw (Table 1). DNA was detected in 94% (15/16) of SCM+BF samples collected from high-quality embryos at 72-h post-thaw and 83% (10/12) in SCM+BF samples collected from high-quality embryos at 48-h post-thaw (Table 1).

3.2. Sex Concordance

Excluding SCM+BF samples with suspected maternal cell contamination and low-level mosaic results, sex concordance between all SCM+BF-derived cf-DNA analyzed using mi-PGT-A and WEs was 73% (8/11) for SCM+BF samples collected at 48-h post-thaw and 100% (35/35) for SCM+BF samples collected at 72-h post thaw (Table 2).

3.3. Ploidy Concordance

General concordance was defined as a concordant normal (euploid) or abnormal (aneuploid) result in both the SCM+BF and WE samples, whereas full concordance was defined as an entirely concordant genetic alteration result in both the SCM+BF and WE samples with respect to the specific number and type of genetic alteration(s) observed. For example, a 48,XX,+X(×2),−Y(×0) SCM+BF result with a corresponding 45,XY,−22(×1) WE result have general concordance, as they would both be reported as abnormal (aneuploid), but they do not have full concordance, as they demonstrate subtle different genetic alterations. The former SCM+BF mi-PGT-A result represents a tetrasomy X, as there are two additional X chromosomes [+X(×2)], whereas the latter result for the WE indicate a monosomy 22, where one copy of chromosome 22 is missing [−22(×1)].

Excluding SCM+BF samples with suspected maternal cell contamination and low-level mosaic results, the general concordance rate between all SCM+BF-derived cf-DNA analyzed using mi-PGT-A and WE was 60% (28/46) (Table 2). Comparing SCM+BF-derived cf-DNA samples collected at 48-h post-thaw to those collected at 72-h post-thaw, there was an improvement in the general concordance from 45% (5/11) to 66% (23/35) (Table 2). However, the full concordance rate did not change in samples collected at 48-h post-thaw (45%; 5/11) compared to those collected at 72-h post-thaw (46%; 16/35).

3.4. mi-PGT-A Diagnostic Performance

A false-positive result was defined as an abnormal (aneuploid) SCM+BF result with a normal (euploid) corresponding WE whereas a false-negative result represents a normal (euploid) SCM+BF result with an abnormal (aneuploid) corresponding WE. A low-level mosaic result represents 20–40% mosaicism in the SCM+BF sample with a discordant (euploid or aneuploid) WE result. Maternal cell contamination was suspected when there was a mosaic dropout of the Y chromosome in the SCM+BF sample but a normal WE result.

Excluding SCM+BF samples with suspected maternal cell contamination and low-level mosaic results, mi-PGT-A yielded a false-positive rate of 21.7% (10/46) and false-negative rate of 6.5% (3/46) (Table 2). Comparing mi-PGT-A results between samples collected at 48-h post-thaw to those collected at 72-h post-thaw, there was a decrease in the number of false positives from 36.3% (4/11) to 17.1% (6/35), and false negatives from 9% (1/11) to 5.7% (2/35) (Table 2). Fewer samples cultured to 72-h post-thaw had low-level mosaicism 6% (2/35) compared to those cultured for 48 h 9% (1/11) (Table 2).

In calculating mi-PGT-A diagnostic performance, the SCM+BF samples collected at 72 h that had suspected maternal cell contamination and low-level mosaicism (four total), were included as false positives, whereas the one low-level mosaicism SCM+BF sample collected at 48 h was found to be a true abnormal result (Table 3). Including samples collected at 72-h post-thaw with low-level mosaicism and suspected maternal cell contamination, the false-positive rate for our mi-PGT-A test changed to 28.5% (10/35), which is still lower than that for samples collected at 48-h post-thaw (36%; 4/11) (Table 3). The positive predictive value (PPV) for SCM+BF samples collected at 48-h and 72-h post-thaw was 71.4% (10/14) and 76.7% (22/43), respectively (Table 3). There was a greater difference in the negative predictive value (NPV) between SCM+BF samples collected at 48-h post-thaw (87.5%; 7/8) and those collected at 72-h post-thaw (92.6%; 25/27). Test sensitivity increased from 90.9% (10/11) in SCM+BF samples collected at 48-h post-thaw to 94.3% (33/35) in SCM+BF samples collected at 72-h post-thaw. Lastly, mi-PGT-A specificity increased from 63.6% (7/11) in SCM+BF samples collected at 48-h post-thaw to 71.4% (25/35) for SCM+BF samples collected at 72-h post-thaw.

4. Discussion

Since the discovery of cf-DNA in both SCM and BF, at least 20 studies seeking to optimize ni-PGT-A have been published [13], with two ongoing multicentered RCTs, including a cumulative 1648 couples [30,34]. However, sample sizes (7–70 couples receiving ART), methodology, DNA detection rates (6.7–100%), and concordance rates (5.9–100%) have varied widely across completed studies [13]. Despite the promise of mi-PGT-A, only five studies seeking to optimize this protocol have been published [34,35,36,38,39], emphasizing the need for further research into this less invasive method.

One of the leading technologies employed for WGA of very small amounts of cf-DNA in SCM samples is MALBAC, which requires very little template DNA and allows for increased genome coverage and lower rates of allele drop-out and amplification bias compared to conventional DNA amplification methods, making it an ideal technology for the amplification of SCM-derived cf-DNA [40]. Xu et al. (2016) reported a 100% DNA detection rate when using MALBAC followed by NGS on SCM samples from 42 vitrified embryos cultured from D3 to D5 [27], whereas Liu et al. (2017) achieved a 91% DNA detection rate with MALBAC and NGS applied to 88 SCM samples [41]. Although most studies demonstrate greater DNA detection rates using MALBAC and NGS on SCM-derived cf-DNA (90–100% WGA) compared to SurePlex and NGS on BF-derived cf-DNA (34.8–82% WGA), only ∼8% of DNA in SCM is embryonic in origin [42,43], potentially confounding the analysis of cf-DNA from SCM. Alternatively, collecting cf-DNA from SCM+BF for mi-PGT-A has been shown to increase DNA quality and quantity compared to SCM or BF alone [34]. However, there is substantially less research optimizing mi-PGT-A [34,35,36,39] compared to ni-PGT-A. Li et al. (2017) was one of the few studies that used MALBAC and NGS for mi-PGT-A, achieving a 97% DNA detection rate when applied to 40 SCM+BF samples [36]. To maximize the amount of assayable DNA and power for analysis [13], we chose to perform mi-PGT-A using MALBAC and NGS on SCM+BF-derived cf-DNA samples. Although TE biopsy is the standard clinical practice for PGT-A, we chose to evaluate mi-PGT-A results in reference to WEs because this better accounted for chimeric embryos and is more representative of the inner cell mass ploidy [44].

4.1. Influence of Culture Time on DNA Detection Rates

The present study identified several advantages to a longer post-thaw embryo culture time prior to SCM+BF-derived cf-DNA collection and testing. DNA detection rates were higher in SCM+BF samples collected at 72 h (97.2%; 35/36) compared to those collected at 48-h post-thaw (68.7%; 11/16). Similarly, Kuznyetsov et al. (2018) achieved an amplification rate of 100% (28/28) from SCM+BF-derived cf-DNA from D5 blastocysts cultured for 24-h post-thaw, but SCM+BF-derived cf-DNA from D5 blastocysts cultured for 8-h post-thaw was not sufficient to produce conclusive NGS results following WGA [34]. We postulate that higher DNA detection rates at 72-h post-thaw may be due to the presence of fewer maternal cells and higher amounts of cf-DNA being extruded by the later-stage, metabolically active embryo. Additionally, ni-PGT-A seems to work better in cases of freeze-thaw embryos. Hu et al. (2023) report that amplification rates increased from 76.6% to 100% using MALBAC on SCM samples from thawed embryos, which further improved among embryos cultured to D6/7 compared to D5 [45].

Compared to SCM+BF samples collected from lower-quality embryos (possessing a C grade on the Gardner Scale), DNA detection rates were higher in SCM+BF samples collected from higher-quality embryos at both 48- (84.6%; 11/13) and 72-h (100%; 16/16) post-thaw. DNA was detected in 95% (19/20) of SCM+BF samples collected from lower-quality embryos at 72-h post-thaw but could not be detected in 3/3 SCM+BF samples collected from lower-quality embryos at 48-h post-thaw. Although some suggest that blastocyst quality does not affect cf-DNA detection in SCM [35,39], more advanced blastocysts of higher morphological grade (more cells and fully expanded) have higher amounts of cf-DNA in the blastocoel cavity [44,46,47], which may explain lower DNA detection rates in the SCM+BF samples from our lower-quality blastocysts. Furthermore, the trend that cf-DNA detection rates were lower in lower-quality blastocysts provides support for hypotheses related to nonapoptotic mechanisms driving cf-DNA release within BF and SCM [48]. The origins and role of embryonic cf-DNA is a point of controversy among researchers seeking to determine the clinical utility of both ni-PGT-A and mi-PGT-A [48,49]. One major concern is that cf-DNA release is primarily driven by apoptosis and that human embryos self-correct by eliminating aneuploid cells, cellular debris, and fragments, meaning that aneuploid cells could be overrepresented in the SCM, which may not reflect the true ploidy status of the embryo [48,49]. If this were true, lower-quality embryos, which inherently have higher degrees of apoptosis, would be expected to yield greater DNA detection rates. Although this has been demonstrated by some groups using SCM for ni-PGT-A [45], our findings and others achieving high concordance rates using SCM+BF-derived cf-DNA from lower-quality embryos [35] support alternative mechanisms. However, our small sample size of low-quality embryos limits the conclusions that can be drawn about the impact of embryo quality on DNA detection in the present study.

Certain studies have suggested that culture conditions, such as the choice of culture media, may impact SCM-derived cf-DNA amplification rates because added supplements including salts, bicarbonate, glucose, pyruvate, amino acids, vitamins, and proteins may influence amplification success [50,51]. One study investigating mt-DNA in the SCM found very low levels of DNA contamination in negative controls, which the authors suggest could have originated from the protein supplement used [49]. However, a recent review of multiple studies comparing cf-DNA amplification across different commercial media choices found very little variation in amplification success [50]. In the present study, the culture media was identical in both our experimental conditions given that the aim of the study was to evaluate the influence of culture time on mi-PGT-A.

4.2. Influence of Culture Time on mi-PGT-A Concordance with WEs

Our temporal analysis revealed that longer embryo culture time post-thaw improved concordance rates between mi-PGT-A results and WE analysis. Sex concordance rates improved from 72.7% (8/11) in SCM+BF samples collected at 48-h post-thaw to 100% (35/35) in SCM+BF samples collected at 72-h post-thaw. Comparing all samples collected at 48-h post-thaw to those collected at 72-h post-thaw, there was an improvement in the general concordance from 45.5% (5/11) to 65.7% (23/35) with longer culture time. However, there was no difference in the full concordance between SCM+BF samples collected at 48-h post-thaw (45.4%; 5/11) and SCM+BF samples collected at 72-h post-thaw (45.7%; 16/35). This may be in part due to maternal cell DNA contamination, which was suspected in 5.7% (2/35) of the SCM+BF samples collected at 72-h post-thaw. Mitigating maternal cell DNA contamination from cumulus cells is critical for ensuring the clinical efficacy of mi-PGT-A and can be achieved through proper denudation, media changes, and serial washes of embryos prior to transfer into culture drops. Although we applied the same post-thaw washing procedure to both groups of embryos, it is possible that the suspected maternal contamination present in the 72-h group resulted from residual cumulus cells within the culture media. This was suspected due to a lack of sex concordance between the embryo and SCM+BF collected. Other methods for identifying maternal cell contamination may include whole-genome DNA methylation sequencing [52], droplet digital PCR (DD-PCR) [53], PCR to identify variable number tandem repeats (VNTR) [54], NGS to identify short tandem repeats (STR) markers [55], and quantitative parental contamination testing (qPCT) [54], but these may substantially increase the cost of ni/mi-PGT-A. To mitigate the risk of maternal DNA contamination, studies seeking to optimize ni-PGT-A have suggested performing five to six serial washes of embryos prior to transfer [56,57], whereas our study performed three washes post-thaw prior to transferring embryos to culture media. Rubio et al. (2020, 2019) outline additional measures for mitigating maternal cell DNA contamination, such as handling embryos individually using new capillaries [8,58], while Kuznyetsov et al. (2020) used fluorescently labeled short tandem repeat markers in SCM+BF-derived cf-DNA to confirm minimal maternal DNA contamination within samples, which could help avoid misdiagnoses using mi-PGT-A [35].

Although concordance rates appear to be better when using mi-PGT-A on frozen-thawed embryos [59], there is still significant variability between studies that may be related to technical differences in mi-PGT-A protocols. Studies reporting the highest general/clinical concordance rates (96–90%) between SCM+BF-derived cf-DNA from frozen frozen-thawed blastocysts and WEs involved thawing D5/D6 blastocysts and collecting SCM+BF samples at either 15-h [38] or 24-h [34] post-thaw. Instead of thawing D5/6 blastocysts, our mi-PGT-A protocol involved thawing D3 embryos and culturing till D5/6 before SCM+BF collection (Figure 1), achieving a lower general concordance rate (60.8%; 23/35). Zhang et al. (2019) is the only other study that collected SCM+BF-derived cf-DNA from D3-thawed embryos and reported similar concordance rates (66.7%; relative to WEs) to our study, suggesting that mi-PGT-A using SCM+BF-derived cf-DNA may be most effective on D5/6 frozen-thawed blastocysts [39]. However, Zhang et al. (2019) collected samples 24 h before blastocyst formation, which may have influenced cf-DNA availability [39]. We used D3 frozen embryos in the present study because this protocol is the most reflective of clinical frozen-thaw embryo transfers and D3 is typically the day that media switchover occurs in fresh-transfer cycles. However, recent research suggests that transferring the growing embryo to new media on day 4 may result in higher concordance rates [43,60]. Ultimately, the choice to freeze on either D3 or D5/6 often depends on the individual characteristics of the embryo’s development, the standard protocols of the fertility clinic, and the preferences or clinical circumstances of the patient. In terms of cf-DNA amplification and analysis, Kuznyestsov et al. (2018) and Jiao et al. (2019) also used SurePlex and a modified version of MALBAC for WGA prior to NGS, respectively [34,38], whereas our study and Zhang et al. (2019) used traditional MALBAC [39].

Our relatively low full and general concordance rates may also be explained by predominantly analyzing SCM+BF samples derived from higher-quality embryos (27/46 3BB or higher) in the present study. Using a similar mi-PGT-A protocol to our study, Zhang et al. (2019) reported a higher full concordance rate (66.7%) between WEs and SCM+BF-derived cf-DNA from predominantly low-quality, frozen-thawed blastocysts (1/27 embryos with a Gardener score > 3CC) [39]. Hu et al. (2023) found that the concordance rates between SCM-derived cf-DNA and TE biopsy from low-quality blastocysts (95.24%) was significantly higher than that in medium-quality blastocysts (67.90%) and high-quality blastocysts (51.85%) [45]. Although Kuznyetsov et al. (2020) reported that the overall concordance of mi-PGT-A results using SCM+BF (compared to TE biopsy) was not different between good 47/48 (97.9%) and moderate/low-quality blastocysts 41/42 (97.9%), their analysis did not include different culture times [35]. These researchers also collected SCM+BF from fresh embryos that were graded using a different system than our study (modified SMART [61]), and their reference point was TE-biopsied cells, which are less representative of ploidy status compared to the WE.

Several studies investigating ni-PGT-A using SCM-derived cf-DNA have suggested that longer culture times increase embryonic cf-DNA quantity, which subsequently improves amplification and concordance rates between TE biopsy, ICM, and WE [8,37,45,54,56,57,58,62,63]. In particular, culturing either frozen-thawed or fresh blastocysts to D6, rather than D5, has been shown to increase the concentration of cf-DNA in the SCM [56,57]. To our knowledge, the present study is the first to demonstrate a similar trend for mi-PGT-A using SCM+BF-derived cf-DNA from D5 or D6 blastocysts generated from embryos thawed on D3. However, extending the embryo culture period for frozen-thaw blastocyst transfer cycles to D6 to optimize mi/ni-PGT-A comes with important clinical considerations. A meta-analysis by Bourdon et al. (2019) found that for frozen-thaw cycles, D5 embryo transfers resulted in a higher CPR and LBR compared to D6 embryos [64], undermining the clinical value of extended culture for genetic testing. In contrast, Kaye et al. (2017) found that CPRs are similar between D5 and D6 frozen-thawed transferred blastocysts [61]. A more recent investigation indicates that the clinical implications of developmental age (D5 vs D6) are largely dependent on embryo morphology for frozen-thaw cycles [65]. Jian et al. (2023) found no difference in CPR, LBR, or miscarriage rate between D5 and D6 AA/AB frozen-thawed blastocysts [66]. However, for BA/BB/BC blastocysts, LBR and CPR were significantly better in D5 compared to D6 [66]. Therefore, the decision to extend culture time to D6 for mi/ni-PGT-A post-thaw should in part depend on the morphological grade of the embryo. Interestingly, we found that mi-PGT-A concordance with WEs improved most dramatically within our high-quality embryo cohort, which further supports extending the culture time for high-quality embryos to improve mi-PGT-A diagnostic performance.

4.3. mi-PGT-A Performance

mi-PGT-A diagnostic performance calculations included the SCM+BF samples collected at 72-h post-thaw that had suspected maternal cell DNA contamination and low-level mosaicism as false positives in Table 3, yielding a cumulative false positive rate (FPR) of 30.4% (14/46) and false negative rate (FNR) of 6.5% (3/46). Among all samples analyzed, mi-PGT-A had a sensitivity of 93.4% (43/46) and specificity of 69.5% (32/46). When analyzed by culture-time, there was a lower number of false positives among SCM+BF samples collected at 72-h post-thaw (28.5%; 10/35) compared to those collected at 42-h post-thaw (36%; 4/11). Similarly, but to a lesser degree, the FNR dropped from 9% (1/11) in SCM+BF samples collected at 42-h post-thaw to 6% (2/35) in samples collected at 72-h post-thaw. Sensitivity and specificity were also better among SCM+BF samples collected at 72-h post-thaw (94.3% and 71.4%) compared to those collected at 48-h post-thaw (90.9% and 63.6%). These results suggest that longer post-thaw culture times may have improved the diagnostic performance of mi-PGT-A. Longer culture time allows for a greater accumulation of cf-DNA that is more representative of the actual genetic makeup of the embryo, improving the accuracy of mi-PGT-A.

False negatives typically result from (1) insufficient or degraded cf-DNA, where abnormalities go undetected; (2) technical artifacts, resulting from errors during sample preparation; and (3) contamination from non-embryonic cells, such as sperm or maternal cumulus cells [37]. The relatively low proportion of false negatives within our samples (3/46) may be attributed to the measures taken to avoid maternal cell contamination, which was only suspected in 2/46 samples in the 72-h group. Additionally, the freeze-thaw process may have diminished adhering paternal and maternal structures such as sperm and cumulus cells. The higher proportion of false negatives SCM+BF samples collected at 48-h post-thaw (1/11) compared to those collected at 72 h (2/25) may have been a result of lower cf-DNA concentrations.

From a clinical standpoint, false positives are of great concern as they may lead to discarding of valuable euploid embryos. Many factors have been associated with false-positive results using ni/mi-PGT-A. Low levels of DNA in BF samples have been shown contribute to false positives, resulting in a lower PPV due to uneven amplification and allele dropout [44,67,68]. Other studies have suggested that FPRs for ni-PGT-A are influenced by maternal age, with fewer false positives among people 35 years and older [68]. However, the most commonly cited cause of false positives is embryo mosaicism. Discrepancies between the FPR in Table 2 (17%) and Table 3 (28.5%) in the 72-h group highlight the influence of embryo mosaicism on accurate mi-PGT-A diagnosis in the present study. Embryo mosaicism has influenced FPRs among other studies investigating mi-PGT-A. Li et al. (2018) reported a similar mi-PGT-A sensitivity (89.5%) and specificity (68.42%) to the present study, attributing false positives (6/38) to mosaicism in their sample (11/40) [36]. That being said, Li et al. [36] discarded embryos with a mosaicism of less than 30% or above 80%, meaning that the actual rate in their study may be higher. Interestingly, the sensitivity of mi-PGT-A in our study (93.4%) exceeded that of the traditional TE biopsy method for PGT-A (89.5%) used as a comparison in Li et al. [36]; however, this was not the case for specificity (73.7%). Zhang et al. (2019) reported a FNR of 7.4% (2/27) and FPR of 14.8% (4/27) [39], which they attributed to embryo self-repair mechanisms during development, whereby abnormal cells are released into the blastocoel liquid or culture medium, increasing the abundance of abnormal cf-DNA [69]. Embryos that are cryopreserved via slow cooling typically have more apoptotic cells than those that are vitrified [70], providing another possible explanation for high FPRs in the present study. To mitigate the impact of apoptotic cells on FPR, Zhang et al. [39] replaced the embryo culture media every 24 h, which may explain why they had a lower FPR compared to our mi-PGT-A protocol. Kuznyetsov et al. (2018) reported a FPR of 3/28, with one mi-PGT-A reading indicating a mosaic result with a corresponding euploid embryo [34]. In their 2020 investigation, the same authors reported three cases of a euploidy mi-PGT-A result and a mosaic PGT-A result and two cases of a euploidy PGT-A result and mosaic mi-PGT-A result [37]. The lower FPR observed in Kuznyetsov et al. [34,35] compared to our study may be attributed to the use of a SurePlex amplification system rather than MALBAC. Deleye et al. (2015) compared SurePlex and MALBAC amplification systems and found that SurePlex resulted in fewer false positives than MALBAC [71]. However, Jiao et al. (2019) also achieved a low FPR (2/21) using a modified MALBAC WGA system for mi-PGT-A that cuts the protocol down from >10 h to 2.5 h [38].

Out of the studies investigating mi-PGT-A [34,35,36,39], Li et al. [36] and our study are the only ones that directly report specificity and sensitivity measurements, which are important parameters that should be considered by future research seeking to optimize mi-PGT-A. Based on the results of our study, further test optimization is clearly required before clinical application can be considered. Researchers have compared SCM- and BF-derived cf-DNA for PGT-A [68] and SCM+BF-derived cf-DNA for mi-PGT-A [34,35,36,39], but there has been no simultaneous evaluation of SCM+BF and each component in isolation within the same study, which would provide critical evidence for determining the optimal embryo selection method.

4.4. Strengths and Limitations

Strengths of our study include a comparison of SCM+BF-derived cf-DNA collected at 48- and 72-h post-thaw of cleavage stage (D3) embryos and corresponding WE results—a comparison that has not been reported previously in the literature. We also compared results with and without mosaic reporting, demonstrating the implications of embryo mosaicism on mi-PGT-A diagnostic performance.

Although concordance rates improved when SCM+BF samples were collected at 72-h post-thaw, these results were still below clinically acceptable ranges and rates found by some studies evaluating ni-PGT-A [56,57,63]. While the present study rinsed embryos three times post-thaw, a recent investigation by Chow et al. (2024) found that sequentially rinsing embryos five times before placing them into fresh culture media improved ni-PGT-A concordance [56]. A similar protocol involving six rinsing stages for frozen-thawed embryo cycles also yielded positive results [57]. That being said, both studies lend support to our findings that longer culture times improved concordance rates due to a higher amount of cf-DNA extrusion by the embryo [56,57]. In contrast to the present study, which analyzed frozen research embryos that were transferred to new media on day 3, other groups suggest that transferring the growing embryo to new media on day 4 results in higher concordance rates [43,58,60]. Small sample sizes in some of our comparison groups (for example, SCM+BF derived from high-quality versus low-quality embryos) limited the usefulness of the analysis but still provide insight into the effectiveness of mi-PGT-A across embryos of different qualities. Our sample sizes may have been better if we had used embryos that were frozen via vitrification, which yields better cryo-survival rates and post-thaw development compared to slow freezing [71]. However, the limited availability of frozen research embryos for this study restricted our choice of freezing method to those that were cryopreserved via slow freezing. Additionally, the relationship between embryo quality and ni/mi-PGT-A concordance remains unclear. The high proportion of false positives requires further study to elucidate the mechanism(s) leading to these results.

5. Conclusions

This study effectively demonstrates that mi-PGT-A, using a combination of spent culture media and blastocoelic fluid-derived cell-free DNA, is a promising alternative to traditional PGT-A methods involving embryo biopsy. Our results revealed that extending the culture time of D3 embryos from 48-h to 72-h post-thaw significantly enhances the DNA detection rate and concordance with WE genetic profiles, suggesting that a longer culture period allows for more representative cf-DNA sampling and potentially reduces diagnostic errors such as false positives and negatives. Ploidy concordance rates also appear to improve when mosaicism is not reported.

The usefulness of mi-PGT-A awaits further optimization and clinical outcome evaluation before it can be successfully applied clinically. Strategies for this may include discriminating between exogenous and maternal cf-DNA and embryonic DNA in SCM to improve ploidy concordance rates. Since aneuploid cells may be extensively eliminated from the growing embryo during the process of self-correction, future clinical studies should focus on the origins of SCM+BF-derived cf-DNA to conclude whether cf-DNA really reflects the embryo genetic status. Future research should also focus on optimizing the protocols for cf-DNA collection and analysis to enhance the accuracy and reliability of mi-PGT-A. Exploring the implications of embryo quality and developmental stage on DNA yield and test accuracy could provide deeper insights into the utility of mi-PGT-A across a broader range of clinical scenarios.

If a reliable method of mi-PGT-A is developed, it could reduce embryo manipulation, reduce PGT-A costs, and improve ART outcomes by using SCM+BF-derived cf-DNA data for better embryo selection compared to current standards.