On the reproductive capabilities of aneuploid human preimplantation embryos

Summary

In IVF cycles, the application of aneuploidy testing at the blastocyst stage is quickly growing, and the latest reports estimate almost half of cycles in the US undergo preimplantation genetic testing for aneuploidies (PGT-A). Following PGT-A cycles, understanding the predictive value of an aneuploidy result is paramount for making informed decisions about the embryo’s fate and utilization. Compelling evidence from non-selection trials strongly supports that embryos diagnosed with a uniform whole-chromosome aneuploidy very rarely result in the live birth of a healthy baby, while their transfer exposes women to significant risks of miscarriage and chromosomally abnormal pregnancy. On the other hand, embryos displaying low range mosaicism for whole chromosomes have shown reproductive capabilities somewhat equivalent to uniformly euploid embryos, and they have comparable clinical outcomes and gestational risks. Therefore, given their clearly distinct biological origin and clinical consequences, careful differentiation between uniform and mosaic aneuploidy is critical in both the clinical setting when counseling individuals and in the research setting when presenting aneuploidy studies in human embryology. Here, we focus on the evidence gathered so far on PGT-A diagnostic predictive values and reproductive outcomes observed across the broad spectrum of whole-chromosome aneuploidies detected at the blastocyst stage to obtain evidence-based conclusions on the clinical management of aneuploid embryos in the quickly growing PGT-A clinical setting.

The type of chromosomal abnormality detected during preimplantation genetic testing affects embryo transfer prognosis. While uniformly aneuploid embryos carry a miscarriage risk of 86.3% and fail to result in chromosomally normal live births in over 98% of transfers, embryos diagnosed with low-range mosaicism show reproductive outcomes similar to euploid ones.

Introduction

Preimplantation genetic testing for aneuploidy (PGT-A) is employed to identify chromosomal aberrations in IVF-generated embryos, ultimately aiming at minimizing the chance of adverse clinical outcomes (i.e., implantation failure, miscarriage, and live birth affected by chromosomal abnormality) by excluding aneuploid embryos from transfer (deselection strategy). Uniform whole-chromosome aneuploidies (i.e., loss or gain of a chromosome copy in all embryonic cells) mainly derive from chromosome segregation errors occurring in maturing oocytes while undergoing meiosis.^{1, 2, 3} The contribution of aneuploid oocytes to the overall gamete output over time appears to follow a non-linear concave distribution, as chromosomal aberration rates are slightly higher at younger age (i.e., <20 years old), reduce during adulthood, and then exponentially increase at later stages of reproductive life (i.e., >35 years old).⁴ The occurrence of this natural phenomenon is attested by both a reduction of overall fertility rates in older women (including lower natural conception and higher miscarriage and abnormal conception rates), as well as a surge in uniform aneuploidies detected in IVF/PGT-A treatments as female age increases.^3,5 Meiotically derived whole-chromosome aneuploidies are homogeneously present in all cells of the human embryo and are unequivocally the most common genetic abnormality found in miscarriages⁶ and IVF-derived embryos.^2,4 Compelling evidence in human genetics shows that, in most cases, meiotically derived aneuploidies are unable to self-correct. Indeed, the product of self-correction events of a meiotically derived trisomy is measurable and characterized by the presence of uniparental disomy (UPD) in a proportion of cells in the developing embryo. When assessed on thousands of blastocyst stage embryos, this outcome was detected at extremely low rates (i.e., 0.06%).⁷ Similar findings are also reported in prenatal and postnatal studies, where UPD is observed at negligible rates in both products of conception, amniocytes and newborns (i.e., <0.001%; see review by Benn 2021⁸).^{9, 10, 11, 12} Accordingly, it is reasonable to consider PGT-A technology an appropriate strategy for deselection of embryos affected by uniform whole-chromosome aneuploidies with predictable negative outcomes.

Late advances in PGT-A testing methodologies (i.e., NGS) now allow the identification of more subtle variations in chromosome copy-number values. The detection of these features in trophectoderm (TE) biopsies is interpreted by some laboratories as evidence of embryo chromosomal mosaicism, the condition for which a secondary cell lineage within the embryo carries a divergent karyotype (e.g., mosaic diploid-aneuploid). In contrast to uniform aneuploidy, mosaic aneuploidy derives from mitotic segregation errors and is rarely detected in miscarriage (i.e., ∼1.3%).¹³ Remarkably, infertility and IVF procedures were shown not to increase the risk for mosaicism in embryos, as mosaicism is detected at similar rates in pregnancies achieved spontaneously or through IVF.^13,14 Nonetheless, the diagnosis of mosaicism and its impact on embryo development and the health of the ensuing fetus/newborn are yet uncertain and difficult to interpret. In fact, the ultimate presence and impact of chromosomal mosaicism on the fetus depends on several variables. These include the imbalance size and its gene content, the developmental stage (e.g., cleavage, blastocyst, early fetal stages) at which the secondary cell lineage arises, and the type and percentage of tissues carrying the secondary karyotype.¹⁵ Although cell differentiation pathway identification may be helpful for understanding the potential impact of the chromosomal abnormality on tissue function (and therefore its clinical relevance), this type of analysis (e.g., transcriptome analysis) is currently not sufficiently robust for clinical use at this stage of development, and its implementation may prove technically challenging. Moreover, laboratory and diagnostic factors, including the occurrence of subtle genetic artifacts, inherent analytical noise, and sampling bias, contribute to the difficulty in reaching definitive chromosomal mosaicism diagnoses and prognoses.^15,16 This aspect especially concerns analyses conducted on specimens with limited amount of cellular material (i.e., embryo biopsies). In addition, the precise characterization of mosaicism (e.g., percentage of affected cells) through PGT-A based on a single biopsy specimen remains equivocal due to sampling bias. Given that mosaic aneuploidy is difficult to accurately detect in the preimplantation embryo (unlike uniform aneuploidy), it is not surprising that the predictive value for actual clinical outcomes has been questioned.¹⁷ In a clinical setting, the inability to reliably discern between uniform and mosaic results could lead to misinforming patients of the dramatically different risks associated with the two types of chromosomal abnormalities, thus hindering the informed decision-making process.

In this perspective, we focus on the evidence gathered so far on PGT-A diagnostic predictive values and reproductive outcomes observed across the broad spectrum of whole-chromosome aneuploidies detected at the blastocyst stage to obtain evidence-based conclusions on the clinical management of aneuploid embryos.

To achieve this, we first discuss intrinsic characteristics of the different types of experimental designs employed to evaluate PGT-A performance and their usefulness in determining clinical validity and utility of PGT-A findings. Next, we review and discuss the best evidence available from clinical studies and how this information can empower individuals’ decision-making regarding the utilization of aneuploid embryos.

Positive and negative predictive values in PGT-A

In the current gold standard approach for PGT-A, a single clinical trophectoderm biopsy (cTE) consisting of 3–10 cells is collected on day 5, 6, or 7 of in vitro development and employed for genetic testing. Although there have been reassuring results from double blastocyst biopsy (i.e., rebiopsy) (e.g., for lack of a result from the first biopsy),^{18, 19, 20} TE biopsy is routinely performed in a single procedure to minimize the inherent cumulative stress imposed by multiple biopsy sessions and cryopreservation procedures. Hence, the chromosomal state of the whole embryo needs to be accurately profiled from the single specimen while accounting for sampling biases (e.g., the biopsy should be representative of the whole embryo) and experimental variability. It is therefore of paramount importance that the analytical result from a single TE biopsy is related to clear and reliable positive and negative clinical predictive values. In medicine, establishing positive predictive value of a diagnostic test is of paramount importance for defining the expected proportion of true affected cases at a given prevalence of the disease and it usually requires an independent assay to confirm or deny the original diagnosis. Since independent follow-up of an aneuploid result is often impossible because a substantial proportion of aneuploid embryos do not implant and not all euploid embryos result in deliveries, the positive predictive value (PPV) of PGT-A diagnosis cannot be calculated in a conventional fashion. In this perspective, although differently used from its general meaning, PPV will indicate the likelihood of an aneuploid result to predict embryonic developmental failure (lethality, as most of aneuploid embryos fail to implant or to progress to term) or, in the case of viable aneuploidies (like trisomies for 13, 18, 21, and sex chromosomes), the presence of a chromosomal abnormality that persists throughout the ensuing pregnancy and newborn. Indeed, most aneuploidies found in PGT-A are not compatible with life (i.e., never persist beyond first trimester), therefore assessing reproductive lethality is the best proxy for deriving clinically meaningful prognostic values after the transfer of an embryo diagnosed as aneuploid via PGT-A. The use of the term PPV in this article is defined as the ratio of the combined number of failed embryo transfers and confirmed aneuploid pregnancies divided by the total number of aneuploid embryos transferred.

More broadly, the PPV of a PGT-A test is defined as the likelihood that an embryo displaying an aneuploidy result, truly carries the abnormality in question, while the negative predictive value (NPV) defines the probability that embryos testing negative (i.e., euploid) are truly free of the aneuploidies under investigation. The diagnostic performance of PGT-A can be measured and related to different developmental time points, thus determining its ability to predict (1) the embryo’s or (2), more meaningfully, the ensuing pregnancy/newborn’s genetic constitution.

While estimates of NPV information are generated during clinical practice, as euploid embryos are routinely transferred and, for those implanting, their downstream clinical outcomes monitored (e.g., analysis of products of conception in case of miscarriage), generating robust clinical PPV data requires experimental studies where individuals in treatment consent to the transfer of aneuploid embryos or blinded ploidy status.

A practical way to measure PGT-A PPV and diagnostic validity consists in performing multiple embryo biopsies including the inner cell mass (i.e., multifocal reanalysis) in a non-clinical context, followed by blinded aneuploidy testing.^{21, 22, 23, 24, 25, 26, 27} A recent review of all studies on rebiopsies published until July 2020 shows that PGT-A analytical reliability is dependent on several factors, including the type of technology employed for analysis and the criteria used to establish concordance rate and PPV/NPVs.²⁸ This review showed that when euploidy or uniform whole-chromosome aneuploidy was detected in the clinical biopsy, the finding was confirmed in >95% of cases, proving very high reliability and reproducibility of uniform euploid/aneuploid classifications.

On the contrary, mosaicism classification showed substantial differences and variable outcomes. Intermediate copy-number categorization ranges from more loose criteria (i.e., reporting mosaicism within the 20%–80% of copy number) to more stringent ones (i.e., 30%–70%). The threshold employed for analysis has substantial impact on the predicted prevalence of mosaicism, and lower stringency criteria lead to higher detection rates and significantly lower confirmation rates in rebiopsies.²⁹ For instance, when less stringent criteria were used for mosaicism classification (i.e., 20%–80%), approximately one third of mosaic embryos were found to be uniformly aneuploid upon reanalysis.^26,28,30

Moreover, the type of classification employed to confirm the outcome of the reference biopsy is also critical and can have a dramatic impact on the prevalence observed.³¹ For example, considering any embryo with discordant karyotype in at least one rebiopsy (less stringent criteria) will predict a higher mosaicism confirmation rate, while accounting for reciprocal aneuploidy events only (if at least one complimentary trisomy/monosomy event is observed in the same blastocyst; conservative criteria) will predict much lower mosaicism confirmation rate.³²

Although not all methodological characteristics (e.g., whole-genome amplification protocols, sequencing technology, and depth) have been systematically investigated and compared, it has been shown that different PGT-A assays can result in variable mosaicism outcomes. For instance, a blinded study evaluated sensitivity and specificity of two commercially available PGT-A platforms, (quantitative polymerase chain reaction and VeriSeq NGS) in detecting aneuploidy in a cell line mixture model mimicking a mosaic trophectoderm biopsy.³² The use of previously published custom criteria for NGS increased sensitivity but also significantly decreased specificity (33% false-positive prediction of aneuploidy).

However, determining PGT-A PPV based on multifocal studies can be impractical and incomplete. In fact, an ideal analysis of PGT-A PPV performance should refer to the ability to predict embryonic reproductive outcomes/lethality or, in the case of viable aneuploidies, the presence of a chromosomal abnormality that persists throughout the ensuing pregnancy and newborn rather than in the biopsy or the preimplantation embryo itself. For instance, if a given PGT-A assay results in high false positive aneuploidy rates for one chromosome, this will be detected as aneuploid in all samples used for multifocal reanalysis, thus confirming high consistency, despite delivering an incorrect diagnosis. Moreover, certain types of aneuploidies (i.e., low grade mosaic or some segmental aneuploidies) might be normal features of the developing embryos, causing no issues for normal development to a healthy baby. Indeed, the depletion of aneuploid cells in experiments on a chimeric mouse model has been explained by competitive growth of euploid cells or selective apoptosis of the abnormal cellular clones.^33,34 Furthermore, recent findings in humans suggest the existence of early embryonic bottlenecks that might enable the normalization of mosaic aneuploidies through selective depletion or confinement to placental lineages.²⁹ Therefore, looking at reproductive outcomes and PPV at the gestational or newborn level avoids technical and experimental problems related with embryo biopsy analysis and provides more reliable data to evaluate clinical validity and utility of the broad spectrum of aneuploidies detected in PGT-A. This information can be obtained from observing the proportion of transferred embryos predicted to be aneuploid that ultimately develop into viable euploid pregnancies as compared to euploid embryo transfers.

Therefore, following up clinical outcomes of transferred chromosomally abnormal embryos showing different types of alterations (e.g., uniform/mosaic and whole/segmental aneuploidies) and construct evidence-based decisional trees is of paramount importance. On this foundation, patients can be individually counseled, reporting the known risks associated and the options available for their specific situation.

Pros and cons of different types of studies investigating the predictive values of aneuploid findings in PGT-A

Several types of study designs can be employed to define PGT-A predictive values, each characterized by specific benefits and limitations, as well as preset outcome measures and intrinsic achievable conclusions (Figure 1). In fact, in standard clinical settings, a diagnosis of uniform aneuploidy typically leads to complete deselection of the embryo from transfer, while mosaicism is usually associated with a deprioritization strategy. Often, strengths and limitations of each study design are poorly acknowledged and discussed.

Figure 1.

Main differences between retrospective, RCT, and non-selection studies when assessing predictive values of PGT-A findings in the broad spectrum of aneuploidies

During the enrollment phase, intervention is performed prior to group assignment (cohort study), after group assignment (RCT), or not performed (non-selection). Outcome measure for RCTs is limited to the clinical utility of the intervention as sufficiently powered post-intervention randomization requires a prohibitive large population. A robust evaluation of embryo’s reproductive potential in cohort studies is prevented by several uncontrolled factors and covariates that affect treatment’s prognosis independently from embryo’s chromosomal status. Non-selection trials offer the possibility to evaluate the predictive value of different types of chromosomal abnormalities including uniform and mosaic aneuploidies. ET, embryo transfer.

In “longitudinal cohort studies,” case enrollment and experimental group assignment is based on the presence of a specific feature. For example, in the context of aneuploidy testing, assignment of the case to the associated experimental group is performed after PGT-A diagnosis but before transfer (i.e., prospective study) or after the transfer has been carried out and the clinical outcome is known (i.e., retrospective study). Assignment of the case to the associated experimental group is performed after PGT-A diagnosis but before transfer (i.e., prospective study) or after the transfer has been carried out and the clinical outcome is known (i.e., retrospective study). However, these studies offer poor control over possible confounding factors and covariates that may significantly diverge across the experimental groups. In the specific context of PGT-A, this is particularly relevant in cases where the genetic information is used for deprioritization (e.g., mosaicism) and when the primary outcome measure is the assessment of reproductive potential. Indeed, it is possible that cases enrolled in the mosaic group are those where only one (mosaic) embryo is available for transfer, while the control group (i.e., euploid) includes cases where multiple embryos are available (possibly also mosaics). In this specific situation, the case with only one embryo available has, by definition, a poorer IVF prognosis compared to the control;^35,36 however, this variable is generally unaccounted for in the analysis. Alternatively, some cases may enter the “mosaic/aneuploid” group only after undergoing (multiple) failed euploid embryo transfers. Similar to the previous example, these cases bear an intrinsically poorer prognosis compared to first timers because good prognosis patients received euploid embryos as first option. For example, in the largest retrospective study on mosaic embryo transfer outcomes currently available, the mosaic groups showed poorer clinical outcomes compared to the euploid group. However, Viotti and colleagues report that in 94% of cases where mosaic embryos were transferred, no euploid embryos were available.³⁷ Although meaningful for other endpoints, a cohort study cannot provide a reliable measure of the impact of aneuploid findings (like mosaicism) on the embryonic reproductive potential. For instance, in the case of the Viotti study, the lower reproductive potential detected in mosaic embryos may be due to a significantly different prognosis across study arms rather than to the genetic finding itself.

On the other hand, prospective and retrospective longitudinal cohort studies are generally less expensive and can include larger populations at a fraction of RCTs’ costs. Overall, although outcomes from retrospective studies should not be used to provide evidence of embryos’ reproductive potential or to develop new policies for embryo selection, they can provide useful information on the genetic risks of aneuploid/mosaic embryo transfer at the pregnancy/newborn level. In this context, it is important to remind that, for certain types of aneuploidies, diagnostic confirmation at later stages does not necessarily mean a negative clinical outcome. For example, a mosaic trisomy identified as confined placental mosaicism (CPM) might have little clinical significance.

“Randomized clinical trials” (RCTs) are prospective studies that minimize population selection bias by performing random blinded experimental group assignment prior to intervention (e.g., in this context, embryo selection based on PGT-A diagnosis). Usually, RCTs are carried out to compare the efficacy of an alternative clinical strategy with current standard of care. In the PGT-A context, RCTs are designed to determine the clinical gain of selecting a euploid embryo for transfer on the economy of IVF treatment cycles (Figure 1). Different primary and secondary outcome measures can be used to power the trial, including live birth rate (LBR) per cycle, LBR at first embryo transfer (ET), and miscarriage rate. In general, RCTs are among the most powerful study designs for the assessment of the relationship between an intervention (e.g., deselection of aneuploid embryos) and its clinical outcome (e.g., sustained clinical pregnancy) and to identify the best population of patients that can benefit from a medical treatment. However, these types of studies are typically underpowered for estimating the PPV of aneuploid diagnoses and their reproductive outcomes/lethality rate, as chromosomally abnormal embryos are not transferred nor followed up by default during the trial. Cumulative LBR per cycle could be used to indirectly assess the impact of false positive diagnosis from PGT-A. However, powering RCTs for this IVF outcome is statistically prohibitive because of the exceedingly high sample size required. Consequently, cumulative LBR analysis per cycle is often limited to a few embryo transfers or a short period of time. For example, in a recently published RCT where LBR was compared across groups receiving PGT-A and not, only young women with a good prognosis for pregnancy were included, follow-up data on cumulative LBR per cycle were limited to three embryo transfer procedures within 1 year after randomization, and the number of embryos subjected to PGT-A was limited to three.³⁸ Furthermore, mosaic embryos were deemed as not transferrable. Because aneuploid/mosaic embryos were not replaced and cumulative LBR was not comprehensively investigated, deriving conclusions on aneuploid diagnoses’ PPV and lethality was limited. Despite the costs and efforts involved in the production of such trials, the evidence on PPV and reproductive outcomes provided could be more efficiently produced through non-selection trials.

“Non-selection trials” are prospective studies that minimize population selection bias by performing random blinded experimental group assignment without intervention (Figure 1). In this design, participants undergo the exact same procedures, while group assignment is revealed after primary outcomes measures are available (in our context, PGT-A diagnosis is available at the time of embryo transfer, however it is not accessed until week 12 of gestation). This approach differs from RCT, as randomization is performed independently from the intervention results. Non-selection trials can be semi-randomized when a non-tested parameter is used for randomization. For example, embryos for transfer could be chosen by their culture number (i.e., full randomization) or on the basis of their morphology (i.e., semi-randomization). In theory, this strategy should further normalize the experimental groups and, when semi-randomized, increase homogeneity for a specific variable (e.g., in this example, embryo morphology). However, semi-randomization may lead to skewed enrolment rates across groups if the parameter used for randomization is associated with the non-selective criteria (e.g., if better morphology were associated with embryo chromosomal mosaicism). Non-selection trials can be developed to investigate the clinical prognostic values of any possible kind of embryonic biomarker. In the PGT-A context, non-selection trials have several advantages and provide, if sufficiently powered, robust conclusions on the PPV and lethality of an aneuploidy diagnosis and extremely valuable data for improving decision making in PGT-A cycles (Figure 1). Compared to longitudinal cohort studies, non-selection trials offer a virtually unbiased population selection process, allowing a more effective detection of PGT-A’s false positive error rate for uniformly aneuploid embryos and reproductive potential estimation for embryos classified as putative mosaics in comparison to euploids, as well as the risk of resulting in a mosaic pregnancy of clinical meaning.

Knowledge of these parameters is extremely important for making clinical decisions in PGT-A cycles presenting with embryos showing a broad spectrum of aneuploidies.

For the reasons detailed above, the following clinical data review and discussion is primarily based on non-selection trials conducted so far on aneuploid and putative mosaic embryos.

Mosaic embryo transfer outcomes in a non-selection design

Three studies reporting clinical outcomes for whole-chromosome mosaic embryos on the basis of a non-selection design are currently available.^26,39,40 This analysis works under the assumption that if the PPV associated with a mosaic diagnosis were substantial and population biases due to deprioritization of mosaics were sufficiently minimized, mosaic embryos should show reduced reproductive potential compared to euploid ones. In the first of these studies,³⁹ the primary aim was to determine the association between uniform aneuploidy diagnosis and failure of successful delivery; therefore the 484 participating embryo transfers were divided into euploid and aneuploid categories whose outcomes are discussed in the following section of the article. Additionally, a minority of the analyzed embryos (n = 72/2,110, 3.4%) showed “secondary findings of uncertain significance” (e.g., mosaicism). Sixteen embryos from this group were transferred, resulting in one (6.3%) negative pregnancy test, two (12.5%) biochemical losses, two (12.5%) clinical miscarriages, and 11 (68.8%) sustained implantations/live births versus 17.9%, 9.0%, 7.4%, and 64.7% in the euploid group, respectively.³⁹ Despite the similar outcomes generated by mosaic and euploid embryos, and the consequent absence of association between detection of mosaicism and reproductive potential, the authors reasonably stated that this analysis was underpowered for reaching significant conclusions. In 2021, Wang and colleagues⁴⁰ reported clinical outcomes of a total of 188 PGT-A embryos that were selected for transfer on the basis of their morphology alone. After transfer, their chromosomal status was disclosed, revealing 135 euploid embryos (110 uniform whole-chromosome euploidy, 12 mosaic < 50%, and 13 segmental mosaic < 50%) and 53 aneuploid (44 showing uniform whole-chromosome aneuploidy and 9 segmental aneuploidy). Positive pregnancy, implantation, ongoing clinical pregnancy, and live birth rates were 69.1%, 59%, 52.7%, and 50.9% in the euploid group and 41.6% for the mosaic group (5/12 transfers progressing through pregnancy from positive result to live birth). Like Tiegs’ study, Wang’s study underpowered to allow a meaningful comparison between euploid and mosaic outcomes. Lastly, Capalbo and colleagues recently published a non-selection trial focused on comparing clinical outcomes from uniformly euploid and putative mosaic blastocysts in the low to medium range (20%–50% of predicted abnormal cells in the clinical TE biopsy).²⁶ This study included 484 transfers of uniformly euploid embryos and 413 putative mosaic embryos (282 with 20%–30% variation and 131 with 30%–50% predicted mosaicism rate). No differences in positive pregnancy, biochemical loss, miscarriage, and live birth rates were detected across the groups. Moreover, post-natal genetic tests in a subset of newborns derived from mosaic embryos showed no evidence of mosaicism or its potential by product (i.e., UPDs). Notably, this trial is the only study available that is designed and powered to detect differences in reproductive outcomes between uniformly euploid and mosaic embryos in the low-medium range. However, a study arm with higher mosaicisms ranges was excluded from the experimental design because of preliminary evidence of extensive uniform aneuploidy detected in re-biopsies of embryos showing this feature.²⁶ Although the reproductive potential across the study groups was shown to be consistent, the positive predictive value thereby calculated provides crucial information on the risk of having a mosaic pregnancy/baby when these embryos are transferred. To this end, retrospective analysis and prospective trials provide equally robust conclusions, as there are no factors known to be associated with the risk of generating a mosaic pregnancy that are worth controlling for in a non-selection or RCT study. The pregnancy follow-up analysis of mosaicism diagnosis in blastocyst stage PGT-A was recently reported by Treff and colleagues, showing 0.003 risk of identifying a consistent mosaic pattern in the ensuing pregnancy/newborn.⁴¹ In only one case, the newborn showed a mosaic profile that was consistent with the analysis previously conducted on the embryo. Of note, this mosaicism case has not so far resulted in a clinically meaningful negative outcome in the baby, as follow-up medical evaluations showed normal values for all the parameters assessed. This risk is even lower than what is expected in pregnancies following both natural and IVF/non-PGT-A conceptions.^13,41 However, it is important to highlight that all the data reported on PPV of mosaic in pregnancies have not been systematically collected during a clinical trial or retrospective analysis (e.g., by comparing amniocentesis samples from euploid and mosaic embryo transfers with increased number of metaphases). Although every PGT-A assay would require clinical validation, with current evidence, the transfer of embryos showing putative chromosomal mosaicism is reasonable when the deviation from the diploid conformation is below 50%. These embryos are likely to produce a normal pregnancy at similar rates as uniformly euploid ones.

Aneuploid embryo transfer outcomes in non-selection trials

Five published papers have investigated the reproductive potential of embryos designated as uniformly aneuploid in a non-selection experimental design and one in a cohort study.^{39,40,42, 43, 44} The first of these studies was conducted in 2012 and utilized SNP-array-based PGT-A, which did not distinguish between uniform and mosaic aneuploidy.⁴³ In this study, four embryos diagnosed as aneuploid resulted in a healthy delivery. However, subsequent polar body analyses of the four aneuploid embryos found no evidence of maternal meiotic origin of the detected aberrations, supporting the likely mitotic (mosaic) nature of the aneuploidies observed.⁴⁵ This suggests that PGT-A methodologies able to accurately distinguish between mitotic and meiotic origin of aneuploidies might improve embryo classification, at least in some specific cases. Overall, a total of 95 uniformly aneuploid embryos were transferred and 95 failed to sustain implantation (100%). In the previously mentioned 2020 study, Tiegs and colleagues observed 100% failure rate following the transfer of 102 uniformly aneuploid embryos diagnosed with a targeted NGS-based PGT-A method.³⁹ In 2021, as mentioned above, using another WGA-based NGS PGT-A methodology, Wang and colleagues observed a 95% failure rate after transfer of 44 uniformly aneuploid embryos, and two aneuploid embryo transfers led to live births.⁴⁰

Another recent report on reproductive outcomes of uniformly aneuploid embryos by Yang et al. showed that six uniformly aneuploid embryos transferred resulted in clinical miscarriage. In this study, the authors analyzed leftover embryonic DNA samples from cases that underwent preimplantation genetic testing for a monogenic condition (PGT-M) and compared clinical outcomes of embryos free of the targeted single-gene disease on the basis of the presence/absence of aneuploidy. Moreover, recent data on transfer outcomes of uniformly aneuploid embryos were reported in a cohort study performed by the Gleicher group.⁴² While the initial publication controversially reported live births from aneuploid embryo transfers, corrected data from Yang and colleagues demonstrated that the transfer of a total of 61 uniformly aneuploid embryos led to zero positive outcomes and several aneuploid miscarriages when using a variety of PGT-A platforms.⁴²

Finally, in the latest case series, Barad and colleagues reported clinical outcomes from 144 transfers involving embryos with aneuploid or inconclusive diagnosis (i.e., degraded DNA from biopsy) from 69 couples.⁴⁴ Considering exclusively embryos in which at least one whole-chromosome aneuploidy in the uniform range had been detected, only one live birth derived from 106 transfers. The remaining six pregnancies resulted in clinical miscarriage.

Based on the evidence gathered so far, these data demonstrate that a PGT-A diagnosis of uniform aneuploidy provides 98% (346/353; 95% CI: 96.0%–99.2%) predictive value for early lethality (Table 1) and 86.3% (44/51; 95% CI: 73.7%–94.3%) miscarriage rate.

Table 1.

Published non-selection and cohort study outcome data involving transfer of full aneuploid embryos for whole chromosomes

Study	Design	Transfers of uniformly aneuploid embryos na	Miscarriage rate % (n, 95% CI)	Lethality rate % (n, 95% CI)
Scott et al., 2012 (45)	blinded	95	33.3% (2/6) (4.3%–77.7%)b	95.8% (91/95) (84.5%–99.4%)
Tiegs et al., 2021 (40)	blinded	102	100% (24/24) (85.8%–100%)	100% (102/102) (96.5%–100%)
Wang et al., 2021 (41)	blinded	44	75.0% (6/8) (34.9%–96.8%)	95.5% (42/44) (84.5%–99.4%)
Yang et al., 2022 (44)	blinded	6	100% (6/6) (54.1%–100%)	100% (6/6) (54.1%–100%)
Barad et al., 2022 (46)	unblinded	106	85.7% (6/7) (42.1%–99.6%)	99.1% (105/106) (94.9%–99.9%)
Total	N/A	353	86.3% (44/51) (73.7%–94.3%)	98.0% (346/353) (96.0%–99.2%)

^aEmbryos with at least one whole-chromosome aneuploidy in the uniform range (non-mosaic).

^bPost-transfer polar body analysis revealed likely mitotic (mosaic) origin of the aneuploidy detected through PGT-A.

Possible explanations of the rare cases of healthy deliveries include true trophectoderm/ICM mosaicism, inadvertent sample mix-up, analytical errors, or even the occurrence of a natural pregnancy. Without the inclusion of genotyping information from the embryo biopsy and newborns, the cause of the discordance between results cannot be further ascertained.

According to this evidence, disposal/deselection of uniformly aneuploid embryos is thus expected to have no meaningful implications for cumulative live birth rate in IVF cycles with PGT-A (i.e., the change of delivery on an intention to treat analysis when all utilizable embryos are transferred). Indeed, the lethality rate following aneuploid embryo transfer described in our literature analysis is corroborated by a recent study reporting outcomes for preimplantation genetic testing in the United States between 2014 and 2018, showing an improvement in cumulative live birth rate in women older than 35 years who used PGT-A compared to those who did not, regardless of the number of oocytes collected.⁴⁶

From a counseling perspective, for women with difficult choices involving only aneuploid embryos, the specific abnormality detected is probably very important and certain types of aneuploidies (e.g., 47,XXX) can be considered safe for transfer and might be associated with equivalent reproductive potential as euploid. However, datasets are yet underpowered to allow chromosome-specific evaluations and will require further investigation going forward. In cases where uniform aneuploidies are detected, the diagnostic outcome can be considered true with high confidence and transfer of affected embryos would most likely result in failed implantation, miscarriage, or the birth of a baby carrying chromosomal syndromes. Because PGT-A is generally undertaken to minimize the chance of such outcomes, the transfer of aneuploid embryos should be avoided.

Progressing PGT-A diagnosis

Based on this evidence, we believe that a few robust and evidence-driven considerations can be drawn. For what concerns the primary objective of any aneuploidy testing strategy in preimplantation embryos, that is avoiding the transfer of uniformly aneuploid embryos, PGT-A is highly accurate. According to the results from the non-selection studies discussed here, whole-chromosome uniform aneuploidy diagnosis is almost never associated with the live birth of a healthy baby. In very rare cases, aneuploid embryos can lead to the birth of an unaffected child, but their exclusion from transfer is not expected to result in a meaningful reduction of the cumulative LBR per cycle as recently shown by a day-to-day general clinical practice analysis.⁴⁶ Although highly discouraged, it can be argued that individuals have an ethical right to transfer an embryo with a genetic abnormality because of reproductive liberty and autonomy.⁴⁷ While there are many important considerations, one key factor is whether the anomaly is lethal. A PGT-A uniform aneuploidy result has now been demonstrated by five studies to successfully predict lethality with over 98% accuracy.^{39,40,42, 43, 44} Furthermore, the indiscriminate clinical use of uniformly aneuploid embryos showed alarming outcomes with all pregnancies resulting in spontaneous miscarriage. This approach is likely to cause additional costs, physical and psychological harm, and a potentially vital loss of precious time for the reproductive journey of the infertile couple.

At minimum, individuals considering IVF treatment options, such as undergoing another IVF cycle, using donor gametes, or transferring a full aneuploid embryo, should be appropriately informed of risks and benefits based upon this evidence. Transfer of embryos diagnosed as uniformly aneuploid for whole chromosomes may also be best considered only within the context of research and under institutional review board approval (i.e., clinicaltrials.gov ID NCT04109846). Failing to do so is not only misleading, but it may also impose unnecessary harm to patients undergoing infertility treatment.

On the other hand, embryos displaying low range mosaicism for whole chromosomes have shown reproductive capabilities somewhat equivalent to uniformly euploid embryos with comparable clinical and gestational outcomes. The clinical validity and utility of reporting these findings in the context of PGT-A is thus highly questionable for several reasons. First, these findings do not predict a significantly different reproductive potential nor, to date, do they provide any benefit over or in combination with morphology. Second, PGT-A findings consistent with mosaicism do not predict higher genetic risks for the ensuing pregnancies.⁴¹ Finally, a considerable likelihood of embryo wastage and impact on cumulative LBR of IVF treatments has been demonstrated in projecting models with chromosomal mosaicism findings as criteria for embryo deselection from transfer.²⁶ Therefore, given their clearly distinct clinical consequences, we believe that careful differentiation between uniform and mosaic aneuploidy is critical in both the clinical setting when counseling on genetic diagnoses and in the research setting when presenting aneuploidy studies in embryology and IVF. Failing to do so in the research setting might lead to the delivery of erroneous information.

Future works will be required to clarify the PPV (in terms of clinically meaningful outcomes) of mosaic diagnosis in pregnancies by using more targeted approaches, such as increased metaphases counts, in a more systematic comparison between euploid and mosaics where all ensuing pregnancies are properly followed up. Finally, further studies are needed to underpin the next challenge in PGT-A, that is defining the prognostic value of segmental aneuploidies detected in clinical TE biopsies, similarly to what has been achieved for uniform and mosaic whole-chromosomes aneuploidies.