IVF outcomes of embryos with abnormal PGT-A biopsy previously refused transfer: a prospective cohort study

Abstract

STUDY QUESTION

What are the outcomes for patients who choose to move embryos diagnosed as abnormal by preimplantation genetic testing for aneuploidy (PGT-A) to a new institution for transfer after the diagnosing institution refused to transfer them?

SUMMARY ANSWER

Many patients seek to have selected embryos with PGT-A abnormal trophectoderm biopsies transferred recognizing that these embryos can still offer a chance of pregnancy and live birth.

WHAT IS KNOWN ALREADY

: PGT-A is a widely practiced method of selecting embryos for transfer based on biopsy of a few cells. Many clinical practices refuse to transfer PGT-A abnormal embryos even when there are no other ‘normal’ embryos available.

STUDY DESIGN, SIZE, DURATION

This is a prospective cohort of 69 couples who, since 2014, moved a total of 444 PGT-A abnormal embryos previously refused transfer at their parent institutions to our practice. Among these, 50 patients have, thus far, undergone 57 transfer cycles of 141 embryos.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Embryos diagnosed at other institutions by PGT-A as abnormal (mostly using next generation sequencing) were moved to our academically affiliated private fertility and research center in New York City. Female age at retrieval was 41.35 ± 3.98 years, 74% were Caucasian, 12% Asian and 10% were of African descent. All embryos identified as PGT-A abnormal among prospectively identified couples were recorded in our center’s registry.

MAIN RESULTS AND THE ROLE OF CHANCE

Among the 144 embryos transferred 102 (72.3%) had only 1 or 2 chromosomal abnormalities, 30 (21.3%) had 3 or more and 9 (6.4%) were ‘undiagnosed’ because of degraded DNA, yet still had been refused transfer. Transfer of PGT-A abnormal embryos resulted in 8 live births, 11 miscarriages and no voluntary terminations. One child was born with a segmental duplication and required repair of coarctation of the aorta as a newborn. Many couples with only PGT-A abnormal embryos are willing to have their PGT-A abnormal embryos transferred and such transfers can result in the establishment of ongoing euploid pregnancies and live births.

LIMITATIONS, REASONS FOR CAUTION

Findings in this case series represent couples who chose to have their embryos transferred after having been refused transfer elsewhere and may not be representative of the wider population of couples undergoing IVF with PGT-A in general. Not all abnormal phenotypes present in the immediate postnatal period so it will be important to continue to follow the development of these children.

WIDER IMPLICATIONS OF THE FINDINGS

PGT-A can result in a clinics refusal to transfer embryos with abnormal PGT-A biopsies, even those with mosaic findings, consequently large numbers of infertile women are prematurely advised that their only chance of motherhood is through third-party egg-donation.

STUDY FUNDING/COMPETING INTEREST(S)

This work was supported by intramural funds from the Center for Human Reproduction and the not-for-profit research Foundation for Reproductive Medicine, both in New York, NY, USA. N.G. and D.H.B. are listed as co-inventors on several U.S. patents. One of these patents (US Patent# 7,615,544) relates to pre-supplementation of hypo-androgenic infertile women with androgens, such as DHEA and testosterone and, therefore, at least peripherally related to the subject of this manuscript. N.G. and D.F.A. also received travel funds and speaker honoraria from several pharmaceutical and medical device companies, though none related to the here presented subject and manuscript. N.G. is a shareholder in Fertility Nutraceuticals and he and D.H.B. receive royalty payments from Fertility Nutraceuticals LLC.

TRIAL REGISTRATION NUMBER

N/A.

preimplantation genetic testing for aneuploidy (PGT-A), preimplantation genetic screening (PGS), in vitro fertilization (IVF), embryo aneuploidy, embryo mosaicism

Introduction

For decades, chromosomal abnormalities have been considered a principal cause of failed IVF embryo implantations and miscarriages (Simpson, 1993; Macklon et al., 2002; Greaney et al., 2018). Preimplantation genetic testing for aneuploidy (PGT-A) is the third iteration of an over 20 years-old hypotheses suggesting that deselection of chromosomal ‘abnormal’ embryos prior to embryo transfer will identify embryos with improved chances for implantation, pregnancy and live birth, thereby improving IVF outcomes (Verlinsky and Kuliev, 1998). This hypothesis remains unconfirmed until now both in meta-analyses and recent clinical trials (Mastenbroek et al., 2011; Munné et al., 2019; Cornelisse et al., 2020; Pagliardini et al., 2020; Yan et al., 2021), with considerable evolving evidence contradicting PGT-A’s validity based on new biologic findings regarding mouse and human preimplantation-stage embryos (Bolton et al., 2016; Yang et al., 2021).

Current PGT-A practice is based on 2016, guidance by the Preimplantation Genetic Diagnosis International Society which defined an ‘abnormal-mosaic’ embryo on trophectoderm biopsy at blastocyst-stage as demonstrating between 20% and 80% aneuploid DNA. Biopsies with <20% aneuploid DNA were considered ‘normal-euploid’, and those over 80% were considered ‘abnormal-aneuploid’ (Preimplantation Genetic Diagnosis International Society, 2016) (http://pgdis.org/docs/newsletter_071816.html).

While a blastocyst-stage embryo has on average 250 cells, those guidelines define embryos based on a single trophectoderm biopsy, made up of only on average 5 to 10 cells. A biopsy of only 5 to 10 cells mathematically, however, cannot fully represent a complete embryo (Gleicher et al., 2017). Moreover, by defining mosaicism within a range of 20–80% of aneuploid DNA in a single 5 to 10 cell biopsy, PGT-A contradicts the universal definition of mosaicism as the presence of more than one cell lineage in a whole organism, not just in a single biopsy of a few cells (Queremel Milani and Chauhan, 2021).

Current PGT-A practice, therefore, ignores the ploidy of remaining unbiopsied cells in a blastocyst-stage embryo. Even a 100% aneuploid trophectoderm biopsy cannot prove that the remaining embryonic cells are all aneuploid. Under PGDIS reporting guidance, a ‘mosaic’ biopsy result, therefore, describes a ‘mosaic’ embryo; however, embryos reported as ‘euploid’ or ‘aneuploid’, may also be ‘mosaic’. PGT-A, as currently offered to the public, may significantly underreport mosaicism in human blastocysts and overreport ‘euploidy’ and ‘aneuploidy’. The ultimate consequence of this may be false-negative and false-positive diagnoses of many embryos (Capalbo et al., 2021; Viotti et al., 2021a). As was first demonstrated in mice (Bolton et al., 2016) and more recently in human embryos and human gastruloids (Yang et al., 2021), aneuploid embryos often self-correct downstream from the blastocyst-stage. In other words, embryos can self-correct after embryo biopsies are taken during PGT-A. Self-correction is, moreover, significantly more pronounced in the embryonic cell lineage (inner cell mass) from which the fetus arises than in the extra-embryonic cell lineage (trophectoderm), from which the placenta develops, and from which embryo biopsies are obtained during PGT-A (Bolton et al., 2016; Yang et al., 2021). A very recent study reconfirmed and further explored the known observation that term placentas are inherently mosaic, characterized by a substantial number of chromosomal abnormalities, even if the fetus is completely euploid (Coorens et al., 2021). An aneuploid trophectoderm biopsy at blastocyst stage, even if correctly diagnosed by PGT-A, therefore, may not accurately predict the ultimate chromosomal fate of the resulting fetus. The clinical utility of PGT-A has, therefore, been increasingly challenged (Gleicher et al., 2021). Thus, false positive results with PGT-A may be caused by a multitude of potential contributing factors: sampling errors, technical errors and failure to acknowledge the potential for embryonic self-correction of aneuploidies downstream from biopsy and/or preferential self-selection of euploid cell lineages over aneuploidies.

In the USA, the clinical utilization of PGT-A has continued to expand. With such growing use of PGT-A, the number of IVF patients completing cycles without having any ‘euploid’ embryos for transfer is growing in parallel. After witnessing several young couples within a short time completing often multiple serial IVF cycles without transferrable embryos and increasingly skeptical of PGT-A, our center in 2014 announced a selective transfer protocol for embryos determined by preimplantation genetic testing to be chromosomal abnormal (Supplementary Data File S1). By 2015, we and colleagues reported five chromosomal normal pregnancies following such transfers (Gleicher et al., 2015), followed shortly thereafter by Italian investigators, reporting an additional six normal live births (Greco et al., 2015). Since then, worldwide, hundreds of pregnancies have been reported without adverse outcomes following transfer of embryos previously diagnosed by PGT-A as ‘abnormal’ (Patrizio et al., 2019).

Many, if not most, US and international IVF centers continue to refuse transfer of such embryos and PGT-A consent forms often require that patients give permission for automatic disposal of all embryos determined by PGT-A to be ‘chromosomal-abnormal’. Because of increasing controversy surrounding the clinical utilization of PGT-A (Gleicher et al., 2021), many IVF centers have recently stopped automatic disposal of such ‘abnormal’ embryos but usually, still, refuse their transfers. Patients, therefore, in growing numbers have chosen to move allegedly ‘abnormal’ embryos for potential transfers to our center. This trend presented us with the opportunity to create a registry that allowed continuous prospective assessment of pregnancy and live birth potential as well as miscarriage risk for such embryos from IVF centers in the USA and Europe, created under a large variety of IVF protocols. The center’s registry, therefore, at least up to this point, appears representative of IVF outcomes independent of protocols used. Our reported findings, therefore, should be widely applicable.

The objective of this study is to report the clinical outcomes from our registry for patients who chose to transfer embryos diagnosed as abnormal by PGT-A after their embryos were initially refused transfer at other institutions.

Materials and methods

Study population

Our center accepts all embryos previously diagnosed by PGT-A as chromosomal-abnormal, whether mosaic or aneuploid. Embryos described in this report were moved to our center or underwent embryo transfer at five independent local practices. The embryos were from patients who, after multiple IVF cycles, had no remaining PGT-A ‘normal’ embryos and who were refused transfer of their embryos at their former IVF centers. Among patients in this cohort 74% were Caucasian, 12% Asian, 10% were of African and 4% were of Hispanic descent.

Embryo selection

Transfer of ‘chromosomal-abnormal’ embryos at our center is selective: patients are advised against transfer of non-lethal trisomies (+13, +15, +18, +21) and of sex-chromosome abnormalities (45, X and 47, XXY), including segmental abnormalities involving sex chromosomes (Sanlaville et al., 2009). In the reported cohort, there were no segmental abnormalities involving sex chromosomes. We transferred embryos classified by PGT-A as ‘mosaic’ or ‘aneuploid’, recognizing that embryos classified as fully aneuploid could in fact be mosaic.

Informed consent

Our center’s consent form for transfer of ‘abnormal’ embryos describes the risks of miscarriages and possibility of ongoing chromosomal-abnormal pregnancies (Supplementary Data File S2). All patients were advised to have early invasive prenatal genetic diagnosis, including microarray analysis. Some patients, having fully embraced the notion that PGT-A diagnosis could be unreliable, requested to transfer embryos diagnosed with potentially viable aneuploidies. As noted previously, patients choosing to transfer embryos diagnosed with non-lethal aneuploidies certify that they did so ‘against medical advice’ acknowledging a willingness to either terminate an affected pregnancy or to accept a child with a non-lethal aneuploidy if the non-lethal aneuploidy expressed in the biopsy is present in the resulting on-going pregnancy.

Diagnostic PGT-A platforms utilized and data transcription

All but three cycles in this series resulting in clinical pregnancy used next generation sequencing (NGS). The three exceptions used Array Comparative Genomic Hybridization (aCGH) or single-nucleotide polymorphism-based microarray analysis (SNP). Reported PGT-A diagnoses were transcribed from original laboratory reports and converted to a single reporting style. For embryos reported to be ‘mosaic’ by PGDIS criteria, we do not report total chromosome counts since alleged percentages are not always reported.

Pregnancy outcomes were collected from medical records and/or correspondence with patients and/or treating physicians. Concordance of pregnancy outcomes with PGT-A diagnoses was assessed based on prenatal diagnostic procedures by chorionic villous sampling, amniocentesis and/or genetic microarray analyses post-miscarriage or delivery. A small number of patients who refused chromosomal testing of their pregnancies are noted in the table. Five earlier pregnancies and deliveries we reported prior to the establishment of this registry are not included in the present report (Gleicher et al., 2015). A first interim analysis of the registry including 5 live births, 4 miscarriages and 23 transfers without implantation was previously reported in a manuscript primarily addressing the self-correction of human embryos (Yang et al., 2021).

Statistical analysis

Patient characteristics are presented as mean and SD, median and quartiles or frequency and percentage, depending on the distribution of the variable. Where outcome data were not normally distributed, analyses were performed using a non-parametric test (independent samples Kruskal–Wallis test). All statistical analysis was carried out with the use of the Statistical Package for the Social Sciences 21.0 (SPSS). P < 0.05 was considered statistically significant.

Institutional review board approval

As noted earlier, in 2014, our center developed and published a new clinical policy, including a detailed informed consent that allowed for transfer of selected ‘abnormal’ embryos, previously designated as aneuploid or mosaic by PGT-A (Supplementary Data Files S1 and S2) (https://www.centerforhumanreprod.com/fertility/possibility-selectively-transferring-embryos-preimplantation-genetic-diagnosis-pgdpgs-determined-chromosomally-abnormal/).

This policy was cited by the ASRM’s Ethics Committee as example for a clinical policy regarding testing and transfer of embryos (Daar et al., 2017). Our center’s IRB declined to review transfer of these embryos as experimental when the registry was established. The IRB noted that our center was already transferring such embryos as part of a published clinical policy. Understanding that some unbiopsied embryos would undoubtedly have had abnormal biopsies if they had then undergone PGT-A, they reasoned that transfer of embryos that might have had a non-euploid biopsy was already a well-established clinical procedure. As part of this informal discussion, the IRB did advise that we use a specific informed consent for this procedure outlining risks and benefits (Supplementary Data File S2). The IRB, at a later meeting, however, did approve the analysis of anonymized clinical outcome data from the center’s anonymized electronic research database (project #ER03302015).

Results

Patients, embryos and IVF cycles

Since December of 2014, 69 patients had 444 embryos, designated by PGT-A as non-euploid and refused transfer in their home institutions, recorded in our registry. In all, 50 women (72.5%) have undergone at least one embryo transfer in a total of 57 transfer cycles, involving 141 embryos (31.8% of all embryos recorded in our registry). Among these were five patients who had transfers at their local treatment centers under our protocols, after their local centers changed their policies. Demographic characteristics of patients in this study are reported in Table I. The average patient age at oocyte retrieval was 41.35 ± 3.98 years. Seventy-seven percent of the 35 patients who had experienced prior pregnancies, had previously experienced at least one miscarriage. Sixty-six percent of biopsies were performed on embryos after Day 5 post-fertilization. There was no difference in patient characteristics between those patients whose transfers resulted in a live birth live-birth, miscarriage or failed implantation except for two categories. Patients with failed implantation had higher BMI (P = 0.031) and those patients in our cohort with miscarriage had experienced a shorter interval trying to conceive (P = 0.005).

Table I

Demographic characteristics of patients.

N	Live birth	Miscarriage	No implantation	Total
	7a	10a	33	50
Age at retrieval—Median (quartiles)	40.9 (35.7, 42.3)	41.7 (38.2, 44.7)	42.3 (39.4, 44.4)	42.0 (39.3, 43.9)
Age at embryo transfer	42.1 (36.5, 44.3)	43.3 (40.5, 45.7)	43.7 (41.0, 46.4)	43.5 (40.9, 46.0)
Biopsy day	5.0 (5.0, 6.0)	6.0 (6.0, 6.0)	6.0 (5.0, 6.0)	6.0 (5.0, 6.0)
Biopsy day > 5, N (%)	2 (29)	9 (90)	33 (67)	33 (66)
Gravidity	2.0 (0.0, 2.0)	1.0 (0.75, 3.0)	1.0 (0.0, 3.0)	1.0 (0.0, 3.0)
Parity	1.0 (0.0, 2.0)	0.0 (0.0, 1.0)	0.0 (0.0, 1.0)	0.0 (0.0, 1.0)
Prior pregnancy ≥ 1, N	5	8	22	35
Prior miscarriagesb  ≥ 1, N (%)	4 (80)	6 (75)	17 (77)	27 (77)
Prior miscarriagesb	1.0 (0.5, 1.0)	1.0 (0.25, 1.75)	1.0 (0.75, 3.0)	1.0 (1.0, 2.0)
BMIc	21.9 (21.1, 23.0)	20.6 (19.9, 26.5)	24.6 (22.2, 26.9)	23.4 (21.3, 26.5)
FSH mIU/ml	7.7 (7.0, 10.3)	8.5 (6.4, 17.6)	11.5 (8.0, 14.2)	10.3 (7.2, 14.2)
AMH ng/ml	0.55 (0.3, 3.4)	1.25 (0.1, 2.0)	0.58 (0.1, 1.1)	0.58 (0.1, 1.4)
Trying to conceive (months)d	34.0 (12.0, 36.0)	12.5 (8.3, 38.4)	46.0 (25.5, 54.0)	36.5 (22.75, 48.5)
Previous IVF attempts	2.0 (2.0, 6.0)	2.0 (1.75, 4.0)	3.0 (1.5, 6.5)	3.0 (2.0, 5.3)
Fertility history, N (%)
Tubal factor	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Endometriosis	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Male factor	1 (14.3)	1 (10.0)	0 (0.0)	2 (4.0)
Ovarian	6 (85.7)	9 (90.0)	32 (97.0)	47 (94.0)
Combined	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Unexplained	0 (0.0)	0 (0.0)	1 (3.0)	1 (2.0)
Race, N (%)
Asian	1 (14.3)	0 (0.0)	5 (15.2)	6 (12.0)
Black	0 (0.0)	1 (10.0)	4 (12.1)	5 (10.0)
Hispanic	0 (0.0)	0 (0.0)	2 (6.1)	2 (4.0)
White	6 (85.7)	9 (90.0)	22 (66.7)	37 (74.0)

a

One patient had two live births; another patient had two pregnancy losses.

b

Per patient with at least one prior pregnancy.

c

P = 0.031.

d

P = 0.005.

IVF cycle outcomes after transfer of chromosomal—‘abnormal’ embryos

Tables II and III present details of individual transfers that led to pregnancies (cycles that did not lead to pregnancy are presented in Supplementary Table SI). These tables include 32 cases previously reported in less detail in a first interim report of the registry (Yang et al., 2021). All but two (Patient #9 and #10, Table III) of reported pregnancies occurred after transfers of embryos with only one to two abnormalities. Two exceptions, both embryos with three abnormalities, miscarried.

Table II

IVF cycles that resulted in live births.

Pt#	Age at biopsy (years)	Cycle	Embryo	Lab	Platform	Year of biopsy analysis	PGT-A diagnosis	Karyotype at outcome	IVF outcome	Outcome date
1	42	1st		ReproGenetics
			1st		aCGH	2015	XY, +12[mos]a	46, XY	Delivery	24 August 2018
			2nd		aCGH	2015	45, XX, -22		—
2a	35	1st		ReproGenetics
			1st		NGS	2016	XY, -6 [mos]	46, XY	Delivery	3 December 2017
2b		2nd		ReproGenetics
			1st		NGS	2017	XY, -17[mos]	46, XY	Delivery	27 December 2019
3	41	1st		ReproGenetics
			1st		NGS	2018	46, XY, del (15) (q24.1-qter)	46, XY	Delivery	21 March 2019
			2nd		NGS	2018	43, XX, -2, -14, -18		—
4	41	1st		Cooper Genomics
			1st		NGS	2018	XX, del (15) (pter-q21.1) [mos]	46, XX	Delivery	13 November 2019
			2nd		NGS	2018	XX, +19 (mos)		—
			3rd		NGS	2018	XY, del (5) (pter-q22.3) [mos], dup (20) (pter-p11.23) [mos]		—
5	36	1st		Cooper Genomics
			1st		NGS	2018	XY, dup (10) (q11.21-q21.1) [mos]	46, XY, dup (10) (11.21–11.23)	Delivery—coarctation	16 May 2020
			2nd		NGS	2018	44, XY, -11, -22		—
			3rd		NGS	2018	44, XY, del (2) (q23.3-qter), -17, -22		—
6	44	1st		Florida Inst Reprod Med Med
			1st		NGS	2019	XX -2p [mos]	46, XX	Delivery	20 March 2020
7	41	1st		Igenomix
			1st		NGS	2017	46 XX +14 -18	46, XX	Delivery	26 February 2021

a

Total chromosome counts are not displayed for biopsies reported as mosaic.

aCGH, Array Comparative Genomic Hybridization; NGS, next generation sequencing; Pt, patient.

Table III

IVF cycles that resulted in miscarriage.

Pt#	Age at biopsy (Years)	Cycle	Embryo	Lab	Platform	Year of biopsy analysis	PGT-A diagnosisa	Karyotype at outcomeb	IVF outcome	Outcome date
1	36	1st		Igenomix
			1st		NGS	2018	XX, +14 [mos]	Declined genetic testing	Miscarriage at 7w 2d	23 April 2019
			2nd		NGS	2018	45, XX, -11
2	39	1st		ReproGenetics
			1st		aCGH	2016	XX, + 9 [mos]	46, XX	Miscarriage at 9w 2d	10 December 2016
			2nd		aCGH	2016	XY, +11 [mos], +22 [mos]	—	—
3	43	1st		InVitae
			1st		NGS	2018	47, XX, +15	47, XX, +15	Miscarriage at 7w 6d	28 May 2019
			2nd		NGS	2018	46, XY, dup (8) (q24.2q24.3)	—	—
4	38	1st		Natera Spectrum
			1st		SNP	2014	47, XX, +12	47, XX, +12	Miscarriage at 10w 0d	21 December 2018
			2nd		SNP	2014	45, XX, -22	—	—
5	40	1st		Cooper Genomics
			1st		NGS	2019	XY, -6 [mos]	46, XYc	Miscarriage of twins at 18 weeks post-amniocentesis (+an empty sac)	27 May 2020
			2nd		NGS	2019	47, XY, +16	46, XY
			3rd		NGS	2019	XX, +1 [mos], -16
			4th		NGS	2019	No DNA	—	—
6	48	1st		Cooper Genomics/RGId
			1st		NGS	2019	XY, dup (10) (q11.21-q11.22) [mos]	46, XY	Miscarriage at 18 weeks sepsis post-amniocentesis	15 March 2020
		2nd		Cooper Genomics/RGI
			1st		NGS	2019	XY, del (7) (q31.2-qter), dup (7) (pter-q31.1) [mos]
			2nd		NGS	2019	45, XX, -21	Declined genetic testing	Twin pregnancy miscarriage at 9w 0d	22 November 2020
			3rd		NGS	2019	XX, dup (10) (q11.21-q11.23) [mos], +11
			4th		NGS	2019	XY, del (7) (q31.33-qter) [mos], +22 [mos]
7	38	1st		Genesis Genetics
			1st		NGS	2016	XY, -4 [mos], -18 [mos]	Declined genetic testing	Miscarriage at 6w 3d	11 April 2020
			2nd		NGS	2016	45, XX, -21
8	44	1st		Igenomix
			1st		NGS	2018	47, XX, + 16	47, XX, + 16	Miscarriage at 7w 6d	27 January 2020
			2nd		NGS	2018	44, XY, -14, -15	—	—
			3rd		NGS	2018	45, XY, -10	—	—
			4th		NGS	2019	44, XY, - 18, -19	—	—
9	52	1st		CCRM
			1st		NGS	2019	49, XY, +8, +13, +15	49, XY, +8, +13, +15	Miscarriage at 7w 4d	26 February 2020
			2nd		NGS	2019	48, XX, +16, +19	—	—
10	44	1st		Igenomix
			1st		NGS	2019	50, XX +1, +8, +10, +11	47, XX, +10	Miscarriage at 6w 4d	14 December 2020
			2nd		NGS	2019	45, XX, -4	—	—
			3rd		NGS	2019	46, XX +11, -15	—	—

a

Total chromosome counts are not displayed for biopsies reported as mosaic.

b

All genetic testing of products of conception was performed with single-nucleotide polymorphism-based microarray analysis with parental controls (NATERA ANORA).

c

DNA zygosity testing identified these as euploid male dizygotic fraternal twins.

d

These embryos were biopsied again 1 month later and the biopsy processed in a different lab. Biopsies at the two labs differed in reporting of segmental duplications and deletions.

aCGH, Array Comparative Genomic Hybridization; d, days; NGS, next generation sequencing; Pt, patient; SNP, single-nucleotide polymorphism-based microarray analysis; w, weeks.

Pregnancy, miscarriage and live birth rates per cycle were respectively, 19/57 (33.3%), 11/57 (19.3%) and 8/57 (14.0%). Per patient, these rates were 19/50 (38.0%); 11/50 (22.0%) and 8/50 (16.0%). Pregnancies included 9/19 first-trimester losses (47.4%, including 1 twin gestation) and 2/19 (10.5%) second-trimester losses shortly after amniocentesis (including 1 twin gestation). In all 11/19 clinical pregnancies (57.9%) ended in miscarriage (Table III). Three patients declined to have genetic testing of their products of conception (POC). The remaining eight pregnancies representing nine gestations (one set of twins) underwent genetic testing of their POCs using SNP. Of these nine lost gestations four were sex concordant but euploid. Four other lost gestations had genetic findings that were concordant with their previous PGT-A biopsy result. One additional loss was aneuploid 47 XX, +10 but was not entirely concordant with the prior PGT-A biopsy result (50, XX +1, +8, +10, +11). Thus, only four of nine (44%) lost gestations with genetic diagnosis of their POCs were entirely concordant with the previous PGT-A biopsy and four others were diagnosed as euploid. The 16-week twin pregnancy loss was further identified by DNA zygosity testing as euploid male dizygotic fraternal twins.

All deliveries were sex-concordant with transferred embryos. Highest ongoing pregnancy/live birth rates (6/8, 75.0%) were achieved with transfers of embryos that were reported to have only a single PGT-A diagnosed ‘mosaic’ chromosomal abnormality.

Specific aneuploidies and their relationship to establishment of pregnancies/live births

The chromosomal composition of 141 transferred embryos is summarized in Table IV: among 141 transferred embryos 102 (72.3%) had only non-complex aneuploidies with 1 or 2 chromosomal abnormalities, 30 (21.3%) had 3 or more abnormalities and 9 (6.4%) were ‘undiagnosed’ because of degraded DNA.

Table IV

Outcome summary for all embryos.

	All	Transferred	%	Pregnancies*	%	Pregnancy loss	%	Live births	%
Patients with PGT-a abnormal embryos	69	50	72.5%	19	38.0%	11	22.0%	8	16.0%
Patient cycles	—	57		19	33.3%	11	19.3%	8	14.0%
Embryos	444	141	31.8%	22	15.6%	14	9.9%	8	5.7%
Non-complex abnormal (2 or less)		102	72.3%**	19	90.4%	12	85.7%	8	100.0%
Full chromosomal abnormalities		76	74.5%***	5	26.3%	4	33.3%	1	12.5%
Non-complex mosaic		15	14.7%	9	47.4%	5	41.7%	4	50.0%
Non-complex segmental		3	2.9%	1	5.3%	0	0.0%	1	12.5%
Non-complex-segmental-mosaic		8	7.8%	5	26.3%	3	25.0%	2	25.0%
Complex abnormal (>2)		30	21.3%	2	9.5%	2	14.3%	0	0.0%
Undiagnosed (insufficient DNA)		9	6.4%	0	0.0%	0	0.0%	0	0.0%

*

Seventeen patients became pregnant. Two patients were pregnant twice. Nineteen pregnancies with 22 gestations (one twin pregnancy and one triplet pregnancy).

**

102/141 = 72.3%.

***

76/102 = 74.5%.

Among 102 embryos transferred with non-complex aneuploidies, 76 had full chromosomal abnormalities (monosomies, trisomies) of which 5 (6.6%) resulted in pregnancy, 4 ended in first-trimester loss and only 1 in live birth of an euploid female child. An additional 26 non-complex aneuploid embryos had been defined by PGT-A as ‘mosaic’ (under PGDIS criteria) and/or as exhibiting segmental copy number variants. Transfer of these embryos led to 14 pregnancies, 5 pregnancies resulted in first-trimester losses (1 with euploid POC), 2 as second-trimester losses (both with euploid fetuses) and 7 (30.4%) resulted in a live birth. Of 30 embryos transferred with complex (>2 chromosomal abnormalities), 28 had no evidence of implantation and 2 resulted in a first-trimester loss. Eight transferred embryos with no DNA signal had no evidence of implantation.

No pregnancy had to be voluntarily terminated. One male embryo with a duplication in chromosome 10 led to pregnancy (#5, Table III). The duplication was not detected in initial routine amniocentesis and karyotyping but was later discovered with a planned SNP which was performed because the treating physicians were aware of the previous PGT-A biopsy diagnosis of a duplication. After genetic counseling, the parents decided to continue the pregnancy even after the fetus was later in utero diagnosed with a coarctation of the aorta. He immediately after birth underwent corrective surgery and is developing normally.

In summary, all eight live births resulted from embryos with only one or two aneuploidies and six of those were under PGT-A PGDIS definition ‘mosaic’.

Discussion

This series of 57 sequential transfers of PGT-A abnormal embryos, previously refused transfers, resulted in births of 8 healthy children, one requiring surgical correction of coarctation of aorta. The losses of two additional euploid pregnancies affecting three fetuses may have been preventable since they occurred post-amniocentesis. Observing the advanced maternal age at the time of retrieval (41.35 ± 3.98 years), large numbers of previous failed IVF cycles (median of 3 per patient), low prior pregnancy rates, high previous miscarriage rates and, finally, a sizable percentage of Days 6 and 7 blastocysts (66.0%), the registry so-far represents mostly poor prognosis patients (Du et al., 2018).

These IVF cycle outcomes, therefore, must be considered surprisingly good. These patients were also judged at their local IVF centers to be poor prognosis patients as best demonstrated by almost all, prior to moving their embryos to our center, being advised that third-party egg donation was their only realistic remaining treatment option.

Most live births in this cohort occurred after transfers of embryos defined as ‘mosaic’ and/or segmental aneuploidies. Among 19 clinical gestations, 17 had genetic diagnosis at either live birth or pregnancy loss. Aneuploid pregnancy losses (7/17, 41.2%) could be considered low given that all transferred embryos had been labeled as ‘chromosomal-abnormal’ and, therefore, unfit for transfer. Euploid pregnancy losses in this cohort after amniocentesis (at different institutions) were unacceptably high. We, therefore, no longer mandate invasive prenatal pregnancy diagnosis by chorionic villus sampling or amniocentesis after transfers of ‘chromosomal-abnormal’ embryos, though, we still recommend such diagnostic procedures.

By early 2019, worldwide over 400 chromosomal-normal children were reported born following transfers of allegedly ‘chromosomal-abnormal’ embryos (Patrizio et al., 2019). In good prognosis patients, investigators reported 50% ongoing clinical pregnancy rates following transfers of embryos with single aneuploidies, and still a 10% rate with complex chromosomal abnormalities involving three or more aneuploidies (Munné et al., 2017). Mosaicism of up to 80% aneuploid DNA in a single biopsy was reported to result in an ongoing embryo implantation rate of 27% (Munné et al., 2020). These findings agree with another study in which positive outcomes were achieved after transfer of mosaicism involving segmental abnormalities to complex aneuploidies affecting three or more chromosomes (Viotti et al., 2021b). The STAR study which recently received wide attention, did not allow transfer of mosaic embryos (Munné et al., 2019), likely worsening the chance of pregnancy for participating patients.

Why most IVF centers, still, don’t routinely transfer selected embryos determined by PGT-A as ‘chromosomal—abnormal’, especially those with mosaic trophectoderm biopsy, is, therefore, difficult to understand. Restricting the diagnosis to only a few cells, makes a ‘mosaicism’ diagnosis a relatively rare finding, explaining why, even to this day, some proponents of PGT-A erroneously consider ‘mosaicism’ in humans to be an unusual and rare phenomenon (Capalbo and Rienzi, 2017). However, if multiple trophectoderm biopsies were taken from the same embryo, the possible diagnosis of mosaicism would, of course, increase.

PGT-A reporting may also exaggerate findings. Assuming a biopsy at random is only derived from a small clonal island of aneuploid cells, the analysis may demonstrate 100% aneuploid DNA and the embryo’s PGT-A diagnosis will, therefore, be ‘aneuploid’. If the rest of the embryo, however, either to minor or major degrees is euploid, under international consensus, this embryo would, despite a 100% ‘aneuploid’ biopsy, be considered truly mosaic. Zernicka-Goetz’s laboratory in Cambridge unsurprisingly identified in human embryos cultured in vitro beyond implantation stage, three mosaic embryos which at earlier blastocyst stage biopsy during PGT-A had been diagnosed as aneuploid (Shahbazi et al., 2020).

As Table IV demonstrates, best chance of pregnancy in the registry was accomplished after transfer of embryos with only ‘mosaic’ and/or segmental chromosomal abnormalities, as based on PGDIS criteria. If no euploid embryos are available, one could establish a hierarchy for transfer, favoring embryos with ‘mosaic’ and segmental abnormalities, followed by single monosomy or single selected trisomy (Viotti et al., 2021a,b). Moreover, the more abnormal cell lineages are detected in a single biopsy sample, the poorer chances will be for implantation. No embryo with more than two abnormal cell lineages detected by PGT-A in the present cohort, indeed, led to live birth suggesting that in some cases a PGT-A diagnosis may have validity.

The number of aneuploid cell lineages must, however, be differentiated from the percentage of aneuploid DNA, which has been said to be predictive of chances for embryo implantation (Scott and Galliano, 2016; Munné et al., 2017). Munné et al. (2017) reported that embryos with 40–80% aneuploid DNA had a lower ongoing implantation rate than mosaic embryos with <40% aneuploid DNA. However, we later reported (Kushnir et al., 2018) that, based on recalculation of that study’s original data, that there was no significant difference in ongoing pregnancy or miscarriage rates among mosaic embryo transfers at any threshold. Others reached a similar conclusion finding that aneuploid cell percentage in trophectoderm biopsies does not correlate with outcomes, but that the type of mosaicism did (Victor et al., 2019). Chromosomal ‘abnormal’ embryos with best chances of implantation, therefore, do not appear defined by percentage of aneuploid DNA within a single trophectoderm biopsy but by number of aneuploid cell lineages seen in a single biopsy: the more there are, the less likely the result represents a false-positive and the less likely, therefore, self-correction will occur.

Chromosomal aneuploidies can be of meiotic or mitotic origin. If arising during meiotic divisions, all cells in the embryos will be aneuploid, representing the only circumstance when a single trophectoderm biopsy will offer an undeniably correct diagnosis. It is also the only circumstance unlikely to be remedied by an embryo’s ability to self-correct downstream. However, most mosaic aneuploidies in human embryos are of mitotic origin and, therefore, are clonal, producing only aneuploid islands of cells (van Echten-Arends et al., 2011; Capalbo et al., 2017; Starostik et al., 2020). Whether such aneuploid cells will be included in a 5-cell trophectoderm cell biopsy, will, therefore, be determined by chance. Failing to recognize the potential for sampling error in a single trophectoderm biopsy, can lead to underestimating the prevalence of ‘mosaicism’ in blastocyst-stage human embryos to be as low as 5% (Capalbo and Rienzi, 2017).

Until 2016 most diagnostic platforms used for PGT-A were unable to detect more than one cell lineage. Consequently, they were unable to detect ‘mosaicism’ (under PGDIS criteria). Since 2016, multiple cell lineages and segmental copy number variants can now be detected with greater sensitivity using NGS and aCGH (Mertzanidou et al., 2012). Yet even NGS detects additional lineages only at concentrations above 20% of total DNA. Biopsies with lower percentages of aneuploid DNA, therefore, are reported as normal-euploid. Thus, an embryo reported by PGT-A as normal-euploid, may still, in fact, be mosaic.

Mosaicism is not only common in human embryos. It represents a quite common phenomenon throughout human life and can involve whole chromosomes, structural or copy-number variants, small or single nucleotide variants or even epigenetic variants (Martínez-Glez et al., 2020). Especially in brain and liver tissues these genetic variants can be found into adult age (Bizzotto et al., 2021; Fasching et al., 2021). Early post-zygotic events leading to genetic variants can be preserved in the inner cell mass of early embryos and, thereafter, in clones of the resulting germ layers. Thus, minor chromosomal abnormalities, including small duplications, deletions or insertions in the normal genetic code exist in humans in abundance. Though many of these appear to be benign, some have been associated with human disorders, including syndromes of intellectual disability (Viñas-Jornet et al., 2018; Truty et al., 2019). Segmental deletions are generally more pathogenic than segmental duplications (Lek et al., 2016) and more likely to lead to implantation failure. These more universal considerations speak against efforts at, even at preimplantation stages, of eliminating all such embryos from transfer.

After the 2016 guidance document from the PGDIS mandating NGS utilization in PGT-A reporting switched from a binary (normal-euploid versus abnormal-aneuploid) to a trinary system (normal-euploid, ‘mosaic’ and abnormal-aneuploid). Consequently, the interpretation of PGT-A results became more complex and, indeed, for many IVF centers too complex and confusing. Despite attempts by professional societies to assist in the interpretation of ‘mosaic’ PGT-A results (Cram et al., 2019; Practice Committee and Genetic Counseling Professional Group (GCPG) of the American Society for Reproductive Medicine, 2020), in the USA a substantial number of IVF centers request that PGT-A laboratories revert to reporting only the original binary classification of normal-and abnormal-aneuploid (Munné et al., 2019). Different genetics laboratories use different limits in calling an embryo mosaic. Laboratories with narrower limits will designate fewer embryos as mosaic. Genetics laboratories also have different policies for which chromosomes or copy number variants they will report as mosaic. This lack of standardization further complicates interpretation of PGT-A reported results.

While current diagnostic PGT-A techniques appear to be technically precise, NGS results vary nevertheless by 11% to 25.7% in interpretation (Surrey, 2021; Viotti et al., 2021a,b). A recent editorial suggested the complete abandonment of the term ‘mosaicism’ (under PGDIS criteria) and replacing it with the phrase ‘intermediate copy numbers’ (Paulson and Treff, 2020). Such a change in nomenclature would recognize that mosaicism represents a much more common phenomenon than has been before appreciated; a simple change in nomenclature, however, while technically correcting the misuse of the term mosaicism, is not likely to change clinical practice. That would require the recognition that euploid/aneuploid embryos with ‘intermediate copy numbers’ can still lead to a viable euploid pregnancies.

Several authors have reported that over 80% of embryos contain at least some aneuploid cells at preimplantation stages (van Echten-Arends et al., 2011; Capalbo and Rienzi, 2017; Starostik et al., 2020). In fact, chromosomal abnormalities (aneuploidy) in preimplantation-stage embryos are now considered so common that some authors find they must be viewed as a normal, physiological presence (Angell et al., 1983; McCoy, 2017). As the presence of aneuploid cells is such a common finding in the peri-implantation period, one also must wonder about its physiological purpose (Gleicher et al., 2021).

How complex the interpretation of PGDIS-defined mosaic results can be, was well demonstrated by two patients in this cohort who, against our advice, requested repeat trophectoderm biopsies of their embryos. In each case, segmental copy number variants and low order mosaicisms were additional new reported findings in the second round of biopsies.

Limitations and conclusions

A limitation of this study is its relative short observation period. To establish the complete cumulative outcome potential for embryos defined by PGT-A as chromosomal-abnormal, using all embryos available, will therefore be an important next step, as over half of embryos brought to our center for transfer are still cryopreserved. This means that the final cumulative pregnancy and delivery potential for this patient cohort could change after more embryos have been transferred. A recent manuscript published as a pre-print found no significant difference in live birth rates of around 40% among single embryo transfers of euploid vs those with <50% mosaicism.

Because of the short post-natal observation in this cohort some abnormal phenotypes may not be apparent in the immediate postnatal period. In view of this fact, it will be important to continue to follow the development of these children.

That these couples came from IVF centers all over the USA and Europe, represents a strength of this study, as transferred embryos were produced in many different IVF centers throughout the USA and Europe, utilizing a large variety of IVF protocols. In this sense, these outcomes may be considered generalizable.

Embryos transferred in this cohort were biopsied over years in which the technology of DNA analysis was steadily changing. As noted above, this improved resolution has potentially complicated interpretation of the biopsy results and, no matter how the technology is improved, it cannot compensate for a potential sampling error. Analysis of spent culture medium has been offered as a way to address the sampling error problem, but that method too, has both advantages and potential problems (Rubio et al., 2021).

A further limitation is that this patient population is limited to patients who actively sought to have their embryos transferred after previously refused transfer at their parent institution, this population may be subject to bias based on the strong motivation of these patients to seek transfer of their embryos and may not reflect the population in general.

Surprisingly, excellent IVF outcomes following transfers of allegedly chromosomal-abnormal embryos have previously been reported (Gleicher et al., 2015; Patrizio et al., 2019; Munné et al., 2020; Viotti et al., 2021a,b), but mostly in only good-prognosis patients. This study demonstrates that, even in poor prognosis patients, transfer of chromosomal-abnormal embryos resulting in a normal pregnancy is possible. Even older women who are often told that all their embryos are chromosomal-abnormal, still, appear to have a reasonable chance of normal pregnancy. That conclusion alone, should rescue many such women from having to prematurely advance into third-party egg donation cycles, and rescue large numbers of human embryos from non-use or even disposal. Considering the increasing utilization of PGT-A worldwide, large numbers of human embryos with still reasonable pregnancy and live birth chances are regrettably disposed of or condemned to permanent cryopreservation.

The Society for Assisted reproductive technology (SART) preliminary data reveals that 129 886 of 293 672 (44.2%) of all ART cycles performed in 2019 used preimplantation genetic testing (of which a minority were PGT-M). Especially in older patients, when PGT-A is used, fewer patients will undergo embryo transfer (Rubio et al., 2017). Clinical evidence supports the conclusion that even in the case of embryos with mosaic biopsy, fewer than 3% are selected for transfer (Munné et al., 2017). Thus, at least some embryos not transferred because of PGT-A results would be mosaic and would have had good potential for implantation and euploid live-birth (Viotti et al., 2021a,b). Thus, it seems that after the expense of an IVF cycle and added expense of PGT-A, patients who produce only a few embryos are being denied transfer of embryos not determined to be fully euploid.

PGT-A may be a useful method of embryo selection when there are sufficient embryos to select from a large embryo cohort. However, PGT-A appears to have a poor track record when used to exclude embryos from transfer, especially in the case of those embryos defined by PGDIS criteria as ‘mosaic’ or with segmental copy number variants when no other embryos are available for transfer.

Supplementary data

Supplementary data are available at Human Reproduction online.

Data availability

The data underlying this article are available in the article and in its online supplementary material.

Authors’ roles

D.H.B.: manuscript conception and design, acquisition of data, and analysis and interpretation of data, manuscript writing, literature search and critical discussions. D.F.A.: manuscript conception and design, literature search and critical discussions. E.M.: acquisition of data and interpretation of data critical discussions, and critical revisions. N.G.: manuscript conception and design, analysis and interpretation of data, manuscript writing, literature search and critical discussions. All authors revised and approved the final version of the manuscript and agree to be accountable for all aspects of the work.

Funding

This work was supported by intramural funds from the Center for Human Reproduction and the not-for-profit research Foundation for Reproductive Medicine, both in New York, NY, USA.

Conflict of interest

N.G. and D.H.B. are listed as co-inventors on several US patents. One of these patents (US Patent# 7,615,544) relates to pre-supplementation of hypo-androgenic infertile women with androgens, such as DHEA and testosterone and, therefore, at least peripherally related to the subject of this manuscript. N.G. and D.F.A. also received travel funds and speaker honoraria from several pharmaceutical and medical device companies, though none related to the here presented subject and manuscript. N.G. is a shareholder in Fertility Nutraceuticals and he and D.H.B. receive royalty payments from Fertility Nutraceuticals LLC.