Mosaic embryo transfer versus additional IVF with PGT-A Cycle: a decision model comparing live birth rate and cost

Abstract

Purpose

To evaluate the relative live birth rate and net cost difference between mosaic embryo transfer and an additional cycle of IVF with PGT-A for patients whose only remaining embryos are non-euploid.

Methods

A decision analytic model was designed with model parameters varying based on discrete age cutoffs (<35, 35–37, 38–39, 40–42, 43–44, >44). Model inputs included probabilities of successful IVF, clinical pregnancy, and live birth as well as costs of IVF with PGT-A, embryo transfer, live birth, amniocentesis, and dilation and curettage. All costs were modeled from the healthcare system perspective and adjusted for inflation to 2023 $USD. Model outcomes were sub-stratified by degree and type of mosaicism.

Results

For patients younger than 43, an additional cycle of IVF with PGT-A resulted in a higher relative live birth rate (<35, +20%; 35–37, +15%; 38–39, +17%; 40–42, +6%; average, +14.5%) compared to mosaic embryo transfer with an average additional cost of $16,633. For patients older than 42, mosaic embryo transfer resulted in a higher live birth rate (43–44, +5%; >44, +3%; average, +4%) while on average costing $9572 less than an additional cycle of IVF with PGT-A.

Conclusion

Mosaic embryo transfers are a superior alternative to an additional cycle of IVF with PGT-A for patients older than 42 whose only remaining embryos are non-euploid. Mosaic embryo transfers also should be considered for patients younger than 42 who are unable to pursue additional autologous IVF cycles. Counseling and care should be personalized to individual patients and embryos.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-024-03027-7.

Introduction

Mosaicism is defined as the simultaneous presence of two or more cell lines with different chromosomal content from one individual [1]. The use of next-generation sequencing (NGS) platforms for pre-implantation genetic testing (PGT) has allowed for more sensitive detection of embryonic mosaicism; current NGS platforms can detect mosaicism when as little as 20% of the biopsied cells are mosaic [2]. As a result, approximately 7% (up to 30%) of all trophectoderm (TE) biopsies are now reported as mosaic [2, 3]. However, there are significant biological and analytical pitfalls in diagnosing an embryo as mosaic based on a TE biopsy and such labeled embryos can in reality be euploid, aneuploid, or mosaic [4].

Despite these limitations, mosaic embryos have been transferred and resulted in successful pregnancies [5–10]. Conclusions from the limited pregnancy outcomes data of mosaic embryos indicate that there are likely higher rates of implantation failure and miscarriage for mosaic embryos compared to euploid embryos, but that overall live birth rates (LBR) may be similar [10–12]. The degree (<50% vs ≥ 50% mosaic cell lines) and type of mosaicism (segmental, whole chromosome, or complex) also appear to affect these outcomes; embryos with <50% mosaic cell lines (low-mosaic) and those with segmental—as opposed to whole-chromosome or complex—mosaicism have better outcomes [8, 13]. To date, while there has been a single reported case of a congenital anomaly in an infant born from a mosaic embryo transfer (MET), there is no evidence to suggest that METs have a higher risk of congenital anomalies that euploid embryo transfers [10, 14]. No long-term data on developmental outcomes of children born from MET is available yet.

The lack of conclusive data on the safety of MET presents a challenge for clinicians and for patients. Particularly, for patients whose only available embryos are non-euploid, the decision to transfer a mosaic embryo or to pursue an additional cycle of IVF with PGT-A is not clear. Our objective is to compare the relative LBR and net cost difference between these two choices for this specific population in a decision analysis. We hope our findings will aid clinicians in their counseling of patients whose only embryos are mosaic regarding whether to perform a MET.

Materials and methods

Model structure

A decision analytic model was created using TreeAge Pro Healthcare (2023R1.1) with two comparative arms, MET versus a cycle of IVF with PGT-A (Fig. 1). A typical global discount rate of 3% was applied [15]. All patients were assumed to have only non-euploid embryos available prior to undergoing simulation in this model.

Fig. 1.

Decision analytic model. Schematic of the model used to evaluate the live birth rate and cost difference between MET and IVF with PGT-A. Rhombuses denote decision nodes. Hexagons denote terminal nodes

For the MET arm, it was assumed that a single mosaic embryo would be transferred that would either result in a clinical pregnancy or be a failed transfer. If a clinical pregnancy occurred, then it would either result in a livebirth or a miscarriage. If a clinical pregnancy did not occur, this was assumed to be a failed transfer; biochemical pregnancies were considered failed transfers. Reported live birth rates excluded ongoing pregnancies. All clinical pregnancies resulting from MET underwent amniocentesis.

For the IVF with PGT-A arm, it was assumed that only a single cycle and a single embryo transfer would be attempted. The model first simulated whether the cycle was successful, defined as having at least one euploid embryo available for transfer. If a euploid embryo was transferred, then it would either result in a livebirth or a miscarriage. If no transfer occurred, then the cost of the euploid embryo transfer was refunded by the model.

For both arms, all miscarriages were assumed to be treated with a dilation and curettage (D&C). This was a conservative assumption to maximize the potential cost of miscarriage management in both arms of the model.

The model was designed to sum the cumulative costs and probabilities for each node and report a total cost and net livebirth for each iteration.

A separate, but identical, decision model was created based on the age group of the patient. The age groups were defined as less than 35, 35–37, 38–39, 40–42, and 43–44, and greater than 44.

Model parameters and data sources

The probability parameters used in the model, as well as their data sources, are listed Table 1. Cost parameters are listed in Table 2.

Table 1.

Model probability parameters

Probabilities by age
Age	Probability	Baseline	Standard deviation	Source
<35	Successful IVF with PGT-A1 (n=808)	0.92	0.020	Institutional data
	Clinical pregnancy—mosaic	0.50	0.038	Viotti [8]
	Livebirth—mosaic	0.27	0.030	Viotti [8]
	Livebirth—euploid	0.39	0.027	Sanders [16] Munné [17]
35-37	Successful IVF with PGT-A1 (n=802)	0.84	0.075	Institutional data
	Clinical pregnancy—mosaic	0.61	0.055	Viotti [8]
	Livebirth—mosaic	0.31	0.038	Viotti [8]
	Livebirth—euploid	0.41	0.029	Sanders [16] Munné [17]
38–39	Successful IVF with PGT-A1 (n=483)	0.69	0.080	Institutional Data
	Clinical pregnancy—mosaic	0.43	0.045	Viotti [8]
	Livebirth—mosaic	0.23	0.024	Viotti [8]
	Livebirth—euploid	0.40	0.020	Sanders [16] Munné [17]
40–42	Successful IVF with PGT-A1 (n=438)	0.49	0.061	Institutional data
	Clinical pregnancy—mosaic	0.52	0.041	Viotti [8]
	Livebirth—mosaic	0.21	0.030	Viotti [8]
	Livebirth—euploid	0.34	0.043	Sanders [16]
43–44	Successful IVF with PGT-A1 (n=238)	0.21	0.026	Institutional data
	Clinical pregnancy—mosaic	0.47	0.038	Viotti [8]
	Livebirth—mosaic	0.19	0.033	Viotti [8]
	Livebirth—euploid	0.22	0.028	Sanders [16]
>44	Successful IVF with PGT-A1 (n=116)	0.09	0.011	Institutional Data
	Clinical pregnancy—mosaic	0.32	0.045	Viotti [8]
	Livebirth—mosaic	0.14	0.024	Viotti [8]
	Livebirth—euploid	0.17	0.021	Sanders [16]

¹Defined as having at least 1 euploid embryo available for transfer, per IVF-PGTA cycle started

Table 2.

Model cost parameters

	Baseline	Range	Source
Costs ($USD)
D&C	1457	957–1957	Murugappan [18]
Amniocentesis	1756	896–2616	Little [19]
IVF with PGT-A1	22,545	7427–376,63	Facadio Antero [20]
Livebirth2	15,385	5385–25,385	Facadio Antero [20]
Frozen embryo transfer	6395	3155–9635	Facadio Antero [20]

¹Includes cost of 1 FET

²Includes peripartum costs of delivery

Costs included in the model were that of a cycle of IVF with PGT-A with a frozen embryo transfer, a live birth, and a D&C. The listed cost of a cycle of IVF with PGT-A included a single frozen embryo transfer; this FET cost was refunded by the model if no euploid embryo was available for transfer. The live birth cost comprised of peripartum costs associated with delivery and was included to realistically represent the additional costs associated with a live birth compared to no pregnancy or a D&C after an embryo transfer. The cost of amniocentesis was also modeled for the MET arm. All costs were from the healthcare perspective, as is the standard with decision and cost-effective analyses; costs were adjusted for inflation using the Consumer Price Index to 2023 $USD. Healthcare perspective costs include all direct costs incurred in the delivery of medical care and excludes costs such as time, transportation, and other intangibles.

IVF with PGT-A outcome data was derived from 3130 cycles performed at a single academic center from 2016 to 2022. These cycles were restricted to those with PGT-A and excluded those using donor gametes. Euploid transfer outcomes were derived from those reported in Sanders et al. [16].

Analysis

A nested probabilistic sensitivity analysis (PSA) was performed to evaluate the model integrity and the effect of individual parameters on the final cost output. First a discrete Monte Carlo simulation was performed to simulate 10,000 patients undergoing IVF with PGT-A through this model. Each virtual patient was simulated through the model 100,000 times, with each iteration varying the value of model parameters using one-way sensitivity analyses for each parameter over a plausible range using 95% confidence intervals (CI), if available, or reported standard deviations.

The effect of including or excluding ongoing pregnancy rate (OPR) in the net LBR calculation was modeled by excluding OPR for the model’s baseline parameters and defining the upper limit of the CI in the PSA based on inclusion of the OPR in the calculated LBR.

All probabilities used in the model were represented by beta distributions and all cost parameters were represented by gamma distributions formed by their corresponding ranges in the PSA.

This work was exempt from review by our institutional review board since institutional data was provided to investigators in aggregate and de-identified format without the ability to link any data to individuals; all other data sources used in this work have been previously published.

Results

In the IVF with PGT-A arm, LBRs declined with advancing patient age from 33% for those less than 35 years ago to 2% for those older than 44 (Fig. 2A). The IVF with PGT-A arm had a higher LBR than the MET arm for patients <35, 35–37, 38–39, and 40–42 (Table 3). The LBR was higher for the MET arm for patients 43–44, and >44 (Table 3).

Fig. 2.

A Live birth rate by age. The live birth rate for each model strategy (MET vs IVF with PGT-A) is displayed for each defined patient age group. B Net cost by age. The net cost accrued for each model strategy (MET vs IVF with PGT-A) is displayed for each defined patient age group

Table 3.

Model results

Maternal age	Strategy	Total cost ($)	Live birth rate	Incremental cost ($)1	Incremental live birth rate1
<35	IVF-PGTA	28,011	0.33	+18,972	+0.20
	MET	9039	0.13
35–37	IVF-PGTA	27,609	0.34	+17,323	+0.15
	MET	10,286	0.19
38–39	IVF-PGTA	25,422	0.27	+16,721	+0.17
	MET	8701	0.10
40–42	IVF-PGTA	22,396	0.17	+13,515	+0.06
	MET	8881	0.11
43–44	IVF-PGTA	18,342	0.04	+9799	−0.05
	MET	8543	0.09
>44	IVF-PGTA	16,968	0.02	+9359	−0.03
	MET	7609	0.05

¹Base case considered to be MET and comparator to be IVF-PGTA

In the IVF with PGT-A arm, net costs decreased with advancing age ($28,011 for those <35 years old to $16,968 for those >44 years old) (Fig. 2B). The trend in cost is due to a lower likelihood of a euploid embryo available for transfer with advancing patient age; in the model, for patients in the IVF with PGT-A arm without a euploid embryo available for transfer, there was no cost for embryo transfer and subsequent D&C or live birth incurred. The results did not show any clear trends in the relative LBR and net cost for the MET arm across different age groups (Fig. 2).

Model outcomes were also stratified by categorizing mosaic embryos as high mosaic (≥ 50% aneuploid cells on TE biopsy) or low mosaic (<50% aneuploid cells on TE biopsy) as well as by type of mosaicism (whole-chromosome, segmental, or complex) derived from the dataset published by Viotti et al. [8] (Supplementary Table 1). The model outcomes did not significantly differ with this stratification except for patients between the ages of 40–42. When only considering MET for segmental mosaic embryos, the live birth rates for MET vs IVF with PGTA were equivalent (16%).

Discussion

Despite the availability of data regarding MET outcomes, the transfer of a frozen euploid embryo should be prioritized over the transfer of a mosaic embryo. However, for patients with only remaining non-euploid embryos, the choice of whether to transfer a mosaic embryo or pursue an additional cycle of IVF with PGT-A is challenging. These patients are open to the possibility of MET with appropriate counseling [21], but clinical scenarios in which MET would be an appropriate option have thus far been unclear. Our findings demonstrate that based on currently available outcomes data, the transfer of a mosaic embryo should be considered for patients with only remaining non-euploid embryos, particularly for those above the age of 42 for whom MET has a higher chance of live birth compared to an additional IVF-PGTA cycle.

Below the age of 43, our model showed that patients are more likely to have a live birth when pursing an additional IVF with PGT-A cycle (average incremental LBR 14.5%) while incurring an average additional cost of $16,633 compared to a MET. Patients below the age of 35 do not benefit from use of PGT-A to improve LBR [22], so when left with only non-euploid embryos, these patients would likely proceed with a non-PGT IVF cycle, and should be counseled as such. For patients between 35 and 40, PGT-A may be beneficial in improving their outcomes, but the available data is inconclusive [22, 23] and non-PGT tested IVF cycles are cost-effective compared to PGT-A cycles [20, 24]; thus, patients in this age group should be counseled individually whether to transfer an untested embryo or biopsy it with PGT-A. MET is still a reasonable option for patients in this age group, particularly if treatment cost is a limiting factor. A cycle of FET will always be cheaper than an IVF cycle and the temporal opportunity cost associated with potentially repeated failed METs may be less significant for young patients. This strategy also has the added benefit of reducing the total number of embryos held perpetually in cryo-storage.

Above the age of 42, the chance of a live birth is higher when pursuing a MET compared to a new cycle of IVF with PGT-A (avg. incremental LBR 4%) despite incurring, on average, $9572 less in costs. These patients should be counseled on the possibility of undergoing MET in line with current society guidelines [12, 25] despite the lack of long-term safety outcomes for MET. Although these patients may also elect for an IVF cycle without PGT biopsy, the rates of aneuploidy are above 75%, and thus untested embryos would be expected to perform poorly upon transfer [26].

Based on currently available data, the type and degree of mosaicism does not change the favorability of IVF with PGTA vs MET for any of the simulated age groups, except for those aged 40–42. In this age group, our model demonstrated that when considering the transfer of only segmental mosaic embryos, the overall live birth rate of MET was comparable to that of IVF with PGTA (16%) at an average cost savings of $13,515. Of note, this subtype of mosaic embryos (n=112) is a much smaller dataset than the overall cohort of mosaic embryos modeled here. Patients in this age group are less likely to transfer an untested embryo than they are to biopsy their embryos, given the reduction in miscarriage rate and time to pregnancy, assuming that they have untested embryos available for possible biopsy [27]. If these patients value cost and time to pregnancy above the higher miscarriage rate, then MET remains a reasonable option for them to consider if they have no euploid embryos remaining.

Our model’s strengths lie in utilizing the most recent and comprehensive MET outcomes data from Viotti et al. [8] divided into patient age groups with high granularity, particularly above the age of 40. IVF with PGT-A outcomes data was also matched to these age groups to provide an accurate comparison group. This is also the first decision model to evaluate the utility of MET compared to an additional cycle of IVF with PGT-A and can clearly demonstrate the merits of MET for patients older than 42.

The limitations of this study are related to the assumptions made in designing the model, which are inherent to any decision or cost-effectiveness model. We restricted the model to a single MET and a single cycle of IVF with PGT-A to limit the computational complexity of the model considering the limited available data on patient decision-making when faced with only remaining mosaic embryos. Up to 35% of such patients do pursue MET and another 7% pursue MET after a failed additional IVF cycle [21]. However, it is not known on average how many additional IVF cycles these patients pursue or how many MET they are willing to undergo before considering alternative options. Our model also exclusively considers autologous cycles and does not account for the possibility of patients pursuing donor oocyte IVF cycles or donor embryo transfer cycles when left with only none-euploid embryos; limited available data suggests that only 3% of these patients would elect for donor oocyte cycles [21].

Future directions for our work include adding a third arm to the model to account for patients who elect to pursue a non-PGT IVF cycle when their only remaining embryos are non-euploid. We also hope to match mosaic and euploid transfer outcomes to individual measures of ovarian reserve as an additional variable for consideration in the model given such measures are often integral to the counseling that patients receive from providers.

Conclusions

Our findings demonstrate that mosaic embryo transfer is a reasonable alternative to an additional cycle of IVF with PGT-A when the only remaining embryos are non-euploid. Controlled, experimental studies are needed to verify these findings clinically. Compared to IVF with PGT-A, MET results in a higher LBR for those above the age of 42 with an average $9572 of cost-savings. Younger patients may also benefit from MET, particularly if additional IVF cycles are cost prohibitive. We hope these findings aid providers in counseling patients on their options for mosaic embryos during their reproductive care.

Author contribution

Dr. Khorshid designed and built the model, performed the data analysis, and wrote the initial manuscript; Dr. Bavan aided in designing the model and assisted in data analysis as well as edited the manuscript; Dr. Chung provided the institutional data used in the model and edited the manuscript; Dr. Lathi was the principal investigator who assisted with overall study design and objectives, with data analysis, and with reviewing/revising the manuscript.

Data availability

Data will be made available to the editors of the journal for review or query upon request.

Declarations

Data regarding any of the subjects in the study has not been previously published unless specified.

Conflict of interest

Esther Chung is an advisory board member of Turtle Health. Ruth Lathi is an advisory board member of Biorad. All other authors have no relevant financial or non-financial interests to disclose.