There are multiple factors that need to be considered when analyzing euploid embryo transfer data and determining when PGT-A is clinically beneficial. One consideration is if PGT-A correctly identifies which embryos will progress to live birth. Although most PGT-A platforms have not performed clinical validation studies, clinical data from one center suggests that aneuploid embryos rarely progress to live birth [1]. Another consideration is if embryo biopsy causes a loss of implantations or live births. One group of investigators found that biopsy protocols can affect live birth rates [17]. It is likely that different biopsy protocols and embryologist experience contribute to variable rates of loss of implantations between clinics. At present, loss of implantations from embryo biopsy seems to be clinically insignificant at some centers and significant at others. This makes it challenging to interpret data from multicenter studies [10]. Centers with significant loss of implantations from embryo biopsy may be less likely to publish their data than other centers where the PGT-A data looks more favorable.
If there is some loss of implantations with embryo biopsy then we would anticipate a smaller than expected increase in ongoing pregnancy for embryos that are euploid by PGT-A testing. In good prognosis patients with high euploid rates, the benefit of selecting euploid embryos may be negated by loss of implantations from the biopsy procedure. However, if there is minimal loss of implantations from the biopsy procedure, we would anticipate increased ongoing pregnancy and live birth rates from euploid embryos at all ages. Looking at our overall analysis we found a 43% live birth rate for untested embryos. With PGT-A and no loss of implantations from the biopsy procedure we would expect this live birth rate to increase proportionally to the euploidy rate. Based on age, approximately 70% of the untested embryos in this study are expected to be euploid [4]. We expect a live birth rate of 0.43/0.70 or 61% for PGT-A tested embryos. Indeed 61% falls within our 95% CI for live birth rate per embryo for PGT-A tested embryos (Fig. 2). Embryo mosaicism factors somewhat into the equation but since less than 2% of embryos are reported as mosaic by our testing platform this has minimal impact on calculating the expected increase in live birth rate with PGT-A [4].
Based on this data, PGT-A is a cost-effective approach for our patients. Going from a 40% to a 60% live birth rate per embryo would decrease the average number of single embryo transfers needed to achieve a live birth from 2.5 to 1.7 transfers as shown in Fig. 3. At a cost of $4,000 per embryo transfer that would be a $3,200 savings on average per live birth. The current cost for embryo biopsy and PGT-A is approximately equal to the cost of one frozen embryo transfer cycle. Many good prognosis patients have more than 1 live birth per retrieval and they would have greater cost savings. Based on the current costs of embryo biopsy and PGT-A, the procedure seems approximately cost neutral at worst and cost effective at best for patients who have enough embryos for more than 1 live birth. Cost effectiveness studies based on data more than 5 or 10 years old or data from multiple centers may not be applicable to current practice at centers proficient in embryo biopsy and PGT-A [18].
There are some other considerations when deciding if PGT-A is a good option for a specific patient. Insurance may not cover PGT-A in younger patients due to lack of published literature supporting PGT-A use at younger ages. This may change with more clinical experience and more data being published each year. Although live birth rate per embryo is currently the main clinical consideration, euploid embryo transfer is also associated with decreased rates of spontaneous abortion and genetically abnormal pregnancy [1–3].
Limitations
Couples at our center who purchase 6 vitrified donor oocytes for a frozen oocyte embryo transfer cycle typically transfer 1 embryo fresh without PGT-A and cryopreserve any additional embryos. On the other hand, couples using fresh donor oocytes are more likely have PGT-A testing performed since there are often excess embryos expected and this testing helps select embryos for transfer. With fresh donor oocyte cycles, typically there are 10–20 oocytes retrieved which in general is expected to result in higher quality embryos than cycles starting with only 6 frozen donor oocytes. There may be some uncorrected confounding between PGT-A and use of fresh oocytes. Despite this, transfer of embryos from donor oocytes using PGT-A actually had worse morphology than embryos from donor oocytes not using PGT-A due to the deselection of morphology that occurs with PGT-A testing and culture of embryos to expansion stage 2 before embryo biopsy (Table 2). Regardless, prospective studies are needed from centers proficient in embryo biopsy and PGT-A to verify our retrospective data.
In this analysis we included use of embryos from fresh and frozen oocytes, fresh and frozen embryo transfer, transfers to gestational carriers, and single and double embryo transfers. Inclusion of these diverse types of transfers makes the data broadly applicable. However, this analysis is less controlled than ideal. Limiting the analysis to single frozen embryo transfers of embryos created from fresh oocytes would add more control but based on our dataset the numbers would have been too small to have sufficient statistical power. In the analysis in Fig. 2 the numerator is the total number of live births and the denominator is the total numbers of embryos transferred. We are essentially assuming that each embryo in a double embryo transfer implants independently of the other. This assumption is reasonable since most major endometrial factors are detected with modern ultrasound monitoring and uterine cavity imaging. We routinely performed saline infusion sonograms on all patients prior to embryo transfer. At least one study supports this conventional thinking that embryos implant independently of other embryos transferred concurrently [19].