Background
Multiple birth is recognised to be an important risk to the health and welfare of children born after in vitro fertilisation (IVF), and can be effectively reduced by transferring only one embryo to those women who are most at risk of having twins [
1]. The current National Institute for Health and Care Excellence (NICE) guideline covering diagnosing and treating fertility problems in the UK recommends state-funding and single embryo transfer (fresh or cryopreserved) in the first full IVF cycle for women under 37 years, and if there are one or more top-quality embryos for women aged 37 to 39 years [
2]. UK state-funding is not available currently for preimplantation genetic screening (PGS).
Current genetic testing techniques for chromosome aneuploidy can test for every chromosome [
3]. Selecting embryos with the highest potential for implantation offers the potential to transfer one embryo at a time in the fewest possible number of transfer procedures to optimise a woman’s chance of achieving a healthy singleton live birth event and reduce the risk of miscarriage due to chromosome aneuploidy. Appropriately powered, well-designed, peer-reviewed randomised control trials, with a live birth outcome measure which goes on to report on child health, are recommended to be the gold standard for evidenced-based IVF medicine [
4]. Although, others have argued for a more pragmatic approach to circumnavigate protracted delay in introducing the highest quality treatment for patients [
5].
Outcome measures which incorporate fresh as well as cryopreserved embryo transfer (cumulative rates) rather than success rates based on only fresh transfer is recognised to be more appropriate for decision making regarding the efficacy of treatment and cost [
6]. However, a prospective intention-to-treat embryo selection study is likely to be costly and take several years to complete, and difficult to justify if it is expected that the cumulative live birth rate (CLBR) with testing will be inferior due to an imperfect genetic test which incorrectly excludes viable embryos [
7]. An inferior CLBR can be avoided if testing is used only to determine the order in which embryos will be transferred [
8].
An aneuploidy screening trial using virtual women and embryos is an idealistic way of investigating the usefulness of different approaches and the cost-effectiveness of testing embryos, free from the principle of equipoise and the constraints of cost and time. The aim of the study presented here is to explore the cost-effectiveness of aneuploidy screening for every woman aged under the age of 40 years when fresh and vitrified-warmed embryos are transferred one at a time in a first full cycle of IVF, comparing selecting out with ranking-only, and with different trial end points.
Discussion
This virtual trial was conducted to provide insight into the cost-effectiveness of incorporating PGS into the first IVF treatment attempt for every woman under the age of 40 years, transferring embryos one at a time. It was envisaged that a 24-chromosome genetic test for aneuploidy could be used to exclude embryos with an abnormal test result from transfer (PGS1), or be used only to rank embryos with the highest potential to be viable (PGS2), with the effect on outcome of prematurely discontinuing treatment also assessed. These approaches were compared with treatment without testing. A hypothetical ideal test was also considered, which could ensure that only a viable embryo, if available, was transferred first (PGS3), with no risk of miscarriage.
Testing, by design, was effective to reduce the occurrence of clinical miscarriage. This was based on a prospective non-selection study [
10] where the clinical miscarriage rate per clinical pregnancy was calculated to be 8.5% (5/59) without genetic testing and 5.1% (3/59) following aneuploidy screening [RR 0.6 (0.15–2.397)
P = 0.4639]. The numbers are small and the confidence interval is wide and caution regarding the effect of testing is advised. In the virtual trial the miscarriage rates without testing were 9.6% (24/251) for younger women (<38 years) and 15% (3/20) for older women (38–39 years) (see also Additional file
1: Table S5). A UK cross-section audit including fresh and freeze-thawed cycles reported to the HFEA for clinics indicated similar rates for younger and older women of 9.0% (1128/12,524) and 15.3% (391/2558) (see also Additional file
1: Table S19). Excluding embryos (PGS1) was shown to be more effective and less expensive than ranking (PGS2); however, relatively few individual women were likely to benefit, whether on an intention-to-treat basis or when more than one embryo was available for testing. Testing did not result in a higher chance of a live birth event, and PGS1 was less effective than PGS2 due to the exclusion from transfer of viable embryos with incorrect abnormal test results (false positives).
Prematurely terminating the trial resulted in a disproportionate exclusion of women without testing who still had embryos available, which substantially reduced the deficit or resulted in an excess of live birth events following testing, with only a marginal effect on the number of clinical miscarriages avoided. Conclusions regarding the effectiveness of PGS for live birth based on clinical trials which include only the first transfer attempt, or that do not take into account women with surplus cryopreserved embryos, are therefore likely to be biased in favour of testing.
The author is aware of one published randomized controlled trial, for older women aged 38 to 41 years, which has attempted to estimate cumulative outcome measures [
18]. After excluding poor prognosis patients, the primary outcome measure was the delivery (live birth event) rate for the first transfer attempt, which should be expected to favour testing [
7]. The study reported a significantly higher live birth rate in the tested group: 52.9% (36/68) vs 24.2% (23/95) [OR 3.522 (1.804–6.873),
P = 0.0002]. Adding live births from cryopreserved embryo transfers during the 6 months following the study recruitment period, the cumulative delivery rate in the tested group was reported to be 37.0% (37/100) vs 33.3% (35/105) [OR 1.175 (0.662–2.085),
P = 0.5285]. It is not clear how many women after this period who had not achieved a live birth event still had cryopreserved embryos available. The cumulative miscarriage rate was 1.0% (1/100) with testing vs 20.0% (21/105) without testing; per 100 women, the cost of avoiding one miscarriage can be estimated to be €13,648 [(€1,075,273–€815,965)/(20–1)], with 19 (19%) women potentially benefiting from testing. Including all the embryos available from a stimulated cycle, it would seem that a woman in this age group is unlikely to increase her chance of having a baby, and around 1 in 5 of good prognosis patients is likely to avoid miscarriage by adding PGS to their treatment. Including fresh and cryopreserved embryo transfers, the authors reported a clinical pregnancy miscarriage rate of 36.2% (21/58) without testing vs 2.6% (1/38) with testing. A UK cross-section audit for older women (38 to 42 years) including fresh and freeze-thawed cycles reported to the HFEA for clinics with PGS activity indicated corresponding rates of 19.1% (859/4508) and 11.8% (14/119) (see also Additional file
1: Table S19). Consequently, the beneficial effect of testing on miscarriage indicated by the trial may be optimistic.
The treatment time, with or without a live birth event, was shortened for PGS1. The treatment time for women following PGS2 who had a baby was shortened, but it was not shortened for women who did not have a baby. Therefore, PGS2 would not enable women who needed more than one stimulated cycle to start the next cycle more quickly. This approach may also be unattractive to women because they would need to accept the possibility of transferring embryos with an abnormal test result to complete their cycle, which may also have some ethical hazard. However, it has been argued that the predictive value of genetic testing, even at the blastocyst stage, is too low for clinical use [
19].
Using the hypothetical ideal genetic test (PGS3), testing was more effective for a live birth event than not testing, and the risk of a clinical miscarriage was eliminated. A woman’s cumulative live birth rate with PGS3 was superior to not testing because only fresh embryos were transferred, which avoided the attrition associated with a subsequent vitrified-warmed cycle [
7]. A limitation of these hypothetical studies is the assumption that the live birth potential is the same for a fresh and warmed embryo transfer. A recent report of a randomized controlled trial [
20], which compared fresh and vitrified-warmed embryo transfer following PGS, did not find a statistically significant difference in the number of live births per fresh (52%, 13/25) and warmed-vitrified (64%, 16/25) single embryo transfer, although the latter was greater with small numbers.
This virtual trial was based on a prospective non-selection study designed to assess diagnostic accuracy [
10], where the value of an aneuploid test result to predict non-viability was high (96%). This may not be realistic if trophectoderm mosaicism is more common than recently appreciated [
21]. However, unrealistically high diagnostic accuracy may be expected to favour testing and therefore testing may be more less-beneficial than indicated by this virtual trial. Using a different approach, another hypothetical trial also demonstrated greater superiority of not-testing over testing in terms of the cumulative live birth rate [
22].
The implantation rate (44%) [
10] of an embryo with a euploid test result used in this virtual trial might be considered to be modest. A higher implantation rate would not be expected to make genetic testing materially more effective than not testing for live birth over a full cycle (testing does not create normal embryos); however, it might be expected to reduce the number of warmed embryo transfer attempts and the cost of testing [
7], and testing may therefore be less expensive than indicated by this virtual trial.
Every transferable embryo in this hypothetical study is assumed to have the same potential for implantation, differentiated only for those that are diploid or aneuploid. Augmenting selection using morphology criteria is outside the scope of this study; however, a mitigating effect on the number of warmed embryo transfers and cost might be expected.