Recent research from IVIRMA has established that the likelihood of a euploid embryo transfer resulting in an ongoing pregnancy is approximately 65% (1). In the search for what causes the remaining one third of embryo transfers to fail, efforts have been focused on the main parameters that seem to contribute to a successful implantation and subsequent pregnancy: obtaining as many high quality euploid blastocysts as possible, and ensuring that the endometrium is receptive to sustained implantation. On the question of whether it is the embryo, the endometrium, or the synchrony between the two that causes an embryo transfer to fail, the answer is likely “all of the above”.
With the issue of embryo-endometrial synchrony being this important, it is no wonder that when the Endometrial Receptivity Array (ERA) test was released 10 years ago it was quickly adopted by reproductive medicine providers. The premise of this test is that, in some women, the window of implantation is displaced, and the endometrium is thus not receptive at the time when the embryo is transferred. Instead, the endometrium is 12-48 hours too early or too late in its transcriptomic profile, rendering the woman infertile and dooming all future transfers performed at the “standard” time to an ominous outcome. Therefore, a test that could first identify whether the endometrium was receptive at the “standard” time or not and then provide guidance on how much to adjust the transfer window for each particular patient would be extremely useful. This would be especially beneficial in patients with recurrent implantation failure (RIF), in which it would be even more likely for this window of implantation displacement to be the cause of the patient’s infertility. According to the manufacturer of the ERA test, 3 in 10 women suffer from a displaced window of implantation, which would make this the leading cause of infertility.
To this date, more than 150,000 ERA tests have been commercially performed. (2) Unfortunately, while the idea is biologically plausible, the correct path for validation was not followed before marketing the final product. Several studies confirmed the technical correctness of the test by establishing that the result of an endometrial sample would always be the same (intra-biopsy reproducibility) and that this result would remain the same during the next cycle in which the embryo transfer would take place (inter-cycle reproducibility). However, this only shows that the technology works as expected, not that its application results in any benefit to the patient.
Ideally, to ensure that the test detects (albeit very accurately) something that actually would be worth addressing, a non-selection study should be carried out. In such a study, endometrial biopsies would be obtained and tested, but the transfer would then be performed at the standard time ignoring the test result. Only if those with an abnormal receptivity result had significantly worse outcomes, we would conclude that the test reliably predicts poor receptivity and the problem warrants solving. At that point, an RCT would be the ideal design to finally determine if the proposed solution works. In summary: 1) intra- and inter-sample reproducibility for validation that the test is accurate, 2) non-selection study to evaluate whether what is detected is actually a problem, 3) RCT to verify that the recommended treatment is effective.
A recent publication by Doyle et al. in JAMA shares the results of an RCT on the use of ERA to guide embryo transfer timing and, perhaps more importantly, includes data to extract the results of a non-selection study as well. (2) The authors performed a double-blind RCT in which 767 patients had a biopsy and ERA testing, followed by a single euploid frozen embryo transfer. 381 patients had a personalized embryo transfer (pET, timed according to the ERA test recommendations) and 386 underwent a standard timing frozen embryo transfer (FET) as the control group. The ERA test result was either receptive or non-receptive, and non-receptive results included a recommendation on how many hours to adjust the embryo transfer timing by. The live birth rate per embryo transfer (LBR) was 58.5% in the pET group vs 61.9% in the standard FET group (not statistically significant). No differences were identified in any of the secondary outcomes either.
Interestingly, of all patients – only 7.6% of whom had a previous failed transfer – 55.5% had a non-receptive ERA result. Several post-hoc analyses provided in the article allow for interpretation of the usefulness of the ERA test in several ways that were previously unavailable:
- Of those patients in the control group, all of which received an embryo transfer at the standard time regardless of ERA result, there was no difference between those with a receptive result and those with a non-receptive one (LBR: 61.2% in receptive vs 62.5% in non-receptive, p=0.8; Chi-square test performed from supplementary data provided).
- Interpretation: a non-selection study within the RCT. If, without treatment, patients with the “disease” have similar outcomes to those without it, the test may be accurate, but what it detects is not detrimental: it is not detecting a pathology.
- When comparing only patients with a non-receptive result (who should benefit from a pET), there was no difference between outcomes after pET vs standard FET (LBR: 54.5% vs 62.5%, respectively, p=0.1).
- Interpretation: those with the “disease” who were treated had similar results as those untreated: the recommended treatment does not improve outcomes.
- When evaluating only those recommended to adjust the timing by more than 24 hours (who should be the least receptive, worst prognosis patients if timing is not adjusted), there was a significantly lower clinical pregnancy rate (-16.5%; 61.8% vs 78.3%, p=0.01) and a significantly higher biochemical pregnancy loss rate (+11.2%; 15.2% vs 4.0%, p=0.02) observed in the pET group compared to standard FET, while LBR were comparable.
- Interpretation: those who should have the worst outcomes (severe “disease”, untreated) had better outcomes than those who should benefit the most (severe “disease”, treated): applying an extreme treatment may cause harm.
All in all, this very well conducted study sheds light on the issue of whether ERA is helpful, merely informative, or harmful. While the test seems reliable in detecting its proprietary transcriptomic signature (similar rates of non-receptive results in both groups), this result does not appear to impact the outcome of embryo transfer. Once this has been elucidated, it is therefore expected that acting upon this result will generally have no impact on the outcome and that, in fact, shifting the transfer timing by 24 hours or more may even be harmful.
On the other hand, proponents of the ERA have argued that this study excluded patients with RIF and these patients are the ones to benefit the most from this test. It is interesting to note that in a population of patients without RIF as a diagnosis, several of them could be assumed to indeed suffer from yet undetected RIF if this were actually a prevalent condition. This can be evaluated in two ways:
- If a population of patients with unknown RIF status undergoes several successive embryo transfers, the rate of failure should increase after each attempt due to concentration of those patients with RIF over an ever decreasing denominator (as those without RIF succeed and are removed from the next transfer attempt). However, this has been clearly showed to not be the case after at least 3 transfers, demonstrating that – if RIF is indeed a clinical entity – it affects less than 5% of the population (given a 95.2% sustained implantation rate after up to 3 consecutive transfers). (1)
- If RIF were indeed a relatively common condition among infertile women, and this was due to a displaced window of implantation detectable and treatable via the ERA, untreated women should have worse outcomes than those treated and those without the condition. In an infertile population such as Doyle’s, one can assume that several patients would eventually have RIF even if undiagnosed as of the time of participation in the study and, therefore, those with a non-receptive endometrium and standard FET timing should have the worse outcomes, which they do not. (2)
This leads to two conclusions: A) RIF may exist as a clinical entity, but if it does, it has a very low prevalence and B) the displacement of the window of implantation, even if it were real, is of no reproductive significance.
While the endometrium and its synchrony with the embryo are key players in the implantation game, the problem to be solved does not lie in adjusting the time of transfer, but likely rather in identifying a pathologic endometrium or embryo-endometrial interaction that may require treatment.
- Pirtea P, De Ziegler D, Tao X, Sun L, Zhan Y, Ayoubi JM, Seli E, Franasiak JM, Scott RT Jr. Rate of true recurrent implantation failure is low: results of three successive frozen euploid single embryo transfers. Fertil Steril. 2021 Jan;115(1):45-53. Epub 2020 Oct 16. PMID: 33077239.
- Doyle N, Jahandideh S, Hill MJ, Widra EA, Levy M, Devine K. Effect of Timing by Endometrial Receptivity Testing vs Standard Timing of Frozen Embryo Transfer on Live Birth in Patients Undergoing In Vitro Fertilization: A Randomized Clinical Trial. 2022 Dec 6;328(21):2117-2125. PMID: 36472596.
Andres Reig, MD