This study presents a novel tool for measuring endometrial receptivity. It is based on a highly quantitative TAC-seq assay that allows precise and cost-effective endometrial receptivity biomarker analysis. A custom computational model for classifying sequenced samples was developed to analyse data generated by the TAC-seq pipeline. As it is based on the expressional profile of a targeted set of genes, the beREADY classification model can be used in high-coverage whole-transcriptome studies. As a result, using endometrial biomarkers discovered from datasets with high predictive power for receptivity22,24, our prediction model allows the construction of the continuous transcriptomic states of endometrial receptivity development.
Previous transcriptomic studies of endometrium have shed light on the complex cross-talk mechanism between implanting embryos and the uterus. This insight has explained why embryo implantation in some women repeatedly fails, regardless of the use of seemingly high-quality embryos3,14,25. Therefore, the focus on helping RIF patients has shifted towards elucidating the maternal factors contributing to unsuccessful IVF cycles. This interest has propagated the development of transcriptomic analysis tools for determining the receptivity status of the endometrium10,18,20,21. Those tests use varying sets of targeted genes and different methodologies. The ERA test analyses 238 receptivity genes using microarray technology, Map®/ER Grade® test, and ERPeakSM test analyses 40 genes and WIN-Test 11 genes with quantitative PCR. The selection of genes in the beREADY model is based on the comprehensive meta-analysis of endometrial receptivity biomarkers22, complemented by eleven additional genes relevant for WOI and four housekeeper genes. To our knowledge, the beREADY model is the only endometrial receptivity testing tool that applies the Unique Molecular Identifiers (UMI) technology, enabling original transcript counts estimation while avoiding the PCR-caused bias in results. Therefore, TAC-seq technology eliminates laboratory-caused PCR duplicates and counts the biomarkers at a single-molecule level23. This approach has already been used to determine the menstrual cycle phases from endometrial tissue26.
For beREADY model development, we used natural cycle endometrial samples from healthy women and women with PCOS diagnosis. Although the involvement of endometrial factors in PCOS-associated infertility has been suggested (reviewed in Piltonen, 2016)13, there are no large-scale and systematic studies simultaneously analysing numerous endometrial receptivity-related genes in different cycle phases in PCOS patients. In addition, the endometrial tissues of PCOS patients have demonstrated altered responses to steroids27,28, but the overall effect of endometrial dysfunction on pregnancy outcomes is still unclear29. Therefore, we tested if PCOS alters the transcriptomic profiles of our selection of receptivity biomarkers. We found no significant differences between healthy controls and PCOS patients in the expression profile of endometrial biomarkers throughout the menstrual cycle from PE to the LSE phase. Based on these findings, we concluded that the selected receptivity biomarkers are unaffected in women with PCOS, and the samples were suitable for use as a reference together with samples from healthy women in the development of the beREADY model. One of the contributing factors to why we did not see significant differences in the expressional profiles of PCOS and healthy samples is that some women in the test development group had lost their PCOS characteristics since their initial diagnosis due to ageing. Nevertheless, it must be noted that PCOS is a lifelong condition despite the effect of ageing and that a retrospective diagnosis is allowed30. Secondly, patients in our study were not obese. When combined, PCOS and obesity can still introduce an additional risk factor for impaired endometrial receptivity31. For these reasons, our conclusions regarding receptivity in PCOS patients are valid for women with PCOS patients within the normal range of BMI.
We validated the beREADY model by analysing endometrial biopsies from healthy women representing the fertile female population, displaying a high similarity between sample collection time and the molecular test result (56/57, 98.2%). This similarity can be explained by the fact that the healthy women in our validation group were all LH tested and had ultrasound confirmation of ovulation and normal hormonal values. Despite this, one MSE sample (1/26, 3.8%) with corresponding tissue morphology suggested the PE phase had shifted WOI according to the beREADY. In that case, the beREADY model classified the sample as pre-receptive, suggesting that the WOI had not yet arrived. However, displaced WOI among oocyte donors and women undergoing IVF without RIF diagnosis has been previously reported11,32. In the MSE group, we also classified six samples as early-receptive (6/26, 23.1%), suggesting a slight shift in WOI that remained within the normal range of receptivity. Therefore, transcriptomic profiling can offer additional accuracy for determining the individual receptivity timing of the endometrium.
The criteria for RIF diagnosis are controversial and include several factors, such as the number of failed treatment cycles, the number of transferred good-quality embryos and maternal age33. Due to its heterogeneous nature, the causes of RIF have remained largely unknown. When focusing solely on endometrial factors, it has been proposed by Sebastian-Leon et al. that RIF is caused by at least two distinct molecular phenomena of displaced (asynchronous) or disrupted (pathological) WOI8. In our study, we applied the beREADY model to detect the rate of displaced WOI in a study group of 44 RIF patients. In total, we detected shifted WOI in 15.9% of RIF cases, which is slightly less than the results described previously by Lessey et al., 25%6, Mahajan et al., 27.5%34, Hashimoto et al., 24%9, and Patel et al. 17.7%25 in women with RIF diagnosis. Compared with the MV group, the RIF study group demonstrated a significantly higher proportion of shifted WOI cases (p < 0.05), indicating that displaced WOI can cause RIF. Moreover, Eisman et al. described a significant decrease in ongoing pregnancy rates in the RIF cohort with asynchronous endometrium compared to control infertile patients with detected WOI displacement35, suggesting that there are additional factors in implantation failure beyond an adjustment in progesterone exposure. Indeed, in RIF cases, additional endometrial-related factors likely contribute to the implantation failure along the displaced or disrupted WOI, like dysbiotic vaginal or uterine microbiome36,37.
Evidence on the clinical effectiveness of endometrial receptivity testing on the IVF pregnancy rate is still controversial. Several studies have demonstrated a significant increase in implantation rate with pET with ERA19,38, Win-test3 and rsERT10 tests. It has also been proposed that receptivity testing before the first embryo transfer increases the implantation and cumulative pregnancy rates when pET is compared with conventional embryo transfers19. Contrary to those reports, recent randomised controlled trials (RCT) and meta-analyses question the effectiveness of endometrial testing and find its impact insignificant39,40,41,42. Although other similar systematic meta-analyses and single cohort studies confirm these results of no benefit of WOI testing in good prognosis patients, they also report that in the RIF subgroup, there is a significant increase in pregnancy rates following pET38,43,44. Another study by Haouzi et al. demonstrated that the implantation rate increased over three times, from 7 to 23%, after using the personalised WOI determination in IVF patients with RIF diagnosis3. Despite generally supporting endometrial testing in RIF cases, these studies also stress low certainty of evidence, thus advocating continued research on this controversial topic38,43,44. One reason for the disparity between studies might be that the actual rate of recurrent implantation failure due to endometrial factors may be lower than previously thought. A large-scale retrospective study involving nearly 4500 patients undergoing IVF with preimplantation genetic testing for aneuploidy (PGT-A) discovered that only around 5% of the infertile patients failed to achieve clinical pregnancy following three consecutive frozen euploid single embryo transfers45. Embryonic and maternal factors such as embryonal chromosomal aberrations, impaired endometrial receptivity, maternal immune dysfunction, and infertility-associated diseases, like endometriosis and male factors, can all affect the chance of embryo implantation. Therefore, due to the challenges related to the high heterogeneity of the study designs and groups, only specific subgroups of women might benefit from the endometrial receptivity testing. In summary, personalised endometrial dating likely increases the chance of implantation in patients with recurrent IVF failure. At the same time, it is considered an unnecessary ‘add-on’ for patients planning their first IVF treatment. However, the overall efficacy of the approach is still largely unclear due to the disparity between the large-scale studies and the transcriptomic tests applied.
The potential clinical utility of endometrial receptivity testing is also fully dependent upon the consistency of the gene expression profile in subsequent cycles following the endometrial biopsy. Indeed, this claim has been approved by the seminal paper of Díaz-Gimeno et al.46. In this study, the authors demonstrated the reproducibility of the endometrial receptivity testing, as seven women underwent ERA testing twice. The second endometrial biopsy was obtained from these women on the same day of their menstrual cycle between 29 and 40 months after the first study cycle. The intercycle variation analysis showed total consistency in the ERA test diagnoses between the same patient’s first and second tests. Therefore, it is believed that the gene expression profile of the endometrial receptivity remains consistent in time, at least for the subsequent couple of months after biopsy, allowing the embryo transfer time to be adjusted according to the endometrial receptivity test recommendations. This personalised embryo transfer would allow to select the time period with the peak of endometrial receptivity for embryo transfer. However, as the initial study involved only seven patients, the question about the endometrial profile’s cycle-to-cycle consistency requires further studies.
Moreover, using the endometrial tissue biopsy for its receptivity testing would exclude the possibility of embryo transfer in the same cycle, thereby prolonging IVF treatment. Considering the possible cycle-to-cycle variation of the window of implantation timing, the endometrial receptivity testing could benefit if it could be analysed non-invasively at the cycle when embryos are planned to be transferred. Receptivity assessment from uterine fluid would circumvent the abovementioned problems, as uterine fluid aspiration can be done before embryo transfer, which is not detrimental to the implantation rate47,48. Recently, we provided proof for RNA biomarker-based minimally invasive endometrial receptivity testing using uterine fluid-derived extracellular vesicles and applying the beREADY endometrial testing model to successfully determine the receptivity status of the samples49. A similar approach would open new possibilities for non-invasive endometrial receptivity testing, which may eliminate the main shortcomings of the endometrial biopsy-based approaches in personalised embryo transfer.
Some limitations of this study should also be noted. Although the MD set consisted of samples with clearly different expressional profiles corresponding to the different menstrual cycle phases, this study suffers from a relatively small sample size with uneven distribution of samples between the receptivity classes. Furthermore, to estimate the clinical usefulness of the model, an extensive RCT must be carried out among the RIF patients, comparing the implantation, pregnancy, and live birth rates between pET and conventional embryo transfer cycles. Ideally, the RCT should be carried out with euploid embryos following PGT-A, which could rule out the genetic defects of IVF embryos as one of the possible causes of implantation failure. PGT-A has been shown to improve the cumulative IVF pregnancy outcome, avoiding the RIF diagnosis in some patients because of the transfer of the euploid embryos only45. Therefore, in our study, the RIF samples with possibly embryo-associated implantation failure were not excluded, likely explaining why in our findings the receptive endometrium was found in around 85% of RIF cases.
In conclusion, our results demonstrate that TAC-seq can be successfully applied for transcriptomic endometrial dating and establishing of the WOI. This assay detected distinct profiles of endometrial samples at pre-receptive, receptive, and post-receptive stages. Additionally, we found no significant difference in the expression levels of receptivity biomarkers in healthy fertile women and women diagnosed with PCOS. Based on these findings, we developed the beREADY endometrial receptivity testing tool with a custom classification model for distinguishing between the transcriptional profiles. In our study, we detected an increased rate of displaced WOI cases in the RIF study group compared to the validation group. Consequently, these findings suggest that applying personalised receptivity testing before embryo transfer could potentially reduce the chance of implantation failure in RIF patients.