Clinical value of histologic dating of the endometrium for personalized frozen-thawed embryo transfer (pFET) in patients with repeated implantation failure (RIF) in natural cycles

Background: Displacement of the window of implantation (WOI) has been proposed as an important factor contributing to RIF. However, histologic dating of the endometrium as a diagnostic tool of endometrial receptivity has been questioned. Methods: This is a prospective intervention trial that entailed 205 infertile patients from July 2017 to December 2017. Endometrial biopsies from 50 good-prognosis patients were conducted on day 3 (n=6), 5 (n=6), 7 (n=26), 9 (n=6) or 11 (n=6) post-ovulation (PO+3/5/7/9/11) of the previous natural cycle before their conventional frozen-thawed embryo transfer (FET) cycle. We conducted endometrial biopsies of 155 RIF patients on day PO +7. Results: The verification of the Noyes criteria for endometrial dating was conducted at different times (PO +3/+5/+7/+9/+11) on 41 good-prognosis patients who achieved an ongoing pregnancy in their first conventional FET cycle after endometrial biopsy. The agreement between two pathologists for endometrial biopsy dating in infertile patients was determined to be acceptable (weighted kappa = 0.672, P < 0.001). The rate of out-of-phase dating on day PO+7 was significantly higher in RIF patients than good- prognosis patients (31.6% vs . 3.8%, P=0.003). pFET was performed in 47 RIF patients diagnosed to be out of phase, and the cumulative live-birth rate was 55.7%. Conclusions: Histologic endometrial dating of RIF patients in natural cycles may be a biomarker for a receptive endometrium in diagnosing the displacement of WOI.

patients were conducted on day 3 (n=6), 5 (n=6), 7 (n=26), 9 (n=6) or 11 (n=6) post-ovulation (PO+3/5/7/9/11) of the previous natural cycle before their conventional frozen-thawed embryo transfer (FET) cycle. We conducted endometrial biopsies of 155 RIF patients on day PO +7. Results: The verification of the Noyes criteria for endometrial dating was conducted at different times (PO +3/+5/+7/+9/+11) on 41 good-prognosis patients who achieved an ongoing pregnancy in their first conventional FET cycle after endometrial biopsy. The agreement between two pathologists for endometrial biopsy dating in infertile patients was determined to be acceptable (weighted kappa = 0.672, P < 0.001). The rate of out-of-phase dating on day PO+7 was significantly higher in RIF patients than good-prognosis patients (31.6% vs . 3.8%, P=0.003). pFET was performed in 47 RIF patients diagnosed to be out of phase, and the cumulative live-birth rate was 55.7%. Conclusions: Histologic endometrial dating of RIF patients in natural cycles may be a biomarker for a receptive endometrium in diagnosing the displacement of WOI.

Background
Repeated implantation failure (RIF) is a particular challenge that is defined upon 4 ultrasonographic examination as the absence of a gestational sac 5 or more weeks after an embryo transfer (ET) subsequent to three previous embryo transfers with high-quality embryos, or after the transfer of ≥10 embryos in multiple transfers [1]. RIF can be caused by both maternal and embryonic factors [2], and blastocyst culture and preimplantation genetic screening can partially improve the outcome of pregnancy through better embryo selection The uterus, an important player in implantation that may be affected by polyps, intrauterine adhesions, uterine fibroids, adenomyosis, endometritis, and uterine malformations, has also been demonstrated to contribute to embryonic implantation failure [4]. Different strategies have been developed to improve the outcome of pregnancy for the aforementioned diseases, but unexplained RIF remains a challenge.
Endometrial receptivity has been frequently evaluated as one of the many uterine factors involved in RIF, but the relationship remains controversial. Several endometrial markers-such as the presence of pinopods, immunohistochemical biomarkers, endometrial blood flow and wave-have been used to determine uterine receptivity. These biomarkers may interact transiently with the embryo at implantation, but they appear to be unreliable in evaluating receptivity, particularly as precision indicators for use as clinical diagnostic tools [5][6][7].
The endometrium becomes receptive as a result of a series of timed hormonal events during the menstrual cycle. The exposure of the endometrium to progesterone after ovulation initiates morphologic and functional alterations that result in a change from a pre-receptive to a receptive endometrium. With the secretion of progesterone, subnuclear vacuoles are found in epithelial cells during the early secretory phase, the secretory products (in vacuoles) within the glandular epithelium cells discharge to the glandular lumen, and stromal edema becomes 5 maximal in the middle secretory phase-all of which contributes to blastocyst adhesion and invasion. Edema is also less marked at this time, and a predecidual reaction begins around the blood vessels, which contributes to embryonic implantation. The morphologic changes observed histologically for each specific day after ovulation were described by Noyes and his colleagues in 1950 (designated as Noyes criteria) [8]. An endometrial biopsy that shows a difference of more than 2 days between the histologic dating and actual day after ovulation is considered to be "out of phase" [9]. However, the clinical application of the Noyes criteria is relatively limited, as an out-of-phase endometrium was found in 5%-50% of fertile patients [10][11][12]. The large variation in researchers' results may be due to the inaccurate determination of the day of ovulation.
Previous investigators demonstrated that the use of classic histologic dating of endometrial biopsy samples to estimate the timing of the window of implantation and to adjust embryonic transfer time may increase the implantation rate in hormone replacement therapy (HRT) cycles of patients with out-of-phase classical dating of the endometrium [13]. The clinical value of histologic endometrial dating in RIF patients in natural cycles is, however, yet to be realized. In the present study, we investigated the clinical effects of pFET in unexplained RIF patients according to the histologic dating of endometrial biopsies under ultrasound-guided ovulation monitoring during natural cycles.

Study population
We evaluated a total of 205 infertile patients and created two phases for our study.
In phase I, a total of 50 good-prognosis patients underwent endometrial biopsy at 6 different time-points (PO+3/5/7/9/11) (Fig. 1). The histologic profiles of goodprognosis patients who were pregnant in their first conventional FET cycle were then collected as fertility parameters. For the good-prognosis patient group we enrolled women aged 20-35 who underwent FET in a natural cycle. In phase II, 155 patients with unexplained RIF were recruited to evaluate their endometrial dating on PO +7 (Fig. 1). According to an ESHRE PGD consortium, RIF was defined as the absence of a gestational sac upon ultrasonographic examination at 5 or more weeks subsequent to three embryo transfers with high-quality embryos, or after the transfer of ≥10 embryos in multiple transfers [1]. Patients with uterine abnormalities (double uterus, bicornuate uterus, unicornuate uterus, and uterine mediastinum), intrauterine adhesions, endometriosis, adenomyosis, hydrosalpinx, or uterine fibroids (submucosal fibroids, non-mucosal fibroids > 4 cm and/or endometrial pressure) were excluded from the unexplained-RIF group. In both groups, patients demonstrated a menstrual cycle length of 24-35 days and an indication for ovarian stimulation before in vitro fertilization/ intracytoplasmic sperm injection (IVF/ICSI). This study was approved by the Ethics Committee of the Reproductive and Genetic Hospital of CITIC-XIANGYA (LL-SC-2017-007) (June 29, 2017). Although we began to recruit the patients on the initial release date of our clinical trial, we discovered an advantage of histologic dating of endometrial biopsy samples when we designed and organized the data of the first clinical trial (NCT03222830). We then increased the sample size of our study, which was of major importance in recruiting RIF patients and good-prognosis patients for histologic dating for the second clinical trial (NCT03312309).

Ovulation monitoring
All patients were monitored throughout a natural cycle with a daily ultrasonographic scan from the 10 th -12 th days of the menstrual cycle when the largest follicular diameter was 16 mm, and until the dominant follicle disappeared. Urinary LH concentrations were assessed simultaneously when the follicular diameter was 16 mm. The day of dominant follicle disappearance was considered to be the day of ovulation (post-ovulation +0, PO+0).

Endometrial biopsy
Endometrial biopsy was performed using a sterile pipelle (Laboratory CCD, China), and the tissue was stored in Hank's Balanced Salt Solution (Life Technologies, Grand Island, NY) on ice for further processing.

Histologic analysis and dating
Endometrial tissue was rinsed in chilled PBS followed by 10% neutral-buffered formalin fixation and paraffin embedding (FFPE). FFPE tissues were sectioned at a 6mm thickness for hematoxylin and eosin (H&E) staining. All H&E-stained endometrial biopsies were analyzed in a blinded manner for the evaluation of endometrial dating and glandular and stromal development. The verification of endometrial dating was established according to the Noyes dating criteria [8].

Personal frozen embryo transfer/conventional frozen embryo transfer protocol
For the FET cycle, no more than two embryos were transferred to each patient.
Embryos were warmed using a commercially available warming solution (Kitazato Biopharma), according to the instructions of the Kuwayama kit [14]. After warming, embryos were transferred to G1.5/G2.5 medium and cultured for 2-6 hours. Only cleavage-stage embryos that exhibited >50% intact blastomeres or blastocysts that re-expanded after warming were considered as surviving and suitable for transfer.
The cleavage-stage embryos or blastocysts were transferred 3 or 5 days, 8 respectively, after ovulation regardless of endometrial dating for good-prognosis patients, or 4-7 days after ovulation depending upon the endometrial dating of the RIF group. We applied luteal support when the dominant follicle disappeared and when we observed satisfactory endometrial development (thickness ≥8 mm as confirmed upon ultrasonographic examination): we administered 40 mg of dydrogesterone (Abbott Biologicals B.V.) until the 28 th day of embryo transfer if a pregnancy test was positive.

Clinical outcomes and statistical analysis
We defined the cumulative live-birth rate of repeated FET cycles during the study period as the probability of a live birth from all cycles during the study period.
Ongoing pregnancy was defined as at least one intrauterine gestational sac with cardiac activity by ultrasonography performed 6 weeks after ET. Biochemical Continuous variables are presented as means ± SD, and comparisons were made using a one-way ANOVA or non-parametric statistical test. Categorical data were presented as a number (N) and percentage (%), and comparisons were made using a Chi-square or Fisher Exact-Probability test. A weighted kappa statistic was calculated to summarize the overall agreement between pathologist A and pathologist B. Bland-Altman plots were drawn to evaluate systematic biases of endometrial dating between pathologists A and B. We used GraphPad Prism7.0 to evaluate the intra-group difference (means ± SD) in the control group. P<0.05 was considered to be statistically significant. 9

Verification of Noyes criteria
Standard parameters of endometrial histologic dating were established from the good-prognosis patients (n=41). With respect to PO/dating +3, gland nuclei were pushed to the center of the epithelial cells with cytoplasm above, with vacuoles below. For PO/dating +4, glandular nuclei returned to the basilar side of the cells.
We noted that wisps of secretory material appeared in the lumina, and some vacuoles were pushed past the nucleus, apparently emptying their glycogen into the lumen. Mitosis and pseudostratification of nuclei were absent. For PO/dating +5, only a few vacuoles remained. For PO/dating +7, tissue edema-though variable in the proliferative phase-was characteristically notable in the mid-secretory stage, becoming evident rather suddenly. For PO/dating +9, the spiral arterioles (which were previously somewhat difficult to distinguish in the edematous stroma), became much more prominent. For PO/dating +11, pre-decidua began to differentiate under the surface epithelium (Fig. S1). These results were consistent with the Noyes dating criteria [8].

The agreement of blinded endometrial dating and the criteria for endometrial dating
All the endometrial datings (n = 205) were evaluated by two experienced pathologists. The inter-observer agreement was determined to be good (weighted kappa = 0.672; 95% CI 0.606-0.737; P < 0.001). As shown in Fig. 2, Bland-Altman (B-A) plots of pathologist A and pathologist B highlighted trends regarding differences in endometrial dating between the two pathologists. The limits of agreement indicated that the difference value for dating an endometrial biopsy was ≤1.76, but that the endometrial dating by both pathologists was clinically consistent when the value was ≤ 2. Thus, the B-A plots suggested that agreement was good between pathologists A and B when the value was ≤ 2 by using the Noyes criteria in the same patients.

Out-of-phase endometrial dating
Endometrial dating standards for different days (PO+3/5/7/9/11) were established in good-prognosis patients who achieved an ongoing pregnancy in their first conventional FET cycle (n=41) (Fig.1). Two experienced pathologists confirmed the endometrial dating by Noyes criteria in the good-prognosis patients group, showing that the endometrial dating at different times was significantly different. In contrast, the inner-group differences were so small that the endometrial dating of most good-prognosis patients showed a mean± SD between the lower and upper limits. The exception was only 1 endometrium of a good-prognosis patient that was biopsied on PO+7 but dating as PO+3, and who was pregnant in a conventional FET cycle (Fig. 3A).
Blinded histologic dating of endometrial biopsies from RIF (n = 155) or goodprognosis patients (n = 26) before frozen-thawed embryo transfer day PO+7. The rate of out-of-phase dating on day PO +7 was significantly higher (31.6% vs 3.8%, P = 0.003) in the RIF group relative to the good-prognosis group (Fig. 3B).

Clinical outcomes in RIF patients with endometrial dating results per pFET
The demographic characteristics and reproductive history of RIF patients-including age, body mass index, duration of infertility, and cause of infertility-are shown in Table 1. Previous failed cycles numbered 3.6 ± 0.7, with minimal and maximal values of 3 and 5 cycles, respectively. pFET was performed in 47 patients whose personal WOI was delayed by 3 (n = 35) or 4 (n = 11) days, or advanced by 3 days

Discussion
This is the first pFET study to entail endometrial dating as verified using the Noyes criteria. Implantation is not a single event, but, is more accurately a cascade of interactions between the embryo and endometrium. The human endometrium is receptive to embryonic implantation during a narrow period of the menstrual cycle referred to as the WOI. The WOI had been assumed to be constant for all women, although investigators have recently demonstrated the existence of "displaced WOI" [15][16][17][18]. The classic method of dating the endometrium using defined histologic criteria was established in 1950 [8], but pFET studies in natural cycles using the Noyes criteria are presently lacking. In the present study we established the endometrial dating criteria only for pregnant patients in their first conventional FET cycle. All our good-prognosis patients were pregnant with histologic dates of days +5-7 on day PO+7, except for one patient whose histologic dating was PO+3, which is consistent with the definition of being in phase [9]. Endometrial dating was simultaneously evaluated by two experienced pathologists, and the inter-observer agreement was statistically determined to be acceptable. Thus, the endometrial dating criteria were easily mastered and utilized according to the Noyes criteria.
Finally, our data indicated that intra-group variation in the good-prognosis patients was so low that we considered our results to be highly reliable. We determined the reproducibility and verifiability of our endometrial dating in the same patients with five RIF patients who failed to become pregnant after the first pFET; these patients underwent a second endometrial biopsy delayed by 1 or 2 days according to the first results of endometrial dating, and showed the expected endometrial dating of the second endometrial biopsy. The second biopsies of the five RIF patients were conducted within 4 months, which suggested that the results of endometrial dating might be repeatable in 4 months. Endometrial dating of a larger sample and longer period of time would still be needed to corroborate reproducibility. Displaced WOI may be detected by endometrial dating, and a subset of the unexplained-RIF patients may benefit from our study data.
Recent data from transcriptomic microarrays (described as endometrial receptivity arrays, ERAs) have shown that the receptive phase of the endometrium may be delayed in about 25% of RIF patients [19][20][21][22][23][24][25]. In these cases, delaying FET according to the endometrial delay (pFET) may lead to an improved clinical pregnancy rate (42%-51.7%), especially in women with multiple-failure cycles and apparently good-quality embryos. However, the widespread use of ERA is hampered by high costs and complex transcriptomic analysis techniques. The recent literature has shown that endometrial transcriptomics point to two potential causes of RIF: molecular displacements and molecular disruptions [26], inducing asynchrony (displacement) and pathology (disruption), respectively, in RIF [26]. It is possible that the remaining 70% of RIF patients whose endometrial dating was in phase in our study would benefit from elucidating the molecular-disruption aspect of RIF using endometrial transcriptomics.
There were some limitations to our research. The primary objective of our study was to use endometrial histology to detect the displacement of WOI in RIF patients, and to evaluate the clinical outcomes by pFET. However, there were no comparative data on clinical pregnancy rates or live-birth rates in women with RIF who had an adjusted FET, or on those who did not. We suggest that a randomized controlled study for RIF patients with out-of-phase dating be undertaken, and that further research be performed to determine the mechanism(s) underlying irregularities in the window of implantation.

Conclusions
We observed an obviously increased percentage in WOI displacement in RIF patients compared with good-prognosis patients, leading to our proposition of pFET as a treatment strategy.

Consent for publication
Not applicable.

Availability of data and material
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.  20 Supplemental File Legends Figure S1. Endometrial specimen dating according to Noyes criteria (X 400). A (dating +3), gland nuclei were pushed to the center of the epithelial cells, with the cytoplasm above and vacuoles below (arrow). B (dating +4), gland nuclei returned to the basilar side of the cells, and some vacuoles (arrow) were pushed past the nucleus to apparently empty glycogen into the lumen. C (dating +5), few vacuoles remained, and the glandular cavity was filled with secretions (arrow). D (dating +7), tissue edema. E (dating +9), glands were highly distorted, jagged, or cauliflowershaped (arrow). F (dating +11), pre-decidua (arrow) began to differentiate under the surface epithelium.

Figures
21 Figure 1 The patients recruited to the control and RIF groups. 22