Single and double endometrial scratching (ES) in infertile women with strict criteria of recurrent implantation failure (RIF).

Controversies surrounding the effect of ES on pregnancy outcome in women with RIF are mostly due to the poorly defined target population. We evaluated the effect of ES on clinical outcomes in women with strict criteria of RIF before IVF/ICSI. We also examined the effect of ES on the expression of markers of endometrial receptivity. Women with failed implantation after transfer of seven or more top quality day 3 embryos or three blastocysts underwent the scratch procedure on exact days of the cycle prior to IVF/ICSI. Results were compared to no scratch control group. Using histopathology, immunohistochemistry, and scanning electron microscopy, we also examined the effect of injury on the endometrial receptivity in a separate series of observations with double ES. Cumulative pregnancy rate was significantly higher in the study group as compared to control (54.8% vs. 29.0%; p < .05). The effect of ES on the clinical outcome was seen during fresh ET, but not on the next FET cycles. ES improves impaired endometrial receptivity by partially normalizing the expression of estrogen and progesterone receptors (ERs, PRs) and pinopodes. We concluded that in a well-defined subpopulation of infertile women with RIF, ES significantly enhances pregnancy rates. ES has a specific impact on endometrial receptivity normalizing the expression of some markers.


Introduction
RIF remains a major factor of unsuccessful treatment of infertility [1]. The use of local traumatization of the endometrial tissue by pipelle biopsy as a means of improving endometrial receptivity was first shown by Barash et al. [2]. Since then, numerous studies have been performed, with the majority but not all showing a positive effect of endometrial scratching on pregnancy outcomes in women undergoing in vitro fertilization (IVF) [3][4][5][6]. The data are still controversial, mostly because of the poorly defined target population, as well as different protocols and timing of the injury. Also, the mechanism of action of ES is unknown [3].
Hence, the aim and objective of our researchare: 1. We have tried to evaluate the effect of endometrial scratching (with strict protocol of the intervention) performed in a carefully defined subpopulation of infertile women on pregnancy outcome after IVF/ICSI. 2. We also aimed to determine how PB affects the markers of endometrial receptivity and what the clinical implementations of these changes are. For this, we conducted a separate pilot study with double endometrial biopsy in women with history of RIF.

Settings
Study was performed in a private IVF clinic affiliated with Yerevan State Medical University.

Participants and study design
Participants were divided into four groups. Group A, 31 women with RIF were allocated to the scratch procedure on day 20 of luteal phase of the cycle prior to IVF/ICSI. Group B, 31 patients with RIF underwent IVF/ICSI with no scratching. Group C, 11 patients with RIF who underwent double scratching in two consecutive cycles prior to IVF/ICSI cycle. Group D, eight healthy donors who underwent ES as control subjects.
The primary outcome of this study was the ongoing pregnancy rate defined as a pregnancy >7 week with positive heartbeat. The secondary parameters of this study were dynamic changes of the markers of endometrial receptivity as a result of scratching.
We used more stringent criteria for RIF than it is defined by ESHRE guideline [7]. Women with failed implantation after transfers of three or more blastocysts or seven or more day 3 top quality embryos were included in the study. Failed implantation was defined as the absence of clinical pregnancy in the full cycle. In order to evaluate the effect of endometrial scratching on the clinical outcome, only patients who underwent two consecutive embryo transfers (fresh and frozen/thaw) were included in the study. All women were aged between 21 and 38 years, with regular menstrual cycle (defined as a cycle length of 21-35 days). Exclusion criteria for all groups were endometrial pathology (polyps, submucosal myomas, sinechia, and hyperplasia); untreated endocrine disorders; and pelvic inflammatory diseases.
In group A, endometrial scratching was carried out once on day 20-22 of the cycle prior to IVF. Days of ES were determined as LH þ 7-9, where the day of LH surge is day 0. In the absence of pregnancy, FET was performed on the next cycle according to the protocols of the clinic. Patients whose ET or FET was delayed for medical or other reasons were excluded. Cumulative pregnancy rate (for two consecutive transfers) was calculated.
The following markers of endometrial receptivity were chosen: estrogen receptors (ERs) and progesterone receptors (PRs) both in glandular and stromal components of endometrial tissue, expression of pinopodes. In addition, we performed a histological evaluation of endometrium, as well as an assessment of endometrial appearance by the means of vaginal sonography. In the study, group C double ES were performed on the same day of two consecutive cycles. Data obtained after the 1st biopsy were compared to the 2nd biopsy data and with control (group D), as well as with clinical outcome. The group D consisted of eight healthy fertile women participating in donor egg programs of the clinic. In this group, endometrial scratching was performed only once on day 20-22 of their natural cycle.
The study design was approved by the ethical committee of the Yerevan State Medical University.
Procedures (see descriptions of all procedures and methods in more details in Supplementary material S1). After endometrial scratching, each sample was divided into two parts and fixed immediately. One part was processed for scanning by electronic microscopy and another for an immunohistochemistry and morphological research.
Transvaginal sonography (TVS). All patients were examined by qualified sonographers with TVS. The women were scanned using TVS with 5-9-MHz transducer. Endometrial thickness and lining, as well as the day of ovulation were documented according to the protocol of our clinic.
A standard controlled ovarian stimulation (COS) with GnRH antagonists and endometrial preparation for FET were used according to the protocols of our clinic. The fresh embryo transfers were scheduled on day 3 or 5. If no pregnancy occurred then the transfers were scheduled for the next month.
Histopathology, immunohistochemistry, and scanning electron microscopy were performed according to manufacturers' recommendations and data of the literature (see Supplementary material S1). For histopathology, we have investigated the date of cycle according to Noyes et al. [8]. In order to semi-quantify the immunoreactivity of nuclear steroid receptors of ER and PR, the method of the histochemical score (H-score) was used.
Pinopodes were identified under scanning electronic microscopy with quantitation of pinopodes number, size, and percent area by a standard technique [9] and using descriptive characteristic of pinopodes expression as fully developed, developing and regressing [10].

Statistical analysis
The statistical analysis was performed using Student's t-test for comparison of mean values of continuous variables and chisquare test for comparison of proportions for categorical variables. For small sample sizes (in the group of double biopsy), we used non-parametric Mann-Whitney's test for continuous variables when comparing two different samples and Wilcoxon's rank test when comparing dynamic changes within one sample, and Fisher's exact test for categorical variables. The results were considered statistically significant at p values below .05. The statistical analysis was performed using SPSS 20.0 computer software (SPSS Inc., an IBM Company, Chicago, IL, USA).

Results
There were no statistically significant differences between groups A and B in terms of patients' age, duration of infertility, numbers of previous IVF/ICSI, and numbers of transferred embryos. In the main group, the mean numbers of ET/FET before ES were 4.39 ± 1.26 (range 2-14), the mean numbers of embryos transferred were 5.26 ± 2.31 (range [3][4][5][6][7][8][9][10][11][12][13][14], and the numbers of IVF/ICSI cycles were 1.4 ± 0.1 and 1.4 ± 0.09. The clinical outcomes in groups A and B are presented in Table 1. It can be seen from Table 1 that ongoing pregnancy rate was significantly higher in group A compared to group B (54.8% vs. 29.0; p < .035). It is noteworthy that difference in pregnancy rates between groups A and B was more pronounced during fresh ET (45.2% vs. 12.9%, p < .05). In group A, there was a dramatic decrease in the rate of pregnancy when comparing first ET to the second FET (45.2% vs. 9.7%, p < .01), which was absent in the no scratch group B (12.9% vs. 16.1%, p>.5). There were no statistically significant differences in pregnancy complications.
The results of the separate pilot study with double endometrial biopsy are presented below. Intensity of immunostaining of ER and PR both in glandular and stromal cells was calculated using H-score (Supplementary material S2). As one can see in Table 2, the levels of ER in glandular cells are significantly higher in the main group compared to the control (141.6 ± 25.7 vs. 84.8 ± 9.1; p < .0001), while PR expression in the stromal cells in the main group was significantly lower (112.2 ± 9.6 vs. 137. 5 ± 10.9, p < .0001). Results of the 2nd biopsy demonstrated that the levels of PR expression in the stroma became non-significantly different from control group (142.5 ± 20.3 vs. 137.5 ± 10.9, p¼.282), whereas those of ER in glands demonstrated tendency toward normalization, remaining however significantly higher (125.6 ± 35.9 vs. 84.8 ± 9.1, p¼.004). At the same time, we observed no significant changes in the levels of ER in stroma and PR in glands. Data of endometrial histopathology demonstrate that ES normalized the morphological structure in six out of 11 women (54.5%), while in three out of 11 (27.3%) women there were no changes, and in two out of 11 (18.2%) there were negative dynamics.
Scanning electronic microscopy pictures of different representative types of pinopodes are shown in Figure 1: developing (DP), fully developed (FDP), secretory (stage of early formation), and regressing (RP) pinopodes.
As expected, in a group of donors more often fields of FDP (25%) and DP (50%) were found, while fields of the RP were seen only in 12.5%. When comparing characteristics of pinopodes during the 1st and 2nd ES the positive dynamics in the endometrial receptivity was found in 4 (36.4%), negative in 5 (45.4%), and no changes in two (18.2%) patients with, however, no significant differences between two groups.
We sought to evaluate the relationship between the patterns of change in the expression of endometrial markers after ES and clinical outcome. For this, we classified the endometrial parameters after the first biopsy as unchanged, worsened, or improved (Table 2).
Totally, from 11 patients with RIF pregnancy was achieved in six (54.5%) women. When comparing the numbers of pregnancy outcomes between three categories of biopsy dynamics, we found that pregnancy developed in none of the patients with worsened parameters (histology, pinopodes, ER, and PR), and the percentages of pregnancies were higher (albeit not significantly) in patients with improved parameters (all four of them) ( Table 2). When we grouped the changes in each parameter into two categories, comparing improvement with no improvement (grouping unchanged and worsened subcategories together), there was a universally observed albeit insignificant trend for higher frequencies of pregnancy in the categories with improvement. However, when we compared the group of patients who had improvement in both ERs and PRs count, with the group with no improvement or improvement of only one type of hormone receptors, the frequency of pregnancy was highly significantly in the former group (85.7% vs. 0%, p¼.015).

Discussion
Understanding the impact of local endometrial injury on endometrial receptivity is important for our ability to prevent RIF. The results of our study clearly demonstrate that ES is highly effective in women with RIF if performed on day 20-22, with an  almost two-fold increase in ongoing pregnancy rate compared to the control group. However, this positive effect of the ES lasts only to the fresh ET cycle, but not second consecutive cycle with FET. Studies with ES have been criticized for high risk of bias because of the absence of strict selection criteria for RIF and large variations in the days of intervention [11][12][13]. A large multicentric randomized study has been published very recently [6].
The results of this study demonstrated that endometrial scratching before IVF/ICSI does not increase live birth rate in women undergoing IVF/ICSI. In this study, however, only 14% of the population meets the criteria of RIF of ESHRE recommendations, while the timing of embryo biopsy was chosen from day 3 of previous cycle to day 3 of the IVF cycle. Our results are different, probably because we used RIF criteria that are even more strict than have been proposed previously [7,14], as well as because we investigated endometrial receptivity at a precise day corresponding to the window of implantation.
To the best of our knowledge, this is the first study with double ES aimed to determine changes in endometrial receptivity as a result of the injury and to compare the relationship of these changes with clinical data. Comparing the data between two successive interventions, we were able to demonstrate that ES tend to normalize the expression of the markers in most but not all endometrial cells. The dynamic changes in histological parameters and expression of pinopodes tend to normalize, however, data did not reach statistical significance, due possibly to the small size of the population. At the same time, level of expression of ER in glandular tissue (but not in stroma) and PR in stromal cells (but not in glands) changed significantly after 1st biopsy. Pregnancy rate was highly significantly related to the positive changes in the ER and PR expression. It is noteworthy that when all parameters (histology, pinopodes, ER, and PR) worsened we observed no pregnancy.
Overall, our data suggest that ES mostly have a positive effect on endometrial receptivity. However, in some cases, endometrial injury may have negative or no effect, which maybe a cause of no pregnancy. Clearly, further studies on a larger scale are needed to address this specific topic.
In conclusion, our data show that ES performed in infertile women with well-defined criteria of RIF prior to the next IVF has significant therapeutic effect. The procedure of ES seems to have a normalizing effect on some markers of endometrial receptivity, hence further promoting pregnancy in women with RIF.