Premature expression of the decidualization marker prolactin is associated with repeated implantation failure

Abstract

Repeated implantation failure (RIF) is a poorly understood reproductive pathology defined by the inability to achieve a clinical pregnancy in at least three consecutive IVF cycles. In this study, we investigated whether the onset of decidualization, marked by prolactin (PRL) expression, is associated with RIF. We performed a retrospective cohort study using endometrial biopsies from women with idiopathic subfertility, that conceived naturally during the same cycle in which the biopsy was taken (group 1; n = 15) conceived naturally within three months after the biopsy was taken (group 2; n = 20), or unsuccessfully underwent six IUI cycles and three IVF cycles with transfer of at least one high-quality embryo (group 3, RIF; n = 20). Our results demonstrated that immunohistochemical PRL-staining was present in 8/15 women from group 1 (53.3%), in 1/20 women from group 2 (5.0%), and in 11/20 women from group 3 (55.0%). Increased proliferation, analyzed by Ki67 expression, was seen in women that were pregnant during the biopsy, compared to all women combined that were not pregnant (p≤.01). In conclusion, our study demonstrates that premature expression of the decidualization marker PRL during the luteal phase is associated with RIF.

摘要

反复种植失败（RIF）是一种了解甚少的生殖病理学, 其定义是在至少3个连续IVF周期中无法临床妊娠。在本研究中我们调查以催乳素（PRL）表达为标志物的蜕膜化是否与RIF相关。我们进行了一项回顾性队列研究, 纳入采用子宫内膜活检的特发性不孕女性, 活检同一周期自然受孕的（组1, n = 15）, 活检后3个月内自然受孕的（组2, n = 20）, 进行6个周期宫腔内人工授精（IUI）和3个IVF周期并种植了至少一个高质量胚胎后未成功受孕的（组3, RIF, n = 20）。我们的结果表明, 免疫组化PRL染色存在于组1中8/15（53.3％）女性, 组2中1/20（5.0％）女性和组3中11/20（55.0％）女性。与所有内膜活检周期内未怀孕的女性相比, 怀孕女性的子宫内膜中观察到Ki67增殖指标表达增加（p≤0.01）。总之, 本研究表明黄体期蜕膜化标志物PRL的过早表达与RIF有关。

The Chinese abstracts are translated by Prof. Dr. Xiangyan Ruan and her team: Beijing Obstetrics and Gynecology Hospital, Capital Medical University, Beijing 100026, China.

Keywords:

• Decidualization
• prolactin
• endometrium
• repeated implantation failure

Introduction

Repeated implantation failure (RIF) is a reproductive pathology defined by the inability to achieve a clinical pregnancy in at least three IVF cycles with transfer of one or more high-quality embryo(s) in each cycle [1,2]. Typically, RIF covers all cases in which no gestational sac is visible on ultrasound, regardless whether initially human chorionic gonadotropin (hCG) was detected or not [1–3].

Different underlying conditions have been suggested as potential cause of RIF, such as obesity, thrombophilia [4,5], or autoimmune conditions [6,7]. In addition, several uterine conditions are associated with RIF, including congenital malformations, fibroids, polyps, adhesions, or hydrosalpinges [1,2]. However, a substantial number of cases of RIF cannot be attributed to a clear cause, and are therefore categorized as unexplained RIF. It has been suggested that unexplained RIF may be caused by defective embryo–endometrial interaction during implantation [8,9].

In preparation for an embryo to implant, human endometrial stromal cells (hESCs) undergo profound changes during a process called decidualization. Starting around the spiral arteries and the sub-epithelial space, hESCs obtain epithelial-like secretory and morphological characteristics [10]. Decidualization of the endometrium is histologically present from around day 18 of the menstrual cycle, when the stroma becomes increasingly edematous, followed by a marked increase in the cytoplasm of stromal cells near the terminal spiral arteries from around day 23 [11]. Embryo and endometrium are thought to start interacting during the ‘window of implantation’, a putative time-frame lasting from day 20 to 24 of the menstrual cycle [12–14].

An important hallmark of decidualization is the expression and secretion of specific proteins by hESCs, including prolactin (PRL) [10]. PRL, reflected by both mRNA and protein expression, has been demonstrated to be present only in late-luteal phase endometrium, either with or without the presence of a pregnancy, as well as during subsequent stages of pregnancy [15–19]. Elevated levels of PRL are not yet seen during the window of implantation, i.e. during the mid-luteal phase.

In case of RIF, it has been suggested that endometrial development is either lagging behind or is progressing too fast, which may lead to disturbed embryo–endometrial interaction and thereby implantation failure [20]. Increasing evidence suggests that the endometrial transcriptome during the window of implantation is often displaced or disrupted in women with RIF, supporting the notion of embryo–endometrial asynchrony [8,21–26]. However, it has not been investigated whether such transcriptomic differences may effectively lead to a different endometrial phenotype, including the decidualization status of hESCs.

Materials and methods

Patients

We searched a database of women that underwent a diagnostic endometrial biopsy as part of a routine fertility work-up between April 2001 and March 2014 at VU University Medical Center (VUmc), Amsterdam, The Netherlands. Eligibility criteria were: idiopathic subfertility (defined as being proven ovulatory, patent fallopian tubes, no luteal phase insufficiency, and a normal sperm count) lasting for at least 12 months, no previous ART attempts, scheduled for IUI. This study has been formally exempted from ethical approval granted by the Institutional Review Board of the VUmc. Eligible women were assigned to one of the study groups:

• Group 1 (pregnant): Women that conceived naturally within the same cycle as the biopsy was taken (n = 15). In all cases, both women and physicians were unaware of the early pregnancy and all women continued to have an ongoing pregnancy.

• Group 2 (ongoing pregnancy, OP): Women that conceived naturally within three months after the biopsy was taken.

• Group 3 (RIF): Women that did not conceive naturally and unsuccessfully underwent six consecutive cycles of IUI and three consecutive embryo transfers in three IVF cycles with transfer of at least one high-quality embryo per transfer.

Hormonal assays

Blood samples were processed to serum immediately after collection and stored at −20°C. Endocrine assessment was performed by The Central Laboratory of the VUmc. Estradiol was assessed by radio immunoassay (Diasorin Biomedica, Saluggia, Italy), and PRL was assessed by immunoluminometric assay (Bayer, Tarrytown, NY).

Endometrial biopsies

Endometrial biopsies were performed during the luteal phase, around 10 days after the rise in basal body temperature occurred, using a Pipelle aspiration catheter. Biopsied material was formalin-fixed and paraffin-embedded. Histological scoring was performed by a qualified gynecological pathologist according to the Noyes criteria [11]. The exact cycle day at which the biopsy was performed was calculated by counting back from the first day of the next menstruation.

Immunohistochemistry

Immunohistochemical staining was performed on 3 μm sections by an automated staining system (750-600, Benchmark Ultra, Roche, Tucson, AZ), using anti-PRL 1:1000 (A0569, Dako, CA) for 16 min or anti-Ki67 1:50 (M7240, Dako, CA) for 32 min. Matched IgGs were used as negative controls. Staining was detected with Optiview Detection (760-700, OptiView DAB IHC Detection Kit, Roche, Tucson, AZ), and photographed using an Olympus BX41 bright-field. PRL staining was scored as either positive or negative in the entire section. Ki67 staining was counted as the percentage of positive cells in three, randomly selected, images from the same biopsy in both the whole stroma and the perivascular space. Images were given a non-traceable identification number, and were analyzed by one independent and blinded investigator.

Statistical analyses

Significance was calculated using one-way ANOVA (IBM SPSS Statistics 23, Armonk, NY). p Values<.05 were considered statistically significant.

Results

Baseline characteristics

A total of 1541 women received a diagnostic endometrial biopsy, from which 55 women met the eligibility criteria to be included in this study. Baseline characteristics between the three study groups were not significantly different (Table 1).

Table 1. Baseline characteristics.

	Group 1 (pregnant) n = 15	Group 2 (OP) n = 20	Group 3 (RIF) n = 20	p Value
Maternal age, years (SD)	31.1 (5.0)	31.8 (3.8)	33.4 (3.8)	.26
Idiopathic subfertility, n	15	20	20	N/A
Cycle length, mean days (SD)	28.7 (1.8)	29.3 (3.2)	28.2 (2.3)	.40
Day of biopsy, mean CD (SD)	N/A	26.1 (3.4)	24.6 (3.2)	.22
Histological score, mean CD (SD)	24.9 (1.0)	25.2 (0.8)	24.8 (1.2)	.41
E2 serum, mean U/l (SD)	117.1 (38.7)	101.7 (39.9)	162.2 (169.3)	.27
PRL serum, mean pmol/l (SD)	0.3 (±0.1)	0.3 (±0.2)	0.2 (±0.1)	.50

OP: ongoing pregnancy; RIF: repeated implantation failure; CD: cycle day; N/A: not applicable; PRL: prolactin.

One-way ANOVA, Tukey’s post hoc test.

Endometrial parameters

Histology

There were no significant differences in the estimated cycle day (histological score) between women from each experimental group (Table 1).

PRL expression

When detectable, PRL staining was mainly localized in the perivascular space but had in some cases advanced into a more general spreading throughout the whole stroma (Figure 1(A)). In group 1 (pregnant), PRL was detectable in 8/15 women (53.3%; Figure 1(B)). In group 2 (OP), PRL was detectable in only 1/20 women (5.0%; OP versus pregnant, p=.005; Figure 1(B)), while in group 3 (RIF), PRL was detectable in 11/20 women (55.0%; RIF versus pregnant, p=.993; RIF versus OP, p=.002; Figure 1(B)). Possible confounders such as maternal age, day of biopsy and histological score did not contribute significantly to a logistic regression model, and were therefore excluded. Negative controls using rabbit IgGs instead of primary antibodies showed no staining (data not shown).

Figure 1. PRL expression in the late-luteal phase. (A) PRL expression was detectable in endometrial biopsies around blood vessels (A; perivascular) or more extensively throughout the stroma (A; stroma). (B) PRL expression was detected in 53.3% of women from group 1 (pregnant), in 5.0% of women from group 2 (OP), and in 55.0% of women from group 3 (RIF). OP: ongoing pregnancy; RIF: repeated implantation failure; PRL: prolactin. *p<.005 (one-way ANOVA, Tukey’s post hoc test).

hESCs proliferation

Endometrial stromal cell proliferation, determined by Ki67 expression, was assessed in two separate areas: the whole stroma and in the perivascular space (Figure 2(A)). No significant difference in Ki67 staining was observed between women from groups 2 and 3 (stroma, p=.28; perivascular, p=.17; Figure 2(B)), while a significant difference was seen when comparing women from groups 1 and 3 (stroma, p=.03; perivascular, p=.01). When all women that were not pregnant during the endometrial biopsy were combined (groups 2 and 3) and compared to women that were pregnant (group 1), we found a significantly lower percentage of Ki67-positive hESCs (stroma, p=.01; perivascular space, p<.01; Figure 2(B)).

Figure 2. Ki-67 expression in the late-luteal phase. (A) Ki67 staining was detectable throughout the stroma and the perivascular space. (B) Ki-67 staining was not significantly different between women from groups 2 and 3 (stroma, upper graph; perivascular space, bottom graph), while more expression of KI-67 was seen in women from group 1 compared to women from group 3. All women that were not pregnant combined (groups 2 and 3), had lower expression levels of Ki67 compared to women that were pregnant (group 1). *p<.05 (one-way ANOVA, Tukey’s post hoc test). OP: ongoing pregnancy; RIF: repeated implantation failure.

Next, we showed that proliferation was not associated with the decidualization status of hESCs, by comparing Ki67 staining inpatients from group 3 with either positive or negative PRL (stroma, 14.4% versus 15.0%, p=.86; perivascular space, 18.3% versus 21.6%, p=.51).

Discussion

In the present study, we demonstrated that endometrial PRL expression during the luteal phase, around cycle day 25, is not detectable in good prognosis women, i.e. group 2 (OP). PRL was detectable in those women that appeared to be pregnant at the time of biopsy, suggesting an embryo- or pregnancy-derived trigger initiating PRL expression. Interestingly, we also detected PRL expression in those women that later on developed RIF, indicating that premature advancement of endometrial decidualization, as reflected by PRL expression, may cause RIF. More specifically, premature advancement of endometrial decidualization may lead to embryo–endometrial asynchrony, resulting in dysfunctional embryo–endometrial interaction and consequently implantation failure [15–18].

The exact timing of the onset of PRL expression in luteal phase endometrium has been controversial. PRL expression was detected in the late-luteal phase, but some studies indicated its onset from as early as day 24 [15,16,27]. However, previous studies did not discern between women with and without RIF. The importance of this distinction is clearly demonstrated by our study, as roughly one third of all women included in this study displayed positive PRL staining around cycle day 25, but almost all of these cases could be attributed to women with RIF.

We also demonstrated that PRL expression, although only present in 53.3%, is associated with early pregnancy. However, this relatively low percentage may also indicate that decidualization is not essential for implantation to occur, but is rather initiated during later stages of pregnancy [19]. Alternatively, PRL expression during early pregnancy is specific only to the distinct site of embryo implantation, and may be missed in case the biopsy is taken elsewhere.

Regarding the proliferative activity of hESCs, we did not see any difference when comparing OP to RIF; although, reduced proliferation has been previously associated with RIF [23]. Possibly, changes in mRNA expression, which were predicted to result in reduced proliferation, may not lead to translational changes, and may therefore not lead to functional changes in proliferation.

Our results are in line with previous studies pointing to a displaced window of implantation as cause of RIF. Here, we demonstrate that RIF may be caused by too early rather than delayed endometrial advancement. Previous transcriptomic analyses of endometrial biopsies from RIF women have suggested that both too early- and delayed endometrial advancement are associated with RIF [8,24]. Remarkably, PRL was not reported as a differentially expressed gene in any of the large-scale gene expression studies of RIF [8,21–25]. One major difference between these studies and ours is that the transcriptomic analyses were performed earlier in the luteal phase. Possibly, at this early stage, only minimal to absent PRL expression is present, which may not be different between RIF and fertile women, whereas potential differences only become clear in later stages of the luteal phase.

In conclusion, our data suggest that premature decidualization during the luteal phase, reflected by PRL expression, may lead to embryo–endometrial asynchrony, and consequently RIF. Altogether, this raises an important question as to why PRL secretion is advanced in women with RIF, and how this may antagonize implantation. Possibly, unknown systemic or local mechanisms are triggering premature decidualization in these women, which then leads to premature secretion of PRL. This may merely be a sign of non-receptive and asynchronous endometrium or, alternatively, PRL may function to directly or indirectly antagonize implantation. To ultimately improve the prognosis of women suffering from RIF, future research should aim to better understand decidualization and its function in implantation.

Disclosure statement

The authors have no conflicts of interest to declare.