Introduction

A universal marker that is solely expressed in senescent cells has not been identified, but most senescent cells express p16INK4A, which is a cyclin-dependent kinase inhibitor binding cyclin D1/cyclin-dependent kinase complex, a key regulatory component in the transition from G1 to S-phase [1, 2]. In addition, p16INK4A is considered to be a tumour suppressor gene that inhibits cell proliferation arresting the cell cycle in the G1-phase [3, 4]. Different studies have shown the usefulness of immunoexpression of p16INK4A as a diagnostic biomarker in gynaecological diseases, including cervical premalignant and malignant lesions [5, 6], primary endometrial adenocarcinoma [7, 8], endometrial stromal sarcoma [9], endometrial tubal metaplasia and endometrial polyps [10].

However, the role of p16INK4A-positive senescent cells in endometrial receptivity remains unclear. It seems that p16-positive cells are responsible for acute cellular senescence remodelling at the time of embryo implantation [11]. A study of endometrial stromal cell (EnSCs) cultures showed that the transcription factor Forkhead box O1 (FOXO 1) induced acute senescence in a subpopulation decidual EnSCs [12]. Also, it was found that the senescence-associated secretory phenotype drives the initial auto-inflammatory decidual response linked to endometrial receptivity, and provided evidence that uterine natural killer (uNK cells) target and eliminate senescent decidual cells as the menstrual cycle progresses [12].

The current study was undertaken to investigate p16 expression in the functional layer of human endometrium in selected endometrial specimens, and to compare the percentage of p16-positive cells among women with unsuccessful embryo implantation, miscarriage and live births. The final aim of this study was thus to evaluate the potential ability of the number of p16-positive cells in the functional layer of the endometrium to serve as a biomarker for miscarriage.

Materials and methods

Study design and participants

Between August 2018 and March 2020, a prospective study was carried out at Nadezhda Women’s Health Hospital in Sofia, Bulgaria. It was hypothesized that the number of p16-positive senescent cells in the endometrial compartments (epithelial cells, glandular cells and stromal cells) could be related to IVF outcome. Therefore, the primary objective of the study was to compare the number of p16-positive cells in these three endometrial compartments in women undergoing IVF resulting in (1) unsuccessful embryo implantation, (2) miscarriage and (3) successful pregnancy and live birth. During the analysis, each patient was included in one of the three groups based on the IVF results from the first embryo transfer performed after the endometrial biopsy.

For the purpose of the study, endometrial biopsies were obtained from women enrolled in standard IVF/intracytoplasmic sperm injection (ICSI) programmes at our institution with history of recurrent implantation failure (RIF), defined as failure to achieve pregnancy after three or more consecutive IVF attempts in which one to three embryos of high-grade quality were transferred in each cycle. In all participants, endometrial biopsies were taken during a natural menstrual cycle using a Pipelle endometrial suction 7 days after luteinizing (LH) hormone surge, and at 2-8 months before initiation of controlled ovarian stimulation and transabdominal ultrasound-guided frozen embryo transfer procedure. Only cases with embryo transfer of frozen good-quality blastocysts (day 5 embryos) were included. Other inclusion criteria were less than 48 years of age, regular ovulatory cycles and both fallopian tubes without hydrosalpinx. Women with ectopic pregnancy, history of pelvic inflammatory disease, polycystic ovary syndrome (PCOS), moderate or severe endometriosis diagnosed by transvaginal ultrasound and male partner with abnormal semen analysis were excluded. Preimplantation genetic testing for aneuploidy (PGT-A) by next-generation sequencing (NGS) was performed on trophectoderm samples in 14.1% (44/311) of all studied cases. In these particular cases, only euploid embryos were used for the embryo transfer.

In this study, an unsuccessful implantation was defined as “negative” beta-hCG test. Miscarriage was defined as pregnancy loss due to expulsion or death of the foetus or embryo weighing < 500 g, and before 20–24 weeks’ gestational age. All miscarriages were visualized in the uterus. Recurrent miscarriage was defined as three or more pregnancy losses (one after the embryo transfer following the endometrial biopsy and at least two miscarriages before the endometrial biopsy) prior to 24 weeks’ gestation.

All experimental procedures and sample procurements were approved by the ethics committee of the hospital. Written informed consent was obtained from all participants.

Immunohistochemistry method

Endometrial biopsies were fixed overnight in 10% neutral buffered formalin at 4°C and wax embedded in paraffin using the Leica TP1020 Semi-enclosed Benchtop Tissue Processor (Leica Biosystems, Wetzlar, Hesse, Germany). Tissues were sliced into 4-μm sections on a manual rotary microtome (Leica Biosystems) and mounted on adhesive slides by 1-h incubation at 60°C. The deparaffinization was done by heated paraffin embedding module (Cat. No. EG1150H, Leica Biosystems).

Immunohistochemical staining was performed on the paraffin sections by the method using a Novolink Polymer Detection System (Leica Biosystems). Briefly, each section was deparaffinized in xylene and rehydrated in graded alcohol. Subsequently, the sections were placed in 0.01 M citrate buffer (pH 9.0) for 30 min at 95°C. Next, 3% hydrogen peroxide was applied to block endogenous peroxidase activity and then the tissue was incubated in 0.4% casein in phosphate-buffered saline to reduce non-specific binding of primary antibody and polymer. The sections were incubated with one rabbit polyclonal antibody against human p16INK4 (MAD-000690QD-7, Master Diagnostica, Granada, Spain). Then, samples were treated with Novocastra Postprimary Block, containing 10% (v/v) animal serum in tris-buffered saline to enhance penetration of the subsequent polymer reagent. Consequently, poly-HRP anti-mouse/rabbit IgG reagent (NovoLink Polymer) containing 10% (v/v) animal serum in tris-buffered saline was applied to localize the primary antibody, and the reaction product was visualized by incubation with the substrate/chromogen, 3,3′-diaminobenzidine (DAB) prepared from Novocastra DAB Chromogen and NovoLink DAB Substrate Buffer (Polymer), as a brown precipitate. Finally, the sections were counterstained with Novocastra haematoxylin stain (0.02%). Appropriate positive control cases were used for p16 antibody. As a negative control, sections were treated with phosphate-buffered saline (PBS) with omission of the primary antibody.

Staining was assessed by using Olympus CKX41 inverted microscope (Olympus, Melville, NY, USA) and 10 high-resolution images of each specimen were captured at ×400 magnification with a UI-1480LE (or Nikon DS Fi1) digital camera (IDS Imaging Development Systems GmbH, Obersulm, Germany). The ratio between p16-positive cells (brown stain) and total endometrial cells in different tissue compartments (blue stain) was assessed both manually and using computer-assisted image analysis with colour deconvolution (Image J software, NIH) for all images. Positive staining was identified by brown-coloured products in the nucleus and/or in the cytoplasm. Nuclear and cytoplasmic staining in each section was evaluated by the percentage of stained cells. For p16, cytoplasmic and nuclear staining were both considered positive; p16-positive cells were quantified separately in glandular epithelium, luminal epithelium and stromal compartments. The stained and unstained endometrial stromal cells and luminal epithelial cells were visually counted in three different representative microscopic fields (0.04 mm2) for each sample. The person who counted the number of stained cells was blinded to the patient IVF outcome. The number of glands with stained cells was determined by enumeration of 100 glands in each sample. The percentage of stained cells was determined by counting the number of stained and unstained cells in each gland (enumeration of ≥ 20 glands with stained cells per sample). In this study, the intensity of staining was not quantified.

Statistical analysis

The sample size was determined using the Raosoft sample size calculator, and it was estimated that at least 60 samples in each study group would be necessary to detect a medium effect size (0.25 standard deviation (SD)) between the groups at a level of significance of 0.05 with 80% statistical power. Continuous data are expressed as mean and SD. The Student’s t test or the Wilcoxon rank-sum test, the Mann-Whitney U test or the Kruskal-Wallis test were applied according to normal or not normal distribution of data. The percentage of p16-positive was also compared between women with miscarriage and those with successful pregnancy and live birth according to women’s age categorized as ≤ 34 years and > 34 years. The relationship between the percentage of p16-positive cells in the endometrial compartments and the women’s age was analysed with the Spearman’s rank-order correlation coefficient (ρ). The optimal cut-off for the percentage of p16-positive cells to predict IVF outcome in terms or miscarriage or successful pregnancy and live birth was assessed by the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Sensitivity and specificity of cut-off values were also calculated. The optimal cut-off was determined based on the minimal absolute value of the difference between the sensitivity and specificity values [13, 14]. The obtained cut-off was verified by second approach, known as the point closest-to-(0,1) corner in the ROC plane which defines the optimal cut-off as the point minimizing the Euclidean distance between the ROC curve and the (0,1) point [15]. Statistical analyses were performed with the SPSS statistical software for Windows, version 21.0 (SPSS, Chicago, IL, USA).

Results

Study population and baseline characteristics

During the study period, endometrial biopsies were taken from 347 eligible women but 26 were excluded because embryo transfer was not performed or poor-quality embryos were transferred. Of the remaining 321 women undergoing embryo transfer (single or double), 10 with biochemical pregnancy were excluded. These women had an initial beta-hCG-positive test but the pregnancy failed to progress to the point of ultrasound confirmation and clinical detection. This specific group was excluded because the sample size of 10 cases was too small. Therefore, the study population included 311 women aged between 22 and 48 years, who were divided into three groups according to IVF outcome, including 151 women with unsuccessful embryo implantation (negative human chorionic gonadotropin (hCG) test), 66 with miscarriage and 94 with live birth. As shown in Table 1, age, body mass index (BMI), indication of IVF, previous IVF attempts, oocytes retrieved, the number of embryos transferred and the number of PGT-A-tested embryos were similar among women in the three study groups.

Table 1 Baseline characteristics of 311 women undergoing endometrial biopsy prior to embryo transfer
Full size table

p16 expression in endometrial compartments

As shown in Fig. 1, all three endometrial compartments, luminal epithelial cells, glandular epithelial cells and stromal cells showed p16-positive cells. p16-positive luminal epithelial cells showed relatively long and well-developed microvilli. Focal p-16 reactivity in the luminal epithelium was also observed (Fig. 1c). However, most stromal cells were p16-negative, although most p16-positive stromal cells were morphologically undistinguishable from the surrounding unstained cells, and some showed a spindle and elongated morphology (Fig. 2).

Fig. 1
figure 1

Representative immunohistochemical staining for p16 expression in the functional layer of human endometrium. a Stromal cells. b Glandular epithelial cells. c Luminal epithelial cells. Scale bar = 200 μm (×400 magnification). d View of the endometrium with p16 positively stained cells in the luminal epithelium, glands and stroma (×200 magnification)

Full size image
Fig. 2
figure 2

Examples of p16INK4a immunohistochemically stained positive endometrial stromal cells with spindled and elongated morphology (×800 magnification)

Full size image

The percentage of p16-positive cells varied between 1.2 and 95.2% in the luminal epithelium, between 0.05 and 73.53% in the endometrial glands and between 0.03 and 4.73% in the endometrial stromal cells. The mean (SD) percentage of p16-positive cells was significantly higher in the luminal epithelial cells (27.44% (24.18%)) as compared with either the glandular epithelial cells (7.5% (6.3%); mean difference 5.24%, 95% confidence interval (CI) 0.72-0.86, P < 0.01) or the stromal cells (0.67% (1.05%); mean difference 5.24%, 95% CI 0.72-0.86, P < 0.01). The percentages of p16-positive cells in the glandular epithelial cells and the luminal epithelial cells showed a significant correlation (ρ = 0.76, P < 0.01), which also correlated significantly (P < 0.01) with the women’s age (ρ = 0.39 for the glandular epithelial cells, ρ = 0.48 for the luminal epithelial cells). In contrast, p16 positivity of stromal cells did not correlate with other endometrial epithelial cells or women’s age.

Relationship between p16-positive endometrial cells and IVF outcome

The percentages of p16-positive cells were significantly higher in women with live birth as compared with those with miscarriage, both for p16 positivity in the glandular epithelial cells (9.30% (8.96%) vs. 2.09% (3.81%); mean difference 5.24%, 95% CI 2.32-8.14, P = 0.01) and in the luminal epithelial cells (35.22% (25.83%) vs. 11.74% (9.98%); mean difference 17.83%, 95% CI 9.15-26.51, P < 0.01) (Fig. 3). Differences in the percentages of p16-positive stromal cells between the groups of live birth and miscarriage were not found (0.57% (0.89%) vs. 1.03% (1.21%); mean difference −0.54%, 95% CI 0.34-1.43, P = 0.23).

Fig. 3
figure 3

a Box plots representing the percentage of p16-positive cells in the luminal epithelial cells, gland epithelial cells and stroma cell of women with successful pregnancy and live birth, unsuccessful implantation and miscarriage. b p16 expression in the endometrial glands of patients with live birth, unsuccessful implantation and miscarriage (×400 magnification). c p16 expression in the endometrial luminal epithelium of patients with live birth, unsuccessful implantation and miscarriage (×200 magnification). The percentage of p16-positive cells was significantly higher in women with live birth compared with women with miscarriage (P < 0.05)

Full size image

Relationship between p16-positive endometrial cells and age

Assessment of p16 positivity in women with successful pregnancy and live birth and women with miscarriage categorized as ≤ 34 years or > 34 years of age showed significantly lower percentages of p16-positive cells in the three endometrial compartments for women with miscarriage in both age groups, except for p16-positive stromal cells (Table 2). Similar findings were observed when women with successful pregnancy and live birth were compared with a subset of 15 women with recurrent miscarriage. Women in the group of recurrent miscarriage also showed percentages of p16 positivity even lower than those in the miscarriage group, particularly regarding the percentage of p16-positive luminal epithelial cells and in both age groups.

Table 2 Percentage of p16-positive cells in the different endometrial compartments according to IVF outcome and women’s age
Full size table

p16 positivity as a predictive marker of miscarriage

The ROC curve using the optimal cut-off percentages of p16-positive cells in glandular epithelial cells (12.5%) and in luminal epithelial cells (3.2%) for predicting miscarriage is shown in Fig. 4. The corresponding AUC values to these optimal cut-off percentages were 0.80 (95% CI 0.73-0.87) for p16-positive glandular epithelial cells, and 0.79 (95% CI 0.72-0.86) for p16-positive luminal epithelial cells. Therefore, a percentage of 12.5% of p16-positive cells in the luminal endometrium was associated with 71.3% (95% CI 68.5-78.2) sensitivity and 74.2% (95% CI 70.6-81.3) specificity for predicting miscarriage after IVF, whereas a percentage of 3.2% p16 positivity in the glandular epithelial cells showed 74.5% (95% CI 71.1-80.9) sensitivity and 71.2% (95% CI 69.2-77.4) specificity.

Fig. 4
figure 4

Receiver operating characteristic (ROC) curve for prediction of miscarriage using percentage of 16-positive cells in glandular epithelial and luminal epithelial as a prognostic factors. Area under the ROC curve is (a) (95% CI): 0.80 (0.73–0.87) for the luminal epithelial p16-positive (blue line) and (b) (95% CI): 0.79 (0.72–0.86) for the glandular epithelial p16-positive cells (green line)

Full size image

Discussion

Current advances in IVF that include special conditions for embryo culture, vitrification, trophectoderm biopsy, followed by preimplantation genetic screening and selection of euploid embryos allow an overall improvement in pregnancy rates [16]. In addition, the choice of transabdominal or transvaginal ultrasound guidance could have an essential effect on live birth success rate. A systematic review and meta-analysis of randomized controlled studies (RCTs) focused on ultrasound-guided embryo transfer [17] showed that transabdominal guidance during embryo transfer, which was also used in our study, compared with conventional clinical touch was more favourable regarding clinical pregnancy and ongoing birth rates, with no differences in miscarriage and ectopic pregnancy rate based on the analysis of 14 RCTs. However, the optimal embryo transfer strategies are still under development and they should also be based on the information obtained from endometrial receptivity assessment [18, 19]. Many studies are focused on the analysis of specific genetic [20, 21] or immunohistochemical biomarkers of endometrial receptivity [22, 23]. In order to understand the role of senescent cells in endometrial receptivity and pregnancy outcome in women with RIF, our study was focused on immunohistochemical quantification of p16-positive senescent cells in the functional layer of human endometrium during the mid-luteal phase.

The present study shows the presence of p16-positive cells in the three endometrial compartments: luminal epithelial cells, glandular epithelial cells and stromal cells. The percentage of p16 positivity, however, was higher in the luminal epithelium as compared to the other two compartments, which was especially scarce in stromal cells. Interestingly, there was a correlation between p16 positivity in the luminal and glandular epithelial cells, and both also correlated with the women’s age. In another study, relatively high quantities of p16-positive epithelial cells were found during the implantation window, and they were significantly more abundant in the luminal than in the glandular epithelium [12]. These results were confirmed in our study where the p16-positive luminal epithelial cells were 4-fold to 6-fold higher compared to the percentage of p16-positive glandular cells. It was proposed that p16-positive luminal epithelial cells play an important role in directing the human embryo to preferential sites of implantation [12, 24, 25].

Our hypothesis that the number of p16-positive senescent cells in the endometrial compartments could be related to IVF outcome was confirmed. The percentage of p16-positive cells in the luminal and glandular epithelial cells was significantly higher in women with successful pregnancy and live birth than in those in whom pregnancy ended in miscarriage. The percentage of p16-positive cells in the luminal epithelium was higher than in the glandular epithelium. However, the association between p16-positive cells in the endometrial stroma and IVF outcome was unclear. It has been suggested that the mechanism driving senescence of endometrial epithelial cells may differ from that in stromal cells [26]. Endometrial epithelial cells express human telomerase reverse transcriptase and exhibit dynamic telomerase activity, which falls dramatically at the mid-luteal phase, preceding the rise in p16-positive epithelial cells [27, 28]. Telomere lengths were found to be also shortest during the mid-secretory phase, suggesting an alternative mechanism of triggering cell cycle exit and senescence of epithelial cells [26, 29].

An interesting finding of the study was an even greater reduction in the percentage of p16-positive luminal and glandular epithelial cells in women with recurrent IVF failure, which additionally emphasizes the impact of the deficit of p16-positive endometrial cells on spontaneous abortion. Therefore, it may be suggested that a decrease in senescent p16-positive cells may contribute to create an unfavourable environment for implantation. According to our findings, the percentage of p16-positive cells in the endometrial luminal and glandular epithelium may be more important in the supportive environment for the establishment and maintenance of pregnancy. However, our preliminary findings should be confirmed in further studies with larger study populations. Recurrent pregnancy loss has been found to be strongly associated with enhanced cellular senescence [25, 30, 31], but counterbalance of senescence decidual cells during the luteal phase by uNK cell-mediated clearance ensures implantation competence of the endometrium [12, 32]. We found out that in the group with recurrent miscarriages, the decrease in the p16-positive cells was even more significant compared with the standard miscarriage group. These results confirm the existence of association between the miscarriage and the change in the percentage of senescent cells in human endometrium during the implantation window. On the other hand, we found that a cut-off of 12.5% of the percentage of p16-positive cells in the luminal endometrium showed a relatively high sensitivity (71.3%) for discriminating an unfavourable miscarriage from a favourable IVF outcome. Accordingly, the percentage of p16-positive luminal epithelial cells could have applicability as a predictive biomarker of miscarriage after IVF.

Limitations and strengths of the study should be considered. In relation to limitations, firstly, the group of 10 patients with biochemical pregnancy was excluded because the sample size was too small and any comparisons with the three study groups (unsuccessful pregnancy, miscarriage, live birth) would have been questionable because of differences in the number of patients that composed each group. However, it would be interesting to assess the presence of p16-positive cells in cases of biochemical pregnancy and to make a comparison with the groups of unsuccessful pregnancy, miscarriage and live birth. Also, control aged-matched women without endometriosis undergoing IVF were not included in the present study. Secondly, this study was conducted in a fertility clinic attending women undergoing IVF to have a baby and focused on the role of p16-positive cells in endometrial biopsies as a marker of miscarriage, although it would have been interesting to have a control group of patients without IVF and a normal pregnancy, in order to assess significant changes in the number of endometrial p16-positive cells compared to the other studied groups (unsuccessful pregnancy, miscarriage, live birth). Thirdly, quantification of senescent (p16-positive cells) was based on immunohistochemical analysis only, and other methods such as flow cytometry or quantitative real-time PCR were not used. Finally, other biomarkers of cell senescence (e.g. β-gal-positive cells, p21-positive cells, p53-positive cells or deiodinase type II expression) were not included in the study. On the other hand, strengths include the prospective design, the relatively large study population, exclusion of potential confounders, such as female disorders, male factor infertility or changes during the menstrual cycle (all biopsies were taken during the mid-luteal phase) and the use of an accurate quantification method to count the number of stained and unstained cells by a blind researcher.

In conclusion, during the mid-luteal phase of the cycle, the endometrial luminal epithelium has the highest percentage of p16-positive senescent cells, followed by the glands and the stroma. There is a significant age-dependent increase in the percentage of p16-positive senescent cells in the luminal epithelium, which is more noticeable than changes in the percentage of p16-positive senescent cells in the endometrial glands. Miscarriage after IVF, and particularly recurrent miscarriage, is associated with a decreased number of p16-positive senescent cells especially in the luminal epithelial cells, but also to a lesser extent in the glandular cells. This feature was confirmed in different age groups and could be applied as a non-invasive biomarker for miscarriage risk assessment.