Chronic endometritis diagnosis and fertility outcomes: an old unresolved question

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

During the inclusion period, 661 infertile patients underwent a hysteroscopy. Among them, 610 were excluded due to a normal appearance of the uterine cavity without systematic endometrial biopsy ( n = 443). The other exclusion criteria were women with hormonal treatment for endometriosis ( n = 61), women with antibiotic therapy for non-gynecologic reasons during the 2 months before hysteroscopy ( n = 53), women with long-term NSAID treatment ( n = 37), women with severe Asherman syndrome ( n = 10), and women over 45 years old ( n = 6). Therefore, the study population was composed of 51 women (7.7%) who underwent both a hysteroscopy and an endometrial biopsy ( Fig. 1 ). The epidemiologic characteristics of the study population are summarized in Table 1 . Median age was 36 years (range: 26–44 years). Concomitant diagnosis of deep infiltrating endometriosis was noted in 12% of patients. Flow chart of the study. Characteristics of the overall study population and according to CE status. Data are presented as n (%) or as median (min–max). BMI, body mass index; ART, assisted reproductive technology. Using CD138 immunostaining, 44 (86.2%) patients had a CE diagnosis and seven had no CE based on negative CD138 expression. The mean number of plasma cells per HPF was 155.9. The distribution of CD138 immunostaining was less than 5, between 5 and 50, and over 50 in 12 (27.2%), 19 (43.1%), and 13 (29.5%) patients, respectively. The correlation coefficient between the first and second reading for the diagnosis of CE (presence or absence of plasma cells) was moderate (Cohen’s Kappa: 0.44 (95% CI: −0.059; 0.767)). A discrepancy in plasma cell counts by the two pathologists of 20% in 54.9% of cases. Therefore, the correlation coefficient between the first and second reading for plasma cell quantification was good (intraclass correlation coefficient: 0.948). Among the 51 patients of the study population, 23 had a normal uterine cavity appearance (45%). For the 28 patients with abnormal uterine cavity, ten (35.7%) had hysteroscopic findings suggestive of CE, mainly corresponding to endometrial hyperemia areas in six patients, strawberry appearance in two patients, and micropolyps in only two patients ( Fig. 1 ). The sensitivity, specificity, PPV, NPV, and accuracy of hysteroscopy to diagnose CE were 22, 100, 100, and 17%, with an infinite OR and an accuracy of 68.6%, respectively. Conventional histologic analysis was blinded to hysteroscopic findings. At conventional histology, the average slide area analyzed was 152 mm 2 (range: 10–427). Using HE, histology allowed to diagnosis of CE in 54% (28/51) of patients. The sensitivity, specificity, PPV, NPV, and the accuracy of conventional histology were 61.4, 100, 100, 29, and 66.7%, respectively. Of the 23 patients with normal uterine cavity appearance, 16 (69.5%) had positive CD138 immunostaining. For the ten patients with hysteroscopic findings suggestive of CE, all had positive CD138 immunostaining. For the remaining 18 (64.2%) patients with uterine cavity abnormality (polyp, myoma, partial synechia, and endometrial hypertrophy) but without hysteroscopic CE signs, as defined in the material and methods, all had positive CD138 immunostaining. The epidemiologic characteristics of the patients according to the CE diagnosis based on CD138 positivity are summarized in Table 1 . No differences were observed between the groups, except for a higher BMI, a lower nulliparous rate, and a shorter duration of secondary infertility observed in the CE group ( Table 1 ). Among the 44 patients with CE, 19 (43%) were treated with antibiotics. The indication for antibiotic treatment was left to the discretion of the ART physician. During follow–up, 27.2% (12/44) underwent a second biopsy, including five patients treated with antibiotics and seven untreated. On the second biopsy in untreated patients, three of seven patients had an increased plasma cell count, while a spontaneous decrease or negative plasma cell count was noted in the remaining four patients (Annex 1 (see section on Supplementary materials given at the end of the article)). On the second biopsy of treated patients, one had a negative plasma cell count and four had a decreased plasma cell count. Median interval between the two biopsies was 12 months (range: 4–17 months) (Annex 1). Among patients with CE, 34% (15/44) achieved pregnancy. The PR was significantly higher in patients treated with antibiotic (53%, 10/19) than in untreated patients (20%, 5/25) with an OR of 4.4 (95% CI: 1.17–16.8; P = 0.03). The median time to achieve pregnancy was 12 months (range: 11–36 months). The plasma cell count at diagnosis was not predictive of achieving pregnancy ( P = 0.65) with an OR of 1.00. Among pregnant patients, the plasma cell count was less than 50 in 80% of patients (12/15).

From January 2018 to August 2020, we conducted a retrospective study from a prospective database at the Centre Hospitalier Intercommunal de Créteil (CHIC), France, to evaluate the diagnosis and impact of CE on pregnancy outcomes in infertile women. The epidemiologic characteristics of the patients were recorded. Inclusion criteria were individuals aged 18–45 years, treated for primary or secondary infertility, who underwent both hysteroscopy and endometrial biopsy, and who were covered by French health insurance and fluent in spoken and written French. Exclusion criteria included recent use (within 2 months before biopsy) of antibiotics or anti-inflammatory medications, need for oocyte or sperm donation, undergoing hysteroscopy without endometrial biopsy or vice versa, a delay of more than 1 month between biopsy and hysteroscopy, and clinical signs suggestive of CE. Clinical data were abstracted from patient charts, including epidemiologic and socio-demographic factors, type of infertility (primary or secondary), smoking, gynecologic symptoms, hysteroscopic findings, and treatment received for CE. In our department, eight gynecology specialists in assisted reproductive techniques (ART) performed hysteroscopies. All physicians had more than 3 years of experience in hysteroscopy. Approval for the study was obtained from the ethical committee of the CNGOF (Collège National des Gynécologues et Obstétriciens Francais) (CEROG: 2022-GYN-0307) and all patients gave their consent for the study. On conventional histology, the CE diagnosis is based on the presence of endometrial stromal plasmacytes (ESPCs), typically appearing as large lymphocytes with a high nucleus/cytoplasm ratio, basophilic cytoplasm, and eccentric nuclei with heterochromatin rearrangement called the ‘spoke-wheel’ or ‘clock-face’ pattern. However, other endometrium-component cell types, such as NK cells, macrophages, and stromal fibroblasts, may exhibit a morphological appearance of ESPCs ( Kitaya et al. 2018 ). In this specific setting, syndecan-1 (also known as CD138), a heparan sulfate/chondroitin sulfate proteoglycan expressed on the plasma membrane of ESPCs, is especially relevant to assess CE diagnosis. Using immunohistochemistry (IHC) for CD138 (IHC-CD138) markedly (odds ratio: 2.8) improves the sensitivity (100 vs 75%), specificity (100 vs 65%), interobserver variability (96 vs 68%), and intra-observer variability (93 vs 47%) in the histologic diagnosis of CE, supporting its use as the most accurate diagnostic tool ( Miguel et al. 2011 , Kitaya & Yasuo 2013 ). As such, the CD138 immunostaining was considered the gold standard for CE diagnosis. Concerning the ESPC cut-off to diagnose CE, no clear consensus exists. Despite some limits, in accordance with previous studies ( Hirata et al. 2021 , McQueen et al. 2021 ), we decided to use the cut-off of ESPC ≥1 in ten HPFs. Histology was performed on endometrial biopsy using a Cornier device during a clinical visit or at the end of the hysteroscopy. For conventional histology examination, specimens were stained by HE. Briefly, for IHC, 4 μm slides were obtained from formalin-fixed paraffin-embedded (FFPE) tissues and were deparaffinized and dehydrated using alcohol. IHC staining was performed using an automated system (Ventana, BenchMark ULTRA, 315485, Roche, Switzerland), and a dilution of 1:250 was used for the mouse anti-CD138 antibodies (760-4248 Roche) in accordance with laboratory recommendations. The immunostaining was carried out with 3, 3′-diaminobenzidine chromogen (DAB) and counterstained with hematoxylin. The sections were then incubated with a 1:250 dilution of mouse anti-CD138 antibody (Gene Tech, clone B-A38, China). Subsequently, the slides were washed and incubated with a horseradish peroxidase-conjugated secondary antibody for 30 min. The slides were read under the microscope at ×40 magnification, and the plasma cell count was performed with the CaloPix software (CaloPix 4.1.0.9, TRIBVN, France). Two independent pathologists read the slides and performed plasma cell counts. In case of discrepancy between pathologists in either CE diagnosis or plasma cell count, a third pathologist reviewed the slides. In accordance with a previous study, the diagnosis of CE was considered positive when at least one plasma cell was observed on each slide at ten HPF ( McQueen et al. 2021 ). The material used was rigid hysteroscopy STORZ, 30°, 4.1 mm, GmbH Germany. In 2019, based on a systematic review and the agreement of the Delphi poll, the International Working Group for the Standardization of CE Diagnosis proposed the following hysteroscopic diagnostic criteria for CE ( Cicinelli et al. 2019 ): Strawberry aspect: first described by Cravello et al. (1997) , recognized as large hyperemic localized or scattered mucosal areas flushed with white central points. Focal hyperemia: small areas of hyperemic mucosa. Hemorrhagic spots: focal reddish mucosa with sharp and irregular borders, possibly in continuity with capillaries. Endometrial micropolyps: first described by Cicinelli et al. (2005 a , b ) , typically visualized as a cluster of less than 1 mm-sized protrusions on the focal or entire mucosal surface with a distinct connective vascular axis. Stromal edema: thick and pale appearance of the mucosa in the follicular phase originating from the stromal compartments (a normal finding during the secretory phase). Strawberry aspect: first described by Cravello et al. (1997) , recognized as large hyperemic localized or scattered mucosal areas flushed with white central points. Focal hyperemia: small areas of hyperemic mucosa. Hemorrhagic spots: focal reddish mucosa with sharp and irregular borders, possibly in continuity with capillaries. Endometrial micropolyps: first described by Cicinelli et al. (2005 a , b ) , typically visualized as a cluster of less than 1 mm-sized protrusions on the focal or entire mucosal surface with a distinct connective vascular axis. Stromal edema: thick and pale appearance of the mucosa in the follicular phase originating from the stromal compartments (a normal finding during the secretory phase). Antibiotic treatment with Roxithromycin for 10–15 days was administered. None of the patients had a pro/prebiotic treatment. No microbiology assessment was performed on the study population. The data were collected on an Excel worksheet (Microsoft Corporation, USA), and all statistical analyses were performed using R studio software (version 1.3.1093, freely available online). Comparison for qualitative variables was assessed by the Chi-square test or Fisher’s exact test and by the Student’s t -test for quantitative variables. A P -value <0.05 was considered to denote a significant difference. Sensitivity is the ability of a test to correctly identify patients with a disease. Specificity is the ability of a test to correctly identify people without the disease. True positive is when the person has the disease and the test is positive. True negative is when the person does not have the disease and the test is negative. Inter-observer agreement analysis was performed using Cohen’s quadratic Kappa test for the diagnosis of CE and the intraclass correlation coefficient for the correlation of plasma cell count. This agreement is considered excellent when Kappa is between 1 and 0.8, good between 0.61 and 0.80, average between 0.41 and 0.6, poor between 0.21 and 0.40, and bad between 0 and 0.20.

Despite some limits of the present study, our results underlined the low relevance of hysteroscopic features, including micropolyps, to suggest CE. However, even in case of normal endometrial appearance, infertile patients might benefit from a systematic evaluation of CE, as antibiotic therapy seems to have a positive impact on the PR. Further studies should clarify the role of antibiotics in improving fertility outcomes in this specific setting.

The present study has demonstrated the low relevance of hysteroscopy to suggest CE, even in the presence of micropolyps. Moreover, conventional histology had low accuracy to assess the CE diagnosis. Antibiotic therapy confirms its relevance to enhance PR. CE diagnosis is a major challenge for infertile women due to the risk of increased implantation failure ( Kushnir et al. 2016 , Cicinelli et al. 2018 ). In the present study, including infertile women without symptoms of uterine infection, the rate of CE on hysteroscopy was only 19.6% (10/51). Moreover, hysteroscopy was considered normal in 36% of patients with CE, supporting systematic CD138 staining in the context of infertility. In the patients with hysteroscopic signs suggestive of CE, hyperemia was the most relevant feature, while the presence of micropolyps was observed in only 20% cases, raising several issues on CE diagnosis criteria. Indeed, micropolyps (<1 mm) diagnosis is easy in case of multi-micropolyps, but difficult in case of isolated micropolyps. Moreover, as previously suggested, the experience of physicians performing hysteroscopy can be discussed ( Cicinelli et al. 2019 ). In our series, all hysteroscopies were performed by physicians with high experience. Our results contrast with those of Cicinelli et al. (2005 a , b ) showing the relation between micropolyps and CE. These authors reported an increased likelihood of CE in women with micropolyps with an OR of 124.2 (95% CI: 50.3–205.4). Their respective sensitivity, specificity, positive and negative predictive values of micropolyps to predict CE were 54, 99, 94, and 89%, and a diagnostic accuracy of 90%. However, our results agree with those of Furui et al. , showing that endometrial congestion was the only hysteroscopic finding significantly associated with CE ( Furui et al. 2024 ). To overcome the diagnostic gap between hysteroscopy findings and histology to assess CE, using logistic regression analysis, Song et al. showed that hyperemia area degree ≥2, micropolyps, polypoid hyperplasia, and history of ectopic pregnancy were independent risk factors for CE, allowing development of a nomogram with a ROC curve of 0.801 (95% CI: 0.742–0.861) ( Song et al. 2019 ). Interestingly, Kitaya et al. developed a deep learning model (ARChival) of hysteroscopic image-based prediction for CE in infertile women, exhibiting a respective accuracy, AUC, sensitivity, and specificity of 0.845, 0.98, 77.97%, and 100%, but requiring external validation ( Kitaya et al. 2024 ). Finally, another crucial data of the present study is the association between CE and the presence of uterine cavity abnormality, underlining the need to manage these women with systematic antibiotic therapy. Recent guidelines on indications and techniques for endometrial biopsy to diagnose CE have recommended the use of a grasp biopsy technique as the first choice in reproductive-aged women ( Vitale et al. 2023 ). However, no clear histologic criteria have been suggested to affirm CE diagnosis. In the present study, low relevance of conventional histology was observed to diagnose CE, with a sensitivity, specificity, PPV, NPV, and accuracy of 61.4, 100, 100, 29, and 66.7%, respectively. This can be linked to difficulties in identifying plasma cells by conventional histology, which is considered the gold standard for CE diagnosis ( Chen et al. 2016 , Li et al. 2021 ). Our results are in line with those of Pérez-Cejuela et al. , showing a low agreement between hysteroscopic findings and conventional histology, with a Cohen’s kappa index of 34% ( Pérez-Cejuela et al. 2023 ). These results underline the usefulness of combining hysteroscopy and endometrial sampling with IHC to assess CE. In the current study, using CD138 immunostaining, a large range of plasma cells from 0 to over 1,000 was observed, with perfect concordance in plasma cell counts according to readers in only one-third of cases. When CE diagnosis was based on the presence of at least one plasma cell, the correlation coefficient between readers was moderate (Cohen’s Kappa: 0.44 (95% CI: −0.059; 0.767)), underlying the need for additional tools such as the contribution of deep learning artificial intelligence analysis ( Greeley et al. 2024 ). However, the correlation coefficient between the first and second reading for plasma cell quantification was good (intraclass correlation coefficient: 0.948). In this specific setting, previous studies have underlined the contribution of multiple myeloma antigen 1 (MUM-1) evaluation, which is expressed on plasma cells, activated B cells, and T cells ( Falini et al. 2000 , Wasco et al. 2008 ). Another concern is how to evaluate the impact of antibiotic therapy on PR. In the present study, despite a small sample size, PR was higher in patients treated with antibiotics vs untreated patients, with an OR of 4.4 (95% CI: 1.17–16.8; P = 0.03). Our results agree with those of a retrospective study showing the relevance of evaluating and treating CE before IVF procedure on PR ( Vaduva et al. 2023 ). Moreover, the randomized trial of Cicinelli et al. demonstrated the positive impact of antibiotic therapy vs no treatment for CE on PR ( Cicinelli et al. 2021 ). Similarly, Dang et al. showed that the PR in the CD138-positive group was lower compared to the CD138-negative group (64.79 vs 81.30%, P < 0.05) ( Dang et al. 2024 ). Using multivariate analysis, these authors demonstrated that CD138 positivity was an independent risk factor to predict embryo implantation failure. However, no evaluation of CD138 changes after antibiotics was performed. In the current study, 12 patients underwent a second endometrial biopsy: five patients after antibiotic treatment and seven untreated. Among patients treated, only one of five had negative CD138 immunostaining, and the remaining four patients had a decrease, while in untreated patients, more than half (four of seven) had spontaneous decrease or negative plasma cells. However, our results of the second biopsy should be analyzed with caution due to the large range of delay between antibiotic treatment and the second biopsy. These data raise issues on how to evaluate the efficacy of antibiotics on CE and factors contributing to spontaneous CD138 immunostaining decrease in untreated patients. A potential explanation is linked to desquamation of endometrium during menstruation ( Lozano et al. 2021 ). Faced with this issue, Lozano et al. evaluated vaginal and endometrial microbiota in patients with and without CE (assessed by CD138 immunostaining), revealing an abundance of Ralstonia and Gardnerella in endometrial samples, and of Streptococcus and Ureaplasma in vaginal samples ( Lozano et al. 2021 ). Moreover, Han et al. identified four microbial vaginal markers of CE (Enterobacter, Prevotella, Faecalibacterium, and Phascolarctobacterium), allowing development of a predictive classifier for CE diagnosis with an AUC of 83.26% ( Han et al. 2024 ). When focusing on infertile women, Tanaka et al. suggested that vaginal microbiota of patients with CE exhibited a reduction in Bifidobacterium and of lactic-acid, not Lactobacillus -dependent, allowing identification of patients with potential benefit from antibiotic treatment ( Tanaka et al. 2022 ). Some limits of the present study deserve to be underlined. First, the retrospective nature cannot exclude all potential biases. Second, the small sample size limits drawing definitive conclusions on systematic hysteroscopy compared to endometrial biopsy alone to assess CE diagnosis. Third, a high variation in the delay to perform a second biopsy was observed due to the absence of a strict protocol. This could be a bias related to potential CE recurrence. Fourth, no attempt was made to evaluate the contribution of MUM-1 immunostaining to improve CE diagnosis. However, previous studies have demonstrated that the plasmacyte marker CD138 is currently the most reliable and time-saving diagnostic method for CE ( Kannar et al. 2012 , Kitaya et al. 2012 ). Finally, further studies are required to identify markers of CE response to antibiotics. In this specific setting, Di Pietro et al. reported that upregulation of miRNA-27a-3p and miRNA-124-3p in endometrium and serum from patients with CE could be new potential molecular markers ( Di Pietro et al. 2018 ). Moreover, Wang et al. suggested that exosomes derived from adipose tissue-derived stem cells could exert an anti-inflammatory effect on endometrial cells via the miR-21/TLR4/NF-kB signaling pathway ( Wang et al. 2023 ).

The concept of chronic endometritis (CE) was introduced at the beginning of the 20th century by Hitschmann and Adler, evaluating the impact of inflammation on morphological endometrial features throughout the menstrual cycle ( Frobenius 1990 ). CE is defined by a localized infectious/inflammatory disorder of the uterine mucosal lining ( Greenwood & Moran 1981 , Crum et al. 1983 ). The prevalence of CE is estimated at 3–72% according to series, raising the issue of the accuracy of diagnostic criteria ( Vasudeva et al. 1972 , Crum et al. 1983 , Han et al. 2024 ). Clinically, the most common symptoms of CE are metrorrhagia, followed by pelvic pain and leukorrhea ( Kitaya & Yasuo 2011 ). However, most patients are asymptomatic and diagnosed in the setting of infertility ( Bouet et al. 2016 , Park et al. 2016 ). Indeed, CE is thought to be involved in infertile women ( Cicinelli et al. 2005 a , b , Kasius et al. 2011 , Kitaya et al. 2012 ), recurrent miscarriage ( Johnston-MacAnanny et al. 2010 , Kitaya & Yasuo 2011 , Zolghadri et al. 2011 , Cicinelli et al. 2014 , McQueen et al. 2015 ), and repeated implantation failure ( Johnston-MacAnanny et al. 2010 , Bouet et al. 2016 , Kitaya et al. 2017 ) in 2.8%–56.8%, 9.3–67.6%, and 7.7–67.5%, respectively. CE diagnosis remains a major diagnostic challenge. Among the various tools to assess CE, hysteroscopy, allowing a direct vision of the uterine cavity, is often recommended, showing hyperemia, a ‘strawberry’ appearance suggestive of extensive hyperemic endometrium with localized central white spots scattered throughout the cavity, stromal edema, and micropolyps ( Tsonis et al. 2021 ). Among hysteroscopic features suggestive of CE, a debate exists on the most relevant ( Furui et al. 2024 ). Micropolyps (less than 1 mm), often associated with stromal edema, endometrial thickening, and periglandular hyperemia, seem the most suggestive of CE ( Cicinelli et al. 2005 a , b , Peng et al. 2022 ). However, a normal hysteroscopy cannot rule out CE diagnosis ( Vitagliano et al. 2021 ). Moreover, a high discrepancy rate in CE diagnosis was found between hysteroscopy and conventional histology using Hematoxylin and Eosin staining ( Guo et al. 2021 ). Finally, a high intra- and inter-observer variability in hysteroscopy CE diagnosis was reported, reinforcing the need for additional consensual diagnostic criteria ( Kitaya & Yasuo 2011 ). Endometrial sampling with conventional histology has been used to assess CE based on the presence of endometrial stroma infiltration by plasma cells, spindled cell alteration, lymphoid follicles, stromal breakdown, or disturbance in normal endometrial growth and maturation ( Vanderstraeten et al. 2015 , Cicinelli et al. 2017 ). Using Hematoxylin and Eosin (HE) staining, plasma cells are characterized by the presence of clockwise chromatin within an eccentric nucleus with a perinuclear halo ( Alonso et al. 2020 ). Immunohistochemistry (IHC) has been suggested to improve CE diagnosis by identifying plasma cells exhibiting expression of syndecan-1 (CD138, a marker of the plasma cell membrane), which is a heparan sulfate proteoglycan involved in inflammation ( Bayer-Garner & Korourian 2001 , Bayer-Garner et al. 2004 ). However, a debate exists on the most adequate measurement method and on threshold determination, especially in the context of infertile women ( Bayer-Garner et al. 2004 , Bouet et al. 2016 , Li et al. 2021 ). Subsequently, McQueen et al. proposed defining CE by the presence of one or more plasma cells per ten high-power fields (HPFs) ( McQueen et al. 2021 ). Therefore, using CD138 positivity as the gold standard of CE, the objectives of the present study were to correlate hysteroscopic and conventional histologic findings with CD138 values for the diagnosis of CE, to evaluate interobserver agreement, and the relation between CD138 expression and fertility outcomes.

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the work reported.

JI and YD had the original idea for the study and contributed to study design. JI, JI, JV, JB, and NM collected data. JI and YD performed statistical analysis. JI, YD, and ED drafted the paper. ED, CT, and BH supervised the study. All authors contributed to the interpretation of the data, revisions, and gave input at all stages of the study. All authors have read and agreed to the published version of the manuscript.