If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
The diagnosis of chronic endometritis in infertile asymptomatic women: a comparative study of histology, microbial cultures, hysteroscopy, and molecular microbiology
Igenomix Foundation-Instituto de Investigación Sanitaria INCLIVA, Valencia, SpainDepartment of Obstetrics and Gynecology, School of Medicine, Stanford University, Stanford, CA
Department of Scientific director, Igenomix SL, Valencia, SpainIgenomix Foundation-Instituto de Investigación Sanitaria INCLIVA, Valencia, SpainDepartment of Pediatrics, Obstetrics and Gynecology, University of Valencia, Valencia, SpainDepartment of Obstetrics and Gynecology, School of Medicine, Stanford University, Stanford, CA
Chronic endometritis is a persistent inflammation of the endometrial mucosa caused by bacterial pathogens such as Enterobacteriaceae, Enterococcus, Streptococcus, Staphylococcus, Mycoplasma, and Ureaplasma. Although chronic endometritis can be asymptomatic, it is found in up to 40% of infertile patients and is responsible for repeated implantation failure and recurrent miscarriage. Diagnosis of chronic endometritis is based on hysteroscopy of the uterine cavity, endometrial biopsy with plasma cells being identified histologically, while specific treatment is determined based on microbial culture. However, not all microorganisms implicated are easily or readily culturable needing a turnaround time of up to 1 week.
Objective
We sought to develop a molecular diagnostic tool for chronic endometritis based on real-time polymerase chain reaction equivalent to using the 3 classic methods together, overcoming the bias of using any of them alone.
Study Design
Endometrial samples from patients assessed for chronic endometritis (n = 113) using at least 1 or several conventional diagnostic methods namely histology, hysteroscopy, and/or microbial culture, were blindly evaluated by real-time polymerase chain reaction for the presence of 9 chronic endometritis pathogens: Chlamydia trachomatis, Enterococcus, Escherichia coli, Gardnerella vaginalis, Klebsiella pneumoniae, Mycoplasma hominis, Neisseria gonorrhoeae, Staphylococcus, and Streptococcus. The sensitivity and specificity of the molecular analysis vs the classic diagnostic techniques were compared in the 65 patients assessed by all 3 recognized classic methods.
Results
The molecular method showed concordant results with histological diagnosis in 30 samples (14 double positive and 16 double negative) with a matching accuracy of 46.15%. Concordance of molecular and hysteroscopic diagnosis was observed in 38 samples (37 double positive and 1 double negative), with an accuracy of 58.46%. When the molecular method was compared to microbial culture, concordance was present in 37 samples (22 double positive and 15 double negative), a matching rate of 56.92%. When cases of potential contamination and/or noncultivable bacteria were considered, the accuracy increased to 66.15%. Of these 65 patients, only 27 patients had consistent histological + hysteroscopic diagnosis, revealing 58.64% of nonconcordant results. Only 13 of 65 patients (20%) had consistent histology + hysteroscopy + microbial culture results. In these cases, the molecular microbiology matched in 10 cases showing a diagnostic accuracy of 76.92%. Interestingly, the molecular microbiology confirmed over half of the isolated pathogens and provided additional detection of nonculturable microorganisms. These results were confirmed by the microbiome assessed by next-generation sequencing. In the endometrial samples with concordant histology + hysteroscopy + microbial culture results, the molecular microbiology diagnosis demonstrates 75% sensitivity, 100% specificity, 100% positive and 25% negative predictive values, and 0% false-positive and 25% false-negative rates.
Conclusion
The molecular microbiology method describe herein is a fast and inexpensive diagnostic tool that allows for the identification of culturable and nonculturable endometrial pathogens associated with chronic endometritis. The results obtained were similar to all 3 classic diagnostic methods together with a degree of concordance of 76.92% providing an opportunity to improve the clinical management of infertile patients with a risk of experiencing this ghost endometrial pathology.
and has brought to light the value of detecting dysbiotic microbiomes to facilitate the improvement of clinical management. Chronic endometritis is a persistent inflammation of the endometrial mucosa caused by the presence of bacterial pathogens in the uterine cavity. The most common infectious agents responsible for chronic endometritis are Enterococcus faecalis, Enterobacteriaceae, Streptococcus species, Staphylococcus species, Gardnerella vaginalis, and Mycoplasma species as well as genital pathogens associated with sexually transmitted infections, such as Ureaplasma urealyticum, Chlamydia trachomatis, and Neisseria gonorrhoeae.
Chronic endometritis: correlation among hysteroscopic, histologic, and bacteriologic findings in a prospective trial with 2190 consecutive office hysteroscopies.
Chronic endometritis diagnosis still depends on the method used. The aim of this study is to develop a new molecular method for the diagnosis of chronic endometritis, overcoming the bias of the current methods.
Key findings
Molecular microbiology can be used to improve diagnosis and management of chronic endometritis in asymptomatic infertile patients.
What does this add to what is known?
Molecular microbiology can detect bacterial pathogens causing chronic endometritis and could be useful to guide a target therapy for this ghost endometrial condition.
Chronic endometritis is often clinically silent and rarely suspected and diagnosed, although it can be accompanied by symptoms like pelvic pain, dysfunctional uterine bleeding, dyspareunia, and leukorrhea.
Even if clinically silent, chronic endometritis has been suggested to diminish the success rates of both spontaneous and assisted reproductive technology conceptions as well as contributing to obstetric and neonatal complications.
Chronic endometritis due to common bacteria is prevalent in women with recurrent miscarriage as confirmed by improved pregnancy outcome after antibiotic treatment.
The diagnosis of chronic endometritis is difficult because there are no typical clinical or ultrasound findings. Classic diagnostic techniques of chronic endometritis rely on histology, which is based on the identification of plasma cells in the endometrial stroma,
but this method is nonspecific and dependent on the date of the menstrual cycle when sampling occurs. Considering these limitations, hysteroscopy and microbial culture are also often used for chronic endometritis diagnosis.
Hysteroscopic diagnosis of chronic endometritis relies on subjective characteristics identified by the reproductive endoscopist such as stromal edema, focal or diffuse epithelial hyperemia, and/or the presence of micropolyps. The identification of endometrial pathogens by microbial culture is the only method that provides objective information for targeted therapy. Its use has resulted in an improvement of reproductive outcome in women with recurrent miscarriage and repeat implantation failure;
Chronic endometritis due to common bacteria is prevalent in women with recurrent miscarriage as confirmed by improved pregnancy outcome after antibiotic treatment.
however, endometrial bacterial culture is not routinely performed because it has a long turnaround time, and not all microorganisms responsible for chronic endometritis are culturable.
To improve and personalize the state of the art for diagnosing and treating chronic endometritis, researchers must determine the identity and pathogenicity of the microbes prone to produce an endometrial infection. Molecular methods have revolutionized the detection and characterization of microorganisms in a broad range of medical fields including virology, mycology, parasitology, microbiology, and dentistry.
For instance, in public health, the screening of Mycobacterium tuberculosis by polymerase chain reaction (PCR) allows for early recognition and optimized treatment.
Along with conventional PCR techniques, real-time (RT)-PCR has an ever-increasing role in clinical diagnostics based on its capacity to detect difficult-to-culture bacteria and generate both qualitative and quantitative results in an accurate and rapid manner.
The aim of this study is to compare, in the same infertile patients, the diagnostic accuracy of the molecular microbiology tool with the traditional chronic endometritis diagnostic methods, ie, endometrial histology, hysteroscopy, and/or microbial culture, by assessing the presence of 9 specific chronic endometritis pathogens by RT-PCR and next-generation sequencing (NGS).
Materials and Methods
Study design
Endometrial samples from 113 patients subjected to chronic endometritis diagnosis using endometrial histology, hysteroscopy, and/or microbial culture were blindly evaluated for the presence of 9 chronic endometritis pathogens by RT-PCR using paired endometrial samples. Then, sensitivity and specificity of the molecular analysis and the classic diagnostic techniques were compared in 65 patients with chronic endometritis results assessed by all 3 recognized classic methods (Figure 1). In parallel, endometrial samples of negative controls based on histology and microbial culture (n = 10) were evaluated for the presence of chronic endometritis pathogens by molecular microbiology.
Figure 1Diagram of study and distribution of population investigated
Molecular diagnosis of chronic endometritis (CE) performed by real-time (RT) polymerase chain reaction (PCR). Classic diagnosis of CE includes histology, hysteroscopy, and microbial culture.
Moreno et al. Molecular microbiology for diagnosing chronic endometritis. Am J Obstet Gynecol 2018.
Participants involved in this study were 21- to 53-year-old infertile patients recruited (E.C.) at the Second Unit of Obstetrics and Gynecology, Department of Biomedical and Human Oncological Science, University of Bari, Bari, Italy, undergoing in vitro fertilization treatment. Patients were diagnosed and treated for chronic endometritis using either a single or various classic diagnostic techniques. Frozen endometrial biopsy specimens were sent blindly to Igenomix SL for molecular microbiology diagnosis by RT-PCR. Also, endometrial tissue from women with negative results for chronic endometritis was analyzed using RT-PCR. This group of negative controls consisted of women undergoing surgery for benign ovarian conditions (oophorectomy for dermoid cyst with exclusion of endometriosis) or operated on due to myomas (myomectomy) (n = 6), and women treated with antibiotics for previous chronic endometritis with no current signs of the disease at the time of sample collection by either histology and microbial culture (n = 4). The ethical committee of the Second Unit of Obstetrics and Gynecology, Department of Biomedical and Human Oncological Science, University of Bari, Bari, Italy, approved the study (register number 4880). All women signed an informed consent form agreeing to the comparative analysis of the techniques performed for the routine diagnosis of chronic endometritis vs the molecular diagnosis.
Histological diagnosis
Patients underwent endometrial sampling using a 3-mm Novak curette connected to a 20-mL syringe. To minimize the risk that endometrial cultures might be contaminated by the vaginal flora, after placing a vaginal speculum, the Novak cannula was inserted under visual control into the uterine cavity avoiding any contact with the vaginal walls. Histological examination was performed blindly by 2 pathologists who were unaware of the results of the other diagnostic techniques. Paraffin-embedded endometrial specimens were used for the study. Samples were fixed overnight in 4% formaldehyde (in phosphate buffer, pH 7.3, Nacalai Tesque, Kyoto, Japan) and embedded in paraffin (Nacalai Tesque). Microsections of 5 μm were stained with hematoxylin-eosin. The inflammatory status of the endometrium was graded based on the presence of stromal infiltrate dominated by lymphocytes and plasma cells, and a spindle cell change of stromal cells.
Infiltration showing aggressive behavior against glands, inflammatory cell-gathering inside glands, or a more structured infiltration was also relevant. Micropolypoid vegetations formed in the endometrium are macroscopic evidence of an inflammation-related process and can be used as markers for this pathology. All cases diagnosed with chronic endometritis by histological criteria were confirmed by antisyndecan-1 (CD138) immunohistochemistry as previously described.
All hysteroscopies were performed by 2 physicians using a lens-based 2.7-mm outer diameter mini-telescope with a 105-degree angle of visual field equipped with a 4.5-mm outer diameter double-flow operative sheath (Karl Storz, Tuttlingen, Germany). Hysteroscopies were performed in the follicular phase (cycle day 7-12). Saline was employed to distend the uterine cavity at a pressure generated by a simple drip from a bag suspended 1 m above the patient. A 300-W light source with a xenon bulb and a high-definition digital camera (Karl Storz) were used. During hysteroscopy, both the anterior and posterior uterine walls were thoroughly examined by passing the hysteroscope parallel to the endometrial surface to identify any surface irregularity.
The following criteria were used for the hysteroscopic diagnosis of chronic endometritis: the presence of stromal edema, focal or diffuse periglandular hyperemia, and micropolyps of <1 mm in size.
The hysteroscopic images were reviewed in the patient database at the Second Unit of Obstetrics and Gynecology, Department of Biomedical and Human Oncological Science, University of Bari, Bari, Italy.
Microbial culture
Endometrial samples were processed according to the current standard for microbiological culture using a separate test to detect the presence of the most prevalent chronic endometritis pathogens. In particular, to detect C. trachomatis, N. gonorrhoeae, U. urealyticum, U. parvum, and Mycoplasma hominis, all of which are noncultivable strains recoverable from the genital tract, a multiplex RT method (AnyplexTM II STI-7 Detection [V1.1]; Seegene, Seoul, Republic of Korea) was used according to the manufacturer’s instructions. To detect the presence of culturable microorganisms (E. faecalis, Escherichia coli, Staphylococcus epidermidis, Staphylococcus aureus, Streptococcus agalactiae, Streptococcus mitis, and yeasts), the endometrial samples were inoculated onto culture media either directly or following enrichment in brain heart infusion medium. These media consisted of columbia-colistin-nalidix acid agar with 5% sheep blood, which is a selective and differential medium commonly used in the isolation of gram-positive organisms from mixed clinical specimens, and MacConkey agar and mannitol salt agar for the isolation of gram-negative bacteria and Staphylococcus aureus. Consequently, the growth of the bacteria indicated above was considered positive when >10,000 colony-forming units were isolated from the culture. In the case of a positive culture, antibiotic sensitivity and resistance were tested by Vitek 2 (BioMérieux). According to microbiology guidelines, the presence of Staphylococcus epidermidis was considered to indicate contamination.
Molecular diagnosis
Frozen endometrial specimens were sent to Igenomix SL for the blind identification of chronic endometritis pathogens by RT-PCR using specific primers for the 9 most common bacteria responsible for causing chronic endometritis (C. trachomatis, Enterococcus species, E. coli, G. vaginalis, Klebsiella pneumoniae, M. hominis, N. gonorrhoeae, Staphylococcus species, and Streptococcus species). A literature search was conducted to identify species- and genus-specific primers for the bacteria of interest, plus 1 primer for the family Enterobacteriaceae including E. coli and K. pneumoniae (Table 1). The primers used were reevaluated and subjected to Basic Local Alignment Search Tool searches using the National Center for Biotechnology Information genomic database. All primers were purchased from Integrated DNA Technologies (Skokie, IL).
Table 1List of primers used for polymerase chain reaction amplification of selected microorganisms
Total DNA was isolated using a QIAamp cador pathogen mini kit (Qiagen, Hilden, Germany) from endometrial biopsies previously treated with enzymatic digestion for difficult-to-lyse bacteria. For this digestion, 25 mg of tissue were cut into small pieces and treated with proteinase K at 56°C for 3 hours under agitation or until the tissue was completely lysed. Then, the sample was mixed with ATL buffer (Qiagen) and disrupted mechanically in a TissueLyser LT (Qiagen) for 5 minutes at 50 Hz, using stainless-steel beads. After these pretreatments, bacterial nucleic acids were purified following the manufacturer’s instructions.
Bacterial DNA template
A panel of DNA templates of bacteria causing chronic endometritis was selected to evaluate the specificity of the RT-PCR assays. A total of 11 DNA templates were obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ, Braunschweig, Germany), including templates of common microorganisms of the disease, Enterococcus faecalis (DSMZ 2570), Escherichia coli (DSMZ 1116), G vaginalis (DSMZ 4944), K pneumoniae (DSMZ 30104), Mycoplasma hominis (DSMZ 25592), Staphylococcus epidermidis (DSMZ 1798), Staphylococcus haemolyticus (DSMZ 20263), Streptococcus agalactiae (DSMZ 2134), and Streptococcus gallolyticus (DSMZ 20065), as well as templates of sexually transmitted disease pathogens, Chlamydia trachomatis (DSMZ 19411) and N gonorrhoeae (DSMZ 15130).
RT-PCR assays
All RT-PCR reactions were performed on a LightCycler 480 II (Roche Diagnostics, Almere, The Netherlands). Reaction mixtures contained 200 ng of DNA isolated from endometrial biopsies or commercial purified bacterial DNA as the template, 2.5 mmol/L MgCl2, 0.25 μmol/L forward and reverse primers, and 1 μL of the LightCycler FastStar DNA Master SYBR Green I 10x (Roche Applied Science, Mannheim, Germany) resulting in a final reaction volume of 20 μL. The cycling program was as follows: an initial denaturation at 95°C for 10 minutes; amplification for 45 cycles of 10 seconds at 95°C, 10 seconds at 57°C, and 50 seconds at 72°C; melting curve analysis for 5 seconds at 95°C, 15 seconds at 65°C and a temperature continuous acquisition up to 95°C (ramp rate of 0.11°C/s); and cooling for 30 seconds at 40°C.
Each experiment included negative and positive controls to determine any possible contamination and unspecific amplification. Positive controls included a mix of all the bacterial DNA templates at a concentration of 10,000 genomes each. Negative controls included all DNA templates except the microorganism to be evaluated for each assay. The genome size of each bacterium was used for the calculation of genomes/ng of DNA.
A melting curve analysis was made after amplification to distinguish the targeted PCR products from nonspecific PCR products. The concentration of microorganism in each sample was calculated comparing the crossing point-PCR-cycle (Cp) values obtained from the sample with the Cp values of the positive control.
16S ribosomal RNA sequencing
Endometrial microbiome profiles were obtained by NGS of the bacterial gene encoding for the 16S ribosomal RNA (rRNA) using the Ion 16S metagenomics kit (ThermoFisher Scientific, Waltham, MA), which includes 2 primer sets (V2-4-8 and V3-6, 7-9) that selectively amplify the hypervariable regions (V2-4-8 and V3-6, 7-9) of the 16S ribosomal subunit. The amplified fragments were sequenced on the Ion S5 XL system (ThermoFisher Scientific). The results were analyzed using the Metagenomics application of Ion Reporter software 5.0 (ThermoFisher Scientific) using default parameters as previously described.
Briefly, the QIIME package (http://qiime.org/) and Greengenes database, Version 13_8 (http://qiime.org/home_static/dataFiles.html), were used for taxonomic assignment. QIIME was used to calculate alpha diversity and rarefaction curves before filtering. Shannon methods were employed to analyze the biodiversity within a group of samples. Positive controls of E. coli DNA and negative controls were included to detect any contamination from reagents.
Statistical analysis
The comparative assessment of the molecular test vs the classic diagnostics methods considered (individually or combined) to be the gold standard were calculated as follows
Sensitivity: percentage of true-positive cases among the total positives in the gold standard test.
•
Specificity: percentage of true-negative cases among the total negatives in the gold standard test.
•
Accuracy: percentage of correct assessments on the total assessments.
•
Positive predictive value: percentage of true-positive cases among the total positives in the molecular test.
•
Negative predictive value: percentage of true-negative cases among the total negatives in the molecular test.
•
False-positive rate: percentage of false-positive cases among the total negatives in the gold standard test.
•
False-negative rate: percentage of negative cases among the total positives in the gold standard test.
Results
Analytical specificity of RT-PCR assays
The analytical specificity of the molecular microbiology assay was verified using specific primers for the most common bacteria causing chronic endometritis using 30 ng of commercial bacterial DNA isolated from chronic endometritis bacteria as a template. All primer sets resulted in specific amplification of their own bacterial DNA templates, with Cp values ranging from 12-20 compared to the negative control of ultrapure water with the lowest Cp value at 35 for N. gonorrhoeae, while the majority of them did not amplify after 40 amplification cycles. When each primer pair was tested against the other bacterial DNAs, insignificant cross-reactivity was detected between them, with a minimum difference of 10 amplification cycles between the specific and nonspecific amplifications (Figure 2, A). Then, the minimum detection limit of each microorganism in the RT-PCR was assessed by amplifying increasing amounts of each template DNA (0-1,000,000 genomes) alone or in a complex mixture of bacterial DNA consisting of 1,000,000 genomes from the other microorganisms included in the molecular method. The detection limits of these RT-PCR reactions showed the high sensitivity of the molecular method for the different bacteria tested ranging from 10-1000 genomes for all microorganisms, except for staphylococci, which showed a minimum detection limit of 10,000 genomes, equivalent to the microbial culture technique in which >10,000 colony-forming unit is considered a positive test (Figure 2, B).
Figure 2Analytical specificity of molecular microbiology
A, Represents amplification cycles (Cp) of commercial bacterial DNA (30 ng) of most common bacteria causing chronic endometritis using specific primers for these bacteria. B, Detection limit of each bacterial DNA in molecular method. Each set of primers was tested against increasing amounts of bacterial DNA (0; 1; 10; 100; 1000; 10,000; 100,000, and 1,000,000 genomes) alone or in complex mix of microorganisms containing 1,000,000 genomes of each remaining bacterial species included in molecular method. C, Graphs obtained from real-time polymerase chain reaction show melting peak profile and melting temperature (Tm) for each amplicon, discriminating specific (positive) and unspecific (negative) amplification signals.
To minimize possible bias introduced by the cross-reactions in subsequent experiments, the positive control was defined as a mix of all the bacterial DNA templates at a concentration of 10,000 genomes each, while the negative control was a mix of all DNA templates except for the bacterial DNA to be detected. Then, for each independent assay, the melting peak profile and melting temperature were determined, defining a specific melting temperature value for each amplicon, thus discriminating between specific (positive) and unspecific (negative) amplification signals (Figure 2, C).
Molecular diagnosis of chronic endometritis using RT-PCR
DNA was isolated from 113 endometrial specimens of women being assessed for chronic endometritis using any of the classic methods. From these, 18 samples did not yield enough standard-quality DNA for analysis. The remaining 95 (set 1) were assessed by RT-PCR with the primers indicated in Table 1 for the microorganisms Enterobacteriaceae, Enterococcus species, E. coli, G. vaginalis, K. pneumoniae, M. hominis, Staphylococcus species, and Streptococcus species. Additionally, other pathogens causing sexually transmitted diseases, such as C. trachomatis and N. gonorrhoeae, were included in the analysis. The amplified signal for each microorganism was considered positive when the threshold cross-point value (Cp) obtained was equal or lower than the Cp value of the positive control and the melting temperature matched with that established in the positive control.
Using this molecular analysis, 42 of the 95 endometrial samples were negative for chronic endometritis (44.21%), while 53 were positive (55.79%) for ≥1 of the pathogens tested (58.5% and 41.5%, respectively) (Figure 3, A). The pathogens most commonly represented in the samples analyzed by RT-PCR were streptococci, while C. trachomatis and N. gonorrhoeae were undetectable in all tested samples (Figure 3, B).
Figure 3Molecular diagnosis of chronic endometritis (CE)
A, Proportion of positive and negative cases of CE using molecular method and percentage of positive cases with detection of ≥1 microorganisms in samples from set 1 (95 samples analyzed by real-time polymerase chain reaction), set 2 (65 samples used for comparison with classic diagnosis), and set 3 (13 samples with concordant results by 3 classic methods). Number of cases in which each targeted pathogen is detected alone or in combination with other bacteria in B, set 1; C, set 2; and D, set 3.
Moreno et al. Molecular microbiology for diagnosing chronic endometritis. Am J Obstet Gynecol 2018.
Comparison of chronic endometritis diagnosis based on molecular microbiology vs classic methods
The results of the molecular microbiology of chronic endometritis were compared with the classic diagnostic methods: histology, hysteroscopy, and microbial culture (Figure 1). For this comparison, 30 patients were excluded for lacking results from ≥1 of the 3 classic methods, so the comparison was finally performed for 65 patients (set 2) with a chronic endometritis diagnosis derived from classic techniques (Table 2).
Table 2Results of chronic endometritis diagnosis by molecular and classic methods
Using molecular analysis, 27 of the 65 endometrial samples were negative for chronic endometritis (41.54%) while 38 were positive (58.46%) for ≥1 of the pathogens tested (32.31% and 26.15%, respectively) (Figure 3, A). The most commonly represented bacteria in the endometria analyzed by RT-PCR were streptococci, which accounted for 45.78% of the total pathogens detected (Figure 3, C).
Molecular microbiology vs histology
Histological diagnosis alone was positive in 25 of 65 patients investigated (38.46%) while the molecular microbiology was positive in 38 cases (58.46%). Concordant results were observed in 30 samples (14 double positives and 16 double negatives) giving a matching accuracy of 46.15%. Interestingly, in 24 (68.57%) of 35 samples with contradictory results, bacterial DNA was clearly identified by RT-PCR while the histological analysis was negative (Table 2).
Molecular microbiology vs hysteroscopy
Hysteroscopic-based diagnosis was positive in 63 of 65 patients investigated (96.92%), while the molecular microbiology was positive in 38 cases. Concordance was observed in 38 samples (37 double positives and 1 double negative) showing a matching accuracy of 58.46%. In 26 (96.29%) of 27 discordant patients, bacterial DNA was not identified in the paired endometrial sample obtained after positive hysteroscopic diagnosis of chronic endometritis. Interestingly, only 1 sample with no signs of chronic endometritis in the hysteroscopy was positive for Streptococcus species (Table 2).
Molecular microbiology vs microbial culture
Microbiological culture was positive in 34 of 65 patients tested for chronic endometritis by either histology or hysteroscopy (52.30%) while molecular diagnosis was positive in 38 of these cases (58.46%). Concordance between molecular microbiology and microbial culture was present in 37 patients (22 double positive and 15 double negative) showing a matching rate of 56.92%. However, when microbiological results were analyzed in detail, we found several cases in which the culture was either contaminated by S. epidermidis (case 28), or presented microorganisms that were not tested in the molecular diagnosis: Ureaplasma species and Candida albicans (specifically cases 20 and 48). Also, the molecular microbiology allowed for the identification of G. vaginalis, which is seldom cultured, causes chronic endometritis, and was not identified by classic microbial culture in some samples (cases 2 and 5). Taking these cases into account, the comparison of these 2 methods demonstrated an accuracy of 66.15% (Table 2). From these 22 double-positive cases, 11 RT-PCR cases were confirmed by isolation of the same pathogens. Moreover, in 4 cases, G. vaginalis DNA was identified, providing additional information over the classic microbial culture.
Molecular microbiology vs histology + hysteroscopy
Because histology and hysteroscopy are subjective methods with the highest discordant results, we compared the molecular diagnostic method to both. From the 65 patients analyzed by all methods (histology, hysteroscopy, and microbial culture), only 27 presented with consistent histology + hysteroscopy results (41.54% concordance rate), with 25 double-positive and 2 double-negative results (Table 2). In this group, 13 cases were also consistent with microbial culture, while 14 differed from microbial culture (Figure 1). From those 27 patients, RT-PCR showed concordant results in 15 cases (14 positives and only 1 negative) with an accuracy of 55.55%. Interestingly, 38 of 65 patients (58.46%) presented with opposite results between histology and hysteroscopy, and in all negative cases for histology were positive based on hysteroscopy. In these cases, the detection of bacterial DNA by RT-PCR (23 of 38 cases) was always coincident with positive hysteroscopy, while negative diagnosis of chronic endometritis using the molecular method (15 of 38 samples) always matched with the negative histological diagnosis.
Molecular microbiology vs histology + hysteroscopy + microbial culture
From the 65 samples with chronic endometritis results from all 3 classic methods, only 13 (20%) had concordant results from all 3 (Figure 1). The molecular analysis of this set of samples showed that 4 of the 13 endometrial samples were negative for chronic endometritis (30.8%), while 9 were positive (69.2%) for ≥1 of the pathogens tested (66.7% and 33.3%, respectively) (Figure 3, A). The most detected pathogen in these samples was Streptococcus species, followed by Enterococcus species and G. vaginalis (Figure 3, D).
The only patient diagnosed as negative based on the 3 classic methods was also negative for the molecular evaluation of chronic endometritis (Figure 4, A). From the remaining 12 that were positive for chronic endometritis based on all 3 classic methods, 9 were also positive based on RT-PCR, while 3 showed discordant results. In total, 10 of 13 cases presented with similar results with an accuracy of 76.92% (Table 2). In 5 of the 9 cases with positive results for microbial culture and RT-PCR, the microorganisms detected were the same, and in 2 of them, G. vaginalis was also detected by RT-PCR together with other pathogens (cases 8 and 15), providing additional information to the microbial culture (Figure 4, A). To confirm these results, endometrial biopsies from these 13 diagnostic concordant patients were subjected to bacterial 16S rRNA sequencing to assess their full endometrial microbiome. The results of NGS for the bacterial 16S rRNA gene confirmed the detection of bacterial DNA in 12 of 13 endometrial samples, since 1 sample did not yield sequencing results (case 24). The number of mapped reads per sample was between 27,178 and maximum of 231,538, with an average value of 117,369. The data of mapped sequences and Shannon index for each sample are detailed in Table 3. The overall analysis of the endometrial microbiome in those 12 samples showed that the most represented genus was Lactobacillus, followed by 2 bacterial pathogens associated with chronic endometritis: Streptococcus and Gardnerella. Also, in the 20 most abundant genera present in this set of samples, other bacteria previously reported to colonize the reproductive tract such as Bifidobacterium, Megasphaera, Parvimonas, Prevotella, Propionibacterium, and Veillonella were found (Figure 5, A).
The microbiome results using NGS were concordant with RT-PCR in 11 of 12 cases and coincided with the microbial culture in 9 of 12 cases, showing an accuracy of 91.67% and 75%, respectively (Table 3). Interestingly, in 2 cases in which the RT-PCR was negative and culture method was positive (cases 17 and 26), the taxonomic assignment obtained in the microbiota confirmed the results of the RT-PCR, with no detection of bacterial DNA of the pathogens isolated in the microbial culture (Figure 5, B). To better clarify this, the percentage of lactobacilli in these samples was also considered, based on previous reports demonstrating that a percentage of Lactobacillus species in the uterine cavity are associated with a healthy endometrial status.
This analysis showed a low percentage of lactobacilli in all those samples with a positive diagnosis of chronic endometritis by all the methods analyzed, with the exception of 1 sample with 93.2% of Lactobacillus and a very low percentage of the pathogens detected by either microbial culture or RT-PCR (case 30). Moreover, in the 2 cases (cases 8 and 15) in which we detected G. vaginalis by RT-PCR, the sequencing results confirmed its presence in the samples (Figure 5).
Figure 4Histological and hysteroscopic diagnosis of chronic endometritis
A, Concordant chronic endometritis results in patients/samples analyzed by 4 methods compared in this study (3 classic methods and real-time [RT] polymerase chain reaction method [PCR]). B, Discordant chronic endometritis results in patients/samples analyzed by 4 methods compared in this study. Black arrows show CD138+ cells.
Moreno et al. Molecular microbiology for diagnosing chronic endometritis. Am J Obstet Gynecol 2018.
A, Microbiome composition profiles showing 20 most-abundant genera and their relative abundances in endometrial samples of patients with consistent chronic endometritis (CE) diagnosis by 3 classic methods. B, Relative abundance of most frequent bacteria causing CE: Ureaplasma, Neisseria, Chlamydia, Staphylococcus, Streptococcus, Gardnerella, Enterococcus, Klebsiella, Escherichia, and Mycoplasma in endometrial samples of patients with consistent CE diagnosis by 3 classic methods.
Moreno et al. Molecular microbiology for diagnosing chronic endometritis. Am J Obstet Gynecol 2018.
By contrast, a high percentage of lactobacilli was detected in the samples that proved negative for the molecular diagnosis of chronic endometritis (cases 10, 17, 26, and 35), supporting the results previously obtained from the RT-PCR analysis (Figure 5, A).
Finally, sensitivity, specificity, positive and negative predictive values, and false-positive and -negative rates were assessed for each individual classic method and their combination in comparison to the RT-PCR diagnosis (Table 4), with the best results obtained when both RT-PCR and NGS diagnosis of chronic endometritis was compared to the 3 concordant classic methods.
Table 4Molecular diagnosis compared to classic diagnostic methods of chronic endometritis
Sensitivity (%)
Specificity (%)
Accuracy (%)
PPV (%)
NPV (%)
FPR (%)
FNR (%)
Histology (n = 65)
56.00
40.00
46.15
36.84
59.26
60.00
44.00
Hysteroscopy (n = 65)
58.73
50.00
58.46
97.37
3.70
50.00
41.27
Microbial culture (n = 65)
71.43
56.67
66.15
65.79
62.96
43.33
28.57
Histology + hysteroscopy (concordant results, n = 27)
Molecular microbiology in patients without chronic endometritis
From the 65 samples analyzed for the 3 classic methods, only 1 presented concordant negative results for all the techniques. Then, to evaluate the efficacy of the RT-PCR method to detect negative cases of chronic endometritis, endometrial samples from 10 control subjects, that proved negative for histology and microbial culture, were subjected to the molecular method. The results of these samples showed negative results for the 9 pathogenic bacteria tested (Table 5) consistent with their negative results for the classic methods analyzed. However, RT-PCR was able to detect very small amounts of DNA from G. vaginalis (case C2) and M. hominis (case C5), but their levels were below the threshold of positive controls. These data support the efficacy of molecular microbiology to discriminate between positive and negative cases of chronic endometritis, based on the detection of 9 pathogens usually causing the disease.
Table 5Molecular diagnosis of endometrial pathogens in negative cases
Subject
Surgery indication/treatment
RT-PCR
Histology
Microbial culture
C1
Myomectomy; usual technique with Novak
Negative
Negative
Negative
C2
Oophorectomy (dermoid cyst); usual technique with Novak
Negative (low detection of Gardnerella vaginalis)
Negative
Negative
C3
Myomectomy; usual technique with Novak
Negative
Negative
Negative
C4
Myomectomy; usual technique with Novak
Negative
Negative
Negative
C5
Myomectomy; usual technique with Novak
Negative (low detection of Mycoplasma hominis)
Negative
Negative
C6
Myomectomy; usual technique with Novak
Negative
Negative
Negative
T1
Previous chronic endometritis, treated with antibiotics
Negative
Negative
Negative
T2
Previous chronic endometritis, treated with antibiotics
Negative
Negative
Negative
T3
Previous chronic endometritis, treated with antibiotics
Negative
Negative
Negative
T4
Previous chronic endometritis, treated with antibiotics
Negative
Negative
Negative
PCR, polymerase chain reaction; RT, real-time.
Moreno et al. Molecular microbiology for diagnosing chronic endometritis. Am J Obstet Gynecol 2018.
Surprisingly, diagnosis of chronic endometritis still depends upon the method used. The current diagnostic gold standard for chronic endometritis is histological examination although is subjective based on different criteria depending on the authors, while hysteroscopy and endometrial culture present controversial reliability in the diagnosis of such condition. Whether histological examination of endometrial tissue, hysteroscopic observation of the uterine cavity, or microbial culture are used has a significant impact on diagnostic accuracy and the different techniques applied to the same patient/sample may yield contradictory results and misleading conclusions (Figure 4, B). The concordant results among the 3 classic techniques in the diagnosis of chronic endometritis in our study was only 20%. Histology and hysteroscopy are highly subjective, unspecific, and rely on the individual observations of the pathologist or endoscopic surgeon.
Also, they cannot discriminate the causal agent of chronic endometritis, which in turn leads to broad-range therapy and likely results in recurrent chronic endometritis. Unlike them, microbial culture is able to identify the culturable microorganisms present in the endometrial tissue. However, some chronic endometritis-causing bacteria such as Ureaplasma species, M. hominis, and G. vaginalis are not culturable in standard laboratory conditions, which leads to false-negative results and/or contamination bias. For the aforementioned reasons, a reliable method for the diagnosis of chronic endometritis based on molecular detection and quantification of DNA from pathogens present in the endometrium is needed to improve clinical management.
In modern medicine, nucleic acid amplification tests that are highly sensitive and specific, and in some cases minimally invasive, are recommended to screen for infectious diseases in men and women.
Neither histology nor endoscopy are being used as part of routine diagnosis for infectious diseases in organs and/or tissues except for the uterine cavity. Chronic endometritis is underdiagnosed and thus rarely considered, although clinically relevant in reproductive medicine because it has been associated with repeated implantation failure and recurrent pregnancy loss.
Chronic endometritis due to common bacteria is prevalent in women with recurrent miscarriage as confirmed by improved pregnancy outcome after antibiotic treatment.
The high prevalence of common bacteria is not surprising considering that 60% of women with pelvic inflammatory disease have nongonococcal/nonchlamydial infection.
Effectiveness of inpatient and outpatient treatment strategies for women with pelvic inflammatory disease: results from the pelvic inflammatory disease evaluation and clinical health (PEACH) randomized trial.
Our panel for molecular microbiology was selected to cover >80% of clinical cases of chronic endometritis. The bacteria primarily responsible for the disease are streptococci, E. faecalis, E. coli, and staphylococci, accounting for 38.5%, 19.3%, 16.3%, and 5.9% of cases, respectively.
Recently, dysbiosis of the endometrial cavity has been associated with poor reproductive outcomes in assisted reproductive treatment patients, suggesting that pathogenic deviations of the Lactobacillus endometrial content (a signature of chronic endometritis) could play a role in infertility.
Our study demonstrates the usefulness of a simple RT-PCR test for the molecular diagnosis of chronic endometritis in endometrial samples using a comprehensive panel of primers to detect the most common microorganisms involved.
Our results show that RT-PCR is a robust and specific technique able to discriminate targeted bacterial DNA even in challenging conditions such as frozen endometrial biopsies. The molecular analysis had a degree of agreement of 76.92% when endometrial samples showed concordant results by all 3 classic methods. In this subset of 13 samples, molecular microbiology had 75% sensitivity and 100% specificity compared to concordant histology + hysteroscopy + microbial culture diagnosis (Table 4). Among them, Streptococcus species were the most abundant bacteria detected (47%), followed by Enterococcus species (15%), E. coli (12%), K. pneumoniae (5%), Staphylococcus species (3%), and M. hominis (2%); these findings are consistent with previously reported microbial culture data.
Interestingly, G. vaginalis, a reproductive tract pathogen usually neglected in chronic endometritis diagnosis, was detected in 7% of the samples analyzed using the molecular method. In contrast, C. trachomatis and N. gonorrhoeae were undetectable in all tested samples, which is in agreement with other works describing a limited role for C. trachomatis and sexually transmitted infection pathogens in the origin of chronic endometritis.
Single use of classic chronic endometritis diagnostic techniques (histology, hysteroscopy, or microbial culture) showed poor diagnostic accuracies (46.15%, 58.46%, and 66.15%, respectively) (Table 4). Moreover, when the overall results were analyzed in detail, we observed that histology and hysteroscopy, the 2 most subjective methods for chronic endometritis diagnosis, yielded contradictory results in 58.46% of cases (Figure 1 and Table 2). In all the inconsistent cases, histology showed negative results while hysteroscopy showed positive results for the same patient, confirming that, compared to other methods, histopathological evaluation usually underdiagnoses chronic endometritis, and hysteroscopy overdiagnoses. The main limitations of histology result from: (1) dependence on the piece of endometrial sample analyzed, (2) variability of staining, (3) observer experience, (4) phase of the menstrual cycle in which the sample was collected,
and (5) unknown clinical relevance of a small amount of plasma cells (below the positive threshold) in the endometrium. In this regard, RT-PCR may be useful for the detection of intrauterine germs when histology is negative, as well as for the decision of a target therapy when histology is positive. Hysteroscopic diagnosis is based on the presence of hyperemia, edema, and/or micropolyps, all of which are endometrial symptoms that could be due to other pathophysiological conditions or a noninfectious inflammation of the uterine cavity, as observed in case 55 that was negative for all the methods used except for hysteroscopy in which the only sign of chronic endometritis was the presence of micropolyps (Figure 4, B). Microbial culture, the most reliable of the 3 classic methods, also presents some limitations mainly represented by contamination of the microbial culture with skin or environmental bacteria (ie, S. epidermidis) and the inability to grow and isolate nonculturable bacteria. The limitations of chronic endometritis diagnosis using individual classic techniques and their misleading results are evident in this study in which only 13 of 65 (20%) of the samples/patients analyzed presented concordant results using all 3 diagnostic methods (Figure 1 and Table 2).
Strengths and limitations
These results indicate that molecular microbiology provides similar results to using the 3 classic methods together, overcoming the bias of using any of the classic methods alone. Other advantages of the molecular method are: (1) RT-PCR is highly sensitive and can identify and quantify very small amounts of bacterial DNA, regardless of their culturable or nonculturable nature, and may be useful in estimating the severity of chronic endometritis; (2) it quantifies bacterial DNA instead of live bacteria, so it is able to detect chronic endometritis pathogens in frozen or fixed samples, facilitating sample collection and storage while avoiding contamination during sample transportation/manipulation;
While the results of this study show that the molecular detection of bacterial DNA in endometrial samples is a reliable method for the diagnosis of chronic endometritis, with a degree of agreement of 76.92%, several limitations must be addressed to improve the validity of this molecular test. The main limitation of molecular microbiology is the relatively low negative predictive value, estimated at 25%, compared to concordant histology+ hysteroscopy + microbiology. This could be biased by the vast amount of positive chronic endometritis cases determined by hysteroscopy, as the negative predictive value of the molecular method compared to histology or microbial culture alone was 59% and 63%, respectively. Also, molecular methods do not guarantee that DNA comes from viable bacteria, and the establishment of the minimum amount of bacterial DNA that causes the disease should be determined, as the presence of such DNA in some women could be innocuous depending on the host response to those pathogens and would not discriminate between acute and chronic endometritis. The copy numbers for the interrogated gene could be variable between different bacterial genera/species and could cause a slight deviation on the number of bacteria estimated by either RT-PCR or 16S rRNA NGS. Also, the topological localization of bacterial growth should be considered; for example, G. vaginalis forms biofilms at the apical surface of the epithelial layer.
This could explain the differential detection of G. vaginalis by the 2 molecular methods obtained in patients 35 and 44, which may depend on the presence of bacterial biofilm on the specific sample of tissue analyzed (Table 3). To overcome these limitations, future work could address the technical and clinical improvement of this molecular tool by increasing the number of microorganisms included in the panel, as well as the transition to molecular microbiology diagnosis of chronic endometritis in endometrial fluid samples for minimally invasive detection of this disease.
Conclusion
In conclusion, this study demonstrates that compared to the commonly used hysteroscopy, histology, or microbial culture diagnostic methods, RT-PCR effectively detects and quantifies bacterial DNA from chronic endometritis-causing pathogens in endometrial samples providing a feasible, faster, and cheaper method for the diagnosis of chronic endometritis.
Furthermore, the microbiome results using NGS were concordant with RT-PCR in 91.67% of cases and coincide with the microbial culture in 75% of samples because it allows for the detection of both culturable and nonculturable bacteria. The high sensitivity of molecular microbiology allows for the detection of endometrial colonization in patients without histological signs of chronic endometritis, providing additional information to improve the current detection of this invisible endometrial pathology in asymptomatic infertile patients.
Acknowledgment
We thank Sheila M. Cherry, PhD, ELS, President and Senior Editor from Fresh Eyes Editing LLC, for her excellent work editing this manuscript.
References
Human Microbiome Project Consortium
Structure, function and diversity of the healthy human microbiome.
Chronic endometritis: correlation among hysteroscopic, histologic, and bacteriologic findings in a prospective trial with 2190 consecutive office hysteroscopies.
Chronic endometritis due to common bacteria is prevalent in women with recurrent miscarriage as confirmed by improved pregnancy outcome after antibiotic treatment.
Effectiveness of inpatient and outpatient treatment strategies for women with pelvic inflammatory disease: results from the pelvic inflammatory disease evaluation and clinical health (PEACH) randomized trial.
This work was partially supported by Igenomix SL and the grant Instituto Valenciano de Competitividad Empresarial (IMIDTA/2017/70) from Generalitat Valenciana. Drs Moreno (Ayudas para contratos Torres Quevedo-13-06133) and Bau (Ayudas para contratos Torres Quevedo-16-08454) are supported by Torres-Quevedo grants from the Spanish Ministry of Economy and Competitiveness. Ms Garcia-Grau is supported by a Formación de Profesorado Universitario grant (FPU15/01923) from the Spanish Ministry of Education, Culture, and Sport. Dr Vilella is supported by Miguel Servet program type I (CP13/00038) and Fondo de Investigación Sanitaria project (PI14/00545) of Spanish Instituto de Salud Carlos III.
Disclosure: Drs Moreno, Bau, Valbuena, and Simon, and Ms Gonzalez are employed by Igenomix SL. The remaining authors report no conflict of interest.
Cite this article as: Moreno I, Cicinelli E, Garcia-Grau I, et al. The diagnosis of chronic endometritis in infertile asymptomatic women: a comparative study of histology, microbial cultures, hysteroscopy, and molecular microbiology. Am J Obstet Gynecol 2018;218:602.e1-16.
30156-X&id=gr1.jpg){kind=link}
30156-X&id=gr2.jpg){kind=link}
30156-X&id=gr3.jpg){kind=link}
30156-X&id=gr4.jpg){kind=link}
30156-X&id=gr5.jpg){kind=link}