Chronic endometritis (CE) presents as a type of persistent and chronic endometrial inflammation, and investigations have shown that the clinical symptoms of patients with CE are typically not obvious or mild. In clinical practice, it is usually considered that CE and recurrent abortion, infertility, and repeated embryo implantation failure are due to persistent and chronic inflammation of the endometrium that may affect endometrial receptivity [1]. The prevalence of CE in infertile women presently varies greatly in different studies, and one of the important reasons for this is the lack of a diagnostic standard and specific quantitative index for the diagnosis of CE.

Unlike acute endometritis, which is typically caused by direct intrauterine pathogen infection, CE is more closely linked to microbiome dysbiosis in the uterine cavity [2,3]. The uterine cavity is not absolutely sterile, and the imbalance of resident microbial communities—rather than single pathogenic organisms—may drive the persistent low-grade inflammation characteristic of CE [3,4]. However, the exact composition of dysbiotic microbiomes and their pathogenic mechanisms in CE remain incompletely elucidated [2]. The confounding effect of microbiome dysbiosis combined with acute pathogenic organisms has stunted progress in CE research to date, which is also reflected in our observation that microbial culture is not recommended for routine CE diagnosis [3].

CE can lead to infertility, as bacterial endotoxins derived from dysbiotic microbiota and subsequent endometrial microenvironment alterations may constitute one of the causes of implantation failure of healthy embryos [5]. In addition, various types of cytokines in the endometrium of CE patients participate in the pathologic process that subserves the promotion of B-cell differentiation and plasma cell functioning in endometrial homeostasis [6]. A majority of clinical investigators agree that the most effective treatment for restoring fertility in CE patients is the application of antibiotics. Studies have shown that curettage and antibiotic therapy reduce the degree of inflammation in CE patients [7], and it has been reported that patients with successfully treated CE demonstrate higher ongoing pregnancy, clinical pregnancy, and implantation rates than patients with persistent CE [8,9].

Histopathologic diagnosis is the gold standard for many diseases. The most specific and key histopathologic feature of CE is endometrial stromal plasma cell (ESPC) infiltration, which is widely recognized as a diagnostic indicator in clinical and research settings. However, the diagnostic criteria for accurately identifying ESPCs and the minimum number of ESPCs required for diagnosis remain ununified. Microscopic identification of plasma cells, monocytes, and plasmacytoid stromal cells is challenging. Late endometrial secretory phase changes, menstrual cycle characteristics, or secondary alterations induced by exogenous progesterone therapy before biopsy may mask plasma cell presence [10]. Moreover, normal endometrial stroma may contain individual plasma cells, while other pathological changes such as stromal spindle cell transformation, superficial mucosal interstitial edema, interstitial disruption, and focal lymphocyte aggregation lack specificity and clear clinical correlation [11,12,13].

Other common diagnostic methods for CE include hysteroscopy and microbial culture. Hysteroscopy is an important reference for CE diagnosis [14,15,16], with key manifestations including: ① diffuse endometrial congestion (the “strawberry sign”); ② endometrial thickening, edema, and focal congestion; and ③ scattered micropolyps on the endometrial surface [17,18]. Its advantage lies in intuitive visualization and overall assessment of endometrial conditions, but it has inherent subjectivity—different observers may interpret features like endometrial edema differently [19]. Although CE onset and progression are closely linked to microorganisms [3], microbial culture is not recommended for routine CE diagnosis. CE can be caused by various microorganisms (not just bacteria), and many bacteria are non-culturable, leading to technical limitations.

While ESPC infiltration is the consensus histopathologic diagnostic basis for CE, the minimum number of ESPCs required for diagnosis remains debated. Some studies propose diagnosing CE with ≥1 plasma cell per tissue section [20], while others require ≥4–6 plasma cells per high-power field (HPF) [18,21]. A nationwide survey in Japan [22] reported varied diagnostic cutoffs ranging from 3–5 cells/20 high-power fields for chronic endometritis. A systematic review and meta-analysis by Riemma et al. [23] further highlighted the variability in hysteroscopic diagnostic criteria across studies, emphasizing methodological inconsistencies in published reports. This discrepancy highlights the lack of standardized quantification criteria.

The use of immunohistochemical (IHC) staining for CD38 and CD138—classic plasma cell markers—has gained attention for CE diagnosis, but their application remains controversial. First, the specificity of these markers is not absolute. CD38 is not exclusive to plasma cells; it is also expressed on pre-B cells, immature B cells, and normal epithelial cells [24]. CD138 shows non-specific expression in secretory endometrial glands, metaplastic squamous epithelium, and trophoblast cells [25], which may lead to misidentification of non-plasma cell populations as ESPCs. Second, technical factors contribute to diagnostic variability. Non-specific cross-reactivity of CD138 IHC frequently occurs in endometrial epithelial cells, which may be misidentified as stromal plasma cells and lead to overdiagnosis of chronic endometritis [26]. Factors such as antigen retrieval time, primary antibody incubation duration, and antibody concentration can exacerbate this problem [27]. For example, prolonged antigen retrieval may increase non-specific binding, while inappropriate antibody dilution can enhance background staining. Third, clinical and histological confounders further complicate interpretation. The menstrual cycle phase affects endometrial morphology—secretory phase changes may mimic inflammatory infiltrates, and exogenous hormone therapy can alter marker expression [10]. Additionally, individual differences in immune status may lead to variable CD38/CD138 expression levels in plasma cells, affecting quantitative accuracy.

These controversies and potential false-positive sources have hindered the widespread acceptance of CD38/CD138 IHC as a standardized CE diagnostic tool. There is an urgent need to clarify the optimal application of these markers, including standardized staining protocols, quantitative thresholds, and combined interpretation strategies, to improve diagnostic reliability.

In this study, we aimed to address these gaps by: (1) optimizing CD38/CD138 IHC protocols to reduce non-specific staining; (2) comparing the diagnostic consistency of single vs. combined marker use; (3) determining optimal quantitative thresholds for CE diagnosis using hysteroscopy as a reference; and (4) validating these thresholds by analyzing marker expression changes before and after anti-inflammatory therapy. This research seeks to establish a hierarchical reference system for CD38/CD138-based CE diagnosis, providing a reliable basis for clinical decision-making.

This was a retrospective, single-center, observational study with a diagnostic test and therapeutic validation design. The study was conducted at the Reproductive Medicine Center of the First Affiliated Hospital of Shantou University Medical College between June 2020 and December 2021.

The research was divided into two core phases: ① Diagnostic threshold exploration phase: A total of 165 infertile women who underwent both hysteroscopy and endometrial biopsy were enrolled. Using hysteroscopic diagnosis as the clinical reference standard, we evaluated the diagnostic performance of different quantitative thresholds of CD38- and CD138-positive cells (≥1 to ≥10 cells/HPF) via IHC staining, aiming to identify the optimal threshold for pathological diagnosis of CE. ② Therapeutic validation phase: Among the 165 enrolled patients, 90 cases diagnosed with CE (based on preliminary pathological criteria) who received standardized anti-inflammatory therapy and underwent post-treatment endometrial biopsy were included. We compared the changes in CD38- and CD138-positive cell counts before and after treatment, and calculated the negative-conversion rate to validate the clinical utility of the identified thresholds.

All study procedures, including patient enrollment, sample collection, IHC staining, result interpretation, and data analysis, were performed in accordance with standardized operating protocols. Two independent pathologists blinded to the hysteroscopic results conducted the IHC staining interpretation and cell counting to minimize observer bias. The statistical analysis plan was pre-specified to assess diagnostic consistency (Kappa test, McNemar test) and therapeutic response (Wilcoxon rank-sum test), ensuring the rigor and objectivity of the study.

We collected pathological paraffin blocks, hysteroscopic findings, and relevant clinical data from 165 infertile patients who visited the Reproductive Medicine Center of the First Affiliated Hospital of Shantou University Medical College from June 2020 to December 2021. Hematoxylin and eosin staining staining and immunohistochemical staining were performed on the paraffin blocks for subsequent analysis.

Our enrollment criteria were as follows:

(1) infertile women aged 20–45 years;

(2) patients who were not previously diagnosed with CE and never received systemic treatment;

(3) patients who were informed as to our study and agreed to undergo hysteroscopy and endometrial tissue pathological biopsy; and

(4) clinical and pathologic case data that were relatively complete.

Exclusion criteria were as follows:

(1) a previous diagnosis of CE;

(2) a history of endometrial malignant tumors;

(3) recent acute reproductive tract or uterine infection;

(4) recent receipt of a large number of antibiotics; and

(5) women who had ever received CE or infertility-related treatment.

In addition, to minimize biological variability caused by menstrual cycle phases, only endometrial tissues sampled during the proliferative phase were included. The proliferative phase was confirmed based on the patient’s last menstrual period and histomorphological features of the endometrium (e.g., glandular structure, stromal cell density, and absence of secretory changes), ensuring uniform hormonal background across all specimens.

Clinical data included the patient’s age, pregnancy history, and the gross appearance of the endometrium during hysteroscopy. All data were collected from clinical medical records and pathology application forms.

Hysteroscopic CE criteria were meeting any or several of the following: ① endometrial mucosa showing diffuse congestion, and a congested endometrial middle with a white center point, such as “strawberry sign”; ② endometrial thickening, edema, and focal congestion; and ③ coverage of the endometrial surface by scattered micropolyps (referred to as multiple polyps <1 mm in diameter).

The chief instruments, antibodies, reagents, and other consumables required for the study are shown in Tables 1,2,3.

The anti-inflammatory treatment administered to the 90 CE patients was standardized empirical antibiotic therapy, consistent with the core recommendations of the Expert Consensus on the Diagnosis and Treatment of Fertility-Related Chronic Endometritis (2025 Edition) [18]. The specific protocol and implementation details were unified to ensure the reliability of therapeutic validation:

Treatment agent and dosage: All patients received the same combined antibiotic regimen: oral doxycycline 100 mg twice daily + oral metronidazole 500 mg three times daily. This regimen was selected based on the consensus that CE is predominantly caused by bacterial infections (e.g., Escherichia coli, Enterococcus faecalis) and mycoplasma [18], and the combination covers both aerobic and anaerobic pathogens with proven efficacy in CE treatment [9,18].

Treatment timing: Therapy was initiated within 1 week after the initial endometrial biopsy and pathological confirmation of CE. To avoid hormonal interference, the 14-day full course was completed within a single menstrual cycle (proliferative phase to early secretory phase), as recommended by the consensus [18].

Treatment standardization and monitoring: All patients were instructed to complete the full course without interruption. Clinical follow-up was conducted at 2 weeks post-treatment to assess adherence and adverse reactions (e.g., gastrointestinal discomfort), with no serious side effects reported.

Post-treatment biopsy timing: Endometrial re-biopsy was performed in the proliferative phase of the next menstrual cycle (7–14 days after the last menstrual period), ensuring a consistent hormonal background with the initial biopsy to minimize variability in CD38/CD138 expression due to cycle phase [18].

The above data were collected and collated, and statistical analysis was performed using SPSS 25.0 (IBM Corporation, Armonk, NY, USA). All statistical tests were two-sided to comprehensively assess potential differences or correlations in both directions, ensuring the objectivity and rigor of the analysis.

Measurement data were analyzed using the Mann-Whitney U test (for comparing CD38 and CD138 positive cell counts), and the Wilcoxon signed-rank test (for analyzing pre- and post-treatment differences), with statistical significance indicated at p < 0.05. Counting data were analyzed using the χ² test, including the paired χ² test (McNemar test) for evaluating consistency between pathological and hysteroscopic diagnoses, and the Kappa consistency test for assessing agreement between the two methods. For the McNemar test, p < 0.05 indicated a significant difference between the two diagnostic methods, while p > 0.05 indicated no statistically significant difference. For the Kappa consistency test: a Kappa value <0.40 indicated poor agreement, 0.40≤ Kappa value <0.75 indicated moderate agreement, and a Kappa value ≥0.75 indicated good agreement.

To further validate the optimal diagnostic thresholds of CD38- and CD138-positive cell counts, sensitivity, specificity, and receiver operating characteristic (ROC) curve analysis were performed. Hysteroscopic diagnosis was used as the gold standard for defining true positive (TP), true negative (TN), false positive (FP), and false negative (FN) cases.

We enrolled a total of 165 female patients with a median age of 32 (23–43 in December 2021), with 42 at or over 35 years of age (25.5%) and 123 under 35. Of these, 50 (30.3%) were clinically diagnosed with primary infertility and 115 (69.7%) with secondary infertility, and 98 (59.4%) patients showed positive hysteroscopic results. Among them, some showed diffuse endometrium with scattered white dots. There were six cases (3.6%) with “strawberry sign” and 65 with endometrial thickening and edema, with/without focal congestion (39.4%). The remaining 27 cases (16.4%) in the CE-positive group presented with multiple small polyps. A total of 67 cases (40.6%) had normal hysteroscopic morphology without any characteristic CE manifestations and were classified as CE negative (see Table 4).

Infiltration of ESPCs is currently accepted as the histologic manifestation of classical CE. However, upon removal of the characteristic ESPC infiltration, nonspecific changes in endometrial stroma exert an effect on CE, and these manifestations may be secondary changes caused by persistent and chronic inflammation [11]. As observed in this study, the changes in endometrial stroma were diverse, including stromal spindle cell changes, interstitial edema, mesenchymal rupture, and focal lymphocyte aggregation (Fig. 2); these changes were masked in the stroma, and ESPCs proved hard to find. Therefore, it was difficult to accurately identify and count ESPCs in H&E-stained sections.

Fig. 2.

Histopathological changes of endometrial stroma in chronic endometritis. (A) Endometrial stroma with extensive congestion. (B) Endometrial stroma with edema and spindle cell-like changes (magnification, 100×). Scale bar: 100 μm.

Two plasma cell-specific immunophenotype antibodies to CD38 and CD138 were used herein to jointly assist in the recognition of ESPCs, and the staining results are shown in Fig. 3. After IHC staining, we counted positive cells in the hotspot of each section at high magnification and then further evaluated the results. The results showed that the median for CD38-positive cell number was 5 (M: 2–7), while the median for CD138-positive cells was 5 (M: 2–6.5) (the specific results are shown in Fig. 4).

Fig. 3.

CD38- and CD138-staining results. (A) and (C) denote CD38 staining, and (B) and (D) denote CD138 staining. (A–D) are magnified in the same field (400×). Scale bar: 20 μm.

Fig. 4.

Results of IHC staining for CD138 and CD38. Note: ns, not significant; orange lines indicate the median and interquartile range.

We subsequently compared results using the Mann-Whitney U test and found no difference in the number of CD38 and CD138 cells in a high-magnification microscopic field in the hotspot (p = 0.630 > 0.050). This indicated that although there were some differences in the cellular profiles regarding the positive expression of the respective cells, the results for CD38 and CD138 positivity in endometrial tissue with respect to ESPC recognition were relatively consistent.

ROC Curve Analysis for CD38/CD138 Mean Value

ROC curve analysis was performed with hysteroscopic diagnosis as the reference standard to evaluate the diagnostic performance of the CD38/CD138 mean value (Fig. 5).

Fig. 5.

ROC curve for the diagnostic performance of CD38/CD138 mean value in chronic endometritis. The blue curve represents the ROC curve of the CD38/CD138 mean value (AUC = 0.642, 95% CI: 0.554–0.730). The green solid line indicates the reference line (AUC = 0.5), corresponding to no diagnostic value. The Y-axis denotes Sensitivity (true positive rate), and the X-axis denotes 1 - Specificity (false positive rate). ROC,receiver operating characteristic; AUC, area under the ROC curve.

AUC: The AUC for the CD38/CD138 mean value was 0.642 (95% CI: 0.554–0.730, asymptotic p = 0.002), indicating moderate diagnostic efficacy.

Optimal threshold by Youden index: The Youden index (sensitivity + specificity – 1) was maximized at a mean value of 4.25 cells/HPF (Youden index = 0.307), with corresponding sensitivity of 0.755, 1-specificity of 0.448, and specificity of 0.552 (Table 5).

Table 5. Diagnostic metrics at key thresholds of CD38/CD138 mean value (based on ROC curve analysis).

Threshold (Mean Cells/HPF)	Sensitivity	1-Specificity	Specificity	Youden index
4.00	0.796	0.493	0.507	0.303
4.25 (ROC-derived optimal)	0.755	0.448	0.552	0.307
5.00	0.735	0.433	0.567	0.302

Note: 1-Specificity corresponds to the false positive rate. Youden Index = Sensitivity + Specificity – 1, representing the balance of diagnostic efficiency.

ROC curve characteristics: The curve exhibited a gradual rise in sensitivity with decreasing specificity across the tested thresholds (ranging from –1 to 26.5 cells/HPF). Key threshold points aligned with the pre-specified diagnostic and treatment monitoring thresholds:

At 4.0 cells/HPF: Sensitivity = 0.796, Specificity = 0.507, Youden index = 0.303.

At 5.0 cells/HPF: Sensitivity = 0.735, Specificity = 0.567, Youden index = 0.302.

The pre-specified primary diagnostic threshold (5.0 cells/HPF) maintained a Youden index close to the ROC-derived optimal value (0.302 vs. 0.307), confirming its balanced sensitivity and specificity for clinical diagnosis.

We analyzed the results of 90 patients who possessed CD38- and CD138-positive cells in 1 HPF in the hotspot area, and we counted them again, averaged them, compared them with the initial diagnostic data, and assessed the number of positive cells before and after treatment (the specific results are shown in Fig. 6). A total of 77 patients showed a diminution in the number of positive cells after the corresponding treatment, with a median difference before vs. after treatment of 5 (M: 1–7). We additionally demonstrated that the number of CD38- and CD138-positive cells before vs. after treatment was statistically significant (p < 0.001) and that the number of CD38- and CD138-positive cells decreased significantly after anti-inflammatory treatment.

Fig. 6.

Comparison of endometrial stromal plasma cells before and after treatment. Note: Data were analyzed using the Wilcoxon signed-rank test..

Entering the above 10 preset quantitative criteria into the data of 90 cases for review, we found that when the quantitative criteria were ≥4 as an average for CD38- and CD138-positive cells/HPF, 60 (78.9%) patients changed from positive before treatment to negative after treatment, and the negative rate was significantly higher than that for the other quantitative criteria (Table 11).

To compare the diagnostic efficiency of ten quantified criteria for the negative conversion rate after standard treatment, a comparison test among groups was conducted. We found no significant differences in criteria 4 through 7 (p > 0.05), but we did in criteria 1 through 3 (p < 0.05). This suggests that when the average value for CD38- and CD138-positive cells ≥4/HPF was used as the diagnostic criterion for CE, the anti-inflammatory treatment rate in terms of the pathologic diagnosis turning negative was the highest. When CE patients were reviewed with this quantitative standard, the maximal number of patients responded to anti-inflammatory treatment, which indirectly suggests that CE may exist when there are ≥4 ESPCs/HPF in hotspots.

Our study has several limitations that should be acknowledged. First, as a single-center retrospective cohort study, potential selection bias exists, and our findings may not be generalizable to diverse populations (e.g., different ethnicities, healthcare settings beyond reproductive medicine centers, or patients with comorbidities like polycystic ovary syndrome). Second, the lack of a placebo control group prevents us from fully excluding the impact of spontaneous fluctuations in endometrial plasma cell counts (low spontaneous resolution) [32] on post-treatment changes, though the high negative-conversion rate (78.9%) far exceeds the reported spontaneous resolution rate (12.7%) [32], supporting that the reduction in plasma cells is primarily treatment-induced. Third, our sample size (n = 165 for diagnostic exploration, n = 90 for therapeutic validation) is relatively modest, which may limit the precision of threshold estimates, especially for subgroup analyses (e.g., primary vs. secondary infertility). Fourth, we did not collect prospective reproductive outcome data for all patients, and the clinical relevance of our thresholds is supported by alignment with outcome-validated criteria from independent studies [22,23,29] rather than direct in-study outcomes. Fifth, the tissue volumes of endometrial specimens varied, which may have introduced bias into cell counting results.

Future multicenter prospective studies are warranted to validate our hierarchical threshold system in more diverse populations and directly link it to reproductive outcomes. Additionally, research incorporating a placebo control group or serial biopsies across multiple cycles (without treatment) would help clarify the independent effects of cycle-to-cycle variability and anti-inflammatory therapy, further enhancing the robustness of conclusions regarding treatment efficacy in CE management. Standardization of IHC staining protocols (e.g., antigen retrieval time, antibody concentration) across institutions is also needed to reduce inter-laboratory variability and promote the widespread adoption of CD38/CD138-based CE diagnosis.