Endometrial Thickness as a Prognostic Marker in Premenopausal Abnormal Uterine Bleeding: Diagnostic Accuracy and Kaplan–Meier Survival Insights

2.1 Study design and setting

A retrospective, longitudinal and analytical study was carried out over 8-years and 11-months, from January 2016 to November 2024, at Gynaecology and Obstetrics Department B, Charles Nicolle Hospital, Tunis, Tunisia.

2.2 Study population

Premenopausal women presenting with AUB who underwent TVUS followed by endometrial biopsy (via hysteroscopy or curettage) and subsequent definitive surgical treatment (hysterectomy) were identified.

AUB refers to bleeding from the uterine corpus that is abnormal in regularity, volume, frequency or duration. It encompasses heavy menstrual bleeding, irregular menstrual bleeding and intermenstrual bleeding.⁹

All TVUS examinations were performed by three experienced operators, each holding an inter-university diploma in gynecologic ultrasound and having more than 10 years of clinical experience in pelvic ultrasound practice. ET was measured in the sagittal plane as the double-layer distance between the two endometrial–myometrial interfaces, at the point of greatest thickness perpendicular to the midline, using appropriately magnified images, in accordance with standard ultrasound assessment protocols used at our institution.

The ultrasound features of different endometrial and other intracavitary pathologies were assessed using the International Endometrial Tumor Analysis (IETA) terminology.¹⁰

The study population consisted of patients who fulfilled the following criteria:

A study flowchart detailing case selection and exclusions has been developed ( Figure 1).

Figure 1. Flowchart of the study population.

2.3 Variables

Data were collected from patient records onto a pre-established form.

2.3.1 Patients’ characteristics

Age, family and personal history, AUB (type, abundance, number of haemorrhage episodes, associated symptoms …), delay before consultation, time to diagnosis …

2.3.2 TVUS features

The primary data extracted from the TVUS reports included¹⁰:

2.3.3 Uterine cavity exploration

Hysteroscopy or curettage, date of endometrial sampling, abundance of the sample, anatomopathological diagnosis from the biopsy.

2.3.4 Definitive anatomopathological diagnosis

The definitive histological diagnosis on the hysterectomy specimen was considered the reference standard.

2.4 Statistical analysis

All analyses were performed using IBM SPSS Statistics (Version 26.0; IBM Corp, Armonk, NY, USA) for data management, descriptive statistics, correlation analysis, and non-parametric tests.

For comparative analysis, variables significantly associated with endometrial malignancy were assessed using the chi-square test or Fisher’s exact test for categorical variables and Student’s t test or Mann-Whitney U test for continuous variables.

Multivariate logistic regression models were constructed to identify independent predictors of endometrial malignancy cases. Variables with a p value ≤ 0.20 in the univariate analysis were included in the model. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were reported. A p value ≤ 0.05 was considered statistically significant.^11,12

These variables comprised: oral contraceptive use, previous uterine endoscopic procedure, number of hemorrhage episodes, vascularization score, bleeding abundance, and endometrial thickness >9 mm (See Table 1).

The diagnostic performance of each sonographic feature for predicting endometrial malignancy was assessed using sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).¹³

A Kaplan-Meier survival analysis was conducted to assess the prognostic value of ET for predicting the timing of a malignant diagnosis.¹⁴

2.5 Ethical considerations

The study protocol received approval from the institutional ethics committee of Charles Nicolle Hospital, Tunis, Tunisia on 6 March 2025 by before conducting the study (approval number FWA 00032748- IORG0011243).

Given the retrospective design and the use of anonymized data, the committee granted a waiver of informed consent.

3.1 Comparative study

The median age was 46 years in benign cases and 48 years in malignant cases (p = 0.217).

The median BMI was 25 for benign and 26 for malignant cases (p = 0.576).

A family cancer history was present in 6% of benign cases and 94% of malignant cases (p = 0.123).

Diabetes was present in 7% of benign cases and 93% of malignant cases (p = 0.338), while hypertension was present in 6% of benign cases and 94% of malignant cases (p = 0.954).

Median age at menarche was 12 years in both groups (p = 0.603).

Gravidity and parity differed, with median gravidity of 3 versus 2 (p = 0.018) and median parity of 3 versus 2 (p = 0.060).

Breast neoplasia, tamoxifen use, and pelvic irradiation were not associated with malignancy.

Bleeding characteristics—including abundance and type—showed limited association, except for the number of hemorrhage episodes (p = 0.027), where patients with 0–1 episode had higher malignancy rates (25–33%) than those with ≥2 episodes.

The median delay before consultation was significantly shorter in malignant cases (6 months) compared to benign cases (12 months, p = 0.011).

Regarding ultrasound features ( Table 2), ET was markedly higher in malignant cases (median 12 mm) than in benign cases (median 5 mm, p < 0.001). Vascularization (p < 0.001), with Score 2 showed 71% malignancy, whereas normal vascularization or Score 1 had much lower rates. The uterine size, the endometrial–myometrial interface and intracavitary images did not show significant differences. Ovarian cysts did not show a significant association with malignancy (p = 0.808).

Presence of fibroids was associated with a lower risk of malignancy (3.8% malignant in patients with fibroids vs. 14% in those without, p = 0.009).

Hemostatic curettage, and hysteroscopy did not show significant differences. Quantity of curettage specimen was significant (p = 0.003), with abundant specimens associated with 10% malignancy compared to 0% for scanty specimens.

3.2 Receiver operating characteristic (ROC) and logistic regression analyses ( Figure 2, Table 3)

Figure 2. Predictive performance of endometrial thickness versus clinical features (gravidity, number of hemorrhage episodes) for endometrial malignancy based on ROC analysis.

ET measured via TVUS was unequivocally identified as the single most robust independent predictor of endometrial malignancy within our studied cohort. Its superior performance, quantified by an excellent Area Under the Curve (AUC) of 0.842 (p < 0.001), underscores its unparalleled utility in the initial triage of patients presenting with AUB.

This diagnostic power significantly surpassed that of other clinical variables, such as gravidity (AUC = 0.690) and the number of hemorrhage episodes (AUC = 0.576, nonsignificant), establishing ET as the cornerstone of non-invasive risk assessment.

Variables showing an association with malignancy in univariate analysis (p < 0.10) were included in a multivariate logistic regression model. The final model identified several independent predictors of malignant endometrial pathology (Table 4).

The derivation of an optimal cut-off >9 mm provides a clear, evidence-based threshold for clinical decision-making. This criterion successfully balances sensitivity (69.2%) and specificity (87.1%), ensuring a high detection rate for malignancies while effectively minimizing false alarms.

The specificity of 87.1% is particularly noteworthy; it signifies that most patients with benign conditions can be correctly identified by this simple measure, preventing anxiety and unnecessary procedures.

However, the most profound clinical implication of our findings lies in the metric of the NPV, which reached 97.7%. This is arguably the most significant result of the analysis. It translates to a powerful clinical rule: an endometrial measurement of ≤9 mm effectively rules out malignancy with a certainty of over 97%. This provides immense reassurance to both the clinician and the patient, creating a safe pathway to conservatively manage a large portion of low-risk individuals. It can justify a strategy of watchful waiting or medical management for these patients, thereby reducing the number of invasive, costly, and potentially uncomfortable procedures like hysteroscopy or dilation and curettage.

Conversely, the PPV of 26.5% for a thickness >9 mm contextualizes its use. It confirms that while a thick endometrium is a strong risk marker mandating histological sampling, it is not diagnostic of cancer on its own. Most positive tests (73.5%) will be due to benign conditions like hyperplasia or polyps. Therefore, the role of ET is not to definitively diagnose cancer but to act as an exceptionally efficient screening gatekeeper—identifying all high-risk patients who require definitive biopsy while safeguarding low-risk patients from undue intervention.

3.3 Kaplan-Meier survival curve analysis

To assess the prognostic value of ET for predicting the timing of a malignant diagnosis, two Kaplan-Meier survival analyses were conducted:

The cohort was stratified by an ET cutoff of >9 mm:

3.3.1 Time-to-consultation analysis

The interval from symptom onset to first consultation differed significantly between groups (Log-rank test χ² = 25.662, p < .0001): ( Figure 3)

Figure 3. Kaplan–Meier survival analyses: time from symptom onset to initial consultation (time-to-consultation).

This indicates that symptoms associated with a thicker endometrium (e.g., heavier bleeding) prompt patients to seek medical attention much sooner.

3.3.2 Time-to-diagnosis analysis

Analysis of the interval from consultation to definitive histological diagnosis also revealed a significant difference (log-rank t χ² = 25.662, *p* < .0001): ( Figure 4).

Figure 4. Kaplan–Meier survival analyses: time from initial consultation to definitive histological diagnosis (time-to-diagnosis).

This finding highlights the rapid clinical progression in the high-ET group compared to the indolent course associated with a thinner endometrium.

In summary, ET exceeding 9 mm is a critical marker that shortens both the patient-related delay in seeking care and the system-related delay in achieving a diagnosis. It effectively identifies a high-risk subgroup of premenopausal women with AUB who require urgent and efficient clinical evaluation.

4.1 Endometrial thickness cutoff in premenopausal AUB

Several studies have attempted to define an optimal cutoff value of ET for predicting endometrial pathology in women with AUB, yet the results remain inconsistent.

Our findings identify an ET >9 mm as the optimal threshold for predicting endometrial malignancy in premenopausal women.

This result aligns with earlier observations but provides more robust statistical validation.

Smith et al.¹⁷ and Tongsong et al.¹⁸ were among the first to suggest a correlation between thickened endometrium and pathology. Smith et al.,¹⁷ evaluating 51 premenopausal women with irregular bleeding, proposed an ET cutoff of 8 mm. In a larger study including 177 peri- and postmenopausal women undergoing vaginosonography, Tongsong et al.¹⁸ found that an ET ≤7 mm reliably excluded endometrial pathology.

Subsequent studies yielded heterogeneous thresholds.

Getpook et al.¹⁹ reported ≤8 mm as rarely malignant. Mayuri et al.²⁰ in a cross-sectional study of 62 perimenopausal women above 40 years of age, proposed an ET cutoff of ≥8 mm.

Luca et al.,²¹ in a cohort of 240 premenopausal women with AUB, identified >11 mm as the best cutoff for predicting endometrial carcinoma. This diagnostic performance was very close to Giannella et al.²² who found ≥11 mm predictive in the presence of obesity or diabetes.

Ruan et al.²³ and Li et al.²⁴ incorporated ET ≥10 mm into multivariate prediction models, emphasizing its central role when combined with clinical variables.

Collectively, our findings bridge the gap between the lower cutoff values suggested in earlier work (7–8 mm) and the higher threshold of 11 mm reported by Luca et al., supporting >9 mm as a pragmatic and evidence-based threshold in premenopausal women with AUB.

4.2 Diagnostic performance of ET and AUC analysis

The diagnostic accuracy of ET in our cohort was high, with an AUC of 0.842. Sensitivity was 69.2% and specificity 87.1%, which is consistent with prior reports.^20,22

Particularly noteworthy was the very high NPV (97.7%), highlighting the reliability of ET in excluding malignancy. This finding is in line with those of Smith et al.¹⁷ and Mayuri et al.,²⁰ who likewise reported NPVs exceeding 94%.

Consistent with our results, Kumari et al.²⁵ reported that a TVS-ET threshold of 10.5 mm achieved the highest diagnostic accuracy, with sensitivity and specificity of 89.5% and 86.3%, respectively, corresponding to an AUC of 0.920 (95% CI 0.846–0.994, p < 0.0001). Their cutoff also demonstrated a high NPV (95.7%), together with a PPV of 70.7% and favorable likelihood ratios, further supporting ET as a robust rule-out tool.²⁴

Conversely, the PPV was modest (26.5%), reflecting the frequent occurrence of benign causes of thickened endometrium such as polyps or hyperplasia. This aligns with the findings of Smith et al.,¹⁷ who reported an even lower PPV of 14%. Taken together with our results, this mirrors the broader literature, where ET is consistently shown to be an excellent rule-out tool but insufficient as a stand-alone diagnostic marker.^19,22,23

Nevertheless, by effectively excluding malignancy in low-risk patients, the cutoff has the potential to substantially reduce unnecessary endometrial biopsies and invasive interventions.

4.3 Integration with clinical risk factors

Although ET is highly informative, its predictive value increases when integrated with clinical parameters. Advanced age, obesity, diabetes, and hypertension are well-established contributors to endometrial carcinogenesis.^23,24 For instance, Giannella et al.²² demonstrated that the risk of cancer is significantly higher in obese women with ET ≥11 mm compared to their lean counterparts.

In our cohort, diabetes and hypertension were uncommon and not significantly associated with malignancy, likely reflecting the relatively small number of cancer cases (n = 13). Nevertheless, the combination of ET measurement with metabolic risk profiling remains promising, particularly in refining personalized risk stratification models.

In conclusion, the paramount importance of ET is its ability to stratify risk with high efficiency. It moves beyond mere association to provide actionable intelligence. By offering an objective, reproducible, and non-invasive metric, it forms the critical first step in a modern diagnostic algorithm for AUB, optimizing resource allocation, minimizing patient burden, and enhancing the overall precision of patient care.

4.4 Kaplan–Meier analysis: A novel prognostic perspective

A major innovation of our study lies in applying Kaplan–Meier survival analysis to ET, an approach not previously reported. It is worth reiterating that time-to-consultation is an indicator of clinical presentation dynamics, not of disease progression.

We demonstrated that women with ET >9 mm had significantly shorter time-to-consultation and time-to-diagnosis compared with those with ET ≤9 mm. This observation has two implications:

Introducing Kaplan–Meier methodology into ET research thus provides a fresh lens for understanding patient trajectories and may inform prioritization strategies in healthcare systems with limited resources.

Our findings also have relevant implications for low-resource settings where access to hysteroscopy is limited or unavailable. In such contexts, the identification of an ET threshold >9 mm with high diagnostic accuracy (AUC 0.842) offers a practical and cost-effective triage strategy. ET measurement by TVUS, which is widely available and comparatively inexpensive, may help prioritize patients who require referral to higher-level facilities for further endometrial evaluation. This approach could optimize resource allocation, reduce unnecessary invasive procedures, and support earlier detection of malignant pathology in constrained healthcare systems.

4.5 Strengths and limitations

Our study has several notable strengths:

However, some limitations must also be acknowledged:

4.6 Clinical implications

4.7 Recommendations for further research