Clinical and Histopathological Correlates of Endometrial Proliferative Lesions in Perimenopausal Women: A Retrospective Study with Internal Validation of a Risk Model

TL;DR

This study identifies diabetes and endometrial thickness as risk factors for advanced endometrial lesions in perimenopausal women, offering a risk model for clinical use.

Contribution

The paper introduces a risk model for predicting advanced endometrial pathology in perimenopausal women, validated internally with bootstrap resampling.

Findings

Diabetes is independently associated with advanced endometrial pathology (aOR 2.75).

Endometrial thickness increases with histopathological severity (ρ = 0.634).

The risk model shows moderate discrimination (AUC 0.68) and acceptable calibration.

Abstract

Background: Endometrial proliferative lesions are common in the menopausal transition and carry a measurable risk of carcinoma. Early risk stratification may guide evaluation and follow-up. Methods: We performed a single-center retrospective study of 315 women aged 45–55 years (May 2021–May 2024) at a private clinic in Bucharest. Lesions were classified per WHO 2014 as hyperplasia without atypia, atypical hyperplasia/endometrial intraepithelial neoplasia (AH/EIN), or adenocarcinoma; “advanced pathology” was defined as AH/EIN or adenocarcinoma. Clinical comorbidities and transvaginal ultrasound endometrial thickness were recorded. Associations were tested with χ2; odds were estimated with multivariable logistic regression (adjusted ORs), with a modified Poisson sensitivity analysis for adjusted relative risk. Thickness differences were compared by one-way ANOVA, and severity correlations by Spearman’s ρ. Internal validation used 1000-bootstrap resampling. Results: Hyperplasia without atypia comprised 74.6% of cases, AH/EIN 20.0%, and adenocarcinoma 5.4% (advanced pathology 25.4%). Diabetes was independently associated with advanced pathology (aOR 2.75; 95% CI 1.14–6.61; p = 0.0237), while a history of non-atypical hyperplasia was inversely associated (aOR 0.31; 95% CI 0.13–0.72; p = 0.0068). Obesity showed a borderline association (aOR 1.79; 95% CI 0.98–3.26; p = 0.058), and long-term oral contraceptive use also approached significance (aOR 0.42; 95% CI 0.18–1.00; p = 0.051). Endometrial thickness increased stepwise with histopathological severity (ANOVA p < 0.0001; η2 = 0.44) and correlated with ordered severity (ρ = 0.634). The multivariable model showed moderate discrimination (AUC 0.68; optimism-corrected 0.66) with acceptable calibration (slope 0.92; Hosmer–Lemeshow p = 0.052) and overall accuracy (Brier 0.18). Conclusions: In perimenopausal abnormal bleeding, metabolic comorbidities—especially diabetes—together with increased endometrial thickness identify women at higher risk of AH/EIN or carcinoma. Histopathology remains the diagnostic reference. The model can aid clinical prioritization but requires external validation and should not be used as the sole basis for decisions.

Figures