Assessing preoperative biopsy accuracy in endometrial carcinoma: A preliminary low‑ and middle‑income country‑based report

Preoperative histopathology is key for guiding the extent of surgical management in endometrial carcinoma (EC). However, the concordance between preoperative endometrial curettage and final histopathology is unclear in low‑ and middle‑income countries. The present pilot study aimed to evaluate the accuracy of preoperative curettage in a Bangladeshi referral center and compare the results with data from middle‑ and high‑income settings. The present cross‑sectional study was performed at a single tertiary center (Department of Gynecological Oncology, National Institute of Cancer Research and Hospital; Dhaka, Bangladesh) to assess agreement between preoperative endometrial curettage and final surgical histopathology for histological subtype and International Federation of Gynecology and Obstetrics (FIGO) grade. Diagnostic performance metrics (sensitivity, specificity, positive and negative predictive values) and Cohen's κ were calculated. A total of 48 patients with EC were included. The mean age was 53.77±13.03 years. For endometrioid vs. non‑endometrioid classification, sensitivity was 93.9 (31/33; 95% CI 79.8‑99.3) and specificity 20.0% (3/15; 95% CI 4.3‑48.1). Accuracy for tumor grading was 77.1% for FIGO grade 1 (95% CI 62.7‑88.0), 79.2% for grade 2 (95% CI 65.0‑89.5) and 93.8% for grade 3 (95% CI 82.8‑98.7), yielding an overall κ‑value of 0.45 (95% CI 0.21‑0.70); 66.7% of presumed grade 1 tumors were ultimately upgraded, underscoring the risk of undertreatment if relying exclusively on preoperative biopsy. Preoperative curettage demonstrated moderate accuracy in predicting final pathology. Given the very small non‑endometrioid numbers and only moderate overall agreement, preoperative curettage findings should be interpreted alongside clinical judgment; adjunct modalities (imaging, immunohistochemical markers) may be helpful where available but were not evaluated in the present study. These preliminary single‑center estimates from a low‑middle income country should be interpreted with caution given the small sample and limited generalizability; larger multicenter studies are warranted.