DIAGNOSTIC ACCURACY OF NON-CONTRAST MRI FOR DETECTING ENDOMETRIAL CARCINOMA: A CROSS-SECTIONAL STUDY AGAINST HISTOPATHOLOGY IN A PAKISTANI TERTIARY-CARE CENTER

Abstract

Background: Endometrial carcinoma is the most common gynecologic malignancy, and timely, accurate diagnosis guides definitive management. In resource-variable settings such as Pakistan, a robust, non-contrast MRI protocol could offer practical diagnostic value alongside histopathology. Objective: To determine the diagnostic accuracy of magnetic resonance imaging (MRI) in detecting endometrial carcinoma, using histopathology as the gold standard. Methods: We conducted a cross-sectional descriptive study at the Department of Radiology, Liaquat University Hospital, Hyderabad, over six months (ethical approval: CPSP/REU/RAD-2018-164-2697). Consecutive women aged 18–65 years with clinical suspicion of endometrial carcinoma underwent pelvic MRI using non-contrast T1-weighted, T2-weighted, and diffusion-weighted (DWI) sequences. Images were reported by a senior radiologist (≥3 years post-training). Histopathology from endometrial biopsy served as the reference standard. Diagnostic indices—sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), overall accuracy—and receiver operating characteristic (ROC) area under the curve (AUC) were calculated. Results: A total of 227 women were included (mean age 49.6 ± 8.7 years); 62.6% were postmenopausal. Histopathology confirmed carcinoma in 102 (44.9%). MRI identified 110 positives, yielding 94 true positives, 16 false positives, 8 false negatives, and 109 true negatives. MRI performance was: sensitivity 92.1% (95% CI ~85.3–96.0), specificity 87.2%(95% CI ~80.2–92.0), PPV 85.5% (95% CI ~77.7–90.8), NPV 93.2% (95% CI ~87.1–96.5), and overall accuracy 89.9%. The ROC AUC was 0.90, indicating excellent discrimination. Subgroup analysis showed slightly higher accuracy in postmenopausal vs premenopausal women (91.5% vs 87.1%). Conclusion: Non-contrast multiparametric MRI (T1/T2/DWI) demonstrated high accuracy for detecting endometrial carcinoma against histopathology, with a strong NPV that can confidently exclude disease in many clinically suspected cases. These findings support MRI as a practical, reliable adjunct to guide diagnostic and preoperative pathways in our tertiary-care setting.