Contrast-enhanced MRI in the preoperative evaluation of endometrial carcinoma.

Original Article

Abstract

Objective: To determine the diagnostic accuracy of preoperative contrast-enhanced magnetic resonance imaging (CE-MRI) for assessing myometrial invasion, cervical invasion, and lymph node metastasis in endometrial carcinoma and to evaluate their association with postoperative histopathological findings.

Methodology: This Prospective Study was conducted in the Department of Obstetrics and Gynecology, Ayub Teaching Hospital, Abbottabad, Pakistan, from January to June 2025. Following institutional Ethical Review Committee approval, 144 patients with biopsy-proven endometrial carcinoma underwent pelvic CE-MRI within one month before surgery. We assessed myometrial invasion, cervical invasion, and lymph node metastasis on T2-weighted and contrast-enhanced T1-weighted images. We used histopathological findings as the definitive reference standard. We used correlation analysis to assess associations between MRI and histopathological findings, and logistic regression to evaluate factors associated with positive pelvic lymph nodes.

Results: For detecting myometrial invasion, cervical invasion, and lymph node metastasis, CE-MRI showed sensitivities of 85.71%, 90.9%, and 85.7% and specificities of 89.47%, 100%, and 79.6%, respectively. Histopathological lymph node metastasis was strongly associated with ≥50% myometrial invasion (100% sensitivity, 97.9% specificity), whereas the association with cervical invasion was poor. In univariate analysis, disease stage, tumor grade and type, positive peritoneal cytology, myometrial invasion, and cervical stromal invasion were significantly associated with lymph node involvement.

Conclusion: CE-MRI demonstrated good diagnostic performance for assessing local tumor invasion and lymph node metastasis. Patients with <50% myometrial invasion and no cervical invasion on CE-MRI may have a lower likelihood of lymph node metastasis, potentially supporting individualized surgical planning.