Reliability of magnetic resonance imaging in assessing myometrial invasion absence in endometrial carcinoma

Objective. This study evaluated the reliability of magnetic resonance imaging (MRI) in assessing the absence of myometrial invasion in endometrial carcinoma patients, to identify candidates for conservative hormonal treatment. Methods. A total of 301 endometrial carcinoma patients who showed Stage I findings on preoperative MRI, and had received staging operations were reviewed retrospectively. We compared MRI findings of myometrial invasion with myometrial invasion levels reported by pathologists when surgical specimens were examined. Design. A descriptive study. Setting. Tertiary care gynecologic oncology clinic, Seoul, Korea. Population. A total of 301 endometrial carcinoma patients who showed Stage I findings on preoperative MRI, and had received staging operations. Main outcome measure. We compared MRI findings of myometrial invasion with myometrial invasion levels reported by pathologists when surgical specimens were examined. Results. Of 301 patients, 17 showed higher stages (FIGO Stages IIA through IIIC) on the final pathology of surgical specimen. Among the 284 patients with Stage I on histology reports, 124 showed no myometrial invasion on preoperative MRI (FIGO Stage IA). The negative predictive value (probability of absence of myometrial invasion) was 49.2%. MRI showed an accuracy of 59.2%, a sensitivity of 68.8%, a specificity of 74.4%, and an 86.9% positive predictive value, for myometrial invasion. Conclusions. MRI had a rather weak predictive value when used to assess absence of myometrial invasion. This should be borne in mind when choosing patients for conservative treatment of endometrial carcinoma.