PURPOSE: To identify potential pitfalls in using magnetic resonance (M
R) imaging to determine the depth of myometrial invasion in patients w
ith clinical stage I endometrial carcinoma. MATERIALS AND METHODS: For
ty women with clinical stage I endometrial carcinoma underwent preoper
ative pelvic MR imaging. Uterine length, tumor signal intensity, appea
rance of the junctional zone, presence of large polypoid tumors, leiom
yomata, and congenital uterine anomalies were analyzed. Univariate log
istic-regression analysis was performed to identify associations betwe
en incorrect MR staging and these variables. RESULTS: MR staging of IA
, IB, and IC disease was 55% accurate (22 of 40 cases); MR differentia
tion of deep myometrial invasion (stage IC) from superficial disease (
stages IA and IB) was 78% accurate (31 of 40 cases). Older age (P =.02
5), presence of polypoid tumors (P =.025), and difficulty in pathologi
c staging (P <.005) were significantly associated with incorrect MR as
sessment. CONCLUSION: When present, large polypoid tumors, leiomyomata
, congenital anomalies, small uteri, and indistinct zonal anatomy may
make it difficult to assess myometrial invasion at MR imaging.