Posterior cul-de-sac obliteration associated with endometriosis: MR imaging evaluation

PURPOSE: To retrospectively evaluate the accuracy of magnetic resonance (MR) imaging in depicting posterior cul-de-sac obliteration in patients with endometriosis. MATERIALS AND METHODS: Institutional review board approval was not required for this retrospective study, but informed consent was obtained from all patients. MR images obtained between January 1989 and December 2000 in 57 women (mean age, 39 years; age range, 26-52 years) with historically confirmed endometriosis were retrospectively evaluated by four radiologists independently. All patients underwent laparotomy or laparoscopy less than 1 month after MR imaging. MR images were evaluated for the presence and location of endometrial implants and adhesions. MR images were also scored for the presence of five findings: retroflexed uterus, elevated posterior vaginal fornix, intestinal tethering or tethered appearance of rectum in direction of uterus, faint strands between uterus and intestine, and fibrotic plaque or nodule covering serosal surface of the uterus. Interobserver agreement for each of the five findings and for the overall diagnosis of cul-de-sac obliteration was calculated. Sensitivity, specificity, accuracy, positive and negative predictive values, and κ statistics were determined. RESULTS: Laparotomy or laparoscopy revealed posterior cul-de-sac obliteration in 30 patients. Overall, the four radiologists had mean accuracies of 89.0% and 76.3% for diagnosing endometrial implants and adhesions, respectively, at MR imaging. Overall, the radiologists achieved mean sensitivity, specificity, accuracy, and positive and negative predictive values of 68.4%, 76.0%, 71.9%, 76.6%, and 68.5%, respectively, in diagnosing posterior cul-de-sac obliteration. The best accuracy (mean value, 64.5%) was obtained with the finding of fibrotic plaque in the uterine serosal surface. Readers agreed on the observations 63.2%-91.2% of the time. For the impression of the presence or absence of posterior cul-de-sac obliteration, interobserver agreement varied between substantial and moderate: Mean interobserver agreement was 78.4% (range, 70.2%-84.2%), and mean κ was 0.57 (range, 0.40-0.67). Mean accuracy of MR imaging for diagnosing posterior cul-de-sac obliteration was 71.9%. CONCLUSION: These results suggest that use of the described MR imaging findings may enable diagnosis of posterior cul-de-sac obliteration.