Magnetic Resonance Imaging Compared with Rectal Endoscopic Sonography for the Prediction of Infiltration Depth in Colorectal Endometriosis

Study objective: To compare the accuracies of magnetic resonance imaging (MRI) and rectal endoscopic sonography (RES) in the prediction of the infiltration depth of colorectal endometriosis.

Design: A retrospective cohort study (Canadian Task Force classification II-2).

Setting: A university teaching hospital.

Patients: Forty patients with symptomatic deep infiltrating endometriosis (DIE) of the rectum who underwent colorectal resection were included.

Interventions: All patients underwent abdominopelvic MRI and RES preoperatively to assess the infiltration depth of colorectal endometriosis, and segmental resection of the rectosigmoid by laparoscopy was performed if RES showed bowel invasion. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive and negative likelihood ratios (LRs), and intermethod agreement were calculated for DIE muscularis and submucosal/mucosal infiltration confirmed by histopathological analysis.

Measurements and main results: For MRI detection of DIE muscularis infiltration, the sensitivity, specificity, PPV, NPV, and negative LR were 68%, 100%, 100%, 20%, and 0.32, respectively. For the MRI detection of DIE submucosal/mucosal involvement, the sensitivity, specificity, PPV, NPV, and positive and negative LRs were 47%, 81%, 69%, 63%, 2.49, and 0.65, respectively. The PPV of RES detection of DIE muscularis infiltration was 93%. For the RES detection of DIE submucosal/mucosal layers, the sensitivity, specificity, PPV, NPV, and positive and negative LRs were 79%, 48%, 58%, 71%, 1.51, and 0.44, respectively.

Conclusion: In the current study, MRI is valuable for detecting endometriosis of the rectum but is less accurate in detecting submucosal/mucosal involvement than RES. Magnetic resonance imaging was not successful for preoperative determination of segmental resection versus a more conservative approach. When bowel involvement is detected by MRI, RES is not essential. When symptoms suggest DIE in patients without intestinal lesions detected by MRI, RES is necessary to exclude bowel invasion.