INTRODUCTION: Endometriosis is a chronic and disabling disease that affects roughly 10% of women worldwide. The current gold standard for diagnosis is laparoscopic visualization of the endometriosis lesions with histopathological confirmation. Multiple modalities have been used to attempt to identify endometriosis, with the most common being transvaginal ultrasound (TVUS) and magnetic resonance imaging (MRI). Currently, MRI is typically used as a tool for pre-surgical planning. However, to date, no large-scale studies have looked at the diagnostic accuracy of the primary MRI after a patient has undergone surgery for endometriosis with inclusion of variables such as age, parity, BMI, and surgical history. Additionally, no studies have confirmed the accuracy of MRI -observed location of endometriosis compared to the location of the endometriosis found in surgery. OBJECTIVE: To determine the utility and accuracy of MRI as a preoperative diagnostic and planning tool for endometriosis surgery. METHODS: This study was a retrospective chart review. The study population included all patients at a single academic medical center who had an MRI for pelvic pain in 2023 and subsequently underwent surgery performed by a member of the OBGYN department for removal of endometriosis. Patients were excluded if they had prior endometriosis surgery. Patient demographics, MRI impression, surgical operative report, and pathology reports were extracted from patient charts. Using R, descriptive statistics were performed to assess the utility of MRI as a diagnostic and planning tool compared to surgical and/or pathology-confirmed endometriosis, with 95% confidence intervals. RESULTS: Overall, 325 patients had an MRI and subsequently underwent endometriosis surgery. Of these patients, 269 had surgical or pathology-proven endometriosis. When used for diagnosing patients confirmed by either pathology or surgery, there was no significant association of MRI as a diagnostic tool (p=0.133). The sensitivity was 0.83, specificity was 0.27, positive predictive value was 0.84, and the negative predictive value (NPV) was 0.25 for MRI use in diagnosing endometriosis. A sub-analysis was done excluding patients who had endometriomas. This analysis included 181 patients, and again, there was no significant association of MRI as a diagnostic tool (p=0.273) and a lower sensitivity (0.77) than the group with endometriomas. Finally, an analysis was done to assess agreement of endometriosis on MRI compared to intraoperatively. 50 locations were coded, and a McNemar’s test was used to determine the rate of agreement. Only 17 of the 50 locations coded were in agreement between MRI imaging and intraoperative findings. CONCLUSIONS: MRI is frequently used to assess pelvic pain and assist with preoperative planning. However, at a single academic medical center, the NPV of MRI was low, suggesting that patients still underwent endometriosis surgery in the setting of a negative MRI. Additionally, the accuracy of MRI in locating endometriosis was low, with only 17/50 coded locations being in agreement via a McNemar’s test. This calls into question the utility of using MRI as a preoperative tool for diagnosis and surgical planning routinely compared to developing designated institutional criteria or a national protocol.Table 1Table 2
Lipschultz et al. (Fri,) studied this question.