The value of MRI in assessing parametrial involvement in endometriosis

STUDY QUESTION What is the accuracy of magnetic resonance imaging (MRI) in the diagnosis of parametrial endometriosis in comparison with surgicopathological findings? SUMMARY ANSWER MRI displayed an accuracy of 96.4% in the preoperative diagnosis of parametrial involvement by deep infiltrating endometriosis (DIE). WHAT IS KNOWN AND WHAT THIS PAPER ADDS MRI is the best technique for preoperative mapping of DIE. This preliminary paper shows that T2-weighted MRI is a valuable tool for the preoperative evaluation of parametrial involvement by endometriosis. DESIGN A retrospective study of an MRI database was used to identify examinations performed in women, who had a clinical suspicion of pelvic endometriosis (n = 666), between 2005 and 2009 in a university medical centre in France. PARTICIPANTS AND SETTING Exclusion criteria were previous surgery for DIE, incomplete surgical evaluation, repeat MRI examinations and incomplete MR protocol. Only symptomatic patients who underwent surgery with a pathological correlation were included (n = 83). An experienced radiologist, blind to the surgical and histological findings, evaluated sagittal, axial and thin-section oblique axial MR images obtained from the 83 patients. DATA ANALYSIS METHOD Descriptive statistics and Fisher exact test were used. MAIN FINDINGS The prevalence of DIE and parametrial endometriosis was 76/83 (91.6%) and 12/83 (14.5%), respectively. The sensitivity, specificity, positive and negative predictive values, accuracy and positive and negative likelihood ratios for the diagnosis of parametrial endometriosis of low signal intensity on T2-weighted MRI, pelvic wall involvement and ureteral dilatation, were 83.3%, 98.6%, 90.9%, 97.2%, 96.4%, 59.2 and 0.17, 58.3%, 98.6%, 87.5%, 93.3%, 92.8%, 41.4 and 0.42 and 16.7%, 100%, 100%, 87.7%, 88%, infinity and 0.83, respectively, with the patient as the unit of analysis. BIAS AND LIMITATIONS The study design was retrospective, and thus prone to bias. Only one experienced reader performed the analysis, so no data are available on intra- or interobserver variability. GENERALISABILITY At present, no consensus exists on the optimal MR protocol to be used for the evaluation of DIE, thus limiting the wider implications of this study. STUDY FUNDING AND COMPETING INTERESTS No funding was obtained for this study. The authors have no conflict of interest.