Evaluation of myometrial invasion in endometrial cancer: comparison of diffusion-weighted magnetic resonance and intraoperative frozen sections

Abstract

To compare the diagnostic accuracy of diffusion-weighted magnetic resonance imaging (DWI) and frozen-section analysis in assessing myometrial invasion in endometrial cancer to guide surgery. In this prospective study, 51 women with diagnosed endometrial cancer were examined using magnetic resonance imaging (MRI) and subsequently underwent hysterectomy with intraoperative frozen-section assessment. The MRI protocol included T2-weighted images (T2WI), a dynamic T1-weighted post-gadolinium series, and DWI sequences acquired with b values of 0, 150, and 800s/mm(2). Apparent diffusion coefficient (ADC) maps were obtained in all the studies. Deep myometrial invasion was defined as involvement of ≥50% of the thickness of the myometrium. The final postoperative pathological evaluation was considered as the reference standard. The accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of DWI for detecting deep invasion of the myometrium were 90.2%, 77.8%, 97%, 93.3%, and 88.9%, respectively. For the intraoperative frozen section, these figures were 90.2%, 73.7%, 100%, 100%, and 86.5%. The precision for both tests was the same using McNemar's test (p=1). In assessing the depth of myometrial invasion, ADC maps show the same accuracy as intraoperative histological studies in endometrial cancers. They also allow for a more precise assessment than conventional MRI sequences. Frozen-section analysis can be avoided if the preoperative MRI study includes DWI sequences and ADC maps.