Objective: Preoperative and intraoperative assessments of myometrial invasion (MI) are commonly used for planning surgical procedures such as dissection of the para-aortic node; however, the assessments often differ from the final diagnosis determined by pathological examination. The present study evaluated the accuracy of preoperative and intraoperative assessments of MI.

Methods: A total of 191 patients with endometrial cancer, who underwent hysterectomy from 1995 to 2007 in Tottori University Hospital, were included in this study. One hundred seventy-four patients underwent endometrial curettage or Pipelle biopsy preoperatively. Histological grade was compared between preoperation and postoperation. Magnetic resonance imaging (MRI) was performed before surgery, and the depth of MI was assessed as 3 levels (no MI, <50%, and >50%). During surgery, the uterine wall was incised at the most invasive part, and then, intraoperative gross assessment was evaluated as less than or greater than 50%.

Results: Histological evaluation revealed that 34 patients had no invasion, 97 had less than 50% MI, and 60 had greater than 50% MI. On MRI assessment, 135 patients had correct diagnoses, and the accuracy was 70.7%. Regarding the diagnosis of greater than 50% MI depth, the accuracy, the sensitivity, and the specificity of the MRI assessment were 83.2%, 75.0%, and 85.7%, respectively. Seventeen patients were overestimated, and 15 patients were underestimated by the MRI assessment. On intraoperative gross assessment, 162 patients had correct diagnoses, 8 patients were overestimated, and the remaining 21 patients were underestimated. The accuracy of the gross assessment was 84.8%, the sensitivity was 65.0%, and the specificity was 93.9%. The preoperative grading accuracy was 71.8% (125/174). A discrepancy between preoperative and postoperative grades was more frequent in a low-grade tumor. The incidence of underdiagnosis was significantly higher in patients with a grade 3 (G3) tumor than in those with a G1 or G2 tumor in both assessments.

Conclusions: The present study suggests that gross assessment may be useful to determine MI of less than 50%, although patients with a G3 tumor were more frequently underestimated.