Objective To compare the diagnostic characteristics of the evaluation of myometrial invasion (MI) retrospectively between preoperative magnetic resonance imaging (MRI) and intraoperative frozen sections. Design A retrospective study. Setting University hospital. Sample 201 women diagnosed with endometrial carcinoma. Methods All women underwent preoperative MRI and 111 of them also underwent intraoperative frozen section assessment. The final pathological evaluation was used as the definitive diagnosis. Main outcome measures In women who underwent MRI and frozen sections (n = 111), the accuracies of detection of MI and of deep invasion (defined as ≥50% invasion) were compared. Results The accuracy, sensitivity, and specificity of MRI for detection of MI were 65.8, 58.8, and 88.5%, and those in frozen sections were 90.1, 90.6, and 88.5%, respectively. The accuracy and sensitivity of frozen sections were significantly higher (p < 0.001, p < 0.001), whereas the specificity of the two methods did not differ (p = 1.000). The accuracy, sensitivity, and specificity of MRI for detection of deep invasion were 83.8, 69.2, and 88.2%, and those of frozen sections were 93.7, 73.1, and 100.0%, respectively. The accuracy and specificity of frozen sections were significantly higher (p = 0.007 and p < 0.001, respectively), whereas sensitivity did not show a significant difference (p = 0.999). Conclusion In assessment of MI, the accuracy of frozen sections was significantly higher than that of MRI. Since the diagnostic characteristics differ between two methods, additional intraoperative frozen sections are recommended for more accurate assessment of MI when MRI is negative for the presence of any MI or positive for the presence of deep invasion.
ASJC Scopus subject areas