Preoperative Magnetic Resonance Imaging versus Intraoperative Frozen Section Diagnosis for Predicting the Deep Myometrial Invasion in Endometrial Cancer: Our Experience and Literature Review

Aim The present study was designed to directly compare the diagnostic performance of preoperative magnetic resonance imaging (MRI) and intraoperative frozen section (FS) diagnoses in predicting deep myometrial invasion (MI) of endometrial cancer. Methods Using MRI findings and FS diagnoses, 194 patients with surgically staged endometrial cancer were evaluated for deep MI between 2006 and 2018. Definitive histological diagnosis of paraffin sections of excised tissues was used as the gold standard approach. Results Of 194 cases, 53 (27.3%) cases were finally diagnosed as having deep MI (≥50%). There was 82% total agreement between MRI and FS diagnoses in predicting deep MI, with a kappa value of 0.54 (95% confidence interval [CI] = 0.40–0.67, moderate agreement). The sensitivity of FS diagnosis (0.66, 95% CI = 0.52–0.78) for predicting deep MI was lower than that of MRI (0.77, 95% CI = 0.63–0.87; p = 0.21), while the specificity of FS (0.98, 95% CI = 0.93–0.99) was significantly higher than that of MRI (0.88, 95% CI = 0.81–0.93; p = 0.001). Overall, the accuracy of FS (0.89, 95% CI = 0.84–0.93) was higher than that of MRI (0.85, 95% CI = 0.79–0.90), although the difference did not reach statistical significance (p = 0.23). The accuracy (0.95, 95% CI = 0.90–0.97) was very high in cases with concordant MRI and FS results. Conclusions MRI and FS showed different diagnostic characteristics for predicting deep MI, with a higher specificity observed for FS and the greatest accuracy obtained in concordant cases. Thus, our findings recommend the addition of FS diagnosis, either alone or in conjunction with MRI, to MI evaluation.