It is not the time to abandon intraoperative frozen section in endometrioid adenocarcinoma: A large‐scale, multi‐center, and retrospective study

INTRODUCTION: Stage IB (deep myometrial invasion) high‐grade endometrioid adenocarcinoma (EA), regardless of LVSI status, is classified into high‐intermediate risk groups, requiring surgical lymph node staging. Intraoperative frozen section (IFS) is commonly used, but its adequacy and reliability vary between reports. Hence, we determined the utility of IFS in identification of high‐risk factors, including deep myometrial invasion and high‐grade. METHOD: We retrospectively analyzed 9,985 cases operated with hysterectomy and diagnosed with FIGO stage I/II EA in postoperative paraffin section (PS) results at 30 Chinese hospitals from 2000 to 2019. We determined diagnostic performance of IFS and investigated whether the addition of IFS to preoperative biopsy and imaging could improve identification of high‐risk factors. RESULTS: IFS and postoperative PS presented the highest concordance in assessing deep myometrial invasion (Kappa: 0.834), followed by intraoperative gross examination (IGE Kappa: 0.643), MRI (Kappa: 0.395), and CT (Kappa: 0.207). IFS and postoperative PS presented the highest concordance for high‐grade EA (Kappa: 0.585) compared to diagnostic curettage (D&C 0.226) and hysteroscope (Hys 0.180). Sensitivity and specificity for detecting deep myometrial invasion were 86.21 and 97.20% for IFS versus 51.72 and 88.81% for MRI, 68.97 and 94.41% for IGE. These figures for detecting high‐grade EA were 58.21 and 96.50% for IFS versus 16.42 and 98.83% for D&C, 13.43 and 98.64% for Hys. Parallel strategies, including MRI‐IFS (Kappa: 0.626), D&C‐IFS (Kappa: 0.595), and Hys‐IFS (Kappa: 0.578) improved the diagnostic efficiencies of individual preoperative examinations. Based on the high sensitivity of IFS, parallel strategies improved the sensitivities of preoperative examinations to 89.66% (MRI), 64.18% (D&C), 62.69% (Hys), respectively, and these differences were statistically significant (p = 0.000). CONCLUSION: IFS presented reasonable agreement rates predicting postoperative PS results, including deep myometrial invasion and high‐grade. IFS helps identify high‐intermediate risk patients in preoperative biopsy and MRI and guides intraoperative lymphadenectomy decisions in EA.