Accuracy of preoperative endometrial biopsy and intraoperative frozen section in predicting the final pathological diagnosis of endometrial cancer.

Abstract Background Histotype and grade of endometrial cancer (EC) are prognostic factors of nodal involvement and thus of survival. Preoperative biopsy (PB) and intraoperative frozen section (FS) are usually used to guide surgical staging on which the choice of adjuvant therapy will be based successively. Objective The aim of this study was to assess the agreement rate between PB and FS with final diagnosis (FD) in a series of surgically resected EC. Materials All patients submitted to hysterectomy for EC or atypical endometrial hyperplasia in the Reggio Emilia Province hospitals from 2007 to 2018 were included. Concordance rate differences in histotype, grading, myoinvasion, risk of recurrence between PB, FS and FD were assessed with Fisher's exact test and Mc Nemar contingency test. Results A total of 352 patients were identified. For 345 patients it was possible to compare PB and FD results. FS examination was performed in 201/352 (57%) cases, while for 21/352 (6%) patients only an intraoperative macroscopic evaluation was done; in the remaining women, FS-exam was omitted. In 14/201 (7%) cases the tumor wasn't grossly identifiable and the random FS-sampling wasn't able to find the tumor site. High diagnostic concordance of tumor type between PB and FD was observed: no significant differences were registered in type 1 and type 2-endometrial cancer identification (83%, 73%, p = 0.121). Significant differences (p = 0.005) were observed comparing FS and FD results: 95% of type 1-ECs were correctly diagnosed by FS, while only 76% of type 2-ECs received a correct diagnosis on FS. PB showed a concordance with FD among tumor grading close to 55% whilst concordance achieved 71% grouping low grade (G1-G2) EC. No significant differences in FS and FD concordance rate were observed between tumor grades. Concordance for low grade was significantly higher than for high grade ECs (89% vs 50%, respectively, p value = 0.014). The concordance rate in evaluating the myoinvasion status between FS and FD was 80% (n: 199 patients), reaching 99% after combining the first 2 groups (0–49% vs ≥ 50%). Twenty-two cases underwent only intraoperative macroscopic evaluation of the myoinvasion, with an accuracy of 91%: only in 1 case the invasion of the cervical stroma was not detected (Stage II), and 1 case the patient was overstaged as Ib. Discrepancies were observed in FS capacity to correctly predict the final ESMO risk group in stage I patients: FS resulted particularly reliable in predicting a low-risk (concordance with FD: 91%) while the accuracy sharply decreased for intermediate- and high-risk patients (62% and 40%, respectively). To investigate the usefulness of FS in EC management, we compared patients who underwent FS (FS-group) or not (no–FS–group). Especially for low risk patients, the FS significantly increased the adequacy of surgical treatment from 53% (no–FS–group) to 72% (FS-group) (p = 0.016). Conclusions FS remains a useful tool to tailor surgery in EC-patients, avoiding secondary surgery to complete staging particularly in patients with AH + AHBA, low and intermediate risk ECs that could benefit from adjuvant therapy.