Diagnostic accuracy of frozen section and patterns of nodal spread in high grade endometrial cancer: a secondary analysis of the SENTOR prospective cohort study

Abstract

<h3>Objectives:</h3> Recent studies have shown that sentinel lymph node (SLN) biopsy accurately detects nodal metastases in patients with high-grade endometrial cancer (EC). SLN biopsy has the potential to replace complete lymphadenectomy for surgical staging in patients with high-grade EC, provided that an SLN algorithm is followed (side-specific pelvic and para-aortic lymphadenectomy for non-mapped hemipelves). The aim of this study was to define the accuracy of intraoperative frozen section for the detection of metastases in SLN biopsy and describe the pattern of lymph node (LN) spread in patients with high-grade EC. <h3>Methods:</h3> We performed a secondary analysis of clinicopathologic data from the SENTOR prospective cohort study evaluating SLN node biopsy at 3 designated cancer centres in Toronto, Canada (ClinicalTrials.gov ID: NCT01886066). Patients with clinical stage I high-grade EC (grade 3 endometrioid, serous, clear cell, carcinosarcoma, undifferentiated, or mixed tumors) undergoing laparoscopic or robotic primary hysterectomy were enrolled. The primary outcome was sensitivity of frozen section of the SLN specimen, compared to a standardized ultrastaging protocol. Secondary outcomes included the pattern and characteristics of LN spread. <h3>Results:</h3> There were 126 patients with high-grade EC with median age 66 years (range: 44-86) and median BMI 26.9 kg/m² (range: 17.6-49.3). Frozen section was performed on surgical specimens from 212 hemipelves; SLNs were identified in 202 specimens (95.7%) and fatty tissue alone was identified in 10 specimens (4.7%). Of the 202 hemipelves in which SLNs were identified, 24 were positive for metastatic disease. Frozen section correctly identified only 12, yielding a sensitivity of 50% (12/24, 95% CI 29.6-70.4) and negative predictive value of 94% (178/202, 95% CI 89-96.5). A total of 24 patients had LN metastases: 16 (13%) had isolated pelvic metastases, 7 (6%) had both pelvic and para-aortic metastases and 1 (0.8%) had an isolated para-aortic metastasis. All para-aortic metastases (n=8) occurred in non-sentinel LNs, among patients with serous (n=5), carcinosarcoma (n=2), and dedifferentiated (n=1) histology. One patient with isolated para-aortic metastases (carcinosarcoma, p53 abnormal) did not have a mapped SLN, but was identified by the SLN algorithm. Nodal metastases occurred in 10% of patients (3/30) with grade 3 endometrioid histology (all pelvic), compared to 33% of patients (15/46) with serous histology (10 pelvic alone, 5 pelvic and para-aortic). <h3>Conclusions:</h3> Intraoperative frozen section of SLNs in high-grade EC patients has poor sensitivity. However, frozen section may still be beneficial to confirm the presence of nodal tissue and ensure adherence to an SLN algorithm, including complete pelvic and para-aortic lymphadenectomy for non-mapped hemipelves. Because isolated para-aortic metastases are rare, para-aortic lymphadenectomy may be omitted in patients in which SLNs were successfully mapped to the pelvis.