Detection and correlation of pre-operative, frozen section, and final pathology in high-risk endometrial cancer

To evaluate sensitivity and specificity of pre-operative and frozen section pathologic evaluation (FSA) in predicting high-risk (HR) histology endometrial cancer.

A retrospective analysis was performed on all patients diagnosed with endometrial cancer at a single institution. Medical records were abstracted for baseline characteristics, surgical reports for staging, and final histology was confirmed by a gynecologic pathologist.

868 patients were identified. Of these, 118 had Grade 3 endometrioid, 36 clear cell carcinoma (CCC), 47 carcinosarcoma (CS), and 84 uterine papillary serous carcinoma (UPSC) histology. Endometrial biopsy (EMB) had an overall sensitivity of 90%, 77% for low grade, 78% for HR, with a specificity of 0%. For dilation and curettage (D&C), overall sensitivity was 85%, 69% for low grade, and 77% for HR. Specificity was 33%. Sensitivities for combined pre-operative testing for G3 endometrioid, CCC, CS, and UPSC were: 56%, 28%, 72%, and 60%, respectively. For frozen section analysis (FSA), overall sensitivity was 77%, and 67% for low and high grade. For G3 endometrioid, CCC, CS, and UPSC, sensitivities were 57%, 20%, 74%, 32%, respectively. Specificity was 95%. FSA identified an additional six patients (8%) with UPSC, CCC or CS that were pre-operatively low risk, providing an 8% improvement in sensitivity but decreased specificity.

Pre-operative EMB and D&C are overall very sensitive for detecting endometrial cancer; however, sensitivity decreases with HR histology. Pre-op testing will miss 28% of HR diagnoses and FSA provides an opportunity to identify some patients with UPSC, CCC, and CS. If pre-operative results suggest HR cancer, the surgeon should proceed with comprehensive surgical staging without an FSA.