Objectives The aims of the study were to evaluate agreement between preoperative endometrial samples and surgical specimens in endometrial carcinoma and to correlate this agreement with sample and patient characteristics.

Methods Patients who received primary surgical treatment for endometrial carcinoma at a tertiary care center and had undergone preoperative endometrial sampling were included. Medical records were reviewed to collect information from pathology reports and data on patient characteristics.

Results The study sample comprised 166 patients (mean age, 64.6 years). The histological results of the biopsies were the following: endometrioid cancer (n = 118), nonendometrioid tumor (n = 38), and hyperplasia (n = 10). The agreement rates were 93.2% for endometrioid and 68.9% for nonendometrioid tumors, with a κ coefficient of 0.73 for tumor cell type. Tumor International Federation of Gynecology and Obstetrics (FIGO) grade was distributed as follows: 37.1% G1, 35.7% G2, and 27.1% G3, with agreement rates of 61.5%, 56%, and 78.9%, respectively. The overall κ coefficient for FIGO grading was 0.46. Only 1.9% of the tumors originally classified as G1 were upgraded to G3, whereas 16% of G2 lesions were upgraded. There was no significant difference in agreement rates for tumor cell type and FIGO grade in relation to any of the studied variables, except that biopsy specimens weighing more than 3 g had significantly better agreement in FIGO grading (P = 0.040).

Conclusions Preoperative biopsy has suboptimal accuracy for prediction of characteristics in the definitive surgical specimen. Caution must be taken when using preoperative information to determine extent of surgical resection, due to the risk of understaging. Additional information must be combined with the biopsy data to help in the decision-making process.