Background  Preoperative tumor grading becomes one of the most important predictors for lymphadenectomy at primary surgery for clinical stage I endometriod adenocarcinoma. However, there is an inconsistency of tumor grade between preoperative curettage and final hysterectomy specimens, and its associated factors are poorly understood. This study aimed to evaluate the accuracy of tumor grade by preoperative curettage so as to achieve a better stratified management for clinical stage I endometriod adenocarcinoma.
Methods  Clinical data of totally 687 patients with clinical stage I endometriod adenocarcinoma who underwent preoperative curettage and primary surgery were retrospectively collected. Compared with final hysterectomy specimens, the sensitivity, specificity, accuracy, positive predictive value, and negative predictive value of tumor grade by preoperative curettage were calculated and their associations with clinicopathologic parameters, including age, status of menopause, position of uterus, location and size of lesion, histological grade, depth of myometrial invasion, cervical invasion, extrauterine spread, peritoneal cytology, metastasis to retroperitoneal lymph node, serum CA125 level, and hormone receptor status, were analyzed.

Results  In final hysterectomy specimens, 139 of 259 grade 1 patients by curettage were upgraded to grade 1 or 2; 31 of 296 grade 2 were upgraded to grade 3, with a significantly discrepant rate of 40.9% (281/687) and an upgraded rate of 24.7% (170/687). The specificity and negative predictive value for grade 3 were 90.7% and 89.9%, while the sensitivity and positive predictive value for grade 1 were 67.1% and 40.9%, respectively.
Conclusions  Preoperative tumor grade by curettage does not accurately predict final histological results, especially in those classified as grade 1. Complete surgical staging seems to be necessary for clinical stage I endometriod adenocarcinoma.