A novel preoperative scoring system for predicting endometrial cancer in patients with complex atypical endometrial hyperplasia and accuracy of frozen section pathological examination in this context: a multicenter study.

OBJECTIVE

To estimate the accuracy of frozen pathological analysis in patients preoperatively diagnosed with endometrial atypical complex hyperplasia (ACH) and to develop a model predicting the probability of the presence of endometrial carcinoma in ACH.

METHODS

Patients (n = 128) who underwent total hysterectomies because of diagnoses of ACH were evaluated at four tertiary centers.

RESULTS

Diagnoses made using frozen sections and permanent sections were in good agreement (ĸ = 0.61, p < 0.0001). A useful scoring system combining weighted risk factors derived from a regression model is as follows: (2 × age ≥50 years) + (3 × BMI ≥30) + (2 × menopausal status) + (2 × diabetes comorbidity) + (3 × endometrial thickness >12 mm). The AUC of this score was 0.793, and the score afforded 80.9% sensitivity, 70.3% specificity, a 75.3% positive predictive value, and a 76.4% negative predictive value, when a score of 5 was used as a cutoff.

CONCLUSIONS

Patients with ACH should be evaluated by gynecological oncologists and intraoperative frozen section analysis should be performed by pathologists specializing in the evaluation of gynecological malignancies, because ACH is closely associated with endometrial cancer (EC). Our novel preoperative scoring system may aid in the detection of patients at increased risk of EC and thus guide general gynecologists.