The accuracy of pre-operative and intra-operative pathologic diagnosis of complex atypical hyperplasia and endometrial cancer (371)

Objectives: The accuracy of frozen pathology to detect endometrial cancer (EC) among patients with a preoperative diagnosis of complex atypical hyperplasia (CAH) directly influences decisionmaking regarding surgical staging. There is little data exploring what factors influence this accuracy. If there are patient factors that decrease the accuracy of the frozen section, then operative decisionmaking should be altered to accommodate these scenarios. We aimed to assess the accuracy of pre, intra, and postoperative diagnosis among patients with an initial histological diagnosis of CAH. In those patients where the pathologic diagnoses did not match, we investigated any common factors among these patients. Methods: An IRB-approved cohort study of 200 patients was conducted among patients with a diagnosis of CAH who underwent hysterectomy between 2019 to 2020. Information about patient demographics, comorbid conditions, preoperative biopsy, intraoperative frozen biopsy, and final pathology was collected. Matching between biopsy results was noted as a percentage, and only an identical diagnosis was considered a match between evaluations. Comparison of clinical variables between matched and mismatched groups was performed. Numeric variables were compared using the independent 2-sample t-test or Wilcoxon rank-sum test. Categorical variables were compared between the two groups using the Chisquare test or Fisher's exact test. Results: The accuracy of intraoperative frozen compared to postoperative pathology for diagnosis of CAH and EC was 75.8% (95% CI: 0.6283-0.8613). The accuracy of preoperative compared to postoperative pathology for diagnosis of CAH and EC was 72.5% (95% CI: 0.636-0.8025). The prediction accuracy for endometrial cancer between the intraoperative frozen specimen and postoperative pathology was reported as sensitivity, specificity, and positive and negative predictive values of 57.6%, 100%, 100%, and 64.1%, respectively. Among all the patient demographics, comorbid conditions, and surgical factors investigated, there were no statistically significant differences between the rate of matched or unmatched diagnosis between the intraoperative frozen specimen and postoperative pathology. Conclusions: Within our institution, sensitivity, specificity, positive predictive, and negative predictive values were comparable to other studies. Although not statistically significant, there was a higher rate of non-matching intraoperative frozen to final pathologic diagnosis in women with a higher BMI, post-menopausal status, current smokers, preoperative biopsy by dilation and curettage, and prior IUD or DMPA use. This study suggests further exploration of these factors and their influence on the accuracy of frozen pathology analysis.