A Selective Biomarker Panel Increases the Reproducibility and the Accuracy in Endometrial Biopsy Diagnosis

Abstract

Grading and histologic typing of endometrial cancer in biopsy material has a direct impact on the decision to perform lymphadenectomy and/or omentectomy in many cancer centers. Endometrial biopsies are among the most common general surgical pathology specimens. Multiple studies have shown that biopsy diagnosis suffers from a lack of reproducibility. Although many biomarkers have been proposed, none have been demonstrated to improve the diagnosis in the biopsy setting. In this study, 70 biopsies with endometrial carcinoma were supplemented with a biomarker panel consisting of ER, PR, P53, and DNA ploidy. A representative H&E slide was scanned digitally and made available to 12 gynecologic pathologists in 4 Nordic countries: Finland, Denmark, Sweden, and Norway. Reviewers diagnosed the cases both before and after being provided with the biomarker results. The interobserver percent agreement and Cohen κ improved from 75.8% (κ=0.52, moderate) to 84% (κ=0.68, substantial) with inclusion of the biomarker panel. Agreement with the subsequent hysterectomy diagnosis also improved from 83.6% (κ=0.67) to 88.7% (κ=0.77). There was no statistical improvement between a reflex (84% agreement) and a reflective testing algorithm (82.9% agreement), suggesting that the selective use of biomarkers is appropriate. Difficult cases were almost exclusively high-grade tumors. Finally, a statistical model indicated that only P53 and DNA ploidy, in conjunction with an H&E review, had an impact on the decision to upgrade or downgrade cases.