Interlaboratory Concordance of ProMisE Molecular Classification of Endometrial Carcinoma Based on Endometrial Biopsy Specimens

Abstract

Molecular classifiers improve the consistency of categorization of endometrial carcinoma and provide valuable prognostic information. We aimed to evaluate the interlaboratory agreement in ProMisE assignment across 3 dedicated Canadian gynecologic oncology centers. Fifty cases of endometrial carcinoma diagnosed on biopsy were collected from 3 centers and 3 unstained sections were provided to each participating site so that immunohistochemistry for MSH6, PMS2, and p53 could be performed and interpreted at each center, blinded to the original diagnoses and the results from other centers. A core was taken for DNA extraction and POLE mutation testing. Overall accuracy and κ statistic were assessed. MSH6, PMS2, and p53 could be assessed for all 50 cases, with agreement for 140/150 results. There was a high level of agreement in molecular classification (κ=0.82), overall. Cases with a discordant result for one of the features used in classification (n=10) were reviewed independently and the most common reason for disagreement was attributable to the weak p53 staining in 1 laboratory (n=4). Interpretive error in PMS2 (n=1) and MSH6 (n=2) assessment accounted for 3 of the remaining disagreements. Interpretive error in the assessment of p53 was identified in 2 cases, with very faint p53 nuclear reactivity being misinterpreted as wild-type staining. These results show strong interlaboratory agreement and the potential for greater agreement if technical and interpretive factors are addressed. Several solutions could improve concordance: central quality control to ensure technical consistency in immunohistochemical staining, education to decrease interpretation errors, and the use of secondary molecular testing.