Loss of B-cell lymphoma 2 immunohistochemical expression in endometrial hyperplasia: A specific marker of precancer and novel indication for treatment: A systematic review and meta-analysis.

Abstract

Endometrial hyperplasia is differentiated into benign or premalignant. Two histological classifications are used for this purpose: World Health Organization (WHO) classification, based on cytological atypia, disregarding glandular complexity, and endometrial intraepithelial neoplasia (EIN) classification, based on several different parameters. B-cell lymphoma 2 (Bcl-2) loss has been studied as immunohistochemical marker with the aim of improving the differential diagnosis between benign and premalignant hyperplasia. We aimed to evaluate: (A) Bcl-2 loss as marker of endometrial precancer, by assessing it in proliferative endometrium, benign hyperplasia, premalignant hyperplasia, and endometrial cancer; (B) the diagnostic accuracy of Bcl-2 in the differential diagnosis between benign and premalignant endometrial hyperplasia; (c) how the results change according to the histological classification and the thresholds of Bcl-2 expression used. Electronic databases were searched from their inception to March 2018. All studies assessing Bcl-2 immunohistochemistry in endometrial specimens were included. In total, 20 observational studies assessing 1,278 specimens were included. Bcl-2 loss rates were not significantly different between proliferative endometrium and benign hyperplasia (P=0.12) and between premalignant hyperplasia and endometrial cancer (P=0.53). Among hyperplasias, Bcl-2 loss was significantly associated with premalignancy, according to both the WHO (OR=4.39; P<0.00001) and EIN classifications (OR=6.07; P=0.01), and also with architecture complexity (OR=2.06; P=0.02). Using the WHO classification, Bcl-2 loss showed low diagnostic accuracy in detecting premalignant hyperplasia (area under the curve [AUC]=0.708), with a sensitivityof 0.41, a specificityof0.81, a positive likelihood ratioof3.22, and a negative likelihood ratioof 0.69. Using the EIN classification, accuracy was high (AUC=0.938), with a sensitivityof 0.18, a specificityof 0.97, a positive likelihood ratioof5.16 and a negative likelihood ratioof0.86. Thresholds of Bcl-2 expression not involving a complete loss showed lower diagnostic accuracy with a slight increase in sensitivity, but a severe decrease in specificity. B-cell lymphoma 2 loss is a marker of endometrial precancer, with a high specificity and high diagnostic accuracy if the EIN classification is used. Thresholds of Bcl-2 expression not involving a complete loss should not be considered. Bcl-2 loss in endometrial hyperplasia may be a novel indication for treatment when precancerous features are ambiguous in a histological examination. Bcl-2 loss correlates better with EIN classification than with the WHO classification, suggesting that glandular complexity is an important precancerous feature.