Objectives: To characterize the relationship between DNA mismatch repair (MMR), ARID1A, and PTEN in endometrial intraepithelial neoplasia (EIN).Methods: Type 1 endometrial carcinoma typically develops from a precursor lesion, endometrial hyperplasia, or EIN. The DNA MMR system prevents tumor progression by correcting errors that occur during DNA replication and is present in a proportion of endometrial cancers. Tumor suppressor genes ARID1A and PTEN are also implicated in the molecular pathogenesis of endometrial carcinoma. Previous studies have shown that in endometrial carcinoma, MMR deficiency may be associated with ARID1A mutations. However, in EIN, the relationship between loss of expression (LOE) of MMR, ARID1A, and PTEN has yet to be established.In a retrospective review, we identified 113 patients with EIN on endometrial biopsy from 2009 to 2014. Tissue microarray blocks were evaluated with immunohistochemical stains using antibodies against MLH1, PMS2, MSH2, MSH6, ARID1A, and PTEN. Statistical analyses were performed, with a P < .05 considered statistically significant.Results: In EIN, rates of LOE of MMR, ARID1A, and PTEN were 4.4% (n = 5), 6.2% (n = 7), and 50.4% (n = 57), respectively. Loss of MMR expression was not significantly associated with expression of ARID1A (P = 1.000) or PTEN (P = .2062). Development of a subsequent invasive malignancy was not significantly associated with LOE of ARID1A (P = .1135), DNA MMR (P = .2424), or PTEN (P = .1670). When combined, LOE of ARID1A and MMR was significantly associated with subsequent cancer (P = .0109).Conclusions: In this series of EIN, the frequency of PTEN LOE (50.4%) agreed with that seen in previous studies. Unlike rates demonstrated in endometrial carcinoma, the proportion of LOE in MMR (4.4%) and ARID1A (6.2%) seen in the precursor lesion were relatively low and thus seem to be nonessential in the development of EIN. There were no cases of concordant LOE of MMR and ARID1A, suggesting independent pathways or differences in temporal patterns in gene expression during the premalignant phase. When combined, LOE of ARID1A and MMR was significantly associated with the development of a subsequent malignancy, suggesting potential prognostic markers, which will require further evaluation. Objectives: To characterize the relationship between DNA mismatch repair (MMR), ARID1A, and PTEN in endometrial intraepithelial neoplasia (EIN). Methods: Type 1 endometrial carcinoma typically develops from a precursor lesion, endometrial hyperplasia, or EIN. The DNA MMR system prevents tumor progression by correcting errors that occur during DNA replication and is present in a proportion of endometrial cancers. Tumor suppressor genes ARID1A and PTEN are also implicated in the molecular pathogenesis of endometrial carcinoma. Previous studies have shown that in endometrial carcinoma, MMR deficiency may be associated with ARID1A mutations. However, in EIN, the relationship between loss of expression (LOE) of MMR, ARID1A, and PTEN has yet to be established. In a retrospective review, we identified 113 patients with EIN on endometrial biopsy from 2009 to 2014. Tissue microarray blocks were evaluated with immunohistochemical stains using antibodies against MLH1, PMS2, MSH2, MSH6, ARID1A, and PTEN. Statistical analyses were performed, with a P < .05 considered statistically significant. Results: In EIN, rates of LOE of MMR, ARID1A, and PTEN were 4.4% (n = 5), 6.2% (n = 7), and 50.4% (n = 57), respectively. Loss of MMR expression was not significantly associated with expression of ARID1A (P = 1.000) or PTEN (P = .2062). Development of a subsequent invasive malignancy was not significantly associated with LOE of ARID1A (P = .1135), DNA MMR (P = .2424), or PTEN (P = .1670). When combined, LOE of ARID1A and MMR was significantly associated with subsequent cancer (P = .0109). Conclusions: In this series of EIN, the frequency of PTEN LOE (50.4%) agreed with that seen in previous studies. Unlike rates demonstrated in endometrial carcinoma, the proportion of LOE in MMR (4.4%) and ARID1A (6.2%) seen in the precursor lesion were relatively low and thus seem to be nonessential in the development of EIN. There were no cases of concordant LOE of MMR and ARID1A, suggesting independent pathways or differences in temporal patterns in gene expression during the premalignant phase. When combined, LOE of ARID1A and MMR was significantly associated with the development of a subsequent malignancy, suggesting potential prognostic markers, which will require further evaluation.