p63 and
cytokeratin (CK) 5/6 are markers of basal and squamous differentiation in several normal epithelia and human
tumors and are also suggested to be markers of progenitor or stem cells in certain stratified epithelia. In
endometrial carcinoma, there is very limited information about the expression pattern of p63 or CK5/6 and no prognostic information. The aim of our study was to examine whether the expression of these markers was associated with a certain
tumor phenotype in terms of other
biomarkers, clinicopathologic characteristics and patient prognosis. Immunohistochemical expression of p63 and CK5/6 was examined using tissue microarrays (TMAs) in a large population-based series of 276
endometrial carcinomas with long and complete follow-up. Selected cases of normal and hyperplastic endometrium were examined for comparison (n = 15). Absence of p63 expression (70%) was significantly associated with nonendometrioid
carcinomas, high histologic grade (FIGO), higher mitotic count and
tumor cell proliferation by Ki-67,
microsatellite instability (MSI) and loss of hMSH6 expression. A tendency toward reduced patient survival was also seen (p = 0.098). Presence of CK5/6 expression was more frequent in endometrioid
tumors with squamous differentiation, while loss of CK5/6 expression (54%) was significantly associated with high FIGO stage, reduced
beta-catenin expression, MSI and reduced patient survival (p = 0.0001); the latter was also found within the endometrioid subgroup (p = 0.0004). Multivariate survival analysis revealed that loss of CK5/6 expression had an independent prognostic impact in addition to well-known prognostic variables. Expression of both markers was increased in simple
hyperplasia compared with normal endometrium. In complex
hyperplasia, p63 expression was also increased, whereas CK5/6 was positive in areas with squamous differentiation only. Thus, loss of p63 or CK5/6 was associated with features of aggressive
tumors, and lack of CK5/6 was significantly associated with reduced survival in multivariate analysis.