Endometrial cancer is a heterogenic group of malignancies with differences in pathogenesis and clinical behavior. Correct risk stratification of these patients is essential for successful allocation of treatment modalities. Currently, classification of endometrial cancer is based solely on clinicopathological parameters. The aim of this thesis was to study the genomic heterogeneity of endometrioid endometrial carcinoma (EEC) and to identify prognostic immunohistochemical biomarkers for disease stratification.

This study is based on patient material derived from 640 patients with up to 30 years of follow-up. Gene expression profiling using two different microarray platforms revealed Apolipoprotein E (APOE) to be the most overexpressed gene in poorly differentiated EEC when compared to well-differentiated carcinoma. Immunohistochemical analysis of early stage EEC specimens suggested that progesterone receptor (PR) has an independent role in prognostication. Further studies suggested that l-asparaginase (ASRLG1) could serve as a novel prognostic biomarker in EEC. In an attempt to produce a clinically useful prognostic panel of biomarkers, immunohistochemical stainings of tissue microarrays combined with a machine learning-based method were employed. The results demonstrate that EEC patients can be stratified into three groups with significantly different clinical behavior using p53 and ASRGL1 stainings.

In summary, the study highlights the importance of PR, p53 and the novel biomarker ASRGL1 in EEC prognostication. The present findings suggest that the panel of tissue biomarkers developed can be used for identification of patients who are at risk of aggressive disease course and an unfavorable outcome of EEC.