Endometrial carcinoma: a step closer to individualized therapy? Exploring transcriptional alterations in relation to prognostic biomarkers

Abstract

Background: Endometrial cancer is diagnosed early and has in general a good prognosis. The more important it is to diagnose and treat the poor-prognosis cases. Therapy of endometrial carcinoma patients per today is to a large extent empirically based. Improvements on therapeutic strategies with more personalized focus are needed. Preparing the ground for later clinical studies, by combining clinico-pathologic and molecular data from preclinical studies and cancer patients, is an important step to individualize therapy in cancer. The incidence of endometrial cancer is increasing and the prognostic categorization used in clinical practices today is suboptimal for separating patients according to survival. Despite a focus on translational research in endometrial cancer for some decades, it has proven difficult to bring new biomarkers to the clinic to improve prognostication and prediction of therapy response in this cancer type. Endometrial cancer is behind other cancer types as breast, lung and colorectal cancer in clinical application of molecular classification of tumors to select patients for targeted therapy. Better tumor biological understanding of subgroups, applicability of prognostic markers in a routine clinical setting, and targets for therapy including markers predicting response to such, is important to improve personalized treatment strategies to benefit the endometrial carcinoma patients. Main objectives: The main objective was to study biomarkers potentially associated with endometrial carcinoma progression, to assess their potential as prognostic markers and explore on targets for therapy associated with pathologic expression of these markers. By this, we aimed to provide a rationale for further testing of candidate markers as prognostic and predictive markers in clinical trials. Also, we wanted to focus on biomarker implementation through an important step in the stair-case from research to clinical use; biomarker validation in independent patient series and in a routine clinical setting. Materials and methods: Overall, subsets of one retrospective and one prospective patient cohort were analyzed (Paper I-IV), in addition to an external gene expression microarray (Paper II) and endometrial cancer data from The Cancer Genome Atlas (TCGA, Paper IV), with comprehensive clinico-pathologic and follow-up annotations for all series. DNA oligonucleotide microarrays were analyzed (Paper I-IV). ERα and pStathmin(S38) immunostaining was performed (Paper II and III, respectively). Also, previously published data were included (e.g. EMT markers, data on vascular invasion, proliferation markers, PIK3CA sequencing data, Stathmin and SNP array data). RNA sequencing for gene expression levels were retrieved from 333 endometrial carcinoma samples in The Cancer Genome Atlas (TCGA). Results: DNA aneuploidy was associated to higher age at diagnosis, non-endometrioid histology and high histologic grade in both series studied, and with independent association with reduced survival in multivariate analyses. We found the research and routine diagnostic series to be comparable, with no significant differences in distribution in standard clinico-pathological variables (Paper I). ERα-low tumors were associated with aggressive endometrial cancer and reduced survival in 4 independent patient series. Transcriptional differences based on ERα status revealed pathways, single genes and transcription factors linked to epithelialmesenchymal transition (EMT) enriched in ERα negative tumors, also validated in an external gene expression data set and validated by mRNA and immunohistochemistry in two independent patient series. ERα-low tumor status was also significantly correlated to various markers for PI3Kinase pathway alterations. Furthermore, the gene expression signatures of PI3K/mTOR inhibitors were correlated to ERα-low gene signatures in two independent patient series (Paper II). High pStathmin(S38) immunostaining associated with an aggressive clinico-pathologic phenotype and reduced survival, in both the investigation and validation cohorts. Gene expression patterns related to cell cycle progression were enriched in pStathmin(S38)-high cases. pStathmin(S38) also correlated with a panel of established markers for tumor cell proliferation: Ki67, mitotic count and S-phase fraction. Gene expression signatures representing effect of PI3K/mTOR and HSP90 inhibitors associated with a pStathmin(S38)-high gene expression signature. High pStathmin(S38) correlated significantly with several potential markers for PI3K activation (Paper III). The 29-gene signature score validated to identify patients with increased risk of recurrence, also in patient subgroups with presumed favorable outcome. The 29-gene endometrial carcinoma recurrence score (ECARS) also associated with clinico-pathologic data of aggressive endometrial cancer. ECARS validated to predict overall survival in 332 cases from The Cancer Genome Atlas (TCGA) database. High ECARS associated with vascular invasion and measures for EMT and potential measures for PI3K pathway activation. Assessing ECARS and an EMT signature in metastatic lesions demonstrated an increase of theses signatures from primary to metastatic tumors (Paper IV). Conclusions: DNA aneuploidy identifies aggressive endometrial carcinoma and predicts poor outcome, also in a routine clinical setting (Paper I). Low ERα in endometrial carcinoma is associated with epithelial-mesenchymal transition, vascular invasion and PI3K alterations (Paper II). High pStathmin(S38) associates with high tumor cell proliferation and measures for PI3Kinase activation in endometrial carcinomas (Paper III). The endometrial carcinoma recurrence score (ECARS) validates to identify endometrial carcinomas with shorter recurrence free survival. ECARS increases from primary to metastatic lesions and is associated with measures for PI3Kinase activation and epithelialmesenchymal transition (Paper IV). Low ERα, high pStathmin(S38) and high ECARS predict aggressive endometrial carcinomas and reduced survival, and may suggest treatment with PI3K/mTOR and or EMT inhibitors in clinical trials (Papers II, III and IV).