Molecular profiling of endometrial carcinoma : closing in on personalized medicine

The ultimate goal of personalized medicine is to offer optimal treatment to each patient while avoiding under- and overtreatment. At present, clinicopathological risk algorithms determine whether lymph node dissection or postoperative radiotherapy/chemotherapy are indicated for a patient with endometrial carcinoma (EC). As traditional risk assessment methods suffer from limited accuracy, researchers are trying to find molecular biomarkers that could help guide the treatment of EC. In 2013, a landmark study by the Cancer Genome Atlas research group (TCGA) identified four prognostic molecular subgroups of EC. The study cohort of this thesis consisted of unselected EC patients who were surgically treated at Helsinki University Hospital between 2007 and 2012. Tumor samples from 842 patients were included in a tissue microarray. The aim of this thesis was to investigate clinicopathological and prognostic significance of L1 cell adhesion molecule (L1CAM) and MLH1 methylation status in EC. In addition, we explored the differences between the two largest molecular subgroups of EC with regard to the prognostic impact of individual risk factors and the expression of an immunotherapy target molecule, programmed death ligand 1 (PD-L1). I: Tumoral L1CAM positivity was associated with more frequent distant relapses and independently predicted poor prognosis in endometrioid EC but not in non-endometrioid EC. L1CAM may enhance the accuracy of post-operative risk stratification algorithms guiding adjuvant treatment of EC. II: Integrating preoperatively assessed tumoral L1CAM with the conventional risk assessment models did not facilitate stratification of patients to lymphadenectomy. III: In multiplex immunohistochemistry, PD-L1 expression was more frequent in intratumoral immune cells (27.7%) than in carcinoma cells (8.6%). Within the molecular subgroups the relative frequency of PD-L1 positivity was higher in POLE-mutated and mismatch repair deficient (MMRd) tumors than in the other molecular groups (p53 abnormal and no specific molecular profile). Further, we found differences in PD-L1 expression across histological subgroups. IV: MMR deficiency correlated with several clinicopathological risk factors. Methylated phenotype was associated with older age and larger tumor size. Methylated MMRd phenotype predicted poor disease-specific survival compared with MMR proficient EC, but the difference with non-methylated MMRd EC was not significant. V: Our study confirmed the survival differences between molecular subgroups originally proposed by TCGA: POLE-mutated EC had an excellent prognosis, followed by NSMP, MMRd, and the more aggressive p53 abnormal EC. In multivariable analysis adjusting for conventional risk factors, NSMP and MMRd molecular subgroups did not present significant survival differences. Interaction analysis confirmed differences between NSMP and MMRd with regard to prognostic impact of individual risk factors. This thesis identifies several molecular factors that may facilitate treatment planning of EC. Given their distinct prognosis and pathogenetic mechanisms, molecular EC subgroups should probably be regarded as disease entities of their own. This way optimal risk stratification algorithms can be determined leading to personalized and cancer type-specific treatment of patients with EC.