| Description |
The incidence rate of endometrial hyperplasia (EH) and endometrial cancer (EC) have doubled over the last two decades going along with a drop in average diagnostic age and a statistically significant reduction in survival rate. The exploration of four genomic clusters by The Cancer Genome Atlas (TCGA) Network provides molecular insights and opens up the opportunities to improve the current management strategies. However, the advancement of genomic-guided treatment is hampered by the lack of appropriate disease models to obtain prospective validation of the relationship between cluster-association and treatment response. Work in this dissertation attempts to develop predictive, clinically relevant mouse models for treatment evaluation through three approaches: 1) providing a new mouse model for EH, 2) establishing and characterizing endometrial cancer patient-derived xenografts (EC-PDXs) for drug screening, 3) refining the current treatment strategy based on genomic classification using PDX preclinical trials. The doubled incidence rate suggests there is a need to develop a reliable, clinically relevant precancer model to study tumorigenesis and identify prevention strategies. In the first approach, an estrogen-induced EH mouse model was developed. Histological analysis of endometrial tissues demonstrated that this mouse model is capable of reflecting different stages of disease development, hormonal receptors status, and genetic alterations as seen in the human disease. In order to provide improved treatment strategies for EC patients based on genomic classification, a predictive model that can provide a broad spectrum of preclinical efficacy is required. In the second approach, a panel of EC-PDXs was developed by orthotopically transplanting patient cancer tissue specimens into murine uteri and propagating them over multiple generations. Established tumor grafts retain crucial histo-pathological characteristics, the capacity to form distant metastasis following known clinical patterns, as well as recapitulating molecular features of the original human tumor specimens. This model serves as a tool to investigate the need for reclassifying EC and feasibility of directing genomic-guided treatment-based on TCGA recommendations. The last approach aims to validate the predictive capabilities of utilizing genomic classification and cluster affiliation in the refinement of current treatment strategies conducting PDX preclinical trials. The results demonstrate that selecting therapeutics based on affiliation to genomic clusters indeed results in improved treatment response and accuracy of prediction, and genomic similarities across various cancer types can be used to direct treatment. The endeavor described in this dissertation aims to provide a more predictive, clinically relevant model for EH and EC. The results demonstrated the utility of genomic classification to enable the prediction of drug response, and thus may facilitate the refinement of current management in a more precise and personalized way. |
