Identification of Genes and Pathways Differentially Expressed in Progestin Responsive Endometrial Cancer and Hyperplasia

One of the oldest and most common therapies for endometrial complex atypical hyperplasia (CAH) and low-stage, low-grade endometrioid endometrial carcinoma (EEC) is the use of progestins. Importantly, the use of progestins remains the only fertility-sparing treatment available. Despite frequent initial response to progestins, relapse rates are high (35-50%). Currently, there are no biomarkers available for predicting a woman’s likelihood of successful progestin therapy. Primary samples (n = 63) were obtained from a total of 31 patients with either CAH or EEC who underwent progestin therapy and were acquired pre- and post-treatment with progestins. Pathological review of the FFPE samples was performed to identify regions of high hyperplastic or neoplastic content for core punches and RNA extraction. RNA-seq was then performed on the FFPE RNA using the TruSeq RNA Exome approach, a method that uses targeted capture to improve sequencing from fragmented samples. Differential expression analysis was performed using two methods: DESeq2 a parametric method and Noiseq a non-parametric method. Both methods were used to obtain an overlapping subset of genes to reduce spurious results due to samples with outlier expression. Analysis of all samples identified 137 genes significantly associated with outcome. These 137 genes were largely increased in post-treatment samples from progestin responders and were highly enriched for progestogen and estrogen responsive genes, indicating a strong hormonal gene expression response to progestin therapy. Importantly, post-treatment samples from non-responding patients did not show this expression pattern, demonstrating that this set of genes may indicate successful hormone response in post-treatment samples. We also identified a 61 gene signature that remains high in non-responders after treatment compared to responders. Overall, we find that responders show a coordinated change in expression during progestin therapy that is missing from non-responders and this signature could be used in the early evaluation of progestin treatment success. Focusing solely on pre-treatment samples, we identified more variable expression differences across tumors, suggesting multiple reasons for progestin success/failure. We found that the combined expression of estrogen receptor alpha and progesterone receptor was predictive of progestin therapy success. In addition, non-responding tumors had increased expression of several immune-related genes that we are currently exploring. Overall, these results show that progestin therapy response could be predicted using gene expression signatures and that multiple factors may underlie progestin success/failure.