Introduction/Background Endometrial cancer incidence and mortality trends are increasing worldwide, regardless of therapeutic advances. The most recent ESGO proposal for risk stratification considers stage, histology, grade, myometrial invasion, magnitude/pattern of lymphovascular involvement, and molecular subtype. However, it does not consider host factors that, maintained over time, can condition therapeutic response, recurrence, and survival rates (e.g., obesity/metabolic disorders). Here we investigated whether the expression of genes linked to obesity and metabolic disorders, measured in the primary tumor (a microenvironment reflex of host condition), could better predict the prognosis than body mass index (BMI).

Methodology We analyzed clinical/genomic data of 589 UCEC-TCGA cases downloaded from UCSC-Xena. A list of 425 genes linked to obesity/lipid metabolism (ORG) was used to cluster patients using non-negative matrix factorization. Differential expression, gene set, KEGG enrichment analyses and Cybersort/Ecotyper were performed. Survival curves and Cox-regression models were also built-up.

• Download figure
• Open in new tab
• Download powerpoint

Abstract #1113 Figure 1

Overall survival curves based on 425 ORG NMF clustering and signature risk score. HR of 6 ORG included in signature and Cox model.

Results Here, we first demonstrate that BMI-defined obesity is not associated with molecular subtypes nor does predict survival. On the contrary, the tumor expression of the 425 ORGs makes it possible to identify four clusters correlating with molecular subtypes and prognosis. Additionally, we demonstrate that a signature composed of at least six ORGs (mainly related with epigenetic regulation) identifies two prognosis risk groups (a score risk independent of BMI, stage, histology, degree of differentiation, molecular subtype, or age of the patient). At the tumor microenvironment significant differences can be found among groups particularly in terms of immune cell states.

Conclusion Here we demonstrate that the expression of ORG at the tumor microenvironment correlate better with tumor behavior (molecular subtypes) and clinical outcomes than BMI does. These findings provide additional evidence on the importance of correcting metabolic disorders, which are not necessarily related to the classic definition of obesity, and which negatively influence therapeutic response and long-term results.

Disclosures This research was supported by Fondecyt 1201083 granted to MC and CONICYT-PFCHA/Doctorado Nacional 2019-Folio 21190421 granted to FG