Abstract 17P
Background
Pre-operative histologic subtype and deep myometrial infiltration at magnetic resonance imaging are strong predictors of high-risk disease in endometrial cancer (EC). Whether molecular subtype in combination with hormone receptor status can refine conventional risk stratification is uncertain.
Methods
A prospectively collected EC cohort including 446 patients was molecularly subtyped using surrogate markers and the WHO-endorsed classification algorithm. Median follow-up was 6.4 years. Estrogen- and progesterone receptor (ER and PR) status was investigated by IHC and scored using the staining index method. Uni- and multivariate analyses to predict disease-specific survival (DSS) were performed. The multivariate model included patient age, preoperative risk groups, molecular subtypes and combined ER/PR status.
Results
Patients were classified as POLE (9%), MMR-D (29%), copy-number low (46%) and copy-number high (16%). Loss of ER and/or PR expression was found in 36% of the tumors. Both molecular type and dichotomized ER/PR expression associated with DSS in univariate analyses (p < 0.001). However, after adjusting for preoperative risk group, loss of ER/PR outperforms molecular class for predicting poor DSS (ER/PR: p = 0.004, MolClass: p > 0.05).
Conclusions
Preoperative loss of ER/PR predicts poor prognosis and outperforms molecular class for improving risk stratification of EC patients.
Legal entity responsible for the study
University of Bergen.
Funding
Norwegian Cancer Society, University of Bergen, Helse Vest, The Research Council of Norway.
Disclosure
All authors have declared no conflicts of interest.