Objective
The prognostic relevance of hormonal biomarkers in endometrial cancer (EC) has been well-established. A refined three-tiered risk model for estrogen receptor (ER)/progesterone receptor (PR) expression was shown to improve prognostication. This has not been evaluated in relation to the molecular subgroups. This study aimed to evaluate the ER/PR expression within the molecular subgroups in EC.

Methods
A retrospective multicenter cohort study was performed and data from the European Network for Individualized Treatment centers and Vancouver, Canada were used. ER/PR immunohistochemical expression was grouped as: ER/PR 0–10 %, 20–80 % or 90–100 %. Molecular subgroups were determined with full next-generation sequencing or combined with immunohistochemistry: POLEmut, mismatch repair deficient (MMRd), p53mut and no-specific molecular profile (NSMP).

Results
A total of 739 patients were included (median follow-up 5.0 years). Tumors were classified as POLEmut in 9.1 %(N = 67), MMRd in 27.6 %(N = 204), p53mut in 20.8 %(N = 154) and NSMP in 42.5 %(N = 314). Among all molecular subgroups, patients with ER/PR 90–100 % expression revealed the best disease-specific survival (DSS). Within p53mut, PR 90–100 % expression showed a 5-year DSS of 100.0 %. ER expression is prognostic more relevant in MMRd and NSMP tumors while PR expression in p53mut and NSMP tumors. Across all molecular subgroups, PR 0–10 %, p53mut, lympho-vascular space invasion and FIGO stage III-IV remained independently prognostic for reduced DSS Whereas PR 90–100 % and POLEmut remained independently prognostic for improved DSS.
​​​​​​​
Conclusion
We demonstrated that ER/PR expression remain prognostically relevant within the molecular subgroups, and that a three-tiered cutoff refines prognostication. These data support incorporating routine evaluation of ER/PR expression in clinical practice.