Abstract Background Endometrial cancer has rising incidence and mortality, with high recurrence rates (10–70%) post-surgery. Discrepancies between preoperative biopsy and postoperative pathology, particularly in estrogen receptor (ER) and progesterone receptor (PR) expression, may misguide treatment. This study explored ER/PR expression dynamics pre- and post-surgery and developed a recurrence prediction model. Methods A retrospective cohort of 600 stage I–III endometrial cancer patients (2017–2021) from a single center was analyzed. Preoperative biopsies (blind vs. hysteroscopy-guided) and postoperative specimens underwent ER/PR immunohistochemical testing. Concordance was assessed via Cohen’s kappa. Survival analysis (Kaplan-Meier), ROC curves, and Cox regression identified prognostic factors. A nomogram integrating PR expression dynamics and clinicopathological parameters was developed and validated. Results ER and PR expression showed moderate-to-substantial overall concordance (86.8%, κ = 0.481; 87.1%, κ = 0.676), with hysteroscopy-guided biopsies demonstrating superior agreement (ER: 94.0%, κ = 0.733; PR: 91.4%, κ = 0.742) versus blind biopsies. Combined pre-/postoperative PR expression improved recurrence prediction (AUC = 0.680). The integrated nomogram (AUC = 0.864) effectively stratified high-risk patients (3-year recurrence-free survival: 53.40% vs. 86.05% in non-high-risk), who benefited from adjuvant therapy. Conclusions Hysteroscopy-guided biopsy enhances ER/PR assessment accuracy. The nomogram integrating PR dynamics and clinical parameters enables precise recurrence risk stratification, aiding personalized adjuvant therapy decisions. Trial registration: Not applicable.
Fan et al. (Thu,) studied this question.