ER and PR expression and survival after endometrial cancer.

Smith, Deborah; Stewart, Colin J R; Clarke, Edward M; Lose, Felicity; Davies, Claire; Armes, Jane; Obermair, Andreas; Brennan, Donal; Webb, Penelope M; Nagle, Christina M; Spurdle, Amanda B

OBJECTIVE To measure association between endometrial carcinoma ER and PR status and endometrial cancer (EC) survival, accounting for inter-observer variation. METHODS The intensity and proportion of tumor cell expression of ER and PR in ECs were assessed independently and semi-quantitatively by two pathologists using digital images of duplicate tumor tissue microarrays (TMAs). Cases with inconsistent initial assessment were reviewed and final scoring agreed. The association between overall and EC-specific survival and hormone receptor expression (intensity, proportion and combined) was assessed using Cox regression analysis. The C-index was used to evaluate model discrimination with addition of ER and PR status. RESULTS Tumor ER and PR analysis was possible in 659 TMAs from 255 patients, and in 459 TMAs from 243 patients, respectively. Initial ER and PR scoring was consistent in 82% and 80% of cases, respectively. In multivariate analyses decreased ER and PR expression was associated with increased tumor-related mortality. Associations reached statistical significance for ER proportion score (P=0.05), ER intensity score (P=0.003), and PR combined score (P=0.04). Decreased expression of combined ER/PR expression was associated with poorer EC-specific survival than decreased expression of either hormone receptor alone (P=0.005). However, hormone receptor status did not significantly improve mortality prediction in individual cases. CONCLUSION ER and PR expression combined, using cut-points that capture variation in scoring and across cores, is significantly associated with EC-specific survival in analyses adjusting for known prognostic factors. However, at the individual level, ER and PR expression does not improve mortality prediction.