Temporary dependency of steroid-receptor prognostic value in breast cancer

BACKGROUND: The influence of prognostic factors in breast cancer may change during follow-up. Our aim was therefore to look for time-related changes in hazard ratios for steroid receptor status and the risk of relapse. PATIENTS AND METHOD: We studied 455 patients with operable primary breast cancer during a mean follow-up period of 72 months (range, 42-130 months). Clinical and histological characteristics were assessed and estrogen receptor (ER) and progesterone receptor (PgR) status were determined and compared to disease-free survival (DFS) and overall survival (OS). Analysis of DFS included fitting a multivariate Cox proportional hazards model, testing for non-proportionality, and examining diagnostic plots. RESULTS: A total of 66.8% of the tumors were ER positive and 54.7% PgR positive. Receptor status is highly associated with menopausal status, histological grade, peritumoral lymphatic or blood vessel invasion and mitotic index. No correlation was found between steroid receptor status and tumor size, node status or tumor stage. In univariate analysis, ER-/PR- vs ER+/PR+ display a hazard ratio of 2.15 (IC del 95%, 1.59-2.99) for DFS (p = 0.001) and 1.95 (IC del 95%, 1.38-2.59) for OS (p = 0.0043). In multivariate analysis, steroid receptor status, node status and mitotic index were independent prognostic factors for DFS and OS. By using Cox regression time-dependent covariates model, we show that the steroid receptor status hazard ratio is 5.6 at diagnosis, decreases after 4 years, and loses its significance after 10 years of follow-up. CONCLUSION: In patients with breast cancer the status steroid receptor have a relatively limited prognostic value and is lost within the first years.