Two distinct local relapse subtypes in invasive breast cancer: effect on their prognostic impact

PURPOSE: Local relapse (LR) remains an important concern in breast cancer surveillance. Given the unique opportunity to study local recurrences in the breast in the absence of irradiation, we tested the possibility of identifying a particular tumor feature that might account for LR. EXPERIMENTAL DESIGN: Archival specimens from 235 patients belonging to the control arm (no radiotherapy) of the Milan 3 Trial were retrieved and included in this study. H&E-stained histologic slides were reviewed for diagnostic reassessment. A panel of biological variables was assessed by immunohistochemical/cytochemical staining. RESULTS: Onset of LR was significantly linked only to patient's age, with a hazard ratio (HR) to relapse reduced by 60% in patients >50 years of age (P = 0.002). Univariate and multivariate Cox analyses in women <or=50 years of age indicated that only expression of hyaluronan in tumor cells was associated with LR risk (HR, 2.48; P = 0.0364 and HR, 2.34; P = 0.0580, respectively). In patients >50 years of age, lymph node and estrogen receptor status were significantly predictive of LR (HR, 3.34; P = 0.0113 and HR, 0.39; P = 0.0424, respectively). Multivariate analysis indicated that apart from lymph node positivity (HR, 3.48; P = 0.0120), none of the variables associated with LR in univariate analysis displayed an independent LR-predictive power. LR had a strong prognostic impact on distant metastasis in patients >50 years of age, whereas in younger women, LR did not affect the risk of metastasis. CONCLUSIONS: These results support the notion that LR arises from two distinct mechanisms, one that is more frequent in young patients, associated with host characteristics, and is not linked with prognosis, and another that is less frequent, but is associated with tumor aggressiveness which represents a peculiar and distinct marker of breast tumor malignancy.