Predictive value of different prognostic factors in breast cancer recurrences: multivariate analysis using a logistic regression model.

The aim of this study was to compare the sensitivity of different pre-operative parameters in patients with breast cancer (BC) recurrence using univariate and multivariate analysis. We retrospectively analyzed a series of 387 women (median age 60 years, range 35-83 years) who underwent curative surgery for pT1-2 BC. The patients were divided into two groups: Group 1: 325 (84.0%) patients with no evidence of disease during a median follow-up of 53 months (range 25-149 months) and Group 2: 62 (16.0%) patients who developed local or distant recurrences. Univariate analysis showed significant (p<0.01) differences between the two Groups in age, size and grading of the tumor and hormone receptor rate. MIB1 proliferation rate, serum markers CEA and CA 15-3, and lymph node status were not useful in predicting relapse. Multivariate analysis using a logistic regression model showed that only age, size of the tumor and hormone receptor rate independently correlate with the onset of recurrences. in conclusion, there is no clear correlation between BC recurrence and the majority of the prognostic factors available. Multivariate analysis of several pre-operative parameters may help to correctly select the high risk population.