Statistical modelling and prognostic analysis in primary breast carcinoma

Abstract

This work reports, for the first time, the results of the first retrospective study on breast cancer in Cyprus. The study which is based on 197 women with primary breast cancer who have undergone modified radical mastectomy in the 5-year-period 1988-1990 in Cyprus, has been performed in order to assess the prognostic importance of some histological and biological variables for the survival time in breast cancer. Note that the relative physical isolation of the island of Cyprus and the social and cultural traditions among the various ethnic groups in Cyprus make such studies uniquely interesting and useful. A number of X2-homogeneity tests were performed to investigate the effects of various prognostic factors and the survival and hazard functions of the population were estimated by Kaplan - Meier estimators. Further analysis of the survival data was performed based on fully parametric models such as the Weibull model and also on the Cox proportional hazards model. In addition to the traditional morphological factors of tumor size, histological grade, lymph node status and the oestrogen and progesterone receptors which are considered significant prognostic indicators in breast cancer, other tumor biological markers such as car B-2 oncogene, p53 protein, and cathepsin D have shown a strong potential of providing additional important prognostic information.