GUARIGIONE DAL TUMORE AL SENO: UN TEMA DIBATTUTO

The cure from breast cancer: a discussed subject Cancer survivorship is a worldwide growing area of research. There are currently two main methods to measure cancer survival: relative survival and cancer-specific survival. Only recently a parametric method was proposed to model cumulative incidences with multiple possible causes of death. The method can be used to evaluate the cure fraction and may be an alternative to relative survival. Parametric crude cumulative incidence estimates are given by transformation models with an improper parametric baseline function. Precisely the originally proposed Gompertz distribution was compared with a more flexible 4 parameters Log-Logistic distribution. The goodness of fit of the two different models can be investigated by bootstrap resampling. The 4 parameters Log-Logistic distribution can model non-monotonic hazard functions and may be more appropriate to model specific deaths with respect to the more simple Gompertz distribution, e.g. in the context of diseases with complex dynamics. The two parametric improper distributions are used in this work to estimate the fraction of cured patients with breast cancer. The literature on the estimation of the cured fraction for breast cancer patients has developed during the last decades thanks to the availability of studies with long follow-up time, which was required by the type of pathology in question. Data from a two-group randomized controlled trial were used. From 1973 to 1980, 701 women were recruited with breast cancer measuring no more than 2 cm in diameter. Were randomly assigned 349 women to radical mastectomy and 352 to breast-conserving surgery, followed by radiotherapy to the ipsilateral mammary tissue. Thanks to a reliable classification of the causes of death, it was possible to analyse data using cause-specific mortality. The 4 parameters Log-Logistic distribution appears to fit better breast cancer specific deaths with respect to the simpler Gompertz distribution. The results of the parametric modelling of competing risks are consistent with the results obtained analysing relative survival. Although there is statistical evidence of a cure fraction, in practical terms, it seems that 30 years after breast cancer the cure cannot be claimed yet.