Predictive modelling in hormone-refractory prostate cancer (HRPC).

Because the evidence is not yet solid enough to strongly recommend whether or not to treat hormone-refractory prostate cancer (HRPC) patients at certain stages of the disease, predictive models might help in decision making. The importance of prognostic models lies in their ability to capture clinically relevant and measurable variables for routine use by clinicians to inform patients, and improve palliation and treatment decisions. Basically this allows for the creation of homogeneous prognostic strata for randomised comparative trials of therapeutic agents. In the last few years different models to predict patient outcome in HRPC have been published in the literature. Recently, based on the phase III randomised trial of docetaxel, a multivariate prognostic model incorporating PSA kinetics has been developed to predict survival at 1, 2 and 5 years in metastatic HRPC men treated with chemotherapy. This novel model includes new independent clinical prognostic factors in addition to PSA-DT such as baseline pain, type of progression at baseline (measurable disease or bone scan compared with PSA only), presence of liver metastases and the number of metastatic disease sites. This nomogram will be a helpful tool to stratify patients for further docetaxel-based trials and could also help us to delineate the potential benefits of chemotherapy at certain points during the natural history of HRPC.