Predictive models before and after radical prostatectomy

Abstract

Context: In the last 10 years, several user-friendly predictive tools have been developed to help clinicians in decision-making process before and after radical prostatectomy.

Objective: To review the most known and used predictive models in pre-operative and post-operative setting.

Evidence acquisition: A structured, comprehensive literature review was performed using data retrieved from recent review articles, original articles, and abstracts. Used keywords were predictive models, nomograms, look-up tables, classification and regression-tree analysis, artificial neural networks, and radical prostatectomy.

Evidence synthesis: A great amount of predictive models has been provided in oncology setting: nomograms, look-up tables, classification and regression-tree analysis, propensity scores, risk-group stratification models, and artificial neural networks. Pre-surgery predictive tools offer the opportunity of getting the most evidence-based and individualized selection of available treatment alternatives. Post-operative predictive models usually provide higher accuracy relative to the pre-surgery models.

Conclusions: Decisions and treatment should be tailored to each individual patient and to the specific characteristics of patients. A number of available predictive models represent a tool to provide accurate prediction of cancer natural history and to improve patients' care.