An Evidence‐Based Analysis of the Management of Localized Prostate Cancer

Abstract

PURPOSEPatients with localized prostate cancer and their doctors face complex trade-offs when deciding on treatment. In this study, we modify the “number needed to treat” method to compare radical prostatectomy with radiotherapy. METHODSA MEDLINE search was performed to identify all studies of radical protatectomy or radiotherapy for prostate cancer. Number needed to treat formulas were modified to account for not only survival but also for complications and their utilities. RESULTSThe unadjusted number needed to treat value for overall survival was 6 favoring prostatectomy (six patients have to undergo prostatectomy to have one more 10-year survivor than if they had undergone radiotherapy). Radiotherapy patients were 4 years older than prostatectomy patients. Because overall survival is strongly linked to patients' age and overall health, the numbers needed to treat for disease-specific and distant metastasis-free survival were analyzed to minimize patient selection bias. The unadjusted number needed to treat values for disease-specific and distant metastasis-free survival were 14 and 18, respectively. When number needed to treat is adjusted for complications and utilities, its value for overall survival is 14, disease-specific survival is -25, and distant metastasis free survival is -22, these last two favoring radiotherapy. CONCLUSIONSUtility-adjusted numbers needed to treat for prostatectomy and radiotherapy are greatly influenced by the likelihood of complications and the utility loss ascribed to them. When literature-reported values are used, radiotherapy is superior, but the differences in outcomes are small. With prostate-specific antigen screening and refined treatment methods, these values will change, and the modified number needed to treat can be used to evaluate, report and compare results. The consequences of treatment in terms of both survival and quality of life determine patient choice and physician recommendations.