Distinguishing prognostic and treatment-predictive information for localized prostate cancer.

Overview

OBJECTIVES: To distinguish the concepts of prognostic and treatment-predictive information for localized prostate cancer. METHODS: We defined a prognostic factor as one that identifies subgroups associated with differing outcomes in untreated patients. A treatment-predictive factor identifies patients with differing outcomes as a consequence of treatment and is best identified in a large, randomized trial. Outside of such a trial, a treatment-predictive factor can be identified in prognostic subgroups or after adjustment for prognostic factors. RESULTS: The distinctions between prognostic and treatment-predictive factors are illustrated by hypothetical examples. CONCLUSIONS: The practical implication of the distinctions is that prognostic information may not provide reliable treatment-predictive information, that is, additional information may be needed before selection of patients for different treatments can be based on prognostic information. Determination of the relative treatment effect in any prognostic subgroup of patients requires a comparative setting. Until now, identified prognostic factors for localized prostate cancer at best can give guidance for clinical decisions on which patients should not be offered local aggressive therapy if the aim of the therapy is to cure the patient of the disease.