Prognostic factors for survival in the phase III TROPIC trial.
102 Background: In TROPIC ( NCT00417079 ) 755 men were randomized (378 cabazitaxel/prednisone [CbzP]; 377 mitoxantrone/prednisone [MP]). Treatment arms were well balanced; ECOG PS 0–1 (93% CbzP vs 91% MP), measurable disease (53% vs 54%), baseline pain (46% vs 45%) and ≤ 6 months from last dose of docetaxel (D) to randomization (62% vs 72%). CbzP significantly improved overall survival (OS) in mCRPC pts who progressed on or after D treatment compared with MP (HR 0.70; CI 0.59–0.83; P < 0.0001). We investigated overall prognosis and performed a multivariate analysis of factors implicated in OS from this robust dataset. Methods: A univariate analysis of a variety of factors followed by a multivariate analysis of all factors was conducted. Interactions with treatment arms were explored. Cox proportional hazard models were used to examine the effect of treatment and prognostic factors on OS. Results: In addition to the significant effect of treatment received, the univariate analysis identified ECOG PS and measurable disease at baseline, time from last dose of D to randomization, time of progression after last D treatment and pain scores at baseline as significant prognostic factors for OS. Interactions of each of these factors with the treatment were not statistically distinct, suggesting that CbzP survival benefit was consistent among the subgroups defined by these factors. After adjustments for all prognostic factors, multivariate analysis identified ECOG PS 2, measurable disease, time of last dose of D to randomization (≤ 6 months vs > 6 months) and presence of baseline pain as statistically significant prognostic factors. Following adjustments, the treatment effect on survival (CbzP vs MP) remained statistically significant (Table). Conclusions: ECOG PS, measurable disease at baseline, time from last D dose to randomization, baseline pain and CbzP treatment predicted OS in patients in the TROPIC study in a multivariate analysis. [Table: see text]