Use of prostate-specific antigen progression (PSA-P) to predict overall survival (OS) in patients (pts) with metastatic prostate cancer (PC): Data from S9346 and S9916

Abstract

5015 Background: PSA decline has been used as an efficacy endpoint in PC trials. PSA-P is an accepted indicator of disease progression in hormone sensitive (HS) and hormone refractory (HR) PC and may play an increasing role as an endpoint. We evaluated different definitions of PSA-P as predictors of OS including Consensus Criteria (CC) (Bubley JCO ’99, Scher ASCO ’07) using data from two phase III trials; S9346 (HS) and S9916 ( HR) PC pts. Methods: 1,015 pts from S9346 treated with continuous androgen deprivation and 573 pts from S9916 treated with docetaxel + prednisone or mitoxantrone + prednisone had adequate PSA and follow-up data for this analysis. SWOG-DSMC approved PSA analysis from S9346 (ongoing trial). PSA-P was defined in CC ’99 as an increase in PSA value of ≥ 25% and an absolute increase of ≥ 5 ng/mL over baseline or nadir; the CC’07 definition is similar except absolute PSA increase is ≥ 2 ng/mL. A time varying approach was used to analyze the association between PSA-P at any time and OS, and a landmark analysis used early PSA-P status at 7 months (ms) for S9346 or 3 ms for S9916 to predict subsequent OS. Tests for modification of this association by whether the pt had a PSA nadir on study were also performed. Results: In the time-varying analysis, the CC ’99 and CC ’07 were both strongly predictive of OS (table). Tests for interaction between PSA-P and PSA nadir were not significant for either definition. In the landmark analyses of pts on S9346 using the CC’99, median OS was 10 ms vs 43 ms in pts who did or did not have PSA-P by 7 ms; in S9916, it was 10 ms vs 18 mos for pts who did or did not have PSA-P by 3 mos. Conclusions: PSA-P defined as an increase of ≥ 25% over baseline or nadir with an absolute increase of at least 2–5 ng/mL is a strong predictor of OS in metastatic HS and HR-PC irrespective of whether pts achieved a PSA nadir or not. PSA-P may be a suitable endpoint for phase II studies in these settings. PSA Progression Definition * S9346, HSPC (N=1015) S9916, HRPC (N=573) Events (%) HR (95% CI) ** Events (%) HR (95% CI) ** Time-Varying Any Increase 629 (62%) 1.82 (1.53, 2.17)+ 441 (77%) 1.20 (097, 1.48) Increase by ≥ 25%, ≥ 2 ng/mL 492 (48%) 2.37 (2.00, 2.80)+ 399 (70%) 1.38 (1.13, 1.67) Increase by ≥ 25%, ≥ 5 ng/mL 414 (41%) 2.31 (1.95, 2.73)+ 380 (66%) 1.40 (1.16, 1.69)+ Landmark *** Increase by ≥ 25%, ≥ 2 ng/mL 196 (19%) 4.53 (3.73, 5.51)+ 163 (28%) 2.05 (1.68, 2.50)+ Increase by ≥ 25%, ≥ 5 ng/mL 156 (15%) 4.45 (3.62, 5.46)+ 152 (27%) 2.10 (1.71, 2.57)+ * For all definitions, PSA-P must have been measured ≥ 7 days post-nadir or baseline and confirmed ≥ 7 days later. ** Hazard ratios from multivariate Cox regression models. For S9346, models were adjusted for age, race, PSA, bone pain, Gleason score, pt site (Europe vs US/Canada), and prior AD. For S9916, models were adjusted for treatment arm, age, race, PSA, performance status, bone pain, prior prostatectomy, and type of progression at study entry (PSA only vs measurable/evaluable disease). + p <.05 *** For landmark analyses, the number of pts who had PSA-P in the first 7 m (S9346) or 3 m (S9916) is reported. Hazard ratios give the increased risk of subsequent death among pts who had PSA-P by the landmark time vs those who did not. No significant financial relationships to disclose.