Identification of prognostic and predictive biomarkers of overall survival (OS) and progression-free survival (PFS) in patients (pts) with metastatic castration-resistant prostate cancer (mCRPC) treated with docetaxel, prednisone (DP) +/- bevacizumab (B) in CALGB 90401 (Alliance).
Halabi, S; Yang, Q; Starr, MD; Brady, JC; Armstrong, AJ; George, DJ; Kelly, WK; Beltran, H; Morris, MJ; Nixon, AB
Published in: Journal of Clinical Oncology
154 Background: CALGB 90401 was a phase III trial of 1050 pts with mCRPC comparing DP+B versus DP alone. While this trial did not show an improvement in OS in the overall population, there were improved PSA response, objective responses and delays in progression suggesting that subsets of men may derive benefit from this combination The purpose of this analysis was to identify and validate plasma angiokines (PAs) that are prognostic or predictive of OS and PFS benefit from bevacizumab in men with mCRPC. Methods: Baseline EDTA plasma samples from 679 consenting pts were analyzed using an optimized multiplex ELISA platform for 25 PAs related to tumor growth, angiogenesis, and inflammation. The data were randomly split into training (n = 462) and testing (n = 217) sets. The proportional hazards model was utilized to test for the prognostic and predictive importance of the PAs in predicting OS and PFS, with and without adjustment for clinical risk score. Analyses were adjusted for multiplicity using false discovery rate (FDR). Results: For the prognostic analysis, 14 PAs (angiopoeitin-2, BMP9, Chromogranin A, HER3, HGF, ICAM-1, IL6, OPN, PIGF, TIMP, TSP2, VEGFA, VEGFR1, and VEGF-R3) were prognostic of OS and 8 PAs (angiopoietin-2, HER3, ICAM-1, IL6, OPN, TIMP, VEGFA and VEGF-R3) were prognostic of PFS in the training set (FDR < 0.05). None of the PAs were statistically significant for OS or PFS when adjusting by the clinical risk score, suggesting that angiokine levels associate with clinical prognostic factors. OPN was predictive of OS in the training set but no other PAs were found to be predictive of PFS improvement with DP+B. Using the testing set, we were unable to validate that OPN is predictive of OS/PFS or any of the PAs are predictive biomarkers of the OS or PFS benefits of DP+B over DP alone in men with mCRPC. Conclusions: While PAs are prognostic for OS and PFS in univariate analysis, we were unable to validate the results in the testing set. Furthermore, we did not identify any PAs that are predictive of benefit from the addition of bevacizumab to docetaxel/prednisone with ADT in this setting. Nevertheless, these remain worthy of further evaluation as potential therapeutic targets.
