Table 3.

Final primary MVA of OS in the subset of PSA-matched African American and Caucasian men with mCRPC treated with sipuleucel-T in PROCEED (n = 657).

Baseline covariate	HR (95% CI)	P valuea
Race: African American versus Caucasian	0.60 (0.48–0.74)	<0.001
Age: >median versus ≤ median	1.26 (1.03–1.53)	0.023
Body weight: >median versus ≤ median	0.97 (0.80–1.17)	0.723
ECOG PS: >0 versus 0	1.33 (1.09–1.61)	0.005
Baseline PSA: >median versus ≤ median	1.74 (1.43–2.12)	<0.001
Baseline ALP: >median versus ≤ median	1.59 (1.27–1.99)	<0.001
Baseline hemoglobin: >median versus ≤ median	0.66 (0.54–0.81)	<0.001
Lymph node only metastases: yes versus no	0.66 (0.49–0.90)	0.009
Prior prostatectomy: yes versus no	0.82 (0.68–1.01)	0.058
Prior abiraterone/enzalutamide: yes versus no	1.64 (1.19–2.25)	0.002
Prior docetaxel/cabazitaxel: yes versus no	1.42 (1.11–1.81)	0.005

ALP alkaline phosphatase, CI confidence interval, ECOG PS Eastern Cooperative Oncology Group performance status, HR hazard ratio, MVA multivariable analysis, OS overall survival, PSA prostate-specific antigen.

^aVariables that were statistically significant on univariable analyses were assessed for clinical relevance and included in the final primary multivariable model if deemed both statistically and clinically significant. For multivariable Cox modeling, the Markov chain Monte Carlo imputation method was used for imputing missing data. Parameters with missing data include ECOG PS, ALP, hemoglobin, weight, prior local therapy, and lymph node only metastases. An MVA of the entire PROCEED population was not undertaken due to a major imbalance between the numbers of African American and Caucasian patients, with concern for an underpowered analysis.