Background: TITAN, a phase 3 PBO-controlled study in patients (pts) with mCSPC, showed APA + ADT improved radiographic progression-free survival and OS vs PBO + ADT. In this exploratory analysis, we report the relationships between biomarkers and OS in TITAN. Methods: Circulating tumor (ct)DNA and genomic aberrations in 17 PC-related genes, including androgen receptor (AR), were assessed at baseline (BL; 114 pts) and end of study treatment (EOST; 129 pts) using next-generation sequencing. ctDNA was assessed qualitatively; genomic aberrations were assessed within ctDNA-positive samples as inactivation (heterozygous/homozygous deletion or single nucleotide variant [SNV]) or activation (amplification or SNV). Associations of detected ctDNA/aberrations at BL or EOST with OS, and biomarkers at EOST with OS on subsequent therapies, were evaluated using univariate or multivariate analyses and Cox proportional hazards model; results were stratified by treatment arm. Results: Among pts from both treatment groups, 36% had detectable ctDNA, of which 27% had any genomic AR aberration and 24% had AR gene amplification at BL; prevalence of these biomarkers increased significantly from BL to EOST (Table). The most prevalent non-AR aberrations at BL (TP53, homologous recombination repair [HRR] pathway, PTEN, RB1, and PIK3CA genes) increased at EOST but not significantly (Table). Among assessed aberrations, presence of ctDNA or any AR genomic aberrations at BL (HR, 1.9 or 6.7; all p < 0.05) and any AR genomic aberrations or PI3K pathway activation at EOST (1.7, p < 0.05 or 2.2, p < 0.001) was significantly associated with poor OS in multivariate analyses from both treatment groups. In univariate analyses of pts who received subsequent therapy (chemotherapy: 106; hormonal therapy: 161), worse OS was associated only with PIK3CA activation, PI3K pathway activation, or TP53 inactivation at EOST (3.7, p < 0.05; 2.4, p < 0.05; 3.0, p < 0.01, respectively) in chemo-treated pts. A small sample size in some biomarker subgroups limits interpretation. Conclusions: These hypothesis-generating data from TITAN show that presence of ctDNA or select AR and non-AR biomarkers at BL or EOST were associated with poor OS. The predictive value of these biomarkers for survival in mCSPC needs further confirmation. Clinical trial information: NCT02489318.

^aInactivation. ^bActivation. ^cIncludes: BRCA1, BRCA2, FANCA, BRIP1, HDAC2, CDK12, ATM, PALB2, and CHEK2.^dEOST vs BL.