Background: Circulating bone biomarkers (BB) are strongly prognostic for OS in castration-resistant PC (CRPC). We prospectively evaluated BB in men with HSPC in S1216, a trial that established new OS benchmarks. We sought to identify patient (pt) subsets with differential OS outcomes as defined by BB. Methods: Markers of bone resorption [CTx;PYD] & formation [CICP;BAP] were assessed. Pts were randomly divided into training (1/3) & validation (2/3) sets. In the training set, recursive partitioning of OS was used to identify the ideal dichotomous cutpoint for each BB & for a combination of biomarker split points to define prognostic groups. In the validation set, Cox PH models were used to assess impact of BB on OS, adjusted for pt & tumor characteristics. Adjusted odds ratios for 3-year OS based on BB & baseline clinical factors were developed using logistic regression to estimate receiver operating characteristic (ROC) curves. Results: Of 1,279 men, 949 had baseline BB. Median age–68y; median PSA-28 ng/dL; Gleason>7: 60%; Zubrod PS 0/1-97%. Values of BB at the median & at cutpoints maximized for OS were identified. For 3 of the BB, the cutpoint was at the ̃85^th %ile; for PYD it was at the median. Recursive partitioning algorithms applied to the training set identified 4 groups with differential OS based on a dichotomous split of CTx in combination with additional CICP splits within each group. Hazard ratios (HR) for OS based on elevated BBs are shown. ROC analysis showed that only BAP & PYD had significantly higher AUC(0.73;0.74) compared to AUC of baseline clinical factors(0.71) w/ p=0.02 and 0.03 respectively. There was no evidence of BB x treatment interaction (all p>=0.2). Conclusions: In men initiating ADT for HSPC, elevated BB are strongly prognostic for worse OS. BB levels alone & in combination with pt/tumor characteristics identify unique subsets of men with high probability of being alive at 3 years from ADT initiation. These results validate the clinical value of BB in the HSPC state, extending BB utility beyond CRPC. Clinical trial information: NCT01809691.

*Adjusted for treatment arm, disease extent, Zubrod PS, PSA, Gleason Score, age, Afr-Amer (Y/N), and visceral mets status; ^p-values from Wald Chi-Square test for Type 3 analysis of effects; ** from logistic model.