Copy number profiles of primary tumors for risk stratification of advanced prostate cancer: A biomarker study embedded in the multicenter STAMPEDE trial.

5021 Background: Men with advanced hormone-sensitive prostate cancer (HSPC) starting long term androgen deprivation therapy (ADT) follow a highly variable clinical course. Treatment intensification with docetaxel or AR targeted therapies improves outcomes but there is a risk of overtreatment, especially in non-metastatic (M0) or metastatic (M1) low volume disease. We established a framework for biomarker evaluation in the STAMPEDE trial. We aimed to evaluate the feasibility and clinical utility of assessing the burden of copy number (CN) aberrations in newly diagnosed advanced HSPC. We hypothesised that increased percentage genome altered (PGA) would associate with higher disease burden and worse prognosis. Methods: We implemented a scalable strategy using low coverage whole genome sequencing (lpWGS) of formalin fixed paraffin embedded (FFPE) diagnostic core biopsies from STAMPEDE participants randomised to the standard of care ADT arm, between 2005 and 2016. Tissue was retrieved from 136 trial sites. 315 cases were randomly selected, aiming for a biomarker population of 300, anticipating an assay failure rate ̃5%. We defined 40% as the minimum histopathologically determined tumor cellularity (TC) for inclusion. We performed a survival analysis investigating PGA at diagnosis as a continuous measure with fractional polynomial specification in Cox models adjusting for disease burden, Gleason grade, pre-ADT PSA (log-transformed), age at randomisation and TC. We pre-specified that all hypothesis tests required evidence at the 5% significance level to consider rejecting the null hypothesis. Results: We successfully CN profiled 300/315 cases. There were no significantly different baseline clinico-pathological features between the full trial comparison n = 3106 and final biomarker population n = 300, 290/300 cases were de novo presentations. PGA in the core with highest Gleason grade and TC was median 18% (range 0%-75%; n = 300). PGA was significantly higher in M1 (n = 169) compared to M0 (n = 131) cases (median: 21% vs 14%; p = 0.00006). 284/300 were subclassified by disease burden into M0 node negative and node positive, and M1 low and high volume. PGA was significantly associated with increased disease burden (p = 0.00002). Increased PGA was significantly and non-linearly associated with an increased hazard of failure-free survival (p = 0.004), progression-free survival (p = 0.002), metastatic progression-free survival (p = 0.003), overall survival (p = 0.045) and prostate cancer-specific survival (p = 0.011). Conclusions: Evaluation of the burden of CN aberrations in archival, poor quality FFPE diagnostic tissue from men randomised in the STAMPEDE trial is feasible using lpWGS and has potential clinical utility to identify better prognosis advanced HSPC patients, who may not require treatment intensification.