Abstract LB-270: Discovery of genomic correlates and tumor purity as an independent clinical factor of poor outcome in advanced prostate cancer lpWGS CNA data

Abstract

Abstract Background: We, and others, have shown that responses to therapies in metastatic castration-resistant prostate cancer (mCRPC) can be monitored using plasma cell-free DNA (cfDNA), and that this can be both quantitative and qualitative. In this study, we explored the utility of low pass whole genome sequencing (lpWGS) of cfDNA in mCRPC patients treated on two large prospectively collected Phase III trials of taxane chemotherapy, FIRSTANA (NCT01308567) and PROSELICA (NCT01308580). Methods: Plasma samples collated from the FIRSTANA and PROSELICA trials were evaluated. cfDNA was isolated and analysed with lpWGS using the Illumina NovaSeq 6000. Reads were aligned using BWA-MEM and quality controlled using Picard and FASTQC. Whole-genome copy number aberration (CNA) profiles were generated using ichorCNA, which also estimated tumor purity and ploidy. Log-rank tests were used for univariable survival comparisons, and Cox proportional hazard models were used for multivariable survival analysis. Results: Overall, lp-WGS data was generated from 528 samples acquired at three different timepoints (Baseline [SCR and C1], C4 and EOS), representing 188 unique patients. We observed CNA profiles in line with emergent mCRPC subtypes including highly aberrant, large-scale events which have been linked to homologous-repair deficient tumors. We found that CNAs in this cohort matched other published studies with frequent amplifications of AR (~30%), copy-gains of PI3KCA (~45%) and copy-loss of NKX3-1 (~75%), RB1 (~70%) and PTEN (~50%). We did not observe changes in baseline tumor purity or ploidy between the FIRSTANA and PROSELICA cohorts. Following calculation of a large-scale transition (LST) score (measuring CNA breakpoints), we observed that this was significantly elevated (Wilcox test, p=0.0073) in cases that had prior exposure to second line androgen deprivation (abiraterone or enzalutamide). Baseline tumor purity values were associated with several mCRPC clinical variables including ECOG performance status. High lpWGS tumor purity values were univariably associated with poorer overall survival in both FIRSTANA and PROSELICA cohorts (log-rank test p=0.0021 and 0.0009), and were also associated with significantly poorer RPFS and PSAPFS. Tumor purity was also associated (HR 1.66, p=0.036) with survival when assessed using multivariable models alongside other mCRPC clinical variables, including cell-free DNA concentration. Further, longitudinal changes in tumor purity on treatment can be illustrative of drug responses. Conclusions: In our analysis of this prospectively collected clinical trial data, we show mCRPC CNA profiles in cfDNA, and that these CNAs may be induced by prior aggressive treatments. We also found that tumor purity estimation can serve as a robust, competitive indicator of both overall survival and, assessed longitudinally, shows clear association with drug response. We envision that these techniques will enable improved strategies for monitoring mCRPC patients. Citation Format: George Seed, Semini Sumanasuriya, Claudia Bertan, Harry Parr, Gemma Fowler, Rossitza Christova, Lorna Pope, Jane Goodall, Maryou Lambros, Pasquale Rescigno, Penelope Flohr, Suzanne Carreira, Wei Yuan, Mustapha Chadjaa, Sandrine Mace, Johann De Bono. Discovery of genomic correlates and tumor purity as an independent clinical factor of poor outcome in advanced prostate cancer lpWGS CNA data [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr LB-270.