Poster Presentation 31st Lorne Cancer Conference 2019

Prognostic and predictive utility of copy number variations (CNVs) in circulating tumor DNA (ctDNA) from metastatic castration-resistant prostate cancer (mCRPC) patients (#175)

Heidi Fettke 1 , Edmond Kwan 1 2 , Jianjun Yu 3 , Amy Wang 3 , Carlos Montesinos 3 , Calvin Wong 3 , Xue Gong 3 , Tiantian Zheng 3 , Peter Pan Du 3 , Shidong Jia 3 , Andrew Mant 4 , Phillip Parente 4 5 , Carmel Pezaro 4 5 6 , Arun Azad 1 2 7
  1. School of Clinical Sciences, Monash University, Melbourne, VIC, Australia
  2. Medical Oncology, Monash Health, Melbourne, VIC, Australia
  3. Predicine Inc., Hayward, California, USA
  4. Medical Oncology, Eastern Health, Melbourne, VIC, Australia
  5. Eastern Health Clinical School, Monash University, Melbourne, VIC, Australia
  6. Sheffield Teaching Hospitals, NHS Foundation Trust, Sheffield, England
  7. Medical Oncology, Peter MacCallum Cancer Centre, Melbourne, VIC, Australia

Background: Plasma circulating tumor DNA (ctDNA) has emerged as a promising minimally-invasive tool with which to interrogate the cancer genome. However, detection of copy number variations (CNVs) in ctDNA has proved challenging. This poses a major problem in mCRPC, which commonly harbours pathogenic CNVs. Here we use an ultra-sensitive next-generation sequencing (NGS) assay to identify CNVs in ctDNA from mCRPC patients and correlate findings with clinical outcomes in men commencing ARSI (abiraterone or enzalutamide) or chemotherapy (docetaxel or cabazitaxel).

 

Methods: mCRPC patients commencing ARSI or chemotherapy were prospectively recruited at two Australian centres. Plasma was collected and platelet poor plasma (PPP) fractions were processed uniformly and cell-free DNA (cfDNA) extracted. Plasma samples were analysed using the PredicineLITE NGS assay, which reports genomic alterations in 90 cancer genes. CNVs from this cohort were correlated with PSA response rate (Fisher’s exact test), PSA progression-free survival (PSA-PFS), clinical/radiographic progression-free survival (clinical/rPFS), and overall survival (OS).

 

Results: Median follow-up was 19.85 months (mo) (IQR 12.5–23.0). In total, 32 pre-treatment samples were analyzed (7 chemotherapy, 25 ARSI). The most common CNVs were PTEN loss (n=12, 38% of cohort), RB1 loss (n=5, 16%) and AR gain (n=12, 38%).  Notably, OS was decreased in patients with PTEN loss (median 9.7 mo vs. not reached; p=0.03) and RB1 loss (median 7.1 mo vs. 17.1 mo; p=0.1), while PSA response rates were also lower in RB1 loss (1/5, 20% vs. 19/27, 71%; p=0.053). In addition, AR copy number gain was associated with decreased clinical/rPFS (median 3.4 mo vs. 10.7 mo; p=0.05) and inferior OS (median 9.7 mo vs. not reached; p=0.05).

 

Conclusion: Using an ultra-sensitive NGS assay, we demonstrate the robust detection of CNVs in plasma ctDNA of patients with mCRPC. Copy number losses of PTEN and RB1, and gains of AR were associated with worse clinical outcomes on chemotherapy and ARSI. Validation is ongoing.