Detection fidelity of AR mutations in plasma derived cell-free DNA

// Alexa Goldstein 1, 2, 3 , Patricia Valda Toro 1, 2, 3 , Justin Lee 2, 3 , John L. Silberstein 1, 3 , Mary Nakazawa 1, 3 , Ian Waters 2, 3 , Karen Cravero 2, 3 , David Chu 2, 3 , Rory L. Cochran 2, 3 , Minsoo Kim 2, 3 , Daniel Shinn 2, 3 , Samantha Torquato 1, 2, 3 , Robert M. Hughes 1, 2, 3 , Aparna Pallavajjala 4 , Michael A. Carducci 2, 3 , Channing J. Paller 2, 3 , Samuel R. Denmeade 2, 3 , Bruce Kressel 2, 3 , Bruce J. Trock 1 , Mario A. Eisenberger 1, 2, 3 , Emmanuel S. Antonarakis 1, 2, 3 , Ben H. Park 2, 3, 5 , Paula J. Hurley 1, 2, 3 1 The James Buchanan Brady Urological Institute, Department of Urology, Johns Hopkins School of Medicine, Baltimore, MD, USA 2 The Department of Oncology, Johns Hopkins School of Medicine, Baltimore, MD, USA 3 The Sidney Kimmel Cancer Center, Johns Hopkins School of Medicine, Baltimore, MD, USA 4 The Department of Pathology, Johns Hopkins School of Medicine, Baltimore, MD, USA 5 The Whiting School of Engineering, Department of Chemical and Biomolecular Engineering, Johns Hopkins University, Baltimore, MD, USA Correspondence to: Paula J. Hurley, email: phurley2@jhmi.edu Keywords: next generation sequencing, droplet digital PCR, circulating tumor DNA, androgen receptor, DNA polymerase Received: September 22, 2016 Accepted: December 25, 2016 Published: January 31, 2017 ABSTRACT Somatic genetic alterations including copy number and point mutations in the androgen receptor ( AR) are associated with resistance to therapies targeting the androgen/AR axis in patients with metastatic castration resistant prostate cancer (mCRPC). Due to limitations associated with biopsying metastatic lesions, plasma derived cell-free DNA (cfDNA) is increasingly being used as substrate for genetic testing. AR mutations detected by deep next generation sequencing (NGS) of cfDNA from patients with mCRPC have been reported at allelic fractions ranging from over 25% to below 1%. The lower bound threshold for accurate mutation detection by deep sequencing of cfDNA has not been comprehensively determined and may have locus specific variability. Herein, we used NGS for AR mutation discovery in plasma-derived cfDNA from patients with mCRPC and then used droplet digital polymerase chain reaction (ddPCR) for validation. Our findings show the AR (tTC>cTC) F877L hotspot was prone to false positive mutations during NGS. The rate of error at AR (tTC>cTC) F877L during amplification prior to ddPCR was variable among high fidelity polymerases. These results highlight the importance of validating low-abundant mutations detected by NGS and optimizing and controlling for amplification conditions prior to ddPCR.