Longitudinal ultrasensitive ctDNA monitoring for high-resolution lung cancer risk prediction
James R. Black, Takahiro Karasaki, Charles W. Abbott, Bailiang Li, Selvaraju Veeriah, Maise Al Bakir, Wing Kin Liu, Ariana Huebner, Carlos Martínez‐Ruiz, Piotr Pawlik, David A. Moore, Daniele Marinelli, Oliver Shutkever, Cian Murphy, Lydia Liu, Charlotte Grieco, Karen Grimes, Fábio C. P. Navarro, Rachel Marty Pyke, Gábor Bartha, Kathleen C. Keough, Steven Dea, Neeraja Ravi, John Lyle, Jason Harris, Katherine D. Brown, Fiona Blackhall, Fatemah Hassani, Dean A. Fennell, Nicholas McGranahan, Jacqui Shaw, Christopher Abbosh, Allan Hackshaw, Mariam Jamal‐Hanjani, Alexander M. Frankell, Sean M. Boyle, Richard O. Chen, Charles Swanton
Abstract
Biomarkers accurately informing prognostic assessment and therapeutic strategy are critical for improving patient outcome in oncology. Here, we apply a whole-genome, tumor-informed circulating tumor DNA (ctDNA) detection approach to address this challenge, leveraging 1,800 variants across 2,994 plasma samples from 431 patients with non-small cell lung cancer (NSCLC) from the TRACERx study. We show that ultrasensitive ctDNA detection below 80 parts per million both pre- and postoperatively is highly prognostic, and combinatorial analysis of the pre- and postoperative ctDNA status identifies an intermediate risk group, improving disease stratification. ctDNA kinetics demonstrate clinical utility during adjuvant therapy, where patients that "clear" ctDNA during adjuvant therapy experience improved outcomes. Moreover, characterization of patterns in postoperative ctDNA kinetics reveals insights into the timing, risk, and anatomical pattern of relapses. By incorporating longitudinal ultrasensitive ctDNA detection, we propose a refined schema for guiding the stratification and treatment recommendations in early stage NSCLC.