Capture-Based Targeted Ultradeep Sequencing in Paired Tissue and Plasma Samples Demonstrates Differential Subclonal ctDNA-Releasing Capability in Advanced Lung Cancer

Abstract

IntroductionCirculating tumor DNA (ctDNA), which represents an unbiased way to assess tumor genetic profile noninvasively, facilitates studying intratumor heterogeneity. Although intratumor heterogeneity has been elucidated substantially in a few cancer types, including NSCLC, how it influences the ability of tumor cells harboring different genetic abnormalities in releasing their DNA remains elusive. We designed a capture-based panel targeting NSCLC to detect and quantify genetic alterations from plasma samples by using deep sequencing. By applying the panel to paired biopsy and plasma samples, we imputed and compared the ctDNA-releasing efficiency in subclones harboring distinct genetic variants. MethodsWe collected 40 pairs of matched biopsy and plasma samples from patients with advanced lung cancer and applied capture-based sequencing using our LungPlasma panel, which consists of critical exons and introns of 168 genes. We derived a normalized relative allelic fraction score (NRAFS) to reflect ctDNA-releasing efficiency. ResultsBy using mutations detected in biopsy samples as a reference, we achieved 87.2% by-variant sensitivity, including for single-nucleotide variants, insertions or deletions, and gene fusions. Furthermore, the by-variant sensitivity for the seven most critical and actionable genes was 96.2%. The average NRAFS for subclones carrying mutations from seven actionable genes was 0.877; in contrast, the average NRAFS for other mutations was 0.658. Mutations from four genes involved in cell cycle pathways had a particularly low NRAFS (0.480) compared with the other two groups (p = 0.07). ConclusionsWe have demonstrated that subclones carrying driver mutations are more prone to release DNA. We have also demonstrated the quantitative ability of capture-based sequencing, paving its way for routine utilization in clinical settings.