Concordance evaluation of plasma whole-exome sequencing (pWES) and tissue whole-exome sequencing (tWES): A pilot study in non-small cell lung cancer (NSCLC)

Background: Circulating tumor DNA (ctDNA) analysis by pWES offers the possibility for comprehensive characterization of tumor-derived alterations and longitudinal monitoring of tumor molecular evolution in a minimally invasive manner. Here, we conducted a pilot study to evaluate the analytical performance of a pWES assay with boosted content compared with tWES in NSCLC. Methods: Cell-free DNA isolated from commercially-acquired plasma samples of patients with stage I-IV NSCLC were analyzed using the PredicineWES+ assay (WES plus a boosted 600-gene panel; Huidu Shanghai Medical Sciences Ltd, China). The copy number burden score (cnbScore; a quantitative measure of genome-wide copy number abnormality as a proxy for tumor fraction [TF]) was derived from low-pass plasma whole-genome sequencing (pWGS). pWES TF was based on the maximum mutation allele fraction of nonsynonymous somatic mutations (single nucleotide variants [SNVs] + indels). Jaccard Index (JI), Spearman correlation (ρ), and Wilcoxon rank-sum test were applied to quantify concordance and differences across measures between pWES and tWES. Results: 39/40 samples were evaluated by low-pass pWGS (stage II-III, 80%). A high TF (cnbScore ≥5.6) was associated with late-stage tumors and squamous cell carcinoma (SCC) histology. Thirty samples (25 highest and 5 lowest cnbScore samples) were selected for pWES. pWES TF was significantly correlated with pWGS cnbScore (ρ = 0.64; P < 0.001). Of 30 samples evaluated by pWES, 14 samples were ctDNA-positive (cnbScore ≥5.6) and were included in concordance analyses with tWES. Adenocarcinoma- and SCC-specific driver mutations were highly concordant between pWES and matched tWES (JI = 0.72). NSCLC-dominant smoking-associated mutational signature was identifiable by pWES and was correlated with tWES (ρ = 0.58). Considerable concordance of mutation calling was observed between pWES and tWES (median JI = 0.47). Blood tumor mutational burden (TMB) was highly correlated with tissue TMB (ρ = 0.69). Compared with nonboosted exome region, boosted region with higher sequencing depth had numerically better TMB correlation (ρ = 0.81 vs 0.64) and SNV concordance (median JI = 0.80 vs 0.44) between pWES and tWES. Disconcordant copy number variant results were observed between pWES and tWES, possibly due to low ctDNA TF and copy number gain in tissue. Conclusions: ctDNA tumor burden metrics by pWES correlated with low-pass pWGS and showed association with tumor stage and histologic subtype of NSCLC. pWES feasibly characterized TMB and SNV in NSCLC especially in ctDNA-positive samples and in the boosted 600-gene WES panel, indicating potential utility for longitudinal tumor profiling and monitoring during treatment.