Comparison and performance of three plasma next-generation sequencing assays versus tissue testing for the detection of actionable driver mutations in advanced non-squamous non-small cell lung carcinoma.

e20018 Background: Liquid biopsy (LB) is a useful tool for the detection of actionable mutations in advanced non-squamous non-small cell lung cancer (NS-NSCLC) at progression. However, the clinical benefits of integrating both LB and tissue biopsy (TB)-based genomic profiling at diagnosis of NS-NSCLC remain uncertain. Methods: We tested three commercially available next-generation sequencing (NGS) gene panel assays in circulating cell-free DNA (cfDNA) using replicate sets of 85 LB samples from advanced NS-NSCLC patients at baseline and compared the results with matched TB for the detection of tier IA/B (AMP/ASCO/CAP classification) alterations. We used methylation testing of cfDNA to assess the tumor fraction (TF). We correlated molecular data with clinico-pathological features and follow-up data. Results: All TB-based NGS results were interpretable. 5 LB-based NGS results for one assay were not performed due to the material exhaustion. Tier IA/B gene alteration was detected by LB-based NGS and TB-based NGS in 18 to 23 (21 to 27%) patients and in 35 (41%) patients, respectively. The positive percent agreement of each LB-based NGS assay method compared with TB-based NGS assay ranged from 51% to 68%. The incremental add provided by first LB-based NGS ranged from 0 to 23% versus 25 to 45% for first TB-based NGS, depending on the LB-based NGS assay. It includes one EGFR uncommon mutation, 9 MET amplifications and 2 HER2 amplifications detected by LB-based NGS assays that were not detected by TB-based NGS. Overall agreement between the three LB-based NGS assays was 44%. Hybrid-capture based assay detects 14 MET/HER2 CNVs, two METex14 skipping mutation and two ALK fusions that were not detected by amplicon-based assay. Amplicon-based assay detects one uncommon EGFR mutation, one KRAS non-G12C and one ALK fusion that were not detected by hybrid-capture based assay. 64 LB samples were tested for cfDNA methylation. Positive predictive value and negative predictive value of the detection of TF by cfDNA methylation profiling was 90% and 92%, respectively. The only false-negative result of TF evaluation by cfDNA methylation profiling was observed for a variant with a cfDNA VAF < 1%. Correlation analyses with clinicopathological data are ongoing. Conclusions: While concordance with tissue is relatively high, the incremental add provided by LB-based NGS at diagnosis of advanced NS-NSCLC compared to TB-based NGS is limited and greatly varies according to the sequencing method used. Thus, LB-based NGS at diagnosis of NSCLC can be used as a complement of TB-based NGS or as an alternative when tissue is not available.