OA06.01 Clinical Utility of Circulating Tumor DNA (ctDNA) Analysis by Digital next Generation Sequencing of over 5,000 Advanced NSCLC Patients

Abstract

Detection of actionable genomic alterations is now required for NCCN guideline-compliant work-up of NSCLC adenocarcinoma. Next-generation sequencing (NGS) of ctDNA, if sufficiently sensitive and specific, could provide a non-invasive, comprehensive genotyping platform relevant to clinical decision-making when tissue is insufficient or at time of progression on targeted therapies. A highly accurate, deep-coverage (15,000x) ctDNA plasma NGS test targeting 54-70 genes (Guardant360) was used to genotype 5,206 advanced-stage NSCLC patients accrued between 6/2014 – 4/2016. The frequency and distribution of somatic alterations in key genes were compared to those described in TCGA (Pearson and Spearman correlations). The clinical impact of ctDNA testing was evaluated by identification of resistance mechanisms emergent at progression on targeted therapies, and through analysis of additional driver mutations detected by ctDNA at baseline in 362 consecutive NSCLC patients with tissue mutation data available. The positive predictive value (PPV) of ctDNA sequencing was assessed in 229 patients with known tumor driver alterations. ctDNA alterations were detected in 86% of cases; EGFR mutations in 25%, KRAS mutations in 17%, MET amplification in 4%, BRAF mutations in 3% and other rare but potentially actionable alterations in 9%. Mutation patterns among driver oncogenes were highly consistent with those from TCGA (Pearson r=0.92, 0.99, 0.99 for EGFR, KRAS, and fusion breakpoint location). PPV of ctDNA-detected variants was 100% for EGFRL858R, 98% for EGFRE19del, 96% for ALK, RET, or ROS1 fusions, and 100% for KRASG12/G13/Q61 mutations. In 362 cases with tissue information available, 63% (229/362) were tissue quantity-insufficient or undergenotyped (QNS/UG). ctDNA analysis identified driver mutations in 51 of the 229 QNS/UG cases, a 38% increase in detection rate over tissue alone. Among 1,111 EGFR-mutant cases, resistance mutations were identified at progression at frequencies consistent with published literature: EGFRT790M 47%, MET amp 5%, ERBB2 amp 5%, FGFR3 fusions 0.4%, ALK/other fusions 1%, BRAF mutations 1.8%, PTEN inactivation 2.5%, NF1 inactivation 3%, RB1 inactivation 3%, KRAS mutations 1.9%. In 143 consecutive NSCLC patients with detailed follow-up and serial analysis seen at the UC Davis Cancer Center, informative driver mutations were observed in 48 (34%). This series represents the largest NSCLC ctDNA study to date. Genotypic patterns of truncal mutations were highly consistent with TCGA in terms of frequency and distribution. At baseline, ctDNA augmented tissue analysis by identifying additional, actionable mutations when tissue was QNS/UG. ctDNA NGS conducted at progression identified emergent resistance mutations that could inform subsequent courses of therapy.