Abstract 4342: Ultra-deep next generation sequencing (NGS) of plasma cell-free DNA (cfDNA) from patients with advanced lung cancers: results from the Actionable Genome Consortium

Introduction: Noninvasive genotyping using plasma cfDNA from cancer patients has the potential to obviate the need for some biopsies while also characterizing disease heterogeneity. This study was undertaken to develop an ultra-deep plasma NGS panel for patients with non-small cell lung cancers (NSCLC). Methods / Results: Plasma was prospectively collected from 51 patients with advanced, progressive NSCLC and a known oncogenic driver from prior tumor genotyping. We performed ultra-deep NGS on extracted cfDNA using a customized Illumina library preparation, hybrid capture panel covering 37 lung cancer related genes (complete exons and partial introns), and ultra-deep sequencing (HiSeq4000). Mean sequencing depth was ∼50,000X (150 million, 150bp reads per sample). After specialized consensus-based error correction for low allele frequency (AF) genomic alterations, the median unique DNA molecules per position were ∼3,500. The mean sequence error rate was reduced by 20-fold to 0.002%, enabling the confident call of a driver mutation as low as 0.03%. In a subset of cases, paired plasma droplet digital PCR (ddPCR) was performed for common EGFR and KRAS mutations using a validated assay. Blinded to tumor genotype, plasma NGS detected SNVs (EGFR, KRAS, BRAF), indels (EGFR, ERBB2), and fusions (ALK, ROS1) as well as significant copy number gains (CNG) (ERBB2, MET). Sensitivity of cfDNA for the detection of known oncogenic drivers was 88% (45/51). A single false positive driver mutation was identified in a case with a known EGFR mutation in tumor; plasma NGS found both EGFR exon 19 del (0.88% AF) and KRAS G12D (2.65% AF), and plasma ddPCR confirmed the presence of both mutations (2.2% and 2.0% AF). Evaluation for an occult second primary is ongoing. In 22 EGFR, ALK, or ROS1 cases with acquired resistance to targeted therapy, plasma genotyping detected a range of potential resistance mechanisms: EGFR T790M and C797S, ALK F1174C, ERBB2 CNG, MET CNG. In 16 cases with paired resistance biopsies, concordance for EGFR T790M status was 94% (15/16). 18 cases with known EGFR or KRAS mutations underwent paired ddPCR. In 14 cases the driver mutation was detected using both assays with high concordance of the%AF (r = 0.91). The remaining 4 cases were negative with ddPCR but 3 were positive with NGS at low AF (0.04%, 0.08%, and 0.29%), and the specificity for each driver was 100%. Conclusions: Ultra-deep plasma NGS can detect a wide range of oncogenic drivers in NSCLC and may be more sensitive than established ddPCR assays. In the setting of acquired resistance to targeted therapy, plasma NGS reliably captured EGFR T790M and additional somatic alterations as potential resistance mechanisms. Citation Format: Bob T. Li, Filip Janku, Pasi A. Janne, Gordon B. Mills, Kiran Madwani, Ryan S. Alden, Cloud P. Paweletz, Marc Ladanyi, Alex Aravanis, Byoungsok Jung, Sante Gnerre, Amy J. Sehnert, David B. Solit, Gregory J. Riely, Geoffrey R. Oxnard. Ultra-deep next generation sequencing (NGS) of plasma cell-free DNA (cfDNA) from patients with advanced lung cancers: results from the Actionable Genome Consortium. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4342.