Abstract 4666: Revolutionizing clinical care using prospective, institution-wide tumor sequencing

Abstract

Abstract Identification of predictive and prognostic biomarkers is central to clinical oncology. Use of targeted next generation sequencing (NGS) is increasing in molecular diagnostics labs, however the feasibility and impact of its routine application across all tumor types is largely untested. We launched an institution-wide effort to generate targeted NGS data (Oncopanel) for invasive tumors of all consenting patients. We hypothesized that this approach could replace traditional targeted testing, generate robust data on copy number alterations and structural variants, and provide novel patient-specific observations to facilitate clinical trial enrollment. Illumina NGS was performed on libraries prepared with Agilent SureSelect custom-designed hybrid capture of 4430 exons from 275 genes plus selected introns of 30 genes. Results were analyzed by an internally-developed computational pipeline for mutations, small insertions/deletions, copy number variation and rearrangements. Variants were characterized according to predictive, prognostic, or diagnostic actionability. Data is available for the first 4291 cases sequenced. Oncopanel succeeded in 96% of specimens with adequate DNA. In a subset analysis performed on the first 1000 cases, assay success ranged from 83-100% according to tumor type; breast carcinoma was most prone to failure (p<0.0001). Median number of mutations per case was 8 (range 0-205) and was lowest in endocrine malignancies and highest in skin malignancies. Three percent of tumors were hypermutated with mutational signatures revealing distinct pathogenic underpinnings, including prior temozolomide therapy, microsatellite instability, and UV exposure. Compared to clinical testing, Oncopanel showed 100% accuracy for detection of KRAS and BRAF point mutations in colon adenocarcinoma, EGFR exon 19 deletion mutations in lung adenocarcinoma, and for EGFR amplification in glioblastoma. Oncopanel was 97.5% sensitive and 87.5% specific for 1p19q deletion as compared to aCGH or FISH and was 80% sensitive and 100% specific as compared to ALK FISH and additionally detected 2 ALK rearranged tumors for which FISH failed. In the overall cohort, 26% of patients had an actionable variant (most commonly KRAS) and 39% had alterations with implications for clinical trial enrollment (most commonly in PI3K/PTEN/AKT pathway). TP53 was the most commonly altered gene overall and the most likely to be co-mutated with oncogenic drivers. High level amplifications were most common for EGFR, MDM2, CDK4, ERBB2, MYC, and CCND1. Two-copy deletions were most common for CDKN2A/B, followed by TP53 and PTEN. In several cases, Oncopanel data uncovered alterations that informed diagnosis and treatment of difficult-to-classify tumors. These efforts demonstrate that high quality, high throughput NGS data can be generated prospectively on an institutional level, thereby informing disease course and therapeutic options at an unprecedented scale. Citation Format: Lynette M. Sholl, Elizabeth Garcia, Yonghui Jia, Matthew Ducar, Bernard Fendler, Priyanka Shivdasani, Frank C. Kuo, Azra H. Ligon, Barrett J. Rollins, Neal I. Lindeman, Laura E. MacConaill. Revolutionizing clinical care using prospective, institution-wide tumor sequencing. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4666. doi:10.1158/1538-7445.AM2015-4666