Abstract 5741: Highly specific copy number variant-flagging algorithm using next-generation sequencing in cancer-predisposition genes

Abstract

Abstract Background: Hereditary cancer accounts for 5% to 10% of all cancers. In addition to sequence variants, copy number variants (CNVs) are a cause of inherited cancer syndromes. Next-generation sequencing (NGS) technology enables simultaneous interrogation of multiple genes in a cost-effective manner. However, accurate CNV detection by NGS remains challenging owing to bias and noise that distort the association between copy number and read-depth. We previously developed and validated an NGS-based CNV detection algorithm using 18 CNV-positive and 85 CNV-negative specimens; the algorithm yielded 100% sensitivity and 98% specificity. In this study, we evaluated the performance of the algorithm on a large group of clinical specimens. Methods: Retrospective CNV analysis was performed on 19,722 specimens submitted for a 34-gene cancer predisposition panel (n=5,726) or 4 sub-panels: BRCA1/2 (n=13,156), a 5-gene Lynch syndrome panel (n=553), a 13-gene colorectal cancer panel (n=205), or a 14-gene women's health panel (n=82). For NGS, target genes were captured using an RNA-bait hybridization method and sequenced on an Illumina NextSeq 500 instrument. CNV-positive specimens were identified using a CNV-flagging algorithm developed in-house that is based on the well-established NGS read-depth approach. All specimens flagged as having CNVs by NGS were reflexed to array comparative genomic hybridization (aCGH) using a custom-designed Agilent array for confirmation. Although highly reliable and widely used, aCGH can only simultaneously process a limited number of specimens. Results: Among the 19,722 clinical specimens, the NGS CNV flagging algorithm identified 485 CNVs; 235 were positive by aCGH and 250 were negative; thus, specificity of the algorithm alone was 98.7% (19,472/19,722) and improved to 100% when aCGH was incorporated. Of 235 aCGH-confirmed CNVs, 224 (95.3%) were pathogenic or likely pathogenic. Approximately 1% of specimens tested by a 34-gene panel (60/5,726) was CNV-positive. BRCA1 (20% of positives) was the most frequently mutated gene followed by CHEK2 (15%), ATM (10%), and PALB2 (10%). From the 4 sub-panels, the 5-gene lynch panel had highest positive CNV rate (4%, 22/553), followed by 13-gene colorectal panel (2%, 4/205), and BRCA1/2 test (1%, 149/13,156). Conclusions: The developed CNV-flagging algorithm enabled CNV detection of clinical specimens, in a variety of panels, with high specificity. Supplementation of aCGH confirmation improved CNV detection specificity to 100%. Our data underscore the important clinical utility of NGS-based CNV detection in the molecular diagnosis of hereditary cancer. Citation Format: Sun Hee Rosenthal, Ke Zhang, Yan Liu, Renius Owen, Alla Smolgovsky, Arlene Buller-Burckle, Felicitas Lacbawan. Highly specific copy number variant-flagging algorithm using next-generation sequencing in cancer-predisposition genes [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5741.