Analytic and Clinical Validation of a Pan-Cancer NGS Liquid Biopsy Test for the Detection of Copy Number Amplifications, Fusions and Exon Skipping Variants

Diagnostics

Madeline Bondy¹  Sarah Kreston¹  Brittany D’Alessio¹  Leisa Jackson¹  Chérie Tschida¹  Audrey Audetat¹  Hestia Mellert¹  Ubaradka G. Sathyanarayana¹  Adam Stephen¹  Niki Givens¹  Gary A. Pestano¹ 

[1] Biodesix Inc., 2970 Wilderness Place Suite 100, Boulder, CO 80301, USA;

关键词: non-small cell lung cancer (NSCLC);    liquid biopsy;    cell-free nucleic acid (cfNA);    next generation sequencing (NGS);    CNA (copy number amplifications);    fusions;   

DOI  :  10.3390/diagnostics12030729

来源: DOAJ

【 摘 要 】

Liquid biopsies are an integral part of the diagnosis of cancer. Here, we have extended previous validation studies of a new targeted NGS panel to include the detection of copy number amplifications (CNAs), fusions, and exon skipping variants. Detection of these gene classes included specimens from clinical and healthy donors and cell lines (fusions: ROS1, EML4-ALK, NTRK1; exon skipping: MET exon 14; CNAs: HER2, CDK6, EGFR, MYC, and MET). The limit of detection (LOD) for fusion/skipping was 42 copies (QC threshold was three copies) and was verified using three additional fusion/skipping variants. LOD for CNAs was 1.40-fold-change (QC threshold = 1.15-fold change) and was verified with three additional CNAs. In repeatability and intermediate precision (within lab) studies, all fusion/skipping variants were detected in all runs and all days of testing (n = 18/18; 100%); average CV for repeatability was 20.5% (range 8.7–34.8%), and for intermediate precision it was 20.8% (range 15.7–30.5%). For CNAs, 28/29 (96.6%) copy gains were detected. For CNAs, the average CV was 1.85% (range 0% to 5.49%) for repeatability and 6.59% (range 1.65% to 9.22%) for intermediate precision. The test panel meets the criteria for being highly sensitive and specific and extends its utility for the serial detection of clinically relevant variants in cancer.

【 授权许可】

Unknown